pubs.acs.org/Langmuir © 2009 American Chemical Society
Membrane Heterogeneities and Fusogenicity in Phosphatidylcholine-Phosphatidic Acid Rigid Vesicles as a Function of pH and Lipid Chain Mismatch Manali Bhagat and Stavroula Sofou* Othmer-Jacobs Department of Chemical and Biological Engineering, Polytechnic Institute of New York University, Six MetroTech Center, Brooklyn, New York 11201 Received July 17, 2009. Revised Manuscript Received September 14, 2009 The role of pH-dependent lipid heterogeneities on the fusogenicity of membranes was evaluated on model lipid bilayers in the form of unilamellar vesicles composed of lipid pairs at a fixed equimolar ratio of phosphatidylcholine (PC) and phosphatidic acid (PA) headgroups. The pH and the hydrophobic composition (lipid acyl tails) of membranes were systematically altered, and their effect on vesicle aggregation, membrane fusogenicity, content release, and content mixing was evaluated. Lowering pH increases the extent of protonated PA headgroups forming phase-separated PA-rich heterogeneities and giving rise to molecular packing defects originating at the phase boundaries. Phase boundaries within the membrane’s hydrophobic portion are affected by the lipid acyl-tail dynamics (fluidity) and the matching or nonmatching lengths of the acyl tails of lipid pairs comprising the membrane. The aggregates’ size increases with lowering pH and is independent of the membrane’s hydrophobic composition. Contrary to aggregation, the initial rates of lipid mixing are proportional to pH and also depend on membrane’s hydrophobic composition. The apparent lipid-mixing rate constants are greater for membranes containing lipid pairs with mismatched acyl-tail lengths, followed by pairs with matching acyl tails in the gel state and by pairs with matching tails where one lipid is close to its transition temperature. Addition of cholesterol conserves the independence of vesicle aggregation from the membrane’s hydrophobic composition. However, it decreases the aggregation rates and inverts the tendency for fusion, among the three types of hydrophobic compositions, suggesting a role of cholesterol’s preferential partition in different regions of membrane’s heterogeneities. We propose a phenomenological model where the membrane phase boundaries containing molecular packing defects act as “sticking points” on the vesicle exterior via which vesicles aggregate upon contact followed by defect merging via intervesicle lipid exchange and mixing. Such heterogeneous bilayers in the form of drug encapsulating liposomes may potentially improve the therapeutic efficacy by fusing with endosomal membranes, thus increasing drug bioavailability.
1. Introduction Fusion of lipid membranes is essential for critical cell functions and is part of the action mechanisms of several diseases or viral infections.1 Fusion in nature is usually the result of orchestrated actions of a variety of molecular factors that are necessarily being regulated by several proteins.2,3 Understanding membrane fusion may ultimately impact the control and therapy of several diseases. In the absence of proteins or peptides inducing fusion, extensive studies in model membranes have described the role of fundamental physicochemical properties of lipids and lipid membranes within the framework of generalized fusion mechanisms.4,5 In the case of lipid membranes in the form of vesicles, the general fusion framework includes vesicle contact followed by decrease of the interbilayer distances that would ultimately lead to lipid exchange or mixing. Contact and membrane dehydration that decreases the interbilayer distances has been reported to be driven by cations6 and nonelectrolyte polymers,7,8 depending on the nature of the *Corresponding author: Ph 718 260 3863; e-mail
[email protected].
(1) Eckert, D.; Kim, P. Annu. Rev. Biochem. 2001, 70, 777–810. (2) Monck, J.; Fernandez, J. Neuron 1994, 12, 707–716. (3) Ohya, T.; Miaczynska, M.; Coskun, U.; Lommer, B.; Runge, A.; Drechsel, D.; Kalaidzidis, Y.; Zerial, M. Nature 2009, 459, 1091–1097. (4) Nir, S.; Bentz, J.; Wilschut, J.; Duzgunes, N. Prog. Surf. Sci. 1983, 13, 1–124. (5) Wilschut, J.; Hoekstra, D. Membrane Fusion; Marcel Dekker, Inc.: New York, 1990. (6) Papahadjopoulos, D.; Nir, S.; Duzgunes, N. J. Bioenerg. Biomembr. 1990, 22, 157–179. (7) Lentz, B. Eur. Biophys. J. 2007, 36, 315–326. (8) Lentz, B.; Lee, J. Mol. Membr. Biol. 1999, 16, 279–296.
1666 DOI: 10.1021/la9026283
lipid headgroups comprising the membrane. Alternatively, phase boundaries on the lipid membrane surface have been proposed by Papahadjopoulos et al.9 to trigger lipid mixing and vesicle fusion. These studies suggested that the presence of fluid/gel phase boundaries which were formed due to formation of crystallized lipid domains;induced by calcium’s electrostatic adsorption on negatively charged lipid membranes;may result in transient (in time) bilayer defects or “defective interfaces” due to molecular packing discontinuities, therefore possibly exposing hydrocarbon regions to water and serving as focal points for fusion following the initial vesicle aggregation. In this work, we use pH to trigger lipid phase separation and formation of fluid/gel and gel/gel phase boundaries in heterogeneous lipid membranes composed of lipids with zwitterionic (phosphatidylcholine, PC) and titratable (phosphatidic acid, PA) headgroups.10,11 The effect of pH on nonideal lipid mixing in membranes composed of lipid pairs with same headgroups (PC and PA) has been studied before.10 We systematically alter the pH and the hydrophobic composition (lipid acyl tails) of membranes, and we evaluate their effect on vesicle aggregation and fusion as expressed in terms of lipid mixing and mixing of encapsulated contents of vesicles. Lowering pH increases the (9) Papahadjopoulos, D.; Vail, W.; Newton, C.; Nir, S.; Jacobson, K.; Poste, G.; Lazo, R. Biochim. Biophys. Acta 1977, 465, 579–598. (10) Garidel, P.; Johann, C.; Blume, A. Biophys. J. 1997, 72, 2196–2210. (11) Karve, S.; Bajagur Kempegowda, G.; Sofou, S. Langmuir 2008, 24, 5679– 5688.
Published on Web 10/08/2009
Langmuir 2010, 26(3), 1666–1673
Bhagat and Sofou
extent of protonated PA headgroups forming phase-separated PA-rich heterogeneities and giving rise to molecular packing defects originating at the phase boundaries.10,11 In an attempt to vary the molecular packing discontinuities within the membrane’s hydrophobic portion, the lipid acyl-tail dynamics (membrane fluidity) and the relative lengths of the acyl tails (matching or nonmatching) of lipid pairs are altered.10,11 The role of cholesterol on altering the lipid packing hierarchy and, therefore, affecting the fusogenicity of lipid vesicles is also investigated. In particular, unilamellar vesicles were composed of lipid pairs at the fixed equimolar ratio of PC and PA lipids with matching and nonmatching acyl tail lengths (n = 16 and 18), and all studies were performed at 37 C. Vesicle suspensions were characterized for (1) formation of phase-separated lipid domains using differential scanning calorimetry, (2) aggregation, (3) fusogenicity (total lipid mixing, content mixing, and inner leaflet lipid mixing), and (4) content release of encapsulated solutes, in the presence and absence of cholesterol at decreasing pH values over time. Measurements in the present study were conducted at the physiologically relevant temperature of 37 C since these vesicles may potentially be utilized as drug delivery carriers to malignant cells.12,13 Acidification of the endosomal pathway14 during cellular uptake of these vesicles may activate phase separation of the vesicle membranes forming phase boundaries of high fusogenic potential. These heterogeneities may result in mixing of vesicle lipids with the endosomal membrane resulting in direct delivery of the encapsulated therapeutic agents into the cytoplasm increasing bioexposure of malignant cells to the delivered therapy. This study aims to contribute to the current understanding of fusion of lipid membranes with emphasis on potential mechanisms involving lipid phase boundaries with molecular packing defects. Our approach to correlate collective properties of heterogeneous membranes to simplified phenomenological mechanisms may ultimately be useful in enabling design and engineering of devices15 performing specific tasks for applications beyond the one mentioned above.
Article
2.1. Materials. 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), and 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine-N-(7-nitro-2-1,3-enzoxadiazol-4-yl) (ammonium salt) (NBD-lipid) were purchased from Avanti Polar Lipids (Alabaster, AL) (all lipids at purity >99%). 1-Hexadecanoyl2-(1-pyrenedecanoyl)-sn-glycero-3-phosphocholine (pyrene-PC lipid), l-aminonaphthalene-3,6,8-trisulfonic acid (ANTS), and N, N0 -p-xylylenebis(pyridinium bromide) (DPX) were obtained from Invitrogen Molecular Probes (Carlsbad, CA). Trizma buffer, sodium dithionite, calcein, cholesterol, EDTA, 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. 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 50-55 C. The lipid suspension (3 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 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). 2.3. 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 90 while samples were incubated at 37 C. Particle size distributions at each angle were calculated from autocorrelation data analysis by CONTIN.16 All buffer solutions used were filtered with 0.22 μm filters just before vesicle preparation. The collection time for the autocorrelation data was 2 min. Vesicles were incubated in different solutions (phosphate buffer at different pH values) at 37 C and were measured over time. The total lipid concentration for DLS measurements was 50 and 100 μM. 2.4. Differential Scanning Calorimetry. Differential scanning calorimetry (DSC) studies were performed using a VP-DSC apparatus (MicroCal, LLC, Northampton, MA). DSC scans of vesicle suspensions (0.5 mL, 3.0 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/h. The thermograms of the corresponding buffers were acquired at identical conditions and were subtracted from the excess heat capacity curves. 2.5. Total Lipid Mixing Assay. Pyrene-labeled lipids were included in vesicle compositions and were used to monitor lipid mixing among vesicles at various pH values at 37 C. The excimer (E; excitation: 340 nm; emission: 470 nm) and monomer (M; excitation: 340 nm; emission: 396 nm) fluorescence spectra of pyrene molecules depend on their distance from other pyrenelabeled lipids parallel to the plane of the lipid bilayer17 and also on the viscosity of their local environment which is assumed not to change significantly in these studies.18 The total lipid mixing assay employs mixing of two vesicle populations: the first containing pyrene-labeled lipids on both lipid leaflets and the second not fluorescently labeled. The assay monitors the average separation of the pyrene-labeled lipids as they are redistributed over the fused vesicle membranes. Vesicle fusion increases the effective distances between fluorophores, thus increasing the contributions attributed to monomers as compared to excimers resulting in an increase in the M/E ratios. Three sets of vesicles were prepared for the assay: (1) equimolar phosphatidylcholine and phosphatidic acid vesicles containing 4 mol % pyrene-PC lipid, (2) vesicles of the same composition but without fluorescent lipids, and (3) equimolar phosphatidylcholine and phosphatidic acid vesicles containing 0.67 mol % pyrene-PC lipid corresponding to 100% lipid mixing of the above two vesicle preparations mixed at 1 to 5 ratio (ratio of labeled to nonlabeled vesicles). For these studies, lipid chloroform solutions were dried under vacuum as described in section 2.2 for 10 min and were then hydrated using PBS with 1 mM EDTA (300 mOsm, pH 7.4). Aliquots from the labeled vesicle suspensions;set numbered (1);and nonlabeled vesicles;set numbered (2);were mixed at 1 to 5 ratio at total lipid concentration of 50 and 100 μM in a 3 mL volume. The ratios of fluorescence intensities (M/E)0 originating from the suspension of labeled-only vesicles at concentrations equal to one-sixth of 50 and 100 μM lipid were set to 0% of total lipid mixing. The ratios of
(12) Alaouie, A.; Sofou, S. J. Biomed. Nanotechnol. 2008, 4, 234–244. (13) Sofou, S. Nanomedicine 2007, 2, 711–724. (14) Mellman, I. J. Exp. Biol. 1992, 172, 39–45. (15) Bajagur Kempegowda, G.; Karve, S.; Adhikari, A.; Bandekar, A.; Khaimchayev, T.; Sofou, S. Langmuir 2009, 25, 8144–51.
(16) Provencher, S. W. Comput. Phys. Commun. 1982, 27, 213–227. (17) Turro, N. J. Modern Molecular Photochemistry; University Science Books: Sausalito, CA, 1991; p 628. (18) Vauhkonen, M.; Sassaroli, M.; Somerharju, P.; Eisinger, J. Biophys. J. 1990, 57, 291–3000.
2. Materials and Methods
Langmuir 2010, 26(3), 1666–1673
DOI: 10.1021/la9026283
1667
Article
Bhagat and Sofou
Figure 1. DSC scans of vesicles containing equimolar ratios of lipid pairs with zwitterionic (phosphatidylcholine, PC) and titratable (phosphatidic acid, PA) headgroups and with acyl tails of matching and nonmatching lengths. Membranes containing (A) lipids with nonmatching acyl-tail lengths DPPC/DSPA (Tm’s: 41 C/75 C), (B) lipids with matching acyl-tail lengths DPPC/DPPA (Tm’s: 41 C/68 C), and (C) lipids with matching acyl-tail lengths DSPC/DSPA (Tm’s: 55 C/75 C). DSC scans from 25 to 85 C at a scan rate of 5 C/h were performed on vesicle suspensions with both lipid leaflets exposed to the same pH value (7.4, 5.5, 5.0, and 4.0). fluorescence intensities (M/E)100 originating from the suspension of labeled vesicles with 0.67% pyrene-PC lipid at concentrations equal to 50 and 100 μM lipid were set to 100% of total lipid mixing. The fraction of total lipid mixing was calculated to be equal to fraction of total lipid mixing ðtÞ ¼
½ðM=EÞt -ðM=EÞ0 ½ðM=EÞ100 -ðM=EÞ0
where (M/E)t is the sample’s M/E ratio at each time t, (M/E)100 corresponds to 100% fusion (measured at each time t), and (M/E)0 corresponds to 0% fusion (measured at each time t). 2.6. Mixing of Vesicle Contents. Two separate vesicle populations were prepared: the first encapsulating the fluorophore ANTS (25 mM ANTS, 40 mM NaCl, 10 mM Tris-HCl at pH 7.4) and the second population the quencher DPX (90 mM DPX, 50 mM NaCl, 10 mM Tris-HCl at pH 7.4).19 Mixing of the encapsulated aqueous contents, due to vesicle fusion, results in quenching of ANTS’s intensity. Aliquots from each type of vesicle suspensions were mixed at 1 to 1 mole ratio at total lipid concentration of 50 and 100 μM in a 12 mL volume and were incubated at 37 C. At different time points aliquots from the parent vesicle suspension were sampled, released ANTS and DPX were removed by a Sephadex G-50 column (of 11 cm length) eluted with phosphate buffer (1 mM EDTA, pH 7.4), and the fluorescence intensities of encapsulated ANTS (excitation: 360 nm; emission: 530 nm) were measured before and after addition of Triton-X 100 that relieves the potential quenching by DPX caused by content mixing among the two populations of vesicles. To assess the maximum ANTS quenching by DPX, vesicles were prepared containing both ANTS and DPX at the same mole ratio as above. The fluorescence intensity of ANTS before and after addition of Triton-X 100 indicated 86% quenching efficiency by DPX. At each time point, the fraction of content mixing was calculated to be equal to fraction of content mixing ðtÞ ¼
½ΔIt ½ΔImax
where ΔIt is the fractional fluorescence quenching at time t of the vesicle suspensions and ΔImax is the maximum fractional quenching (86%) corresponding to 100% of content mixing. For 0% of content mixing, the fractional quenching is zero. 2.7. Content Retention during Vesicle Fusion. To evaluate the release of encapsulated contents during vesicle fusion, vesicles containing self-quenching concentrations of calcein (55 mM in PBS at pH 7.4) were incubated in phosphate buffer at 37 C over time at different pH values (pH 7.4, 5.5, 5.0, and 4.0). The (19) Ellens, H.; Bentz, J.; Szoka, F. C. Biochemistry 1984, 23, 1532–1538.
1668 DOI: 10.1021/la9026283
concentration of lipids during incubation was 50 and 100 μM. Release of calcein from vesicles and its dilution in the surrounding solution result in increased fluorescence intensities due to relief of self-quenching. Calcein release was measured over time for up to 5 days by removing 0.5 mL of sample volume from the parent suspension and monitoring the fluorescence intensities upon dilution in 3 mL of phosphate buffer (1 mM EDTA, pH 7.4). The ratios of calcein fluorescence (excitation: 495 nm; emission: 515 nm) after addition of Triton-X 100 over the intensities before addition of Triton-X 100 were measured using a Fluoromax-2 spectrofluorometer (Horiba Jobin Yvon, Edison, NJ) and were equated to calcein’s quenching efficiency, Q. The percentage of retained contents with time was calculated as follows: Qt -Qmin % calcein content retention ðtÞ ¼ 100 Qmax -Qmin where Qt is calcein’s quenching efficiency at the corresponding time point t, Qmax is the maximum quenching efficiency in phosphate buffer (at pH 7.4) at room temperature immediately after preparation of vesicles, and Qmin is the minimum quenching efficiency equal to unity.
3. Results 3.1. Extent of Formation of Lipid Heterogeneities Is Affected by Changes in pH. The extent of formation of lipid heterogeneities was studied on vesicles containing equimolar ratios of lipid pairs with zwitterionic (phosphatidylcholine, PC) and titratable (phosphatidic acid, PA) headgoups and with acyl tails of matching and nonmatching lengths at the pH values of 7.4, 5.5, 5.0, and 4.0. Parts A, B, and C of Figure 1 show that all lipid pairs studied, DPPC/DSPA (Tm’s: 41 C/75 C), DPPC/DPPA (Tm’s: 41 C/68 C), and DSPC/DSPA (Tm’s: 55 C/75 C), respectively, exhibit a monotonic shift of their thermal spectra toward higher temperatures with decrease in pH from 7.4 to 4.0. In addition, all membrane compositions exhibit increase of the width of their thermal responses with lowering pH from 7.4 to 4.0. All thermal spectra exhibit a clear contribution of two major thermal transitions whose spectral distance increases with lowering pH. These distinct thermal transitions indicate heterogeneous membranes with distinct lipid phases. The observed pHdependent response is attributed to the protonation of the phosphate group of PA that possibly results in hydrogen bonding between protonated and nonprotonated PA headgroups,20 forming, therefore, heteorogeneous domains rich in phosphatidic acid lipids. Attractive van der Waals interactions among the matching (20) Boggs, J. M. Biochim. Biophys. Acta 1987, 906, 353–404.
Langmuir 2010, 26(3), 1666–1673
Bhagat and Sofou
Article
Figure 2. Vesicle aggregation is enhanced with lowering pH. Membranes containing (A) lipids with nonmatching acyl-tail lengths DPPC/ DSPA, (B) lipids with matching acyl-tail lengths DPPC/DPPA, and (C) lipids with matching acyl-tail lengths DSPC/DSPA, at 50 μM total lipid concentration. Aggregation studies were performed on vesicle suspensions in PBS at 37 C using DLS. All vesicles were prepared with encapsulated PBS at pH 7.4, and at t = 0 the suspension’s pH was adjusted to (b) pH 7.4, (1) pH 5.0, and (4) pH 4.0. Error bars correspond to standard deviations of repeated measurements (two vesicle preparations, one sample per preparation). Lines are guides to the eye.
Figure 3. Vesicle aggregation as a function of lowering pH for vesicles containing 5 mol % cholesterol. Membranes containing (A) lipids with nonmatching acyl-tail lengths DPPC/DSPA, (B) lipids with matching acyl-tail lengths DPPC/DPPA, and (C) lipids with matching acyl-tail lengths DSPC/DSPA, at 50 μM total lipid concentration. Aggregation studies were performed on vesicle suspensions in PBS at 37 C. All vesicles were prepared with encapsulated PBS at pH 7.4, and at t = 0 the suspension’s pH was adjusted to (b) pH 7.4, (1) pH 5.0, and (4) pH 4.0. Error bars correspond to standard deviations of repeated measurements (two vesicle preparations, one sample per preparation). Lines are guides to the eye.
acyl-tail lengths of phosphatidic acid lipids could also favorably contribute to the observed response in membrane reorganization.21 The thermograms of lipid membranes after their incubation at 37 C for 4 days exhibit similar response with respect to pH. For lipid pairs with matching acyl-tail lengths the thermal spectra at this later time point are shifted toward higher transition temperatures (see Figures S1A, S1B, and S1C in the Supporting Information). Addition of cholesterol at 5 mol % does not alter the general shift of thermal transitions toward higher temperatures with lowering pH, but it does flatten the transitions resulting in smoother thermograms (see Figures S1D, S1E, and S1F in the Supporting Information). 3.2. Aggregation of Vesicles with Heterogeneous Membranes Increases with Lowering pH and Is Independent of Acyl-Tail Composition. All vesicle suspensions composed of lipid pairs as in section 3.1, forming lipid heterogeneities with lowering pH, exhibit increased aggregation upon decrease of the suspension’s pH from 7.4 to 4.0 for nonmatching (Figure 2A) and matching lipid acyl-tail lengths (Figure 2B,C). The initial rates of aggregation for all vesicles composed of equimolar ratios of PC and PA lipids increase with lowering pH and, interestingly, are not a function of the acyl-tail composition (acyl-tail length and/or acyl-tail mismatch). After 24 h of incubation, the extent of aggregation is greater at the lowest pH value for all vesicle compositions. For all vesicle compositions and at pH 4.0 and 5.0, the extent of aggregation increases with time, reaching an (21) Tokutomi, S.; Ohki, K.; Ohnishi, S. I. Biochim. Biophys. Acta 1980, 596, 192–200.
Langmuir 2010, 26(3), 1666–1673
average measured size after 48 h that is greater than 900% of the initial vesicle size which was ∼110 ( 16 nm in diameter, at t = 0. The measured aggregate sizes at t = 48 h are not included in Figure 2 since they are beyond the measuring range of DLS. Doubling of the vesicle concentration, from 50 to 100 μM total lipid, accelerates the aggregation rates on all three vesicle compositions in a similar way. At the lower vesicle concentration shown in Figure 2, the limit of detection by DLS is reached after 48 h of incubation at 37 C and at pH 4.0. At the double vesicle concentration at pH 4.0 and 37 C, this limit is reached only after 3 h of incubation (see Figures S2A, S2B, and S2C in the Supporting Information). The increase in aggregate size at pH 7.4 both at the low and high vesicle concentration did not exceed 214% for all vesicle compositions over the period of 4 days. Addition of cholesterol at 5 mol % does not cancel the dependence of aggregation on pH as shown in Figure 3. Furthermore, addition of cholesterol conserves the independence of aggregation from the hydrophobic composition of the vesicle’s membrane. However, cholesterol significantly decreases the extent of aggregation with respect to pH and time for all three vesicle compositions in a similar manner: the maximum aggregation size that is reached after 4 days of incubation at 37 C and at pH 4.0 is ∼312 ( 10 nm for all three vesicle compositions. Vesicle suspensions of compositions not including cholesterol form large aggregates starting after 48 h of incubation at the lowest pH value of 4.0. These aggregates are of the order of millimeters, they are visible by the naked eye, and they look like suspended cotton balls. Measurements from this state of aggregation were not included in this study. In the presence of cholesterol, DOI: 10.1021/la9026283
1669
Article
Bhagat and Sofou
Figure 4. Total lipid mixing of vesicles increases with lowering pH. Membranes containing (A) lipids with nonmatching acyl-tail lengths DPPC/DSPA, (B) lipids with matching acyl-tail lengths DPPC/DPPA, and (C) lipids with matching acyl-tail lengths DSPC/DSPA in the gel phase, at 50 μM total lipid concentration. Total lipid mixing studies were performed on vesicle suspensions in PBS at 37 C. All vesicles were prepared with encapsulated PBS at pH 7.4, and at t = 0 the suspension’s pH was adjusted to (b) pH 7.4, (O) pH 5.5, (1) pH 5.0, and (4) pH 4.0. Error bars correspond to standard deviations of repeated measurements (two vesicle preparations, three samples per preparation). Lines are guides to the eye.
Figure 5. Total lipid mixing versus pH of vesicles containing 5 mol % cholesterol and (A) lipid pairs with nonmatching acyl-tail lengths DPPC/DSPA, (B) lipid pairs with matching acyl-tail lengths DPPC/DPPA, and (C) lipid pairs with matching acyl-tail lengths DSPC/DSPA in the gel phase, at 50 μM total lipid concentration. (b) pH 7.4; (O) pH 5.5; (1) pH 5.0; (4) pH 4.0. Error bars correspond to standard deviations of repeated measurements (two vesicle preparations, three samples per preparation). Lines are guides to the eye.
formation of suspended large aggregates was not detected for any of these compositions over the time period of 4 days at 37 C. 3.3. Total Lipid Mixing among Vesicles with Heterogeneous Membranes Increases with Lowering pH and Is Dependent on Acyl-Tail Composition. Upon lowering of the suspension’s pH from 7.4 to 4.0, vesicles composed of lipid pairs with nonmatching acyl-tail lengths (Figure 4A) exhibit increasing total lipid mixing. The pH dependence in total lipid mixing, but at a lower extent, is also exhibited between pH 4 and the other pH values by vesicles with membranes containing lipid pairs with matching acyl tails of DPPC/DPPA lipids where DPPC is close to its transition temperature (Figure 4B). Vesicles composed of lipid pairs with matching acyl tails in the gel phase (DSPC/DSPA) exhibit increase in total lipid mixing with time which is pH-independent. In particular, after 24 h of incubation lipid vesicles with nonmatching acyl tails exhibit the following extents of total lipid mixing: 35 ( 3%, 37 ( 2%, 49 ( 4%, and 71 ( 6% at pH 7.4, 5.5, 5.0, and 4.0, respectively. At this late time point, the extents of total lipid mixing of vesicles composed of matching acyl-tail lipids do not exceed 50 ( 3% for DPPC/DPPA (Figure 4B, pH 4.0) and 43 ( 4% for DSPC/DSPA lipids in the gel phase (Figure 4C, pH 4.0). Doubling of the vesicle concentration, from 50 to 100 μM total lipid, accelerates the initial total lipid mixing rates on vesicle compositions with nonmatching acyl-tail lengths at all pH values studied. After 2 h of incubation, the extents of total lipid mixing from pH 7.4 to 4.0 range from 37 ( 3% to 50 ( 3% (data not shown). 1670 DOI: 10.1021/la9026283
Figure 5 shows that pH dependence in total lipid mixing is conserved by all vesicle compositions upon addition of 5 mol % cholesterol. However, except for the results at pH 4.0, none of the vesicle compositions studied exhibits extents of total lipid mixing greater than 40% even after 4 days of incubation. Interestingly, the tendency for fusion is reversed in the presence of cholesterol at pH 4.0 and 37 C, with the vesicle membranes containing lipid pairs with stiff, matching acyl tails (Figure 5C showing DSPC/ DSPA with Tm’s: 55 C/75 C) to exhibit the greatest lipid mixing that reaches 75 ( 0% (at pH = 4.0) after 4 days of incubation followed by vesicles with membranes containing lipid pairs with matching and more fluid matching acyl tails of DPPC/DPPA lipids (60 ( 2% at pH = 4.0, Figure 5B, Tm for DPPC is 41 C) and, last, by vesicles composed of DPPC/DSPA lipid pairs with nonmatching acyl tails (43 ( 1% at pH = 4.0, Figure 5A). Vesicles formed by uniform membranes not bearing explicit charge, containing only the zwitterionic lipid component DPPC, exhibit insignificant aggregation and minimal total lipid mixing at the same experimental conditions (see Figures S3A and S3B in the Supporting Information). 3.4. Content Mixing of Vesicles with Heterogeneous Membranes and Nonmatching Acyl-Tail Lengths Increases with Lowering pH. Vesicles composed of lipids with nonmatching acyl-tail lengths show extents of content mixing between 2 and 24 h of incubation at 37 C that increase from 15.3% to 16.5% at pH 5.5 and from 9.6% to 12.5% at pH 7.4. Content mixing measurements at the two lowest pH values of 5.0 and 4.0 did not result in meaningful results due to extensive release of encapsulated contents (vide infra). Langmuir 2010, 26(3), 1666–1673
Bhagat and Sofou
Article
3.5. Permeability of Vesicle Membranes with Heterogeneous Membranes Increases with Lowering pH. Lowering pH results in increase in the release of encapsulated calcein for vesicles composed of lipid pairs with nonmatching (DPPC/DSPA) and matching acyl-tail lengths (DPPC/DPPA and DSPC/DSPA) (see Figures S4A, S4B, and S4C, respectively, in the Supporting Information). Addition of cholesterol at 5 mol % generally increases the content retention at all time points, pH values, and vesicle compositions studied (see Figures S4D, S4E, and S4F, respectively, in the Supporting Information).
4. Discussion In the present study, we explore the role of pH-induced lipid heterogeneities with defective phase boundaries on triggering fusion of lipid membranes. These discontinuous interfaces, formed due to molecular packing defects among lipids,9 are suggested to originate at the boundaries between the laterally phase-separated lipid domains10 and have previously been shown to alter other collective properties of model lipid bilayers such as permeability.11 In particular, on intact vesicles kept autonomous by lipid PEGylation, enhancement in membrane permeability was observed with lowering pH, and the rates of content release were shown to correlate linearly with the calculated “total length” of these defective transient interfaces formed between the growing phase-separated lipid domains.11 These leaky interfaces were attributed to transient (in time) defects in molecular packing among the hydrophobic tails of lipids residing at these phase boundaries.9 To a first approximation, the “total length” of these defective phase boundaries was suggested to be proportional to the square root of the number of protonated PA lipids at each pH value assuming formation of small, circular phase-separated domains rich in PA lipids20 (the number of protonated PA lipids being proportional to the surface area of phase-separated domains). Although the proposed model, and “total length” approximation, is an oversimplification, it adequately describes the observed behavior of membrane permeability.11 In the present studies, model membranes, in the form of unilamellar vesicles without PEGylation, are formed by mixtures of pairs of lamellar-forming lipids. Lipid headgroups are chosen to be of two types: the first one is titratable within the pH range of interest (phosphatidic acid (PA), in this study), and the second type is zwitterionic and not titratable (phosphatidylcholine, PC). Decrease in pH increases the extent of protonation on PA lipids, which then form hydrogen bonds with other PA headgroups.20 Therefore, it is suggested that lowering of pH results in increased extents of phase-separated lipid phases rich in PA lipids and, to a first approximation, in formation of longer transient defective interfaces/phase boundaries between the phase-separated lipid domains.20 Lipid acyl tails are chosen to be also of two types: the first one has a length of 16 carbons (palmitoyl) and the second type a length of 18 carbons (stearoyl), therefore affecting the corresponding lipid fluidity at the working temperature of 37 C. We introduce the notion of the “degree” of interfacial discontinuities that is expected to be affected by the fluidity and match (or mismatch) of the acyl-tail lengths of lipids residing within the phase boundaries. Accordingly, slowing down the lipid acyl-tail dynamics at the working temperature (37 C) and introducing lipid pairs with nonmatching acyl tails is suggested to increase the spatiotemporal extent of transient lipid packing discontinuities at the interfaces and, therefore, to enhance the “degree” of defects. We observe that vesicle aggregation increases with lowering pH and is independent of the composition of the acyl-tail component of lipids. Vesicle fusion, however, as expressed by lipid mixing, is Langmuir 2010, 26(3), 1666–1673
dependent both on the pH and on the composition of the hydrophobic component of the bilayers. Fusion increases with lowering pH and with enhancement of the “degree” of interfacial defects that were previously correlated to the dynamics of lipid acyl tails residing at these defective interfaces.11 In the present studies, content release from vesicles was found to increase with lowering pH and is attributed to two parallel processes: (a) phase separation and formation of defective (leaky) interfaces on each vesicle membrane11 and (b) vesicle fusion. Increasing extents of vesicle aggregation with lowering pH in vesicles composed of PA and PC lipids is not primarily attributed to the decrease of electrostatic repulsion among negatively charged vesicles, since (almost neutral) vesicles composed of lipids with only phosphatidylcholine headgroups did not exhibit significant fusion at the same experimental conditions. The zeta potential of vesicles composed of PA and PC lipids was, as expected, independent of the lipid acyl-tail composition and ranged from -31 ( 3 mV at pH 7.4 to -27 ( 1 mV at pH 4.0 (see Figure S5 in the Supporting Information). The zeta potential of vesicles composed of PC lipids was approximately -3 mV (see Figure S6 in the Supporting Information). We suggest, in agreement with previous hypotheses,9 that the observed aggregation is driven mostly by the formation of “defects” on the outer surface of vesicles originating at the discontinuous interfaces/phase boundaries between the pH-triggered phase-separated lipid domains. Figure 6A supports this suggestion by showing the same strong linear correlation for all three membrane compositions (independent of the composition of the hydrophobic region of the membranes) between the aggregates’ size and the “total length” of transient interfaces on the vesicles’ surface. Only one measured value deviates from this trend that corresponds to the aggregates’ size of vesicles composed of DPPC/DPPA at pH 5.0 at the high lipid concentration (open triangles). Data are replotted from Figure 2 and Figure S2 (Supporting Information). For the calculation of the “total length” of the transiently defective interfaces, to a first approximation as before, formation of small, circular PA-rich domains of a similar size is assumed with total area proportional to the number of protonated PA lipids at each pH (7.4, 5.0, and 4.0).11 Therefore, an estimation of the “total length” of defective interfaces would be proportional to the square root of this quantity that is identical across all three vesicle compositions shown in Figure 6A. For a general mass action reaction of irreversible aggregation between essentially identical primary particles, an increase in vesicle concentration decreases the distances between vesicles and, therefore, is expected to increase the initial aggregation rates as shown in Figure 6A (closed symbols for 50 μM lipid concentration vs open symbols for 100 μM).4 In particular, for the simplified case where the average aggregate size Æl(t)æ is proportional to the dimensionless time parameter τ = C11X0t (the initial aggregation reaction is characterized by a dimerization reaction with C11 being the forward dimerization rate constant and X0 the initial vesicle concentration),4 such that Æl(t)æ = 1 þ 2τ, then the ratios of the two linear fits (the two linear functions) in Figure 6A for the two vesicle compositions (50 and 100 μM total lipid) should be proportional to (1 þ 2C11X0t)/(1 þ 4C11X0t) = (1 þ A)/ (1 þ 2A). The calculated ratio of the two fitted linear functions shown in Figure 6A is proportional to (1 þ A)/(1 þ 1.9A), which is in close agreement with (1 þ A)/(1 þ 2A) and, therefore, with this simplified aggregation scheme. This correlation, which is uniform for all acyl-tail pairs studied, supports the hypothesis that aggregation is driven mostly by the “total length” of the transient DOI: 10.1021/la9026283
1671
Article
Bhagat and Sofou
Figure 6. (A) Size of vesicle aggregates at early time points is linearly proportional to the total length of transiently defective interfaces on the membranes and is independent of the hydrophobic component of the membranes. Data are replotted from Figure 2 and Figure S2 (Supporting Information), and symbols are as follows: circles represent membranes with nonmatching acyl tails (DPPC/DSPA), triangles represent membranes with matching acyl tails (DPPC/DPPA), and squares represent membranes with matching acyl tails (DSPC/DSPA, in the gel phase) at 50 μM (closed symbols) and 100 μM (open symbols) lipid concentration. Lines are least-squares linear fits for all data points at the same lipid concentration. (B) Rate of total lipid mixing at early time points depends on the total length and the degree of transiently defective interfaces. Data are calculated from Figure 4, and symbols are as follows: (b) membranes with nonmatching acyl tails (DPPC/DSPA), (1) membranes with matching acyl tails (DSPC/DSPA) in the gel phase, (O) membranes with matching acyl tails (DPPC/DPPA) with one type of lipid not being in the gel phase, at 50 μM total lipid concentration. Lines are least-squares linear fits. For the calculation of the total length of the transient defective interfaces, to a first approximation, formation of small, circular PA-rich domains of a similar size is assumed with total area proportional to the number of protonated PA lipids at each pH (7.4, 5.0, and 4.0). See text for details.
defective phase boundaries between phase-separated domains on the vesicles’ outer surface and that aggregation is independent of the composition of the hydrophobic portion of membranes. Contrary to aggregation, lipid mixing among apposing vesicles depends both on the “total length” and the “degree” of transient defective interfaces. Assuming a second-order reaction scheme to describe registration between the defects of two apposing membranes in order to result in lipid mixing, Figure 6B shows the initial rate of total lipid mixing, within the first 2 h of incubation, to be linearly proportional to the square of the “total length” of transient defects for each vesicle composition (initial rates are calculated from Figure 4). The slope of the linear fit for each vesicle composition, which is suggested to correspond to an apparent lipid mixing rate constant, is greater for membranes containing lipid pairs with mismatched acyl-tail lengths (DPPC/ DSPA, closed circles), followed by lipid pairs with matching acyltail lengths in the gel state (DSPC/DSPA, closed triangles) and by lipid pairs with matching acyl tails where one lipid is close to its transition temperature (DPPC/DPPA, open circles). The greater fusogenicity, i.e., the higher slope, observed for the membranes with mismatched acyl-tail lengths could be attributed to the greater “degree” of defects at the discontinuous transient phase boundaries of these heterogeneous membranes.9 We have previously correlated the “degree” of defects to the permeability of free-standing membranes independently of fusion, and we observed that the membrane permeability follows the same order with respect to lipid (mis)matching and to lipid relative fluidity as mentioned above regarding the membrane fusogenicity.11 A mechanistic explanation aiming to rationalize these observations correlates the “degree” of defects to the relative dynamics of the lipid acyl tails that correspondingly affect the molecular packing at the boundaries of phase-separated lipid domains:11 gel phase acyl tails would result in transient discontinuities that may persist longer in time. For the vesicle composition with DPPC/ DPPA matching tails, where the DPPC lipid is not entirely in 1672 DOI: 10.1021/la9026283
the gel state at 37 C and the interfacial defects are not expected to be significant, the measured slope is too small to be accurately described by this experimental technique (Figure 6B, open circles). In addition to the simplistic phenomenological suggestion of increasing formation of defective lipid phase boundaries with decreasing pH, merging (ripening) of phase-separated lipid domains over time is possible to occur, therefore decreasing the “total length” of defects over time. However, experimental observations indicate that the phenomenon of domain ripening is either too fast or too slow compared to the time frame of our experimental measurements and/or the sensitivity of the measuring techniques. Otherwise, the content release rates11 obtained over the entire observation time of 4 days would not have been linearly proportional to the approximate “total length” of phase boundaries that is not calculated to change over time. All measurements indicate a unimodal mechanism over time. It may also be possible that, at the experimental conditions used, these phase-separated heterogeneities are kinetically trapped at a size smaller than their equilibrium size.22 The extent of vesicle fusion (complete vs hemifusion) was studied using an inner lipid-leaflet mixing assay (see Supporting Information for experimental details). Vesicles composed of lipid pairs with mismatched acyl-tail lengths (DPPC/DSPA) that display the greater extents of total lipid mixing are the only membranes that exhibit after 24 h of incubation significant inner lipid-membrane mixing at pH 5.0 and 4.0. These are the only pH values at which the corresponding total lipid mixing assay indicated more than 50% extent of mixing (Figure 4). In particular, the following pH dependence on the extent of inner leaflet-lipid mixing was observed: 5.5%, 8.0%, 14.8%, and 28.0% at the pH values of 7.4, 5.5, 5.0, and 4.0 at 24 h, respectively. At the two highest pH values of 7.4 and 5.5 after (22) Veatch, S. L.; Keller, S. L. Biophys. J. 2003, 85, 3074–3083.
Langmuir 2010, 26(3), 1666–1673
Bhagat and Sofou
24 h of incubation, the content-mixing assay also indicated mixing of the encapsulated aqueous contents of fusing vesicles confirming the inner lipid mixing results and excluding a hemifusion mechanism. Unfortunately, the substantial content release from vesicles at the two lowest pH values of 5.0 and 4.0 did not allow use of the content-mixing assay and its comparison to the inner lipid mixing results. The present studies on heterogeneous lipid membranes suggest that vesicle aggregation and fusion may possibly act toward minimizing the extent of transient (in time) defective phase boundaries that are triggered to form by lowering the pH. We propose that the “total length” of defective phase boundaries is a property of the vesicle exterior. These outer defective interfaces are acting, therefore, as “sticking points” via which vesicles aggregate upon contact. The total lipid mixing findings suggest that the activation energy of the irreversible process of lipid mixing23,24 decreases with enhancement of the “degree” of lipid packing defects at these phase boundaries. A greater “degree” of lipid packing defects within the membrane’s hydrophobic region may potentially create energetically unfavorable molecular configurations due to the longer transient exposure of hydrophobic regions of the membrane to water molecules that is compensated, therefore, by greater vesicle fusogenicity.9 These transiently defective phase boundaries should also be unfavorable to hydration which is a major opposing force to lipid membrane apposition and fusion.25 It is possible that upon vesicle apposition registration occurs of the defective phase boundaries from different vesicles, resulting in merging of defects from different vesicles via lipid exchange and mixing. This suggested phenomenological model could provide a plausible mechanistic explanation for the observed fusogenicity of these pH-dependent heterogeneous lipid membranes. In our studies, addition of cholesterol conserves the property of the initial vesicle aggregation rates to be mostly dependent on the “total length” of transient defective phase boundaries that are suggested to be located on the exterior of vesicles and to be independent from the hydrophobic component of the membranes. Interestingly, however, addition of cholesterol inverts the tendency for fusion as previously described in terms of the “degree” of these defective phase boundaries. This observation could possibly be related to the variable preference of cholesterol to partition within the defective phase boundaries altering, therefore, their tendency for lipid exchange and mixing. A potentially preferential partition of cholesterol in PC-rich, PA-rich lipid domains or in the defective phase boundaries that would induce changes in the molecular packing of lipids affecting, therefore, the membrane fusogenicity cannot be determined by our measurements. Cholesterol’s lipid miscibility has been extensively studied and been reported to depend on its interactions both with the acyl chains of neighboring lipids and with their headgroups.26-31 This (23) Hui, S.; Stewart, T.; Boni, L.; Yeagle, P. Science 1981, 212, 921–923. (24) Cevc, G.; Richardsen, H. Adv. Drug Delivery Rev. 1999, 38, 207–232. (25) Zimmerberg, J.; Vogel, S.; Chernomordik, L. Annu. Rev. Biophys. Biomol. Struct. 1993, 22, 433–466. (26) Van Dijck, P. W. M. Biochim. Biophys. Acta 1979, 555, 89–101. (27) Almeida, P. F.; Vaz, W. L.; Thompson, T. E. Biophys. J. 1993, 64, 399–412. (28) Silvius, J. R.; del Giudice, D.; Lafleur, M. Biochemistry 1996, 35, 15198– 15208. (29) Korlach, J.; Schwille, P.; Webb, W. W.; Feigenson, G. W. Proc. Natl. Acad. Sci. U.S.A. 1999, 96, 8461–8466. (30) McMullen, T. P. W.; Lewis, R. N. A. H.; McElhaney, R. N. Biophys. J. 2000, 79, 2056–2065.
Langmuir 2010, 26(3), 1666–1673
Article
complex partition of cholesterol could explain the interesting inversion in fusogenicity observed in these studies. Elucidation of cholesterol’s preferential partition in the membranes studied here is expected to be further convoluted given the locally heterogeneous lipid composition of membranes and the presence of phase boundaries that are all expected to affect differently cholesterol’s partition functional. Systematic studies (probably of calorimetric nature) toward answering these questions would be of great interest and significance. The property of pH-dependent fusogenicity in lipid vesicles can be potentially applied in liposomes composed of these heterogeneous membranes and used as drug delivery carriers with the aim to increase the bioexposure of malignant cells to delivered therapeutics.12,13 Following appropriate labeling of these carriers to induce their endosomal uptake,32 the acidification of the endosomal pathway can be exploited to trigger fusion of vesicles with the endosomal membranes with the ultimate goal to release the delivered therapeutic contents directly into the cytoplasm. Preliminary fusion studies between heterogeneous vesicles composed of DPPC/DSPA lipids and vesicles composed of lipid mixtures characteristic of the endosomal membrane (phosphatidylcholine, phosphatidylethanolamine, sphingomyelin, cholesterol)33 show that lowering pH results in increasing extents of total lipid mixing among the different vesicle populations (see Figures S7A and S7B in the Supporting Information). This technology could potentially result in therapeutic outcomes while simultaneously lowering the required administered doses and, therefore, resulting in relief of related toxicities to normal organs. These vesicles were chosen to contain lipids in the gel phase at the working temperature of 37 C to conserve long circulation times in the blood in order to increase the probability of greater accumulation at the malignant sites (enhanced permeability and retention (EPR) effect).11,34,35 This study is a proof of principle, since successful performance of these vesicles in real applications would require more complex lipid mixtures and possibly additional physicochemical adjustments. Further engineering and optimization of these heterogeneous membranes is required to potentially lead to useful technologies for the advancement of human health. Acknowledgment. This study was supported by the Wallace H. Coulter Foundation (Early Career Award in Translational Research), the New York State Office of Science, Technology and Academic Research (J.D. Watson Investigator Award), the Susan G. Komen Foundation (Career Catalyst Award), and the Polytechnic Institute of New York University. Supporting Information Available: Additional experimental details, DSC scans, vesicle aggregation, lipid mixing studies, and detailed release profiles. This material is available free of charge via the Internet at http://pubs.acs.org. (31) Haran, G.; Cohen, R.; Bar, L. K.; Barenholz, Y. Biochim. Biophys. Acta 1993, 1151, 201–215. (32) Sofou, S.; Sgouros, G. Expert Opin. Drug Delivery 2008, 5, 189–204. (33) Bergstrand, N.; Arfvidsson, M. C.; Kim, J. M.; Thompson, D. H.; Edwards, K. Biophys. Chem. 2003, 104, 361–379. (34) Maeda, H.; Wu, J.; Sawa, T.; Matsumura, Y.; Hori, K. J. Controlled Release 2000, 65, 271–284. (35) Noguchi, Y.; Wu, J.; Duncan, R.; Strohalm, J.; Ulbrich, K.; Akaike, T.; Maeda, H. Jpn. J. Cancer Res. 1998, 89, 307–314.
DOI: 10.1021/la9026283
1673