SHEARING A MEMBRANE
Shear is the stress involving a deformation of the material in two parallel opposite directions at constant surface area (Figure 2-8). A membrane resists such a deformation only if the relative positions of its constituent molecules are fixed by some lattice structure. Thus, fluid membranes are, by definition, unable to sustain shear deformations (Zeman, Engelhard et al. 1990). But shear deformations can become relevant in the case of gel-phase bilayers or if the membrane is coupled to an external lattice structure such as the cytoskeleton. Areal density of energy (Hshear) associated to shear stress can be deduced from Hook’s law: 𝐻Shear=(12)μ (𝜆2+𝜆−2−2) (2-2). Where λ= (L0 + ΔL)/L0 is the lateral extension rate, μ the shear modulus (expressed in J.m-2). In the case of fluid membranes, shearing deformations are negligible compared to stretching and bending and is ignored.
BENDING A MEMBRANE
Membrane bending is the dominant deformation for fluid lipid bilayers. To describe membrane bending, one must relate to the notion of membrane curvature, which corresponds to any deformation out of the membrane plane (Figure 2-8). The bending energy derives from the curvature of the membrane: at a given point of the surface, one can define two perpendicular radii of curvature R1 and R2. The two principal curvatures are then defined as the inverse of these radii, with a positive or negative sign, which corresponds to the 2 radii oriented in the same or opposite direction relative to the surface, respectively. The two principal curvatures are thus C1 = 1/R1 and C2 = 1/R2. Thus, if the radius is very large (i.e., the membrane is nearly flat), the curvature is small, and vice versa. The sum of the principal curvatures C1 and C2 is the mean curvature J = C1 + C2. And, the product of the principal curvatures is the Gaussian Curvature K = C1 x C2. The mean curvature J and the Gaussian Curvature K are local parameters that describe the membrane shape.
For example, for a sphere of radius R, the mean curvature is 2/R and the Gaussian curvature is 1/R2, whereas for an infinite cylinder of radius R, the mean curvature is 1/R and the Gaussian curvature is equal to zero at every point of the cylinder surface. Finally, a saddle point has the particularity of having two curvatures of opposite signs, a positive and a negative curvature. Thus a saddle point has a negative Gaussian curvature (Figure 2-8).
To define the capacity of a membrane to bend, two intrinsic parameters must be considered: κ, the bending rigidity modulus (or bending stiffness of the membrane) ranging from 10 to 100 kBT and κG, the Gaussian bending rigidity modulus (or Gaussian Curvature modulus). Both depend on the membrane composition and they represent the energetic cost to generate principal curvature (by increasing J) and Gaussian curvature (by increasing K). The bending modulus depends on the aliphatic chain length and degree of unsaturation (Evans and Rawicz 1990; Rawicz, Olbrich et al. 2000; Marsh 2006; Rawicz, Smith et al. 2008), for instance it increases from 13 to 30 kBT when the PC lipid chains contain 22 carbon atoms instead of 13 and decreases to 10 kBT for cis-polyunsaturated PC lipid (Rawicz, Olbrich et al. 2000). The corresponding areal density of energy (Hcurvature) is given by: 𝐻curvature=(12)𝜅 (𝐽−𝐶₀)2+𝜅G 𝐶₁ 𝐶₂ (2-3)
Where C1 = 1/R1 and C2 = 1/R2 are the two principal local membrane curvatures describing the shape of the membrane at a given point, C0 is the spontaneous curvature (i.e. the curvature of the membrane in the absence of any external stress). κ is the bending rigidity modulus (expressed in J or in kBT unit) and κG is the Gaussian bending rigidity modulus (expressed in J or in kBT unit).
As described previously, cellular membranes undergo continuous rearrangements. Many cellular processes such as vesicular trafficking, exocytosis, endocytosis, cell division, entry and release of enveloped virus, etc., involve membrane budding and fission events. Intracellular vesicular trafficking is used to transfer cargoes between membranes in the secretory and the endocytic pathways. The generation of these vesicles occurs in three steps: cargo sorting from the donor compartment, membrane budding (or tubulation) and finally membrane separation from the donor compartment by a fission event. Membrane fission was first studied and discussed in the context of dynamin-induced membrane fission on clathrin-coated vesicles at the plasma membrane. But, there are many scission processes at the surface of organelles that are dynamin-independent. This is the case for the scission of COPI and COPII coated-vesicles at the Golgi apparatus and the ER, respectively, and of course, of the scission of vesicles in MVBs by the ESCRT complexes.
When a bud is formed at the surface of a membrane, fission reaction proceeds as follows: (i) constriction of the budding vesicle with formation of a highly curved neck (Figure 2-9 / A and B), which can be mediated by different means (ii) merge of the contacting monolayers in a stalk intermediate (hemifission) (Figure 2-9 / C) and (iii) disappearance of the fission stalk and completion of the reaction (Figure 2-9 / D). Hemifission (similarly to the hemifusion-like pathway) is an intermediate stage of the fission reaction, where the opposing internal leaflets of the neck are fused, but not the external ones; it requires a transient membrane disruption which is opposed by the hydrophobic forces preserving the integrity and continuity of the lipid assembly (Chernomordik and Kozlov 2003; Kozlovsky and Kozlov 2003; Chernomordik and Kozlov 2005; Kozlov, McMahon et al. 2010). The formation of this hemifission state allows the accomplishment of membrane fission without compromising the integrity of the bilayer by exposure of the content to the external milieu or even content leakage (Matsuoka, Orci et al. 1998; Takahashi, Kishimoto et al. 2002; Frolov, Dunina-Barkovskaya et al. 2003).
ESCRT MACHINERY IN SACCHAROMYCES CEREVISIAE
The endosomal sorting complex required for transport (ESCRT) complexes were discovered and named, in 2001 by Scott Emr’s group, for their main role in ubiquitin-dependent sorting from endosomes to lysosomes in the multivesicular body biogenesis (MVB) (Figure 3-1) (Katzmann, Babst et al. 2001).
The machinery was first identified in yeast by means of genetic isolation of mutants that cause defective protein sorting to the vacuole, the functional yeast equivalent of the lysosome (Bankaitis, Johnson et al. 1986; Rothman, Howald et al. 1989). These mutants, called “class E vps (Vacuolar Protein Sorting) mutants”, caused a major morphological change of the vacuole (Raymond, Howald-Stevenson et al. 1992). Most of the class E vps genes were later found to act in succession to concentrate trafficking cargoes and include them in forming late endosomes, also termed multivesicular bodies (MVB). The latter, also called multivesicular endosomes (MVE), are specialized compartments within endosomes that are delivered into lysosomes for protein degradation (Katzmann, Babst et al. 2001). They consist of a limiting membrane and small intraluminal vesicles (ILVs).
Like all vesicle budding reactions, the formation of intraluminal MVB vesicles requires three successive steps, respectively, cargo recognition and sorting, membrane budding and, vesicle separation from a donor membrane which in this case is the endosome (Adell and Teis 2011). In opposition to the formation of secretory and endocytic vesicles, where membrane budding and fission occur into the cytosol, the MVB formation requires budding away from the cytosol (Katzmann, Babst et al. 2001).
ESCRT ROLE IN TERMINAL STAGES OF CYTOKINESIS
Cytokinesis, the last step of cell division, involves large-scale cleavage of the plasma membrane. This process is characterized by the constriction of an acto–myosin contractile ring leading to the ingression of the plasma membrane at the cell equator, which partitions two cytoplasmic domains of emerging sister cells that remain connected by a membrane tube about 1 μm wide, called the intercellular bridge (Figure 3-6) (Eggert, Mitchison et al. 2006; Steigemann and Gerlich 2009; Guizetti and Gerlich 2010; Green, Paluch et al. 2012). This process divides the organelles and most of the cytoplasm equally between the two daughter cells, but the microtubules forming the spindle remain in the intercellular bridge. To separate the daughter cells and finalize the cellular division process, the microtubules, mostly enriched at the center of the intercellular bridge, in a dark zone region named the “midbody”, must be severed and the plasma membrane must be sealed (Figure 3-6). Furthermore, Cryo-EM measurements showed that cell separation does not take place at the midbody site itself but rather at two peripheral sites located about 1 μm away from the midbody center (Elia, Sougrat et al. 2011; Guizetti, Schermelleh et al. 2011).
ESCRT ROLE IN HIV-1 BUDDING
Many enveloped viruses, such as HIV, hijack the cellular ESCRT machinery to the cytoplasmic leaflet to promote their own egress from infected host cells (Morita and Sundquist 2004; Martin-Serrano and Neil 2011). Retroviruses such as HIV replicate and leave the cell through a process called budding. This means that the virus uses part of the host cell plasma membrane to enclose itself and bud out of the cell before proceeding to a new host. The complete process can be divided into a series of steps.
For the budding to be more efficient, it is also desirable to gather the viral proteins into defined budding spots. This is thought to be defined by specific lipid domains in the cell membrane called lipid rafts, where the lipids are arranged in a more ordered state. Following the assembly of the viral proteins at the budding sites, the budding is initiated by locally deforming the cell membrane. The membrane is then further deformed, making the buds grow to a usually defined size after which they are finally cut from the cell membrane allowing the now fully enveloped virus to exit the cell. In many enveloped viruses that replicate through budding, scission is performed by the host cell machinery, the most common being the ESCRT complexes.
HIV assembly and budding require the clustering of viral Gag proteins at the plasma membrane. Gag proteins are the major structural proteins of retroviruses (Ganser-Pornillos, Yeager et al. 2008), and fluorescence microscopy experiments have shown that the HIV virions are fully assembled when the recruitment of Gag molecules stops (Jouvenet, Bieniasz et al. 2008). The resulting virus bud is connected to the plasma membrane by a narrow membrane neck; HIV buds have typically diameters ranging from 100 nm to 200 nm (von Schwedler, Stuchell et al. 2003; Morita, Sandrin et al. 2011). Thus, formation of the cell membrane-attached HIV bud occurs independently of ESCRT proteins. However, its detachment from the PM of the cell necessitates the ESCRT complexes to achieve the membrane scission step (Figure 3-13).
Studies have found that ESCRT-III filaments surround Gag assemblies at the PM in Vps4 depleted cells (Hanson, Roth et al. 2008). More recent super-resolution studies detect endogenous ESCRT proteins or low expression HA-tagged ESCRT components in clusters (diameter 60 – 100 nm) at the base of or inside viral necks, but not inside the viral particle (Van Engelenburg, Shtengel et al. 2014). This suggests that membrane scission by ESCRT-III filaments and Vps4 occurs within the bud neck, which is consistent with their role in other biological processes.
Table of contents :
Chapter 1. Introduction
Chapter 2. Physics of Biomembranes
2.1 Structure of lipids
2.2 Membrane mechanics
2.3 Membrane tension
2.4 Membrane fission
Chapter 3. The ESCRT-dependent membrane remodelling processes
3.1 ESCRT machinery in Saccharomyces Cerevisiae
3.2 ESCRT machinery in Homo sapiens
3.2.2 ESCRT role in HIV-1 budding
3.2.3 ESCRT role in neuronal pruning
3.3 ESCRT-III crystal structure and cycling
3.4 ESCRT-III polymer structures in vivo and in vitro
3.4.1 ESCRT-III polymers form flat spirals
3.4.2 ESCRT-III polymers form helices and tubes
3.5 Theoretical models for membrane scission by the ESCRT-III polymers
3.5.1 Spiral spring (buckling) Model for ESCRT-III mediated membrane scission
3.5.2 Theoretical Dome Model for ESCRT-III mediated membrane scission
3.6 Objective: characterization of CHMP2B and determination of its role within the ESCRT-III machinery
3.6.1 CHMP2B is specific to higher organisms
3.6.2 CHMP2B is implied in the diversification of ESCRT functions
3.6.3 CHMP2B mutation leads to a neurological disorder: Fronto-Temporal Dementia
3.6.4 Thesis objective: study of CHMP2B using model membranes in vitro
Chapter 4. Material and methods
4.1 Protein purification
4.2 Model membrane systems
4.2.2 Lipid mixtures
4.2.3 GUVs preparation
4.2.4 Making LUVs and SUVs
4.2.5 Making SLBs
4.3 Fluorescence microscopy
4.3.2 Spinning disk confocal microscopy
4.3.3 Fluorescence recovery after photobleaching assay (FRAP)
4.3.4 Fluorescence-activated cell sorting (FACS)
4.4 Cryo-electron microscopy
4.4.1 Cryo-EM principle
4.4.2 Experimental conditions
4.5 Micropipette aspiration assay
4.5.1 Micropipette aspiration principle
4.5.2 Experimental conditions
4.6 Quartz Crystal Microbalance with Dissipation monitoring
4.6.1 QCM-D principle
4.6.2 Typical experiment
4.7 Atomic Force Microscopy (AFM)
4.7.1 Principle of AFM
4.7.2 Experimental conditions
Chapter 5. Results
5.1 Optimization and characterization of CHMP2B protein interaction with model membranes
5.1.1 Study of CHMP2B protein stability
5.1.2 CHMP2B proteins bind preferentially to PI(4,5)P2-containing membranes
5.1.3 Encapsulation of CHMP2B proteins inside GUVs to mimic ESCRTs inverted topology .
5.1.4 CHMP2B proteins interaction with PI(4,5)P2 lipids is irreversible
5.1.5 CHMP2B proteins form a reticular-like structure on GUVs
5.1.6 CHMP2B assembles into ring-like structures at the nanoscale
5.2 CHMP2B polymers modulate membrane elastic properties
5.2.1 Investigation of CHMP2B mechanical properties by applying osmotic shocks
5.2.2 Study of CHMP2B mechanical properties by micropipette aspiration
5.2.3 Study of CHMP2B mechanical properties by AFM
5.2.4 Mobility of CHMP2B supramolecular assembly on GUVs
5.2.5 Diffusion of membrane-associated protein on GUVs covered by CHMP2B assemblies
5.3 CHMP2A and CHMP2B display opposite properties on model membranes
5.3.1 Study of CHMP2A protein interaction on model membrane
5.3.2 CHMP2A and CHMP2B proteins display opposite mechanical properties on membrane
5.3.3 CHMP2A + CHMP3 supramolecular assembly on membrane is dynamic in contrast with CHMP2B
5.4 CHMP3 perturbs CHMP2B polymerization and assembly on membranes
5.4.1 CHMP3 blocks CHMP2B polymerization on membranes
5.4.2 CHMP2B + CHMP3 supramolecular assembly is not dynamic
5.4.3 CHMP3 modulates the mechanical properties of CHMP2B polymers
5.5 CHMP2A and CHMP2B modulate CHMP4B assembly on membranes
5.5.1 CHMP4B assembly on membranes
220.127.116.11 CHMP4B alone forms spirals on flat membranes
18.104.22.168 Mechanical properties of GUVs coated with CHMP4B
5.5.2 CHMP2B disorganizes CHMP4B spirals on flat surfaces
5.5.3 CHMP2A and CHMP2B induce deformations on CHMP4 assembly on membrane tubes
Chapter 6. Conclusions and perspectives