Effect of Cyclodextrin and Membrane Lipid ... - ACS Publications

Oct 31, 2013 - Guangtao Li, JiHyun Kim, Zhen Huang, Johnna R. St. Clair, Deborah A. Brown, Erwin London. Efficient replacement of plasma membrane oute...
0 downloads 0 Views 1MB Size
Article pubs.acs.org/Langmuir

Effect of Cyclodextrin and Membrane Lipid Structure upon Cyclodextrin−Lipid Interaction Zhen Huang and Erwin London* Department of Biochemistry and Cell Biology, Stony Brook University, Stony Brook, New York 11794-5215, United States S Supporting Information *

ABSTRACT: Methyl-β-cyclodextrin (MβCD) can be used to exchange membrane lipids between different vesicles in order to prepare model membrane vesicles with lipid asymmetry. To help define what factors influence lipid exchange, we studied how lipid interaction with cyclodextrins (CDs) was affected by lipid and CD structure. The decrease in light scattering upon CD-induced vesicle solubilization and the change in Förster resonance energy transfer of labeled lipids upon vesicle solubilization and lipid exchange were used to detect phospholipid−CD interaction. Of the CDs examined, MβCD, hydroxypropyl-α-cyclodextrin (HPαCD), and hydroxypropyl-β-cyclodextrin (HPβCD) were the three with the most suitable phospholipid interaction properties. Only MβCD was observed to dissolve lipid vesicles (at least at CD concentrations below 125 mM). Solubilization of lipid vesicles was half complete at 10−80 mM MβCD with progressively higher MβCD concentrations required as phospholipid acyl chain length increased from 14 to 22 carbons. Phospholipid acyl chain unsaturation and lipid headgroup structure also affected the amount of MβCD needed for solubilization. All three CDs studied were able to carry out phospholipid exchange. MβCD, which retained the ability to carry out lipid exchange below MβCD concentrations needed for solubilization, exchanged lipid more efficiently than HPαCD or HPβCD. However, the ability of HPαCD to exchange phospholipids, coupled with its inability to interact with cholesterol, indicates that it will be useful for preparing asymmetric vesicles with controlled amounts of cholesterol.



was found to be in the order βCD ≫ γCD > αCD.8 It has also been reported that HPβCD and MβCD are best at forming complexes with cholesterol, while βCD, γCD, and αCD do not interact well with cholesterol, in the latter case presumably because its cavity is too small.4 In agreement with a poor interaction of αCD with cholesterol, in studies of cholesterol exchange between vesicles, βCD, γCD, and MβCD were found to be all effective, while αCD was not.14 CDs also act interact with phospholipids, but the specificity of CDs for phospholipids differs from that for cholesterol. For example, in a study of ring size effects, CDs partially removed phospholipids from human erythrocytes in the order αCD > βCD ≫ γCD. This differs from the order for cholesterol (see above) where αCD is not effective.8 The difference between αCD and βCD interactions presumably reflects the inability of cholesterol to fit into the cavity of αCD, in contrast to a single acyl chain, which should fit well into the cavity of both αCD and βCD. In agreement with this difference in αCD specificity for phospholipid relative to cholesterol, in an in vitro model of the blood brain barrier, it was shown that αCD could remove phospholipids while βCD and γCD removed both cholesterol

INTRODUCTION Cyclodextrins (CDs) are a family of compounds made up of glucose molecules bound by glycosidic bonds into macrocycles.1 They form a somewhat conical cylinder, with a relatively nonpolar cavity.2,3 This allows the formation of inclusion complexes of CDs with a wide variety of hydrophobic guest molecules, including membrane lipids such as cholesterol and phospholipids.4,5 CDs have found wide application in modifying the lipid composition of cells and model membranes. Most studies have involved the interaction of CDs with cholesterol. The ability of CDs to remove cholesterol from cells, or (when preloaded with cholesterol) deliver cholesterol to cells, has been used to study the cellular function of cholesterol in many studies.6−10 CDs have also been used to study the strength of cholesterol association with various lipids.11−13 One method to do this involves CD-catalyzed sterol exchange between two populations of vesicles with different lipid compositions.14 Other studies have shown that both CD ring size and derivatization affect CD interaction with cholesterol. In terms of the effect of derivatization, one study showed that βCD, HPβCD, and MβCD all were able to extract significant amounts of cholesterol from isolated stratum corneum, but the ability of βCD to extract cholesterol was limited.6 In terms of the effect of ring size, the extent of cholesterol removal from erythrocytes © 2013 American Chemical Society

Received: August 20, 2013 Revised: September 27, 2013 Published: October 31, 2013 14631

dx.doi.org/10.1021/la4031427 | Langmuir 2013, 29, 14631−14638

Langmuir

Article

and phospholipids.15 Similar behavior was found for hydroxypropyl CDs. Irie and co-workers showed that HPαCD preferentially solubilized phospholipids into solution relative to cholesterol, HPβCD solubilized cholesterol better than phospholipids, and HPγCD was nonspecific in its solubilization properties.16 Leventis and Silvius demonstrated that βCD and γCD accelerated the rate of cholesterol exchange between vesicles by a larger factor than the transfer of phospholipids, whereas αCD and MβCD accelerated the rate of phospholipid exchange by a larger factor than the transfer of cholesterol.14 The effect of CD structure upon CD−phospholipid interactions has also been examined to some degree in prior studies. Both αCD and MβCD were found to be slightly more effective in exchanging dipalmitoylphosphatidylcholine (DPPC) than βCD or γCD.14 The ability to extract phospholipids and protein was found to be in the order HPαCD < αCD < DMαCD (dimethyl-α-cyclodextrin) for red blood cells.17 Carboxylethylated-γ-cyclodextrin (CEγCD) was found to be effective in enhancing the transfer of fluorescent phospholipids from vesicles to cells.18 Using differential scanning calorimetry, lipid vesicles made of DPPC were chosen as a model to test interaction with CDs. It was found that αCD forms an insoluble complex with DPPC, while MβCD has the highest ability to form soluble complexes with DPPC.5 In the case of MβCD, it has been shown that there is a strong enough interaction such that lipid vesicles can be dissolved, being replaced by soluble CD−lipid complexes.19 Other studies have examined the effect of lipid structure upon interaction with CD. In one study, the ability of αCD to extract phospholipids from red blood cells was found to follow the order PS ≥ PE ≫ SM > PC.20 (αCD has also been observed to interact with PI.21) In contrast, no specificity of αCD (or γCD) for SM over PC was observed in brain capillary endothelial cells, although preferential extraction of SM over PC was observed for βCD.15 A study using unilamellar vesicles made of phospholipids with varying acyl chain lengths (dimyristoylphosphatidylcholine, DMPC, DPPC, and distearoylphosphatidylcholine, DSPC) monitored carboxyfluorescein (CF) leakage to examine interaction with CDs. The results showed the effect upon CF leakage was in the order DMβCD > αCD > TMβCD for DPPC liposomes, in the order DMβCD > TMβCD > αCD in DSPC liposomes, and in the order αCD > DMβCD > TMβCD in DMPC liposomes. It was also found that βCD, HPβCD, and γCD scarcely cause leakage in these vesicles.22 These differences in leakage may reflect both differences in the effect of lipid structure upon the extent of lipid extraction and upon sensitivity of the vesicles chosen to leakiness. Our group has developed CD-catalyzed lipid exchange between different model membrane vesicles to build model membrane vesicles in which, as in many natural biological membranes, the lipid compositions of the inner and outer leaflet are different, i.e., in which vesicles have lipid asymmetry. MβCD-catalyzed exchange using high concentrations of MβCD was employed to prepare small unilamellar vesicles (SUVs),23 large unilamellar vesicles (LUVs),24 and giant unilamellar vesicles (GUVs)25 with lipid asymmetry. However, whether other CDs could interact with a wide range of lipids and catalyze the lipid exchange needed to prepare asymmetric vesicles has not been investigated. Therefore, we have carried out studies of lipid−CD interaction and lipid vesicle solubilization as a function of lipid and CD structure. The

results provide important clues for optimization of lipid interaction and lipid exchange.



MATERIALS AND METHODS

Materials. 1,2-Dimyristoleoylphosphatidylcholine (DMoPC or di C14:1 PC); 1,2-dipalmitoleoylphosphatidylcholine (DPoPC or di C16:1 PC); 1,2-dioleoylphosphatidylcholine (DOPC or di C18:1 PC); 1,2-dierucoylphosphatidylcholine (DEuPC or di C22:1 PC); 1palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC or C16:0C18:1 PC); 1,2-dipalmitoyl-sn-glycero-3-phosphocholine (DPPC or di C16:0 PC); 1,2-diphytanoyl-sn-glycero-3-phosphocholine (DiphyPC); 1,2-dioleoyl-sn-glycero-3-phospho-(1′-rac-glycerol) (DOPG or di C18:1 PG); 1-palmitoyl-2-oleoyl-sn-glycero-3-phospho-(1′-racglycerol) (POPG or C16:0-C18:1 PG); 1,2-dioleoyl-sn-glycero-3phospho-L-serine (DOPS or di C18:1 PS); 1-palmitoyl-2-oleoyl-snglycero-3-phospho-L-serine (POPS or C16:0-C18:1 PS); 1,2-dioleoylsn-glycero-3-phosphoethanolamine-N-(7-nitro-2-1,3-benzoxadiazol-4yl) (NBD-DOPE); 1,2-dipalmitoyl-sn-glycero-3-phosphoethanolamine-N-(7-nitro-2-1,3-benzoxadiazol-4-yl) (NBD-DPPE); 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine-N-(lissamine rhodamine B sulfonyl) (rhod-DOPE); and 1,2-dipalmitoyl-sn-glycero-3-phosphoethanolamine-N-(lissamine rhodamine B sulfonyl) (rhod-DPPE) were purchased from Avanti Polar Lipids (Alabaster, AL). Lipids were dissolved in chloroform and stored at −20 °C. The concentrations of unlabeled lipids were determined by dry weight and that of fluorescent lipids by absorbance using εNBD‑DOPE = εNBD‑DPPE = 21 000 M−1 cm−1 at 460 nm in methanol and εrhod‑DOPE = εrhod‑DPPE = 95 000 M−1 cm−1 at 560 nm in methanol. Methyl-β-cyclodextrin, 1.6−2 methyls per glucose (MβCD), hydroxypropyl-α-cyclodextrin (HPαCD) average MW 1180, and sulfo-α-CD were from Sigma-Aldrich (St. Louis, MO). Other CDs, α-cyclodextrin (αCD), γ-cyclodextrin (γCD), and hydroxypropyl-β-cyclodextrin (HPβCD), were purchased from Tokyo Chemical Industry (Portland, OR). All CDs were dissolved in distilled water and (except for αCD which was prepared freshly) and were stored at −20 °C. Vesicle Preparation. Multilamellar vesicles (MLVs) were prepared in glass tubes similarly to as described previously.23 Vesicles containing the desired lipid mixtures were dispersed at 70 °C in phosphate-buffered saline (1X PBS = 10 mM sodium phosphate, 150 mM sodium chloride, pH = 7.8 ± 0.2) at the desired lipid concentration and vortexed in a multitube vortexer (VWR International) at 55 °C for 15 min. They were then cooled down to room temperature and briefly revortexed prior to preparing individual samples. Ethanol dilution small unilamellar vesicles (SUVs) were prepared similarly to as described previously.23 The desired lipids were dried under nitrogen followed by high vacuum for at least 1 h, dissolved in ethanol, and then dispersed by 50-fold dilution at 70 °C in PBS. Measurement of Vesicle Solubilization by Light Scattering. Light scattering (optical density) measurements were made in a Beckman 640 spectrophotometer (Beckman Instruments, Fullerton, CA) using quartz cuvettes at a wavelength of 300 nm. A 4.25 mL MLV sample, prepared as described above, was divided into eight 500 μL aliquots. This was used to prepare a series of samples with different CD concentrations. Because different CDs stock solutions were used, to do this first an amount of distilled water was added so that every sample for a particular CD titration with a specific lipid would have the same final PBS concentration, lipid concentration, and volume, and then the CD was added. After a 30 min incubation at room temperature light scattering measurements were made. PBS with the same PBS concentration as in the lipid-containing samples was used as the reference sample. (MβCD, HPαCD, and HPβCD did not scatter light when dispersed in PBS, so to conserve CDs they were left out of the reference samples.) Final lipid and PBS concentrations were DMoPC (192 μM, 0.96X PBS), DPoPC (181 μM, 0.91X PBS), DOPC (175 μM, 0.88X PBS), DEuPC (125 μM, 0.63X PBS), POPC (175 μM, 0.88X PBS), DPPC (149 μM, 0.75X PBS), and Diphy PC (185 μM, 0.93X PBS). Even though the lipid and PBS concentrations are different from lipid to lipid, it did not affect the results significantly 14632

dx.doi.org/10.1021/la4031427 | Langmuir 2013, 29, 14631−14638

Langmuir

Article

Figure 1. Effect of lipid acyl chain structure upon the solubilization of MLV by MβCD assayed by light scattering. (A) Effect of lipid acyl chain length upon solubilization. DMoPC = C14:1 acyl chains, DPoPC = C16:1 acyl chains, DOPC = C18:1 acyl chains, and DEuPC = C22:1 acyl chains. (B) Effect of differences in acyl chain saturation, saturation (DOPC, DPPC, and POPC), and branching (Diphy PC) upon solubilization. (C, D) Effect of lipid headgroup upon solubilization: C, DO lipids; D, PO lipids. Samples composed of MLVs dispersed in PBS were mixed with various concentrations of MβCD. The final lipid and PBS concentration varied slightly (within 1.5-fold) for different lipids (see Materials and Methods.) This did not affect the results (see Supporting Information Figure 1). The y-axis value is the ratio of optical density at 300 nm in the presence of MβCD to that in the absence of MβCD. Separate samples were prepared at each MβCD concentration. The average (mean) of four samples and standard deviations are shown. (see Supporting Information Figure 1). For the comparison of different polar headgroups, the actual lipid and PBS concentrations were DOPG, DOPC, POPC (175 μM, 0.88X PBS); POPG (183 μM, 0.92X PBS); and DOPS, POPS (167 μM, 0.83X PBS). Fö rster Resonance Energy Transfer (FRET) Measurements of Lipid Exchange and Vesicle Solubilization. Fluorescence was measured in a SPEX FluoroLog 3 spectrofluorometer (Jobin-Yvon, Edison, NJ) using quartz semi-microcuvettes (excitation path length 10 mm; emission path length 4 mm) according to previously described protocols.23 For FRET in labeled vesicles, F samples had a mixture of unlabeled lipid, lipid labeled with NBD, and lipid labeled with rhodamine, while Fo samples had a mixture of unlabeled lipid and lipid labeled with NBD. Background for F samples contained unlabeled lipid with same amount of acceptor as in F samples. Background samples for Fo contained pure unlabeled lipid. Two donor−acceptor pairs were used: NBD-DOPE/rhod-DOPE and NBD-DPPE/rhodDPPE. The mol % of labeled lipids in the “F sample” labeled vesicles was either 0.48 mol % NBD-DOPE/1.0 mol % rhod-DOPE or 0.19 mol % NBD-DPPE/1.2 mol % rhod-DPPE. There were two different types of FRET experiments. In one, all of the vesicles contained labeled lipids. In the other, the samples and backgrounds contained mixtures of labeled vesicles and unlabeled vesicles. The inner filter effect arising from acceptor absorbance was found to be negligible, at most a few percent (not shown). For FRET experiments, a sample with a scaled-up volume (3.2−7.2 mL) was prepared as described above and then divided into eight aliquots. Next, distilled water was added to adjust the volume to ensure each sample would have the same final lipid and PBS concentration after CD was added. An aliquot from MβCD (390 mM) or HPαCD (384 mM) or HPβCD (287 mM), dissolved in water, was then added to the samples to prepare a series of samples with increasing CD concentrations. Unless otherwise noted, the final concentrations in the samples were 0.8X PBS, 80 μM lipid for FRET in labeled vesicles, or 0.8X PBS, 20 μM labeled lipids plus 140 μM unlabeled lipids for FRET

in mixtures of labeled and unlabeled vesicles. Final volumes were in the range 0.8−1 mL. After 30 min incubation at room temperature (unless otherwise noted), NBD fluorescence for F, Fo, and background samples were measured at an excitation wavelength of 460 nm and an emission wavelength of 534 nm.



RESULTS Vesicle Solubilization by MβCD Assayed by Light Scattering: Effect of Lipid Chemical Structure. High concentrations of MβCD have the ability to bind to lipids and dissolve lipid vesicles.19 To investigate MβCD−lipid interaction, the decrease light scattering (optical density) upon solubilization of lipid vesicles with different lipid compositions was measured. Figure 1 shows that when MβCD was added to MLV composed of various lipids, light scattering decreased as MβCD concentration increased above a threshold value, with scattering reaching a minimal value at high MβCD concentrations. The concentration of MβCD needed to solubilize MLV (measured by the concentration at which the decrease in optical density was half of the maximal decrease) was dependent upon lipid composition. In Figure 1A, the dependence of solubilization upon acyl chain length was studied using PCs with monounsaturated acyl chains of various lengths. There was a gradual change in MβCD concentration needed for solubilization as acyl chain length increased, with a 6-fold increase (from 10 to 60 mM MβCD) from di C14:1 PC to di C22:1 PC. In Figure 1B, the effect of acyl chain saturation and acyl chain branching upon solubilization was examined. The concentration of MβCD (45 mM) needed to dissolve vesicles composed of di C16:0 PC, which has two saturated acyl chains, was higher than that (30 mM) needed to dissolve di 14633

dx.doi.org/10.1021/la4031427 | Langmuir 2013, 29, 14631−14638

Langmuir

Article

Table 1. Residual Light Scattering at the High MβCD Concentrationsa lipids

DMoPC

DPoPC

DOPC

DEuPC

POPC

DPPC

Diphy PC

initial OD final OD ratio

0.38 ± 0.04 0.02 ± 0.00 0.04 ± 0.00

0.31 ± 0.01 0.02 ± 0.01 0.07 ± 0.02

0.25 ± 0.04 0.04 ± 0.00 0.14 ± 0.01

0.29 ± 0.02 0.12 ± 0.01 0.42 ± 0.07

0.27 ± 0.05 0.05 ± 0.01 0.18 ± 0.06

0.25 ± 0.01 0.07 ± 0.01 0.29 ± 0.02

0.39 ± 0.02 0.03 ± 0.01 0.07 ± 0.01

a

Ratio is the value of optical density (OD) at 300 nm at highest MβCD concentrations used in Figure 1 (“final OD”) divided by the OD of MLVs without MβCD (“initial OD”). Errors are standard deviation from four samples.

Figure 2. Vesicle solubilization by various CDs detected using FRET. Left: solubilization of NBD-DOPE and rhod-DOPE labeled SUVs. F samples composed of DOPC with 0.48 mol % NBD-DOPE and 1.0 mol % rhod-DOPE. Right: solubilization of NBD-DPPE and rhod-DPPE labeled SUVs. F samples composed of DOPC with 0.19 mol % NBD-DPPE and 1.2 mol % rhod-DPPE. Fo samples lacked rhodamine lipid. Background samples lacked NBD lipid. Samples contained labeled SUVs dispersed in PBS with various concentrations of CDs. The final concentrations in all samples are 0.8X PBS, 80 μM total lipid. F/Fo is the ratio of donor fluorescence in the presence of acceptor to that in its absence. Separate samples were prepared at each CD concentration. The average (mean) of four samples and standard deviation are shown.

C18:1 PC, which has two unsaturated chains, but the difference between vesicles composed of di C18:1 PC and C16:0-C18:1 PC, which has one saturated and one unsaturated chain, was very small. Interestingly, a relatively low MβCD concentration (15 mM) was needed to dissolve vesicles composed of diphytanoyl PC, which has four methyl branches per acyl chain. Figures 1C and 1D compare solubilization by MβCD for lipids with different headgroups. Figure 1C shows di C18:1 PG dissolved at lower MβCD concentrations (18 mM) than did di C18:1 PC (28 mM) or di C18:1 PS (25 mM). As in the case of PC, PG and PS with di C18:1 chains or C16:0-C18:1 chains solubilized at similar concentrations of MβCD. It is noteworthy that there was a residual light scattering in the lipid-containing samples at the highest MβCD concentrations used. Residual light scattering was especially high for di C22:1 PC (Figure 1 and Table 1). This may be incomplete solublization or (more likely) reflects formation of lipid−CD complexes or aggregates large enough to scatter a significant amount of light (see below and Discussion). This residual light scattering was not observed in MβCD in the absence of lipids. We also examined lipid solubilization by HPαCD and HPβCD using light scattering. They did not dissolve lipid vesicles over a very wide range of CD concentrations up to 125 mM (see below). Other CDs were found to have other unfavorable properties (see below) and were not examined for their ability to solubilize lipids. Comparison of Solubilization Catalyzed by MβCD, HPαCD, and HPβCD Using FRET. To investigate CD−lipid interactions in more detail, we used FRET. In FRET the excited state energy of a donor is transferred to a nearby acceptor (within 10−100 Å depending on donor and acceptor), resulting in quenching of donor fluorescence. To examine CD−lipid interaction using FRET, vesicles containing a small amount of donor (NBD-labeled lipid) and acceptor (rhodamine-labeled lipid) were prepared. FRET should be abolished, and thus donor fluorescence enhanced, when vesicles dissolve and donor

and acceptor labeled lipids reside in separate CD complexes (see schematic illustration in Supporting Information Figure 2). Figure 2 shows the effect of CDs upon donor fluorescence (F/Fo) in FRET-probe labeled vesicles. F/Fo is the fraction of donor fluorescence not quenched by FRET (i.e., when F/Fo = 0 there is 100% FRET while when F/Fo = 1 there is 0% FRET). In agreement with light scattering results, in labeled DOPC vesicles high concentrations of MβCD were able to largely abolish FRET, indicating vesicle solubilization, while HPαCD and HPβCD had no effect on FRET, indicating no solubilization. FRET was very similar when the donor and acceptor labeled lipids had saturated (dipalmitoyl) or unsaturated (dioleoyl) acyl chains, indicating that the type of labeled lipids did not influence solubilization. Figure 3 shows that FRET-detected solubilization by MβCD also had a similar dependence on acyl chain length as did solubilization as judged by light scattering, with di C14:1 PC dissolving at the lowest MβCD concentration, while di C22:1 PC dissolved at the highest MβCD concentrations. This confirms that FRET is responding to solubilization of the unlabeled lipids. The concentration of MβCD necessary induce half-maximal loss of FRET was similar to but slightly higher than (by about 5−10 mM) that needed to induce a half-maximal decrease in light scattering. Supporting Information Figure 3 shows that this was partly due to a difference between the amount of MβCD needed to dissolve MLV (used in the light scattering experiments) and SUV (used in most FRET experiments). Additional factors that give rise to the difference between the light scattering and FRET assays are described in the Discussion. Interestingly, there was significant residual FRET at high MβCD concentrations, consistent with the residual light scattering seen at high MβCD concentration after maximal solubilization. This residual FRET was somewhat variable from experiment to experiment. As in the case of residual light scattering, we attribute this residual FRET to aggregation of 14634

dx.doi.org/10.1021/la4031427 | Langmuir 2013, 29, 14631−14638

Langmuir

Article

which there is no solubilization, there can be lipid exchange such that the labeled lipids in the outer leaflet of the labeled vesicles become diluted in the outer leaflets of the unlabeled vesicles, so that the average distance between outer leaflet donor and acceptor increases, which decreases FRET. Since CDs do not have access to the inner leaflet, donor and acceptor labeled inner leaflet lipids will not exchange, and there will be strong residual FRET between inner leaflet lipids. This residual FRET should be abolished if enough CD is added to solubilize the vesicles. Figure 5 shows effect of CDs upon FRET in mixtures of labeled and unlabeled DOPC vesicles using the exchangesensitive FRET assay. At MβCD concentrations (as low as 10 mM or less) there is a partial loss of FRET due to MβCD catalyzed lipid exchange between labeled and unlabeled vesicles. There is a slight increase in FRET near 50 mM MβCD. This

Figure 3. Effect of acyl chain length upon solubilization assayed by FRET. Solubilization of NBD-DPPE and rhod-DPPE labeled MLVs composed of PC with different acyl chain lengths. F samples composed of PC with 0.19 mol % NBD-DPPE and 1.2 mol % rhodDPPE. Fo samples lacked rhodamine lipid. Background samples lacked NBD lipid. Samples composed of labeled SUVs dispersed in PBS with various concentrations of CDs. Final concentrations of lipids and PBS were the same as in the light scattering experiments. Separate samples were prepared at each CD concentration. The average (mean) of four (DMoPC and DOPC) or six (DEuPC) samples and standard deviation are shown.

CD−lipid complexes, which would bring some donor and acceptor lipids into close proximity (see Discussion). FRET, which can be carried out at lipid concentrations too low for accurate light scattering by optical density measurements, was also used to determine if lipid concentration affects solubilization. As shown in Figure 4, solubilization of DOPC

Figure 4. Effect of lipid concentration in FRET solubilization assay. DOPC SUVs labeled with 0.48 mol % NBD-DOPE as donor and 1.0 mol % rhod-DOPE as acceptor were used. The final lipid concentration in the samples was 20, 80, or 160 μM. Other conditions as in Figure 2 (left). Single samples were used at 20 μM. For 80 μM average of four samples and standard deviations are shown. For 160 μM the average and range of duplicates are shown.

vesicles was somewhat dependent upon lipid concentration, with the concentration of MβCD needed for solubilization doubling from 25 to 50 mM when lipid concentration increased from 20 to 160 μM. Comparison of Lipid Exchange Catalyzed by MβCD, HPαCD, and HPβCD Using FRET. In cases in which CDs cannot dissolve vesicles, they may still bind enough lipid to exchange lipids between vesicles.14 A FRET assay using a mixture of labeled and unlabeled vesicles was devised to detect lipid exchange. As shown schematically in Supporting Information Figure 4, when vesicles containing donor and acceptor labeled lipids are incubated with an excess of vesicles containing only unlabeled lipid and CD under conditions in

Figure 5. Effect of various CDs upon lipid exchange detected using FRET with a mixture of labeled and unlabeled vesicles: top, MβCD; middle, HPαCD; bottom, HPβCD. F samples composed of 140 μM DOPC SUVs and 20 μM DOPC SUVs labeled with (circle) 0.48 mol % NBD-DOPE and 1.0 mol % rhod-DOPE, (triangle) 0.19 mol % NBD-DPPE and 1.2 mol % rhod-DPPE. Fo samples lacked rhodamine lipid. Background samples lacked NBD lipid. Samples contained labeled and unlabeled SUVs dispersed in PBS with various concentrations of CDs. The final concentrations in all samples was 0.8X PBS. Separate samples were prepared at each CD concentration. The average (mean) of four samples and standard deviation are shown. 14635

dx.doi.org/10.1021/la4031427 | Langmuir 2013, 29, 14631−14638

Langmuir

Article

Figure 6. Effect of temperature upon lipid exchange. Sample conditions as in Figure 5 top except after 30 min incubation and measurement at 23 °C or 55 °C: left, MβCD; right, HPαCD.

energy when acyl chains are in contact with water. If two CD molecules interact with each fully extended acyl chain in CD− lipid complexes,19 they would cover a length of ∼16 Å, which is about the length of a fully extended 14 carbon acyl chain. Longer acyl chains may protrude into aqueous solution when bound to CD. This is energetically unfavorable and thus may make it more difficult to extract longer acyl chain lipids from the lipid bilayer. It was also observed that DPPC vesicles, which have saturated acyl chains, required a relatively high concentration of MβCD for solubilization relative to lipids with unsaturated acyl chains. Although DPPC lacks double bonds, this by itself does not explain its behavior, as diphytanoyl PC, which also lacks double bonds, dissolved at relatively low MβCD concentrations. Instead, resistance of DPPC to solubilization may be due to the fact that unlike diphytanoyl PC and lipids with unsaturated acyl chains, DPPC is in the tightly packed, and thus highly stable, gel state at room temperature. Lipid headgroup also affected the amount of MβCD solubilization to some degree. The effect of headgroups may be limited because they only interact weakly, if at all, with the hydrophobic cavity of MβCD. Finally, decreased vesicle size, i.e., increased vesicle curvature, slightly inhibited solubilization. Similar results were found by Hatzi and coworkers, who suggested that the curvature of liposomes affects the initial contact of CDs with the lipid membrane in a fashion that is important for CD−lipid interaction.26 The concentration of MβCD needed for solubilization of MLV was slightly lower in the light scattering assay than that observed with the FRET assay. One possibility is that the labeled lipids used in FRET do not dissolve up as easily as the unlabeled lipids in which they are embedded. That would increase the MβCD concentration needed to abolish FRET. Another possibility may arise from the fact that light scattering intensity can decrease due to a reduction in vesicle size as well as because of solubilization. If MβCD addition decreases vesicle size as well as solubilizes vesicles, undissolved vesicles present after partial vesicle solubilization will scatter less light per lipid molecule than those present before MβCD addition. This would lead to an overestimate of the extent of solubilization. Interpretation of light scattering was also complicated by the residual light scattering observed after the extent of solubilization plateaued at the highest MβCD concentrations. We speculate that there is formation of clusters of CD−lipid complexes which are large enough to scatter light. This would also explain the residual FRET seen at high MβCD concentration. This possibility is supported by the observation that αCD forms highly insoluble complexes with lipids in aqueous solution (data not shown). (The alternative that there

may be due to vesicle clustering. Removal of sufficient outer leaflet lipid could cause vesicles to stick to one another. This would give rise to increased FRET due to intervesicular energy transfer, as donors in one vesicle come into proximity to acceptors in another. It should be noted that exchange of outer leaflets by MβCD results all vesicles being labeled in their outer leaflet, so vesicle clustering would all involve interactions between labeled vesicles. There is a further decrease in FRET above about 50 mM, which corresponds to vesicle solubilization (see Figure 1 and 160 μM lipid curve in Figure 4). The exchange-sensitive FRET assay was also carried out in the presence of HPαCD and HPβCD (Figure 5, middle and bottom). Both of these CDs exhibited an ability to catalyze lipid exchange as shown by a decrease in FRET, although the extent of exchange was weaker than that observed for MβCD, requiring a much higher concentration of hydroxypropyl CDs to obtain significant levels of lipid exchange. Results were very similar when the donor and acceptor labeled lipids had saturated (dipalmitoyl) acyl chains or unsaturated (dioleoyl) acyl chains, indicating that the structure of the labeled lipids did not influence lipid exchange. We previously carried out lipid exchange to prepare asymmetric vesicles at elevated temperatures.23 To determine if the use of elevated temperatures aids lipid exchange, results using exchange-sensitive FRET assay were compared at 23 and 55 °C. Figure 6 shows that the extent of exchange using HPαCD, and of exchange and/or solubilization using MβCD, increases at 55 °C. The interactions of other water-soluble CDs with lipids were also tested using the exchange-sensitive FRET assay. There was no effect of sulfo αCD upon FRET, while a loss of FRET was observed with αCD, γCD, and CMαCD (not shown). However, for the last three CDs there was a significant increase in light scattering at higher CD concentrations (at >20 mM for the αCD and at >80 mM for γCD). This suggests that these CDs might induce vesicle aggregation and thus might not be suitable for preparing asymmetric vesicles using lipid exchange.



DISCUSSION Effect of Lipid Structure upon Interaction with MβCD. In this work, the interaction of various phospholipids with CDs differing in macrocycle size or substituent was studied. MβCD, HPαCD, and HPβCD were chosen for detailed studies because they had the most favorable solubility and lipid interaction properties. The concentration of MβCD required to interact with phospholipids, as judged by vesicle solubilization, increased with acyl chain length. This may be due the unfavorable free 14636

dx.doi.org/10.1021/la4031427 | Langmuir 2013, 29, 14631−14638

Langmuir

Article

enough to interfere with lipid exchange. In recent studies in which we prepared asymmetric vesicles via lipid exchange, varying both lipid acyl chain27 and polar headgroup,28 efficient lipid exchange was always observed. In contrast, lipid specificity will be particularly useful for applications of HPαCD-catalyzed lipid exchange. The reason for this is that HPαCD interacts poorly, if at all, with cholesterol.16 This has allowed us to use HPαCD to exchange phospholipids and prepare cholesterol-containing vesicles with an asymmetric phospholipid concentration starting from cholesterol-containing symmetric vesicles while circumventing complications arising from cholesterol exchange [Lin, Q.; London, E., data not shown].

are subpopulations of vesicles that cannot dissolve in MβCD seems very unlikely.) The observation that residual light scattering increased as acyl chain length increased suggests acyl chains protruding from CD−lipid complexes may promote interactions between CD−lipid complexes. Given the effect of lipid structure upon interaction with CDs, an interesting question is: why is there no effect of labeled lipid acyl chain structure the FRET assay? The key issue is that binding of CD to lipid is affected by both the structure of the molecules AND the environment (solvent) in which they are dispersed. In the case of lipid vesicles, we have the unusual situation that lipid is both solute and “solvent”. For this reason, labeled lipid extraction from vesicles depends on the structure of the unlabeled lipids from which the labeled lipid is extracted. The observation that labeled lipid acyl chain structure does not affect extraction indicates that the lipid environment can dominate lipid extraction by CDs. This conclusion is in agreement with the observation that the FRET and optical density data for different acyl chain length lipids are in good agreement, both showing a strong and similar dependence on unlabeled lipid acyl chain length. If the labeled lipid structure determined extraction, then the FRET curves would have been independent of unlabeled lipid acyl chain length. Effect of CD Structure upon Interaction with Lipid. CD−lipid interaction was also greatly affected by the nature of the substituent on the CD molecule, with the methyl substituent resulting in a much stronger interaction with lipid as judged by both the ability of MβCD to dissolve vesicles and its ability to catalyze efficient lipid exchange at lower concentrations than did HPβCD. Perhaps the methyl substituent increases the hydrophobicity of MβCD sufficiently to promote its interaction with lipid acyl chains. Interestingly, phospholipid exchange by HPαCD and HPβCD seemed to be similar, despite the difference in their ring sizes. We speculate that this is because single acyl chains fit in the cavity of both of αCD and βCD. Another interesting question is: why does MβCD dissolve vesicles only at high MβCD concentrations? It is possible binding of MβCD to lipid may be highly cooperative, so that its binding to lipids increases greatly at high MβCD concentrations.19 Implications for CD-Catalyzed Lipid Exchange. The differences in lipid interaction as a function of lipid and CD structure have implications for the preparation of asymmetric acceptor vesicles via lipid exchange between vesicles having two different lipid compositions. The relationship between lipid binding and the efficiency of lipid exchange is complex. Overall, the desired situation is one in which there is reversible binding of the lipids of interest to CD, so that they will be delivered to vesicles by CD as readily as they are extracted from vesicles by CD. In this case lipid exchange will equilibrate the lipid compositions of (excess) “donor” and “acceptor” vesicles. A lipid bound too weakly to CD would not be extracted from a vesicle, preventing its exchange. A lipid bound too strongly to CD would not be released from CD back into a vesicle, also preventing its exchange. Thus, an intermediate strength of CD−lipid interaction is most desirable. A related question is whether a difference in the relative binding of two different lipids to CD impacts lipid exchange. If one lipid binds to a CD much more strongly than the lipid with which it is being exchanged, there would be a tendency for CD to bind and release just the tighter binding lipid species, which would slow down the net rate of exchange. The differences in MβCD−lipid interaction observed in this report do not seem to be large



ASSOCIATED CONTENT

S Supporting Information *

Supplemental Figures 1−4. This material is available free of charge via the Internet at http://pubs.acs.org.



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected] (E.L.). Notes

The authors declare no competing financial interest.

■ ■

ACKNOWLEDGMENTS This work was supported by NSF grant DMR 1104367. ABBREVIATIONS PC, phosphatidylcholine; PS, phosphatidylserine; PE, phosphatidylethanolamine; PG, phosphatidylglycerol; PI, phosphatidylinositol; DMoPC (di C14:1 PC), 1,2-dimyristoleoylphosphatidylcholine; DPoPC (di C16:1 PC), 1,2-dipalmitoleoylphosphatidylcholine; DOPC (di C18:1 PC), 1,2dioleoylphosphatidylcholine; DEuPC (di C22:1 PC), 1,2dierucoylphosphatidylcholine; POPC (C16:0-C18:1 PC), 1palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine; DPPC (di C16:0 PC), 1,2-dipalmitoyl-sn-glycero-3-phosphocholine; DiphyPC, 1,2-diphytanoyl-sn-glycero-3-phosphocholine; DOPG (di C 18:1 PG), 1,2-dioleoyl-sn-glycero-3-phospho-(1′-racglycerol); POPG (C16:0-C18:1 PG), 1-palmitoyl-2-oleoyl-snglycero-3-phospho-(1′-rac-glycerol); DOPS (di C18:1 PS), 1,2dioleoyl-sn-glycero-3-phospho-L-serine; POPS (C16:0-C18:1 PS), 1-palmitoyl-2-oleoyl-sn-glycero-3-phospho-L-serine; NBDDOPE, 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine-N-(7nitro-2−1,3-benzoxadiazol-4-yl); NBD-DPPE, 1,2-dipalmitoysn-glycero-3-phosphoethanolamine -N-(7-nitro-2−1,3-benzoxadiazol-4-yl); rhod-DOPE, 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine-N-(lissamine rhodamine B sulfonyl); rhod-DPPE, 1,2-dipalmitoyl-sn-glycero-3-phosphoethanolamine-N-(lissamine rhodamine B sulfonyl); CDs, cyclodextrins; αCD, αcyclodextrin; βCD, β-cyclodextrin; γCD, γ-cyclodextrin; DMαCD, dimethyl-α-cyclodextrin; CMαCD, carboxylmethylα-cyclodextrin; HPαCD, hydroxypropyl-α-cyclodextrin; MβCD, methyl-β-cyclodextrin; HPβCD, hydroxypropyl-βcyclodextrin; DMβCD, dimethyl-β-cyclodextrin; TMβCD, trimethyl-β-cyclodextrin; CEγCD, carboxylethylated-γ-cyclodextrin; HPγCD, hydroxypropyl-γ-cyclodextrin; PBS, phosphate-buffered saline; FRET, Förster resonance energy transfer; MLVs, multilamellar vesicles; SUVs, small unilamellar vesicles; LUVs, large unilamellar vesicles; GUVs, giant unilamellar vesicles; OD, optical density. 14637

dx.doi.org/10.1021/la4031427 | Langmuir 2013, 29, 14631−14638

Langmuir



Article

(22) Nishijo, J.; Shiota, S.; Mazima, K.; Inoue, Y.; Mizuno, H.; Yoshida, J. Interactions of cyclodextrins with dipalmitoyl, distearoyl, and dimyristoyl phosphatidyl choline liposomes. A study by leakage of carboxyfluorescein in inner aqueous phase of unilamellar liposomes. Chem. Pharm. Bull. 2000, 48, 48−52. (23) Cheng, H. T.; Megha; London, E. Preparation and properties of asymmetric vesicles that mimic cell membranes: effect upon lipid raft formation and transmembrane helix orientation. J. Biol. Chem. 2009, 284, 6079−6092. (24) Cheng, H. T.; London, E. Preparation and properties of asymmetric large unilamellar vesicles: interleaflet coupling in asymmetric vesicles is dependent on temperature but not curvature. Biophys. J. 2011, 100, 2671−2678. (25) Chiantia, S.; Schwille, P.; Klymchenko, A. S.; London, E. Asymmetric GUVs prepared by M beta CD-mediated lipid exchange: an FCS study. Biophys. J. 2011, 100, Lo1−Lo3. (26) Hatzi, P.; Mourtas, S.; Klepetsanis, P. G.; Antimisiaris, S. G. Integrity of liposomes in presence of cyclodextrins: effect of liposome type and lipid composition. Int. J. Pharm. 2007, 333, 167−176. (27) Son, M.; London, E. The dependence of lipid asymmetry upon phosphatidylcholine acyl chain structure. J. Lipid Res. 2013, 54, 223− 231. (28) Son, M.; London, E. The dependence of lipid asymmetry upon polar headgroup structure. J. Lipid Res. 2013, in press.

REFERENCES

(1) Uekama, K.; Hirayama, F.; Irie, T. Cyclodextrin drug carrier systems. Chem. Rev. 1998, 98, 2045−2076. (2) Martin Del Valle, E. M. Cyclodextrin and their uses: a review. Process Biochem. 2004, 39, 1033−1046. (3) Dodziuk, H. Cyclodextrins and Their Complexes; Wiley-VCH Verlag GmbH & Co. KGaA: Weinheim, 2006. (4) Somogyi, G.; Posta, J.; Buris, L.; Varga, M. Cyclodextrin (CD) complexes of cholesterol–their potential use in reducing dietary cholesterol intake. Pharmazie 2006, 61, 154−156. (5) Nishijo, J.; Mizuno, H. Interactions of cyclodextrins with DPPC liposomes. Differential scanning calorimetry studies. Chem. Pharm. Bull. 1998, 46, 120−124. (6) Legendre, J. Y.; Rault, I.; Petit, J. Effects of β-cyclodextrins on skin: implications for the transdermal delivery of piribedil and a novel cognition enhancing-drug. Eur. J. Pharm. Sci. 1994, 3, 311−322. (7) Niu, S. L.; Mitchell, D. C.; Litman, B. J. Manipulation of cholesterol levels in rod disk membranes by methyl-beta-cyclodextrin: effects on receptor activation. J. Biol. Chem. 2002, 277 (23), 20139− 20145. (8) Ohtani, Y.; Irie, T.; Uekama, K.; Fukunaga, K.; Pitha, J. Differential effects of α-, β- and γ-cyclodextrins on human erythrocytes. Eur. J. Biochem. 1989, 186, 17−22. (9) Christian, A. E.; Haynes, M. P.; Phillips, M. C.; Rothblat, G. H. Use of cyclodextrins for manipulating cellular cholesterol content. J. Lipid Res. 1997, 38, 2264−2272. (10) Zidovetzki, R.; Levitan, I. Use of cyclodextrins to manipulate plasma membrane cholesterol content: evidence, misconceptions and control strategies. Biochim. Biophys. Acta 2007, 1768, 1311−1324. (11) Ohvo, H.; Slotte, J. P. Cyclodextrin-mediated removal of sterols from monolayers: effects of sterol structure and phospholipids on desorption rate. Biochemistry 1996, 35, 8018−8024. (12) Radhakrishnan, A.; McConnell, H. M. Chemical activity of cholesterol in membranes. Biochemistry 2000, 39, 8119−8124. (13) Niu, S. L.; Litman, B. J. Determination of membrane cholesterol partition coefficient using a lipid vesicle-cyclodextrin binary system: effect of phospholipid acyl chain unsaturation and headgroup composition. Biophys. J. 2002, 83, 3408−3415. (14) Leventis, R.; Silvius, J. R. Use of cyclodextrins to monitor transbilayer movement and differential lipid affinities of cholesterol. Biophys. J. 2001, 81, 2257−2267. (15) Monnaert, V.; Tilloy, S.; Bricout, H.; Fenart, L.; Cecchelli, R.; Monflier, E. Behavior of alpha-, beta-, and gamma-cyclodextrins and their derivatives on an in vitro model of blood-brain barrier. J. Pharmacol. Exp. Ther. 2004, 310, 745−751. (16) Irie, T.; Fukunaga, K.; Pitha, J. Hydroxypropylcyclodextrins in parenteral use. I: Lipid dissolution and effects on lipid transfers in vitro. J. Pharm. Sci. 1992, 81, 521−523. (17) Motoyama, K.; Arima, H.; Toyodome, H.; Irie, T.; Hirayama, F.; Uekama, K. Effect of 2,6-di-O-methyl-alpha-cyclodextrin on hemolysis and morphological change in rabbit’s red blood cells. Eur. J. Pharm. Sci. 2006, 29, 111−119. (18) Tanhuanpäa,̈ K.; Somerharju, P. Gamma-cyclodextrins greatly enhance translocation of hydrophobic fluorescent phospholipids from vesicles to cells in culture. Importance of molecular hydrophobicity in phospholipid trafficking studies. J. Biol. Chem. 1999, 274, 35359− 35366. (19) Anderson, T. G.; Tan, A.; Ganz, P.; Seelig, J. Calorimetric measurement of phospholipid interaction with methyl-β-cyclodextrin. Biochemistry 2004, 43, 2251−2261. (20) Debouzy, J. C.; Fauvelle, F.; Crouzy, S.; Girault, L.; Chapron, Y.; Göschl, M.; Gadelle, A. Mechanism of alpha-cyclodextrin induced hemolysis. 2. A study of the factors controlling the association with serine-, ethanolamine-, and choline-phospholipids. J. Pharm. Sci. 1998, 87, 59−66. (21) Fauvelle, F.; Debouzy, J. C.; Crouzy, S.; Göschl, M.; Chapron, Y. Mechanism of alpha-cyclodextrin-induced hemolysis. 1. The two-step extraction of phosphatidylinositol from the membrane. J. Pharm. Sci. 1997, 86, 935−943. 14638

dx.doi.org/10.1021/la4031427 | Langmuir 2013, 29, 14631−14638