Lipid self-assembly and lectin-induced reorganization of the plasma membrane

Abstract

The plasma membrane represents an outstanding example of self-organization in biology. It plays a vital role in protecting the integrity of the cell interior and regulates meticulously the import and export of diverse substances. Its major building blocks are proteins and lipids, which self-assemble to a fluid lipid bilayer driven mainly by hydrophobic forces. Even if the plasma membrane appears—globally speaking—homogeneous at physiological temperatures, the existence of specialized nano- to micrometre-sized domains of raft-type character within cellular and synthetic membrane systems has been reported. It is hypothesized that these domains are the origin of a plethora of cellular processes, such as signalling or vesicular trafficking. This review intends to highlight the driving forces of lipid self-assembly into a bilayer membrane and the formation of small, transient domains within the plasma membrane. The mechanisms of self-assembly depend on several factors, such as the lipid composition of the membrane and the geometry of lipids. Moreover, the dynamics and organization of glycosphingolipids into nanometre-sized clusters will be discussed, also in the context of multivalent lectins, which cluster several glycosphingolipid receptor molecules and thus create an asymmetric stress between the two membrane leaflets, leading to tubular plasma membrane invaginations.

This article is part of the theme issue ‘Self-organization in cell biology’.

1. The plasma membrane

The plasma membrane physically separates the cytoplasm of living cells from the extracellular environment and hence maintains the physical integrity of the cell. It acts as a barrier that is selectively permeable to ions and organic molecules, and regulates transport processes into and out of the cell [1]. The plasma membrane is involved in a multitude of cellular processes, such as signalling and adhesion, among others [2,3]. It also helps to hold the cytoskeleton in place to preserve the cell shape [4]. With respect to all these diverse functions, it sounds amazing that the plasma membrane represents a lipid matrix with a thickness of only 4–6 nm, with embedded integral and peripheral proteins [5,6]. The principles of how membrane lipids self-assemble into cell membranes and how lipid–lipid interactions lead to the formation of small, transient domains within the membrane will be outlined in the following.

2. Self-organization of lipids in aqueous solutions

The plasma membrane consists of a complex mixture of components, among which lipids are the major building blocks. Plasma membranes contain a large number of different types of lipids that form a lipid bilayer with an inner (cytosolic) and an outer (extracellular) membrane leaflet [7]. Lipids exhibit an impressive diversity of properties and can be classified by their geometry [8], as well as by the type of assembly they form in contact with water.

Since lipids contain a hydrophilic (polar head group) and a hydrophobic part (fatty acyl chain) they are considered as amphiphilic molecules. Exposed to water, lipids compact their hydrophobic parts in the energetically most favourable way, forming a hydrophobic core [9]. This simple principle underlies the self-organization of lipids in an aqueous environment, e.g. in living cells as well as in artificial membrane systems.

As indicated by Israelachvili et al. [10,11], the lipid behaviour in aqueous solution can be easily predicted by taking into consideration the geometrical parameters of its hydrophobic and hydrophilic parts. These are v, the volume of the nonpolar part, l, the length of the nonpolar part and a₀, the optimal surface area of the polar head group. The parameters v and l can be simply determined by the lipid size and shape [11,12]. The determination of the parameter a₀ is less straightforward. The possible repulsive interactions between head groups and parameters of the aqueous environment (ionic strength, pH) have to be taken into account. This factor is particularly critical for charged lipids, e.g. phosphatidylserine (PS) or phosphatidylglycerol (PG), where the head groups have to be far enough from each other that Coulomb's repulsive forces are equilibrated by lipid–lipid interactions.

One of the potential assembly modes of lipid molecules in water is the formation of spherical micelles (figure 1) [12]. If a spherical micelle is built up of N lipid molecules, the total micelle surface area a_M and the volume of the micelle V_M can be determined as

and

Hence, the radius of the micelle R_m can be expressed as

As R_m ≤ l for spherical micelles, we obtain the following equation:

Figure 1.

Self-assembly of lipids in contact with water. Lipids are characterized by their critical packing parameter v/la₀. (a) Inverted cone-shaped lipids form micelles (micellar phase) of positive curvature. (b) Cone-shaped lipids assemble into different structures of hexagonal (H_II) phase. (c) Cylinder-shaped lipids form a lamellar phase.

The v/la₀ expression is called the critical packing parameter.

The interconnection between these three parameters determines the geometry of the lipid molecule and its behaviour in water. Using the same logic, the value of the critical packing parameter for a lipid bilayer can be obtained.

Lipids can be classified by their shapes into three major groups (figure 1)

• (a) Inverted cone-shaped lipids (v/la₀ < 1/2), which form micellar phases with positive curvature in water. This group comprises mainly lipids with a single fatty acyl chain, such as lysophospholipids. To assemble as spherical micelles, the critical packing parameter of lipids has to be below one-third, whereas lipids with 1/3 < v/la₀ < 1/2 assemble as cylindrical micelles. The relevance of this type of lipids in native cell membranes is most obvious for pore formation [13] or vesicle budding [14].

• (b) Cone-shaped lipids (v/la₀ > 1), which self-organize in so-called hexagonal micellar phases of negative curvature (in acidic environment). This group of lipids includes mainly unsaturated lipids, which participate in uptake processes where the cell membrane is bended inwards [15].

• (c) Cylinder-shaped lipids (1/2 < v/la₀ ≤ 1), which assemble into a lipid bilayer that forms a lamellar phase when assembled in stacks. These lipids that are characterized by two fatty acyl chains of different length and saturation degree represent probably the most important group of membrane lipids and form the fluid lipid matrix of the cell membrane. They can assemble into flexible lipid bilayers of liposomes (1/2 < v/la₀ < 1) as well as planar lipid bilayers (v/la₀ ≈ 1).

In addition, factors such as the ion composition of aqueous solutions, pH or temperature can alter the lipid geometric parameters, and as a result, the assembly of lipids [11].

A lipid bilayer can undergo solid–liquid phase transition, also referred to as main phase transition or chain-melting transition [16,17]. Depending on the lipid, the main phase transition temperature varies from −17°C for unsaturated lipids, like dioleoylphosphatidylcholine (DOPC), up to 37°C for sphingomyelin (SM) with long saturated acyl chains [18]. Glycolipids with sugar moieties have even higher solid–liquid phase transition temperatures [19]. Lipid phase transitions have already been intensively studied for more than 50 years [20–22] using a large variety of techniques, such as X-ray diffraction [23], calorimetry [22], atomic force microscopy (AFM) [24] or fluorescence techniques on different membrane model systems [25,26]. The length and saturation degree of the fatty acyl chain determines not only the temperature of the solid–liquid phase transition but also the degree of packing in the liquid phase. Quantitatively, this degree of packing can be described using the so-called acyl chain order parameter. Lipids with a low value of order parameter form a bilayer phase of low order and high mobility, known as liquid-disordered (Ld) phase [7]. Lipids that form Ld phases are mainly unsaturated phospholipids (figure 2). By contrast, bilayers formed of long-chain saturated lipids, like SM, adopt a solid-like phase of high order and low mobility. Addition of sterols, such as cholesterol, renders the solid-like membrane more mobile, turning it into a liquid phase. This phase is called the liquid-ordered (Lo) phase as the order still remains high. Interestingly, the addition of cholesterol enhances acyl chain order in Ld membranes [7,27,28].

Figure 2.

Lipid bilayer phases. Depending on the lipid composition, temperature and order parameters, the lamellar phase can adopt solid or fluid phases. (a) Lipid bilayer of solid phase composed of SM. (b) Lipid bilayer of fluid phases can exist as liquid-ordered (Lo) (composed of SM and cholesterol) or Ld (composed of DOPC).

A lipid bilayer can contain spatially separated domains of different lipid phases. In binary mixtures composed of lipid species of low-melting temperature (e.g. DOPC) and of high-melting temperature (e.g. SM), the bilayer exhibits spatial phase separation into solid and Ld phase domains. The temperature is the crucial factor for this process; phase separation can occur only at temperatures in between the main phase transition temperatures of the two lipids. Moreover, the coexistence of all three—Ld, Lo and solid phase—domains is possible by the addition of cholesterol [29]. The preference of cholesterol to associate with sphingolipids and partition in SM-containing Lo phases is remarkable [30–32]. Furthermore, phase separation can be reversed by increasing the lipid entropy (e.g. by increasing the temperature).

3. Higher-order organization of lipids within the plasma membrane

To describe the structure of biological membranes, Singer & Nicolson introduced the fluid mosaic model in 1972 ([33]; see also [34]). According to this model, the cell membrane is considered as a two-dimensional liquid arranged as a lipid bilayer with homogeneously distributed proteins. Lipids in a lipid bilayer are in continuous motion. In addition to translational diffusion (microseconds) and rotational movement (picoseconds–nanoseconds), they can flip from one leaflet to the other (milliseconds–seconds) or exhibit undulatory movements (milliseconds–seconds), which are induced by transmembrane proteins or cytoskeleton [35–37]. Further studies of biological membranes based on the use of detergents revealed the separation of the so-called detergent-resistant membrane (DRM) fractions, enriched in particular with SM and cholesterol, from detergent-soluble membrane fractions enriched with rather unsaturated membrane phospholipids [38]. Based on these and other important findings, the ‘lipid raft’ theory was introduced by Simons & Ikonen in 1997 [39]. According to this theory, the membrane is no longer a homogeneous fluid where all components are randomly distributed, but contains highly ordered lipid (micro-) domains stabilized by SM and cholesterol (figure 3a). Lipid rafts are also enriched with (transmembrane) proteins and glycolipids, and are thought to play a key role in cell signalling, uptake and release [40–42]. Further studies of membrane asymmetry revealed even more complicated structures of raft and non-raft membranes [43]. SM-enriched rafts form mainly in the extracellular membrane leaflet while the cytosolic leaflet contains in total less than 10% of all membrane SM molecules [44–46]. Consequently, the cytosolic leaflet is less viscous and does not sustain the formation of lipid rafts [47]. In general, the cytosolic leaflet is enriched with lipids, such as phosphatidylinositol (PI), phosphatidylethanolamine (PE) and PS. Remarkably, lipid raft-embedded glycosylphosphatidylinositol (GPI)-anchored proteins in the extracellular leaflet interdigitate with PS clusters in the cytosolic membrane leaflet [48]. Saturated phosphatidylcholine (PC) lipids are found in the extracellular membrane leaflet.

Figure 3.

From lipid rafts in native plasma membranes to Lo phases in synthetic membrane systems. (a) Composition of raft and non-raft types of native plasma membranes. The membrane is asymmetrical—the extracellular membrane leaflet is enriched with SM and saturated phosphatidylcholine (PC) lipids, while phosphatidylserine (PS), phosphatidylinositol (PI) and phosphatidyl ethanolamine (PE) lipids associate mainly to the cytosolic leaflet. (b) Synthetic membrane systems composed of liquid membrane domains to mimic native membrane phases. The Lo domain mimics a lipid raft membrane and the Ld domain mimics a non-raft membrane. In synthetic membrane systems, lipids self-assemble into a lipid bilayer mainly in a symmetric manner. GPI, glycosylphosphatidylinositol; SM, sphingomyelin.

The lipid raft theory is in line with data obtained from viral membranes. It was found that the HIV envelope is enriched in host cell sphingolipids and cholesterol, the typical lipid raft components [49]. This suggested that the assembly of HIV Gag proteins at the cytosolic leaflet of the cell membrane induces the accumulation of lipid rafts on the extracellular leaflet, so that the HIV envelope consists mainly of raft-like membrane after virus release [50]. This strengthens the significance of lipid rafts as important signalling platforms for diverse cellular processes [6,51].

The size and lifetime of lipid rafts are far below the spatial and temporal resolution of most conventional microscopy techniques, which makes a direct visualization of these domains in cell membranes nearly impossible [52]. Though impressive progress in fluorescence microscopy has been made with the development of advanced techniques that offer high temporal and spatial resolution [52–56], membrane nano-domains can hardly be visualized.

Plasma membrane organization can be also explored in less dynamic systems with compositions close to native ones. These types of membrane systems, e.g. giant plasma membrane vesicles [57] or plasma membrane sheets [58], can be extracted from living cells by cell membrane disruption.

For elucidating the role of distinct membrane components, synthetic membrane systems (figure 3b), such as unilamellar vesicles [59,60], supported [61] or pore-suspending [62] lipid bilayers are employed. The main advantage of synthetic membrane systems is the possibility to adjust the lipid and/or protein content in order to prepare membranes of different properties. These systems are widely used to study the organization of lipid bilayer phases (e.g. Lo and Ld) [63], membrane pore formation [64] and uptake processes [65]. Reconstitution of Lo domains using lipids found in DRMs (e.g. SM and cholesterol, in distinct ratios) enables mimicking some properties of lipid rafts at the microscopic scale. Various cellular processes (e.g. transport across the plasma membrane or induction of signalling events) require local reorganization of the plasma membrane. For instance, extracellular (e.g. pore-forming proteins [66–68] and cell-penetrating peptides [69,70]) and cytosolic (e.g. actin [71,72], clathrin [73], BAR domains [74,75]) components can alter the lipid distribution within the lipid bilayer.

In this review, we focus on the membrane reorganization driven by specific interactions of carbohydrate-binding proteins (so-called lectins) with glycosphingolipids, their host cell receptors.

4. Glycosphingolipids self-organize in small clusters

Glycosphingolipids (GSLs) play crucial roles in intercellular communication and uptake processes, among others [76]. The GSL content in cell membranes varies from 5% to more than 20% [77]. GSLs are mainly found in the extracellular membrane leaflet—the hydrophobic ceramide backbone being embedded in the lipophilic part of the membrane and the carbohydrate moieties being exposed to the extracellular environment [78]. The majority of GSLs are reported to be enriched in lipid rafts [5,39,79–81], where they associate with SM and cholesterol. In addition, AFM studies as well as molecular dynamics simulations have revealed the capability of GSLs to form separate clusters in lipid raft-like membranes [82–85]. Variations in the length and saturation degree of the fatty acyl chain of the ceramide backbone affect the orientation and partitioning of GSLs in the lipid monolayer [86,87], and hence the exposure of the carbohydrate group rendering it partially inaccessible for ligands [88,89]. GSL assemblies can locally modify membrane order and curvature, which is of high significance for various uptake processes. The long ceramide backbones of GSLs are also important elements for transbilayer coupling and communication since they interdigitate the two membrane leaflets [90].

5. Lectin-induced glycosphingolipid clustering reorganizes the plasma membrane

Besides their natural functions, GSLs are hijacked by various pathogens (e.g. bacteria and viruses) [91–93] and pathogenic products (e.g. toxins, lectins) [94,95]. Lectins [96] can either be soluble, prearranged as a viral capsid or attached to the outer bacterial membrane. They bind to the carbohydrate moieties of GSLs [97], either by targeting preformed GSL domains or by inducing domains by local recruitment of receptor molecules underneath their oligomeric structures [94]. Thereby, asymmetric stress between the extracellular and cytosolic membrane leaflets can lead to inward bending of the cell membrane and growth of tubular plasma membrane invaginations [94].

(a). Toxins/lectins act as clustering devices

Based on cellular as well as synthetic membrane systems, it has been demonstrated that oligomeric toxins/lectins bind to and cluster host cell GSLs, which leads to the formation of tubular plasma membrane invaginations as endocytic transport intermediates. For instance, the homopentameric B-subunit of Shiga toxin (StxB) from Shigella dysenteriae recognizes specifically the carbohydrate moieties of the GSL globotriaosylceramide (Gb3) on the extracellular membrane leaflet. Up to 15 Gb3 receptor molecules can interact with one toxin pentamer [98]. This multivalent binding leads to Gb3 clustering and membrane reorganization [99]. Thus, the entropy of the system decreases together with lipid mobility [100], while the membrane order increases locally [85,101]. StxB-induced Gb3 clustering modifies the area per lipid in the outer membrane leaflet. This creates asymmetric stress in the bilayer, and consequently a membrane tubule nucleates [102]. Such a mechanism is probably common for the cellular uptake of many oligomeric lectins [94,103,104]. Nevertheless, the mechanism of lectin clustering on the extracellular membrane surface remains debated. Different reports claim that this process is either driven by capillary effect [105], by entropy modification of the lipid bilayer hydrophobic bulk [102,106] or by local lectin-induced curvature [107]. However, Pezeshkian et al. [108] suggested that neither of these factors is critical and that multi-molecular lectin clustering is more likely related to membrane fluctuation-induced forces. Multiple carbohydrate-binding sites of lectins ensure their efficient binding to GSLs and clustering on the membrane surface to induce cellular signalling and uptake processes. In order to explore the impact of lectin valency and binding site geometry on the induction of cellular processes, distinct carbohydrate-binding sites of lectins have been modified by site-directed mutagenesis [94,99,103,109], and tailor-made neolectins have been designed [110]. In comparison to the natural Ralstonia solanacearum lectin (RSL), which is trimeric and hexavalent, a divalent, monomeric neolectin, which is highly similar in structure, was sufficient to induce tubular membrane invaginations on giant unilamellar vesicles by binding to a fucosylated glycolipid receptor. Amazingly, only the divalent neolectin with two carbohydrate-binding sites nearby was able to form membrane invaginations, whereas divalent neolectins with more distant binding sites did not, even though their binding constants were quite similar [110].

(b). Toxins/lectins generate membrane curvature

Membrane curvature changes are the response to external or internal perturbations of the membrane, such as osmotic effects [111], membrane fusion [112], vesicle budding [113] or endocytosis [113–115]. For instance, lectin–GSL interactions induce spontaneous curvature leading to tubular plasma membrane invaginations, which are energetically favourable. The efficiency of curvature induction and tubule formation depends on membrane rigidity and composition. In the case of membranes with low rigidity, small protein clusters can initiate tubule nucleation, while in the case of membranes with higher rigidity larger protein clusters are required to bend the membrane [116,117]. Once initiated, the endocytic membrane tubule grows and sequesters lectin molecules on its negatively bended surface. With increasing lectin concentrations inside the tubule, a highly curved bud forms [114] that finally undergoes scission. In living cells, bud formation and scission can be significantly facilitated by the assembly of clathrin, actin [115] or BAR proteins at the cytosolic leaflet of the membrane [75,118].

6. Lectin- and curvature-driven lipid sorting as a membrane reorganization process

Lipid sorting was reported for multicomponent lipid mixtures [119], where an interplay between membrane entropic forces and lipid–lipid interactions takes place. In such systems, like ternary mixtures close to phase separation [119,120], lipid sorting can be triggered by a change in temperature or lipid composition. When an oligomeric lectin binds to GSLs, it acts like a clustering device (figure 4a,b) that can activate lipid sorting and induce phase separation in otherwise homogeneously mixed membranes [101,120–123]. Local asymmetric stress imposed by lectin–GSL interactions and subsequent membrane deformation can change locally the lipid organization of the bilayer. Pinot et al. [124] highlighted the importance of polyunsaturated, cone-shaped phospholipids in those uptake processes. Such lipids can adapt their conformation to the membrane curvature and may support endocytosis [65,124]. Besides, curvature-driven lipid sorting may trigger the redistribution of membrane lipids based on their molecular geometry. However, numerous studies [119,125–129] demonstrated that such curvature-driven lipid sorting is effective only in the presence of membrane-bound or transmembrane proteins, otherwise entropy homogenizes the lipid distribution even in highly curved membranes. The lipid composition of plasma membrane tubules is still elusive. Pulling a membrane tether from giant unilamellar vesicles using a micropipette [119,120,130] showed that tethers are enriched with unsaturated lipids, strongly arguing for their implication in tubule formation [130]. However, as GSL clusters are associated with SM, lectin-induced tubules may also contain saturated sphingolipids [101,130]. Hence, successful tubule formation should involve an interplay between GSL clustering and attraction of saturated lipids (figure 4c).

Figure 4.

Lectin- and curvature-driven lipid sorting. (a) Lectins recognize GSL receptors and bind to the membrane surface. (b) Lectins cross-link the GSL-enriched clusters. These clusters induce an asymmetric stress in the membrane, which leads to tubule nucleation and plasma membrane invagination. (c) Lectin-induced plasma membrane tubules are enriched with sphingolipids due to their association with GSL clusters.

7. Conclusion

The plasma membrane is an exciting example of self-organization in nature. Although many discoveries have been made concerning lipid self-assembly into a bilayer and the organization of distinct lipids into functional units within the plasma membrane, the exact membrane organization and its dynamics are still ill-defined. Synthetic membrane models are helpful tools in mimicking, and hence understanding plasma membrane organization and dynamics, but one should not forget that the plasma membrane is a highly complex, asymmetric bilayer system consisting of a huge variety of lipid species that fulfil specific functions. Further improvements in microscopy techniques, in particular in super resolution live cell microscopy, as well as in the development of novel lipid labelling and detection methods are required to better decipher the principles of dynamic membrane organization.

Data accessibility

This article has no additional data.

Competing interests

We declare we have no competing interests.

Funding

The position of T.S. was funded by the German Research Foundation grant [IRTG 1642 - SoftMatter Science: Concepts for the design of functional materials].