Biology and physics of membranes

Get Complete Project Material File(s) Now! »

The Lipid bilayer

Eukaryotic cells are divided into compartments. These compartments, as well as cells themselves, are delimited by membranes. In the middle of the 20th century, the membrane was described as a two dimensional fluid structure (the lipid bilayer) which led to the fluid mosaic model (Singer and Nicolson, 1972). In this model, the membrane is a fluid matrix of lipids in which peripheral and integral proteins diffuse. Since then, this model has been refined and it appears that lipids are not immobile structures holding proteins together. Membranes are crowded and heterogeneous environments with lipids and proteins diffusing laterally allowing the formation of regions which vary in thickness and composition (Engelman, 2005).

From a lipid molecule to a bilayer

The basic components of membranes are lipids. Lipid molecules are amphiphilic and are thus composed of a hydrophilic part, the head, and a hydrophobic part, the tail. The hydrophilic head of the lipid defines the lipid type and can be neutral or charged. The hydrophobic tail is most often made of two aliphatic chains (can vary from 1 to 4 chains) of various length and degree of unsaturation. Due to their amphiphilic structure, lipids have the property to self-assemble in such a way that the heads are accessible to the solvent, whereas the tails (more hydrophobic) are buried in the core of the membrane. This spontaneous rearrangement of the lipids is of entropic origin and the result of a competition between the hydrophobic attraction (also called hydrophobic effect), which tends to aggregate the molecules together therefore reducing the interfacial area; and the repulsion of the hydrophilic head groups which tends to increase the interfacial area (Israelachvili, 1992).
This competition between hydrophobic attraction and headgroup repulsion results in a constant area per lipid. Depending on this parameter as well as on the geometrical shapes of lipids, lipid assemblies will exhibit different morphologies (Figure 1.1). Most common lipid assemblies in water are micelles (also called hexagonal phase) (Figure 1.1A) and bilayers (lamellar phase) (Figure 1.1B), which are respectively mainly composed of inverted cone-shaped and cylinder-shaped lipids. Micelles are globular structures where the lipid head groups form a spherical shell protecting the lipid tails (Figure 1.1A). Bilayers are composed of two monolayers in which the lipids are parallel to each other and the tails of each monolayer are facing each other in the core of the membrane (Figure 1.1B). On average bilayers have a thickness of around 5 nm with an area per lipid of around 0.7 nm2 (this will of course depend on the type of lipid). Lipids with a conical shape have the possibility to self-assemble into an inverted micelle structure (also called inverted hexagonal phase) (Figure 1.1C).
Figure 1.1: Lipids can self-assemble into different structures depending on their molecular shapes. (A) Inverted-conical lipids will tend to self-assemble into micelles, (B) cylinder-shaped lipids into bilayers and (C) cone-shaped lipids into inverted micelles. *Cardiolipins (or Diphosphatidylglycerol) contain 4 acyl chains and consequently display a high conical shape.
It is important to mention that the shape (also called packing parameter) of a particular lipid is not fixed. External parameters such as hydration, temperature and pH can modulate the headgroup effective area as well as the apparent chain volume (Pomorski et al., 2014).

Different classes of lipids

The most abundant membrane lipids in eukaryotic cells are phospholipids. They are usually composed of two fatty acids (their hydrophobic tails) which are linked to the headgroup, made of a phosphate group and another group such as choline. The two major classes of phospholipids are glycerophospholipids and sphingolipids (Figure 1.2).
Figure 1.2: Membrane lipids. Phospholipids comprise two subclasses of lipids: glycerophospholipids containing a diacylglyerol backbone and sphingolipids which contain a ceramide backbone. These subclasses are further divided into lipid species that display different hedgroups. Here we show the four types of glycerophospholipids (Phosphatidylcholine, Phosphatidylethanolamine, Phosphatidylserine, Phosphatidylinositol), and one type of sphingolipid (Sphingomyelin) used in the study. Glucosylceramide contains a ceramide backbone and is the key precursor for most Glycosphingolipids. Cholesterol is the major sterol found in eukaryotic cells. Of note glycerophospholipids are mostly composed of unsaturated acyl chains whereas sphingolipids mostly exhibit saturated tails. Adapted from (van Meer et al., 2008).
Polar lipids from the glycerophospholipid family are the main eukaryotic membrane lipids. They are based on a diacylglycerol backbone supplemented by a phosphate group (cone-shaped phosphatidic acid, PA) (Figure 1.2). Phosphatidylcholine (PC), the major species of this family is formed by the addition of a choline on top of the PA. PC lipids are zwitterionic (contain both a positive and a negative charge at physiological conditions) and usually have a rather cylindrical shape. Other groups can be added instead of the choline to form a phosphatidylethanolamine (PE), a species with a conical shape due to the small size of its headgroup. Phosphatidylserine (PS) is a negatively charged and cylinder-shaped phospholipid mainly concentrated in the inner (cytosolic) leaflet of the plasma membrane. Phosphatidylinositol lipids (PI) are present in smaller amounts and bear a negative charge. PI lipids are known to be phosphorylated by several different kinases and their derivatives are known to be involved in a multitude of signaling processes, mainly at the plasma membrane and at the endocytic compartments (Di Paolo and De Camilli, 2006; van Meer et al., 2008).
Sphingolipids have a ceramide (Cer) backbone (sphingosine base amid-linked to a fatty acid) (Figure 1.2). Sphingomyelin (SM) is the most abundant species of this family of lipids, and is composed of a phosphate-choline headgroup. Another important class of ceramide based lipids is the Glycosphingolipids (GSLs), consisting of a ceramide molecule attached to monosaccharides or polysaccharides. SM and GSL are found on the non-cytosolic (extracellular or luminal) leaflet of the plasma membrane (van Meer et al., 2008).
Sterols constitute another major class of lipids present in cellular membranes. The presence of the OH group on the lipid headgroup suggests that they are slightly polar and their specific structure suggests that they are non-bilayer forming molecules (even though they are able to insert in membranes). Cholesterol (Figure 1.2) is the major sterol present in mammalian cells and its level increases from the ER to the plasma membrane (Ikonen, 2008; Mesmin and Maxfield, 2009).

Lipid synthesis and distribution in cells

Lipids are not homogeneously distributed across cellular membranes (Figure 1.3). Additionally, the two leaflets of the Golgi, endosomal and plasma membranes all exhibit asymmetric lipid compositions (Devaux, 1991; Verkleij et al., 1973; Wood et al., 2011).
The endoplasmic reticulum (ER) is the main lipid and protein biosynthetic organelle (Bell et al., 1981). The presence of ribosome complexes, which gives its rough aspect to the ER, is responsible for protein synthesis. This compartment is mostly composed of Glycerophospholipids (PC and PE). The ER produces the bulk of structural phospholipids and cholesterol. Ceramide, the precursor of complex sphingolipids is also produced at the ER. Even though sterols and complex sphingolipid precursors are synthesized in the ER, these products are rapidly leaving this compartment and are transported to other organelles via vesicular transport but also via a non-vesicular route involving membrane contact sites (Jackson et al., 2016) (also discussed in chapter 2).
Figure 1.3: Lipid synthesis and distribution in cells. Lipid composition is shown as graphs and expressed as the percentage of total phospholipid content (blue for mammals and light blue for yeast). The figure shows the sites of synthesis of the major phospholipids (in blue) and that of signaling lipids involved in membrane recognition (in red). From (van Meer et al., 2008).
The Golgi apparatus is a lipid-based sorting station. It is composed of several compartments called cisternae. The cisterna on the ER side is called the cis-Golgi and the cisterna on the plasma membrane side is called the trans-Golgi. The main processes that take place at the Golgi apparatus are the post-translational modifications of newly synthesized proteins coming from the ER (known as protein maturation). The Golgi is the place where significant level of lipid synthesis occurs. This compartment is specialized in the synthesis of complex sphingolipids (SM, GSLs) (Futerman and Riezman, 2005). PC and PE synthesis can also take place at the Golgi apparatus. Cholesterol levels are higher as compared to the ER membrane. The PI derivative PI(4)P, which acts as a signaling lipid, is enriched at the trans-Golgi network.
The plasma membrane has very different lipid content as compared to the other intracellular membranes (Figure 1.3). It is enriched in sphingolipids and sterols which are packed at higher density than glycerophospholipids and can resist mechanical stress. The plasma membrane bilayer is highly asymmetric as the outer leaflet mainly contains SM and PC while the inner leaflet contains mostly PE, PS but also the PI derivative PI(4,5)P2 (van Meer et al., 2008). The inner leaflet is thus highly negatively charged.
The plasma membrane is not involved in the autonomous synthesis of structural lipids but numerous PI lipid derivatives (PI(3,4)P2, PI(4,5)P2, PI(3,4,5)P3, PI(4)P), involved in signaling cascades, were shown to be synthesized or degraded there (Di Paolo and De Camilli, 2006). SM synthesis was also shown to take place at the plasma membrane (Tafesse et al., 2007).
While the composition of early endosomes is very similar to that of the plasma membrane, late endosomes have a quite different composition as the amounts of sterols and PS decrease whereas the levels of bis(monoacylglycero)phosphate (BMP) increase (Kobayashi et al., 2002). BMP is a cone-shaped and negatively charged lipid which was shown to act in multivesicular body generation (endosome containing internal vesicles that originate from inward budding), fusion processes and sphingolipid hydrolysis (Gallala and Sandhoff, 2011; Matsuo et al., 2004). The endocytic compartments recruit specific sets of kinases and phosphatases that allow the regulation of phosphoinositide content (Di Paolo and De Camilli, 2006). Thus, early endosomes are composed of PI(3)P, whereas PI(3,5)P2 is mainly found on late endosomes.
As a general important comment, the PI lipid derivatives previously mentioned act as signaling lipids which allow membrane identification and subsequently the recruitment of cytosolic proteins involved in vesicular transport (Di Paolo and De Camilli, 2006).

Membrane domain formation

Different states of membranes

As previously mentioned, cellular membranes are composed of mixtures of many different lipid species which display various geometrical shapes. This lipid structure is dependent on their physical properties (size of the headgroup, degree of unsaturation, aliphatic chain length) but also on external parameters such as temperature or hydration (Pomorski et al., 2014). As any physical system, lipid bilayers can exist in different phases depending on the overall structures of their lipid content.
The membrane is fluid at high temperature and in different liquid-crystal phases at lower temperatures (Los and Murata, 2004). The two most extreme phases are the gel or solid ordered phase (So) and the liquid disordered phase (Ld) (Figure 1.4). In the So phase, the lipid acyl chains can undergo trans-isomerization which leads to their extension and more Van der Waals interactions. Stronger interactions lead to more ordered lipid packing which prevents any lateral lipid diffusion (Seu et al., 2006). The Ld phase is usually characterized by the presence of unsaturated lipids (one or more double bonds in the acyl chain) and their irregular packing. Unsaturation of the acyl chains leads to kinks in their structure which reduces the surface area accessible to other lipids and thus weakens Van der Waals interactions. Consequently, the Ld phase is a highly fluid state in which individual lipids can freely diffuse (Seu et al., 2006).
Under physiological conditions, intracellular bilayers tend to exist in a fluid phase and can undergo phase transition under correct environmental conditions. The temperature at which a membrane lipid can undergo phase transition from the gel to the liquid state is the melting temperature (Tm). The Tm can vary between lipids due to their different structural properties (acyl chain length and degree of unsaturation) (Cevc, 1991). Lipids that exhibit longer acyl chains will have higher surface areas as compared to lipids exhibiting smaller chain length, resulting in stronger Van der Waals interactions between aliphatic chains and thus increased Tm. As previously mentioned, increasing unsaturation leads to weaker Van der Waals interactions and thus to lower Tm of the lipid. As the Tm strongly depends on the amount of unsaturation and on the length of the acyl chains, one can thus make the distinction between high Tm lipids that will be in a solid state at physiological conditions, and low Tm lipids that will be in a liquid state under the same conditions.
As an example, sphingolipids usually carry saturated or trans-unsaturated (linear) aliphatic chains whereas the acyl chains of glycerophospholipids are often unsaturated. Membranes composed of sphingolipids therefore adopt a more tightly packed structure (solid-like phase) as compared to glycerophospholipids containing membranes that form less ordered domains (Ld phase).
Figure 1.4: Mechanism of lipid domain formation. Membranes can exist as a fluid state or as a solid state at high and low temperature, respectively. The temperature at which a lipid can transition from one phase to another is the melting temperature (Tm). The Tm is dependent of the specific structure of the lipid (acyl chain length, degree of unsaturation). At physiological conditions, high Tm lipids (long saturated acyl chains) will localize to the solid phase whereas low Tm lipids (short unsaturated acyl chains) will transition to the liquid-disordered state (Ld). Cholesterol can induce the formation of an intermediate liquid-ordered phase (Lo).
Cholesterol is another key component of eukaryotic cellular membranes and was shown to drastically affect the physical properties of membranes, through lateral order disruption of the gel phase and ordering of the Ld phase (Henriksen et al., 2006; Ipsen et al., 1987). The presence of cholesterol in membranes can lead to the appearance of an intermediate state between gel phase and Ld, the liquid ordered (Lo) phase (Figure 1.4). This Lo phase is characterized by a tight lipid packing, as in the gel solid-ordered phase, but also by a rapid lateral diffusion rate, as in Ld membranes (London, 2002; M’Baye et al., 2008). Lo membranes are usually thicker, stiffer and less permeable than Ld membranes (Rawicz et al., 2008). The lateral diffusion coefficient is 2-3 fold less in the Lo phase as compared to the Ld phase (Veatch and Keller, 2005).

READ Mitotic and meiotic spindle dynamics comparison in fission yeast

Phase state of cellular membranes

Cellular membranes display inhomogeneous lipid distribution at different levels (van Meer et al., 2008). The lateral heterogeneity in cellular membranes can be described by the lipid raft hypothesis (Simons and Ikonen, 1997) which has been a matter of great debate in the last 20 years. In biomembranes, lipid rafts, which are characterized by a tight packing of saturated lipids and cholesterol, likely exist in a Lo state and behave like islands floating in a sea of loosely-packed Ld domains of unsaturated glycerophospholipids (M’Baye et al., 2008). Lipid rafts have received a huge attention as they are believed to be involved in many cellular processes such as signal transduction, lipid trafficking and regulation of membrane protein activity (Jacobson et al., 2007; Lingwood and Simons, 2010; Owen et al., 2012).
Another less described heterogeneity is related to the differential composition and organization of the plasma membrane as compared to intracellular membranes. Cholesterol and sphingolipid levels were found to be increased from the ER to the plasma membrane and this seems to be due to the directed anterograde transport and the absence of retrograde transport of these lipids (Brugger et al., 2000; Klemm et al., 2009). As previously mentioned, even though raft components are synthesized in the ER and the Golgi complex, they are quickly leaving these compartments and are transported towards the plasma membrane. As an example, sphingomyelin is mainly found at the plasma membrane even though ceramide, the hydrophobic backbone of sphingolipids, is synthesized in the ER and the assembly of the sphingolipids headgroups to ceramide takes place at the Golgi complex (van Meer and Lisman, 2002).
Thus, lipid raft components are moved toward the plasma membrane where they concentrate but also spread into the endocytic recycling pathways (Mukherjee and Maxfield, 2000). This seems to be consistent with studies showing that more than 60% of all cellular cholesterol is located to the plasma membrane whereas intracellular membranes such as the ER or the Golgi membranes exhibit low levels of cholesterol (Ikonen, 2008; Mesmin and Maxfield, 2009).
As cholesterol is known to promote phase separation (Ohvo-Rekila et al., 2002; Silvius et al., 1996), this process is thought to mainly occur at the cholesterol-rich plasma membrane. Multiple studies, in particular the ones using fluorescent probes such as Laurdan (Bagatolli, 2006; Owen et al., 2012) and its derivatives (Kim et al., 2008; Sezgin et al., 2014) have focused on lipid raft organization in model and cell membranes. These probes have the ability to change fluorescence color and intensity in response to changes in membrane hydration and solvent relaxation (Demchenko et al., 2009), two parameters linked to lipid order.
Niko and coworkers recently developed a new probe based on pyrene which demonstrates enhanced photophysical properties over Laurdan (Niko et al., 2016). Their study confirms clear differences in lipid order among the different cellular membranes with the plasma membrane mainly composed of Lo domains while intracellular membranes are much closer to Ld phases. This is consistent with the higher amount of sphingomyelin and cholesterol, both responsible for the formation of lipid rafts, at the plasma membrane.

Membrane deformations

Biological membranes are two-dimensional surfaces with two principal curvatures C1 = 1/R1 and C2 = 1/R2 (with R1 and R2 referred to as the principal radii of curvature) along two perpendicular directions (Zimmerberg and Kozlov, 2006) (Figure 1.5). The total curvature of the membrane is C = C1+C2. In the case of a spherical vesicle of radius R, the membrane deforms equally in both directions leading to C1 = C2 = 1/R and a total curvature Cv = 2/R. In the case of a cylindrical tube of radius R, which is curved only in one direction and flat in the other, C1 > 0 and C2 = 0 yielding a total curvature Ct = 1/R. When proteins interact with the membrane, it is said that a protein can sense positive curvature if it senses the convex side of the membrane whereas the curvature is negative if it senses its concave side.
Lipid bilayers have an average thickness in the order of 4-5 nm (Marquardt et al., 2016). Logically enough, this value also corresponds to the lower limit of radii to which a bilayer can be bent. Thus, in the perspective of a protein, if a membrane has a curvature radius in the same range (10-50 nm) it will be considered as highly curved whereas if the radius is superior to 50 nm it will be considered as weakly curved.

Membrane curvature in cells

Due to their heterogeneous lipid and protein compositions, intracellular compartments exhibit membrane regions with both high and low curvature (Figure 1.6). For example, the ER and the Golgi form complex networks of interconnected flat sheets and highly curved tubules (30-60 nm diameter) (Shibata et al., 2009; Voeltz and Prinz, 2007). Mitochondria also display many curved invaginations (30 nm in diameter), called cristae, which are directed towards the mitochondrial matrix and thus greatly increase the total surface area necessary for chemical reactions (Voeltz and Prinz, 2007). Endosomes were also shown to exhibit tubular regions of high curvature and globular regions of low curvature (Sonnichsen et al., 2000).
These tubular structures are often highly dynamic. They can undergo continuous fission and fusion events and get continuously rearranged by moving along the actin or microtubule cytoskeleton.

Table of contents :

From cell discovery to membrane binding properties of RAB GTPases
1 Biology and physics of membranes
1.1 The Lipid bilayer
1.1.1 From a lipid molecule to a bilayer
1.1.2 Different classes of lipids
1.1.3 Lipid synthesis and distribution in cells
1.2 Membrane domain formation
1.2.1 Different states of membranes
1.2.2 Phase state of cellular membranes
1.3 Membrane deformations
1.3.1 Membrane curvature in cells
1.3.2 Mechanisms of membrane deformation
1.3.3 Protein curvature sensing
1.4 In vitro experimental approaches
1.4.1 Model membranes for in vitro experiments
1.4.2 Phase separation from living cells to model membranes
1.4.3 Curvature sensing on model membranes
2 RAB GTPases
2.1 RAB discovery and evolution
2.2 RAB sequence and structure
2.2.1 G-domain
2.2.2 RAB specific sequence motifs
2.2.3 RAB C-terminal region
2.3 RAB posttranslational modifications
2.3.1 RAB activation cycle
2.3.2 RAB membrane insertion and extraction
2.4 Membrane targeting of RAB GTPases
2.5 RAB GTPases and vesicular transport
2.5.1 General mechanism of intracellular transport
2.5.2 RABs and membrane tethering
2.6 Focus on the RAB proteins used in this study
2.6.1 RAB1 and the ER-Golgi intermediate compartment
2.6.2 RAB6 and the Golgi
2.6.3 RAB4 / RAB5 / RAB11 and the endosomal system
2.6.4 RAB35 and the plasma membrane
3 Materials and Methods
3.1 Protein synthesis and modification
3.1.1 Protein expression and purification
3.1.2 In vitro modifications of RAB and GST proteins
3.2 Experimental studies with GUVs
3.2.1 Synthesis of giant unilamellar vesicles
3.2.2 Generalities of the experimental approach
3.2.3 Curvature sensing experiments with GUVs
3.3 Experimental studies with purified Golgi membranes
3.3.1 Purification of Rat Liver Golgi stacks
3.3.2 Experimental chamber
3.3.3 Pulling tubes with kinesins
3.3.4 Immunofluorescence on Golgi membranes
4 Article: RAB proteins bind lipid packing defects
5 RAB4 and RAB11 binding requirements
5.1 Description of the in vitro approach
5.2 RAB4 and RAB11 recruitment to GUV membranes
5.2.1 RAB4 and RAB11 are not recruited to PC-containing membranes
5.2.2 RAB4 and RAB11 are not recruited to GUVs of various lipid composition
5.2.3 Membrane curvature has no effect on the recruitment of RAB4 and RAB11
5.3 RAB4 and RAB11 recruitment to purified Golgi fractions
5.3.1 RAB4 and RAB11 are positively recruited through their prenyl group
5.3.2 RAB4/RAB11 membrane recruitment does not depend on the presence of effector proteins
5.4 Monoprenylated RAB proteins are mislocalized to the same membrane structures
5.4.1 Monoprenylated RAB proteins localize to the same membrane structures
5.4.2 Monoprenylated RAB proteins do not localize to Golgi or recycling endosomal structures
5.5 Discussion
6 RAB6-induced membrane tethering
6.1 Specificities of RAB6-induced membrane tethering
6.1.1 Vesicle tethering is a RAB6-specific effect
6.1.2 Vesicle tethering is nucleotide and concentration dependent
6.1.3 RAB6-induced vesicle tethering is mediated by a RAB-RAB dimerization in trans
6.1.4 The RAB-RAB interaction is dynamic
6.2 Involvement of the Switch regions
6.2.1 RAB6A mutant induces vesicle tethering
6.2.2 Unprenylated RAB6A does not interact with membrane-bound RAB6A
6.2.3 Bivalent αRAB:GTP antibodies promote vesicle tethering
6.2.4 Effect of monovalent RAB6 effector proteins
6.3 Discussion
Concluding remarks
References