Heterogeneous Domains and Membrane Permeability in

(7) To address the requirement for long blood circulation times, vesicles need .... two major lipid components with matching and nonmatching chain len...
0 downloads 0 Views 250KB Size
Langmuir 2008, 24, 5679-5688

5679

Articles Heterogeneous Domains and Membrane Permeability in Phosphatidylcholine-Phosphatidic Acid Rigid Vesicles As a Function of pH and Lipid Chain Mismatch Shrirang Karve, Gautam Bajagur Kempegowda, and Stavroula Sofou* Othmer-Jacobs Department of Chemical and Biological Engineering, Polytechnic UniVersity, Brooklyn, New York 11201 ReceiVed June 10, 2007. ReVised Manuscript ReceiVed March 5, 2008 Heterogeneous lipid membranes tuned by pH were evaluated at 37 °C in the form of PEGylated vesicles composed of lipid pairs with dipalmitoyl (n ) 16) and distearoyl (n ) 18) chain lengths. One lipid type was chosen to have the titratable moiety phosphatidic acid on its headgroup, and the other lipid type was chosen to have a phosphatidylcholine headgroup. The effect of pH on the formation of lipid heterogeneities and on membrane permeability was studied on vesicles composed of lipid pairs with matching and nonmatching chain lengths. The formation of lipid heterogeneities increases with decreasing pH in membranes composed of lipid pairs with either matching or nonmatching chain lengths. Increased permeability with decreasing pH was exhibited only by membranes composed of lipid pairs with nonmatching chain lengths. Permeability rates correlate strongly with the predicted extent of interfacial boundaries of heterogeneities, suggesting defective packing among nonmatching acyl chains of lipids. In heterogeneous mixtures with one lipid type in the fluid state (n ) 12), the dependence of membrane permeability on pH is weaker. In the presence of serum proteins, PEGylated gel-phase vesicles containing lipid pairs with nonmatching chain lengths exhibit faster release rates with decreasing pH compared to measured release rates in phosphate buffer, suggesting a second mechanism of formation of separated phases. PEGylated vesicles composed of lipid pairs with nonmatching chain lengths labeled with internalizing anti-HER2/neu antibodies that target overexpressed antigens on the surface of SKOV3-NMP2 ovarian cancer cells exhibit specific cancer cell targeting, followed by extensive internalization (more than 84% of bound vesicles) and fast release of contents intracellularly. These PEGylated vesicles composed of rigid membranes for long blood circulation times that exhibit pH-dependent release of contents intracellularly could become potent drug delivery carriers for the targeted therapy of solid tumors.

1. Introduction Interfacial boundaries between lipid domains in heterogeneous membranes may contain areas of mismatches in molecular packing, resulting in increased membrane permeability.1 Such interfacial boundaries are observed in membranes of single lipids or of lipid mixtures at the transition temperatures where regions of gel domains and fluid domains coexist.2–4 This molecular origin of membrane permeability suggests that, in membranes composed only of lipids in the gel state, any molecular packing defects between different heterogeneous gel domains could result in more pronounced membrane permeability. Triggered formation of heterogeneities with leaky interfacial boundaries that results in triggered increase of membrane permeability can be useful in liposome-based drug delivery. Needham et al.5 studied the increased permeability of membranes close to their phase transition temperatures and developed temperature-sensitive liposomes in conjunction with local hy* To whom correspondence should be addressed. Address: CBE, Polytechnic University, 6 MetroTech Center, Brooklyn, NY 11201. E-mail: [email protected]. Telephone: 718 260 3863. Fax: 718 260 3125. (1) Mouritsen, O. G.; Zucherman, M. J. Phys. ReV. Lett. 1987, 58, 389–392. (2) Carruthers, A.; Melchior, D. L. Biochemistry 1983, 22, 5797–5807. (3) Clerc, S. G.; Thompson, T. E. Biophys. J. 1995, 68, 2333–2341. (4) Papahadjopoulos, D.; Jacobson, K.; Nir, S.; Isac, T. Biochim. Biophys. Acta 1973, 311, 330–348. (5) Needham, D.; Dewhirst, M. W. AdV. Drug DeliVery ReV. 2001, 53, 285– 305.

perthermia at the tumor sites. They demonstrated enhanced release of encapsulated therapeutics from the liposome carriers at the tumor sites with successful therapeutic outcomes.6 However, this approach is an antivascular approach limited by the short circulation times of liposomes and intended for superficial tumors that are accessible by external heating. We are interested in drug carriers with long blood circulation times to increase the probability of accumulating into any vascularized tumor, and in using a tumor endogenous stimulus to trigger the release of therapeutic contents after internalization of liposome carriers by the cancer cells that constitute the tumor.7 To address the requirement for long blood circulation times, vesicles need to be composed of gel-phase lipids8 and also need to be PEGylated.9,10 The studied stimulus for triggered release of encapsulated contents is the pH due to the acidification occurring during the endosomal pathway after antibody-mediated uptake of targeting vesicles by cancer cells. (6) Hauck, M. L.; LaRue, S. M.; Petros, W. P.; Poulson, J. M.; Yu, D.; Spasojevic, I.; Pruitt, A. F.; Klein, A.; Case, B.; Thrall, D. E.; Needham, D.; Dewhirst, M. W. Clin. Cancer Res. 2006, 12, 4004–4010. (7) Simoes, S.; Moreira, J. N.; Fonseca, C.; Düzgünes, N.; Pedroso de Lima, M. AdV. Drug DeliVery ReV. 2004, 56, 947–965. (8) Gabizon, A.; Papahadjopoulos, D. Proc. Natl. Acad. Sci. U.S.A. 1988, 85, 6949–6953. (9) Allen, C.; Dos Santos, N.; Gallagher, R.; Chiu, G. N. C.; Shu, Y.; Li, W. M.; Johnstone, S. A.; Janoff, A. S.; Mayer, L. D.; Webb, M. S.; Bally, M. B. Biosci. Rep. 2002, 22, 225–250. (10) Sofou, S. Nanomedicine 2007, 2, 711–724.

10.1021/la800331a CCC: $40.75  2008 American Chemical Society Published on Web 05/10/2008

5680 Langmuir, Vol. 24, No. 11, 2008

It has been shown that pH can be used to control nonideal mixing forming heterogeneities in lipid membranes containing phosphatidylcholine (PC) and phosphatidic acid (PA) mixtures.11 We are interested in evaluating the effect on membrane permeability of pH-sensitive heterogeneities in mixtures of rigid membranes with PC and PA lipid pairs of matching and nonmatching chain lengths. We are interested in evaluating the roles of nonideal lipid mixing and of lipid acyl chain mismatches across the bilayer at the interfacial boundaries of these heterogeneities. In this study, the pH-dependent formation of lipid heterogeneities and the pH-dependent change in membrane permeability have been investigated for PEGylated phospholipid bilayers in the form of unilamellar vesicles at the physiologically relevant temperature of 37 °C. Vesicles were composed of lipid pairs with matching and nonmatching chain lengths (n ) 16 and 18) and with headgroups of PC and PA. Vesicles were characterized for: (1) content release of encapsulated solutes of variable size (calcein, fluorescent dextrans) in the presence and absence of cholesterol, (2) formation of membrane heterogeneities, and (3) aggregation and potential fusion in phosphate buffer at decreasing pH over time. In order to evaluate the performance of these systems in environments closer to the physiological milieu, antibody-labeled vesicles were also studied in serum supplemented media and in Vitro.

2. Materials and Methods 2.1. Materials. Trisodium 8-hydroxypyrene-1,3,6-trisulfonate (HPTS) was purchased from Sigma Aldrich Chemical Co. (Milwaukee, WI). The lipids 1,2-dipalmitoyl-sn-glycero-3-phosphocholine (DPPC), 1,2-dipalmitoyl-sn-glycero-3-phosphate (monosodium salt) (DPPA), 1,2-distearoyl-sn-glycero-3-phosphocholine (DSPC), 1,2-distearoyl-sn-glycero-3-phosphate (monosodium salt) (DSPA), 1,2-dilauroyl-sn-glycero-3-phosphocholine (DLPC), 1,2dioleoyl-sn-glycero-3-phosphoethanolamine-N-[methoxy(polyethylene glycol)-2000] (ammonium salt) (DOPE-PEG), 1,2-dipalmitoylsn-glycero-3-phosphoethanolamine-N-[methoxy(polyethylene glycol)2000] (ammonium salt) (DPPE-PEG), 1,2-distearoyl-sn-glycero3-phosphoethanolamine-N-[methoxy(polyethylene glycol)-2000] (ammonium salt) (DSPE-PEG), 1,2-dipalmitoyl-sn-glycero-3-phosphoethanolamine-N-(lissamine rhodamine B sulfonyl) (ammonium salt) (rhodamine-lipid), and 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine-N-(7-nitro-2-1,3-benzoxadiazol-4-yl) (ammonium salt) (NBD-lipid) were purchased from Avanti Polar Lipids (Alabaster, AL) (all lipids at purity >99%). Texas Red dextrans (3000 MW and 10 000 MW) were obtained from Invitrogen Molecular Probes (Carlsbad, CA). Fetal bovine serum was purchased from Omega scientific (Tarzana, CA). RPMI 1640 was purchased from Cellgro Mediatech Inc. (Herndon, VA). Lysozyme from chicken egg white (lyophilized powder), calcein, cholesterol, Triton X-100, and phosphate buffered saline (PBS) were purchased from Sigma Aldrich Chemical Co. (Milwaukee, WI). 2.2. Preparation of Vesicles. Lipids in chloroform were combined in a 25 mL round-bottom flask. Compositions contained 5% mol cholesterol unless otherwise stated and 2% mol PEGylated lipids. Two types of PEGylated lipids were used for each composition with lipid chain lengths matching the chain lengths of the two types of phospholipids in each composition (dilauroyl, dipalmitoyl, or distearoyl). For DLPC containing vesicles, DOPE-PEG lipids were used for matching the lauroyl phase. Chloroform was evaporated in a Buchi rotavapor R-200 (Buchi, Flawil, Switzerland) apparatus for 10 min at 55 °C followed by evaporation under N2 stream for 5 min. The dried lipid film was then hydrated in 1 mL of phosphate buffer (PBS with 1 mM EDTA, pH ) 7.4) for 2 h at 50 °C. The lipid suspension (10 mM total lipid) was then extruded 21 times through two stacked polycarbonate filters of 100 nm pore diameter (Avestin Inc., Ottawa, Canada). Extrusion was carried out at 80 °C in a water bath. For calcein quenching efficiency measurements, the lipid film (11) Garidel, P.; Johann, C.; Blume, A. Biophys. J. 1997, 72, 2196–2210.

KarVe et al. was hydrated in 1 mL of phosphate buffer containing 55 mM calcein (pH ) 7.4, isosmolar to PBS). After extrusion, unentrapped calcein was removed at room temperature by size exclusion chromatography (SEC) using a Sephadex G-50 column (of 11 cm length) eluted with phosphate buffer (1 mM EDTA, pH ) 7.4). For dextran retention measurements, the dried lipid film was hydrated with 1 mL of phosphate buffer containing 3 kDa dextrans (0.5 mg/mL) or 10 kDa dextrans (0.62 mg/mL). 2.3. Cryo-Transmission Electron Microscopy (TEM). Vesicles were imaged using a FEI Tecnai 20 cryo-transmission electron microscope. Measurements were performed by the staff of the Analytical Imaging Facility at the Albert Einstein College of Medicine, Yeshiva University. Samples were frozen, and thin frozen sections were imaged without staining. 2.4. Evaluation of Vesicle Size Distributions. Dynamic light scattering (DLS) of vesicle suspensions was studied with an N4 plus autocorrelator (Beckman-Coulter, Fullerton, CA) equipped with a 632.8 nm He-Ne laser light source. Scattering was detected at 30.2, 62.6, and 90°. Particle size distributions at each angle were calculated from autocorrelation data analysis by CONTIN.12 The average vesicle size was calculated to be the y-intercept at zero angle of the measured average particle size values versus sin2(θ).13 All buffer solutions used were filtered with 0.22 µm filters just before vesicle preparation. The collection times for the autocorrelation data were 1-10 min. Vesicles were incubated in different solutions (phosphate buffer and serum supplemented media) at 37 °C, and vesicle fractions were removed from the parent suspension at different time points and were diluted in phosphate buffer (pH ) 7.4) before measurement. 2.5. Differential Scanning Calorimetry. Differential scanning calorimetry (DSC) studies were performed using a VP-DSC instrument (MicroCal, LLC, Northampton, MA). DSC scans of vesicle suspensions (0.5 mL, 2.5 mM lipid) for different pH values were performed on vesicles with encapsulated phosphate buffer at the same pH as the surrounding phosphate buffer solution (7.4, 5.5, and 4.0) from 10 to 85 °C at a scan rate of 5 °C/hr or 60 °C/hr. The thermogram of the corresponding buffer was acquired at identical conditions and was subtracted from the excess heat capacity curves. DSC scans of vesicle suspensions with externally added lysozyme (at 3.12 mg protein/mL) were performed at a scan rate of 60 °C/hr. In these studies, vesicle suspensions with encapsulated phosphate buffer (pH ) 7.4) were incubated for 2 h at 37 °C in the presence or absence of lysozyme in solutions of different pH values (7.4, 5.5, and 4.0) before acquisition of thermograms. 2.6. Release of Encapsulated Contents from Vesicles. To evaluate the release of calcein, vesicles containing self-quenching concentrations of calcein (55 mM) were incubated in phosphate buffer or serum supplemented media at different pH values at 37 °C over time. The concentration of lipids for incubation was 0.20 mM. The release of calcein from vesicles and its dilution in the surrounding solution results in an increase in fluorescence due to the relief of self-quenching.14 Calcein release was measured at different time points by adding fixed quantities of vesicle suspensions into cuvettes (with cross section 1 cm × 1 cm) containing phosphate buffer (1 mM EDTA, pH ) 7.4). Calcein fluorescence (ex ) 495 nm, em ) 515 nm) before and after addition of Triton-X 100 was measured using a Fluoromax-2 spectrofluorometer (Horiba Jobin Yvon, NJ), and was used to calculate the quenching efficiency defined as the ratio of fluorescence intensities after and before addition of Triton-X 100. The percentage of retained contents with time was calculated as follows:

% calcein content retention )

(

)

Qt - Qmin × 100 Qmax - Qmin

where Qtis the calcein quenching efficiency at the corresponding time point t, Qmax is the maximum calcein quenching efficiency in phosphate buffer (at pH ) 7.4) at room temperature immediately after (12) Provencher, S. W. Comput. Phys. Commun. 1982, 27, 213–227. (13) Teraoka, I. Polymer Solutions: An Introduction to Physical Properties, 1st ed.; John Wiley & Sons: New York, 2002; pp 188-191. (14) Allen, T. M.; Cleland, L. G. Biochim. Biophys. Acta 1980, 597, 418–426.

Membrane Permeability in PC-PA Rigid Vesicles

Langmuir, Vol. 24, No. 11, 2008 5681

Figure 1. DSC thermograms showing the effect of pH (7.4, 5.5, and 4.0) on the formation of heterogeneous (nonideal) mixing on equimolar mixtures of lipids with phosphatidyl choline and phosphatidic acid headgroups and acyl tails of matching (B,C) and nonmatching (A) lengths. All membranes were in the form of extruded vesicles containing 5% mol cholesterol and 2% mol PEGylated lipids incubated in phosphate buffer at the indicated pH at 37 °C for two hours. (A) DPPC and DSPA, (B) DPPC and DPPA, and (C) DSPC and DSPA lipids. Thermograms were acquired at 60 °C/hr.

separation of vesicles by SEC, and Qmin is the minimum quenching efficiency equal to unity. This experimental approach underestimates the actual retention of contents when the encapsulated calcein content retention falls below 18% of the initial value of 55 mM. This is because the relief of calcein self-quenching occurs essentially at concentrations below 10 mM. However, for the present studies, the calcein selfquenching method should be accurate for calculating the rates of content release. This is because content retention by these vesicles is significantly above 18%. Only a few measured points, and only at later incubation times, exhibit content retention values of less than 18% and should not influence the calculated content release rates. These values were observed only for equimolar DPPC and DSPA vesicles at pH ) 5.0 (at 72 h of incubation and longer times) and pH ) 4.0 (at 48 h of incubation and longer times) (see the Supporting Information). To evaluate the release of dextrans, vesicles containing fluorescent dextrans were incubated in phosphate buffer at different pH values at 37 °C. The concentration of lipids during incubation was 1.25 mM in phosphate buffer to increase the concentration of encapsulated dextrans and to improve the efficacy of their detection due to the low dextran passive entrapment efficiency by the vesicles. At various time points, vesicle fractions were removed from the parent suspension and released dextrans were separated from vesicles by SEC using a Sepharose 4B column (of 11 cm length) eluted with phosphate buffer (1 mM EDTA, pH ) 7.4) at room temperature. Free dextrans and dextrans encapsulated in vesicles were quantitated by fluorescence spectroscopy (ex ) 595 nm, em ) 615 nm). 2.7. Intervesicle Fluorescence Resonance Energy Transfer (FRET). FRET (ex ) 496 nm, em ) 590 nm) was monitored to evaluate possible fusion among vesicles at decreasing pH values.15 Two populations of vesicles were prepared: the first population containing NBD-lipids (energy donor) and rhodamine-lipids (energy acceptor), each at 0.5% mol fraction, and the second population not containing fluorophores. Vesicle fractions from each population were mixed at equal volumes at various pH values, and their fluorescence intensity was monitored over time. In this experimental setup, fusion increases the effective distances among fluorophores, resulting in lower emission intensities of the energy acceptor. To account for quenching or photobleaching effects unrelated to fusion, the fluorescent intensities of samples were normalized by the intensities of samples containing only the fluorescent vesicles. 2.8. Labeling of Vesicles with Antibodies. The protocol to immunolabel vesicles (immunoliposomes) is described elsewhere.16 Briefly, Trastuzumab purified from Herceptin (Genentech, South San Francisco, CA) was activated with Traut’s reagent (1 mL of 0.5 mg/mL antibody in PBS buffer (1 mM EDTA, pH ) 8) for 1.5 h), purified by using a PD-10 column (Amersham Biosciences, Piscataway, NJ) eluted with PBS buffer (1 mM EDTA, pH ) 7.4), and then reacted with maleimide-lipids contained in vesicles (2% mol of total lipid) in a nitrogen (15) Struck, D. K.; Hoekstra, D.; Pagano, R. E. Biochemistry 1981, 20, 4093– 4099. (16) Kirpotin, D. B.; Park, J. W.; Hong, S. K.; Zalipsky, S.; Li, W.-L.; Carter, P.; Benz, C. C.; Papahadjopoulos, D. Biochemistry 1997, 36, 66–75.

atmosphere for 40 min at room temperature (the ratio of vesicles to antibody was 0.5 mg of activated antibody to 5 µmol lipids in a total volume of 2 mL). Unbound antibodies were separated from immunoliposomes by size exclusion chromatography (SEC) using a Sepharose 4B column (11 cm length) eluted with PBS buffer (1 mM EDTA, pH ) 7.4). In purified immunoliposome suspensions, a protein assay was used to quantify the concentration of antibodies, and lipid concentrations were determined by the fluorescence intensity of rhodamine-lipids contained in liposomes. Using the measured values of conjugated antibodies per lipid, the average number of antibodies per immunoliposome was calculated16 using for input values the mean size of vesicles (as measured by DLS) and the headgroup surface area per lipid (70 Å2).17 2.9. Cell Line. The metastatic ovarian carcinoma cell line SKOV3NMP2 was derived from serial passage of the parental SKOV3 cell line in nude mice18 and was propagated at 37 °C in 5% CO2 in RPMI 1640 media supplemented with 10% fetal bovine serum, 100 units/ mL penicillin, and 100 µg/mL streptomycin. Cell concentrations were determined by counting trypsinized cells with a hemocytometer. 2.10. Cell Binding and Internalization of Liposomes. Harvested SKOV3-NMP2 cells were suspended in media (RPMI 1640/10% FBS) at a density of 3.5 × 106 cells/mL. HPTS-encapsulating liposomes (0.12 mM final lipid concentration) were added to 5.4 mL of cell suspension to give a final volume of 8.1 mL, and two 500 µL samples were immediately taken and processed as described below. The cells were then placed in a humidified 37 °C incubator with 5% CO2, where they were periodically swirled and sampled at 0, 0.5, 1, 2, 4, and 6 h. The cells were washed twice with 2 mL of ice-cold PBS, and then 1 mL of acidic striping buffer (50 mM glycine, 100 mM NaCl, pH ) 2.7) was added for 5 min at room temperature to eliminate the surface bound immunoliposomes.19 To quantitate cell bound and internalized liposomes, the pellets from centrifugation before and after stripping, respectively, were measured using a fluorescence spectrophotometer.

3. Results 3.1. Lipid Membrane Heterogeneities Detected by DSC: Effect of pH and of Mismatch of Lipid Chain Lengths. The effect of pH on the formation of membrane heterogeneities was studied on vesicles containing two major lipid components with matching and nonmatching chain lengths of n ) 16 and 18. Figure 1 shows the thermal scans of PEGylated vesicles with equimolar DPPC and DSPA lipids (Figure 1A), equimolar DPPC and DPPA lipids (Figure 1B), and equimolar DSPC and DSPA lipids (Figure 1C) in phosphate buffer of pH ) 7.4, 5.5, and 4.0 (17) Lasic, D. D. Liposomes from Physics to Applications; Elsevier: Amsterdam, 1993; 555. (18) Mujoo, K.; Maneval, D. C.; Anderson, S. C.; Gutterman, J. U. Oncogene 1996, 12, 1617–1623. (19) Sofou, S.; Kappel, B. J.; Jaggi, J. S.; McDevitt, M. R.; Scheinberg, D. A.; Sgouros, G. Bioconjugate Chem. 2007, 18, 2061–2067.

5682 Langmuir, Vol. 24, No. 11, 2008

KarVe et al.

Figure 2. Phase diagrams exhibiting the effect of pH 7.4 (A), 5.5 (B), and 4.0 (C) on nonideal lipid mixing of membranes composed of DPPC and DSPA lipid pairs with nonmatching chain lengths. All membranes were in the form of extruded vesicles containing 5% mol cholesterol and 2% mol PEGylated lipids incubated in phosphate buffer at 37 °C for 2 h before acquisition of DSC thermograms with a scanning rate of 5 °C/hr. The dotted lines are guides to the eye.

Figure 3. Effect of chain length mismatching on the release of encapsulated calcein as a function of pH from vesicles composed of equimolar lipid pairs with phosphatidyl choline and phosphatidic acid headgroups incubated in phosphate buffer at 37 °C. Membranes were composed of (A) DPPC and DSPA, (B) DPPC and DPPA, and (C) DSPC and DSPA lipids, and all contained 5% mol cholesterol and 2% mol PEGylated lipids. pH ) 7.4 (b), pH ) 5.5 (O), pH ) 5.0 (1), and pH ) 4.0 (3). The error bars correspond to standard deviations of repeated measurements (two vesicle preparations, three samples per preparation per time point). The solid lines are guides to the eye.

after 2 h of incubation at 37 °C. Membranes containing lipid pairs with mismatched chain lengths (Figure 1A) exhibit multipeak thermal spectra indicative of heterogeneous phases. The system appears to be not ideally miscible. Decreasing pH, from 7.4 to 4.0, results in increasing contributions from thermal transitions at higher temperatures. Higher thermal transitions at lower pH values suggest increasing formation of lipid phases that should be rich in clustered protonated DSPA lipids.11 A decrease of the fraction of the titratable DSPA lipid (25% and 10% mol) resulted in similar contributions at higher thermal transitions with decreasing pH but at a lower extent. The phase diagrams for the DPPC-DSPA system at varying pH values of 7.4, 5.5, and 4.0 are shown in Figures 2A, B, and C, respectively. The phase diagrams were constructed from initiation and completion temperatures of the observed calorimetric transition curves.20 These results suggest nonideality of lipid mixing that increases with decreasing pH. Vesicles containing lipid pairs with matching chain lengths (DPPC and DPPA) exhibited distinctive thermal peaks with decreasing pH, suggesting phase separated lipid domains (Figure 1B, black arrows). Interestingly, the combination of lipid pairs with matching acyl tails of longer length (DSPC and DSPA) exhibited a more complex behavior of heterogeneous lipid formations that were also pH-dependent (Figure 1C, black arrows). 3.2. Membrane Permeability Altered by pH. 3.2.1. Role of Formation of Membrane Heterogeneities Induced by pH. The effect of lipid heterogeneities induced by pH on increasing the (20) Mabrey, S.; Sturtevant, J. M. Proc. Natl. Acad. Sci.U.S.A. 1976, 73, 3862– 3866. (21) Pott, T.; Maillet, J. C.; Dufourc, E. J. Biophys. J. 1995, 69, 1897–908.

permeability of bilayer membranes was studied on vesicles containing two major lipid components. Membrane permeability was studied by monitoring the extent of encapsulated calcein release from vesicles composed of equimolar DPPC and DSPA lipids (Figure 3A), equimolar DPPC and DPPA lipids (Figure 3B), and equimolar DSPC and DSPA lipids (Figure 3C) at different pH values over time. Only the membranes composed of lipid pairs with nonmatching chain lengths, dipalmitoyl and distearoyl, exhibit a pH-dependent release of contents that increases with decreasing pH (Figure 3A). The release of encapsulated contents from vesicles composed of lipid pairs with matching chain lengths, DPPC and DPPA lipids (Figure 3B), does not depend on pH. Membranes composed of matching acyl tails of longer length (DSPC and DSPA lipids) exhibit a weak dependence of content release on pH (Figure 3C). DSC studies show that all the above lipid membranes exhibit formation of heterogeneities with decreasing pH that includes nonideal mixing, lipid phase separation, or both. However, in order to increase the dependence of membrane permeability on pH, the formation of lipid heterogeneities needs to be combined with lipid pairs of nonmatching chain lengths. In content release studies, all vesicle compositions exhibit an initial drop in content retention within the first minutes of incubation that is similar for all pH values studied. The release of contents during this interval should indicate melting of grain boundaries on the vesicle membranes due to the instantaneous heating of vesicles from room temperature to 37 °C, and it should be independent of the extent of protonation of titratable lipids (see next section (3.2.2)). The existence of grain boundaries could be due to poor molecular packing of lipid tails (packing

Membrane Permeability in PC-PA Rigid Vesicles

Langmuir, Vol. 24, No. 11, 2008 5683

Table 1. Average Size of Vesicles Measured by DLS in Phosphate Buffer (Top Panel) and in 10% Serum Supplemented Media (Lower Panel) at 37 °C in Different pH Conditions for Various Incubation Timesa DPPC-DSPA (mol ratio 1:1) time (days)

pH ) 7.4

pH ) 5.5

DPPC-DSPA (mol ratio 3:1)

DPPC-DSPA (mol ratio 9:1)

pH ) 4.0

pH ) 7.4

pH ) 5.5

pH ) 4.0

pH ) 7.4

pH ) 5.5

pH ) 4.0

in phosphate buffer 189 ( 15 188 ( 14 199 ( 16 198 ( 18 190 ( 10 193 ( 13

185 ( 16 195 ( 20 195 ( 13

167 ( 13 174 ( 15 176 ( 17

157 ( 13 168 ( 15 185 ( 17

159 ( 17 168 ( 15 180 ( 14

in 10% serum supplemented media 124 ( 15 130 ( 21 224 ( 13 130 ( 14 163 ( 20 376 ( 19 130 ( 14 186 ( 12 421 ( 31

125 ( 11 141 ( 13 133 ( 10

141 ( 12 167 ( 13 181 ( 11

236 ( 21 336 ( 23 344 ( 22

1 2 3

167 ( 13 180 ( 10 177 ( 10

166 ( 8 178 ( 7 180 ( 10

174 ( 8 183 ( 8 178 ( 11

1 2 3

161 ( 9 171 ( 11 169 ( 9

175 ( 12 194 ( 14 221 ( 11

210 ( 17 379 ( 21 400 ( 26

a Vesicles were composed of different ratios of DPPC and DSPA lipids, 5% mol cholesterol, and 2% mol PEGylated lipids. The errors correspond to standard deviations of repeated measurements (two vesicle preparations, one sample per preparation measured five times per scattering angle).

Table 2. Parameters of the Double Exponential Decay y(t) ) a exp(-bt) + c exp(-dt) Used to Fit the Content Release Profiles from Vesicles Composed of Different Mole Ratios of DPPC and DSPA Lipids with 5% Mol Cholesterol and 2% Mol PEGylated Lipidsa b × 102

c

d × 104

pH

a

7.4 5.5 5.0 4.0

38 ( 1 40 ( 1 49 ( 2 66 ( 2

DPPC-DSPA (mole ratio 1:1) 2.19 ( 0.15 62 ( 1 2.93 ( 0.27 61 ( 1 2.62 ( 0.30 52 ( 2 2.36 ( 0.23 41 ( 2

0.54 ( 0.04 1.69 ( 0.08 2.17 ( 0.15 3.79 ( 0.33

7.4 5.5 5.0 4.0

18 ( 3 26 ( 4 27 ( 3 28 ( 2

DPPC-DSPA (mole ratio 3:1) 7.09 ( 3.48 82 ( 2 6.77 ( 2.62 74 ( 2 1.26 ( 4.93 73 ( 2 3.91 ( 0.92 72 ( 2

0.83 ( 0.09 0.94 ( 0.12 1.02 ( 0.12 0.97 ( 0.09

7.4 5.5 5.0 4.0

15 ( 2 21 ( 3 26 ( 3 23 ( 4

DPPC-DSPA (mole ratio 9:1) 10.8 ( 5.37 85 ( 2 3.55 ( 1.37 78 ( 2 4.07 ( 1.36 74 ( 2 3.64 ( 1.55 75 ( 3

0.57 ( 0.06 0.52 ( 0.09 0.47 ( 0.10 0.64 ( 0.12

Figure 4. Cryo-TEM image of vesicles composed of equimolar DPPC and DSPA lipids, 5% mol cholesterol, and 2% mol PEGylated lipids (scale bar is 100 nm).

a The first exponential decay (a, b) is attributed to the sudden temperature difference at the onset of measurements from room temperature to 37 °C. The second exponential decay (c, d) is attributed to poor lipid packing resulting in leaky interfacial boundaries of the pH-dependent membrane heterogeneities.

defects) or due to premelting of DPPC-lipids whose Tg is 41 °C and very close to the working temperature of 37 °C.1,5 PEGylated vesicles composed only of DPPC lipids and 5% mol cholesterol exhibited an initial drop in content retention within the first few minutes of incubation at 37 °C similar to all vesicles shown above that was then followed by stable retention of 60% of encapsulated contents for at least 4 days at all pH values studied. PEGylated vesicles composed only of DSPA lipids and 5% mol cholesterol exhibited no drop in content retention at the beginning of incubation at 37 °C and retained more than 80% of their contents for longer than 5 days at all pH values studied. 3.2.2. Role of the Extent of Leaky Interfacial Boundaries Induced by pH. PEGylated vesicles were studied composed of different ratios of DPPC and DSPA lipids of average sizes (diameter) that ranged from 167 ( 13 to 189 ( 15 nm in PBS at pH ) 7.4 the day of preparation (Table 1, top panel). This value is in agreement with the average diameter of 189 ( 38 nm observed in cryo-TEM images on a total population of 52 vesicles composed of equimolar DPPC and DSPA lipids, 5% mol cholesterol, and 2% mol PEGylated lipids (Figure 4). Figure 4 also indicates that vesicles were unilamellar. Membrane permeability with decreasing pH was studied by monitoring the release of calcein from these unilamellar vesicles

composed of lipids with nonmatching chain lengths. All vesicle compositions exhibited increased content release rates with decreasing pH (see the Supporting Information). Vesicles composed of equimolar DPPC and DSPA lipids exhibited faster pH-dependent content release compared to vesicles with 25% and 10% mol DSPA lipid. After 30 days of incubation, contents were completely released from all vesicle compositions. A double exponential function was used to fit the content release profiles from the above lipid membrane compositions (Table 2). The faster release mechanism is attributed to the instantaneous temperature difference introduced at the onset of measurements (see Discussion section) with a rate value that is approximately the same for all three lipid membrane compositions and is pH-independent (Table 2, second column). The slower release mechanism (rates shown in Table 2, fourth column) is attributed to the poor lipid packing within the lateral interfacial boundaries of the pH-dependent lipid heterogeneities. At decreasing pH, preferential association between protonated DSPA lipids is suggested to occur (section 3.1). This increasing lipid association could be affecting the total length of interfacial boundaries of lipid heterogeneities. In the simplest approximation of small circular domains of similar size rich in protonated DSPA lipids, the total perimeter of interfacial boundaries should be proportional to the square root of the number of protonated DSPA lipids. The pKa value of 4.5 was used for phosphatidic acid to

5684 Langmuir, Vol. 24, No. 11, 2008

Figure 5. Effect of the extent of leaky interfacial boundaries on the permeability of heterogeneous rigid membranes composed of lipid pairs with nonmatching chain lengths. Membrane permeability is indicated by the rate of calcein release in vesicles composed of DPPC and DSPA lipids at different mole ratios ((O) 1:1, (0) 3:1, (3) 9:1) with 2% mol PEGylated lipids at 37 °C in various pH conditions (7.4, 5.5, 5.0, and 4.0). The square root of the number of protonated DSPA lipids is proportional to the extent of interfacial boundaries for small, circular, and similar in size lipid heterogeneities rich in DSPA. Open symbols correspond to membranes containing 5% mol cholesterol. Closed symbols denote membranes without cholesterol. The error bars correspond to standard deviations of repeated measurements (two vesicle preparations, three samples per preparation per time point). The solid lines are guides to the eye.

calculate the fraction of protonated DSPA lipids. For all three lipid fractions, Figure 5 shows that the rate of content release attributed to discontinuities at the interfacial boundaries is linearly proportional to the square root of protonated DSPA lipids and therefore proportional to the total perimeter of the discontinuities’ interface (open symbols). For comparison, identical lipid membranes lacking cholesterol (closed symbols) exhibit similar dependence on the square root of protonated DSPA lipids but with faster release rates. The pKa value of 4.5 is the lowest value above the reported value of 4.021 that results in arrangement of all data points in Figure 5 in a straight line. This deviation of the pKa value can be attributed to several factors. First, the proposed model of small, circular domains is certainly an oversimplification (see Discussion). Second, because the apparent pKa of PA should depend on the surface charge density,22 the values of apparent pKa should be composition-dependent. In addition, incubation of vesicles at 37 °C in phosphate buffer with pH values ranging from 7.4 to 4.0 does not cause significant increase on the vesicle size distributions over three days, suggesting minimal vesicle aggregation. Also, no membrane fusion was observed that could also contribute to changes in membrane permeability (Table 1, top panel). 3.2.3. Effect of Probe Size. To assess the size of transient discontinuities at the interfacial boundaries attributed to poor lipid packing, fluorescently labeled dextrans of molecular sizes larger than calcein were encapsulated in vesicles and their release from vesicles was measured at decreasing pH values. Vesicles composed of equimolar DPPC and DSPA lipids with nonmatching chain lengths exhibit pH-dependent release of 3 kDa dextrans (Table 3, also plotted in a figure in the Supporting Information). The extent of content release increases with decreasing pH. For equimolar DPPC and DSPA mixtures, the 3 kDa dextran release rates are slower than the calculated release rates of calcein. Membranes containing lower fractions of the titratable lipid DSPA (22) Trauble, H.; Teubner, M.; Woolley, P.; Eibl, H. Biophys. Chem. 1976, 4, 319–342.

KarVe et al.

(25% and 10% mol DSPA content, Table 3) do not exhibit pHdependent permeability to 3 kDa dextrans. Larger dextrans with a molecular weight of 10 kDa were stably retained (>93%) by all vesicle compositions for 30 days. 3.3. Interaction of Vesicles with Serum Proteins. Vesicles composed of high Tg lipids with pH-responsive membranes may be useful in drug delivery applications, and particularly in cases where the release of encapsulated contents is triggered at decreasing pH values related to the endosomal pathway. 3.3.1. Vesicle Aggregation in Serum Supplemented Media. Incubation of PEGylated vesicles composed of DPPC and DSPA lipids at 37 °C in 10% serum supplemented media at pH values ranging from 7.4 to 4.0 causes changes on the vesicle size distributions (Table 1, lower panel). Vesicle aggregation increases with decreasing pH values. Aggregation increases with time and with increasing DSPA content. 3.3.2. Induction of Lipid Heterogeneities by Proteins. Lysozyme was added to vesicle suspensions, and the system thermal transitions were evaluated. Lysozyme was used as a model protein because of its high denaturation temperature (72 °C) that is not interfering with the temperature window defined by the thermal transitions of DPPC and DSPA lipids of 41 and 75 °C, respectively. Figure 6A shows the thermograms of equimolar DPPC and DSPA containing vesicles at neutral pH in the presence and absence of lysozyme. The split of the main transition peak of lipid membranes in the presence of lysozyme probably suggests lateral phase separation caused by lateral clustering of negatively charged DSPA lipids upon protein adsorption.23 The shift of the protein denaturation peak to lower temperatures also suggests adsorption of lysozyme onto the lipid membrane.23 Thermograms at lower pH values (5.5 and 4.0) in the presence of lysozyme resulted in multiple splits on the broad lipid transition peak that could probably be correlated to the more than one transition temperatures that were detected on free lysozyme at the corresponding pH values. 3.3.3. pH-Controlled Content Release from Vesicles in Serum Supplemented Media. Figure 6B shows the pH-dependent release of fluorescent contents from vesicles in 10% serum supplemented media. The release kinetics was significantly faster compared to release rates in phosphate buffer. In particular, vesicles containing equimolar DPPC and DSPA lipids release within 30 min of incubation at the endosomally relevant pH values of 5.5 and 5.07 (49% and 70% of their contents, respectively), compared to pH 7.4. The observed content release within the first 10 min of incubation appears to depend on two mechanisms: a pH-dependent mechanism and a second mechanism possibly due to thermal or osmotic membrane destabilization at the onset of incubation as mentioned above (section 3.2.1). Vesicles containing smaller fractions of the titratable DSPA lipid exhibit a slower release of contents at every pH value studied. 3.3.4. Labeling of Liposomes with Antibodies. Cell Binding and Internalization. The conjugation reaction resulted in 50 ( 7 antibodies per anti-HER2/neu PEGylated liposome. Leakage of entrapped fluorescent HPTS contents due to conjugation was not detected. Anti-HER2/neu PEGylated liposomes composed of DPPC and DSPA lipids (at 9:1 mol ratio) exhibited high binding specificity to HER2/neu overexpressing SKOV3 ovarian cancer cells (Figure 7A) compared to nontargeted liposomes with identical lipid composition. Anti-HER2/neu liposomes were extensively internalized (more than 84% of the total cell-bound liposomes). Content release from internalized liposomes was indicated by the ratio of fluorescence intensities at 450 and 413 nm of HPTS. (23) Raudino, A.; Castelli, F. Colloid Polym. Sci. 1992, 270, 1435–1536.

Membrane Permeability in PC-PA Rigid Vesicles

Langmuir, Vol. 24, No. 11, 2008 5685

Table 3. Effect of the Extent of pH-Dependent Membrane Heterogeneities with Leaky Interfacial Boundaries on the Permeability of Nanometer-Sized Solutes. Percentage Retention of Fluorescent Dextrans (3000 MW) versus pH by Vesicles Composed of Different Mole Ratios of DPPC and DSPA Lipids with Nonmatching Chain Lengths, With 5% Mol Cholesterol And 2% Mol PEGylated Lipids in Phosphate Buffer at 37 °C for Various Incubation Timesa DPPC-DSPA (mol ratio 1:1) time (minutes)

DPPC-DSPA (mol ratio 9:1)

pH ) 7.4 pH ) 5.5 pH ) 5.0 pH ) 4.0 pH ) 7.4 pH ) 5.5 pH ) 5.0 pH ) 4.0 pH ) 7.4 pH ) 5.5 pH ) 5.0 pH ) 4.0

0 100 ( 0 10 95 ( 2 30 93 ( 0 60 89 ( 1 1440 (day 2) 88 ( 2 5760 (day 5) 68 ( 2 a

DPPC-DSPA (mol ratio 3:1)

100 ( 0 85 ( 3 81 ( 4 79 ( 3 68 ( 3 59 ( 3

100 ( 0 86 ( 3 80 ( 2 75 ( 4 66 ( 2 60 ( 3

100 ( 0 79 ( 4 71 ( 4 68 ( 2 61 ( 2 51 ( 3

100 ( 0 100 ( 0 100 ( 0 97 ( 4 99 ( 0 98 ( 1 98 ( 1 94 ( 10 93 ( 8 96 ( 2 98 ( 1 97 ( 0 97 ( 2 94 ( 6 95 ( 2 95 ( 2 94 ( 2 93 ( 1

100 ( 0 94 ( 1 93 ( 5 95 ( 1 93 ( 6 90 ( 3

100 ( 0 95 ( 3 90 ( 3 91 ( 2 93 ( 2 91 ( 4

100 ( 0 97 ( 1 93 ( 2 94 ( 4 95 ( 3 94 ( 2

100 ( 0 98 ( 1 95 ( 2 97 ( 0 96 ( 4 96 ( 1

100 ( 0 96 ( 2 94 ( 6 95 ( 1 96 ( 2 94 ( 0

Errors correspond to standard deviations of repeated measurements (two vesicle preparations, two samples measured per preparation and per time point).

Figure 6. Effect of proteins on inducing the formation of lipid heterogeneities on vesicle membranes (A) and on increasing the rates of encapsulated content release from vesicles (B) composed of equimolar DPPC and DSPA lipids with nonmatching chain lengths, with 5% mol cholesterol and 2% mol PEGylated lipids. (A) DSC thermographs on the effect of lysozyme on lipid phase separation in vesicle membranes incubated for 2 h in phosphate buffer at pH 7.4 at 37 °C. (B) Retention of encapsulated calcein as a function of pH by vesicles incubated in 10% serum supplemented media at 37 °C. pH 7.4 (b), pH 5.5 (O), pH 5.0 (1), and pH 4.0 (3). The error bars correspond to standard deviations of repeated measurements (two vesicle preparations, two samples per preparation per time point). The solid lines are guides to the eye.

Figure 7. Specific cell binding, internalization, and content release of pH-sensitive targeting vesicles. (A) Binding to SKOV3-NMP2 ovarian cancer cells of anti-HER2/neu (triangles) and nontargeted (circles) PEGylated vesicles composed of DPPC and DSPA lipids with 5% mol cholesterol. Closed symbols correspond to total cell associated vesicles. Open symbols correspond to cell internalized vesicles. (B) Calculated pH of the immediate environment of encapsulated HPTS from cell internalized anti-HER2/neu vesicles. The error bars correspond to standard deviations of duplicate measurements.

Figure 7B shows that HPTS is collectively experiencing a fast decrease in pH from 7.4 to 6.7 after cell internalization of liposomes, suggesting that encapsulated contents are rapidly released from internalized liposomes following the endosomal uptake. 3.4. Vesicle Fusion Evaluated by FRET. For vesicles composed of equimolar DPPC and DSPA lipids, no change in the FRET intensities was observed over 4 days both in phosphate

buffer and in 10% serum supplemented media for all pH values studied (7.4, 5.5, and 4.0), suggesting that no fusion occurs even after vesicle aggregation in media

4. Discussion Interfacial boundaries in heterogeneous lipid membranes can potentially contain areas of poor lipid packing resulting in increased membrane permeability.2 Changes in membrane

5686 Langmuir, Vol. 24, No. 11, 2008

permeability due to the formation of such interfaces on the plane of the membranes is of particular interest given the role of heterogeneous domains such as lipid rafts in biological phenomena including membrane trafficking, signaling, and protein transport.24,25 Our interest is the triggered formation of leaky interfaces that can lead to a triggered increase in membrane permeability. This property can be useful in liposome-based drug delivery. Lipid membranes containing heterogeneities tuned by pH were evaluated in the form of vesicles composed of two types of long saturated lamellar-forming lipids. One lipid type was chosen to have the titratable moiety phosphatidic acid on its headgroup (PA), and the other lipid type was chosen to have a phosphatidylcholine headgroup. The lipids were chosen to have either dipalmitoyl (n ) 16) or distearoyl (n ) 18) chain lengths, and membranes with lipids of matching and nonmatching chain lengths were studied. We demonstrate that lowering of the pH causes an increasing formation of lipid heterogeneities that do not always affect the membrane permeability. Only vesicles composed of lipid pairs with nonmatching chain lengths exhibit strong pH-responsive membrane permeability. Antibodyconjugated vesicles composed of nonmatching lipid acyl tails are shown to become endocytosed after selectively targeting ovarian cancer cells and to effectively release their contents as a response to the decreasing endosomal pH. In membranes composed of lipid pairs with different chain lengths, the formation of lipid domains or “model rafts” may occur. In particular, the domain hypothesis states that lipids, when in the gel state, can phase separate in the plane of a membrane that also contains lipids with fluid hydrocarbon chains. The separation is driven by intermolecular van der Waals attractions between rigid chains of the same length and hydrogen bonding of adjacent lipid headgroups.26–28 In membranes composed of lipid pairs with smaller differences in chain lengths and titratable headgroups, Garidel et al.11 demonstrated that nonideal mixing and phase separation occur and that they were shown to be affected mainly by pH. In particular, in membranes containing phosphatidylcholine and phosphatidic acid lipids with dimyristoyl (n ) 14) and dipalmitoyl (n ) 16) chain lengths, nonideal mixing was shown to increase with lowering pH due to its effect on reducing the electrostatic repulsion between PA headgroups and on increasing the attractive interactions between them via hydrogen bonds. The chain length differences between lipids were also shown to contribute to the formation of heterogeneities. In our studies, a similar effect of pH on the formation of lipid heterogeneities was observed by DSC. Our focus is however on the effect of these heterogeneities on membrane permeability. All constituent lipids were chosen to be in the gel state (chain lengths n ) 16 or 18) at the working temperature of 37 °C, because rigid lipid membranes in liposome-based drug delivery result in increased blood circulation times. In addition, all vesicles contained 2% mol PEGylated lipids (MW ) 2000) that contributes to increased circulation times in ViVo. In membranes composed of lipids with nonmatching acyl chains, the lipid with the longer chain length was chosen to have a titratable acidic headgroup. DSC measurements demonstrated that lowering of the pH results in the formation of lipid heterogeneities of increasingly high transition temperatures, suggesting the formation of domains (24) Fielding, C. J. Lipid Rafts and CaVeolae; Wiley-VCH Verlag GmbH & Co.KGaA: Weinheim, 2006; p 278. (25) Simons, K.; Ikonen, E. Nature 1997, 387, 569–572. (26) Dietrich, C.; Bagatolli, L. A.; Volovyk, Z. N.; Thompson, N. L.; Levi, M.; Jacobson, K.; Gratton, E. Biophys. J. 2001, 80, 1417–1428. (27) Feigenson, G. W.; Buboltz, J. T. Biophys. J. 2001, 80, 2775–2788. (28) Samsonov, A. V.; Mihalyov, I.; Cohen, F. S. Biophys. J. 2001, 81, 1486– 1500.

KarVe et al.

rich in the protonated longer chain lipids. In lipid mixtures with mismatched chain lengths, longer incubation over 5 days at 37 °C in low pH resulted in thermograms with similar temperature profiles as those in samples with 2 h incubation but with higher overall enthalpy changes suggested by the larger calorimetric peaks (see the Supporting Information). This result could possibly indicate the formation of aggregated or merged lipid heterogeneities at longer incubation times.29 Aggregation of lipid domains would decrease the total interfacial boundaries that contribute to lower transition temperatures due to lipid packing defects at the interface. At the same time, aggregation would increase the thermal contributions from lipids residing at the core of the heterogeneities that contribute to higher transition temperatures due to better lipid packing. Our content release studies show that membranes composed of lipids with mismatched chain lengths exhibit increased membrane permeability with lowering pH that is suggested to occur along the interfacial boundaries of the formed heterogeneities. In these membranes, lowering of the pH increases the extent of heterogeneities as well. In particular, content release studies show that the pH-dependent permeability of membranes does not only require the formation of lipid heterogeneities which were demonstrated by DSC to occur at decreasing pH for all systems studied, but in addition to this the chain lengths of the constituent lipids should be mismatched. This mismatch could possibly result in poor lipid packing at the interfacial boundaries of lipid heterogeneities causing extensive discontinuities (transient in time) that alter the membrane permeability. In membranes composed of lipids with matching chain lengths, the permeability rates were not strongly dependent on pH. Membrane permeability rates of vesicles composed of different ratios of lipids with nonmatching chain lengths were shown to be linearly proportional to the square root of the number of protonated DSPA lipids for each pH and composition studied. This is the simplest approximation of relatively small, circular heterogeneities. Other geometric models would possibly be more accurate as is suggested by fluorescent studies of giant lipid vesicles with separated domains where heterogeneities assemble in various shapes that are definitely nonregular or circular.27 The lateral diffusion coefficient of fluorescent lipid probes in supported pure DPPC bilayers (Tg ) 41 °C) has been reported to be 2 × 10-12 cm2/s at 17 °C below the Tg, 10-10 cm2/s at 37 °C, and 5 × 10-8 cm2/s above the Tg.30 In our studies, not only are the membranes composed of pure DPPC lipids that would decrease the lipid diffusion coefficient by 2 orders of magnitude at the working temperature of 37 °C, but in addition they contain DSPA lipids with a Tg of 75 °C. In this study, lateral diffusion of individual protonated DSPA lipids is suggested to result in clustering with other protonated DSPA lipids to increasingly form small heterogeneities at decreasing pH. We do not have information that correlates the diffusion rates of lipids forming membrane discontinuities with the permeability rates of membranes due to these discontinuities. We speculate that the slow diffusion rates of protonated DSPA lipids result in the slow formation of DSPA-rich phases with leaky interfacial boundaries. The content release rates for these leaky interfaces are measured to be 2 orders of magnitude slower than the release rates attributed to content release from packing imperfections that melt instantaneously at temperatures below 37 °C. We also have no information on the diffusion of these heterogeneous domains on the membrane surface or on the kinetics of their clustering and potential ripening. Assuming the rates of content release are (29) Veatch, S. L.; Keller, S. L. Biophys. J. 2003, 85, 3074–3083. (30) Tamm, L. K.; McConnell, H. M. Biophys. J. 1985, 47, 105–113.

Membrane Permeability in PC-PA Rigid Vesicles

indicative of the total perimeter of the leaky interfacial boundaries, slower leakage rates over time would suggest clustering. However, the content release measurements indicate a unimodal release mechanism over the time length of days. It may be possible for the heterogeneities to be kinetically trapped at a size smaller than the equilibrium size.29 The presence of cholesterol, even at the small fraction of 5% mol of total lipid, seems to dramatically decrease the permeability rates of these heterogeneous membranes without altering, however, the linear responsiveness toward the extent of leaky interfacial boundaries. We speculate, and we are currently investigating, that this is probably due to the formation of different types of heterogeneous lipid phases with altered interface boundaries or/and due to possibly preferential partition of cholesterol into the discontinuities along the interfacial boundaries. In the latter case, the lipid packing discontinuities within the interfacial boundaries may provide to cholesterol molecules additional packing space that is combined with the protective phospholipid headgroups acting like umbrellas.31 For pure DPPC PEGylated vesicles, even though the working temperature of 37 °C is very close to the lipid transition temperature of 41 °C, only a 10–20% content release was observed to occur in the first few minutes of incubation at 37 °C. These results are not in disagreement with previous studies that show an increase of membrane permeability close to the membrane Tg, since these reports refer to smaller than calcein water-soluble molecules.3,4 In addition, the vesicle membranes studied in the present work were designed to contain 5% mol cholesterol. Increased membrane permeability attributed to interfacial discontinuities between gel and fluid domains was reported by Clerc et al.3 in mixtures of phosphatidylcholine lipids with different chain lengths (n ) 14 and 16). We studied the formation of pH-induced lipid heterogeneities and membrane permeability in vesicles composed of equimolar DLPC and DSPA lipids that have nonmatching chain lengths and with DLPC (n ) 12) being in the liquid crystalline state at the working temperature of 37 °C. These vesicles exhibit a weak dependence of the content release rates on decreasing pH (see the Supporting Information). Also, these release rates are slower compared to the release rates of vesicles composed of equimolar DPPC and DSPA lipids with both lipids in the gel state at 37 °C. Mixtures of DLPC and DSPC lipids are reported to exhibit phase separation.20 In our studies, the headgroup of the longer acyl chain lipid was altered from phosphatidic choline to phosphatidic acid (DSPA). It is expected that, at least at pH ) 4.0, DSPA would be highly protonated and phase separation should occur due to a reduction of the electrostatic repulsion between PA headgroups and an increase in attractive interactions between them via hydrogen bonds. Even at this acidic pH, the measured membrane permeability was lower than the corresponding permeability in rigid membranes composed of DPPC and DSPA lipids. A possible explanation could be the fluidity of DLPC (Tg ) –1 °C) that, due to the faster dynamics of DLPC’s acyl chains, possibly does not allow for persistent (in time) membrane discontinuities at the interfacial boundaries of heterogeneities. In content release studies of membranes composed of DPPC and DSPA lipids, two independent release mechanisms were hypothesized and a double exponential function was used to fit the measured release rates. To verify that the higher exponential release rate is only due to the sudden temperature difference from room temperature to 37 °C that the system experiences at the beginning of measurements, content release profiles at decreasing pH values were evaluated for vesicles with and without (31) Huang, J.; Feigenson, G. W. Biophys. J. 1999, 76, 2142–2157.

Langmuir, Vol. 24, No. 11, 2008 5687

Figure 8. Effect of preincubation at 37 °C on the membrane permeability rates attributed to leaky interfacial boundaries in heterogeneous membranes composed of lipid pairs with nonmatching chain lengths. Membrane permeability was evaluated as the rate of calcein release at 37 °C versus pH in vesicles composed of equimolar DPPC and DSPA lipids with 5% mol cholesterol and 2% mol PEGylated lipids. (b) Content release rates of vesicles preincubated at 37 °C in phosphate buffer at pH ) 7.4 for 1 h before measurement of pH-dependent content release at 37 °C. (O) Content release rates at 37 °C of vesicles without preincubation. The error bars correspond to standard deviations of repeated measurements (two vesicle preparations, three samples per preparation per time point). The solid lines are guides to the eye.

preincubation at 37 °C. A single exponential function was adequate to fit the content release profiles of preincubated vesicles. In Figure 8, the exponential release rates attributed to the formation of leaky interfacial boundaries between heterogeneities due to decreasing pH are compared for vesicles with preincubation (open symbols) and without preincubation (closed symbols). Good agreement is observed at each pH value, verifying that the initial high membrane permeability observed for all vesicle compositions is due to the effects of the instantaneous temperature difference that is uncoupled from the effect of pH on the membrane permeability. To better characterize the discontinuities at the interfacial boundaries between the pH-tuned heterogeneities, the pHdependent release of fluorescent dextrans of larger sizes was studied. Smaller dextrans (MW ) 3 kDa, approximately 4.4 nm in diameter32) were released in a pH-dependent manner only from vesicles that contained the maximum fraction (50% mol) of DSPA lipid. These results suggest that for the equimolar lipid mixture the leaky interfacial boundaries are comparable to the thickness of the phospholipid bilayer membrane (4–5 nm).33 Large dextran particles (MW )10 kDa, approximately 5.7 nm in diameter32) were not released from any composition at all conditions studied, indicating an upper cutoff size on the discontinuities at the interfacial boundaries. These findings suggest that the size of membrane discontinuities could be a function of the extent of domain formation that depends on the total number of available DSPA lipids that are protonated and available to partition into domains. The ζ-potential of dextrans was measured at the pH values studied and was found to range between –14.1 and –6.5 mV, and –3.6 and –4.1 mV for the 3 kDa and 10 kDa dextrans, respectively, suggesting that a release mechanism other than direct diffusion across the membrane transient discontinuities would be highly unlikely. However, it is surprising that, contrary to the presence of these relatively large transient discontinuities, (32) Lang, I.; Scholz, M.; Peters, R. J. Cell Biol. 1986, 102, 1183–1190. (33) Israelachvili, J. Intermolecular and Surface Forces, 2nd ed.; Academic Press: London, 1992; p 450.

5688 Langmuir, Vol. 24, No. 11, 2008

vesicles retain their structure in solution and do not collapse into larger aggregates as is verified by DLS measurements. Such behavior could be possibly explained by the formation of heterogeneous domains on both membrane leaflets that are not mirror images of each other across the membrane bilayer but rather constantly diffuse in the plane of each membrane monolayer and occasionally coregister, or/and by the formation of heterogeneous domains which are not solely surrounded by a continuous well-differentiated interfacial periphery but their boundaries also contain areas less well-defined which are composed of lipids mixed with the surrounding membrane matrix, forming less differentiated structures that would sustain the intact vesicle structure. We are interested in membranes in the form of vesicles composed of lipids in the gel phase at 37 °C for potential use in drug delivery that can be triggered to release their contents using external stimuli such as pH. These vesicles can be ideal for targeted cancer therapy given that the interstitial space of solid tumors and intracellular compartments of cancer cells are shown to exhibit characteristic acidic gradients that could be used to trigger the release of encapsulated therapeutic agents from vesicles.34,35 In the presence of serum proteins, vesicles composed of DPPC and DSPA lipids exhibit faster release rates at decreasing pH. This could indicate more extensive lipid heterogeneities and therefore longer interfacial boundaries with decreasing pH that could depend on two mechanisms. The first mechanism is due to the protonation and clustering of DSPA lipids driven by the deletion of electrostatic repulsion that was shown to occur in phosphate buffer in the absence of serum proteins. The second mechanism could be due to the clustering of still charged DSPA lipids that are electrostatically attracted by the charged serum proteins adsorbing onto the vesicle surface. This is supported by the DSC thermographs using lysozyme as a model protein. Increasing content release with decreasing pH in media containing serum proteins could potentially be attributed to the higher effective positive charge of serum proteins at lower pH values.36,37 Although vesicle aggregation in the presence of serum proteins was detected with decreasing pH and with time, no fusion was observed among the aggregated vesicles that could potentially contribute to the observed faster release rates. In Vitro, immunolabeled vesicles composed of DPPC and DSPA lipids demonstrated greater specific binding to human ovarian carcinoma cells (SKOV3-NMP2) compared to nontargeted vesicles. In addition, they exhibited fast and extensive content release into the endosomal compartment. The extent of bound immunolabeled vesicles did not exceed 10% of the total vesicles, that agrees with the maximum reported bound fractions on antibody-labeled vesicles for comparable antigen densities on targeted cells.16 This could be due to the relatively large size of vesicles that may interfere with the extracellular matrix by obstructing free diffusion toward the cell surface. The pH-sensitive content release was retained by these vesicles in Vitro, that makes them promising candidates for the delivery of therapeutic agents at potentially high bioavailability levels. pH-sensitive liposomes for triggered content release intracellularly have been extensively studied. To retain liposome stability through rigidity and to combine triggered content release, polymerized mixed liposomes have been suggested.38 In these (34) Drummond, D. D.; Zignani, M.; Leroux, J. C. Prog. Lipid Res. 2000, 39, 409–460. (35) Martin, G. R.; Jain, R. K. Cancer Res. 1994, 54, 5670–5674. (36) Cann, J. R.; Brown, R. A.; Kirkwood, J. G. J. Biol. Chem. 1949, 181, 161–170. (37) Tiselius, A. Biochem. J. 1937, 31, 313–317.

KarVe et al.

systems, polymerized lipids are phase-separated from nonpolymerized lipids in the plane of the membrane, and the release of encapsulated contents is subsequently triggered by pH that causes uncorking of the liposome membrane. However, these systems are limited to small water-soluble encapsulated molecules. Another approach entails liposomes with membranes that form inverted hexagonal phases at acidic pH, causing membrane fusion and content release.39,40 These liposomes however are composed of fluid lipid membranes that limit the blood circulation times of these drug carriers. Addition of PEGylated lipids improves the circulation times of liposomes but decreases their pH-sensitive character that requires membrane contact for membrane fusion and content release.41 The liposome membranes studied in this work are rigid, composed of gel-phase lipid pairs with nonmatching chain lengths. They respond to the acidic endosomal pH by releasing their contents fast and to a great extent by a mechanism that does not require fusion and is therefore not influenced by the presence of PEGylated lipids. Therefore, these rigid pH-sensitive liposomes could be promising as drug carriers to vascularized tumors.

5. Conclusion Gel-phase membranes containing lipid pairs with variable chain lengths (n ) 16 and 18) and titratable headgroups were studied in the form of vesicles. Lowering of the pH increases the formation of lipid heterogeneities that exhibit enhanced membrane permeability only for lipid pairs with nonmatching chain lengths. It is suggested that this is due to defective packing at the interfacial boundaries of lipid heterogeneities between lipid pairs with nonmatching chain lengths. In phosphate buffer, this is supported by the following findings: (1) content release from vesicles is shown to be pH-dependent and to correlate with the formation of lipid heterogeneities on the vesicle membranes (DSC studies); (2) vesicle membranes composed of lipids with matching chain lengths exhibit pH-induced formation of heterogeneities without a strong pH-dependence in membrane permeability; and (3) no vesicle aggregation or vesicle fusion was shown to occur at decreasing pH conditions that could indicate a different leakage mechanism. Content release rates from vesicles in serum supplemented media are pH-sensitive and are faster probably due to the accelerated formation of lipid heterogeneities driven by the adsorption of charged proteins on the vesicle surface. These vesicles retain their pH-controlled content release intracellularly after antibody-mediated binding to and internalization by SKOV3-NMP2 ovarian cancer cells. These vesicles may have potential as delivery carriers for the targeted therapy of solid tumors. Acknowledgment. We are grateful to the Wallace H. Coulter Foundation (Early Career Award in Translational Research) and to Polytechnic University for their support. Supporting Information Available: Content release versus pH in DPPC/DSPA vesicles; content release versus pH in DLPC/DSPA vesicles; DSC versus pH and versus time of DPPC/DSPA membranes; and content release of fluorescent dextrans from DPPC/DSPA vesicles. This material is available free of charge via the Internet at http://pubs.acs.org. LA800331A (38) Ringsdorf, H.; Schlarb, B.; Venzmer, J. J. Angew. Chem., Intl. Ed. Engl. 1988, 27, 113–158. (39) Bergstrand, N.; Arfvidsson, M. C.; Kim, J. M.; Thompson, D. H.; Edwards, K. Biophys. Chem. 2003, 104, 361–379. (40) Ellens, H.; Bentz, J.; Szoka, F. C. Biochemistry 1984, 23, 1532–1538. (41) Slepushkin, V. A.; Simoes, S.; Dazin, P.; Newman, M. S.; Guo, L. S.; Pedroso de Lima, M.; Düzgünes, N. J. Biol. Chem. 1997, 272, 2382–2388.