INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION APPENDIX REFERENCES

Phospholipids, with saturated acyl chains in the sn-1 and unsaturated acyl chains in the sn-2 glycerol backbone position, represent the most abundant membrane lipid class (White, 1973). Physico-chemical studies have largely dealt with lipids composed of either symmetric saturated chains or those with sn-2 chains containing only a single double bond. Polyunsaturated lipids with multiple double bonds in the sn-2 chain, however, constitute a considerable fraction of lipids in neuronal tissues and the retina. Particularly, near-native levels of 50% of docosahexaenoyl (DHA, 22:63) chains in retinal rod outer segment disk membranes are required for proper function of the visual receptor rhodopsin (see Mitchell et al., 1998, and references therein). The content of DHA in gray brain tissue is consistently 20-30% in 45 animal species (Crawford et al., 1977). The fact that DHA accumulates at high concentration in neural membranes raises the possibility that it may alter membrane biophysical properties important for function.

There exists an extensive literature concerned with the importance of DHA for human health, but comparatively little research has been done on the physical properties of this important molecule. Early theoretical (Albrand et al., 1994; Applegate and Glomset, 1986) and spectroscopic (Litman et al., 1991; Paddy et al., 1985; Salmon et al., 1987) studies deal with conformational and packing properties of polyunsaturated chains in membranes. Investigations were considerably intensified over the last few years to find out how such oxidation-prone, exotic lipids affect membrane molecular architecture and dynamics (Everts and Davis, 2000; Holte et al., 1995; Separovic and Gawrisch, 1996) as well as mechanical properties such as lateral compressibility (Huster et al., 1999; Koenig et al., 1997; Rawicz et al., 2000) and water permeability (Huster et al., 1997; Olbrich et al., 2000).

An essential outcome of these studies has been that polyunsaturation loosens chain packing and decreases the strength of cohesive interactions in the membranes. As a consequence, bending stiffness is decreased (Rawicz et al., 2000) and water permeability increased in comparison with membranes of saturated and monounsaturated lipids (Huster et al., 1997; Olbrich et al., 2000). Moreover, it was proposed that polyunsaturated chains, when paired with saturated ones, cause subtle changes of membrane lateral organization and interfacial properties, which may provide the key to understanding the effect of multiple cis double bonds in lipid acyl chains (Holte et al., 1995). It was observed that chain order parameters of saturated chains are lower when paired with polyunsaturated DHA. In particular, order of the segments from the second half to the chains near the bilayer center was lower, perhaps, indicating an altered profile of lateral tension across the bilayer. Polyunsaturation results in a thinner bilayer and an increased area per lipid in almost fully hydrated membranes (Koenig et al., 1997).

Hydration studies on lipids not only provide information on water-binding properties but also probe elastic properties of lipid aggregates (Binder et al., 1999b; Koenig et al., 1997; Parsegian et al., 1979) and lipid phase behavior (Binder et al., 1997, 1999a). Phase transitions between lamellar and nonlamellar, solid gel or subgel, and liquid-crystalline phases reveal microscopic characteristics such as preferred shape, flexibility/rigidity, and hygroscopicity of the molecules and the intermolecular forces acting between them on a macroscopic level.

The degree of hydration of amphipathic structures is thermodynamically determined by the chemical potential of water in the system. In combination with temperature, the chemical potential of water represents an independent thermodynamic degree of freedom, and its variation opens up novel opportunities to study physico-chemical properties of lipids. The degree of hydration of lipid films can be easily varied by exposing the sample to an atmosphere of definite relative humidity (RH). Recently we used this method to study the effect of conjugated double bonds in lipid acyl chains on phase behavior and membrane architecture of diene lipids (Binder et al., 1997, 1999a,c, 1998; Binder and Kohlstrunk, 1999). The main purpose of the present investigation was to obtain the RH-T phase diagram of SDPC and to determine the mean dimensions of lipid bilayers as a function of hydration. We combined infrared (IR) linear dichroism, x-ray, and NMR techniques that are well suited to study lipid phases in terms of aggregate morphology, local interactions, and molecular ordering. The monounsaturated 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC) was included in the study for the sake of comparison between poly- and monounsaturated lipids. Perdeuterated lipid analogs were used to obtain specific signals of the acyl chains in sn-1 and sn-2 positions.

The most interesting observation of this study has been the detection of a novel liquid-crystalline lamellar phase of SDPC that expands laterally upon dehydration. This unusual behavior contradicts typical phase diagrams of membranes of disaturated and monounsaturated lipids. A hypothesis that may explain this surprising tendency is presented.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION APPENDIX REFERENCES

Materials

The lipids SDPC and POPC and their perdeuterated analogs (SDPC-d35 and POPC-d31) were purchased from Avanti Polar Lipids (Alabaster, AL). The stearic acid chain is ~98% deuterated (Holte et al., 1995), except for the C₂ methylene segment that is deuterated to ~50% only. To minimize oxidation of SDPC and SDPC-d35, the lipids were stored as methylene chloride stock solutions with butylated hydroxy toluene (BHT) added at a molar lipid-to-BHT ratio of 250:1. Lipid purity was checked by matrix-assisted laser desorption and ionization time-of-flight mass spectrometry (MALDI-TOF-MS), which detects lipid peroxidation products with high sensitivity. Concentration of peroxidation products in freshly prepared samples was lower than the detection limit, corresponding to lipid purity of better than 98% (Schiller et al., 1998). Lipid samples were studied by infrared spectroscopy over a period of up to 2 weeks after preparation. Initially sharp lyotropic phase transitions become broad after a few days of storage in the sample cell of the spectrometer owing to degradation of the lipid. We used the sharpness of the gel-to-liquid crystalline phase transition of freshly prepared SDPC and SDPC-d35 as a criterion of the integrity of the lipid (see below). The sharpness of the phase transition was tested after each series of Fourier transform infrared (FTIR) measurements. X-ray measurements were performed within 1 day after preparation.

Infrared measurements

Appropriate amounts of the stock solution were spread on the surface of a ZnSe-attenuated total reflection (ATR) crystal for sample preparation (angle of incidence, 45°; six active reflections). The solvent evaporates under a stream of nitrogen within a time of less than 1 min. The amount of material corresponds to an average thickness of the lipid films of >3 µm, or equivalently, to a stack of >500 bilayers in the lamellar phase. Consequently, the films represent bulk samples despite their macroscopic orientation on the surface of the ATR crystal (vide infra). The ATR crystal was immediately mounted inside a sample chamber that protects the lipid film from visible light and from aerobic conditions (Binder et al., 1997). The sample chamber was placed into a BioRad FTS-60a FTIR spectrometer (Bio-Rad, Cambridge, MA) equipped with a wire grid polarizer. Polarized absorption spectra, A() and A(), were recorded with light polarized parallel and perpendicular with respect to the plane of incidence (128 accumulations each). Band positions were analyzed by means of their center of gravity (COG) in the weighted sum spectrum, A() = A() + 2.55 A() (Binder and Schmiedel, 1999).

The lipid films were hydrated by blowing a definite RH, high-purity nitrogen gas through the sample chamber. The temperature (T) and RH were adjusted by means of a flowing water thermostat (Julabo, Seelbach, Germany) and a moisture generator (HumiVar, Leipzig, Germany) with an accuracy of ±0.05 K and ±0.5% RH, respectively. The RH was increased in steps of RH = 3% from 5% to 98% (hydration scan) or decreased in the opposite direction (dehydration scan) at constant T. The sample was allowed to equilibrate for 10 min before measurement at a given RH. No significant hysteresis effects between hydration and dehydration scans were detected, confirming that lipid hydration reached equilibrium. The hydration studies were also conducted as a function of RH of D₂O to improve the spectral analysis in the C==O stretching region, which overlaps with the H₂O bending band.

Determination of infrared order parameters

The IR order parameter of an absorption band, S_IR  P₂(_µ), is defined as the second-order Legendre polynomial, P₂(_µ)  0.5(3cos²_µ  1), of the angle _µ between the respective transition moment, µ, and the ATR normal, n. The angular brackets denote spectroscopic averaging over all groups in the sample that absorb in the respective frequency range where the contribution of each group to the mean value is weighted by its integral absorption coefficient. The IR order parameter was calculated from the dichroic ratio of the integral, baseline-corrected polarized absorbances, R = A/A using Harrick's thick-film approximation (Harrick, 1967; see also Binder and Schmiedel, 1999, for details):

(1)

Selective deuteration of the saturated acyl chains of the lipids allows evaluation of the molecular ordering of the stearoyl (SDPC-d35) and palmitoyl (POPC-d31) chains using the IR order parameters of the symmetric and antisymmetric CD₂ stretching bands near 2090 cm¹ and 2195 cm¹, S_IR(_s(CD₂)) and S_IR(_as(CD₂)), respectively. Their transition moments are assumed to point perpendicular one to another and perpendicular to the fiber axis of the polymethylene chain. Consequently, both parameters can be combined to yield the apparent longitudinal order parameter of the deuterated acyl chains according to (Binder and Schmiedel, 1999)

(2)

which provides a measure of the mean orientation of the chain axis with respect to the ATR normal according to S  P₂(_z). The angle _z is enclosed between n and the interconnecting line between the midpoints of two successive C-C bonds. The longitudinal chain order parameter depends to a high degree on the conformational order of the methylene segments as indicated by a marked drop of S at the chain melting transition of the lipids (see below). Additional effects such as tumbling motions of whole chains, imperfect alignment of the membranes on the ATR surface, fluctuations of the local director due to undulations of the bilayers, a permanent tilt of the chain axes, and also vibrational coupling along the chains can interfere with conformational ordering in lamellar phases. Moreover, bend monolayers in nonlamellar phases decrease IR order parameters considerably (Binder et al., 1999a; Binder and Pohle, 2000). All these effects can be considered in terms of a set of nested uniaxial order parameters if they act independently of each other:

(3)

The indices i = m, d, n refer to the angles between a set of axes, the ith of which distributes in a symmetrical fashion about axis i + 1. For example, _m is the angle between the chain axis m and the local director d whereas _d defines the angles between d and the ATR normal n.

The mean ordering of the proteated unsaturated chains of POPC-d31 and SDPC-d35 is also accessible in the IR linear dichroism experiment. The longitudinal order parameter of the polymethylene fragments of the oleoyl chains were determined from the IR order parameters of the symmetric and antisymmetric CH₂ stretches near 2852 cm¹ and 2922 cm¹, respectively:

(4)

The IR order parameter of the C-H bending mode of the vinyl groups of the dososahexanoyl chains near 1388 cm¹ yields the longitudinal order parameter of the polyunsaturated chain:

(5)

This choice was motivated by the fact that S_IR((C-H)) nearly reaches its maximum possible value of unity in the L phase of SDPC-d35 (vide infra). Consequently, the corresponding transition moments must point along the chain axis for symmetry reasons. A detailed discussion of the linear dichroism of SDPC will be given elsewhere.

Gravimetric measurements

The stock solution was spread on the surface of a circular quartz slide (diameter 15 mm) and allowed to dry under a stream of nitrogen. The sample was placed into a twin microbalance system (Sartorius, Göttingen, Germany) that has been equipped with a moisture-regulating device (see above and Binder et al., 1997). The RH was adjusted by flowing moist, high-purity N₂ gas through the sample chamber. Before starting the experiments, the lipid was dried for 12-24 h at RH = 0% until the mass of the sample attained some constant value (~0.5 mg). The mass increment due to the adsorption of water was recorded in the continuous mode by scanning RH at a constant rate of <±10% RH per hour covering the range of 0-98% RH. The mass increment yields the sorption isotherms presented as the molar water-to-lipid ratio, R_W/L.

X-ray measurements

For x-ray investigations, oriented multibilayer stacks of the lipids were prepared by spreading appropriate amounts of the stock solution on glass slides (20 × 25 mm). Subsequently, the organic solvent was evaporated. The slides were positioned into a sealed thermostatted (±0.5 K) chamber mounted at a conventional Philips PW3020 powder diffractometer (Philips, Eindhoven, The Netherlands). X-ray diffractograms were obtained with Ni-filtered Cu K radiation (20 mA/30 kV) by _x-ray/2 _x-ray scans monitoring the s-range s = 0.1-1.1 nm¹ (s = 2sin/,  = 0.154 nm). The intensity was detected with a proportional detector system. Nitrogen of definite RH was continuously streamed through the sample chamber using a moisture-regulating unit (see above). The samples were investigated at discrete RH values and equilibrated for at least 1 h before measurements. Repeat distances of the lamellar phase, d = ns¹ (n is an integer), were determined with an accuracy of ±0.1 nm using the Bragg peaks of up to fourth order (n = 1-4). The mean area requirement per lipid in the membrane plane was calculated by means of

(6)

where v_L and v_W denote the molecular volumes of lipid and water, respectively.

The mean thickness of the water gaps in the multibilayer stacks was estimated assuming nonpenetrating water and lipid layers (Luzzati, 1968):

(7)

The thickness of the hydrophobic core of the bilayer was estimated by d_hc = d_L  d_pol, where d_L = d  d_W denotes the thickness of the bilayer and d_pol = 2v_pol/A_L is the thickness of the polar part of the bilayer. v_pol is the volume of the nonhydrated polar part of the lipids, which includes the glycerol and carbonyl moieties. We used a value of v_pol = 0.325 nm³ (see (Nagle and Tristam-Nagle, 2000, and references therein). Consequently, the thickness of the hydrophobic core of one-half of the bilayer is L_hc = 0.5d_hc.

NMR experiments

NMR samples were prepared by removing the solvent under a stream of argon with subsequent brief application of vacuum. SDPC in excess water was prepared by adding 50 wt % of deuterium-depleted water. The lipid was transferred to a 4-mm glass sample tube that was sealed with a ground glass joint and cap. An SDPC sample at reduced water content was prepared by blowing argon gas with RH of 33% over a thin layer of dried lipid in a glass vial. The RH was adjusted by blowing the argon through a saturated salt solution of MgCl₂ that was thermostatted at 30°C. The hydrated lipid was collected at the bottom of the tube by centrifugation at 50,000 × g. The sample was sealed with a ground glass joint and cap as above. All preparation procedures were conducted in a glove box (Plas-Labs, Lansing, MI) that was filled with a 90% nitrogen/10% hydrogen gas mixture. Oxygen content was reduced to nondetectable levels by catalytically burning hydrogen. The resulting water vapor was adsorbed in a column. Samples were immediately investigated after preparation. Lipid integrity was verified by hydrolyzing small quantities of the sample and transmethylating the fatty acids. The ratio of saturated to polyunsaturated fatty acids was checked by gas chromatography. Any loss of DHA was less than the resolution of GC peak intensities (±1%).

The ³¹P and ²H NMR spectra were acquired on a Bruker DMX300 spectrometer using a high-power probe with a 4-mm solenoid sample coil tunable to both ³¹P and ²H resonance frequencies. Proton-decoupled ³¹P NMR spectra at a resonance frequency of 121.4 MHz were collected with a Hahn echo sequence with a 1.8-s 90^o pulse, a between-pulse delay of 50 µs, and a repetition rate of one acquisition per second. The spectral width was 125 kHz. Proton-noise decoupling resulted in sample heating of less than 1°C. The ²H NMR spectra were acquired at a resonance frequency of 46.0 MHz using a quadrupolar echo pulse sequence with a 2.2-µs 90^o pulse, a 50-µs delay between pulses, and a repetition rate of two acquisitions per second. De-Paked spectra (Sternin et al., 1983) were calculated using the algorithm of McCabe and Wassall (1995). Smoothed order parameter profiles of the stearic acid chain were computed according to the method of Lafleur et al. (1989).

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION APPENDIX REFERENCES

Lyotropic gel/liquid-crystalline phase transition of POPC

For comparison with SDPC-d35, we first studied the lyotropic phase behavior and molecular order of monounsaturated POPC-d31. The choice of POPC-d31 is motivated by the similar main transition temperatures of fully hydrated SDPC (T_m = 3.8 ± 1.8°C) and POPC (T_m = 2.5 ± 2.4°C) (see (Koynova and Caffrey, 1998, and references therein) that enables comparison of phase behavior without introducing a reduced temperature scale.

Fig. 1 a depicts the center of gravity of the symmetric methylene stretching band of the deuterated palmitoyl chain (COG(_s(CD₂))) as a function of RH at different temperatures. The graphs show the characteristics of a lyotropic chain melting transition between the lamellar gel (L) and liquid crystalline (L) phase. This event is characterized by a sigmoidal increase of COG by 1-3 cm¹. RH scans at increased temperatures shift the RH of the phase transition to smaller RH values.

View larger version (28K): [in this window] [in a new window]   FIGURE 1   Center of gravity of the symmetric CD2 stretching band (a) and longitudinal IR order parameter (S) (b) of the deuterated palmitoyl chains of POPC-d31 as a function of RH. The temperature of the hydration scans was increased from T = 15°C () to 25°C () in steps of 3 K. (c) Longitudinal IR order parameter of the proteated oleoyl chain in the sn-2 position of POPC-d31.

Fig. 1, b and c, depicts the longitudinal order parameter, S, of the acyl chains of POPC-d31. It was calculated from the IR order parameters of the antisymmetric and symmetric methylene stretching bands of the deuterated palmitoyl and the proteated oleoyl chain by means of Eqs. 2 and 4, respectively. Also, the RH dependence of S reveals chain melting by a distinct decrease in the intermediate RH range (Fig. 1). The significantly smaller values of S(oleoyl) suggest reduced conformational order of the monounsaturated chain due to two effects: 1) the bent conformation that acyl chains typically adopt in the sn-2 position of 1,2-diacyl-glycero-lipids near the C2 atom and 2) the disordering effect of the cis double bond.

We found equal linear dichroism of antisymmetric and symmetric methylene stretching modes at all conditions (i.e., S_IR(_as(CD₂))  S_IR(_s(CD₂)) for the palmitoyl chains within the error limits (|S_IR| < 0.03). This result indicates cylindrical symmetry caused by rotations about the chain axis and/or the absence of a uniform tilt of the extended chains in contrast to the arrangement of the palmitoyl chains in the L_' phase of dipalmitoylphosphatidylcholine (DPPC) (Binder, 1999). Probably, the monounsaturated oleoyl chains prevent transverse ordering and/or tilting of the chains in the L phase of POPC.

IR evidence of lyotropic phase transitions of SDPC

The graphs of the center of gravity of the symmetric methylene stretching band of the deuterated stearoyl chain of SDPC-d35 (COG(_s(CD₂))) as a function of RH show the characteristics of a lyotropic gel (L)-to-liquid-crystalline (L) chain melting transition at temperatures T  15°C (Fig. 2 a). The positive values of the longitudinal IR order parameter S(stearoyl) > 0.3 indicate that the lipid membranes preferentially align parallel with the ATR surface (Fig. 2 b). The relation S_IR(_as(CD₂))  S_IR(_s(CD₂)) indicates the absence of transverse ordering of the stearoyl chains, i.e., rotational symmetry about the chain long axis and/or nontilted chain axes with respect to the bilayer normal as was observed also for the palmitoyl chains in the L phase of POPC. The existence of the CH₂ wagging progression of fully proteated SDPC at small RH unequivocally gives evidence of the all-trans conformation of a predominant fraction of the stearoyl chains in the gel phase (not shown).

View larger version (33K): [in this window] [in a new window]   FIGURE 2   Center of gravity of the symmetric CD2 stretching band (COG(s(CD2)); a) and longitudinal IR order parameter (S; b) of the deuterated stearoyl chains of SDPC-d35 as a function of RH. The temperature of the hydration scans was increased from T = 0°C () to 45°C () in steps of 5 K.

Also at higher temperatures, T > 15°C, both spectral parameters, S(stearoyl) and COG(_s(CD₂)), change in a sigmoidal fashion, however, into the opposite direction when compared with the changes at T < 15°C (Fig. 2). A similar behavior was observed at the transition from a nonlamellar phase of inversely curved aggregates (water inside) into the lamellar L phase (Binder et al., 1999a). In this case the increase of S reflects a change in the degree of membrane alignment at the solid surface, perhaps triggered by changes in aggregate morphology (see, e.g., Binder and Pohle, 2000). For example, the order parameter of aggregate morphology S_d (see Eq. 3) is expected to increase by the factor 4 when the system transforms from the H_II (S_d(H_II)  0.25) into the lamellar phase (S_d(lam)  1) (Binder et al., 1999a; Binder and Pohle, 2000). It is important to note that S_d scales the IR order parameters of all vibrational modes of the lipid, and thus the formation of nonlamellar structures is expected to reduce absolute S_IR values of other IR bands of SDPC-d35 as well.

Fig. 3 depicts the chain order parameter of the DHA chains, S(DHA), and Fig. 4 the IR order parameter of the carbonyl and phosphate groups of SDPC-d35 as a function of RH. On the one hand, S(DHA) suggests that the conformation and molecular order of the DHA chains vary in a similar fashion as the conformation and order of the saturated chains (Fig. 3 b). The phase transitions obviously involve the conformation and molecular order of the stearoyl and of the DHA chain.

View larger version (35K): [in this window] [in a new window]   FIGURE 3   Center of gravity (COG((CH)); a) and IR order parameter (SIR((CH)); b) of the C---H deformation mode of the vinyl groups of the DHA chains of SDPC-d35 as a function of RH. Conditions were as in Fig. 2.

View larger version (32K): [in this window] [in a new window]   FIGURE 4   IR order parameters of the antisymmetric PO (above) and P-(OC)2 (middle) stretching modes near 1230 cm1 and 820 cm1, respectively, and of the C==O stretching band (below) near 1735 cm1 of SDPC-d35 (left) and POPC-d31 (right) as a function of RH at different temperatures.

In contrast, the S_IR values of the _as(PO) and the (C==O) bands of SDPC-d35 remain nearly constant under all conditions studied (Fig. 4). This result is important because it indicates that the macroscopic orientation of the lipids on the ATR crystal remains virtually unchanged, and thus the decrease of chain order parameters cannot be explained by formation of a nonlamellar phase. The observed variation of the IR order parameter of the methylene and vinyl groups predominantly originates from changes of the chain conformation. We suggest that a new lamellar phase with highly disordered chains forms at low RH and T > 15°C. It will be designated as L'.

The latter conclusion is confirmed by the center of gravity of the _s(CH₂) and (CH) bands, which change in a similar fashion as the respective IR order parameters (Figs. 1 a, 2 a, and 3 a). The mean frequencies of these modes are sensitive to the conformation of the saturated and polyunsaturated chains, respectively. The slightly higher mean frequency of the _s(CH₂) mode at small RH is compatible with a more disordered conformation of the polymethylene chains when compared with their conformation in the L phase. Direct proportionality between the _s(CH₂) frequency and the respective IR and ²H-NMR order parameters was previously reported for fluid lipid membranes (Kodati and Lafleur, 1993; Le Bihan and Pezolet, 1998). Recently we found that COG(_s(CH₂)) is directly related to the mean area that the chain occupies in the membrane plane (Binder et al., 1999b). Hence, the IR frequencies are expected to reflect changes of the lateral packing of the acyl chains.

Water sorption characteristics

The lipids POPC and SDPC progressively hydrate with increasing RH as indicated by the increasing number of water molecules per lipid, R_W/L (Fig. 5). Comparison of the sorption isotherms of both lipids (T = 25°C) shows that SDPC imbibes slightly more water at RH > 60% than POPC (Fig. 5). The step of the sorption isotherm of POPC at the L-L phase transition reflects a slightly increased hydration potency of the fluid phase (Binder et al., 1999e). No change of hydration behavior was observed at the L'-L phase transition of SDPC.

View larger version (17K): [in this window] [in a new window]   FIGURE 5   Molar water-to-lipid ratio, RW/L, of SDPC and POPC as function of RH at T = 25°C. The vertical dotted line marks the L-L phase transition of POPC and the L'-L transition of SDPC, which occur in both lipids at virtually equal RH.

The interaction of water with the polar moieties of the lipid can be characterized by means of IR spectroscopy. For example, the mean frequency of the C==O stretching vibration of the carbonyl group, (C==O), can serve as a sensitive marker band that characterizes the hydration of the carbonyl groups in lipid assemblies (Binder et al., 1999a; Blume et al., 1988; Pohle et al., 1998). The center of gravity of the C==O band, COG((C==O)), shifts typically toward smaller wave numbers upon hydration due to the formation of H bonds between the water and the carbonyl moieties (Fig. 6). The break of the COG((C==O)) graph at the onset of the chain melting transition of POPC was attributed to the increased water uptake of the carbonyl groups in the fluid L phase (Fig. 6 and Binder et al., 1999e). The mean (C==O) frequency of SDPC shows the same behavior at T < 15°C (see Fig. 6 for T = 10°C). The break of the COG((C==O)) graph of SDPC disappears at T > 15°C (see Fig. 6 for T = 20°C). The carbonyl groups obviously dehydrate more strongly in the RH range that has been attributed to the L' phase when compared with the L phase. In other words, the progressively bigger COG((C==O)) values of SDPC indicate that the carbonyls are more effectively screened from the water than in POPC membranes where a certain amount of water remains trapped near the C==O groups.

View larger version (26K): [in this window] [in a new window]   FIGURE 6   Center of gravity of the C==O stretching band of SDPC-d35 (left) and POPC-d31 (right) at two temperatures as a function of RH.

Interestingly, the IR absorption bands of the phosphate group of SDPC and POPC give no indication of different water binding characteristics of this moiety in the different phases (not shown). The strongly hygroscopic character of the phosphate group obviously prevents significant modifications of its hydration by changes of chain ordering in the investigated systems.

Small-angle x-ray diffraction characteristics of SDPC

X-ray diffractograms of SDPC show a strong first-order Bragg peak and a series of weak, equally spaced peaks up to fifth order (see Fig. 7). This result confirms the existence of multibilayer stacks at all conditions studied. The lamellar gel phase (L) in most cases coexists with a second lamellar phase that shows the characteristics of L' (see below). The intensity of the respective Bragg peaks decreases distinctly after equilibration of the samples for 12 h. We conclude that the gel corresponds to thermodynamic equilibrium whereas L' is metastable at RH < 50% and T < 15°C. Only diffractograms that were measured within 10 h after preparation of the lipid films were considered for further analysis because of potential degradation of the DHA chains at a longer time.

View larger version (23K): [in this window] [in a new window]   FIGURE 7   Representative x-ray diffraction pattern of SDPC-d35, which was measured at 5°C (above) and 25°C (below) at different RHs.

The repeat distance of SDPC-d35 in the L phase is nearly independent of RH (Fig. 8). However, it decreases considerably with hydration in the L' phase. The gel phase of the lipids is characterized by distinctly bigger d values in comparison with those of the fluid systems owing to the stretched conformation of the chains (vide infra).

View larger version (28K): [in this window] [in a new window]   FIGURE 8   Dimensions of the SDPC and POPC bilayers as a function of RH at T = 10°C (SDPC) and 25°C (SDPC and POPC). , Repeat distance d; , thickness of the bilayer dL; , area per lipid AL (Eq. 6); , Lhc = 0.5dhc, thickness of the hydrophobic core of one-half of the bilayer.

Bilayer dimensions

The mean dimensions of the membranes were estimated by means of the simple approach of nonpenetrating lipid/water layers proposed by Luzzati (1968) (Fig. 8). Removal of water from the lipid layers decreases the mean area per lipid, A_L, in the L and in the L phase. That means the lipid bilayers contract laterally upon dehydration. The nearly constant repeat distance shows that the reduction of the interbilayer water gap is almost entirely compensated by the thickening of the bilayers.

In sharp contrast to this behavior, the bilayers expand laterally and contract vertically in the L' phase upon further dehydration. This result seems to be quite atypical. In the discussion section we propose a molecular interpretation for this behavior. Comparison of the dimensions of SDPC and POPC membranes shows that at identical hydration levels in the L phase, area per lipid increases with the number of double bonds in the sn-2 chain. The same behavior was observed when comparing stearoyloleoylphosphatidylcholine (SOPC) with SDPC (Koenig et al., 1997).

Characterization of the L' phase using ³¹P and ²H NMR

The ³¹P NMR spectra of SDPC were recorded over the temperature range 0-50°C. The sample containing excess water was in the L phase, characterized by an anisotropy of chemical shift of 45 ppm at the temperature of 40°C (see Fig. 9 a). The sample prepared at RH of 33% had a phase transition into the L' phase at T  15°C, in good agreement with the IR measurements. The ³¹P anisotropy of chemical shift of L' is considerably smaller than the value of L, just 34 ppm at 40°C. The small peak to higher field from the 90^o orientation shoulder is most likely caused by the presence of a few percent of an L phase (Fig. 9 a).

View larger version (10K): [in this window] [in a new window]   FIGURE 9   NMR characteristics of sn-1 chain perdeuterated SDPC-d35. Proton-decoupled 31P NMR (a) and 2H NMR (b) powder spectra in the L (below, hydrated with 50 wt % water, T = 40°C) and L' phase (above, RH = 33%, T = 40°C). 31P NMR line shapes are characteristic for liquid-crystalline lamellar phases. The anisotropy of chemical shift in the L phase (45 ppm) is larger than the anisotropy of chemical shift of the L' phase (34 ppm). 2H NMR quadrupolar splittings of stearic acid methylene segments is considerably smaller in the L' phase compared with the L phase. (c) Respective smoothed SCD order parameter profiles in the L () and L' () phases calculated from 2H NMR quadrupolar splittings of the stearic acid chains. The solid line in c represents L'order parameters multiplied by a factor of 1.33.

Not only the ³¹P anisotropy of chemical shift was smaller in the L' phase but also the ²H NMR quadrupolar splittings of the stearic acid hydrocarbon chains. The spectra and the corresponding smoothed chain order parameter profiles are shown in Fig. 9, b and c. Hydrocarbon chains in both the L and L' phases have an order parameter plateau in the first half of the chain near the glycerol. Order in the second half of the chain decreases rapidly toward the bilayer center. In first approximation both order parameter profiles differ from each other by a constant factor of 1.33 with lower order in the L' phase (see Fig. 9 c).

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION APPENDIX REFERENCES

RH-T phase diagrams

On the basis of the phase transition data of the FTIR and x-ray diffraction experiments, we are able to construct the RH-T phase diagrams of SDPC and POPC (Fig. 10). The substitution of H₂O by D₂O does not significantly affect the phase transitions. Deuteration of the sn-1 chain in SDPC-d35 shifted the L-L phase transition by 2-4 K to lower temperature. Both lipids have a lyotropic chain melting transition between the L and L phases. Its temperature increases almost linearly with decreasing RH. Lyotropic chain melting transitions of lipids with PC headgroups have been well studied previously (Binder et al., 1999e; Janiak et al., 1979; Jürgens et al., 1983; Pohle et al., 1998), and their existence in SDPC and POPC water dispersions is not surprising.

View larger version (24K): [in this window] [in a new window]   FIGURE 10   RH-T phase diagram of SDPC-d35 (below) and POPC-d31 (above). The solid and open circles are obtained from the center points of the sigmoidal changes of the IR spectral parameters at the L-L and L'-L phase transitions, respectively. The larger symbols correspond to hydration experiments with H2O and the smaller ones to D2O. Hydration and dehydration scans show no significant hysteresis. The transition data of SDPC-d35 in excess water were taken from Holte et al. (1995) (). , L-L phase transitions of the proteated species, SDPC and POPC; , transition data of temperature scans at constant RH.

In contrast to the monounsaturated POPC, polyunsaturated SDPC converts to a liquid-crystalline lamellar phase, L', at higher temperatures and/or lower hydration. The existence of a second lamellar phase at these conditions was not expected. Because dehydration causes a reduction of the headgroup volume, phosphatidylcholines typically form nonlamellar phases of inverse symmetry at low hydration and higher temperatures with a high degree of negative curvature strain in their monolayers. For example, mono- and dihydrates of disaturated PCs (di-C14, di-C16, and di-C18) transform from the gel state to nonlamellar phases of ribbon-like (P), cubic (Q), and/or inverse hexagonal (H_II) symmetry at T > 60°C (Dörfler and Brezesinski, 1983; Janiak et al., 1979; Jürgens et al., 1983). A L-to-nonlamellar transition of nearly dry POPC (RH < 15%) was found at T > 30°C (Binder, unpublished results). Dioleoyl PC (DOPC) exhibits this event upon dehydration at room temperature near RH  40% (Binder et al., 1999d). Interestingly, a lyotropic phase transition between two lamellar liquid-crystalline phases was found for 1,2-diphytanoyl-3-glycero PC (DPhPC), which possesses branched methylene groups in each of the hydrocarbon chains (Hsieh et al., 1997). Also, DPhPC membranes expand laterally at low RH (He et al., 1996). This similarity with SDPC suggests existence of a common underlying molecular mechanism.

The L-L' transition temperature is virtually independent of water chemical potential at RH < 30%. This was measured by means of temperature scans at constant RH (triangles; IR data are not shown). Hydration/dehydration scans at constant temperature do not detect this phase transition.

Thermodynamics of the lyotropic phase transitions

The RH-T phase diagram can be transformed to a µ_W-T representation (see Fig. 11) where

(8)

is the difference of the chemical potential of water adsorbed to the lipid at a given RH and bulk water, µ_W^bulk = µ_W(RH = 100%). R denotes the gas constant. The chemical potential is defined as the partial molar Gibbs free energy of water in the two-component mixture (water plus lipid), µ_WG/R_W/L (G is the Gibbs free energy per mole of lipid). The slope of the transition lines in the µ_W-T phase diagram is directly related to the difference of the partial molar entropy (and partial molar enthalpy) of water at the phase transition and the molar entropy (and enthalpy) of bulk water (see Appendix and Binder et al., 2000):

(9)

with s = s_W(RH^tr)  s_W^bulk (s_W  S/R_W/L) and h = h_W(RH^tr)  h (h_W  H/R_W/L; S and H are the entropy and enthalpy per mole of lipid, respectively; see also Binder et al., 1999e). In other words, the µ_W-T phase diagram yields information about the change of Gibbs free energy, entropy, and enthalpy that accompanies the adsorption of water at the respective phase transition. The results of this analysis for two hydration ranges can be summarized as follows (see Table 1):

View larger version (20K): [in this window] [in a new window]   FIGURE 11   µ-T phase diagram of SDPC-d35 (below) and POPC-d31 (above).

1) The different sign of the slopes of the L'-L and L-L transition lines clearly indicate that the hydration-induced chain melting transition is driven by entropy whereas the transition between the L' and L phase is driven by enthalpy. Chain melting is endothermic because the system loses energy owing to weaker dispersion forces between the acyl chains and/or because of energetically less favorable conformations of the acyl chains in the fluid phase (vide infra). Contrarily, at the L'-L transition the inter-chain interaction energy obviously gains due to denser lateral packing and increased chain ordering in L membranes.

2) The enthalpy and entropy changes compensate each other nearly completely. For example, the weakening of molecular interactions of the gel phase at the chain melting transition transforms into molecular disorder in the fluid state to a high degree. Note that the alteration of composition owing to the adsorption of water is accompanied by the change of Gibbs free energy, and thus enthalpy-entropy compensation is not required in the thermodynamic process studied (Binder et al., 1999e)

3) The partial molar quantities are defined as the change of enthalpy/entropy upon differential adsorption of water. Their absolute values increase considerably with decreasing water activity. In other words, the more direct a water molecule interacts with the lipid, the stronger it affects the properties of the water and lipid. This tendency appears plausible because the interaction strength of water with the PC moieties is expected to increase with dehydration (Binder et al., 1999e; Rand and Parsegian, 1989).

4) h_W and Ts_W of SDPC are significantly bigger than those of POPC. That means that adsorption of a water molecule to the polyunsaturated lipid more strongly affects the enthalpy and entropy of the system than adsorption of the same amount of water to the monounsaturated lipid. In a more general context, this result shows that perturbing a membrane of polyunsaturated lipids affects a wider range of entropic and energetic (i.e., enthalpic) states than a similar perturbation of lipids with monounsaturated and/or saturated chains. This difference between the mono- and polyunsaturated lipids becomes clearly evident at smaller RH at which SDPC transforms from the L into the L' phase with increasing T. The virtually horizontal transition line between the L and L' phase is equivalent to relatively big absolute values of Ts_W and h_W (Table 1). The change of systems enthalpy upon transformation between the solid and fluid phases is given in a first-order approximation by the endothermic heat of chain melting, and thus it can be assumed to be virtually similar for both melting transitions, H^LL  H^LL'. Consequently, the considerably bigger partial molar enthalpy of water at the L/L' transition, h  H^LL'/R^L' h  HLL/R > 0, must be attributed to a considerably smaller increase of hydration, R R. Indeed, the mean frequency of the C==O stretching band, COG((C==O)) (Fig. 6), indicates a relatively weak degree of hydration of the carbonyl region in the L' phase compared with that in the L phase.

Solid membranes of SDPC seem to be less stable than membranes of disaturated and monounsaturated lipids under identical conditions. Information on phase transition temperatures of fully hydrated lipids supports this view. The substitution of cis monounsaturated C18:19 chain for stearoyl chain in the sn-2 position of di-C18:0 PC brings the depression of main phase transition temperature, T_m, by 58 K (Holte et al., 1995; Ichimori et al., 1998). An additional cis double bond in the cis-di-unsaturated chain of C18:0/C18:26-PC decreases T_m further by ~13 K. A third double bond in C18:0/C18:33-PC has nearly no additional effect on T_m. Also, longer polyunsaturated chains such as the DHA chain in SDPC (C18:8/C22:63-PC) leave T_m nearly unchanged when compared with C18:0/C18:26 PC. These facts seem to indicate that the conditions of chain melting/freezing of saturated-polyunsaturated mixed-chain lipids are mainly determined by the saturated chains. With respect to chain melting, the polyunsaturated chains appear to weaken the mean field dispersion energy in the hydrophobic core of the bilayer. A similar conclusion was previously drawn on the basis of an analysis of chain ordering in polyunsaturated C16/C22:6 PC (Salmon et al., 1987). On the other hand, the transition temperature also remains virtually unchanged after substituting the stearoyl chain by a palmitoyl one in C16:0/C18:26 PC (T_m  3°C (Litman et al., 1991)). Hence, the polyunsaturated chains also to some degree contribute to the stability of the lamellae.

Molecular origin of the lyotropically induced gel phase

The cohesive interfacial tension at the hydrocarbon/water interface tends to contract the membrane area. Repulsive forces between neighboring lipids due to steric and entropic effects in the polar and apolar parts of the bilayer counterbalance the attractive forces at the interface. Desorption of water from the lipid assemblies results in an additional lateral tension that compresses lipid bilayers (Koenig et al., 1997). In a simplified view, this effect can be rationalized by the fact that neighboring lipids can approach each other more closely after removing water from the polar region of the bilayer. As a consequence, the area per lipid molecule decreases with dehydration in the L phase (Fig. 8). The reduction of the average cross-sectional area of the lipids is paralleled by an increase of the mean length of their acyl chains, L_hc (Fig. 8), and an increase of the longitudinal chain order parameters, S (Figs. 1 b, 2 b, and 3 b). A denser lateral arrangement of acyl chains causes lowering of the mean (negative) dispersion energy between the chains that is roughly proportional to the longitudinal chain ordering. The shift of the center of gravity of the methylene stretches to smaller wave numbers reflects this tendency (Binder et al., 1999b) (Figs. 1 a and 2 a). At a critical value of the chemical potential of water, POPC and SDPC (at T < 15°C) undergo the lyotropic transition from the L into the L phase. At the transition, the chains convert to a more extended conformation (see also Fig. 12 for illustration). The appearance of the methylene wagging band progression in the spectra of proteated POPC and SDPC in the gel phase provides unambiguous evidence that the all-trans conformation of saturated chains is predominant (data not shown). The half-width of the hydrophobic core of POPC and SDPC bilayers, L_hc  2.0 nm and 2.3 nm, respectively, slightly exceeds the length of the corresponding saturated chains in the all-trans conformation, L_max  1.91 nm (palmitoyl) and 2.16 nm (stearoyl), calculated by L_max  n_CH2·0.127 nm (n_CH2 is the number of methylene groups per chain). Note that the length of extended conformations of the DHA chain, such as the angle iron (2.26 nm) and helical (2.27 nm) conformation (Jandacek and Broering, 1989), are comparable with L_hc of SDPC in L phase.

View larger version (45K): [in this window] [in a new window]   FIGURE 12   Schematic illustration of the different phases of SDPC. The arrows across the phase boundaries (· · ·) point in the direction of increasing chain ordering. Saturated and polyunsaturated chains are indicated by thick and thin lines, respectively. Only one monolayer of the bilayer is shown. See text for details.

Upon transformation into the gel state, the mean thickness of the hydrophobic core of SDPC bilayers increases by ~25% whereas the thickness of POPC membranes increases only by ~12% (Fig. 8). Hence, membranes of polyunsaturated SDPC deform more drastically at this event compared with bilayers of monounsaturated POPC. Presumably these differences are a reflection of the differences in conformational degrees of freedom between saturated, monounsaturated, and polyunsaturated chains

Possible molecular interpretation of the L' phase

A new lamellar phase called L' forms upon dehydration of SDPC at T > 15°C. It is characterized by a larger area per lipid molecule compared with the preceding L phase. The lamellae expand laterally with dehydration in this novel lamellar phase. Hence, the removal of water effectively decreases the absolute value of lateral tension. That means the decrease of volume of the polar region must be overcompensated by a considerable change of membrane architecture that increases the cross section of the lipid molecules.

This unusual behavior may have been caused by substantial interdigitation of the acyl chains from both leaflets of the bilayer. Several arguments strongly contradict this explanation. 1) The area increase of A_L < 0.05 nm² is considerably smaller than the minimum cross section of a chain, a_min > 0.18 nm²; one would expect A_L  2a_min (Pascher et al., 1992). 2) Interdigitation typically appears in solid lipid phases with increasing hydration and not with dehydration of fluid bilayers (Kim et al., 1987). 3) Last but not