Influence of the Vesicular Bilayer Structure on the Sorption of

pubs.acs.org/Langmuir © 2009 American Chemical Society

Influence of the Vesicular Bilayer Structure on the Sorption of Ethylbenzyl Alcohol Pieter Saveyn,*,† Joke De Geeter,† Davy Sinnaeve,‡ Paul Van der Meeren,† and Jos
e C. Martins‡ †

Particle and Interfacial Technology Group, Faculty of Bioscience Engineering, Ghent University, Coupure links 653, B-9000 Gent, Belgium, and ‡NMR and Structure Analysis Unit, Department of Organic Chemistry, Faculty of Sciences, Ghent University, Krijgslaan 281 S4, B-9000 Gent, Belgium Received February 12, 2009. Revised Manuscript Received July 22, 2009

The influence of the physicochemical properties of the vesicular bilayer on the sorption of poorly water soluble compounds was investigated with pulsed field gradient 1H nuclear magnetic resonance (PFG-NMR) for the case of phosphatidylcholine and dioctadecyldimethylammonium bromide (DODAB), using 4-ethylbenzyl alcohol as a model compound. Hereby, the effect of bilayer thickness at a constant physicochemical state was studied using a range of phosphatidylcholines of varying chain lengths, whereas DODAB was preferred to check the influence of the bilayer physicochemical state since this cationic lipid is characterized by three different states within the studied temperature range. When the phospholipid alkyl chain length was changed, no differences were observed in the sorption which was linked to the surface-mediated sorption. On the other hand, when the chemical composition was preserved but the temperature and thus the physical state of the bilayer were changed, the sorption in dioctadecyldimethylammonium bromide (DODAB) vesicles changed dramatically. From those experiments, a strong relationship between the ordering of the surfactant molecules and the sorption can be assumed.

1. Introduction Vesicles consist of one or multiple bilayer shells surrounding an aqueous core. Because of their amphiphilic character, such vesicles have been used as delivery systems for many drugs with low water solubility. Commercial examples of vesicular dosage forms include AmBisome, DaunoXome, and DepoCyt; clinical studies of many other active ingredients are on the way.1 The properties of both the drug and the membrane affect partitioning of the drug into lipid bilayers.2-6 Below the gelto-liquid-crystalline phase transition temperature (Tm), the surfactant molecules in the vesicle bilayer possess more solid phase-like alkyl chains. Above Tm, the surfactant molecules are in the liquidcrystalline state and conformational disorder predominates in the alkyl chains. In general, a more organized structure of the lipid bilayer often leads to a decreased level of partitioning of the drug toward the vesicle bilayer. This explains why sorption is in general lower below Tm. On the other hand, Saveyn et al. showed that sonication of charged lipids results in incomplete lipid chain freezing7 below Tm and, therefore, favors solubilization of benzyl alcohol derivatives.8 Moreover, some β-blockers showed a decreased sorption with an increase in temperature.4,5 Sorption is generally also reduced when the bilayer contains cholesterol, which is known to enhance the ordering of lipids in the bilayer.2,5,8 Several literature reports have also investigated the effect of the *To whom correspondence should be addressed. E-mail: Pieter.Saveyn@ Ugent.be.

(1) Torchilin, V. P. Nat. Rev. Drug Discovery 2005, 4, 145–160. (2) Korten, K.; Sommer, T. J.; Miller, K. W. Biochim. Biophys. Acta 1980, 599, 271–279. (3) Ma, L.; Ramachandran, C.; Weiner, N. D. Int. J. Pharm. 1991, 70, 209–218. (4) Beigi, F.; Gottschalk, I.; Hagglund, C. L.; Haneskog, L.; Brekkan, E.; Zhang, Y. X.; Osterberg, T.; Lundahl, P. Int. J. Pharm. 1998, 164, 129–137. (5) Liu, X. Y.; Yang, Q.; Kamo, N.; Miyake, J. J. Chromatogr., A 2001, 913, 123–131. (6) Word, R. C.; Smejtek, P. J. Membr. Biol. 2005, 203, 127–142. (7) Saveyn, P.; Van der Meeren, P.; Cocquyt, J.; Drakenberg, T.; Olofsson, G.; Olsson, U. Langmuir 2007, 23, 10455–10462. (8) Saveyn, P.; Cocquyt, E.; Sinnaeve, D.; Martins, J. C.; Topgaard, D.; Van der Meeren, P. Langmuir 2008, 24, 3082–3089.

11322 DOI: 10.1021/la9005295

alkyl chain length on drug partitioning. Liu et al. reported that lipids with short chains or more unsaturation provide a more fluid membrane environment that favors drug partitioning in phosphatidylcholines.5 Word et al. reported a constant membranewater molar partitioning with a decrease in alkyl chain length or, in other words, equal sorption with a decrease in alkyl chain length when the molar concentration of the lipids is kept constant.6 To investigate whether sorption can be predicted well by a water-bilayer partition coefficient in which the lipid volume is the main factor, we investigated the effect of the bilayer structure on the sorption behavior of 4-ethylbenzyl alcohol (EBA) in either phosphatidylcholine or DODAB vesicles. Phosphatidylcholines were selected because different chain lengths are commercialy available in highly pure form. DODAB was selected because of its complex but well-characterized thermal behavior.9-11 EBA was selected as an amphiphilic model compound because many drugs contain benzyl or similar aromatic chemical moieties. The partitioning was studied in a quantitative fashion by means of pulsed field gradient 1H nuclear magnetic resonance spectroscopy (PFGNMR). This technique allows us to determine the diffusion coefficient of each component of interest from which the partitioning into the lipid bilayer can subsequently be derived directly in situ in a nondestructive way.12-17 The density of the lipid (9) Cocquyt, J.; Olsson, U.; Olofsson, G.; Van der Meeren, P. Langmuir 2004, 20, 3906–3912. (10) Cocquyt, J.; Olsson, U.; Olofsson, G.; Van der Meeren, P. Colloid Polym. Sci. 2005, 283, 1376–1381. (11) Saveyn, P.; Van der Meeren, P.; Zackrisson, M.; Narayanan, T.; Olsson, U. Soft Matter 2009, 5, 1735–1742. (12) Stilbs, P. J. Colloid Interface Sci. 1982, 87, 385–394. (13) Stilbs, P.; Arvidson, G.; Lindblom, G. Chem. Phys. Lipids 1984, 35, 309– 314. (14) Waldeck, A. R.; Kuchel, P. W.; Lennon, A. J.; Chapman, B. E. Prog. Nucl. Magn. Reson. Spectrosc. 1997, 30, 39–68. (15) Momot, K. I.; Kuchel, P. W. Concepts Magn. Reson., Part A 2003, 19A, 51–64. (16) S€oderman, O.; Stilbs, P.; Price, W. S. Concepts Magn. Reson., Part A 2004, 23A, 121–135. (17) Leal, C.; R€ognvaldsson, S.; Fossheim, S.; Nilssen, E. A.; Topgaard, D. J. Colloid Interface Sci. 2008, 325, 485–493.

Published on Web 09/01/2009

Langmuir 2009, 25(19), 11322–11327

Saveyn et al.

Article

bilayer was measured in parallel using densimetry.18-20 Both techniques were chosen because their application does not perturb the system of interest. This represents a major advantage when studying equilibrium systems.

2. Materials and Methods 2.1. Materials. Dioleoylphosphatidylcholine (DOPC), dipalmitoylphosphatidylcholine (DPPC), dimyristoylphosphatidylcholine (DMPC), and dilauroylphosphatidylcholine (DLPC) were used as purchased from Avanti Polar Lipids (Alabaster, AL) and were >99% pure. Dioctadecyldimethylammonium bromide that was >99% pure was used as purchased from Acros Organics (Geel, Belgium). The model component 4-ethylbenzyl alcohol (EBA) was used as purchased from Sigma-Aldrich and was 99% pure. Deuterium oxide (D2O) with >99.8 at. % D was used as purchased from Armar Chemicals. Tm values of the different vesicular lipids with their corresponding molecular masses are listed in Table 1. 2.2. Preparation of Vesicular Dispersions. We prepared all lipid dispersions by first hydrating the lipids in D2O and gently stirring them at 60 °C (above Tm) for 2 h, using a magnetic stirrer. Subsequently, the phospholipid dispersions were sonicated using a Sonifier 250 (Branson) tip sonicator. The 13 mm tip was immersed approximately two-thirds of the sample height. The power monitor indicated 20%. In the beginning and after every 2 min sonication interval, the sample was left to rest in a water bath at 55 °C for 1 min. A 50% duty cycle was selected to prevent heating of the sample. The sonication time was always 10 min in total. After sonication, the sample was cooled in a water bath at room temperature for at least 1 h. The DODAB dispersions were extruded five times with a TEX060 apparatus (Northern Lipids Inc.) at 65 °C through double-stacked 25 mm polycarbonate membranes with a 200 nm pore size (Track-Etch, Whatman) with pressurized nitrogen gas. No sonication was applied on the DODAB dispersions since sonication can affect the thermal behavior,10 with the occurrence of additional phenomena such as incomplete chain freezing. 2.3. Proton Nuclear Magnetic Resonance (1H NMR). All NMR experiments were performed on a Bruker DRX spectrometer operating at a 1H frequency of 500.13 MHz. A TXI-Zgradient probe with a maximum gradient strength of 55.4 G/cm was used throughout. The temperature was controlled within (0.1 °C with a Eurotherm 2000 VT controller. A 10 min delay was respected between measurements when the temperature was changing. Diffusion coefficients were measured by PFG-NMR with a double-stimulated-echo experiment22 using monopolar gradient pulses to avoid convection effects. The gradient pulse durations (δ) were between 1 and 1.6 ms and had a smoothened square shape instead of the more typical sine shape to maximize the attenuation. To avoid strong loss of signal due to relaxation, a moderate diffusion delay (Δ) of 80 ms was used. More experimental details can be found elsewhere. A detailed description of the PFG-NMR method is available from the review by Johnson.23 2 2

I ¼ I0 exp½ -Dγ G δ  0:81ðΔ -0:5884δÞ

ð1Þ

I ¼ I0 expð -DkÞ

ð2Þ

2

where I is the echo intensity with a gradient, I0 is the echo intensity at zero gradient, D is the diffusion coefficient, γ is the gyromagnetic (18) Nagle, J. F.; Wilkinson, D. A. Biophys. J. 1978, 23, 159–175. (19) Nagle, J. F.; Wilkinson, D. A. Biochemistry 1982, 21, 3817–3821. (20) Uhrı´ kov
a, D.; Ryb
ar, P.; Hianik, T.; Balgav
y, P. Chem. Phys. Lipids 2007, 145, 97–105. (21) Koynova, R.; Caffrey, M. Biochim. Biophys. Acta 1998, 1376, 91–145. (22) Jerschow, A.; M€uller, N. J. Magn. Reson. 1997, 125, 372–375. (23) Johnson, C. S. Prog. Nucl. Magn. Reson. Spectrosc. 1999, 34, 203–256.

Langmuir 2009, 25(19), 11322–11327

Table 1. Tm Values and Molecular Masses of the Different Lipids Used9,21 lipid DODAB DOPC DPPC DMPC DLPC

chain length

Tm (°C)

molecular mass (Da)

18:0 18:1c2 16:0 14:0 12:0

44 41 41 24 0

631 786 734 680 622

ratio, G is the gradient strength, δ is the duration of the gradient, and Δ is the diffusion time. Resonances used to establish the diffusion coefficient of EBA were limited to those in the aromatic region to avoid potential contributions to the decay from broadened lipid resonances that contribute only to the aliphatic region of the spectrum. The determination of the diffusion coefficient with the corresponding 67% confidence interval was based on the fitting of a monoexponential curve to the echo decay of the peak integral of the selected resonances using the Monte Carlo procedure.24 This fitting procedure was repeated 100 times for each experiment; according to Alper and Gelb,24 60 fits are sufficient to produce a constant confidence interval. 2.4. Determination of the Sorption into Bilayers. When a poorly soluble component is added to a vesicular dispersion, it can either occur freely in the aqueous phase or be inserted in the vesicular bilayer. Both environments are characterized by a different chemical shift and diffusion coefficient. Experimentally, only a single set of resonances, each characterized by a single translational diffusion coefficient, could be observed for EBA. This demonstrates that the rate of exchange between the aqueous phase and the bilayer is fast on both the frequency and diffusion time scale. As a result, the diffusion coefficient Dobs of EBA in the presence of vesicles represents an average value given by13 

Dobs ¼ pDvesicle þ ð1 - pÞDfree

ð3Þ



p ¼

Dfree - Dobs  Dfree - Dvesicle

ð4Þ

where Dobs is the observed diffusion coefficient, Dvesicle is the diffusion coefficient of the vesicles, Dfree * is the free diffusion coefficient, corrected for obstruction and retardation,8 and p is the fraction absorbed by the vesicular bilayer. In the calculations, Dvesicle can be neglected because Dvesicle , Dfree. A more detailed description of the determination of the partitioning into bilayers is given in the study by Saveyn et al.8 2.5. Densimetry. Density measurements were performed on a DMA 5000 oscillating U-tube density meter (Anton Paar). The temperature was controlled by a Peltier element within 0.002 °C. The scan rate was ∼0.12 °C/min. The density of the vesicle bilayer (Fbilayer) was calculated by linear regression based on density measurements on a dilution series, using -1 -1 -1 -1 Ftotal ¼ ðFbilayer - Fsolvent ÞX þ Fsolvent

ð5Þ

where X represents the mass fraction of the lipid material. The mass fraction after extrusion was determined from the dry matter upon freeze-drying. The total density of the sample (Ftotal) is measured. The density of the solvent (Fsolvent) can be measured separately or calculated via extrapolation of eq 5 to X = 0.

3. Results and Discussion 3.1. EBA Solution Characterization. In this study, 4-ethylbenzyl alcohol (EBA) was used as a model component to study (24) Alper, J. S.; Gelb, R. I. J. Phys. Chem. 1990, 94, 4747–4751.

DOI: 10.1021/la9005295

11323

Article

Saveyn et al.

Table 2. Diffusion Coefficients (Dfree) with Confidence Intervals of the Different Chemical Moieties of EBA and HDO at Different Temperatures diffusion coefficient (10-10 m2 s-1) T (°C)

aromatic group

benzyl-CH2-OH

benzyl-CH2-CH3

methyl group

HDO

8 27 45 50

3.78 ( 0.01 7.15 ( 0.04 11.15 ( 0.02 13.1 ( 0.1

3.77 ( 0.01 7.05 ( 0.04 11.16 ( 0.04 13.2 ( 0.2

3.75 ( 0.02 7.11 ( 0.05 11.17 ( 0.03 13.1 ( 0.1

3.76 ( 0.01 7.09 ( 0.05 11.14 ( 0.03 13.1 ( 0.1

11.17 ( 0.04 19.84 ( 0.08 29.3 ( 0.4 35.7 ( 0.4

Figure 1. Example of the determination of the diffusion coefficient based on the echo decay of the aromatic resonances of 14 mM EBA at 45 °C as a function of the parameter k (s/m2). The raw data are represented by empty circles, while the solid line is the monoexponential fit from which the diffusion coefficient was derived.

the sorption behavior of amphiphilic components in vesicular bilayers. To quantify its sorption by PFG-NMR as described in section 2.4, the values of the free diffusion coefficient (Dfree) of EBA and HDO were first determined in the temperature interval between 8 and 50 °C (Table 2). A typical example of the signal decay of the aromatic group of EBA at 45 °C with the corresponding monoexponential fit is given in Figure 1 3.2. Influence of the Lipid Alkyl Chain Length. The sorption behavior of EBA in sonicated and extruded dioctadecyldimethylammonium chloride (DODAC) vesicles was recently studied by Saveyn et al.,8 who reported that membrane fluidity was the most relevant factor. To study the influence of the alkyl chain length on the sorption behavior of EBA, we performed the experiments above the chain melting temperature of the different lipids (Table 1). A typical 1H NMR spectrum of a sonicated DPPC dispersion below (a) and above (b and c) Tm is shown in Figure 2. The dispersions were sonicated because small vesicles in the liquid-crystalline state result in narrow line widths in the 1H NMR spectra. Hence, the behavior of the separate chemical moieties of the lipids can be investigated while overlap with the EBA signal is avoided. In addition, sonicated zwitterionic dispersions are less prone to gravity effects. In large multilamellar vesicles, rotation as well as lateral diffusion can interfere with the quantification of EBA diffusion and hence calculated sorption. Whereas sonication may lead to incomplete lipid chain freezing below Tm in charged lipids, it was shown that sonication does not affect the lipid packing in sonicated vesicles of zwitterionic phosphatidylcholines.7 This explains why the alkyl chain signal is completely broadened below Tm (Figure 2a). Therefore, no additional effects from the sonication process were expected. In Figure 2c, a shoulder on the left side of the HDO signal around 4.8 ppm can be observed. This shoulder corresponds to the benzyl-CH2-OH resonance of EBA, which was slightly broadened as a result of the partial sorption on the vesicles. Therefore, to determine the diffusion rate of HDO, we selected a narrow peak 11324 DOI: 10.1021/la9005295

Figure 2. 1H NMR spectra of a sonicated 7.6 mM DPPC dispersion at 23 (a), 45 (b), and 45 °C in the presence of 5.1 mM EBA (c). The intensities were normalized with respect to the area of the choline methyl groups, as indicated with an asterisk. The arrow indicates the shoulder on the left side of the HDO resonance, originating from the benzyl-CH2-OH resonance of EBA.

area and at the same time used a biexponential fitting procedure to cancel out the contribution of the EBA signal.8 First, the sorption was determined at 45 °C where all lipids are in the liquid-crystalline state. The results are summarized in Table 3. Since the different vesicular dispersions have the same molar concentration within each series, these results indicate that sorption of EBA in phosphatidylcholines is independent of the alkyl chain length of the lipids. In contrast, if we consider a constant partition coefficient, the sorption of EBA would depend on the volume of the alkyl chains and decrease with a decrease in chain length as listed as the theoretical sorption in Table 3, based on a constant lipid density. However, this was clearly not the case. From Figure 3, it is clear that mainly the resonances of the PC headgroup are shifted, which indicates that EBA is located close to the phospholipid headgroup region. This explains why the sorption of EBA depends on the number of lipid molecules and thus the molar lipid concentration, rather than the lipid volume determined by the alkyl chain length. 3.3. Influence of Lipid Packing. The complex thermal behavior of DODAB has been observed for many years10,25,26 but has only been understood just recently.11 Above Tm, the lipids in the lipid bilayer possess a high mobility. When cooled below Tm, the lipids lose most of their mobility, stretch, and pack in a hexagonal lattice. When cooled further below the gel-to-subgel transition temperature (Tc), the lipids stretch further and pack into a denser triclinic form in which the mirror planes of the alkyl (25) Blandamer, M. J.; Briggs, B.; Cullis, P. M.; Green, J. A.; Waters, M.; Soldi, G.; Engberts, J. B. F. N.; Hoekstra, D. J. Chem. Soc., Faraday Trans. 1992, 88, 3431–3434. (26) Feitosa, E.; Barreleiro, P. C. A.; Olofsson, G. Chem. Phys. Lipids 2000, 105, 201–213.

Langmuir 2009, 25(19), 11322–11327

Saveyn et al.

Article Table 3. Sorption of EBA in the Presence of Sonicated PC Vesicles at 45 °Ca C18:1c2

-10

C16:0

C14:0

C12:0

m /s) 6.67 ( 0.03 6.81 ( 0.03 6.76 ( 0.03 6.61 ( 0.04 Dobs (10 sorption (%) 40.2 ( 0.4 38.9 ( 0.4 39.3 ( 0.4 40.7 ( 0.6 -10 2 m /s) 28.7 ( 0.3 28.7 ( 0.2 28.4 ( 0.2 28.3 ( 0.2 DHDO (10 6.82 ( 0.06 6.97 ( 0.05 6.99 ( 0.05 6.84 ( 0.05 correction Dobs (10-10 m2/s) sorption after correction (%) 39.0 ( 1.0 37.5 ( 0.8 37.3 ( 0.9 39.0 ( 1.0 39.0 36.4 33.7 30.9 theoretical sorptionb (%) a The lipid and EBA concentrations were 7.6 and 5.1 mM, respectively. b The theoretical sorption was determined on the basis of a constant partition coefficient equal to the partition coefficient in C18:1c2 and a constant lipid density. 2

Figure 3. Chemical shift change of the different chemical moieties of DOPC (black bars) and DMPC (white bars) in a 7.6 mM extruded dispersion after addition of 5.1 mM EBA at 45 °C.

chains are all parallel, restricting the lipid mobility even more. The different phase transitions can be clearly observed as sharp changes in the bilayer density in Figure 4. It was reported that the kinetics of the subgel transition are fast upon heating while they are very slow upon cooling9 which explains why the subgel transition occurs around 36 °C upon heating, while upon cooling this starts only around 23 °C, i.e., a hysteresis of 13 °C. The hysteresis is clearly observed in Figure 4 when the upscan is compared with the downscan. In view of the importance of membrane fluidity in the sorption behavior of EBA, the different types of lipid packing were expected to influence the sorption of EBA in DODAB vesicles. In a first experiment, an extruded DODAB dispersion was diluted with an unsaturated EBA solution to DODAB and EBA concentrations of 6 and 3 mM, respectively. The DODAB dispersion was diluted at 8 °C, which is below Tc, and thus, the DODAB bilayer was in the dense subgel state. Subsequently, the sample was heated to 50 °C, cooled to 8 °C, and again heated to 50 °C. From the corresponding 1H NMR spectra in Figure 5, it is clear that when the system was heated to 50 °C (c and g), the EBA resonances shifted and broadened as a result of the increased sorption compared to that at the lower temperatures. Another remarkable observation is that the resonances are broader in panel d compared to panels b and f, although all spectra were recorded at 27 °C. Again from the peak shift at 50 °C shown in Figure 6, it was clear that EBA is also situated in the headgroup region of DODAB. Similar results for the interaction between dioctadecyldimethylammonium chloride (DODAC) and EBA were reported previously.8 During the temperature cycle, the EBA sorption was quantified simultaneously and is shown in Figure 7. At 8 °C, the DODAB bilayer was in the dense subgel state which explains the low initial Langmuir 2009, 25(19), 11322–11327

Figure 4. Density of the DODAB bilayer in D2O in the absence (b) and presence (O) of EBA. The lipid and EBA concentrations were 6.0 and 3.0 mM, respectively: (a) downscan and (b) upscan, with a scan rate of 0.12 °C/min. Panel c represents the differential density of panel a, clearly showing the different transitions. The subgel and main transitions are indicated with dotted and dashed lines, respectively. DOI: 10.1021/la9005295

11325

Article

Figure 5. Stacked plot representation of the 1H NMR spectra of an extruded 6.0 mM DODAB dispersion in the presence of 3.0 mM EBA at 8 °C (a), heated to 27 °C (b), heated further to 50 °C (c), cooled to 27 °C (d), cooled further to 8 °C (e), again heated to 27 °C (f), and heated further to 50 °C (g). The intensities were normalized with respect to the area of the aromatic group of EBA. A detailed view of the aromatic resonances is shown at the right (not normalized).

sorption (4.1 ( 0.9%). Here, the line width is dominated by the free state. After the sample had been heated to 27 °C, the bilayer was still in the subgel state and hence the sorption remains similar (4.8 ( 0.8%). As sorption has not changed, the slight sharpening of the resonances with an increase in temperature reflects line narrowing, as usual. When the sample was heated further to 50 °C, however, very broad lines were apparent for EBA (Figure 5c-g). Now, both the subgel and main transitions were crossed, and the bilayer was in the loose liquid-crystalline state. The broad signals with correspondingly lower intensity made the quantification of the EBA sorption difficult. However, on the basis of Figures 5 and 7, the sorption at 50 °C is clearly higher despite the higher degree of uncertainty. The broad line width results from the increased sorption, thereby increasing the contribution of the bound, rapidly relaxing state to the overall line width. Because of the slow kinetics of the subgel transition upon cooling, we expected the DODAB bilayer to be in the gel state when the system was cooled to 27 °C. As a result of the denser packing of the lipids in a hexagonal lattice, the sorption dropped to 17 ( 2%, leading to line narrowing. When the system was further cooled to 8 °C, the bilayer was expected to be in the subgel state again. However, the sorption decreased to only 14 ( 3%. When the sample was heated again to 27 °C, the sorption further dropped to 6.3 ( 0.8%. Concomitantly, the EBA resonances are now (Figure 5f) considerably more narrow than before (Figure 5d) at this temperature. This decrease in sorption is inconsistent with the increase in temperature, since the bilayer is expected to remain in the subgel state at both temperatures. In our opinion, this observation may be explained by the fact that upon cooling, subgel formation as well as the desorption kinetics is relatively slow, while the melting of the subgel into the gel phase and the sorption kinetics are expected to be fast. Thus, it could be that equilibrium was not yet reached upon cooling. To verify this, we performed a complementary experiment in which the EBA solution was added at 27 °C to three DODAB dispersions that were previously kept overnight at 4 °C (to ensure the subgel state) and to three DODAB dispersions that were first heated to 60 °C. 11326 DOI: 10.1021/la9005295

Saveyn et al.

Figure 6. Chemical shift change of the different chemical moieties of DODAB in a 6.0 mM extruded dispersion after addition of 3.0 mM EBA at 50 °C.

Figure 7. Sorption of EBA in extruded DODAB vesicles in D2O at different temperatures. The sample was heated from 8 to 50 °C (b), cooled to 8 °C (0), and heated again to 50 °C ([) as indicated by the arrows. The lipid and EBA concentrations were 6.0 and 3.0 mM, respectively.

The sorption of all the samples was quantified at 27 °C. The sorption in the DODAB dispersions that were kept below Tc was 4.7 ( 0.9%, while the sorption in the dispersions that were heated above Tm prior to the measurements was 15 ( 2%. This indicates that during the previous temperature cycle experiment that included cooling from 50 to 27 °C, equilibrium was reached, while this was not the case upon further cooling to 8 °C. Thus, EBA sorption in DODAB bilayers in the gel state was 3-fold higher than in the subgel state, pointing again to the relevance of the lipid packing on the sorption of amphiphilic components. The density of the DODAB bilayer in D2O was measured, both in the absence and in the presence of EBA (Figure 4). In the absence of EBA, the density increase upon cooling below the main and subgel transitions was centered around 40 and 20 °C, respectively. However, in the presence of 3 mM EBA, both transitions were downshifted by 4 and 3 °C, respectively (Figure 4a). This clearly indicates that the amphiphilic molecules disturbed the packing processes of the lipids in the bilayer upon cooling. This fluidizing effect of EBA was reported previously by Saveyn et al.7 Another consequence of this phenomenon is that the subgel transition is only completed around 12 °C, while this is already Langmuir 2009, 25(19), 11322–11327

Saveyn et al.

completed at 17 °C in the absence of EBA, at the same cooling rate. This may also contribute to the nonequilibrium state at 8 °C upon cooling from the gel state. In the upscan in the absence of EBA, the subgel and main transitions are still clearly separated from each other and centered around 37 and 43 °C, respectively (Figure 4b). On the other hand, in the presence of EBA, it seems that the density decreased more when the subgel transition was crossed compared to the case without EBA. A possible explanation is that since both transitions were broadened, part of the bilayer was already in the liquid-crystalline state, and equilibrium was not yet reached.

4. Conclusions We investigated the effect of bilayer alkyl chain length and lipid packing on the sorption behavior of EBA with PFG-NMR. It was shown that the sorption of EBA into lipid bilayers depends on the molar lipid concentration, rather than on the length of the alkyl chains, which indicates that EBA sorption is surface-mediated. NMR peak shifts indeed revealed that EBA preferentially accumulates near the PC headgroup and is largely absent in the terminal methyl region. When DODAB was taken to be a model system, the sorption of EBA was shown to be highly dependent on lipid packing, which in turn affected the bilayer density. In the subgel state,

Langmuir 2009, 25(19), 11322–11327

Article

corresponding to a dense (triclinic) lipid packing, sorption was significantly lower than in the gel state, in which the lipids are packed in a looser (hexagonal) lattice.11 In the liquid-crystalline state, the sorption increased tremendously as a result of the liquidlike behavior of the lipids in the bilayer. A similar effect was earlier observed as a result of incomplete chain freezing upon sonication of charged lipids in general.7 Densimetry revealed that sorption of EBA in the DODAB bilayer had a bilayer fluidizing effect and hence both broadened and lowered the two transition temperatures. As a consequence, we conclude that it is important to know the sample history of lipids with a pronounced hysteresis effect in their thermal behavior when they are used in model studies. Sample history effects can be avoided by careful sample annealing in reaching the equilibrium state. Acknowledgment. We thank the Research Foundation Flanders (FWO) for a Ph.D. fellowship to P.S. and D.S. (FWO aspirant). J.C.M. and P.V.d.M. acknowledge the Fund for Scientific Research-Flanders (FWO-Vlaanderen) for various research and equipment grants (G.0365.03, G.0064.07, and G.0678.08 to J.C.M. and G.0102.08 to J.C.M. and P.V.d.M.). Daan Curvers and Dries Huygens from Ghent University are acknowledged for the very fruitful discussions.

DOI: 10.1021/la9005295

11327