ARTICLE pubs.acs.org/Langmuir
Making Unilamellar Liposomes Using Focused Ultrasound Roberto Tejera-Garcia,† Sanjeev Ranjan,† Vladimir Zamotin, Rohit Sood, and Paavo K. J. Kinnunen* Helsinki Biophysics and Biomembrane Group, Department of Biomedical Engineering and Computational Science, School of Science, Aalto University, P.O. Box 12200, FI-00076 Espoo, Finland
bS Supporting Information ABSTRACT: Several techniques are available for making large unilamellar vesicles (LUV) with an average diameter of approximately 100 nm, widely employed as model biomembranes as well as vehicles for drug delivery. Here we describe the use of adaptive focused ultrasound (AFU) for the preparation of LUV from multilamellar vesicles (MLV) and studied the effects of ultrasound intensity and number of cycles per burst (CPB) on the average size of vesicles produced. CPB determines the duration of the intermittent pressure wavetrains emitted by the transducer, and the corresponding relaxation periods. Preliminary experiments indicated that CPB controls the size of vesicles assembling after the disruption of MLV by ultrasound and optimum values for obtaining LUV could be iterated. The sizes and lamellarity of LUV were assessed by dynamic light scattering (DLS), cryogenic transmission electron microscopy (cryo-TEM), and fluorescence quenching. AFU provides a simple and easy to use approach for making liposomes with several advantages: it is minimally invasive and involves no loss of material. Precisely controlled wavelengths are employed with a significant reduction in the presence of hot spots, which could destroy some biological materials of interest.
’ INTRODUCTION In addition to their use as model biomembranes, liposomes continue to be intensively developed as carriers for drug delivery. Three commonly employed types of liposomes are multilamellar vesicles (MLV), small unilamellar vesicles (SUV), and large unilamellar vesicles (LUV). Compared to MLV, LUV provide a number of advantages as drug carriers, including efficient encapsulation of water-soluble and insoluble drugs, economy of lipid consumption, and reproducible drug release rates.1 Lipid vesicles form spontaneously when lipid films are hydrated and stacks of liquid crystalline bilayers become fluid and swell. Vesicle formation requires layer separation and bending. In early stages of vesicle formation, the normal forces that cause repulsion between lipid layers, and the tangential forces that bend the lipid layers, play crucial roles.2 The hydrated lipid sheets detach during agitation and close to form large MLV, which prevents contacts of water with the hydrocarbon core of the bilayer at the edges. Once MLV have formed any reduction of size requires energy. Accordingly, MLV suspensions can be disrupted by several freezethaw cycles or by extrusion through, for example, polycarbonate large pore size filters (typically 0.21.0 μm), yielding unilamellar vesicles with average diameters determined by the pore size.3 Extrusion is now frequently employed for preparing LUV. As with most procedures for downsizing MLV dispersions the extrusion has to be done above the lipid main phase transition temperature (Tm) because, due to the rigidity of gel state bilayers, liposomes below Tm cannot pass through the filter pores. Extrusion through filters with 100 nm pores has been reported to yield LUV with mean diameters of 7080 nm.3 This particle average size depends also on the lipid composition and is generally reproducible. r 2011 American Chemical Society
The major drawback of extrusion is a loss of material in the filters, which can be significant for some lipids. Ethanol injection is suitable for the formation of liposomes having a small average radius, SUV. In this method, a stock solution of a phospholipid in ethanol is rapidly injected into an aqueous solution.4 Despite many advantages, that is, fast, reproducible, and no need for additional operations, there are two potential disadvantages. First, the liposome suspension obtained contains ethanol, which may have to be removed. Second, since large volumes of aqueous solution are used, it is necessary to use relatively high amounts of any solute that is to be trapped into the liposomes.4 SUV obtained by this method have small diameter (approx 20 to 50 nm) and a high positive curvature of the outmost lipid layer resulting in strain and causing the vesicles to be rather unstable. A number of methods for the preparation of unilamellar vesicles involve the disruption of MLV using ultrasound.5 These approaches produce almost exclusively SUV with average diameters in the range of 1550 nm.6,7 Resulting vesicles can be employed for the preparation of supported lipid bilayers8,9 and for making complexes for gene delivery and encapsulating drug. The most common instruments for the preparation of vesicles using ultrasound are bath and probe sonicators. Both employ low frequencies with an unfocused and incompletely controlled energy output. Probe sonicators transmit higher energies but present several disadvantages. To obtain effective dispersion, these instruments involve direct contact with the sample, releasing titanium Received: May 9, 2011 Revised: June 14, 2011 Published: July 08, 2011 10088
dx.doi.org/10.1021/la201708x | Langmuir 2011, 27, 10088–10097
Langmuir
ARTICLE
particles, which need to be removed by centrifugation. The main obstacle of the above sonication methods is that the acoustic signal is released in the form of pressure wave packages that are partially reflected by the walls of the container used. This produces irregular interference patterns due to the random appearance of constructive and destructive interferences in different regions of the sample. Consequently, the samples are not exposed homogeneously to the ultrasound during the treatment. Another disadvantage is heating of the samples due to the random presence of hot spots. For unsaturated phospholipids this may cause oxidation of unsaturated acyl chains or hydrolysis to lysophospholipids and free fatty acids.10 The above issues are likely reasons for the scarcity of publications reporting the preparation of LUV using ultrasound, merely used as a complementary technique.11 It has been suggested that sonication could be an effective method for the production of liposomes with a defined size if its effects on the lipid membrane were better characterized.12 Traditionally, the formation of vesicles by sonication has been attributed to two different mechanisms, namely, (a) simple budding off or excision of smaller vesicles from MLV surface due to the acoustic pressure waves produced by ultrasonic energy and (b) random shattering of larger MLV into small phospholipid bilayer fragments (PBF), which then assemble and fold up into thermodynamically stable liposomes.2,1315 In most of these mechanistic models, the role that acoustic cavitation plays in the formation of liposomes is incompletely understood. Cavitation is the formation of gas bubbles as cavities in a liquid, resulting from pressure gradients. In acoustic cavitation these local changes in pressure are produced by sound waves. The gas phase inside a cavitation bubble may derive from the liquid phase itself, when the local pressure decreases below the liquid vapor pressure, or from the nucleation of gas molecules dissolved in the liquid phase. In the absence of thermal effects, the dependence of the cavitation bubble dynamics with the variation of the local pressure can been described by the Rayleigh-Plesset equation:16 pv ðT∞ Þ p∞ ðtÞ pG0 R0 3k € ðtÞ þ ¼ RðtÞR FL FL RðtÞ _ 3 4νL RðtÞ 2S 2 _ þ þ þ ðRðtÞÞ 2 FL RðtÞ RðtÞ
ð1Þ
where R(t) is the bubble radius at time t, pv(T∞) is the vapor pressure of the liquid at temperature T∞, and p∞(t) is the local solvent pressure; FL, νL, and S are the solvent density, viscosity, and surface tension, respectively; k, is the polytropic constant of the gas phase inside the cavitation bubble; and pG0 is the partial pressure of liquid-dissolved gases condensed into a bubble with radius R0. At t = 0, if the system is at equilibrium, pG0 can be expressed by16 pG0 ¼ p∞ ð0Þ pv ðT∞ Þ þ
2S R0
ð2Þ
Depending on the media and ultrasound characteristics, the emerging cavitation bubbles may collapse, releasing mechanical energy in the form of secondary shock waves and producing high temperature hot spots, or may oscillate, expanding and shrinking in a continuous way.1720 In traditional sonication techniques, both types of cavitation usually appear simultaneously with
different fractional predominance depending on the input signal. The formation of unilamellar vesicles from sonicated MLV could initially be associated with the presence of high speed (>100 m/sec) short-range jets in the proximities of collapsing cavitation bubbles.21 However, when studying how the predominance of stable or transient types of cavitation affect the mean size of sonicated vesicles, Richardson et al. showed that the presence of transient cavitation bubbles cannot be correlated with changes in the dispersion mean size.22 Instead, when stable cavitation bubbles are predominant, there is a very significant decrease in the average vesicle diameter.12,22 Radiation pressure has been proposed as the main effect responsible for the reduction of the vesicle size.23,24 This radiation pressure produces eddy currents on the bubble surface, generating microstreaming, and attracts particles with a density higher than the surrounding liquid toward stable cavitation bubbles. According to Richardson et al., microstreaming around oscillating bubbles should create sufficient shear to reduce the size of the vesicles during the acoustic treatment of MLV. They integrated numerically eq 1, correlating R0, the initial equilibrium bubble radius, with the resonance frequency f, which equals the frequency of the applied acoustic signal.16 This allows the calculation of the maximum velocity of the bubble walls, ug, and use it to estimate the shear flow in the proximity of a pulsating cavitation bubble as G ≈ UL/δ, where UL = ug2/2πfR0 is the streaming velocity near the bubble and δ = (νL/πpLf)1/2 is the thickness of the acoustic-streaming boundary.25 As mentioned above, one of the mechanisms for downsizing of vesicles by sonication assumes a random shattering of MLV into PBF by the acoustic energy. In this case, it is likely that shear flow in the proximity of a pulsating cavitation bubble is the main factor responsible for MLV disruption.25 As a consequence, the formation and properties of PBF should depend on parameters such as pressure, temperature, density, viscosity, and surface tension of the liquid phase, on the amount and characteristics of dissolved gases, and finally, on the frequency, amplitude, and duration of the employed acoustic energy input. However, the presence of heterogeneous and variable interference patterns inherent to sonication methods similar to bath sonicators can obscure the correlation of experimental results with the former calculations, in particular, due to the error associated with p∞(t) in eq 1. By means of superposition, the presence of variable interferences resulting from reflection can be minimized by adjusting the shape of the transducer so that the wavetrains from its different sections converge into a focal point of maximum mechanical energy. The single element transducer in the commercially available AFU instrument used by us possesses a concave surface that determines the volume of the focal point and its distance from the transducer surface (Figure 1). Due to the stability of the acoustic signal emitted by the transducer, the latter can adapt iteratively to the frequency of the electric input on the transducer, in the proximity of its central resonance frequency, thus, achieving maximum power for transmission. The availability of commercial equipment with a well-focused ultrasound transducer prompted us to investigate if LUV could be generated by this approach. When the sample to be sonicated is placed at the focal point, the augmented resonance acoustic signal induced is more stable, and therefore, the formation of PBF should be more precisely controlled. The properties of the vesicles formed by the fusion of PBF should further depend on the relaxation time between the wavetrains transmitted into the sample. 10089
dx.doi.org/10.1021/la201708x |Langmuir 2011, 27, 10088–10097
Langmuir
Figure 1. Schematic illustration of the AFU acoustic field during an ON period in the transducer, showing the focus of the ultrasound emitting from the concave surface of the transducer. Both the transducer and the sample are contained in a thermostatted water bath.
’ MATERIALS AND METHODS 1,2-Dipalmitoyl-sn-glycero-3-phosphocholine (DPPC), 1-stearoyl-2hydroxy-sn-glycero-3-phosphocholine (LysoPC), 1-palmitoyl-2-oleoylsn-glycero-3-[phospho-rac-(1-glycerol)] (POPG), and 1,2-distearoyl-snglycero-3-phosphoethanolamine-N-[methoxy(polyethylene glycol)-2000] (DSPE-PEG-2000) were from Lipoid GmbH (Ludwigshafen, Germany), 1-stearoyl-2-oleoyl-sn-glycero-3-phosphocholine (SOPC), 1,2-dipalmitoyl-sn-glycero-3-[phospho-rac-(1-glycerol)] (DPPG), 1-hexanoyl-2-[6[(7-nitro-2,1,3-benzoxadiazol-4-yl)amino]hexanoyl]-sn-glycero-3-phosphocholine (NBD-PC), and cholesterol (Chol) were from Avanti Polar Lipids (Alabaster, AL), and N-2-hydroxyethylpiperazine-N-2-ethanesulfonic acid (Hepes) and ethylenediaminetetraacetic acid (EDTA) were from Sigma-Aldrich. The purity of lipids was checked by thin layer chromatography on silicic acid coated plates (Merck, Darmstadt, Germany) developed with chloroform/methanol/water mixture (65:25:4, v/v/v). Visual examination of the plates after iodine staining or upon UV illumination revealed no impurities. Lipid concentrations were determined gravimetrically with a high precision microbalance (SuperG, Kibron, Espoo, Finland). Mole fraction of the indicated materials is used to define their content in the lipid dispersions. In brief, mole fraction X for NBDPC = 0.01 means that of the total lipid present one mole % is the fluorescent lipid analog. Other chemicals were of analytical grade and from standard sources. Unless otherwise indicated all experiments were conducted in 20 mM Hepes, 0.1 mM EDTA, and pH 7.4, at 25 °C. Preparation of Unilamellar Vesicles. Appropriate amounts of the lipid stock solutions were mixed in chloroform to obtain the desired compositions. The solvent was removed under a stream of nitrogen, and the lipid residues were subsequently maintained at reduced pressure overnight. The dry lipid film was then hydrated at 60 °C for 1 h in 20 mM Hepes, 0.1 mM EDTA, pH 7.4 buffer, with stirring. MLV with different lipid compositions, at a final concentration of 100 μM, were subjected to acoustic energy using a commercial AFU sonicator (S2, Covaris, Inc., Woburn, MA) under conditions described in detail in Results and Discussion section.
ARTICLE
Dynamic Light Scattering. Hydrodynamic particle diameters (Zav) and polydispersity indexes (PDI) of lipid vesicles were determined by dynamic light scattering at 25 °C (Zetasizer Nano ZS, Malvern Instruments Ltd., U.K.). The instrument employs photon correlation spectroscopy to characterize the dynamics of the particles on the basis of a Brownian motion model,26 collecting backscattering at an angle of 173°. By cumulant analysis of the measured intensity autocorrelation function, PDI was obtained as the relative variance of the average diffusion coefficient, D. Zav was inferred from D, using the Stokes Einstein relation, which for spherical particles takes the form Zav = kT/ 3πηDZ, where k is the Boltzmann constant, T is the absolute temperature, and η is the shear viscosity of the solvent. In the present study, each data point represents the mean of three independent measurements. Unless otherwise stated, lipid concentration was 100 μM. Differential Scanning Calorimetry (DSC). Lipid samples were loaded into the microcalorimeter (VP-DSC, Microcal Inc., Northampton, MA) calorimeter cuvette before and after AFU treatment. The instrument was operated at a heating rate of 0.5 degrees/min, interfaced to a PC, and the data analyzed using the routines of the software provided by the equipment manufacturer. Dithionite Assay and Fluorescence Measurements. To check for the lamellarity of the liposomes the dithionite quenching assay was employed.27 For this assay a fluorescent phospholipid analog NBD-PC (X = 0.01) was incorporated into the vesicles (100 μM total lipid concentration). In brief, NBD can be reduced by dithionite to the nonfluorescent product, 7-amino-2,1,3-benzoxadiazol-4-yl, allowing for the assessment of the amount of this probe in the outer leaflet of the bilayer. Fluorescence intensity for the lipid solution was recorded before and after the addition of an aliquot of a freshly prepared, ice cold, 50 mM dithionite solution in 20 mM Hepes, 0.1 mM EDTA, pH 7.4, using a Perkin-Elmer LS 50B spectrofluorometer (Wellesley, MA) interfaced to a computer, with 4 nm band passes and excitation and emission wavelengths of 470 and 540 nm, respectively. Intensity values for stable fluorescence readouts were used to calculate the percentage of quenching of NBD fluorescence. Cuvette temperature was maintained at 25 °C with circulating water bath and the contents were agitated by a magnetic stirrer bar. Unilamellarity is assumed when ∼50% quenching upon the addition of dithionite is measured. The initial rapid decline in the emission intensity was followed by a very slow fluorescence loss due to flip-flop of the NBD probe from the inner leaflet. Preparation of Vitrified Specimens and Electron Microscopy. The vitrified samples of vesicles generated by AFU were prepared on Protochips C-Flat 224 grids as previously described.28 A Gatan 626 cryostage was used to observe the sample in a FEI Tecnai F20 field emission gun transmission electron microscope at 200 kV under low-dose conditions and 180 °C. The images were recorded using a Gatan Ultrascan 4000 CCD camera at a nominal magnification of 68000.
’ RESULTS AND DISCUSSION The instrument employed by us works with a center frequency of around 500 kHz, corresponding to a wavelength of about 2 mm. The main parameters controlling the acoustic signal that can be adjusted in this equipment are Duty cycle (DC): Percentage of the time, during one On/Off transducer cycle, where the transducer is “On”, emitting acoustic energy. The instrument used has a maximum DC of 20%. Cycles per burst (CPB): Number of pressure waves contained in the wavetrain produced during each “On” period in the transducer. In AFU at ultrasound frequency f, this parameter controls the duration of the “On” periods in the transducer, 10090
dx.doi.org/10.1021/la201708x |Langmuir 2011, 27, 10088–10097
Langmuir
ARTICLE
Figure 2. Schematic illustration of the AFU transducer output employing (A) intensity 10, 2 cycles per burst with 20% duty cycle, and (B) intensity 10, 4 cycles per burst with 10% duty cycle.
TOn = CPB/f, and also determines the length of the Off period as follows:
TOff
CPB 100 1 ¼ f DC
ð3Þ
Intensity: The acoustic power transmitted by the transducer per unit area. This parameter controls the amplitude of the acoustic pressure waves produced by the transducer and in the instrument employed here its value can be varied between 0 and 10 on an arbitrary scale.
Treatment time (t): The duration of the sonication procedure, which has been shown to be an important parameter that can be correlated with a decrease in the mean size and polydispersity of the final dispersion of vesicles.12 The significance of DC and CPB is illustrated by the schematic examples depicted in Figure 2. Apart of these sonication parameters, the vesicle characteristics are influenced by factors such as lipid concentration, temperature, and sample volume,29 as well as by the properties of the lipids used, such as, Charge: This is directly related to the membrane properties of the formed vesicles, affecting their size, and the final colloidal stability. 10091
dx.doi.org/10.1021/la201708x |Langmuir 2011, 27, 10088–10097
Langmuir
Figure 3. Dependence of Zav (A) and PDI (B) on CPB during 20 min AFU at 25 °C of 5 mL samples of 0.1 mM DPPC MLV at intensity 8 (O) or 10 (b), and 0.1 mM SOPC MLV at intensity 8 (4) or 10 (2).
Lipid packing: The phase state and organization of the saturated or unsaturated lipids in the membrane can be readily expected to influence the average vesicle size and system thermodynamic stability. Packing is in part controlled by the effective shapes of the lipids,30,31 for two or more lipid components, the molar ratio also being a key parameter to be considered. Dispersant characteristics: The characteristic of the solvent phase such as its density, viscosity, and surface tension are directly related to the impact of the propagating energy on the MLV. As there was essentially no prior art in the use of AFU for obtaining LUV, we had to establish initial conditions by trial and error. When different lipid compositions are used, the parameters investigated in our first preliminary trials were those controlling the output of AFU, recording impact on liposome size distribution by DLS. We could quickly abandon the use of low duty cycles yielding no clear effects on the size distribution when changing the other parameters. We then focused on the impact of intensity and CPB on the formation of vesicles composed of DPPC or SOPC, treated for 20 min. As mentioned above, CPB controls the length of the transducer On/Off periods modulating the duration of the pressure wavetrains and the relaxation times between them. Depending on the total treatment time and lipid composition, the results revealed optimum values for the duration of the transducer On/Off periods for obtaining LUV with Zav ≈ 100 nm. Initial experiments revealed that intensity settings below 8 did not yield vesicles with average diameters smaller than 500 nm (data not shown). This could be due to (a) the inability of these energies to disrupt MLV into PBF or (b) to the formation of large
ARTICLE
Figure 4. Heat capacity scans (Cp) for (A) 1 mM DPPC MLV and (B) AFU LUV. The calibration bars correspond to 2 and 1 kJ/deg mol1 in (A) and (B), respectively.
PBF unable to form small liposomes. At higher intensities, downsizing of MLV into smaller vesicles could be seen. For DPPC, there was a clear difference between intensity settings 8 and 10. At intensity 8, the particles were larger than those obtained at intensity 10, yet with similar PDI (Figure 3). In contrast, for SOPC, both Zav and PDI were similar at these intensities (Figure 3). The above is likely to reflect the different phase states of these phospholipids at the treatment temperature, 25 °C, when DPPC is below its main phase transition temperature, Tm, and SOPC in the liquid disordered (fluid) phase. For solid ordered DPPC, higher intensities may give rise to smaller PBF, which then assemble into smaller vesicles. As intensity decreases, the dynamics of formation of cavitation bubbles change (eq 1), producing microstreaming of different strengths. Due to higher membrane elasticity, for lipids such as SOPC (above Tm), the formation of similar PBF is likely to be achieved also with lower microstreaming energies. Some values of CPB were observed to produce monodisperse DPPC and SOPC vesicles with Zav ≈ 100 nm (Figure 3). Under these conditions, it is likely the presence of stable cavitation bubbles causing microstreaming and creating shear enough to produce smaller PBF.22 When maintaining constant values for DC, the value of CPB controls the length of the transducer On/Off periods (eq 3). Three effects by CPB can be considered (Figure 3): (i) Low CPB values may not transmit enough acoustic energy to cause the formation of PBF and a simple excision of polydisperse large vesicles from the MLV may take place. (ii) At high enough CPB, PBF are generated with subsequent fusion into small vesicles during long enough transducer Off periods. In our experiments, optimum values for CPB, 10092
dx.doi.org/10.1021/la201708x |Langmuir 2011, 27, 10088–10097
Langmuir
ARTICLE
Figure 5. Dependence of Zav and PDI on AFU CPB and on the molar ratio of the constituent phospholipids in 0.1 mM DPPG:DPPC MLV at intensity 8 (A, B) and at intensity 10 (C, D). Similar data but for POPG/DPPC at intensity 8 (E, F) and at intensity 10 (G, H). Samples were in 5 mL of buffer. Total time for sonication was 20 min at 25 °C. All liposomes additionally contained NBD-PC (X = 0.01).
yielding Zav ≈ 100 nm, were observed between 100 and 500 CPB. (iii) High CPB values increase the duration of the transducer On periods, which could decrease the size of the formed PBF. Accordingly, at constant DC, the corresponding longer Off periods upon increasing CBP may allow PBF assemble into a more polydisperse collection of larger vesicles, as observed (Figure 3). Our results demonstrate that, by varying the intensity and CPB, depending on the saturation of the lipid acyl chains, it is possible to modulate the size of the vesicles and to diminish their polydispersity.
The presence of unsaturated acyl chain eliminates differences between size distributions obtained at intensity settings 8 and 10 and has no effect on the trends observed for different CPB. The dithionite assay revealed the vesicles to be largely unilamellar. Differential Scanning Calorimetry. The commonly employed membrane models MLV and LUV differ significantly in their thermal phase behavior.32 Proper understanding of the physicochemical properties of LUV obtained by AFU is essential for their possible use as model membranes or as drug or gene delivery vehicles. DSC was used to check the differences in the 10093
dx.doi.org/10.1021/la201708x |Langmuir 2011, 27, 10088–10097
Langmuir thermal phase behavior of DPPC MLV and LUV prepared by AFU (Figure 4). Even when the total enthalpy for the AFU LUV is more or less the same as for MLV, LUV made by AFU revealed a broader asymmetric endotherm than MLV, peaking at 41 °C, corresponding to the main phase transition temperature. The lower melting component could be due to the presence of PBF in the samples (see below). MLV exhibit two well discernible endothermic phase transitions at 35.3 and at 41.3 °C, corresponding to the pretransition between the gel (Lβ) and rippled (Pβ) phases, and to the main (chain melting) transition between the rippled and liquid crystalline (LR) phases, respectively. The pretransition is weak for pure DPPC LUV,33 and it has been suggested that the formation of the Pβ phase by MLV requires interlamellar coupling, which is absent for the separated bilayers in LUV.32 Binary Liposomes. To evaluate the influence of surface charge on the formation of LUV by AFU, we prepared MLV composed of DPPG and DPPC at different molar ratios, namely, 0.05:0.94, 0.2:0.79, 0.4:0.59, and 0.99:0.00. NBD-PC (X = 0.01) was additionally included to allow to assess the lamellarity of the liposomes. MLV were treated for 20 min at intensity 8 or 10, varying CPB. No differences in Zav and PDI were evident upon changing CPB or lipid composition (Figure 5AD). Accordingly, an increase in the negative surface charge due to the inclusion of DPPG is without effect. Saad et al. demonstrated that increasing the content of DPPG in a mixture with DPPC up to 50% does not change the elasticity of mixed monolayers.34 To vary the membrane elasticity of liposomes, we repeated the above experiment using the unsaturated POPG instead of DPPG. POPG/DPPC MLV were made at molar ratios of 0.05:0.94, 0.2:0.79, 0.4:0.59, and 0.99:0.00. Pronounced impact on Zav and PDI by CPB and the lipid stoichiometry could now be observed (Figure 5E,H). Importantly, the same values of CPB yielding for zwitterionic lipids low PDI now yielded large aggregates with high polydispersity. At the two intensities applied, 8 and 10, and for CPB close to 200, both Zav and PDI increased with XPOPG > 0.2. This inversion in the change of Zav and PDI with respect to CPB for vesicles containing POPG indicates a possible synergetic effect of increasing surface charge and increasing membrane elasticity due to the unsaturated acyl chain. Because of the latter, a more loose packing and increased membrane elasticity and less resistance to stretching and bending is expected. For the zwitterionic liposomes, the optimum values of CPB (∼200) for obtaining LUV with low PDI are likely to reflect the formation of properly sized PBF fusing and assembling during sufficiently long relaxation periods. For liposomes containing POPG, these same AFU parameters result in the formation of larger and more polydisperse PBF, able to fuse and assemble into a highly polydisperse collection of vesicles. The different durations of the transducer On/Off periods, above or below 200 CPB, may produce either monodisperse, smaller PBF with longer Off assembling times or the excision of smaller PBF from MLV at lower CPB. This is consistent with the formation of SUV (Zav = 50 ( 5 nm, PDI = 0.20 ( 0.05) at CPB above 400, with a minimum in PDI for POPG/DPPC (0.2:0.8, molar ratio) in all cases. Higher contents of DPPC yield again smaller diameters and less polydispersity, as seen for neat DPPC at 200 CPB. At intensity 10, both Zav and PDI changed with CPB and lipid composition similarly as when using intensity 8. These changes occurred in a narrow range of CPB and lipid ratios, appearing at XPOPG above 0.5 and at CPB between 50 and 400 (Figure 5G,H).
ARTICLE
Figure 6. Dependence of the Zav and PDI (A, B) on the AFU treatment time, varied from 4 to 20 min. The samples (5 mL of 100 μM total lipid) were SOPC, sonicated at intensity 10 and CPB 200 (O), DPPC/Chol/ DSPE-PEG-2000 (75:20:5, molar ratio), treated at intensity 8 and CPB 200 (b), POPG/DPPC/Chol (65:5:30, molar ratio) at intensity 10 and CPB 100 (4), and POPG/DPPC/Chol (50:20:30, molar ratio) at intensity 8 and CPB 500 (2). Temperature was maintained at 25 °C.
Higher ultrasound energy seems to yield less polydisperse PBF, with a smaller change in liposome size depending on the elasticity of the lipid compositions and the relaxation times of AFU. Effect of AFU Treatment Time. In general, longer AFU treatment decreased Zav irrespective of the other parameters used. For SOPC, Zav decreased upon prolonging the sonication time while values for PDI remained unaltered (Figure 6). This was evident also for compositions forming more rigid membranes, such as POPG/DPPC/Chol (65:5:30, molar ratio) and POPG/DPPC/Chol (50:20:30, molar ratio), for which Zav decreased, whereas PDI was unaffected. For these compositions, cholesterol may well be responsible for the lack of changes in PDI. Differences in acyl chain saturation, treatment intensity, or CPB do not seem to affect the above findings (Figure 6). DPPC/Chol/DSPE-PEG-2000 (75:20:5, molar ratio) corresponds to the liposome formulation Doxil developed for anticancer drug delivery.35 This composition involves high membrane rigidity due to the presence of cholesterol and saturated lipids. For this lipid mixture, constant values of Zav and decrease in PDI for a 20 min treatment were evident (Figure 6). We can conclude that the relation between longer treatment time and reduced vesicle size does not hold for all lipid compositions and depends on the parameters describing lipid packing and on the stability of PBF and the assembling vesicles. Stability of Liposomes. Liposome stability is a key parameter controlling the shelflife of the preparations. To check the stability of lipid vesicles prepared using AFU, they were stored at 4 °C for several days and changes in their size distributions upon aging 10094
dx.doi.org/10.1021/la201708x |Langmuir 2011, 27, 10088–10097
Langmuir
ARTICLE
Figure 7. Dependence of the Zav (A) and PDI (B) on CPB and sample age for 100 μM (lipid) DPPC/Chol/DSPE-PEG-2000 (75:20:5, molar ratio) liposomes in 5 mL and treated at intensity 8 for 20 min at 25 °C.
were assessed by DLS. For the Doxil formulation, we observed an increase in the mean size and polydispersity with time. This was particularly clear during the first two days of storage (Figure 7). As in the former experiments, it was again possible to observe optimum CPB values corresponding to minimum mean sizes (100 nm, Figure 7A). Interestingly, these optimum values (500 CPB) did not influence the PDI values, which increased linearly with CPB. These results may relate to the steric stabilization imparted by the PEGylated lipid upon the formation and assembly of PBF, together with the enhanced stability of the vesicles formed. Cryo-TEM of Liposomes. Vesicles generated by AFU were analyzed also by cryo-TEM. DPPC MLV (0.1 mM) treated during 20 min at 20 DC, intensity 8, and 100 CPB (Figure 8A) were unilamellar and showed sizes on the range of those obtained by extrusion of 1 mM MLV through 200 nm pore polycarbonate filters (Figure 8B). Their Zav was slightly above 100 nm and they exhibited a dimpled surface compared to the smooth appearance of the extruded liposomes. Accordingly, the AFU-generated LUV are likely to be tension free. Micrographs consistently revealed
PBF (Figure 8A, white arrows). When a higher (20 mM) concentration of DPPC/LysoPC/DSPE-PEG-2000 (87:9:4, molar ratio) MLV was sonicated by AFU using the former parameters, the number of PBF increased. LysoPC may stabilize the contour of these PBF while the PEG conjugate reduce their assembly rate. In this way it was possible to observe more PBF after the AFU treatment and also to analyze their mechanism of assembly into unilamellar vesicles (Figure 8CF). Possible fusion of two PBF is likely to be taking place in Figure 8C (white arrow). Self-closure of bigger membrane fragments produced by PBF fusion was occasionally seen (Figure 8D, white arrows). Below this self-closing vesicle there is an already formed LUV. PBF trapped inside a LUV during the vesicle closure was also captured (Figure 8E, white arrow), with possible selfassembly when two or more PBF were trapped in this way Figure 8F (white arrow). The phospholipid bilayer fragments illustrated in the Cryo-TEM photographs (Figure 8DF) were consistently seen for a large number of samples analyzed. These cryo-TEM micrographs corroborate the mechanism for formation of unilamellar vesicles by sonication after the disruption of 10095
dx.doi.org/10.1021/la201708x |Langmuir 2011, 27, 10088–10097
Langmuir
ARTICLE
We studied the effects of AFU parameters intensity and CPB on the formation of unilamellar vesicles with different lipid compositions. Intensity determines the size and shape of the generated PBF and CPB modulates their self-assembly by controlling the duration of the transducer On/Off. In this way, both parameters influence the mean size and polydispersity of the forming vesicles. A suitable protocol to adjust these key AFU parameters to obtain LUV was iterated. For most lipid compositions we were able to obtain a combination of parameters, yielding monodisperse unilamellar vesicles with a Zav ≈ 100 nm (see example in Figure 9). In conclusion, we employed AFU to obtain unilamellar vesicles from MLV composed of anionic and zwitterionic lipids. The vesicles formed are free of organic solvents and detergents and are generated in an isothermal environment. The procedures described could be used on a large scale as well as a continuous flow process. This method is currently under further development in our laboratory for the preparation of LUV encapsulating drugs and siRNA.
’ ASSOCIATED CONTENT
bS
Supporting Information. Additional figure and supporting data. This material is available free of charge via the Internet at http://pubs.acs.org.
’ AUTHOR INFORMATION Corresponding Author
*Fax: +358 9 470 23182. E-mail: paavo.kinnunen@aalto.fi. Figure 8. Cryo-TEM images of (A) 0.1 mM DPPC MLV after 20 min AFU at intensity 8, DC 20 and CPB 100; (B) 1 mM DPPC MLV passed through polycarbonate filters of 200 nm pore; (CF) 20 mM DPPC/ LysoPC/DSPE-PEG-2000 (87:9:4, molar ratio) MLV after 20 min AFU at intensity 8, DC 20 and CPB 100.
Figure 9. Size distributions (vol %) for 100 μM DPPC MLV after lipid film hydration (2), after extrusion through 100 nm polycarbonate filter (b), and after 20 min AFU at 25 °C (O, intensity 8, CPB 100).
MLV into PBF and their subsequent fusion and self-assembly into LUV.14
’ CONCLUSIONS There are several methods for preparing unilamellar vesicles yet all have certain limitations. AFU overcomes some of these limitations being very fast and noninvasive.
Author Contributions †
Equal contribution.
’ ACKNOWLEDGMENT This work was supported by EU FP7 integrated projects Nanoear (NMP4-CT-2006-026556) and Sonodrugs (NMP4LA-2008-213706), Finnish Academy, and Sigrid Juselius Foundation. The authors thank Kristiina S€oderholm for technical assistance. ’ ABBREVIATIONS AFU, adaptive focused ultrasound; Chol, cholesterol; CPB, cycles per burst; cryo-TEM, cryogenic transmission electron microscopy; DC, duty cycle; DLS, dynamic light scattering; DPPC, 1,2-dipalmitoyl-sn-glycero-3-phosphocholine; DPPG, 1,2-dipalmitoyl-sn-glycero-3-phospho-(10 -rac-glycerol); DSC, differential scanning calorimetry; DSPE-PEG-2000, 1,2-distearoyl-sn-glycero3-phosphoethanolamine-N-[methoxy(polyethylene glycol)2000] (ammonium salt); EDTA, ethylenediaminetetraacetic acid; Hepes, N-2-hydroxyethylpiperazine-N-2-ethanesulfonic acid; LUV, large unilamellar vesicles; LysoPC, 1-stearoyl-2-hydroxy-snglycero-3-phosphocholine; MLV, multilamellar vesicles; NBD, nitro-2,1,3-benzoxadiazol-4-yl; NBD-PC, 1-hexanoyl-2-[6-[(7nitro-2,1,3-benzoxadiazol-4-yl)amino]hexanoyl]-sn-glycero-3phosphocholine; PBF, phospholipid bilayer fragments; PDI, polydispersity index; POPG, 1-palmitoyl-2-oleoyl-sn-glycero3-[phospho-rac-(1-glycerol)]; SOPC, 1-stearoyl-2-oleoyl-snglycero-3-phosphocholine; SUV, small unilamellar vesicles; Tm, main phase transition temperature; Zav, average hydrodynamic particle diameter 10096
dx.doi.org/10.1021/la201708x |Langmuir 2011, 27, 10088–10097
Langmuir
ARTICLE
’ REFERENCES (1) Betageri, G. V.; Jenkins, S. A.; Parsons, D. L. Liposome Drug Delivery Systems; Technomic Publishing Co. Inc.: Lancaster, 1993; Chapter 1. (2) Lasic, D. D. Biochem. J. 1988, 256, 1–11. (3) MacDonald, R. C.; MacDonald, R. I.; Menco, B. P.; Takeshita, K.; Subbarao, N. K.; Hu, L. R. Biochim. Biophys. Acta 1991, 1061, 297–303. (4) Batzri, S.; Korn, E. D. Biochim. Biophys. Acta 1973, 298, 1015–1019. (5) Chatterjee, S.; Banerjee, D. K. In Liposome Methods and Protocols; Basu, S. C., Basu, M., Eds.; Humana Press: Totowa, 2002; Vol. 199, p3. (6) Morrissey lab protocol for preparing phospholipid vesicles (SUV) by sonication, http://tf7.org/suv.pdf. Accessed on 20/04/2011. (7) Saunders, L.; Perrin, J.; Gammack, D. J. Pharm. Pharmacol. 1962, 14, 567–572. (8) Domanov, Y. A.; Kinnunen, P. K. J. J. Mol. Biol. 2008, 376, 42–54. (9) Domanov, Y. A.; Kinnunen, P. K. J. Biophys. J. 2006, 91, 4427–4439. (10) Uhumwangho, M. U.; Okor, R. S. J. Med. Biomed. Res. 2005, 4, 9–21. (11) Zhang, H.; Zhang, L. J. Colloid Interface Sci. 1997, 190, 76–80. (12) Woodbury, D. J.; Richardson, E. S.; Grigg, A. W.; Welling, R. D.; Knudson, B. H. J. Liposome Res. 2006, 16, 57–80. (13) Lasic, D. D. J. Liposome Res. 1995, 5, 431–441. (14) Lasic, D. D. J. Theor. Biol. 1987, 124, 35–41. (15) Fromherz, P.; Ruppel, D. FEBS Lett. 1985, 179, 155–159. (16) Brennen, C. E. Cavitation and Bubble Dynamics; Oxford University Press: New York, 1995; Chapter 1. (17) Cavitation in Microgravity (Full Movie), http://www.youtube. com/watch?v=gR0YBAhY2PQ. Accessed on 20/04/2011. (18) Suslick, K. S. J. Acoust. Soc. Am. 1991, 89, 1885–1886. (19) Putterman, S. J. Sci. Am. 1995, 272, 46–51. (20) Didenko, Y. T.; Suslick, K. S. Nature 2002, 418, 394–397. (21) Lauterborn, W.; Ohl, C. Ultrason. Sonochem. 1997, 4, 65–75. (22) Richardson, E. S.; Pitt, W. G.; Woodbury, D. J. Biophys. J. 2007, 93, 4100–4107. (23) Miller, D. L.; Nyborg, W. L.; Whitcomb, C. C. Science 1979, 205, 505–507. (24) Nyborg, W. L. Br. J. Cancer Suppl. 1982, 5, 156–160. (25) Nyborg, W. L. J. Acoust. Soc. Am. 1968, 44, 1302–1309. (26) Koppel, D. E. J. Chem. Phys. 1972, 57, 4814–4820. (27) McIntyre, J. C.; Sleight, R. G. Biochemistry 1991, 30, 11819–11827. (28) Baker, T. S.; Olson, N. H.; Fuller, S. D. Microbiol. Mol. Biol. Rev. 1999, 63, 862–922. (29) Johnson, S. M.; Bangham, A. D.; Hill, M. W.; Korn, E. D. Biochim. Biophys. Acta 1971, 233, 820–826. (30) Janmey, P. A.; Kinnunen, P.K. J. Trends Cell Biol. 2006, 16, 538–546. (31) Thuren, T.; Virtanen, J. A.; Kinnunen, P.K. J. Chem. Phys. Lipids 1986, 41, 329–334. (32) Parry, M. J.; Hagen, M.; Mouritsen, O. G.; Kinnunen, P.K. J.; Alakoskela, J. I. Langmuir 2010, 26, 4909–4915. (33) Heimburg, T. Biophys. J. 2000, 78, 1154–1165. (34) Saad, S. M. I.; Policova, Z.; Acosta, E. J.; Hair, M. L.; Neumann, A. W. Langmuir 2009, 25, 10907–10912. (35) Lasic, D. D. Nature 1996, 380, 561–562.
10097
dx.doi.org/10.1021/la201708x |Langmuir 2011, 27, 10088–10097
