Article pubs.acs.org/Langmuir

New Micellar Transfection Agents Christian Wölk,*,† Dorota Pawlowska,‡ Simon Drescher,† Anna Auerswald,† Annette Meister,§ Gerd Hause,∥ Alfred Blume,⊥ Andreas Langner,† Gerald Brezesinski,‡ and Bodo Dobner*,† †

Institute of Pharmacy, Martin-Luther-Universitaet (MLU) Halle-Wittenberg, Wolfgang-Langenbeck-Strasse 4, 06120 Halle (Saale), Germany ‡ Max Planck Institute of Colloids and Interfaces, Science Park Potsdam-Golm 14476 Potsdam, Germany § Center for Structure and Dynamics of Proteins (MZP), Biocenter, MLU Halle-Wittenberg, Weinbergweg 22, 06120 Halle (Saale), Germany ∥ Biocenter, MLU Halle-Wittenberg, Weinbergweg 22, 06120 Halle (Saale), Germany ⊥ Institute of Chemistry, MLU Halle-Wittenberg, von-Danckelmann-Platz 4, 06120 Halle (Saale), Germany S Supporting Information *

ABSTRACT: Two novel micelle-forming amino-functionalized lipids (OT6 and TT6) bearing two alkyl chains connected to a large positively charged hexavalent headgroup, which might be interesting polynucleotide transferring agents with the advantage of an easy and reproducible production of micelle dispersions, have been characterized. The critical micelle concentration (cmc) of both lipids has been determined by two different methods, namely, isothermal titration calorimetry (ITC) and 1,6-diphenyl-1,3,5-hexatriene (DPH) fluorescence experiments. In addition, the lipid dispersions were studied as a function of temperature using differential scanning calorimetry (DSC), dynamic light scattering (DLS), Fourier-transform infrared (FT-IR) spectroscopy, and cryo-transmission electron microscopy (cryo-TEM). The OT6 and TT6 micelles effectively complex DNA as determined by ITC and DSC measurements. In addition, DLS and ζ-potential measurements were performed to determine lipoplex formulations that exhibit colloidal stability. Finally, the structures of OT6/ DNA complexes were investigated by means of X-ray scattering and TEM.

1. INTRODUCTION

depending on the application. However, liposomes exhibit stability problems due to fusion processes.12 Furthermore, the up-scaled and reproducible production is sophisticated for both systems. For example, extrusion of cationic liposomes is a critical step due to the adsorption of cationic lipids to polycarbonate membranes. A promising alternative are micellar gene transfer systems. Micellar dispersions are easy to prepare and have a higher colloidal stability compared to liposomes and SLNs. To date, only a few micellar transfection systems are described.13−15 Previously, we described the synthesis of a new class of amino-functionalized lipids: the malonic acid diamides of the second generation.16 Only the lipids with the largest headgroup were able to form micelles in aqueous dispersions. In the present study, two representatives of these lipids, the micelleforming compounds OT6 and TT6, are intensively characterized. The chemical structure of both lipids is presented in Figure 1. The lipids exhibit a large headgroup with three lysine molecules bearing six (hexavalent) primary amino groups for

Efficient delivery of polynucleotides into cells is a task of utmost importance in the laboratory and in medicine. Gene therapy has become an interesting field in medicine,1,2 and gene silencing has become an important tool in biological research.3−5 Since “naked” polynucleotides are sensitive against RNase and DNase, and additionally not internalized by cells in an effective manner, vehicle systems (vectors) are required for the genetic cargo. There are two major classes of polynucleotide vehicles: viral and nonviral vectors. Although the virusbased vehicles are the more effective ones, nonviral vectors came into the focus of research due to the drawbacks of viral gene transfer, such as immunogenicity, size limitations of inserted polynucleotides, limitations for scale up procedures.6,7 Beside cationic polymers, cationic lipid (cytofectin) based vectors became an important tool in nonviral polynucleotide transfer.8 Common lipid mediated polynucleotide transfer is based on cationic liposomes or solid lipid nanoparticles (SLN).8,9 While liposomes were prepared as preformed dispersions before loading them with the genetic materiala characteristic that allows the development of commercially available liposome formulations for flexible use (e.g., Lipofectin, Lipofectamine)10,11SLNs have to be prepared differently © 2014 American Chemical Society

Received: January 13, 2014 Revised: April 3, 2014 Published: April 3, 2014 4905

dx.doi.org/10.1021/la404860w | Langmuir 2014, 30, 4905−4915

Langmuir

Article

Ultrapure water was produced with a Milli-Q Advantage A10 system (Merck Millipore, Billerica, MA, USA). The buffer systems used are HEPES buffer, pH 7.3 (20 mM 2-[4-(2-hydroxyethyl)piperazin-1yl]ethanesulfonic acid (HEPES), 0.1 mM ethylenediaminetetraacetic acid (EDTA)), MES buffer, pH 7.3 (20 mM 2-(N-morpholino)ethanesulfonic acid (MES), 0.1 mM EDTA), Tris buffer, pH 7.3 (20 mM 2-amino-2-hydroxymethylpropane-1,3-diol (Tris), 0.1 mM EDTA), and carbonate buffer, pH 10 (10 mM). Salmon sperm DNA (10 mg mL−1 = 30.3 mMnucleotides in water) was purchased from Sigma-Aldrich (Steinheim, Germany) and diluted with water or buffer to the appropriate concentration. 2.2. Methods. 2.2.1. Micelle Preparation. A lipid dispersion (OT6: 0.84 mM, TT6: 0.88 mM) was prepared with water or buffer as solvent. Although the dispersions were clear and transparent, they were vortexed and treated 5 min at 60 °C with ultra sound (37 kHz). The lipid dispersions were stored 24 h at 7 °C before use. If lower concentrations were needed for the experiments, the lipid dispersions were diluted with the appropriate amount of solvent and vortexed. 2.2.2. cmc Determination. Method 1 − ITC: The experiments were performed as described in section 2.2.3. The sample cell was filled with HEPES buffer (pH 7.3), and the syringe was filled with lipid dispersion (0.84 mM OT6, 0.88 mM TT6). Method 2 − DPH inclusion: 199 μL lipid dispersion in HEPES buffer (with increasing concentration of the cationic lipid) were mixed with 1 μL DPH solution (1 mM in THF) and incubated for 30 min at 25 °C under light exclusion. The fluorescence intensity was measured (λex: 355 nm; λem: 460 nm). The measurements were performed three times. 2.2.3. ITC. Experiments were performed in buffer (HEPES, MES, Tris) with a CSC Model 5300 Nano-ITC III (TA Instruments, New Castle, DE, USA). Sample cell volume was 950 μL, and the syringe volume was 250 μL. Experiments were performed at 25 °C while stirring with 200 rpm. The samples were degassed for 10 min before each experiment. The following concentrations were used. Lipid-toDNA titrations: DNA − 0.45 mMnucleotides; lipid dispersion − 0.84 mM OT6, 0.88 mM TT6; DNA-to-lipid titrations: DNA − 4.85 mMnucleotides; lipid dispersion − 0.26 mM OT6, 0.27 mM TT6. Lipid or DNA were injected in steps of 10 μL in an interval of 600 s (after this time, the cell feedback system has been returned to the baseline). Heats of dilution (control titrations) were evaluated by injection of lipid or DNA into the cell filled with pure buffer and used for heat correction of the ITC results. The heat correction by dilution heats measured at the end of the scan showed comparable results. Data were evaluated with the NanoAnalyze software version 2.2.0 (TA Instruments, New Castle, DE, USA). The reference cell was filled with pure buffer. The titrations were performed twice ensuring reproducibility. 2.2.4. Sample Preparation for DLS- and ζ-Potential Measurements of Lipoplexes. The lipoplexes were prepared in a one-step procedure using concentrations comparable to the ITC experiments. The following titration sequences were used: (a) lipid-to-DNA titration: (1) DNA stock solution, (2) buffer, (3) lipid stock solution; (b) DNA-to-lipid titration: (1) lipid stock solution, (2) buffer, (3) DNA stock solution. Incubation time was 15 min at 25 °C. 2.2.5. DLS. Size and size distributions were characterized with a Zetasizer Nano ZS ZEN3600 (Malvern Instruments, Worcestershire, UK). The scattering angle was 173°, and the temperature was kept at 25 °C. Three measurements consisting of 15 runs with a duration time of 20 s for each run were performed. For the calculations, a viscosity η = 0.8872 mPa s and a refractive index of 1.33 were assumed. The autocorrelation function was evaluated by ALV-Correlation software version 3.0 using an exponential regularized fit. 2.2.6. ζ-Potential. The electrophoretic measurements were carried out by means of the laser Doppler electrophoresis technique using a Zetasizer Nano ZS ZEN3600 (Malvern Instruments, Worcestershire, UK). The measurements were performed at 25 °C. Three measurements consisting of 30 runs with a voltage of 50 V were performed. For the calculations, a viscosity η = 0.8872 mPa s, a dielectric constant ε = 78.5 F m−1, and a refractive index of 1.33 were assumed. The analysis was done with the Zetasizer Software 6.34. The mobility μ of

Figure 1. Chemical structures of transfection lipids OT6 and TT6. Theoretical pKa values of the lysine moieties: ε-amino group ∼ 10.3; α-amino group ∼ 8.9.

DNA complexation. The structural difference between the two lipids is the alkyl chain pattern: while OT6 has two different alkyl chains (namely, oleyl and tetradecyl), TT6 exhibits two tetradecyl chains. A detailed characterization of the lipid dispersions of OT6 (N-{2-[N,N-bis(2-{N-[(2S)-2,6-diamino-1oxohexyl]amino}ethyl)amino]ethyl}-N′-(6-{N-[(2S)-2,6-diamino-1-oxohexyl]amino}-1-[N-(9Z)-octadec-9-enylamino]-1oxohexan-(2S)-2-yl)-2-tetradecylpropandiamide) and TT6 (N{2-[N,N-bis(2-{N-[(2S)-2,6-diamino-1-oxohexyl]amino}ethyl)amino]ethyl}-N′-(6-{N-[(2S)-2,6-diamino-1-oxohexyl]amino}1-oxo-1-(N-tetradecylamino)hexan-(2S)-2-yl)-2-tetradecylpropandiamid) by means of differential scanning calorimetry (DSC), dynamic light scattering (DLS), Fourier-transform infrared (FT-IR) spectroscopy, X-ray scattering, and cryotransmission electron microscopy (cryo-TEM) will be shown. The critical micelle concentration (cmc) values for both lipids were determined by two different methods, namely, isothermal titration calorimetry (ITC) and 1,6-diphenyl-1,3,5-hexatriene (DPH) inclusion. Furthermore, the characterization of the DNA-complexing behavior of micellar dispersions of OT6 and TT6 is described in detail. The understanding of this process is of utmost importance for the optimization of systems for gene delivery applications. Detailed ITC measurements gave insights in the complex formation process (titration end point (TEP), process enthalpies). Additionally, ζ-potential measurements, DLS studies, and DSC experiments were performed to investigate the complex formation process. Finally, the structure of OT6/DNA complexes has been investigated by X-ray scattering and TEM studies representative for the new class of micellar transfection systems. The studies prove a little-noticed lamellar lipoplex structure with DNA entrapped between interdigitated lipid bilayers (alkyl chain interdigitation). This work presents a very detailed physicochemical characterization of the lipid aggregates as well as the lipid/ DNA complexes with the aim of possible application as transfection systems. First investigations have shown effective gene transfer (data not shown). Nevertheless, detailed investigations of the biocompatibility of the micellar gene carrier systems are necessary to quantify possible bioeffects of such nanocarriers.17−21 Therefore, transfection efficacy and cytotoxic effects of OT6 and TT6 will be presented elsewhere.

2. EXPERIMENTAL SECTION 2.1. Materials. If not mentioned otherwise, chemicals were purchased from Sigma-Aldrich (Steinheim, Germany). Fluorescence measurements were carried out in 96 well plates from Grainer Bio-One GmbH (Frickenhausen, Germany) at a POLARstar Omega (BMGLabtech, Ortenberg, Germany). 4906

dx.doi.org/10.1021/la404860w | Langmuir 2014, 30, 4905−4915

Langmuir

Article

the diffusing aggregates was converted into the ζ-potential using the Smoluchowski relation ζ = μη/ε. 2.2.7. DSC. Micelle dispersions were prepared in buffer as mentioned above. For the lipoplex dispersions, the DNA concentration was fixed at 3.03 mMnucleotides and the needed amount of lipid results from the different nlipid/nnucleotide ratios. The lipid dispersion was prepared with 810 μL HEPES buffer and stored 1 day. Afterward, 90 μL DNA (30.3 mMnucleotides) was added and incubated for 15 min at 25 °C. The DSC measurements were performed on a MicroCal VP-DSC (MicroCal Inc., Northampton, MA, USA). The heating rate was 60 K h−1, each heating and cooling scan was repeated confirming reproducibility. The first scan was abolished except for DSC experiments with DNA and lipoplexes. The scanned temperature range was between 2 and 95 °C. The reference cell was filled with pure solvent. The buffer−buffer baseline was subtracted from the thermograms of the samples, and the DSC scans were evaluated using MicroCal Origin 8.0 software. 2.2.8. TEM, Small-Angle X-ray Scattering (SAXS) and FT-IR. See Supporting Information (SI).

(except the OT6 dispersion at pH 10). The size species ascribed to micelle aggregates in the intensity weighted distribution disappear in mass weighted curves. Hence, the particles with radii >10 nm are in a negligible amount, but are clearly noticed in the intensity weighted size distribution due to their higher scattering effect compared to the much smaller micelles. The mass weighted size distribution of the OT6 dispersion (0.84 mM) in carbonate buffer at pH 10 shows a small second size species with radii between 10 and 30 nm. Cryo-TEM images of an OT6 dispersion (1.68 mM) in carbonate buffer at pH 10 only show small aggregates seen as black dots (see Figure 2B, white arrows). These aggregates are ascribed to micelles. In conclusion, micelles are the predominant lipid assemblies at these conditions, and larger assemblies, such as aggregates of micelles, are formed in a negligible amount. The sizes of the OT6 and TT6 micelles at pH 7.3 and 10 (radius between 5 and 10 nm) are relatively large for spherical micelles assuming a molecule length of ∼4 nm for a fully expanded lipid. Nonspherical micelle shapes (e.g., discoidal or prolate ellipsoidal) or a large hydration shell can explain the higher experimental values considering that the DLS analysis of the correlation function yields the radius of equivalent hydrodynamic spheres.22 3.1.2. Determination of the Critical Micelle Concentration (cmc). The cmc values were determined by two different methods in HEPES buffer (physiological pH of 7.3). The cmc determination by ITC is based on the detection of the demicellization heat before reaching the cmc in the sample cell (see SI Figures S1−S4).23,24 The second method based on fluorescence dye (DPH) inclusion in the micelles formed above the cmc (see SI Figure S5). DPH, as a lipophilic fluorescence dye, changes its emission maximum in a lipophilic environment.25 The determined cmc values are in the range between 15 and 85 μM (OT6: 49 μM (DPH), 85 μM (ITC); TT6: 15 μM (DPH), 54 μM (ITC); see also SI Table S1.). The cmc values determined by ITC are higher compared to the DPHfluorescence method. This discrepancy could be explained by the evaluation method. The Boltzmann fit used for ITC data determines the transition midpoint (point of inflection; see SI Figure S2,S4), while the linear extrapolation used for DPH data rather determines the beginning of the micellization process (see SI Figure S5). Furthermore, lipophilic foreign molecules like DPH can shift the cmc. A shift to higher cmc values would be expected if the foreign molecule disturbs the micelle assembly. Contrary, lipophilic substances also could support micelle formation. In the case of the investigated lipids, the increasing DPH fluorescence before reaching the cmc indicates the formation of premicellar aggregates between DPH and the lipids (see SI Figure S5). The smooth transitions of the properties measured for both methods (see SI Figures S1−S5) point to a low aggregation number.26 3.1.3. DSC and FT-IR. The temperature-dependent aggregation behavior of the lipid dispersions (OT6: 0.84 mM, TT6: 0.88 mM) at different pH values was studied in the range between 2 and 95 °C by means of DSC (see SI Figure S6A). The DSC heating scans show a very broad and weak transition in the temperature range between 20 and 70 °C in carbonate buffer at pH 10, where most of the primary amino groups are uncharged. In contrast, this transition is not present in HEPES buffer at pH 7.3 (see SI Figure S6A) as well as in acetate buffer at pH 5 (data not shown), where most of the primary amino

3. RESULTS 3.1. Aggregation Behavior of Pure Lipids in Aqueous Dispersions. 3.1.1. Characterization by DLS and Cryo-TEM. To get information about the size of aggregates formed, OT6 and TT6 dispersions (OT6: 0.84 mM, TT6: 0.88 mM) have been investigated by DLS at pH 10 and pH 7.3. Figure 2A

Figure 2. Intensity and mass weighted size distribution given as hydrodynamic radius (r) of OT6 and TT6 samples (OT6: 0.84 mM, TT6: 0.88 mM) in carbonate buffer at pH 10 and HEPES buffer at pH 7.3 measured by DLS (A). Curves are shifted vertically for clarity. Cryo-TEM image of an OT6 dispersion (1.68 mM) in carbonate buffer (pH 10) prepared at 25 °C (B). White arrows point on small aggregates representing micelles. The bar corresponds to 100 nm.

shows the intensity and mass weighted size distribution (hydrodynamic radius (r)) of OT6 and TT6 at pH 7.3 and pH 10. All distributions exhibit a size species with radii smaller than 10 nm, which is attributed to micelles. Beside this small particles, larger size populations with radii between 10 and 1000 nm appear in the intensity weighted size distributions of both lipids at both pH values. These size populations are ascribed to aggregates of micelles. For TT6, these aggregates are larger at pH 10 compared to pH 7.3, and for OT6 these aggregates disappear for the most part at pH 7.3 compared to pH 10. This observation is attributed to an electrostatic stabilization of the micelle dispersions due to the protonation of the amino groups of the lipids by decreasing the pH value. The mass weighted size distributions, which are more representative for polydisperse samples, are monomodal 4907

dx.doi.org/10.1021/la404860w | Langmuir 2014, 30, 4905−4915

Langmuir

Article

Figure 3. (A/B) Integrated heats of reaction (open circles: uncorrected values, filled squares: control titration corrected values) normalized to the amount of injected lipid vs molar ratio of the lipid-to-DNA titration of salmon sperm DNA (0.45 mMnucleotides) with OT6 (0.84 mM) (A) and TT6 (0.88 mM) (B) in HEPES buffer (pH 7.3) at 25 °C. The dotted line marks the previously determined cmc of the pure lipids. (C/D) Integrated heats of reaction (open circles: uncorrected values, filled squares: control titration corrected values) normalized to amount of injected DNA vs molar ratio of the DNA-to-lipid titration of OT6 (0.26 mM) (C) and TT6 (0.27 mM) (D) with salmon sperm DNA (4.85 mMnucleotides) in HEPES buffer (pH 7.3) at 25 °C. A section of the green marked part of the heat flow (Φ) curves is shown for OT6 (E) and for TT6 (F). The diagram insets show the time dependent raw heat flows of the titration process. Positive peaks are exothermic events.

ray scattering patterns show broad halos that are characteristic for lipid alkyl chains in the fluid state (data not shown), which is in line with the above-mentioned FT-IR measurements. Molten chains, i.e., a high amount of gauche-conformers, are shorter than chains in all-trans conformation but this fact cannot explain the observed small d value. The theoretical length of the lipid molecule amounts to ∼40 Å for a fully expanded lipid and ∼34 Å for a single energy minimized lipid. The only reasonable explanation for the small d values is an interdigitation of the fluid alkyl chains within the lipid bilayer. The tendency to alkyl chain interdigitation is given by the larger distance between the two alkyl chains attached to the backbone (see Figure 1). The distance is more comparable to 1,3-diester phospholipids, which are known to form bilayers with interdigitated alkyl chains,29 as with 1,2-diester phospholipids. The SAXS experiments were carried out at high concentrations (factor 2000 above the determined cmc of OT6). In this high concentration region the formation of isotropic micellar phases is implausible, and the existence of liquid crystalline phases is supposable. As mentioned, the SAXS pattern indicates weakly ordered lamellar stacks while we decided to use the GAP-fit. We could not determine whether the lamellar stacks are stacks of discoidal micelles or other lamellar structures. At low concentrations, the existence of micelles was confirmed (see section 3.1.1). 3.2. Characterization of Lipid/DNA Complexes. 3.2.1. ITC Experiments. For a better understanding of the complex formation process, the DNA/cytofectin-interactions were investigated by ITC. This method is commonly used for the investigation of binding processes and allows a quantitative and thermodynamic evaluation of these processes, e.g., peptide/ ligand-interactions.30 It can also be used for the investigation of complex formation between cationic lipids and DNA.31−37 The

groups are in the protonated state (positively charged ammonium groups). Temperature dependent FT-IR measurements (10 °C − 80 °C) have been performed with two OT6 dispersions (42 mM in D2O and 84 mM in carbonate buffer 300 mM at pH 10). The observed wavenumbers (υ) of the asymmetric (as) methylene stretching vibrational band are typical for fluid alkyl chains with a high amount of gauche-conformers (carbonate buffer: υ[CH2]as ≈ 2924−2925 cm−1; D2O: υ[CH2]as ≈ 2925−2926 cm−1; see SI Figure S6B,C). The wavenumbers increase slightly with increasing temperature. Due to these observations, the broad and weak transition in the DSC curves at pH 10 is not attributed to a distinct change in the fluidity of the alkyl chains. Instead, conformational changes in the headgroup region or aggregation effects between the micelles could explain the broad transition in the DSC heating curves at pH 10. Temperature-dependent DLS measurements of OT6 and TT6 dispersions (OT6: 0.84 mM, TT6: 0.88 mM) in carbonate buffer at pH 10 and HEPES buffer at pH 7.3 show micelle aggregation with increasing temperature at pH 10, while the dispersion at pH 7.3 does not tend to aggregate due to electrostatic stabilization (data not shown). 3.1.4. SAXS. The SAXS pattern of OT6 dispersed in HEPES buffer (pH 7.3) to 168 mM at 25 °C shows one broad peak (see SI Figure S7A). The diffuse scattering curve results from weakly correlated fluid bilayers. The peak was fitted with GAPfit (GAP 1.3 program by Georg Pabst) based on a modified Caille theory.27,28 This theory takes the bending fluctuation of fluid bilayers into account and is applicable to the Lα and SmA phases. As a result of the fit, an electron density profile is obtained. The d value (thickness of a lipid bilayer with its hydration layer) calculated from the electron density profile is 46.3 Å (see SI Figure S7B). The corresponding wide-angle X4908

dx.doi.org/10.1021/la404860w | Langmuir 2014, 30, 4905−4915

Langmuir

Article

Table 1. Comparison of the Enthalpies and Titration End Point (TEP) Determined by ITC with the Isoelectric Point (IEP) Determined by ζ-Potential Measurements of OT6 and TT6a ΔHTot ± s [kJ mol−1]b

ΔHComp [kJ mol−1]b,c

bound protons per mole titrating agentd

(ITC) TEPe [nlipid/ nnucleotide]

(ζ) IEPf [nlipid/ nnucleotide]

0.45 DNAnucleotides 0.26 lipid

−26.5 ± 3.6

−48.6

1.5

0.31

0.33

−2.2 ± 0.4

−4.6

0.2

0.37

0.31

0.45 DNAnucleotides 0.27 lipid

−35.8 ± 2.7

−56.2

1.5

0.25

0.23

−3.1 ± 0.9

−6.1

0.2

0.28

0.28

lipid

csyringe [mM]

ccell [mM]

OT6

0.84 lipid

TT6

4.85 DNAnucleotides 0.88 lipid 4.85 DNAnucleotides

The results are given for lipid-to-DNA and DNA-to-lipid titrations. The solvent was HEPES buffer pH 7.3. bkJ mol−1lipid when syringe was filled with lipid, kJ mol−1nucleotide when syringe was filled with DNA. cProcess enthalpies without buffer dissociation enthalpies (Comp = complex building process). Calculated with eq 2. dCalculated with eq 1. eTitration end point determination; see SI Figure S8 and Figure S9. fDetermined by sigmoidal Boltzmann fit; see SI Figures S10−S13. a

method is applicable for the lipoplex formation due to the favorable time scale of this process (seconds to few minutes). For slower process (minutes to hours), ITC is not appropriate due to the instrumental drift.32 First, the lipid-to-DNA titrations of OT6 and TT6 were studied in HEPES buffer at pH 7.3. The raw heat flow-timeprofiles, raw titration heats, and control-corrected titration heats are presented in Figure 3A,B. The corresponding total process enthalpies (ΔHTot) and the titration end points (TEP) are summarized in Table 1. At the beginning of the titration, only exothermic processes occur, the reaction enthalpy being much more negative than the one observed for lipid injection into buffer. This indicates an exothermic reaction for the complex formation. At nlipid/nnucleotide near the TEP endothermic processes arise. The shapes of the titration curves of lipid OT6 and TT6 are comparable, but, both lipids differ in their TEP and ΔHTot values (see Table 1). We also performed the reversed titration experiment. Figure 3C−F summarizes the raw heat flow-time-profiles and the measured process heats of the DNA-to-lipid titrations of OT6 (Figure 3C,E) and TT6 (Figure 3D,F). The processes are exothermic for the whole titration curve up to the TEP. As in the above-mentioned lipid-to-DNA titration, OT6 and TT6 also differ only in the TEP and in the ΔHTot values (see Table 1). In contrast to the lipid-to-DNA titration, every injection before reaching the TEP consists of two different processes: A fast exothermic process overlapping with a slower endothermic process (see Figure 3E,F). The DNA-to-lipid titrations and the lipid-to-DNA titrations result in different values of ΔHTot, TEP, and different shapes of titration curves. The two titration modes vary in the concentration ratios between DNA and lipid at the beginning of the titration. If the lipid is titrated to the DNA solution, a small number of lipids interact with an excess of DNA, so that the cationic lipids occupy only a small amount of the DNA binding sites. In the progress of the titration, more and more cationic lipids bind to the free binding sites of the DNA resulting in a lipoplex formation over the whole titration process until the TEP is reached. In contrast, during the DNAto-lipid titration, a small number of DNA molecules interact with an excess of lipids (monomers and micelles). As a consequence, a DNA molecule is compacted in a one-step mechanism by the lipids including all interaction and rearrangement processes. This complexation experiment seems to require more cationic lipids compared to the lipidto-DNA titration.

Similar overlapping exo- and endothermic peaks as observed in the DNA-to-lipid titrations were already observed before by Pector et al.34 They explained the biphasic behavior in the following way: The fast exothermic part reflects the electrostatic component of the interaction (electrostatic attraction between phosphate groups of DNA and protonated amino moieties of the cationic lipid) and the second endothermic part reflects the lipid and DNA rearrangement, which results in the final lipoplex structure. However, we suggest that in our experiments the mentioned rearrangement process is the coagulation and precipitation of charge neutral lipoplexes. This scenario can be adjusted to the lipid-to-DNA titration: In such experiments, charge neutral complexes occur only at the TEP, where endothermic process heats were observed. At the end of the titration experiment, a precipitate was found in the reaction vessel, as often mentioned in the literature, too.31,33,34 In most ITC studies on DNA interaction with cationic lipids, endothermic processes are observed.31−33,35,36 One exception is described for DOTAP/DOPE vesicles.32 In this case, the exothermic total process enthalpies were attributed to the protonation of the phosphatidylethanolamine amino group of the DOPE. The above-described ITC experiments also result in exothermic process enthalpies. To check for possible changes in protonation state of the lipid during the lipid/DNA complex formation, additional ITC titration experiments were performed in MES and Tris buffer. MES, HEPES, and Tris have different buffer dissociation enthalpies (ΔHDiss) (MES: ΔHDiss = 12.7 kJ mol−1, HEPES: ΔHDiss = 16.4 kJ mol−1, Tris: ΔHDiss = 47.3 kJ mol−1).38 Figure 4 shows the comparison between the lipid-to-DNA titrations (Figure 4A) and the DNA-to-lipid titrations (Figure 4B) of OT6 in the three different buffer systems. TT6 shows a comparable behavior (data not shown). For both titration types, the curves of the three buffer systems differ in the measured total heats until the TEP is reached. While the ΔHTot values of corresponding titrations in HEPES buffer and MES buffer are exothermic, the titrations in Tris buffer show endothermic ΔHTot values. The TEP is identical for comparable titrations (lipid-to-DNA or DNA-to-lipid) in all the three buffer systems. Such a behavior indicates a proton transfer during the complex formation while the proton activity in the solution is kept constant by the buffer system. But how do we explain such protonation effects during complex formation? A transfer of the lipid amino and the DNA phosphate groups, respectively, from the aqueous phase into a medium of lower dielectric constant, such as in the lipid/DNA complex itself, destabilizes the charged and stabilizes the uncharged form of such (de)protonable groups resulting in 4909

dx.doi.org/10.1021/la404860w | Langmuir 2014, 30, 4905−4915

Langmuir

Article

lipid headgroup during the lipid/DNA complex formation can break such hydrogen bonds, which results in a free α-amino group available for protonation reactions. 3.2.2. ζ-Potential and DLS Measurements. ζ-potential studies were carried out to prove whether the TEP, determined by ITC, is identical with the IEP, where all negative charges of DNA are neutralized with lipids. In order to compare the results with ITC experiments, ζ-potential studies were performed with the same DNA and lipid concentrations. Figure 5 compares the heats determined by ITC (squares) with the ζ-potential (circles) as a function of the molar ratio for

Figure 4. Integrated control corrected heats of reaction between OT6 and DNA in three different buffer systems at 25 °C at pH 7.3: HEPES buffer (filled squares), MES buffer (open circles), and Tris buffer (filled triangles). (A) heats normalized to amount of injected lipid vs molar ratio of the lipid-to-DNA titration of salmon sperm DNA (0.45 mMnucleotides) with OT6 (0.84 mM). (B) heats normalized to amount of injected DNA vs molar ratio of the DNA-to-lipid titration of OT6 (0.26 mM) with salmon sperm DNA (4.85 mMnucleotides).

changes of the apparent pKa value (pKaapp). Furthermore, negatively charged surfaces, such as DNA, accumulate protons. The binding of amino-functionalized lipids to the DNA binding sites will change the degree of protonation. A plot of the overall reaction enthalpy (ΔHTot) versus the buffer dissociation enthalpy (ΔHDiss) allows the determination of the number of protons that migrate during the complex formation process (n)39 using the following equation: ΔHTot = nΔH2 + ΔH1 + nΔHDiss

Figure 5. Comparison of ζ-potential (open circles) and heats measured by ITC (filled squares) of the lipid-to-DNA titration of OT6 (A) and TT6 (B) as well as the DNA-to-lipid titration of OT6 (C) and TT6 (D) in dependence of the molar ratio in HEPES buffer pH 7.3. The horizontal line marks the neutral point of the ζ-potential.

(1)

Herein, ΔH2 is the dissociation enthalpy of a reaction partner taking up/releasing the proton and ΔH1 the association enthalpy. The complex formation process is accompanied by an uptake of protons (positive n) for the both titration modes (see Table 1). During the lipid-to-DNA titration, approximately 1.5 protons were bound per titrated lipid molecule at pH 7.3, whereas during the DNA-to-lipid titration around 0.2 protons were bound per nucleotide. Theoretically, the amino groups of the lipids can be protonated. The amino groups of DNA-bases are integrated within hydrogen bonds in the DNA interior core, where the conditions should not be affected by the complexation process and, in consequence, they should not be responsible for proton uptake and phosphate groups of DNA (pKa value of phosphoric acid diester ∼2) should be deprotonated at pH 7.3. The εamino group of lysine (pKa value ∼10.3) should be protonated at pH 7.3. In contrast, the protonation degree of the α-amino group (pKa value ∼8.9) can be affected by slight changes in the microenvironment, so this amino group is probably responsible for the proton uptake. Furthermore, one lipid exhibits three αamino groups in lysine moieties. We have to consider differences between their pKaapp values due to conformational effects. For example, hydrogen bonds with neighboring carbonyl oxygen are possible. Conformational changes in the

both lipids (OT6 and TT6). For both, the lipid-to-DNA titration (see Figure 5A,B) as well as the DNA-to-lipid titration (see Figure 5C,D), the ζ-potential curves have a sigmoidal shape and intersect the 0 mV line (IEP). In all four cases, the lipoplexes exhibit a positive or negative overall charge at molar ratios before and after the IEP. The positive plateau is at about 30 mV if the lipid is in excess. The negative plateau is at about −40 mV if the DNA is in excess. These kind of overcharging phenomena are also described for other lipid/DNA complexes.40,41 The results of the ζ-potential measurements clearly demonstrate that the TEP determined by ITC (for lipid-toDNA as well as for DNA-to-lipid titrations) is identical with the formation of isoelectric lipid/DNA complexes (IEP) (see Figure 5 and Table 1). The curves can be used to determine molar ratios at which lipoplex dispersions were colloidal stable. These are molar ratios in the plateau regions of the ζ-potential curve. According to the DLVO theory,26,42 low charged and neutral lipoplexes tend to coagulate. DLS measurements are in line with this theory (see SI Figure S14 and S15). Around the IEP, particle sizes above 2 μm were obtained, and precipitation 4910

dx.doi.org/10.1021/la404860w | Langmuir 2014, 30, 4905−4915

Langmuir

Article

occurs. Molar ratios in the range of the plateaus of the ζpotential curves lead to particles with radii smaller than 200 nm in nearly all cases. Theoretically, the incorporation of excess lipid into the neutral lipid/DNA complexes should be detectable in ITC measurements (IEP ≠ TEP). However, Caracciolo et al. demonstrated that only lipoplexes generated in a single mixing step can incorporate excess lipid.43 In the ITC experiments, multiple-step complex formation processes take place for the lipid-to-DNA titration resulting in charge neutral lipoplexes that coagulate and precipitate (particle sizes above 1 μm where measured after the ITC experiments). As a consequence, no overcharging occurs and IEP = TEP. In the DNA-to-lipid ITC titration, every titration step is a single lipoplex formation process in huge lipid excess. Therefore, we would expect positive overcharging and colloidal stable lipoplexes. However, a precipitate with particle sizes above 1 μm occurs after the ITC experiments. This can be explained by lipid exchange between positive lipoplexes with lipid excess and uncomplexed DNA, which is added during the progress of the titration. Finally, this lipid exchange results in neutral lipoplexes which coagulate, and, as consequence, in the observation IEP = TEP. 3.2.3. DSC Investigations of Lipid/DNA Complexes. The DSC curves of pure salmon sperm DNA and of lipoplexes with different nlipid/nnucleotide ratios (0.08, 0.17, 0.33, 0.67, and 1, respectively) are shown in Figure 6. The corresponding Tm

not shown). No transition can be seen in the subsequent heating scans (scan 2−5) showing that the single strands do not anneal into double strands in the time scale of the performed DSC experiments. The ability of DNA to anneal to dsDNA after melting in DSC experiments depends on the cooling rate and the equilibration time before the next heating cycle. The parameters of our DSC experiments allow detecting shifts in the DNA melting temperature of the investigated lipoplexes by comparing the first heating scans (scan 1) of different lipid/DNA-mixtures. In addition, conclusions about the annealing time can be made by examination of the subsequently performed heating scans (scan 2−5). The addition of lipid to DNA changes the DSC curves in a similar way for both lipids. At the nlipid/nnucleotide ratios of 0.08 and 0.17, two endothermic transitions around 60 and 80 °C (see Figure 6 and SI Figure S16) are detected in the first heating scan. The only differences between the two lipids are the ΔH values (transition enthalpies) (see SI Figure S16). Scan 5 shows only two broad and weak transitions in the range between 40 and 90 °C. In the first heating scan at nlipid/nnucleotide ratio of 0.33, the lower temperature transition (∼60 °C) disappears while the high temperature transition remains with Tm values different for both lipids (80 °C for OT6 and 85 °C for TT6). Scan 5 still shows the two broad and weak transitions in the range between 40 and 90 °C. At nlipid/nnucleotide ratio of 0.67 and 1, the DSC curves are the same for both lipids: An endothermic transition with a Tm value of ∼87 °C, which clearly appears in all 5 heating scans is observed. Only the transition enthalpy decreases in the first scans but remains constant for scans 3−5 (data not shown). Additionally, a broad and weak transition, which was also observed for the other nlipid/nnucleotide ratios, still appears. We assume that the observed endothermic transitions show the melting of salmon sperm DNA. The broad and weak transitions in the range between 40 and 90 °C appearing in scans 2−5 are assigned to the melting of partially annealed DNA strands after cooling. The lipoplex formation of DNA with the hexavalent lipids leads to an increasing melting temperature. This assumption is supported by the two transitions observed at nlipid/nnucleotide ratios of 0.08 and 0.17. By comparing the Tm values, it is obvious that the transition at lower temperature is comparable with the melting temperature of the pure and uncomplexed DNA. The high temperature transition can be attributed to the DNA in complex with lipid, a phenomenon that is also described in the literature for cationic lipids and polymers.45,46 The area under the high temperature peak increases while the area under the low temperature peak decreases with ongoing complexation process. Additionally, this observation is in line with the results of the ITC and ζ-potential measurements, indicating that at nlipid/nnucleotide ratios of 0.08 and 0.17, the DNA is not completely complexed (compare with TEP and IEP in Table 1). At these nlipid/nnucleotide ratios, DNA in complex with lipid and free bulk DNA coexist. Furthermore, the observed transitions are not reversible in the time scale of the DSC experiment. This indicates that the DNA in complex with lipid also does not anneal to dsDNA under the chosen DSC conditions after denaturation. At an nlipid/nnucleotide ratio of 0.33, the entire DNA is complexed with a transition around 80 °C. This nlipid/nnucleotide ratio is close to the IEP of the lipid/DNA complexes (see Table 1). However, the complexed DNA still denatures irreversibly under the conditions of the DSC experiments. Another

Figure 6. DSC heating scans from 2 to 95 °C of salmon sperm DNA pure and in complex with different amounts of OT6 and TT6 in HEPES buffer pH 7.3. The DNA concentration was 3.03 mMnucleotides. The amount of lipid is given with the nlipid/nnucleotide ratio. Curves of each nlipid/nnucleotide ratio are shifted along the y-axis for clarity. The curves of each nlipid/nnucleotide ratio are not shifted among themselves. In the right of the curves, the scan number is given. Scan 5 is shown as a representative of scans 2−5. The scanning rate was 60 K h−1 for heating as well as cooling (data not shown) scans, and the equilibration time before each scan was 10 min. Cp is the molar heat capacity under constant pressure.

(peak maxima) and enthalpy values of the observed transitions are given in the SI (Figure S16). The DSC curves of pure salmon sperm DNA exhibit only a broad endothermic transition with a maximum at 60 °C in the first heating scan, which is in line with the dsDNA2000 (double stranded DNA with 2000 base pairs) melting temperature reported in the literature.44 UV−vis experiments support this observation (data 4911

dx.doi.org/10.1021/la404860w | Langmuir 2014, 30, 4905−4915

Langmuir

Article

Figure 7. (A) SAXS pattern of OT6/DNA complex (nlipid/nnucleotide = 0.17) dispersion in HEPES buffer (pH 7.3) at 25 °C. The Bragg peaks at q001 = 0.115 Å−1 and q002 = 0.232 Å−1 result from the multilamellar structure of the OT6/DNA complex (Lαc). The Bragg reflection at qDNA = 0.180 Å−1 corresponds to the spacing between aligned DNA strands intercalated between lipid bilayers. (B) Schematic representation of the OT6/DNA complexes. The spacing d(Lαc) = 55 Å and d(DNA) = 35 Å result from the SAXS pattern (A), the DNA diameter of 20 Å is a theoretical value from literature.51 (C) TEM image of an OT6/DNA complex (nlipid/nnucleotide = 0.5) dispersion in HEPES buffer (clipid = 42 μM). The sample was stained with uranyl acetate. The bar corresponds to 50 nm.

situation is found at nlipid/nnucleotide ratios of 0.67 and 1, where a reversible endothermic transition at around 85 °C is observed. In contrast to the lower nlipid/nnucleotide ratios, these complexes of DNA with the cationic lipids enable an annealing into dsDNA after melting under the chosen DSC conditions.45 A higher degree of organization after DNA melting seems to enable an easier reconstitution to native dsDNA. 3.2.4. X-ray Scattering Measurements. To obtain further information about the structure of the lipoplexes formed with the novel micellar gene transferring agents, OT6/DNA complexes were investigated by small-angle X-ray scattering technique (SAXS). Because of comparable microstructures observed by TEM (see SI Figure S19) and similar DSC curves, the investigations were restricted to OT6 as representative for both lipids. Figure 7A shows the SAXS pattern of the OT6/DNA complex (nlipid/nnucleotide ratio of 0.17) in HEPES buffer (pH 7.3) at 25 °C. Three Bragg peaks appear. The Bragg peaks at q001 = 0.115 Å−1 and q002 = 0.232 Å−1 result from the multilamellar structure of OT6 (Lαc) with DNA intercalated between the lipid bilayers. The Bragg reflection at qDNA = 0.180 Å−1 corresponds to the distance between parallel aligned DNA strands. SAXS measurements of the OT6/DNA complex at nlipid/nnucleotide ratio of 0.5 (complete DNA complexation) result in a comparable scattering curve with slightly shifted q values for the lamellar phase (q001 = 0.110 Å−1 and q002 = 0.220 Å−1, qDNA was not affected; see SI Figure S20). The SAXS data of complexes at nlipid/nnucleotide = 0.17 result in the following d (repeat distance) values (q = 2π/d): 54.6 Å for the lipid bilayer thickness with incorporated DNA (d(Lαc)) and 34.9 Å for the DNA lattice spacing (d(DNA)). These values result in a sandwich like lamellar lipoplex model with interdigitation of the lipid alkyl chains (see Figure 7B). The lamellar structure model was described in the literature previously but not for interdigitated lipid bilayers in the liquid-crystalline phase.47,48 Koynova et al. and Fielden et al. describe a lamellar lipoplex with interdigitated alkyl chains in the gel phase.49,50 The adsorbed DNA acts as a condensed counterion lattice. Assuming that the lipid bilayer thickness is comparable to that calculated for pure OT6 dispersion (46.3 Å) then only 8.3

Å are left for the DNA layer. This is too small for DNA with a diameter of about 20 Å.51 However, it has to be considered that the 46.3 Å for pure OT6 include the water layer between the lipid bilayers. Additionally, the headgroup is quite flexible (three separate lysine moieties). Finally, we assume a tight attachment of the DNA to the lipid headgroups with a negligible water layer (the lipid replaces the surrounding water layer of the DNA). The interaxial spacing of 34.9 Å also indicates a quite tight in-plane packing of the DNA strands within OT6/DNA complexes.41 The comparison between two different nlipid/nnucleotide ratios (0.17 and 0.5) of the OT6/DNA complexes shows only slight differences. While the spacing between two DNA strands was not affected, the repeat distance of the lamellar structure increases slightly from 54.6 Å (nlipid/nnucleotide = 0.17) to 57.1 Å (nlipid/nnucleotide = 0.5). The ζ-potential measurements show that OT6/DNA complexes with a nlipid/nnucleotide ratio of 0.17 are negatively charged, while complexes with an nlipid/nnucleotide ratio of 0.5 are positively charged (see section 3.2.2). In the present case, the incorporation of excess cationic lipid in the bilayers of the lamellar complex results in an increased repulsion and, in consequence, in an increase of the water layer thickness between the bilayers. The DNA-lattice spacing is not affected, indicating that the electrostatic repulsion between the aligned DNA strands is equally screened at both nlipid/nnucleotide ratios. The lamellar structure of the lipoplexes is also proved by TEM, but is little noticed. Figure 7C shows the lamellar ordering in an OT6/DNA complex (nlipid/nnucleotide = 0.5) in HEPES buffer. Measurements result in distances between 50 and 60 Å and matches with the lamellar spacing determined by SAXS (55 Å). TEM images of the OT6/DNA complex (nlipid/ nnucleotide = 0.17) also exhibit distinct lamellar assemblies (see SI Figures S17 and S18). A possible reason for the rare observation of these structures could be the orientation of the lipoplexes. The most representative structures found in TEM images of OT6 and TT6 lipoplex dispersions do not show such distinct lamellar ordering (see SI Figure S19). However, these assemblies show very short lamellar orderings. The layers are weakly correlated, resulting in a structure with small lamellar regions (Figure 7C), but not in an onion-like 4912

dx.doi.org/10.1021/la404860w | Langmuir 2014, 30, 4905−4915

Langmuir

Article

value of about 120 Å2 and that only the three ε-amino groups are protonated results in a high charge density (ρ) of about ρ = e−(40 Å2)−1. DNA has a smaller charge density of around ρ = e−(110 Å2)−1.52 (ii) Electrostatic forces lead to an attraction between cationic lipids and DNA as a first step of the complex formation process. Furthermore, the gain of entropy caused by counterion release is often discussed as the main driving force.31,36,52 DNA exhibits multiple binding sites: (a) isolated phosphate groups and (b) phosphate groups contiguous to an already occupied site.37 For the latter one, we have to consider the interaction of the neighboring lipids, e.g., hydrogen bonds in the backbone region, electrostatic repulsion, and hydrophobic interaction between the alkyl chains, in addition to the electrostatic interactions between lipids and DNA. Finally, the process of lipid binding to DNA becomes more complex because of the helical structure of DNA with minor and major grooves and the conformational flexibility of the lipid headgroup. (iii) The interaction between the alkyl chains results in a rearrangement of the complex and a compact lipoplex structure. As more cationic lipids bind to DNA the alkyl chains interact with alkyl chains of other DNA bound lipids on the same as well as on neighboring DNA helices. As a result, the DNA helices coalesce. This interaction brings multiple DNA strands together due to the reduced electrostatic repulsion between DNA strands and results in the compact sandwich like structure with DNA between the lipid layers with interdigitated alkyl chains (see Figure 7B). This multifaceted aggregation and complexation behavior only allows model-free methods for the ΔH and TEP determination (see section 3.2.1.), and hence, the process entropy is not accessible. The enthalpy determined by integration of the heat flow peaks (ΔHTot) includes the enthalpy contribution of the buffer dissociation. Equation 1 allows us to separate the complexation ascribed enthalpy (ΔHComp = nΔH2 + ΔH1) from ΔHDiss:

arrangement of the layers (highly correlated layers). This lower degree of bilayer correlation is in agreement with the broadening of the Bragg peaks and the disappearance of the third-order peak of the lamellar phase in SAXS (see Figure 7A). A comparable disordering in lamellar lipoplexes was also described by Zidovska et al. for octavalent lipids.41

4. DISCUSSION The two investigated lipids OT6 and TT6 form micellar aggregates in aqueous dispersions and lamellar aggregates when complexed with DNA. Micelles are formed by lipids with positive curvature while lamellar aggregates are formed by lipids with vanishing curvature.26,52 This behavior can be described with the molecule shape concept of Israelachvili et al.,53 using information about the headgroup area (Ah) as well as the alkyl chain length (lc) and volume (vc) to calculate the packing parameter P. Due to the good solubility of the lipids in buffer, monolayer experiments to determine the molecular area by pressure/area isotherms could not be performed. Therefore, Ah has to be estimated. In a previous work, monolayers of a lipid (malonic acid diamide of the first generation) with one lysine moiety as headgroup and two tetradecyl chains were studied, and a headgroup area of 40 Å2 in the condensed state at pH 4 and 8 was determined.54 For OT6 and TT6 with three terminal lysine moieties Ah is assumed to be roughly 120 Å2. The same value can be obtained by using the 1D lattice distance of DNA (SAXS) and the equation of Farago et al. (dDNA = Ah·(2 × 1.7 Å)−1).55 The lc and vc values can be estimated26 for rigid alkyl chains and P amounts to 0.32 for OT6 and to 0.35 for TT6. In the present case, the chains of the two lipids are disordered (including a high amount of gauche conformers) and therefore shorter leading to larger P values. P < 0.33 is typical for spherical micelles, and 0.33 < P < 0.5 is typical for nonspherical micelles.26 The determined P values are rather indicative of nonspherical micelles and supports the assumption made in section 3.1.1. Hence, the question arose, why can these lipids form lamellar structures? The addition of DNA to the OT6 and TT6 micelles results in a decrease of the headgroup area by the partial loss of the hydration shell and counterions due to the interaction with highly charged DNA. Additionally, SAXS measurements show that the alkyl chains interdigitate in the bilayer, even if they are in the fluid state, which leads to effectively four chains per headgroup compensating the huge mismatch between the required in-plane area of the headgroup and the alkyl chains in a lamellar ordering. The cmc values of OT6 and TT6 are between the ones of common micelle-forming (10−2−10−5 M) and bilayer-forming lipids (10−6−10−10 M).26 The investigated lipid/DNA complex formation is a multiplestep process with the interplay of different intermolecular forces and processes: (i) The process of demicellization occurs during the first injections. The measured heats in the lipid-to-DNA titration experiments clearly show that the monomers bind to the DNA. If only micelles would complex the DNA, one would detect only dilution and demicellization heats until the cmc in the cell is reached.35 This is not the case in the experiments with OT6 and TT6 (see Figures 3 and 4). Furthermore, it is also possible that the micelles interact with DNA if the micelle concentration is high enough.56 Assuming the above-mentioned theoretical Ah

ΔHTot = ΔHComp + nΔHDiss

(2)

Both lipids exhibit exothermic ΔHComp values (Table 1), an evidence for an enthalpic contribution to the lipid/DNA complex formation. The TEP and IEP values of both lipids show that not all six primary amino groups of the lipids are involved in the complexation process. If all six primary amino groups would complex DNA, the theoretic neutral point would be nlipid/ nnucleotide = 0.17. Complete complexation is reached only at higher nlipid/nnucleotide ratios (0.25 (TT6) and 0.31 (OT6); see Table 1). This means that the lipids have an effective charge (qeff) of qeff = 4 (TT6) and qeff = 3 (OT6) in HEPES buffer (pH 7.3). Obviously, the variation of only one alkyl chain, leading to different packing densities, has drastic consequences on the headgroup charge and finally the DNA complexation behavior. It seems that TT6 binds the DNA stronger than OT6, indicated by the higher enthalpy ΔHComp of the DNA complexation process and by the lower nlipid/nnucleotide ratio (∼0.25 vs ∼0.33) of complete DNA complexation (TEP and IEP; see Table 1). In consequence, if DNA is bound too strong in the complexes, the intracellular DNA release decreases.57,58 4913

dx.doi.org/10.1021/la404860w | Langmuir 2014, 30, 4905−4915

Langmuir

Article

(10) Felgner, P. L.; Gadek, T. R.; Holm, M.; Roman, R.; Chan, H. W.; Wenz, M.; Northrop, J. P.; Ringold, G. M.; Danielsen, M. Lipofection: A highly efficient, lipid-mediated DNA-transfection procedure. Proc. Natl. Acad. Sci. U. S. A 1987, 84 (21), 7413−7417. (11) Chytil, A.; Ciccarone, V. C.; Gebeyehu, G.; Hawley-Nelson, P.; Jessee, J. A. Cationic lipids US5334761 A, 1994. (12) Torchilin, V.; Weissig, V. Liposomes, 2nd ed.; Oxford University Press: 2003; p 154−155. (13) Pitard, B.; Aguerre, O.; Airiau, M.; Lachagès, A.-M.; Boukhnikachvili, T.; Byk, G.; Dubertret, C.; Herviou, C.; Scherman, D.; Mayaux, J.-F.; Crouzet, J. Virus-sized self-assembling lamellar complexes between plasmid DNA and cationic micelles promote gene transfer. Proc. Natl. Acad. Sci. U. S. A. 1997, 94 (26), 14412−14417. (14) Candiani, G.; Pezzoli, D.; Cabras, M.; Ristori, S.; Pellegrini, C.; Kajaste-Rudnitski, A.; Vicenzi, E.; Sala, C.; Zanda, M. A dimerizable cationic lipid with potential for gene delivery. J. Gene Med. 2008, 10 (6), 637−645. (15) Johnsson, M.; Wagenaar, A.; Engberts, J. B. F. N. Sugar-based gemini surfactant with a vesicle-to-micelle transition at acidic pH and a reversible vesicle flocculation near neutral pH. J. Am. Chem. Soc. 2003, 125 (3), 757−760. (16) Wölk, C.; Drescher, S.; Meister, A.; Blume, A.; Langner, A.; Dobner, B. General synthesis and physico-chemical characterisation of a series of peptide-mimic lysine-based amino-functionalised lipids. Chem.Eur. J. 2013, 19 (38), 12824−12838. (17) Vlachy, N.; Touraud, D.; Heilmann, J.; Kunz, W. Determining the cytotoxicity of catanionic surfactant mixtures on HeLa cells. Colloids Surf., B 2009, 70 (2), 278−280. (18) Russo, L.; Berardi, V.; Tardani, F.; La Mesa, C.; Risuleo, G. Delivery of RNA and its intracellular translation into protein mediated by SDS-CTAB vesicles: Potential use in nanobiotechnology. BioMed Res. Int. 2013, 2013, 6. (19) Nogueira, D. R.; Carmen Morán, M.; Mitjans, M.; Martínez, V.; Pérez, L.; Pilar Vinardell, M. New cationic nanovesicular systems containing lysine-based surfactants for topical administration: Toxicity assessment using representative skin cell lines. Eur. J. Pharm. Biopharm. 2013, 83 (1), 33−43. (20) Aiello, C.; Andreozzi, P.; La Mesa, C.; Risuleo, G. Biological activity of SDS-CTAB cat-anionic vesicles in cultured cells and assessment of their cytotoxicity ending in apoptosis. Colloids Surf., B 2010, 78 (2), 149−154. (21) De Angelis, I.; Barone, F.; Zijno, A.; Bizzarri, L.; Russo, M. T.; Pozzi, R.; Franchini, F.; Giudetti, G.; Uboldi, C.; Ponti, J.; Rossi, F.; De Berardis, B. Comparative study of ZnO and TiO2 nanoparticles: Physicochemical characterisation and toxicological effects on human colon carcinoma cells. Nanotoxicology 2013, 7 (8), 1361−1372. (22) Pencer, J.; Hallett, F. R. Effects of vesicle size and shape on static and dynamic light scattering measurements. Langmuir 2003, 19 (18), 7488−7497. (23) Paula, S.; Sues, W.; Tuchtenhagen, J.; Blume, A. Thermodynamics of micelle formation as a function of temperature: A high sensitivity titration calorimetry study. J. Phys. Chem. 1995, 99 (30), 11742−11751. (24) Birdi, K. S. Calorimetric determination of the enthalpy of micelle formation in aqueous media. Colloid Polym. Sci. 1983, 261 (1), 45−48. (25) Chattopadhyay, A.; London, E. Fluorimetric determination of critical micelle concentration avoiding interference from detergent charge. Anal. Biochem. 1984, 139 (2), 408−412. (26) Israelachvili, J. N. Intermolecular and Surface Forces., 3rd ed.; Academic Press Elsevier: Waltham, MA, 2011; p 503−577. (27) Pabst, G.; Rappolt, M.; Amenitsch, H.; Laggner, P. Structural information from multilamellar liposomes at full hydration: Full qrange fitting with high quality X-ray data. Phys. Rev. E 2000, 62 (3), 4000−4009. (28) Pabst, G.; Koschuch, R.; Pozo-Navas, B.; Rappolt, M.; Lohner, K.; Laggner, P. Structural analysis of weakly ordered membrane stacks. J. Appl. Crystallogr. 2003, 36 (6), 1378−1388.

5. CONCLUSION In summary, the aggregation behavior and DNA-complexation process of two novel lipids, namely OT6 and TT6, which are promising candidates for new micellar transfection systems, were characterized. A panoply of different physicochemical methods was used. It could be demonstrated that the complete DNA complexation occurs at nlipid/nnucleotide ratios above the theoretical neutral point (nlipid/nnucleotide ratio = 0.17) for both lipids. Additionally, we have proved for the first time by ITC experiments that a proton uptake occurs during the lipoplex formation. The presented investigations clearly demonstrate that the complex preparation process determines the characteristics of the final lipid/DNA complexes, namely, the total amount of complexed DNA, particle charge, and particle size.

■

ASSOCIATED CONTENT

S Supporting Information *

Supplemental methods, details of the cmc-determination, DSC and FT-IR curves of pure lipid dispersions, TEP determination, curve fits of the IEP determination by ζ-potential measurements, additional DLS-results, supplemental data of the DSC investigations of lipoplexes, TEM-images, and SAXS data. This material is available free of charge via Internet at http://pubs. acs.org.

■

AUTHOR INFORMATION

Corresponding Authors

*(B.D.) Tel: +49-345-55-25120. Fax: +49-345-55-27018. Email: [email protected]. *(C.W.) Tel: +49-345-55-25098. Fax: +49-345-55-27018. Email: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS We are grateful to Dr. Hartmut Metzger (Max Planck Institute of Colloids and Interfaces, Potsdam) for making it possibile for us to use the BESSY II facility, and Dr. Chenghao Li and Dr. Stefan Siegel for support during the X-ray measurements.

■

REFERENCES

(1) El-Aneed, A. An overview of current delivery systems in cancer gene therapy. J. Controlled Release 2004, 94 (1), 1−14. (2) O’Connor, T. P.; Crystal, R. G. Genetic medicines: Treatment strategies for hereditary disorders. Nat. Rev. Genet. 2006, 7 (4), 261− 276. (3) Medema, R. H. Optimizing RNA interference for application in mammalian cells. Biochem. J. 2004, 380 (3), 593−603. (4) Iyer, L.; Kumpatla, S.; Chandrasekharan, M.; Hall, T. Transgene silencing in monocots. Plant Mol. Biol. 2000, 43 (2−3), 323−346. (5) Ramachandran, P.; Ignacimuthu, S. RNA interferenceA silent but an efficient therapeutic tool. Appl. Biochem. Biotechnol. 2013, 169 (6), 1774−1789. (6) Kay, M. A.; Glorioso, J. C.; Naldini, L. Viral vectors for gene therapy: The art of turning infectious agents into vehicles of therapeutics. Nat. Med. (N. Y., NY, U. S.) 2001, 7 (1), 33−40. (7) Seow, Y.; Wood, M. J. Biological gene delivery vehicles: Beyond viral vectors. Mol. Ther. 2009, 17 (5), 767−777. (8) Mintzer, M. A.; Simanek, E. E. Nonviral vectors for gene delivery. Chem. Rev. 2008, 109 (2), 259−302. (9) Wan, C.; Allen, T. M.; Cullis, P. R. Lipid nanoparticle delivery systems for siRNA-based therapeutics. Drug Delivery Transl. Res. 2014, 4 (1), 74−83. 4914

dx.doi.org/10.1021/la404860w | Langmuir 2014, 30, 4905−4915

Langmuir

Article

(29) Zumbuehl, A.; Dobner, B.; Brezesinski, G. Phase behavior of selected artificial lipids. Curr. Opin. Colloid Interface Sci. 2014, http:// dx.doi.org/10.1016/j.cocis.2014.01.003. (30) Perozzo, R.; Folkers, G.; Scapozza, L. Thermodynamics of protein−ligand interactions: History, presence, and future aspects. J. Recept. Signal Transduction 2004, 24 (1−2), 1−52. (31) Kennedy, M. T.; Pozharski, E. V.; Rakhmanova, V. A.; MacDonald, R. C. Factors governing the assembly of cationic phospholipid−DNA complexes. Biophys. J. 2000, 78 (3), 1620−1633. (32) Pozharski, E.; MacDonald, R. C. Thermodynamics of cationic lipid−DNA complex formation as studied by isothermal titration calorimetry. Biophys. J. 2002, 83 (1), 556−565. (33) Barreleiro, P. C. A.; Olofsson, G.; Alexandridis, P. Interaction of DNA with cationic vesicles: A calorimetric study. J. Phys. Chem. B 2000, 104 (32), 7795−7802. (34) Pector, V.; Backmann, J.; Maes, D.; Vandenbranden, M.; Ruysschaert, J.-M. Biophysical and structural properties of DNA· diC14-amidine complexes: Influence of the DNA/lipid ratio. J. Biol. Chem. 2000, 275 (38), 29533−29538. (35) Zhu, D.-M.; Evans, R. K. Molecular mechanism and thermodynamics study of plasmid DNA and cationic surfactants interactions. Langmuir 2006, 22 (8), 3735−3743. (36) Patel, M. M.; Anchordoquy, T. J. Contribution of hydrophobicity to thermodynamics of ligand−DNA binding and DNA collapse. Biophys. J. 2005, 88 (3), 2089−2103. (37) Spink, C. H.; Chaires, J. B. Thermodynamics of the binding of a cationic lipid to DNA. J. Am. Chem. Soc. 1997, 119 (45), 10920− 10928. (38) McGlothlin, C. D.; Jordan, J. Thermodynamic parameters of some biochemically significant buffers. Anal. Lett. 1976, 9 (3), 245− 255. (39) Seelig, J. Titration calorimetry of lipid−peptide interactions. Biochim. Biophys. Acta, Rev. Biomembr. 1997, 1331 (1), 103−116. (40) Koltover, I.; Salditt, T.; Safinya, C. R. Phase diagram, stability, and overcharging of lamellar cationic lipid DNA self-assembled complexes. Biophys. J. 1999, 77 (2), 915−924. (41) Zidovska, A.; Evans, H. M.; Ewert, K. K.; Quispe, J.; Carragher, B.; Potter, C. S.; Safinya, C. R. Liquid crystalline phases of dendritic lipid−DNA self-assemblies: Lamellar, hexagonal, and DNA bundles. J. Phys. Chem. B 2008, 113 (12), 3694−3703. (42) Derjaguin, B.; Landau, L. Theory of the stability of strong charged lyophobic sols and of the adhesion of strongly charged particles in solution of electrolytes. Acta Physicochim. URSS 1941, 14, 633−662. (43) Caracciolo, G.; Pozzi, D.; Caminiti, R.; Amenitsch, H. Formation of overcharged cationic lipid/DNA complexes. Chem. Phys. Lett. 2006, 429 (1−3), 250−254. (44) Michanek, A.; Kristen, N.; Höök, F.; Nylander, T.; Sparr, E. RNA and DNA interactions with zwitterionic and charged lipid membranes  A DSC and QCM-D study. Biochim. Biophys. Acta, Biomembr. 2010, 1798 (4), 829−838. (45) Tarahovsky, Y. S.; Rakhmanova, V. A.; Epand, R. M.; MacDonald, R. C. High temperature stabilization of DNA in complexes with cationic lipids. Biophys. J. 2002, 82 (1), 264−273. (46) Burova, T. V.; Grinberg, N. V.; Tur, D. R.; Papkov, V. S.; Dubovik, A. S.; Grinberg, V. Y.; Khokhlov, A. R. Polyplexes of poly(methylaminophosphazene): Energetics of DNA melting. Langmuir 2011, 27 (18), 11582−11590. (47) Koltover, I.; Salditt, T.; Rädler, J. O.; Safinya, C. R. An inverted hexagonal phase of cationic liposome−DNA complexes related to DNA release and delivery. Science 1998, 281 (5373), 78−81. (48) Rädler, J. O.; Koltover, I.; Salditt, T.; Safinya, C. R. Structure of DNA−cationic liposome complexes: DNA intercalation in multilamellar membranes in distinct interhelical packing regimes. Science 1997, 275 (5301), 810−814. (49) Koynova, R.; MacDonald, R. C. Columnar DNA superlattices in lamellar o-ethylphosphatidylcholine lipoplexes: Mechanism of the gel− liquid crystalline lipid phase transition. Nano Lett. 2004, 4 (8), 1475− 1479.

(50) Fielden, M. L.; Perrin, C.; Kremer, A.; Bergsma, M.; Stuart, M. C.; Camilleri, P.; Engberts, J. B. F. N. Sugar-based tertiary amino gemini surfactants with a vesicle-to-micelle transition in the endosomal pH range mediate efficient transfection in vitro. Eur. J. Biochem. 2001, 268 (5), 1269−1279. (51) Podgornik, R.; Rau, D. C.; Parsegian, V. A. The action of interhelical forces on the organization of DNA double helixes: Fluctuation-enhanced decay of electrostatic double-layer and hydration forces. Macromolecules 1989, 22 (4), 1780−1786. (52) May, S.; Ben-Shaul, A. Modeling of cationic lipid−DNA complexes. Curr. Med. Chem. 2004, 11 (2), 151−167. (53) Israelachvili, J. N.; Marčelja, S.; Horn, R. G. Physical principles of membrane organization. Q. Rev. Biophys. 1980, 13 (2), 121−200. (54) Dittrich, M.; Böttcher, M.; Oliveira, J. S. L.; Dobner, B.; Möhwald, H.; Brezesinski, G. Physical-chemical characterization of novel cationic transfection lipids and the binding of model DNA at the air−water interface. Soft Matter 2011, 7 (21), 10162−10173. (55) Farago, O.; Grønbech-Jensen, N. Computational and analytical modeling of cationic lipid−DNA complexes. Biophys. J. 2007, 92 (9), 3228−3240. (56) Wang, Y.; Dubin, P. L.; Zhang, H. Interaction of DNA with cationic micelles: Effects of micelle surface charge density, micelle shape, and ionic strength on complexation and DNA collapse. Langmuir 2001, 17 (5), 1670−1673. (57) Dittrich, M.; Heinze, M.; Wölk, C.; Funari, S. S.; Dobner, B.; Möhwald, H.; Brezesinski, G. Structure−function relationships of new lipids designed for DNA transfection. ChemPhysChem 2011, 12 (12), 2328−2337. (58) Ahmad, A.; Evans, H. M.; Ewert, K.; George, C. X.; Samuel, C. E.; Safinya, C. R. New multivalent cationic lipids reveal bell curve for transfection efficiency versus membrane charge density: Lipid−DNA complexes for gene delivery. J. Gene Med. 2005, 7 (6), 739−748.

4915

dx.doi.org/10.1021/la404860w | Langmuir 2014, 30, 4905−4915