Thermotropic and lyotropic phase properties of glycolipid diastereomers: role of headgroup and interfacial interactions in determining phase behaviour

https://doi.org/10.1016/j.cocis.2004.01.009Get rights and content

Abstract

Recent advances in the area of glycobiology have been paralleled by progress in our understanding of the physical properties of glycoglycerolipids (GGLs). These advances have been accelerated by interest in the new found roles of these simple glycolipids in nature, by advances in synthetic procedures, and by an interest in the technological application of a group of amphiphiles with unique physical and chemical properties. Here, we consider the phase properties of some GGL/water systems containing either a single hexopyranoside or pentopyranoside headgroup. Recent calorimetric and X-ray diffraction measurements of some GGL diastereomers suggest that both headgroup and interfacial hydration play a major role in determining both lyotropism and mesomorphic phase properties as the chemical structure of the lipid headgroup, interface and hydrocarbon chains are systematically altered. For GGLs of a given chain length, interactions between the headgroup/interface and water determine whether or not a highly ordered, lamellar crystalline phase is formed, the number of such phases and their rate of formation and, in some cases, the nature of the molecular packing of those phases. In the liquid crystalline phases, the hydrocarbon chains determine the area per molecule in the lamellar liquid crystalline phase, but it is the cross-sectional area of the hydrated headgroup and the penetration of water into the interface which determines the nature of the non-lamellar phases, probably through small changes in interfacial geometry as the lateral stresses in the headgroup region increase.

Introduction

Glycolipids belong to a larger group of substances termed the glycoconjugates. In nature, three types of glycolipid are commonly found, the glycosphingolipids (GSLs), which contain a sphingosine backbone [1], the glycoglycerolipids (GGLs), which contain a glycerol backbone, and the glycosyl phosphopolyprenols, which contain a phosphate bridge linked to a polyisoprenoid hydrocarbon chain [2]. The two former types of glycolipid are key structural molecules in the membranes of animals, plants and bacteria, whereas the latter are important lipid-linked components of biosynthetic intermediates whose glycosyl residue is readily transferred to some appropriate acceptor which is usually a glycosylated polymer [2]. The GSLs are normally found in the membranes of animals and fungi, where they play both structural and antigenic roles. In contrast, the GGLs are commonly found as structural components in the membranes of Gram-positive bacteria, halophylic archaebacteria, the photosynthetic membranes of higher plants and cyanobacteria, in the seeds of cereals such as wheat and oats [3], [4], and in smaller amounts in specialized tissues of the central nervous system (CNS) in vertebrates (see reviews [5], [6], [7]). Those GGLs containing a single monosaccharide moiety attached through the anomeric carbon atom by means of an α- or β-linkage to either a 1,2 (2,3)-di-O-acyl or 1,2 (2,3)-di-O-alkyl-sn-glycerol are generally termed monoglycosyl diglycerides (MGDGs; Fig. 1).

Until recently, the majority of these simple GGLs were thought to function solely as structural components in the aforementioned membrane systems, where they may comprise up to 75% of the total polar membrane lipid fraction [6], [8]. However, in the last 20 years, the known roles played by simple GGLs in plant, animal and microbial membranes have grown substantially. In green plants, high levels of polyunsaturated 1,2-di-O-acyl-3-O-(β-d-galactopyranosyl)-sn-glycerols in the chloroplast thylakoid membrane have been correlated with the maintenance of membrane fluidity and with the prevention of chilling injuries [9], [10], whereas the binding of certain native galactose-specific lectins with the corresponding diacyl digalactoside (Galα1-6Galβ1-3DAG) in photosynthetic membranes has been shown to decrease freeze-thaw-induced cellular damage [11], and several monoglycosyl glycerols, which are components of many plant cell walls, are known to play a role regulating the osmotic strength in some marine microorganisms [12]. The prevention of membrane damage by these GGLs is supported by the observation that in a psychrophilic bacterium, Listeria monocytogenes, glycolipids were only found in cultures grown at low temperature, and were accompanied by a corresponding decrease in the ratio of 17:0ai to 15:0ai fatty acyl chains [13]. This suggests that some organisms modify the headgroup composition and acyl chain structure of their membranes in order to maintain both a liquid-crystalline bilayer and an optimal ratio of lamellar-to-non-lamellar preferring lipids in order to maintain optimal membrane function.

In photosynthetic membranes, the importance of these glycolipids in the maintenance of photosynthesis has been corroborated by the recent observation that the addition of a neutral, synthetic glycolipid to phospholipid liposomes containing a reconstituted cyanobacterial photosystem II complex resulted in a six-fold increase in the light-induced oxygen-evolving activity [14]•, although the long term stability of these complexes remains uncertain. X-ray crystallography experiments have recently shown that the glycolipids present in a bacterial photosynthetic reaction center are closely associated (within 3.5Å) with the bacteriochlorophyll molecule [15]. Additional support for a close association between membrane glycolipids and the PSII complex has been found in a phytoflagellate, Chrysochromulina sp., where a molecule of chlorophyll C2 esterified to a monogalactosyl diacyl glycerol has been isolated [16]. Other strong associations between glycolipids and proteins in the chloroplast membrane have also been reported. These include the CF0/CF1 H+-ATPase, which is strongly associated with the 1,2-di-O-acyl-3-O-(6-sulfono, 6-deoxy-α-d-glucosyl)-sn-glycerol (6-SD-α-d-GlcDAG) in the chloroplast membrane ([17], [18] and refs therein) and chlorophyllase, which is associated with β-d-GalDAG [19]. It has also recently been suggested that the GGLs themselves may be directly involved in proton transport through the thylakoid membrane by means of a hydrogen bonding process possibly through a water–wire mechanism [20], although the rate of transport may not be significantly different from that seen in phospholipid vesicles.

Although the precise roles of both glycolipids and phospholipids in the energy transduction process have yet to be fully elucidated [21], studies of photosynthetic membrane systems lacking 6-SD-α-d-GlcDAG [22], [23], and genetically engineered mutants deficient in genes for either PG [24], [25], [26], [27], [28] or 6-SD-α-d-GlcDAG [26], [27], [28], have shown that both lipids are required for the full functioning of the PSII reaction centre. In the absence of 6-SD-α-d-GlcDAG, the PSII complex may remain functional, but with a reduction in O2 evolution [29], where as in the absence of PG, in addition to the substantially lower O2 evolution, there are marked structural changes in the membrane morphology [25]. In addition, 6-SD-α-d-GlcDAG has been shown to be important for the proper functioning of the PSII complex under phosphorous-limiting growth conditions [30]•[31], [32] and at high growth temperatures when the system is under heat stress [33], [34], [35], [36]. This latter observation supports the results of work on the membranes of a Microbacterium sp. [35], [36], where the galactoglycerolipids have also been shown to limit the oxidative damage which occurs during heat stress. It has also been suggested that these molecules with oxygen scavenging properties may be virulence factors in microbial pathogens, such as Mycobacterium and Leishmania spp. [37]. The anti-oxidant properties of GGLs are also supported by the observation that the 6-SD-α-d-GlcDAG counteracts the effects of the superoxide anion generated in leukocytes primed with phorbol myristate acetate [38]. However, the mechanism behind this effect is not presently understood.

In the marine environment, some algal blooms have been found to contain highly polyunsaturated, non-ionic β-d-GalDAGs which have been found to be toxic to fish when present at high levels in water [39], [40], [41], [42], as well as being toxic to some microorganisms [43]. In marine invertebrates, these lipids have important secondary roles, such as messengers responsible for initiating feeding behavior [44], [45] and in the selection of habitat in the sedentary life stages of both sea urchins and mussels [46], [47].

The roles of GGLs in vertebrates is restricted by their limited distribution in tissues of the reproductive and central nervous systems [48], [49]. Typically in such membranes, the major antigenic roles are performed by complex GSLs, which, by virtue of their interfacial amide group, possess a tighter interfacial packing than do the corresponding GGLs [50], [51], [52]. Thus, the GGLs are limited to other specialized roles. One typical example is the ‘seminolipid’ (1-O-alkyl, 2-O-acyl-3-O-(3′-sulfo-β-d-galactosyl)-sn-glycerol; 3-S-β-d-GalAAG) isolated from mammalian sperm membranes. After ejaculation, the seminolipid is converted by arylsulfatase A present in the seminal fluid to 1-O-alkyl, 2-O-acyl-3-O-(β-d-galactosyl)-sn-glycerol, which has been implicated as a major contributor to membrane fusion during the process of fertilization [53], [54]•. In addition to its structural role in the mammalian spermatozoon membranes at fertilization [55], this 3-S-β-d-GalAAG has also been shown to be involved in the binding and transfection of viruses, including AIDS [53], [54]•[56], [57], [58], [59]•, influenza virus type A [60] and certain species of Mycoplasma that are known to cause infertility [61], [62]. Recent work suggests that this same lipid may be responsible for the binding of the HIV glycoprotein 120 to spermatozoa and subsequent transmission of HIV to the sexual partner [57], [59]•. Thus, MGDGs are of importance in both an immunological and a structural capacity in the membranes in which they are found.

Recent trends in glycoconjugate chemistry have seen the development of GGLs for numerous biomedical and physical applications. Several GGLs have been found to potentially inhibit virus replication, including the HIV virus through the HIV reverse transcriptase [63], [64], [65], [66], [67], [68]. Similar GGLs have also been found to inhibit several DNA polymerases [66], [67], [69], [70], [71], [72], [73]•[74], [75], [76]•[77]•[78]•[79], [80], [81], [82]•. The inhibition of HIV reverse transcriptase and of specific mammalian DNA polymerases has been shown to be dependent on the chemical structure of the GGL analogues including contributions from the hydrocarbon chains [65], [69], [70], [77]•[78]•[79], the headgroup [65], [72] and the presence or absence of a 6-deoxy, 6-sulfonate group [65], [82]• and the chirality of the glycerol backbone [80], [81]. However, it is not presently clear from these studies whether the physical properties of the GGLs are an important regulating factor in the mechanism of action in each case [78]•[82]•.

More traditional applications of GGLs which utilize the thermotropic and lyotropic phase properties of these lipid molecules include their development as: liquid crystals [83]•[84]•[85]•[86]•[87]•, as surfactants [88]•, a matrix for protein reconstitution [89]•[90]•, liposome delivery systems [91] and, for the native lipids, as emulsifiers in food processing [92]. They may also have a potential application in the crystallization of membrane proteins [93], [94]. Several of these areas have recently received attention in this journal and so here we will focus on our own primary interest, which is the physical properties of structural lipids in biological membranes [95]•[96]••.

Early physical studies of GGLs, which were aimed at understanding their structural diversity and consequently their role in biological membranes, were confined to a few native β-d-GalDAGs isolated from photosynthetic organelles [97], [98], [99], [100], their hydrogenated derivatives [101], [102], [103], [104] and the native α-d-GlcDAGs isolated from the membranes of Acholeplasma laidlawii [105], [106], [107], [108], [109], [110]. In order to extend this limited number of experimental observations, our group and others initiated synthetic programs aimed at preparing ranges of GGLs differing in their hydrocarbon linkage, chain length and structure, headgroup configuration and glycerol chirality in sufficient quantities to enable physical studies of their phase behaviour [111], [112], [113], [114]. The application of suitable carbohydrate synthetic techniques (which continue to evolve), have enabled a more focused, systematic approach to studies on the phase properties of model synthetic MGDGs, as well as to specific studies of their lamellar/non-lamellar phase preference in both model and biological membranes [95]•[96]••. These investigations have utilized a multidisciplinary approach using a combination of low sensitivity and high sensitivity differential scanning calorimetry (DSC), small-angle (SAX) and wide-angle (WAX) X-ray diffraction, Fourier transform infrared (FTIR) spectroscopy and monolayer film techniques to provide both thermodynamic and structural information on aqueous lipid dispersions and to identify patterns of phase behaviour common to all double-chained GGLs. In order to evaluate contributions arising from the changes in the chemical configuration of a lipid molecule to its thermotropic and lyotropic phase behaviour, it is advantageous to investigate the effects of altering a single structural component, such as hydrocarbon chain length, independently. This approach provides a better indication of the contributions of the various molecular components of the GGL molecule to the overall pattern of lamellar solid-state and lamellar/non-lamellar phase polymorphism [96]•• in terms of the component forces and interactions rather than simply considering the ‘size’ [105], [106] or ‘shape’ [115] of the lipid molecule. While from a physical perspective, it is probably not legitimate to divide an amphiphilic lipid molecule into regions consisting of its various structural components, nonetheless, such delineations are useful in that they allow comparisons of molecules in which a single structural component remains constant in a wide diversity of lipid chemical configurations. Here, for comparative purposes, we have partitioned the GGL molecule into three parts, the hydrocarbon chains, the headgroup and the glycerol backbone (interfacial) regions in order to discuss the contributions arising from each structural component.

Calorimetric investigations of aqueous dispersions of synthetic hexopyranoside-containing GGLs with either two acyl or two alkyl chains containing 10 to 20 carbon atoms show a complex pattern of thermotropic and lyotropic phase behavior on heating, which is dependent on chain length and on the thermal history of the sample. Previously heated samples of GGLs with short linear, unbranched, saturated hydrocarbon chains (n<14 alkyl, n≤16 acyl) in excess water, exhibit highly energetic lamellar gel/lamellar liquid crystalline (Lβ/Lα or Lβ/Lα) phase transitions at lower temperatures (Fig. 2), with Lα/inverted cubic (QII) and QII/hexagonal (HII) phase transitions at higher temperatures. With increases in chain length (n=14–17 alkyl, n≥16–18 acyl), the Lβ/Lα phase transition temperature (Tm) [and enthalpy (ΔH)] increase, whereas the Lα/QII or Lα/HII phase transition temperatures (Th) decrease (Fig. 2, Fig. 3). Thus, the temperature range (ΔTL/NL) over which stable Lα and QII phases are evident gradually decreases with increasing chain length until, at a specific chain length (n=14 for dialkyl-β-d-GlcDAGs and n≅16 for the diacyl-β-d-GlcDAGs), the Lα phase converts directly to a HII phase [116], [117], [118], [119], [120]••[121]•[122]••[123]•[124]•. At still longer chain lengths, the Lα window disappears completely and there is a direct Lβ/HII phase transition (Fig. 3; [96]••). These phase assignments were confirmed by SAX diffraction and show characteristic diffraction lines in the ratios, 1:2:3:4 (lamellar), 6: 8: 14: 16: 20: 22: 24: 26 (QII, Ia3d), 2: 3: 4: 6: 8: 9: 10: 11: 12: 14: 16 (QII, Pn3m or Pn3 lattices), or 1: 3: 4: 7: 9: 12: 13 (HII lattice). In the majority of MGDGs, the above basic pattern of mesophase events is reversible on cooling, although at shorter chain lengths (n<14 alkyl, n<16 acyl), more than one bicontinuous cubic phase may be observed on heating, but only a single QII phase may be seen on cooling ([119], [124]•[125]••[126]•• and Mannock et al., unpublished results). In addition, the QII/Lα phase transitions are often substantially supercooled and may entirely obscure the Lα phase region in the cooling regime.

On nucleation and annealing of the diacyl GGLs at low temperatures, the Lβ phase may convert to one or more highly-ordered, lamellar crystalline (Lc) phase(s). The formation of these Lc type phases is detected calorimetrically as highly energetic endothermic events either above or below the Lβ/Lα phase transition temperature. Typically, the phase transition temperatures of these events increase with increasing chain length and may show evidence of odd–even alternation, either in the kinetics of Lc phase formation or in the phase transition temperatures themselves [116], [127]•[128]•, which have been presumed to originate from the influence of ‘end group effects’ on molecular packing. In the diacyl-β-d-GlcDAGs, there is evidence of the formation of two Lc phases. At lower chain lengths (n≤15 carbon atoms), both the intermediate and equilibrium Lc phases are found above Tm, whereas at longer chain lengths (n>16 carbon atoms), only a single Lc phase is found which undergoes an Lc/Lβ phase transition below the Tm. FTIR spectroscopy showed that the chain packing of the Lc phase formed at longer chain lengths is the same as that of the intermediate phase formed at shorter chain lengths [116], [128]•. However, in the corresponding α-d-GlcDAGs, only a single Lc phase hydrocarbon chain packing mode is seen throughout the saturated straight chain series, suggesting that the headgroup anomeric configuration, as well as hydrocarbon chain length, is determining the Lc phase structure [129]•[130]•.

There have been many reports of glycolipids isolated from the membranes of halotolerant and acidothermophillic archaebacteria in which the acyl chains are replaced by highly-branched isoprenoid or alicyclic alkyl chains [7]. Such alkyl substitutions are thought to provide improved membrane stability at high temperatures, low pHs and high osmotic strengths, where acylated lipids would be expected to undergo rapid deacylation (but see [131]). Changing the hydrocarbon chain linkage from acyl to alkyl reduces the polarity of the interfacial region and produces a tighter interfacial packing, enhancing the differences in headgroup configuration. Calorimetric and SAX diffraction studies of model GGLs containing two straight, saturated alkyl chains linked to positions O-1 and O-2 of the glycerol backbone [96], [117], [118], [119], [126]••[132]••[133], [134]•• have been found to possess higher Tms and lower Ths (Fig. 3, Fig. 4) than the diacyl compounds of the same chain length, in agreement with earlier observations of the corresponding PEs [135], [136], [137]•. A comparison of the β-d-GlcDAGs containing either 2 acyl or alkyl chains [116], [117], [118], [119], [124]•[125]••[126]•• found that the Tm is higher and the Th is lower in the dialkyl compounds and that the reduction in Th is consistently greater than the increase in Tm (Fig. 4). This produces a different chain length dependence of, for example, the Lα/HII phase transition in the diacyl and dialkyl lipids (Fig. 3, Fig. 4). Similar patterns of thermal events are observed in diacyl- and dialkyl-β-d-GalDAGs ([96]••[127]•[133], [134]••[138] and Mannock et al., unpublished results) and for the corresponding PEs [135], [136]. This behaviour has been rationalized by suggesting that the absence of carbonyl groups results in a closer packing of the lipid molecules in the lamellar phase together with a corresponding decrease in the penetration of water into the interfacial region, thus shifting the pivotal plane towards the headgroup and the aqueous phase, thereby increasing the tendency to form non-lamellar phases ([96]•• and refs cited therein). For example, at a chain length of twelve carbon atoms, two transitions were found by DSC and SAX diffraction in the diacyl-β-d-GlcDAG below 100°C [26.0°C (Lβ/Lα), 57.7°C (Lα/QII); [116]], whereas in the corresponding dialkyl-β-d-GlcDAG, at least three thermal events [32.6°C (Lβ/Lα), 54.2°C (Lα/QIIa), 64.2°C (QIIa/QIIb), 83.8°C (QII/HII)] were detected (Fig. 5; [117], [118], [119], [126]••). These measurements also clearly demonstrated that QII phases could be formed in a single amphiphile/water system as thermodynamically stable phases over a wide range of temperature. In a related paper, Turner et al. [125]•• confirmed the existence of QII phases in the rac-di-O-dodecyl-β-d-GlcDAG belonging to the Ia3d and Pn3m/Pn3 space groups and reported that their thermal stability was dependent on the direction of temperature change and on the lipid concentration. Similar results have been obtained for both the dialkyl-β-d-GalDAGs ([126]••[132]••[133], [134]•• Mannock et al., unpublished experiments) and the dialkyl PEs [135], [136], [137]•.

One of the structural parameters which has been used as an indicator of the lamellar/non-lamellar phase preference of a lipid molecule in the literature is the area per molecule ([115] and refs therein.). However, it has been reported from measurements of the molecular area of lipid monolayer films at the air–water interface that the di-O-palmitoyl-α- and β-d-GlcDAGs have larger molecular areas than that of DPPE [139], yet the former have a greater tendency to form non-lamellar structures indicating that other factors besides geometric shape play a role in lamellar/non-lamellar phase transitions. In order to investigate the mechanism of the Lα/HII phase transition, we examined the effect of altering the acyl chain structure on the lamellar/non-lamellar phase transition temperatures of some synthetic PEs and MGDGs containing two saturated, straight-chain, two cis or trans-monounsaturated straight-chain, or two branched-chain fatty acids with the different chemical configurations, including methyl-iso (N:0i), methyl-anteiso (N:0ai), ethyl-anteiso (N:0eai), dimethyl-iso (N:0dmi), or an alicyclic fatty acid, ω-cyclohexyl (N:0ch)) (Fig. 1), using a combination of DSC and SAX diffraction measurements [140]••[141]••[142]. The effective chain length (ECLs) of these fatty acids was calculated using standard CC bond distances to provide chain lengths equivalent to saturated, straight chain 16 or 18 carbon atoms fatty acids, respectively. In this way, the contributions to both the Tm and the lamellar/non-lamellar phase transition temperature (TNL) arising from differences in chain structure could be determined independently of chain length. We found that for any effective chain length, modification of the n-saturated hydrocarbon chain structure, through the introduction of a branch or a double bond, lowered both the Tm and TNL. Both the Tm's and TNL's of these MGDG's decreased in the order: linear saturated>methyl-iso>ω-cyclohexyl>dimethyl-iso≅methyl-anteiso>trans-monounsaturated>ethyl-anteiso->cis-unsaturated>cis-diunsaturated. Tm and TNL also became progressively lower as the size of substituent groups at the hydrophobic termini of the chains increased (18:0>18:0i>21:0ch), as the number of substituents at any given position increased (18:0>19:0ai>20:0dmi), and as the position of the substituent was moved towards the center of the chain (19:0i>19:0ai) [140]••[141]••[142].

An appraisal of the Tm, TNL and ΔTL/NL values for the range of chain lengths and chain structures in both the PE's and α-d-GlcDAG's indicates that the Tm's and Th's of the α-d-GlcDAG's are lower on the absolute temperature scale than those of the PE's, but covered a similar span of temperatures in each lipid (Tm.'s: PE's 54.2°C, α-d-GlcDAG's 51.9°C; Th.'s: PE's 47.2°C, α-d-GlcDAG's 44.0°C). The above three parameters followed a similar order in both the PE's and α-d-GlcDAG's as follows: saturated straight chain>methyl-iso->ω-cyclohexyl->methyl-anteiso->dimethyl-iso->trans-monousaturated>ethyl-anteiso-branched, confirming that different hydrocarbon chain structures have similar effects on the lamellar and non-lamellar phase properties in both classes of lipids (Fig. 3, Fig. 6). In the PE's, the ΔTL/NL values were frequently larger and covered a narrower range of values than those of the α-d-GlcDAG's (ranging from 5 to 25°C in the 18-carbon ECL α-d-GlcDAG's and 25–35°C in the 18-carbon ECL PE's; Fig. 6), indicating that the PE's have a lower tendency to form HII phases than the corresponding α-d-GlcDAGs. These results also suggest that the PE headgroup has a greater ability to counteract variations in the hydrocarbon chain packing properties, such as the hydrocarbon chain cross-sectional area and the chain-averaged orientational order with changes in chain structure, than does the α-d-Glc headgroup. Support for this idea was provided by the observation that, in this selection of PE and α-d-GlcDAGs variants, the ΔTL/NL values for lipids with linear, unbranched-chains were significantly smaller than those with branched-chain structures (solid symbols and empty symbols, respectively, in Fig. 6). Although the corresponding values for the PE's were also at the lower end of the ΔTL/NL scale, they were not substantially different from the PE's with other chain structures [141]••.

Corresponding SAX diffraction measurements of these lipids indicate that the hydrocarbon chain length and structure significantly altered the dimensions of both the Lα and HII phases in both the PE's and the MGDG's independently of the Th [140]••[141]••. The magnitude of the d-spacings in the Lα phases of the 16- and 18 carbon ECL α-d-GlcDAG's (Fig. 7) increased in the order: cis-monounsaturated≥trans-monounsaturated>methyl-anteiso-≥methyl-iso-≥linear saturated>dimethyl-iso->ethyl-anteiso->ω-cyclohexyl-branched, whereas those of the HII phases increased in the order: trans-monounsaturated>cis-monounsaturated>methyl-anteiso-≥methyl-iso-≥ω-cyclohexyl->dimethyl-iso->ethyl-anteiso-branched [140]••[141]••. Interestingly, the spacings of the HII phases were smaller for the 18 ECL α-d-GlcDAGs than for the corresponding PEs (Fig. 7c–f), suggesting that the sugar headgroup may have a greater cross sectional area when confined than the PE headgroup, in agreement with NMR and computer modeling calculations [143].

In order to investigate which part of the lipid molecule was responsible for these different dimensions, Harper et al. [140]•• performed Fourier reconstructions of the 18 carbon ECL PE's and obtained the unit cell basis vector lengths (d), the water thickness (w and 2r), the average lipid thickness (〈l〉) and the average area per lipid headgroup, (A). A comparison of d-spacings and l (Fig. 7e,f,h) on the reduced temperature scale found that the HII phase was not formed at a critical Lα phase bilayer thickness, as had previously been suggested [144]•. Moreover, in the Lα phase, the value of l (Fig. 7h) decreases with increasing temperature for all hydrocarbon chain structures, while A is forced to increase. At the same time, w in the Lα phase gradually decreases and is within 1Å for all hydrocarbon chain structures at the phase transition (Fig. 7g). The water layer, w, shrinks at roughly the same rate as A expands, suggesting that the aspect of the headgroup area that sets w varies inversely with A (Fig. 7g,i). At the Lα/HII phase transition, w(r) abruptly increases while A shows a dramatic decrease. At the same time, l shrinks by approximately 1.5Å irrespective of the hydrocarbon chain structure (Fig. 7g,h). With continued heating above Th, A remains constant, while w(r) decreases for all chain structures examined (Fig. 7g,i). These results have been interpreted in terms of a simple picture of molecular events [140]••[141]••, where the number of gauche rotamers in the hydrocarbon chains increases with increases in temperature in the lamellar phase, resulting in a gradual decrease in the lipid length which is accompanied by an increase in headgroup area as the hydrocarbon chains splay outward. At a nearly constant volume per lipid molecule, the one-dimensional geometry of the lamellar phase has only 1 d.f., so a reduction in the lipid length necessarily results in an increase in headgroup area. At the Lα/HII phase transition, the energetic cost of maintaining this increase in headgroup area becomes too great and the system is forced to adopt the HII phase. Since the two-dimensional geometry of the HII phase has 2 d.f., the lipid monolayer thickness can change independently of the area per lipid headgroup and thus the lipid chains and headgroups are allowed to assume their respective desired states. The fact that the headgroup area in the HII phase does not depend on the hydrocarbon chain structure is clearly shown in Fig. 7i. This in turn suggests that the unfavorable change in lipid headgroup area, enforced by the change in the lipid layer thickness in the lamellar phase, may be viewed as driving the phase transition [145]. From these observations, it is evident that the hydrocarbon chains determine the area per molecule in the lamellar phases, whereas the headgroup determines the area per molecule in the HII phase [140]••[141]••. This picture of molecular events seems reasonable, but it does not explain why the GGLs have larger molecular areas, but a greater tendency to form non-lamellar structures than the corresponding PEs [139].

The relationship between the GGL headgroup chemical structure and phase behavior has been considered primarily in terms of the effects of altering the anomeric configuration of the glycosidic bond (α vs. β), or the isomeric configuration (d-Gal, d-Glc, d-Man, etc), or the charge and hydrophobicity of the sugar headgroup. In general, the differences in the mesophase pattern of events are most evident at shorter hydrocarbon chain lengths since, with increasing chain length, the properties of the GGLs are dominated by the hydrocarbon chain-melting process (Fig. 2, Fig. 3, Fig. 4; [96]••). In both the diacyl-β-d-Glc- and GalDAGs, changing the anomeric configuration from β to α reduces Tm and raises Th (Fig. 2, Fig. 3; [116], [127]•[129]•), increasing ΔTL/NL and thus reducing their non-lamellar propensity. These observations are consistent with a decrease in the area per molecule at the air–water interface (α:42Å2, β:41Å2), although the PEs have a smaller molecular area (40Å2) and higher values of Th (Fig. 3, Fig. 4; [139]), suggesting that other contributions may be stabilizing the Lα phase in the PEs. Typically, in the small-angle region, the d-spacing of both the Lβ and Lα phases of a homologous series of specific GGLs differing only in their hydrocarbon chain length each fall on a straight line. With changes in headgroup configuration, there are small differences observed in the Lβ phase d-spacing values, probably arising from small variations in hydrocarbon chain tilt and headgroup orientation, but for a given chain length, those of the Lα phases are remarkably uniform [96]••[117], [126]••[127]•, indicating that these polyhydrated mesophases are motionally averaged [96]••[136]. The above results also suggest that in double chained amphiphiles, differences in headgroup configuration and glycerol chirality only exert their influence in phases where intimate headgroup and interfacial contact occurs, either due to loss of water from the interface with subsequent Lc phase formation [122], or a change in monolayer curvature leading to the formation of non-lamellar phases [140]••[141]••.

2H-NMR spectroscopic studies of the α- and β-anomers of 1,2 ditetradecyl-3-O-(d-Glc)-sn-Gro ([146], [147], [148] and references therein) found that the preferred orientation of the β-d-Glc headgroup was almost parallel to the bilayer surface in the Lα phase, whereas the preferred orientation of the β-d-Glc headgroup was extended away from the bilayer surface. In the β-anomer, they also found that there was a large change in the amplitude of angular motions of the sugar headgroup about its preferred axis of reorientation at the Lα/HII phase transition, but similar changes were not observed with the corresponding α-d-Glc headgroup at Th. This change in amplitude produces a substantial decrease in cross-sectional area, estimated to be ≅90Å2 in the Lα phase vs, ≅55Å2 in the HII phase [147]. Monolayer studies have shown that DTGL has a much larger surface area (102±4Å2) than does dimyristoylphosphatidylethanolamine (82Å2), and it has been suggested that at low surface pressures, the practically unhindered rotation around the glycosidic bond permits the glucose head group to assume conformations with a larger space requirement than that found at high packing densities [118]. This is supported by NMR and computer modeling calculations, which suggest that the sugar head group, by virtue of its bulky ring structure, would be conformationally restricted at higher lateral pressures [143].

It is possible that the changes which occur with the β-anomer may be a ‘conformational adjustment’ to the increased steric crowding of the sugar headgroup as a result of the marked change in monolayer curvature which occurs at the Lα/HII phase transition. Given this, then the fact that a similar change does not occur at the Lα/HII phase transition of the α-anomer suggests that there may be energetic barriers preventing a similar conformational change and that such energetic barriers contribute to the lower non-lamellar-forming propensity of the α-d-GlcDAGs. Jarrell and coworkers also reported 2H-NMR measurements of the 1,2-sn-ditetradecyl-α-d-ManDAG. DSC and 2H-NMR spectroscopic studies found that the α-d-Glc derivative has Lβ/Lα and Lα/HII phase transitions at approximately 52°C and 59°C, respectively, whereas the α-d-Man derivative exhibited an Lβ/HII phase transition just below 48°C [148]. Thus, the reorientation of the ring OH-2 group from equatorial to axial, while the OH-4 remains equatorial, substantially decreases Th, increasing the tendency to form non-lamellar phases. As well, changes in the orientation of the 2- or the 4-OH groups to either the d-manno- or d-galactopyranosyl configuration alters the magnitude and direction of the non-lamellar phase preference. A comparison of the β-d-Glc and α-d-ManDAGs shows that the α-d-Man configuration also has a greater tendency to form non-lamellar structures than the β-d-Glc configuration. These authors found that both the α-d-Man and the β-d-Glc headgroups have orientations which extend away from the bilayer surface, but with significant differences in the rate and amplitude of the angular motions about their preferred orientations [147], [148]. Furthermore, they suggested that these differences probably originate from an assortment of headgroup and glycerol backbone steric and motional contributions influencing the overall interfacial conformation [147], [148]. They calculated that the amplitude of angular motion about the preferred axis of orientation of the α-d-Man headgroups described a semicone angle of some 24°, a value which is significantly smaller than the 37° described by the β-d-Glc headgroups. An increase in the amplitude of headgroup angular motions inevitably increases the time-averaged headgroup effective cross-sectional area, suggesting that the greater non-lamellar phase preferrence of the α-d-ManDAG can be explained in terms of a smaller headgroup cross-sectional area than the corresponding d-GlcDAG anomers. Thus, the non-lamellar propensity of these GGLs depends both on the anomeric and stereoisomeric configuration of the headgroup and its ability to adopt a compact headgroup conformation which may, in some cases, be aided by a change in headgroup orientation.

To test the hypothesis that there is a desired headgroup area for a given headgroup in the 18 carbon ECL PEs and α-d-GlcDAGs, Mannock and co-workers [141]•• presented a simple lateral stress model of the Lα/HII phase transition in which they calculated the desired headgroup area for the PEs and then calculated the expected values of Th in a reverse application. In order to test the general applicability of the mathematical model, they used a headgroup area obtained from a phase diagram of the di-dodecyl-β-d-GlcDAG [125]•• to obtain the theoretical values of Th assuming the same packing constraints in the GGL. The values of Th obtained for both the PE and α-d-GlcDAG were in good agreement with the actual measurements, supporting the hypothesis that it is the hydrocarbon chains which determine the molecular area in the lamellar phase, but that is the headgroup which determines the area per molecule in the HII phase. This outcome also supports, but by no means proves that the general mechanism of the L/NL phase transition is the same in both lipids, despite the obvious differences in the chemical configuration of their headgroups [96]••, and is in agreement with recent SAX diffraction temperature-jump measurements of intermediates formed at the Lα/HII phase transition in both the GGLs and PEs [121]•[149]••.

This conclusion is also supported by recent investigations of the effect of hydrostatic pressure on both the hydrated PC/FA and PE systems, which have shown that with increasing pressure the HII phase destabilizes in favour of the bicontinuous cubic phases [150]•[151]•. Unfortunately, at this time there are no comparable investigations of corresponding GGLs containing hexopyranoside headgroups. However, similar measurements of some β-d-XylDAGs have shown that the micellar cubic Fd3m phase is destabilized in favour of the HII phase with increasing pressure [151]•, suggesting that the effects of pressure and temperature on monolayer spontaneous mean curvature and bending rigidity, and thus the lipids non-lamellar phase preference, may be explained by the same mathematical models despite differences in headgroup chemical configuration [151]•[152]••[153]••.

Additional support for the idea that the hydrocarbon chains play the major role in determining the molecular area in the lamellar phases comes from recent investigations of GGLs in which the hexopyranoside headgroup (d-Glc, d-Gal, d-Man) has been replaced by a pentopyranoside headgroup (d-Xyl, d-Arab, d-Rib). DSC and SAX diffraction measurements of some 1,2-dialkyl-3-O-(d-xylopyranosyl)-sn-glycerols (d-XylDAGs) have found that the chain-melting phase transition of the di-14:0-β-d-XylDAG occurs at 50.9°C (52.1°C for the β-d-GlcDAG), but SAX diffraction measurements show that it is an Lβ/HII phase transition [154]•, rather than the Lβ/Lα phase transition seen in the corresponding β-d-GlcDAG [118], [120]••[126]••[133]. Similar measurements of the corresponding d-ArabDAGs (approx. 49°C; Mannock et al., unpublished experiments), and d-GalDAGs (51.8°C; [126]••[132]••[133], [134]••) and the d-RibDAG (approx. 46–8°C; Mannock et al., unpublished experiments) and α-d-ManDAG (approx. 48°C; [148]), also support this argument and suggest that removing the hydroxymethyl group from the pyranoside ring does not markedly effect the temperature of transitions from the Lβ phase, although it does destabilize the Lα phase in favour of the HII and micellar cubic phases.

There have also been several reports of differences in the pattern of solid-state polymorphism in GGLs differing in their headgroup chemical configuration. On annealing for extended periods of time, some but not all GGLs form ordered Lc phases. Whether or not such phases are formed depends on the hydrocarbon chain length and structure, headgroup stereochemistry and anomeric configuration, and the chirality of the glycerol backbone. The effects of increasing the hydrocarbon chain length have been discussed above. Typically in diacyl and dialkyl GGLs containing a single hexopyranoside headgroup, the β-d-GalDAGs form Lc phases more rapidly than the corresponding β-d-GlcDAGs [116], [118], [119], [127]•[128]•[133], [138], whereas no Lc phases are formed in the α-d-ManDAGs [148]. In both the diacyl-d-Glc- and -GalDAGs, the α-anomers form Lc phases much more rapidly than the β-anomers ([129]•[130]•; Fig. 2). Typically, the β-anomers show evidence of intermediate Lc1 phases en route to the stable Lc2 phase which may be chain length dependent [128]•, while the α-anomers form only a single Lc phase which is isostructural for the linear, straight hydrocarbon chain series ([129]•[130]•; Fig. 2). FTIR and XRD measurements of the Lc1 phases in the β-d-Glc- and β-d-GalDAGs have found that they are structurally similar [122]••[127]•[128]•[138], [155], [156], whereas the WAX measurements suggest that the Lc phase observed in the α-d-GlcDAGs is structurally similar to the Lc2 phase evident in the β-d-GalDAGs [130]•. A similar correlation may exist between the Lc phases of the dialkyl-β-d-Glc- and -β-d-GalDAGs [119]. Thus, while the headgroup structure does seem to moderate the structure and rate of formation of these Lc phases, because of differences in their hydrocarbon chain packing subcells, the rates of Lc phase formation cannot be considered as definitive measurements.

Recent DSC and XRD measurements of dialkyl GGLs with pentopyranoside headgroups have found that the β-d-ArabDAGs form Lc phases more rapidly than the β-d-XylDAGs and the α-d-RibDAGs, which form Lc phases only very slowly or not at all ([154]• and Mannock et al., unpublished experiments). These observations suggest that removing the hydroxymethyl group from the pyranoside ring does not significantly alter the stability of the headgroup hydration sphere in these lipids. Although these studies are presently incomplete, the measurements of hydration number for the Lβ and Lc phases [122]••[155], [156] and the headgroup-dependent loss of water with the resulting formation of an Lc phase, appear to follow the order β-d-Gal>β-d-Glc>α-d-Man and β-d-Arab>β-d-Xyl=α-d-Rib. These observations correlate well with the stability of the hydration shells in the corresponding methyl glycosides [157] and provide an interesting insight into the role of the sugar moiety in the dehydration/hydrocarbon chain packing rearrangement in the GGL bilayer.

Other clues to the relationship between glycolipid headgroup structure and their phase properties have emerged from studies of aqueous dispersions of GGLs with chemically modified β-d-Glc headgroups. It was recently found that the 1,2-dialkyl-3-O-(β-d-glucuronosyl)-sn-glycerols (β-d-GlcUADAGs) have Tms similar to those of the corresponding β-d-GlcDAGs, but they exhibited no discernible non-lamellar phases when dispersed in excess water over the range of pH from 1.5 to 10 [158]. That the oxidation of the primary alcohol of the β-d-GlcDAGs to a carboxylic acid significantly decreases the non-lamellar propensity, even when the pH of the aqueous medium exceeded the pKa of the carboxyl function, is not unexpected. Under such conditions the headgroup carboxyl function would be ionized and the repulsion between the negatively charged headgroups, in combination with an increase in headgroup hydration caused by the polarizing effects of the charged carboxyl group on the headgroup hydration shell, should present a substantial energy barrier preventing an increase in monolayer curvature and the formation of non-lamellar phases. However, it is surprising that the β-d-GlcUADAGs do not form any type of non-lamellar phase even under conditions in which the carboxyl group should be fully protonated [158], in contrast to the negatively charged phospholipids such as dioleoylphosphatidylserine, which when protonated forms non-lamellar phases at moderate temperatures [159]. It is unlikely that the behaviour of the β-d-GlcUADAGs can be attributed an increase in headgroup size per se or to changes in the capacity for interheadgroup hydrogen bonding, because the carboxyl group is only slightly larger than the primary alcohol and, even when the former is fully protonated, it has a similar capacity for hydrogen–bonding interactions. Estimates of the water of hydration in the lamellar gel phase of the β-d-Glc and -GlcUADAGs have found that the β-d-GlcDAG binds less water than the β-d-GlcUADAG (approx. 3.6 waters per glucosyl headgroup vs. ∼5.3 per glucuronosyl headgroup; [158]). Recent measurements of PE/DGDG bilayers using atomic force microscopy have found that the zwitterionic PEs possess a long-range electrostatic double-layer force, which is absent in the nonionic glycolipid ([160]• and refs cited therein) and this might also explain why the lamellar phase is more stable in the PEs than in the MGDGs as well as partly explaining the differences observed between the Glc- and GlcUADAGs.

Studies of the effects of acylation or alkylation of the OH-3 group of the d-Glc headgroup have also been reported. Upon methylation of the OH-3 group of 1,2-sn-ditetradecyl-β-d-GlcDAG, the hydrated lamellar mesophases become thermodynamically unstable and direct interconversions between Lc and inverted non-lamellar phases are observed at ∼60°C on heating and ∼40°C on cooling [161]. The HII phase of this lipid was also replaced by one or more inverted micellar phases at higher temperatures. Thus, this O-methylated GGL exhibits potent non-lamellar-forming tendencies despite the fact that O-methylation of the β-d-Glc OH-3 group results in a small increase in the steric bulk of the headgroup while reducing its ability to donate hydrogen bonds. N-methylation of the PE headgroup also increases headgroup size and reduces nominal headgroup hydrogen–bonding capacity while decreasing the tendency of the N-methyl-PE to form non-lamellar phases. In contrast, O-methylation of the β-d-GlcDAG at OH-3 markedly increases the headgroup hydrophobicity, decreasing its aqueous solubility and perturbing the stability and structure of its hydration shell. The resulting decrease in headgroup hydration greatly outweighs the opposing effects of increases in headgroup steric bulk and decreases in hydrogen-bonding capacity. In a closely related study, a 1,2-diacyl-3-O-[3-O-acyl-α-d-Glc)]-sn-glycerol isolated from A. laidlawii A membranes was found to exhibit a direct Lc to an inverted micellar phase transition at ∼80°C [162]. Thus, the esterification of the OH-3 of the α-d-GlcDAG with a long chain fatty acid also increases the tendency of the GGL to form non-lamellar phases. This result is in marked contrast to the behaviour seen upon N-acylation of the PEs with a similar long chain fatty acid [163]. In principle, the enhanced non-lamellar-forming tendencies exhibited by the O-acylated glucolipid can be largely rationalized using the same arguments presented for the 3-O-methylated GGL described above. Moreover, since the long acyl chain of the 3-O-acyl-α-d-GlcDAG will almost certainly partition into the hydrophobic region, one can also suggest that the hydrophobic packing requirements of this additional acyl chain will accentuate the formation of an inverted non-lamellar phase. The absence of similar behaviour in the corresponding N-acylated PEs can be attributed to the fact that such acylations destroy the positive charge on the nitrogen, thereby converting the zwitterionic PE into a negatively charged lipid species. With the latter, the attenuation of non-lamellar-forming tendencies can be attributed to the combined effects of repulsion between the negatively charged headgroups and the reduction of their hydrogen-bonding capacities.

While diacyl GGLs isolated from eucaryotic cell membranes have the 1,2-sn-glycerol configuration, the dialkyl GGLs which are found naturally in halotolerant archaebacteria are exclusively 2,3-sn (for reviews, see [5], [6], [7]). Although these disparities in glycerol chirality almost certainly reflect differences in the lipid biosynthetic pathways of these various organisms, it raises the interesting question of whether such differences provide any physical advantage in the real world. While the effect of changing the chemical configuration of the headgroup and hydrocarbon chains in both phospholipids and glycolipids has received considerable attention, there have been relatively few reports of the effects of changes in the chemical configuration of the interfacial region in these glycerolipids [96]••. The main observation from physical studies of the phospholipids is that both the enantiomeric and racemic configurations have similar mesophase properties as judged by DSC and XRD measurements, but differ in their Lc phase properties [164], [165], [166], [167]. However, in the GGLs the issue of chemical configuration is complicated by the existence of at least two chiral centres, one at C-2 of the glycerol and another at C-5 of the hexopyranose ring, which means that molecules chiral at both centres are diastereomers, rather than enantiomers like the phospholipids, PC and PE. Most early reports of the thermotropic properties of the dialkyl GGLs [118], [133], [146], [147], [148], [168], [169] focused on the thermotropic properties of the 1,2-rac and 1,2-sn GGLs. As discussed above, the pattern of thermotropic and lyotropic behaviour in these MGDGs was found to be chain-length dependent. On heating aqueous dispersions of the 1,2-di-O-dodecyl-β-d-GlcDAGs, the DSC thermograms were found to exhibit a sharp, energetic endothermic peak at ∼32°C followed by two broad, poorly energetic endothermic events at ∼50–60°C and 78–88°C (Fig. 5a) for both the diastereomers and their mixture [117], [118], [119], [123]•[126]••. SAX diffraction measurements of these GGLs exhibited several characteristic diffraction patterns which were consistent with the existence of lamellar, QII and HII phases (see above). Plots of the lattice parameter (Fig. 5b) for these phases as a function of temperature for each configuration identified the events seen by DSC as Lβ/Lα, Lα/QII and QII/HII phase transitions, respectively. The Tms in all three di-O-dodecyl-β-d-GlcDAGs configurations showed little variation with changes in the chirality of the glycerol backbone (Fig. 5a), in agreement with calculations of Tm for bilayers of various MGDGs derived from monolayer film measurements [120]••. Although the d-spacings of the lamellar phases were almost identical (Fig. 5b), there were significant differences in the temperatures of both the Lα/QII and QII/HII phase transitions and in the magnitude of the lattice spacings just above the Lα/QII phase transitions in all three interfacial configurations (Fig. 5b). The Lα/QII phase transition temperatures decreased in the order 58°C (1,2-sn)>56°C (1,2-rac)>50°C (2,3-sn), and the same order was evident in the corresponding QII/HII phase transition temperatures 89°C (1,2-sn)>87°C (1,2-rac)>77°C (2,3-sn). DSC measurements obtained in the cooling direction showed subtle differences from the heating thermograms. Both the chain solidification and the HII/QII phase transition were reversible, but there was no evidence of a QII/Lα phase transition (Fig. 5a). XRD measurements confirmed that the Lα phase was absent on cooling. The SAX diffraction cooling profiles obtained from the two GGL isomers and their mixture each had almost identical lattice parameter values in both the lamellar and non-lamellar phases, although on the temperature scale the Lα/QII and QII/HII phase transitions in the 2,3-sn diastereomer were significantly reduced. Thus, changes in the interfacial configuration do not seem to drastically alter the dimensions of the shared phases. In this regard, the magnitude and slope of the Pn3m QII phase lattice parameters seen on cooling was also very similar in each diastereomer and differed substantially from the values seen on heating in the QII phase region just above the Lα/QII phase transition in both the 1,2-sn and 1,2-rac compounds, but not in the 2,3-sn isomer. On further investigation, an additional cubic phase belonging to space group Ia3d was identified just above the Lα/QII phase transition in both the 1,2-sn and 1,2-rac-β-d-GlcDAGs. The existence of an Ia3d phase is supported by a phase diagram of the 1,2-rac-di-dodecyl-β-d-GlcDAG, in which the Ia3d phase was found from 8 to 38 wt.% water content [125]••. At high water contents, phase coexistence between the Ia3d and Pn3m phases was observed, but the Pn3m phase was absent at water contents below 28wt.% and the Ia3d phase was displaced to higher temperature by an Lα phase at water contents of between 16 and 28wt.%. It has been suggested that this ‘re-entrant’ phase behaviour is the result of a balance between the lipid monolayer curvature and hydrocarbon chain packing free energies with changing hydration [125]••[170]•.

DSC and XRD measurements of the didodecyl-β-d-GalDAGs typically show three mesophase transitions on heating; an Lβ/Lα phase transition at 30–32°C; an Lα/QII phase transitions at 58–60°C; and a QII/HII phase transition at ∼95°C. Although the mesophase behaviour of the β-d-GalDAGs is similar to that of the β-d-GlcDAGs, the former form Lc phases at much faster rates than the latter. This presents significant difficulties in comparing the effects of changes in the orientation of OH-4 (axial=Gal, equatorial=Glc, Fig. 1) and at C-2 of the glycerol backbone on lipid phase properties, especially at such short chain lengths. In the di-dodecyl-β-d-GlcDAGs, no Lc phases are evident in either the 1,2-rac or 2,3-sn GGLs and the rate of Lc phase formation in the 1,2-sn compounds can readily be monitored without resorting to either temperature jump experiments or other special conditions (Fig. 5). In the di-dodecyl-β-d-GalDAGs, Lc phases are formed rapidly in the order 1,2-sn>1,2-rac>2,3-sn (Mannock et al., unpublished experiments). In the case of the 1,2-sn-β-d-GalDAG, the rate of Lβ/Lc phase conversion is so fast that it is very difficult to obtain the Lβ phase using standard protocols. Because of experimental difficulties, initial reports of the phase behaviour described for the 1,2-sn-di-dodecyl-β-d-GalDAG identified the thermal event seen at ∼60°C as either an Lc/Lα [126]•• or an Lc/HII [120]••[133] phase transition. More recent measurements have found that the transition at this temperature is an Lc/QII phase transition, but at this time it is unclear whether the QII phase belongs to space group Pn3m or Im3m or whether a region of phase coexistence exists at ∼60°C (Mannock et al., unpublished experiments). At higher temperatures, above the Lβ/Lα phase transition, a Pn3m/HII phase transition is initially observed at 105°C and then decreases to 90–95°C on subsequent heating. On cooling, both the HII/QII phase transition and the Pn3m/Lα transition are supercooled to ∼84°C and ∼40°C, respectively. Corresponding measurements of the 1,2-rac and 2,3-sn-di-dodecyl-β-d-GalDAGs also show evidence of a region of phase coexistence above ∼60°C, followed at ∼95°C by a Pn3m/HII phase transition. On cooling, all di-dodecyl-β-d-GalDAGs exhibited a HII/Pn3m phase transition at 82–84°C, a Pn3m/Lα phase transition at ∼40°C and then transitions to either an Lc phase (1,2-sn) or an Lβ phase (1,2-rac and 2,3-sn) at 30–35°C.

In the 1,2-sn-di-tridecyl-β-d-GalDAGs (Fig. 8a), only a single Lc/HII phase transition is found on heating at ∼64°C. On cooling, three transitions are evident with decreasing temperature at ∼60°C (HII/Pn3m), ∼50°C (Pn3m/Lα) and 49°C (Lα/Lc). In the 2,3-sn and 1,2-rac-di-tridecyl-β-d-GalDAGs, the conversion to the Lc phase is slower and so on heating four phase transitions are visible in both lipids at ∼40°C (Lβ/Lα), ∼60°C (Lα/Im3m), ∼65°C (Im3m/Pn3m), and ∼70°C (Pn3m/HII) (Fig. 8b; Mannock et al., unpublished experiments). On cooling, two transitions are observed at 60–65°C (HII/Pn3m) and ∼40°C (Pn3m/Lβ). The transition to the HII phase in the 2,3-sn-di-tridecyl-β-d-GalDAG is found at a higher temperature than those of the 1,2-rac and 1,2-sn. In addition, neither the 1,2-rac (not shown) or the 2,3-sn exhibit an Lα phase on cooling, suggesting that the 1,2-sn compound and to a lesser extent the 1,2-rac compound are significantly affected by the stability of the headgroup hydration in these β-d-GalDAGs (Fig. 8; Mannock et al., unpublished experiments). As the number of heating and cooling cycles increases, so the phase transitions seen in the 1,2-sn diastereomer shift downward (Fig. 8a) as was seen in the di-dodecyl compound, whereas in the 2,3-sn diastereomer, the changes are less marked both in the chain-melting and L/NL phase transitions (Fig. 8b). It seems reasonable to suggest that the shift in Lα/QII and QII/HII phase transition temperatures observed reflects changes in the stability of these phases with changing hydration as was evident in the phase diagram of the di-dodecyl-rac-β-d-GlcDAG [125]••, partially as a result of differences in the influx and efflux of water into those non-lamellar phases [171], [172], [173].

At still longer chain lengths, the effect of differences in the chirality of the glycerol backbone on the thermotropic phase behaviour becomes less evident as the interface has a progressively smaller influence on the phase properties with increasing hydrocarbon chain length [96]••. In the di-tetradecyl-β-d-Glc- and GalDAGs (Fig. 9a), the Tms’ are all found at ∼52°C, whereas the Ths’ are found at ∼57°C (β-d-Glc) and ∼63°C (β-d-Gal), respectively, [117], [124]•[126]••[132]••. For diastereomers with either the β-d-Glc or β-d-Gal configuration of the same hydrocarbon chain length, the d-spacings of the lamellar mesophases have similar values (Fig. 9b), supporting the suggestion that these motionally averaged phases are structurally similar [126]••. On nucleation, an Lc phase is rapidly formed in both the β-d-Glc- and β-d-GalDAGs in which the hydrocarbon chains are packed either on a monoclinic [132]••, or a hybrid orthorhombic/triclinic subcell [122]••. In the β-d-GalDAGs, a second Lc phase which is not calorimetrically detected may also be formed in both the 1,2-sn and 1,2-rac lipids (but not the 2,3-sn diastereomer, or any of the didodecyl or ditridecyl-β-d-GalDAGs) above 65°C and has been found to have up to four lines in the WAX diffraction region at 4.42, 4.33, 4.20 and 4.06Å [122]••[132]••[155], [156]. The nature of the hydrocarbon chain packing of this Lc2 phase is uncertain at this time [124]•[132]••[155], [156], however, both Lc1 and Lc2 phases show differences in their long spacing peak intensities indicating subtle differences in the long-range ordering of these phases with changing interfacial packing.

On the basis of DSC and XRD measurements (Fig. 4, Fig. 8, Fig. 9), it is also evident that the L/NL phase transitions of the 2,3-sn-didodecyl- to ditetradecyl-β-d-GalDAG diastereomers are consistently found at higher temperatures than those of the corresponding β-d-GlcDAGs regardless of the interfacial chirality of the latter ([126]••[132]••[134]•• and Mannock et al., unpublished experiments). However, because of the effects of hydration on the stability of these phases in the 1,2-sn-β-d-GalDAGs, it is impossible to make a direct comparison of phase properties between the lipid diastereomers, as was done earlier for the β-d-GlcDAGs [126]••. Nevertheless, the combined observations of the faster Lβ/Lc phase conversion rates in the β-d-GalDAGs (which is often misleadingly termed the ‘galactose stabilizing effect’) and the generally higher L/NL phase transition temperatures, suggest that the β-d-Gal headgroup hydration is less stable than that of the corresponding β-d-Glc headgroups [155], [156], but that the overall headgroup cross-sectional area is greater in the β-d-Gal- than in the β-d-GlcDAGs. Indeed, it is interesting that the values of Th decrease in the order 1,2-sn>1,2-rac>2,3-sn in the didodecyl- and ditridecyl-β-d-Glc diastereomers, but the order is reversed in the β-d-Gal diastereomers. The fact that the rate of Lc phase formation in the diastereomers has a similar pattern in both the β-d-Gal- and GlcDAGs suggests a common mechanism by which the loss of water is initiated from the lipid interface. In the 1,2-sn-glycerol configuration, it seems likely that this process is initiated by the close juxtaposition and, consequently, the competition between interfacial water and the glycerol O-2 for hydrogen bonding to the ring OH-2 [126]••. This observation is supported by the fact that the corresponding α-d-ManDAGs, in which the OH-2 group is axial (Fig. 1), do not appear to form Lc phases at all [124]•[148]. Such disparities in GGL phase properties most likely originate from the differences in the structure of bound water, which is evident from measurements of their methyl glycosides [157], suggesting that differences in hydration properties also influence the values of Th.

One further example of the effects of hydration on the non-lamellar phase morphology may be the difference in the identification of the QII phase space group in the didodecyl-β-d-GlcDAG. It has been suggested that the QII phase seen in both the didodecyl-β-d-Glc and -β-d-GalDAGs at low temperature in samples containing excess water is the Im3m phase [124]•[133], [134]••. However, our unpublished work (Mannock et al.) supports the co-existence of the Im3m and Pn3m phases in the β-d-GalDAGs in excess water. Typically in lipid-water systems forming QII phases, three QII phases are often found in phase diagrams. The Ia3d (G surface) is always found at the cool, dry end, the Im3m (P surface) is usually found at the cool, wet end; and the Pn3m/Pn3 (D surface) is usually at the warm, wet end of the phase diagram. The reason for the different QII phase space group identifications is not immediately clear, but may arise from any number of chemical or physical origins. Potential chemical contributions affecting Tm and Th may arise from small amounts of chemical impurities which have not been removed during purification or which have been produced by degradation of the lipid during the course of the experiment. Potential physical contributions such as variations in the lipid/water ratio (increases in lipid content lower Th, except when Lc phases are rapidly formed) and the rate of temperature change (which may markedly affect the resolution of phases on the temperature scale as well as affecting the measurements of the phase transition temperature) have been mentioned above. Studies of PC/fatty acid mixtures with a chain length of 12 carbon atoms have found that a Pn3m phase is present in a narrow region of lipid concentration between the Ia3d and Im3m phases [152]••[153]•• on heating. Furthermore, Duesing et al. [151]• working on DDPE found that the Im3m phase existed at lower temperatures than the Pn3m phase, but in a region of phase co-existence with the Lα phase and that it required longer equilibration times to isolate the Im3m phase to permit a confident identification. Thus, the reported differences in the QII phase region of the β-d-GlcDAGs may arise from small differences in either: lipid concentration (perhaps from differences in sample preparation); the thermal history of the sample; or the equilibration times used during the experiment. On the basis of XRD measurements of the above PC/FA mixtures, Templer and co-workers have developed energetic models which consider hydrocarbon chain packing energy of inverse phases in terms of the elastic energy of extension of the hydrocarbon chains [174]• or in terms of the molecular splay [153]••, and have suggested that the conversion from one QII lattice to another may be the result of small changes in interfacial geometry [153]••. On the molecular level, a recent study of some liquid crystals has suggested that similar QII/QII phase transitions are probably caused by small differences in the order of the hydrocarbon chains relative to other, more rigid parts of the molecule [175]. One can then surmize that in the GGLs, the headgroup and interface are motionally restricted relative to the two hydrocarbon chains, as is indeed suggested by NMR [146], [147], [148] and monolayer measurements [118] and by computer modeling experiments [143]. This conclusion is broadly consistent with the glycolipid thermotropic phase behaviour summarized here.

From the presently available data, we can say that changing the headgroup or interfacial stereochemistry for a given hydrocarbon chain length in a GGL does affect the patterns of both lamellar solid-state and lamellar/non-lamellar phase polymorphism. Those observations suggest that in lipids with small headgroup areas, the chain melting phase transition temperature is not greatly affected, but the temperature of the lamellar/non-lamellar phase transition, or the nature of that transition may be significantly altered. In addition, it is primarily the hydrocarbon chains which drive the phase transitions and it is not until the headgroup and interfacial region are motionally restricted, either by loss of water to the bulk phase and the adoption of a more highly ordered molecular packing, or through an increase in the hydrocarbon chain dynamic volume with a concommitant increase in lateral compression in the polar region that the chemical structure of the headgroup exerts a ‘significant influence’ on the lipid phase properties. If this perspective is valid, then one possible advantage of the 2,3-sn-glycerol configuration in the GGLs is that under the extreme conditions of pH 2 and above 50°C, the mesophase region of such lipids are likely to remain more hydrated, while the corresponding 1,2-sn glycerol configurations might potentially dehydrate and phase separate from complex lipid mixtures under such conditions. In combination with branched and alicyclic chains, an organism utilizing the 2,3-sn-glycerol lipid configuration would be better able to maintain a fluid, hydrated membrane barrier between itself and its environment while maintaining the elasticity necessary to form highly curved phases.

The connections between structural, thermodynamic, membrane dynamics and surface force observations are still developing and their assembly into workable, coherent models is just beginning. The developing picture is one from which a more sophisticated view of the balance between the hydrocarbon chain, headgroup/interface and aqueous contributions to the lateral stresses within a lipid bilayer is gradually emerging. We hope that the observations presented here from our own work, and that of our collaborators and our colleagues will contribute to further progress in this endeavor.

Section snippets

Acknowledgements

This work was supported by operating and major equipment grants from the Canadian Institutes of Health Research and the Alberta Heritage Foundation for Medical Research. We would also like to take this opportunity to acknowledge the many contributions from our collaborators without whose excellent support most of this work would not have been possible.

References (177)

  • M.S. Webb et al.

    Comparison of different hydrophobic anchors conjugated to poly(ethylene glycol): effects on the pharmacokinetics of liposomal vincristine

    Biochim Biophys Acta, Biomembranes

    (1998)
  • B.M. Gadella et al.

    Glycolipids as potential binding sites for HIV: topology in the sperm plasma membrane in relation to the regulation of membrane fusion

    J Reprod Immunol

    (1998)
  • J.E. Parks et al.

    Lipid composition and thermotropic phase behaviour of boar, bull, stallion and rooster sperm membranes

    Cryobiology

    (1992)
  • A. Brogi et al.

    New insights into the interaction between the gp120 and the HIV receptor in human sperm

    J Reprod Immunol

    (1998)
  • Y. Tsuji et al.

    Sperm immobilizing antibodies react to the 3-O-sulfogalactosyl residue of seminolipid on human sperm

    J Reprod Immunol

    (1992)
  • K. Takahashi et al.

    Analysis of anti-rotavirus activity of extract from Stevia rebaudiana

    Antiviral Res

    (2001)
  • S. Hanashima et al.

    Synthesis of sulfoquinovosylacylglycerols inhibitors of eukaryotic DNA polymerase alpha and beta

    Bioorg Med Chem

    (2001)
  • D. Colombo et al.

    Chemoenzymatic synthesis and antitumor promoting activity of 6′- and 3-esters of 2-O-β d-glucosylglycerol

    Bioorg Med Chem

    (1999)
  • D. Colombo et al.

    Anti-tumor-promoting effects of glycoglycerolipid analogues on two-stage mouse skin carcinogenesis

    Cancer Lett

    (2000)
  • D. Colombo et al.

    Glycoglycerolipid analogues active as anti-tumor-promoters: the influence of the anomeric configuration

    Eur J Med Chem

    (2000)
  • D. Colombo et al.

    Inhibitory effect of stabilized analogues of glycoglycerolipids on Epstein-Barr virus activation and mouse skin tumor promotion

    Cancer Lett

    (2002)
  • C. Murakami et al.

    A novel DNA polymerase inhibitor and a potent apoptosis inducer: 2-mono-O-acyl-3-O-(α d-sulfoquinovosyl)-glyceride with stearic acid

    Biochim Biophys Acta: Proteins Proteomics

    (2003)
  • C. Murakami et al.

    Structure-function relationship of synthetic sulfoquinovosyl-acylglycerols as mammalian DNA polymerase inhibitors

    Arch Biochem Biophys

    (2002)
  • K. Ohta et al.

    Specific interaction between an oligosaccharide on the tumor cell surface and the novel antitumor agents sulfoquinovosylacylglycerols

    Biochem Biophys Res Commun

    (2001)
  • V. Vill et al.

    Carbohydrate liquid crystals: structure–property relationship of thermotropic and lyotropic glycolipids

    Curr Opin Colloid Interface Sci

    (2002)
  • C. Tschierske

    Amphotropic liquid crystals

    Curr Opin Colloid Interface Sci

    (2002)
  • C. Tschierske

    Liquid crystalline materials with complex mesophase morphologies

    Curr Opin Colloid Interface Sci

    (2002)
  • S. Diele

    On thermotropic cubic mesophases

    Curr Opin Colloid Interface Sci

    (2002)
  • R. Templer

    Thermodynamic and theoretical aspects of cubic mesophases in nature and biological amphiphiles

    Curr Opin Colloid Interface Sci

    (1998)
  • S.T Hyde et al.

    Novel surfactant mesostructural topologies: between lamellae and columnar (hexagonal) forms

    Curr Opin Colloid Interface Sci

    (2003)
  • M. Hato et al.

    Self-assembly of synthetic glycolipid/water systems

    Adv Colloid Interface Sci

    (1999)
  • M. Hato

    Synthetic glycolipid/water systems

    Curr Opin Colloid Interface Sci

    (2001)
  • J.N. Coupland

    Crystallization in emulsions

    Curr Opin Colloid Interface Sci

    (2002)
  • M.C. Wiener

    Existing and emergent roles for surfactants in the three-dimensional crystallization of integral membrane proteins

    Curr Opin Colloid Interface Sci

    (2001)
  • G.G. Shipley

    Bilayers and non-bilayers: structure forces and protein crystallization

    Curr Opin Struct Biol

    (2000)
  • R.N.A.H. Lewis et al.

    Membrane lipid molecular structure and polymorphism. In: Lipid Polymorphism and Membrane Properties, R.M. Epand (Ed.)

    Curr Top Membr

    (1997)
  • E. Rivas et al.

    Polymorphism of polar lipids and galactolipids of maize chloroplasts in the presence of water

    J Mol Biol

    (1969)
  • D.A. Mannock et al.

    Phase behaviour of the membrane lipids of the thermophilic blue-green alga Anacystis nidulans

    Biochim Biophys Acta

    (1985)
  • Makita A, Taniguchi N. Glycosphingolipids. In: Wiegandt H. (Eds.) ‘Glycolipids’, new comprehensive biochemistry. NY:...
  • P.W. Hemming

    In: Glycosylphosphopolyprenols

  • T.A. MacMurray et al.

    Composition of wheat-flour lipids

    J Sci Food Agric

    (1970)
  • M.R. Sahasrabudhe

    Lipid Composition of Oats (Avena sativa L.)

    J Am Oil Chem Soc

    (1979)
  • I. Ishizuka et al.

    In: Glycoglycerolipids

  • Ratledge C, Wilkinson SG, editors. Microbial lipids. New York: Academic Press,...
  • M. Kates

    Handbook of lipid research, 6: glycolipids, phosphoglycolipids and sulfoglycolipids

    (1990)
  • P.J. Quinn et al.

    The structural role of lipids in photosynthetic membranes

    Biochim Biophys Acta

    (1983)
  • G.V. Novitskaya et al.

    The composition of galactolipids of chloroplast. Membranes of hardened and non-tempered winter wheat plants

    Prikl Biokhim Mikrobiol

    (1982)
  • A.R. Wellburn

    Environmental factors and genotypic changes in lipids contribute to the winter hardiness of Norway spruce (Picea abies)

    New Phytologist

    (1997)
  • D.K. Hincha et al.

    Galactose-specific lectins protect isolated thylakoids against freeze-thaw damage

    Plant Physiol

    (1993)
  • H.P. Kreuzer et al.

    Role of α-galactosidase in osmotic regulation of Poterioochromonas malhamensis

    Planta

    (1980)
  • Cited by (33)

    • Self-assembly and emulsifying properties of biomass derived surfactant from branched alkyl chain xyloside

      2024, Colloids and Surfaces A: Physicochemical and Engineering Aspects
    View all citing articles on Scopus
    View full text