The multiple faces of self-assembled lipidic systems
Lipids, the building blocks of cells, common to every living organisms, have the propensity to self-assemble into well-defined structures over short and long-range spatial scales. The driving forces have their roots mainly in the hydrophobic effect and electrostatic interactions. Membranes in lamellar phase are ubiquitous in cellular compartments and can phase-separate upon mixing lipids in different liquid-crystalline states. Hexagonal phases and especially cubic phases can be synthesized and observed in vivo as well. Membrane often closes up into a vesicle whose shape is determined by the interplay of curvature, area difference elasticity and line tension energies, and can adopt the form of a sphere, a tube, a prolate, a starfish and many more. Complexes made of lipids and polyelectrolytes or inorganic materials exhibit a rich diversity of structural morphologies due to additional interactions which become increasingly hard to track without the aid of suitable computer models. From the plasma membrane of archaebacteria to gene delivery, self-assembled lipidic systems have left their mark in cell biology and nanobiotechnology; however, the underlying physics is yet to be fully unraveled.
PACS Codes: 87.14.Cc, 82.70.Uv
KeywordsLipid Bilayer Phase Transition Temperature Cationic Lipid Packing Parameter Giant Vesicle
Lipids are the building blocks of cellular compartments. By self-assembling into bilayers, they form fluid membranes that act as relatively impermeable barriers to the passage of most water-soluble molecules. Lipid membranes enclose the cell machinery and protect it from the extracellular environment . They likewise maintain the characteristic differences between the contents of each compartments and the cytosol. They accommodate a number of specialized molecules performing crucial functions to the life of the cell: ion channels pumping protons across the plasma membrane , nuclear pore complexes controlling access to and from the nucleus [3, 4], or rotary motors synthesizing ATP [5, 6]. Several of membrane proteins and glycosphingolipids are used as receptors by viruses and pathogens, including the Alzheimer's associated amyloid peptide [7, 8, 9].
Lipids share with other amphiphilic molecules the ability to self-assemble in solution into more or less complex aggregates, provided their density exceeds a certain critical micellization concentration (cmc) which depends upon their chemical structure and the ions present [10, 11]. A typical cmc value for bilayer-forming lipids ranges from 10-10 to 10-6 M while micelle-forming lipids require 10-5 up to 10-2 M in the bulk solution . The traditional view of the aggregation of amphiphilic molecules is based on the poor solubility of hydrocarbons in water, leading to what is known as the hydrophobic effect . The presence of hydrocarbon residues induces the formation of a cavity in the water structure which causes an increased degree of order and consequently a significant decrease in the entropy of water [14, 15]. When hydrocarbon residues meet upon an effective long-range attractive force [16, 17], the cavities fuse with one another and expel water from the interface releasing entropy to the solution. This leads to the spontaneous formation of stable aggregates .
The hydrophilic headgroup – although not driving the aggregation – is responsible for the formation of an interface with water, and contributes to determine, in principle, the size and the shape of the aggregates through the interactions between the molecules. Simple geometric packing considerations allow the prediction of the final aggregate conformation given some elementary structural information on the amphiphilic molecules . For this purpose, a geometric factor can be conveniently used, the dimensionless packing parameter p, defined as p ≡ v/a0l c where v is the hydrocarbon volume, a0 the optimal headgroup area, and l c the critical chain length beyond which the hydrocarbon chain can no longer be considered as fluid . This parameter determines whether the amphiphiles will form spherical micelles (p < 1/3), non-spherical micelles (1/3 <p < 1/2), vesicles or bilayers (1/2 <p < 1), or inverted structures (p > 1). This heuristic picture holds as long as only one amphiphilic component enters the system. Otherwise, the interactions between components – electrostatic interactions, van der Waals forces, or hydrogen bonding – may reorganize the system following a complex phase diagram. For example, the mixing, in the absence of added salt, of cationic and anionic surfactants with different packing parameters, yields a segregation of the amphiphiles and gives rise to unexpected aggregates such as nanodiscs, punctured planes, and facetted icosahedra, depending on stoichiometry [20, 21, 22, 23].
Due to their natural occurrence in living organisms, lipids, and the assemblies they generate, are of special interest not only for the understanding of the many biological functions they are involved in, but also in regard of their applications as biocompatible carriers of drug and gene for pharmaceutical and biomedical purposes [24, 25]. Another reason for this interest lies in a high potential in material science and nanobiotechnology, for example, by constructing intricate nanoscale networks of enzymatic reactors [26, 27], or by arranging inorganic materials with the liquid-crystalline regularity of lipid complexes used as templates .
This article gives an overview of the various structures and arrangements based on lipids which have attracted the attention of biophysicists in the last few years. The functions of particular lipid structures within the cell are presented and the applications in therapeutic treatments or nanobiotechnology are mentioned whenever applicable. Emphasis is also given to the underlying physics that governs self-assembly processes and vesicle formation. The review begins with lamellar membranes along with a discussion on the phase separation occurring in raft microdomains. Afterwards, the varying forms of non-lamellar phases are described. The long-range (> 20 nm) organized structures come next, including liposomes, exotic vesicles and tubular objects. The last section encompasses complexed systems where lipids are associated with other entities, namely biological polyelectrolytes and inorganic materials. The review ends with a short section which highlights the benefits given by computer simulations in complementing experimental data to visualize and to understand the mesoscale structure of self-assembled lipidic systems.
2. Various aspects of lipid membranes
2.1 Lipid bilayer and lamellar phase
Bilayers are certainly the most common structure formed by lipids as they are present in every cellular organisms. They can take various shapes within the cell: fairly flat in the plasma membrane, spherical and tubular for the components involved in vesicular transport, or with an intricate geometry in the endoplasmic reticulum and Golgi apparatus. In this section, we focus on the short-range (over a few nanometers) organization of lipid bilayers.
Pure lipid bilayers are fluid at high temperatures but undergo a phase transition when the temperature decreases below a critical value . The phase transition temperature is -2°C for POPC depicted in Figure 1B. According to its state, a lipid bilayer is said to be: in L α liquid-crystalline phase when it is fluid with melted hydrocarbon chains; in L β gel phase below the phase transition temperature; in Open image in new window tilted phase when the gel phase tilts relative to the layer normal; and in Open image in new window phase for tilted phase distorted by a periodic asymmetric ripple with a wavelength of the order of 100 Å [37, 38]. The fluidity of lipid bilayer allows the membrane to reorganize spontaneously over a short time upon external stimulation: for instance, in response to an intense external electric field, biological membranes form submicrometer pores provided their transmembrane potential exceeds a critical breakdown value [39, 40]. With no longer electrical stimulation, the pores reseal over a period ranging from milliseconds to a few seconds depending on the membrane dynamics. This technique, known as electroporation , is used to inject plasmid DNA across the plasma membrane of cells.
Notice that several bilayers can pile up with a thin layer of water solution separating each of them; such a structure is referred to as lamellar phase, denoted L α when the bilayers are fluid (Figure 1D). They are quite common with phosphatidylcholine (PC) lipids .
2.2 Phase separation and raft microdomains
A mixture of lipids in different phases – L α and L β , or liquid-disordered and liquid-ordered phases  for example – can phase-separate and give rise to the formation of raft microdomains in the bilayer. Each of the microdomains is enriched with lipids in the same liquid-crystalline phase. The size of microdomains ranges typically from a few nanometers to a few micrometers. Based on the properties of lipids in liposome membranes, domain models have long been proposed for native cell membranes [43, 44]. The lateral segregation of lipids is believed to play a crucial role as it may govern a number of fundamental cellular processes such as signal transduction and inter and intracellular trafficking [45, 46, 47, 48]. The self-organization into distinct domains permits the concentration of raft-associated specific receptors of proteins, which promotes the uptake of these proteins via the endocytic pathway. For example, a peptide sequence common to the Alzheimer's disease-associated Aβ peptide, the HIV-1 gp120 glycoprotein and the Prion protein was found to bind preferentially to raft-associated glycosphingolipids [7, 49, 50]. Such a peptide conjugated with a fluorophore constitutes a good raft marker for live cell imaging .
Whether such idealized situations are transposable to live cells is still lively debated. Experiments on native lipid mixtures extracted from pulmonary membranes have shown the separation of two fluid phases , but the direct observation on live cell remains almost impossible due to the presence of anchored proteins and receptors that cover the membrane surface. Moreover, it seems that different experimental methods probe their own typical available length scales and therefore result in biased data. In an interesting computer study, Yethiraj and Weisshar  modeled a binary lipid mixture by using an Ising model on a square lattice comprising obstacles that mimic proteins anchored to the cytoskeleton. They reported that even at 5–10% by area of protein obstacles, the phase separation of lipids was dramatically reduced. This finding might bring the size of possible raft microdomains in live cell down to a few nanometers at best. However, another recent study reported that at physiological temperature, raft microdomains in the plasma membrane of an epidermoid carcinoma cell line coalesce upon the binding of cholera toxin B subunit to raft-associated ganglioside GM1, leading to the formation of raft clusters of a few micrometers in size .
2.3 Non-lamellar membrane structures
Lipids with packing parameter p ~1 form preferentially bilayers, or more generally, a lamellar phase made of bilayer sheets. For other classes of lipids and mixtures of lipids, the three-dimensional polymorphism can be quite diverse, accompanied by a complex phase diagram depending on temperature, pressure, molecular structure and concentration of components . The pioneering work in this field was carried out by Luzzati and coworkers who studied lipid-water systems by x-ray scattering techniques and found a number of non-lamellar liquid-crystalline arrangements which can be categorized into hexagonal and cubic phases [62, 63, 64].
There are many evidences that lipid membranes in cubic phase are ubiquitous in the biological world. They have been observed in the plasma membrane of archaebacteria , as well as in the endoplasmic reticulum and mitochondria of mammalian cells . In structural biology, lipid cubic phases can be employed as matrices to crystallize membrane proteins enabling diffraction and thereby reconstruction with a high resolution [81, 82, 83].
3. Vesicular and tubular shapes
Submicrometer liposomes can be obtained with a narrow size distribution. Given that their membrane is biocompatible and impermeable to hydrophilic molecules, they can be conveniently used as nanocapsules. Consequently, submicrometer liposomes have attracted a strong interest in the biomedical and pharmaceutical sectors for their applications in drug delivery [24, 88, 89, 90]. Liposomes are not just merely passive capsules transferring drugs into cells, their membrane can be engineered, for example so as to release the cargo inside a low pH environment such as in the endosome . Many kinds of site-specific ligands such as antibodies, receptors and peptides can be anchored to the membrane, directing the cargo to designated cell types . The grafting of poly(ethylene glycol) (PEG) at the surface of a liposome carrier enables an extended circulation lifetime in the body . Other applications include the use of liposomes as marker for ultra sensitive detection of biological toxins [94, 95], or in binding assays of peptides to membrane receptors [51, 96].
Giant vesicles occupy a privileged place in biophysics because their micrometer size allows a direct observation under optical microscope. They can be conveniently used as model of cell membrane for investigating biological processes in controlled environment, as well as for bioanalytical purposes [27, 97]. The electrically-induced fusion of giant vesicles gives insights into the response of biological membranes to electric fields; it reveals for instance the existence of a threshold intensity related to the critical transmembrane potential [39, 98, 99, 100, 101]. The activity of particular ion channels embedded into giant liposomes can be recorded via patch-clamp methods . Giant vesicles also constitute a good biomimetic environment for monitoring enzymatic reactions [103, 104], and more fundamentally, they can be envisioned as a minimal system for constructing an artificial cell assembly expressing genes [105, 106, 107].
3.2 Exotic vesicles
The shape of lipid vesicles at equilibrium is not limited to a sphere. A large variety of shape deformations on giant vesicles are achievable by changing the external constraints on the membrane, namely the osmotic pressure difference between the interior and exterior of the vesicle  and the temperature . The equilibrium shape of the vesicle can be fully determined by the area difference elasticity (ADE) model [110, 111], which implies an additional term to the curvature energy of the membrane. This energetic term arises from the deviation in the total area difference between the inner and outer leaflets. Minimizing the thus-obtained free energy leads to the final shape. Complex two-dimensional phase diagrams can be numerically calculated, and the morphology of vesicles is then obtained as a function of a dimensionless measure of the volume-to-area ratio and of the intrinsic area difference which indicates the preferred curvature of vesicles .
In the case where the membrane of vesicles experiences a phase separation into raft microdomains, a term arising from the line tension at the phase boundary must be added to the free energy . It results in complicated morphologies where the domains impose locally their preferred curvature and generate budding portions on the surface of vesicles [116, 117, 118]. Two photon fluorescence microscopy on giant vesicles provides a direct way to visualize lipid domains labeled with distinct fluorescent dyes. It gives information about the deformations induced on the vesicles (Figure 7D) [119, 120, 121, 122] and allows to evaluate quantitatively the dynamics of raft microdomains [114, 123, 124].
3.3 Lipid nanotubes
Electron microscopy has revealed the existence of tubulo-vesicular elements interposed between the endoplasmic reticulum and the Golgi apparatus in pancreatic rat cells . It was suggested that these lipid nanotubes, abundant around the Golgi complex, interconnect adjacent Golgi elements and are involved in the transport of membrane outward along microtubules . The directed transport of small membrane blebs along a lipid nanotube has been observed in red blood cells as well , supporting the idea of the general interconnection of cellular compartments by lipid nanotubes.
Lipid nanotubes consist of multiple lipid bilayers rolled up in a long cylinder [128, 129]. Their inner diameter ranges from ~10 nm with synthetic lipids to hundreds of nanometers for natural phospholipids, and their length can reach up to several centimeters .
Because most phospholipids do not self-assemble into tubular shapes upon simple dispersion, phospholipid nanotubes must be obtained for example by pulling on the membrane of immobilized giant vesicles with a micropipette [131, 132]. In doing so, complex tubulo-vesicular networks in which the transport of specific molecules between compartments is assured by controlled diffusion can be constructed in view of bioanalytical applications [27, 132, 133]. In other protocols, the lipid tube growth is induced by a fluid flow guided in microfluidic channels [130, 134], or by the binding of steptavidin to biotinylated membranes .
4. Lipid-based complexes
Lipids can be complexed with virtually any materials provided that the electrostatic interactions are favorable. It is therefore impossible to review all the existing structures in a comprehensive manner. We will limit ourselves to the systems that have been extensively studied or that present a particular interest to biophysicists.
4.1 Lipid-DNA complexes or lipoplexes
Mesoscale assemblies made of lipids and DNA are perhaps the most documented self-assembled lipid-based complexes because they are a good case study of the intermolecular interactions between lipids and polyelectrolytes, and most importantly because they hold great promises for the future of gene therapy and protein delivery into cells [142, 143, 144, 145, 146, 147]. Lipid-DNA complexes, also called lipoplexes, were first introduced some 20 years ago by mixing cationic liposomes with DNA, and allowed the effective transfer and expression of genes in cultured cells . The encapsulation of DNA was by far more efficient than previous techniques involving liposomes because the cationic charge of the synthetic lipids enabled a 100-%-efficiency association with the negatively-charged DNA.
Cationic lipids scarcely occur in cell membrane, they are only found in tiny amounts in neuronal tissues as cationic glycosphingolipids for instance . As a result, the injection of synthetic cationic lipids into cells induces a number of toxic effects, often lethal, the more so as the lipid-to-DNA charge ratio of lipoplexes increases . To address this issue, non-cationic phospholipids have been used in association with multivalent cations. By binding to the lipid headgroup, multivalent cations are able to turn the overall headgroup charge positive , making the complexation with negatively-charged DNA electrostatically favorable. In doing so, the usual complexed liquid-crystalline structures are recovered, namely lamellar [160, 161, 162] and inverted hexagonal [163, 164], the cations being intercalated so as to bridge the phospholipid headgroups and the DNA rods. Such systems have been proven as efficient as cationic lipids to transfer genes in cultured cells depending on the concentration and the valence of cations . Monte Carlo calculations have shown that phospholipid-DNA complexes are the more stable in terms of free energy as the cation valence increases but this stabilization saturates beyond the value +4 .
The efficacy of gene transfection into cells depends upon a large number of variables and to date no clear picture has been drawn relatively to the requirements for an optimal delivery of genes. It is commonly admitted that lipoplexes are internalized by endocytosis after binding to the negatively-charged cell surface thanks to their cationic charge . The charge may also play a role in promoting the fusion necessary to escape the endosome. However, at high lipid-to-DNA charge ratio, the DNA may be so strongly coupled to the lipids that it cannot be released toward the nucleus . The liquid-crystalline structure appears to be critical for an efficient release of DNA. A good configuration seems to start from a stable lamellar Open image in new window lipoplex, which turns into a non-lamellar – possibly non-complexed hexagonal or cubic phase – upon mixing with the anionic lipids of the endosomal membrane [168, 169].
4.2 Other lipid-polyelectrolyte complexes
Mixtures of cationic and neutral lipids that yield membranes in lamellar phase have been used in association with negatively-charged filamentous bacteriophage M13 virus and cytoskeletal filamentous actin (F-actin). The two polyelectrolytes have similar diameters, ~6.5 nm for the former and 7.5 nm for the latter, but different surface charge densities, 1 e-/256 Å 2 and 1 e-/625 Å 2 respectively. Like DNA, M13 virus and lipids form a complexed lamellar phase Open image in new window with an inter-M13 spacing of 8.2 nm, slightly larger than the diameter of the M13 virus . In contrast, F-actin and lipids result in the formation of ribbon-like nanotube structures with an average width of 250 nm and length up to 100 μm, consisting of lipid bilayers sandwiched between two layers of actin . This difference of structure is attributed to the charge-density-matching mechanism: because the F-actin lattice of low charge density cannot compensate the charge density of the lipid membrane (1 e+/251 Å 2), the system self-assemble into a superlattice structure where one layer membrane is matched against two layers of F-actin.
Another unconventional complexed lamellar structure is produced with poly-L-glutamic acid (PGA) polypeptides. Small angle x-ray scattering data revealed a "pinched lamellar" structure where anionic PGA and cationic lipids formed localized pinched regions; in between, the adjacent quasi-neutral bilayers swelled into large pockets of water stabilized by hydration repulsion .
4.3 Association with inorganic materials
Lipid bilayers immobilized on solid supports have become very popular for mimicking the basic processes occurring on a real cell membrane (see the section dedicated to raft microdomains) and for biotechnological applications [176, 177]. A number of coupling techniques have been developed over the past decades including polymer-cushioned lipid bilayers [178, 179], hybrid bilayers , tethered lipid bilayers  and physically self-assembled lipid monolayers , with the possibility to pattern the membranes on the micron scale by using photolithography . The simplest route though is by the spreading of small lipid vesicles on hydrophilic substrates , employing if necessary divalent cations to bridge the like charges of lipids and substrate [185, 186].
5. Perspectives on computer simulations
Most of the studies described above relied on experimental methodologies to get structural information about the system of interest, often indirectly. Electron microscopy, x-ray scattering, atomic force microscopy, all these techniques give only certain features of the structure – symmetry or periodicity -, and must be supplemented with careful interpretations to reconstruct the detailed arrangement. The prediction of the final structure for a given system is challenging because a huge number of molecular interactions usually come into play. With the refinement of statistical mechanics models and the increasing rapidity of modern computers, fine structural calculations and dynamic over long time scale become accessible, for system complexity up to a limited extent though. We shall give a few words about the possibilities offered by computational techniques to self-assembled lipidic systems.
A lot of the underlying physics can be obtained by phenomenological Hamiltonians which conceive of the lipid systems as an assembly of thin interfaces characterized by their elastic constants. This description permits to deal with large systems, considering the collective behavior of lipid molecules possibly in interaction with polyelectrolytes. It is a very convenient approach to predict the equilibrium shape adopted by exotic vesicles . We can also calculate the complete phase diagram of cationic lipid-DNA complexes as a function of the lipid composition and the lipid-to-DNA charge ratio [194, 195]. This method, easy to implement numerically, yet requires an a priori knowledge of the system and of its behavior through the choice of suitable parameters. Density-Functional Theory (DFT) proceeds in a similar way, namely by assuming that the organized structures satisfy a local minimum of the free energy, this latter being represented in terms of molecular density-functionals [196, 197]. Based on coarse-grained models of lipid molecules, DFT is able to reconstruct the phase diagram of lipid bilayers, predicting the transition from dilute bilayers to lamellar phase . Applied to self-assembled systems, DFT is still in its infancy but holds many promises as it provides a rather fine structural description at a low computational cost.
The ultimate refinement in molecular simulation is achieved by atomistic models in which the molecular structure and the interactions of components are described faithfully, including chemical bonds (bond, angle and dihedral potentials), electrostatic and van der Waals interactions . Extremely accurate, they cannot, however, deal comfortably with mesoscale systems (extending beyond 10 nm) or track most of the self-assembly processes taking place beyond microseconds given the insufficient power of nowadays computers. Atomistic models have been up to now well suited for investigating the atomic interactions between lipids and proteins , lipids and DNA , or the configuration of lipids sticking around carbon nanotube , from a pre-assembled system close to its equilibrium, but they might reveal themselves as the method of choice for unraveling the full dynamic of self-assembly processes at atomic scale as soon as the computer technology will allow it.
If Biology finds naturally more interest in active compounds, that is, proteins and enzymes with dynamic and vital functions to the cell, Physics still remains intrigued and inquiring about the fundamental principles that drive the lipid self-organization into mesoscale structures covering three orders of magnitude in the nanometer range. Not only lipids may tell us about the spark that gave birth to primitive living organisms – Life is after all, a self-assembly process -, but also, by harnessing the building blocks of Life, we may be able to mimic, or even trick, Nature, and design Life-like systems performing specific tasks in a better way. No matter what the applications may be, protein crystallization or gene delivery, it is a safe bet that understanding self-assembled lipidic systems will continue to enrich the biological and medical research.
Several figures were prepared with the POV-Ray raytracer package http://www.povray.org. This study was supported by the Agency for Science, Technology and Research (A*STAR) in Singapore.
- 1.Alberts B, Johnson A, Lewis J, Raff M, Roberts K, Walter P: Molecular biology of the cell. Garland Science, New York. 2002Google Scholar
- 10.Hunter RJ: Foundations of colloid science. 2001, Oxford University Press, New YorkGoogle Scholar
- 12.Israelachvili J: Intermolecular and surface forces. 1991, Academic PressGoogle Scholar
- 13.Tanford C: The hydrophobic effect. Formation of micelles and biological membranes. 1980, Wiley, New YorkGoogle Scholar
- 31.Helfrich W: Z Naturforsch [C]. 1973, 28: 693-703.Google Scholar
- 43.Jain MK, White HB: Adv Lipid Res. 1977, 15: 1-60.Google Scholar
- 44.Klausner RD, Kleinfeld AM, Hoover RL, Karnovsky MJ: J Biol Chem. 1980, 255: 1286-1295.Google Scholar
- 52.Henderson RM, Edwardson JM, Geisse NA, Saslowsky DE: Lipid rafts: feeling is believing. News Physiol Sci. 2004, 19: 39-43.Google Scholar
- 56.Lin W, Blanchette CD, Ratto TV, Longo ML: Lipid domains in supported lipid bilayer for atomic force microscopy. Methods in membrane lipids. Edited by: Dopico AM. 2007, Humana Press, Totowa, 503-513.Google Scholar
- 85.Lasic DD: Giant vesicles: a historical introduction. Giant vesicles. Edited by: Luisi PL, Walde P. 2000, John Wiley & Sons, Chichester, 11-24.Google Scholar
- 87.Torchilin VP, Weissig V: Liposomes. 2003, Oxford University Press, OxfordGoogle Scholar
- 110.Miao L, Seifert U, Wortis M, Döbereiner H: Phys Rev E Stat Phys Plasmas Fluids Relat Interdiscip Topics. 1994, 49: 5389-5407.Google Scholar
- 115.Jülicher F, Lipowsky R: Phys Rev E Stat Phys Plasmas Fluids Relat Interdiscip Topics. 1996, 53: 2670-2683.Google Scholar
- 125.Sesso A, de Faria FP, Iwamura ES, Corrêa H: J Cell Sci. 1994, 107 (Pt 3): 517-528.Google Scholar
- 144.Gardlík R, Pálffy R, Hodosy J, Lukács J, Turna J, Celec P: Med Sci Monit. 2005, 11: RA110-21.Google Scholar
- 159.Tatulian SA: Surface electrostatics of biological membranes and ion binding. Surface chemistry and electrochemistry of membranes. Edited by: Sørensen TS. 1999, Marcel Dekker, New York, 872-922.Google Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.