Christophe
Tribet
and
Florent
Vial
Physico-chimie des Polymères et Milieux Dispersés, CNRS UMR 7615 and Université Paris 6, ESPCI, 10 rue Vauquelin, F-75005 Paris, France
First published on 11th October 2007
This review summarizes recent investigations on the association of macromolecules on lipid bilayers. Hydrophilic and flexible polymers can form soft coronae tenuously adsorbed or anchored on the lipid membrane. Other synthetic macromolecules are embedded in the apolar region of the membrane. Recent experimental and theoretical works focus on the perturbation of lipid properties achieved depending on the nature and strength of binding. Of importance to biomimicry, to tethered model membranes, and drug carriers, the effects achievable include modulation of the lateral diffusivity of lipids, shape distortions, lateral segregations, formation of well-defined nanopores and ultimately the stimuli responsive disruption of the membrane.
![]() Christophe Tribet | Christophe Tribet received his Ph.D. in chemistry from the University Paris VI (1993) and pursued post-doctoral research at ULB (Center of Non-Linear Phenomena, Bruxelles) and at the Institute of Biology and Physical Chemistry, Paris. He joined the Dept of polymer science at ESPCI in 1994, and since 2005 has been visiting professor at the University de Montreal. His current research interests relate to stimuli- and light-responsive polymers for the handling of lipid membranes and proteins, sp. integral membrane proteins. |
![]() Florent Vial | Florent Vial studied chemistry at the Ecole Normale Superieure de Cachan & University Paris XI Orsay, France. He completed his Ph.D. at the University Paris VI, on the controlled permeabilization of lipid membranes by stimuli-responsive amphiphilic copolymers. He is now teaching physics and chemistry. |
The association of macromolecules with membranes has proven to be an efficient means of achieving remarkable effects. Most often, polymers are expected to form a protective (repulsive) corona that enhances the circulation time of liposomal formulations in vivo, and substitutes for glycolipids and glycoproteins. The adsorbed macromolecules confer also on the layer above the membrane, high viscosity and/or elastic properties that differ markedly from those of bulk water. Experimental evidence9 and theoretical models10 show that yet at low coverage (isolated chains) anchored polymers modify the mechanical properties of the lipid membrane (e.g. increase the bending rigidity). Not surprisingly, obvious effects are reported on the mechanical response of vesicles,11,12 buddings and deformations of soft bilayers.13 The mechanical coupling between complex flows in a gel-like layer and the deformation of membranes enables one to mimic with giant vesicles the behaviour of red-blood cells in vessels14 or their disk-like shapes.15,16 To the benefit of the development of biosensors, tethered lipid membranes can be deposited on a cushion of hydrophilic polymers improving mechanical stability, but also reducing mobilities of ions in the cushion as compared to water.17 When it is tightly anchored in a membrane, a polymer may also control the formation of lateral lipid domains,18 transversal asymmetry in the composition of the leaflets,19 permeabilization and formation of pores.20,21
Ringsdorf et al.22 have pioneered the chemistry of polymerized lipids and liposomes, aiming at the stabilization of hollow capsules against rupture and de-assembly (e.g. in presence of other amphiphiles or at high dilution below the critical association concentration). Since that time, the expectation for new medical applications based on liposomes has motivated most studies of drug delivery by polymer-containing vesicles. Polymerization of lipids,23 a wide variety of hydrophilic macromolecules anchored on lipid vesicles,3,24 and novel polymer self-assemblies and polymeric membranes (polymersome)5,25,26 have been designed as promising nano-carriers with enhanced circulation-time in vivo. The reviews cited above discuss the formulation and potentialities of such systems for the purpose of therapeutic application. Likewise it is emphasized on the protection against aggregation and resistance to disruption/digestion of the carriers in biological fluids. Slow kinetics of dissociation provide for instance an apparent “stability” that is not distinguished from a low free energy of formation (e.g. kinetically frozen assemblies of diblock copolymers and polymersomes).25 The chemistry of such nano-carriers and delivery issues are not the purpose of the present review, and the reader is referred to the papers cited above. Here we focus on the interaction of polymers with bilayers and the resulting changes in the energy and shape of the membranes.
The main features of macromolecules compared to molecular additives come from the covalent string of monomers whose global behaviour averages the properties of each “molecular” subunit together with constraints in their accessible spatial distributions. As a consequence, one domain of the chain with enough affinity for the lipids would bring the rest of the chains in the vicinity of the membrane irrespective of the monomer–membrane (possibly repulsive) interaction. The reduced translational entropy of the monomers leads to abrupt adsorption and collapse of polymer chains as soon as the monomer–surface or monomer–monomer interactions are weakly attractive.27 Some practical consequences are listed below: (i) water-soluble polymers are usually too hydrophilic to pass through the bilayer, and most often stick to a single leaflet. This may explain their mild effect on membranes and the low toxicity of chains made of a majority of hydrophilic monomers. Likewise, “cushions” of hydrophilic polymers appeared as promising mild supports to deposit tethered bilayers.28,29,30 (ii) A few monomers with an affinity for a surface provide enough driving force for tight adsorption of a polymer chain. The multipoint attachment by a few anchors achieves dilution-independent binding. In addition, a possible inter-polymer connectivity would impart the layer with new elastic properties. (iii) An isolated polymer chain may perturb (or protect) a large area containing tens to several thousands of lipids. The typical size of a polymer coil can be increased from ∼5 nm2 to 104 nm2 with increasing the polymer chain length. (iv) The properties of polymers, including hydrophobicity, can be abruptly switched. Parameters such as pH, salt concentration, or temperature can easily trigger the monomer–monomer intrachain interaction between (weakly) attractive and repulsive conditions. Such a slight change in solvent conditions results in collapse of a chain or inter-chain aggregation, which in turn makes mixed membranes responsive to stimuli.8,31
We illustrate and summarize the various systems recently investigated as follows. Section 1 summarizes the nature of polymer–lipid interactions and the structures of macromolecules that have been shown to be efficient at anchoring the chains onto/into membranes. In section 2, we illustrate practical aspects of controlling the association on flat cushions of macromolecules (tethered model bilayers) or in solution to obtain liposome-filled gels. At conditions of tighter association, polymers induce the destabilization of liposomes. The coupling with membrane stability is described in section 3, including general thermodynamic considerations as recently proposed by Marsh et al32 to clarify the magnitude of perturbations achieved by an hydrophilic polymer layer and the case of stimuli-responsive chains. The last two sections present the effects of polymers on the shape of liposomes, their curvature (section 4), and the stabilization of complex inclusions (domains and pores, sections 5). Over the last few years, progress in the micromanipulation and the direct microscopic visualization of both supported membranes and giant vesicles has made possible new investigations on domain formation, protrusion, adhesion, and fusion events. The works summarized here illustrate representative use of polymer–lipid mixed assemblies in such experimental set-ups, and the present understanding that has emerged about the coupling between membrane properties and polymer properties. We will particularly emphasize the impact and characteristic features of synthetic polymers that trigger non-specific effects. Indeed, peptide and protein interactions with membranes have recently been reviewed.33
![]() | ||
Fig. 1 Some chemical structures representative of macromolecules forming associates with lipid membranes. (a) PEO with a hydrophobic end, (b) PEO–PPO–PEO triblock, (c, d) polycations of adjustable content of ionic groups, alkyl chains, and hydrophilic segments, (e,f) amphiphilic derivatives of acrylic or glycolic acids (g) complexes of organic phosphate and polyarginin. |
The physical origin of predominantly electrostatic binding is the entropic gain of translational freedom of the counter ions released upon association of the polymer and the membrane. This free-energy gain balances the loss of freedom of multi-charged assemblies having limited conformational space even in their unbound state, namely the lipids in a membrane and monomers of the polyelectrolyte chains. Charged peptides and proteins with even less intrachain mobility of the charges bind to membranes because of similar effects, and detailed calculations of the electric potential enabled a quantitative estimation of the association constant in this case.58,59 A polyelectrolyte (or a protein) adsorbed in the surface of charged vesicles could be completely removed from the membrane by an increase in simple salt concentration or by competition with a soluble polyelectrolyte of opposite charge.35,60 On almost neutral membranes, the electrostatic interaction may not be strong enough to provide attachment. The zwitterionic heads of phospholipids, though neutral, are oriented by self-assembly and their negative phosphate ions lay closer to the hydrophobic core than their positive ammonium moieties. This preferred orientation does not however favour the interaction with anionic surfaces. Pure phosphocholine liposomes are not adsorbed in cationic surfaces, as shown on multilayers of polystyrensulfonate and polyallylamine hydrochloride, but in contrast adhere to the anionic ones, possibly because of the hydrophobicity of the polystyrene backbone.61 The generic binding of polycation on biomembranes is presumably not due to electrostatic association with phospholipids, but most likely driven by the presence of anionic lipids or negatively charged proteins.
Of practical relevance to construct vesicles with controlled shapes and loading, lipids can be adsorbed onto colloidal particles coated with polyelectrolytes and polyelectrolyte multilayers.62 Charged lipids, such as DPPA, form bilayers while zwitterionic lipids, such as DPPC, may adsorb in the form of more than one bilayer. The reduced permeability imparted to those complex particles by the lipid layer, and the reduced mobility of lipids in contact with polyelectrolyte cushion have been extensively studied by Mohwald and colleagues.63,64 Somehow surprisingly, the diffusivity of zwitterionic lipids (phosphocholine heads) was more hindered than that of charged lipids. An obvious strong coupling between head groups and polyelectrolyte cushions (PHA/PSS) was therefore revealed irrespective of the total charge of the lipid head.63
But, on neutral and hydrophilic cushions, lipid membranes have been shown to be unstable, presumably because of very weak adhesion.78 In contrast, cushions made of cationic macromolecules,78,79 and/or containing hydrophobic groups,28,29,75,80 specifically lipid-like anchors, are very effective at rapidly adsorbing liposomes (in less than one minute) and at preserving the stability of the membrane afterward, even against rinsing and drying (Fig. 2).81 Reversibility of the adhesion and its exquisite control has been achieved by modulating the (anionic) charge density of the polymer, using thermo-responsive macromolecules,82,83 or other polymers imparted with stimuli-triggered hydrophobicity.84 The “sticky” cholesterol or lipid tails can be, for instance, replaced by a photo-responsive chain that partitions into lipid bilayers in the dark but, when exposed to UV light, switches to a more polar state and releases the attached vesicles. It seems reasonable that the lateral density of hydrophobic anchors affects the properties and stability of supported membranes. Well-defined telechelic polymers (i.e. one hydrophobe at one end, and functionalized to be adsorbed into the substrate at the other end) provide good orientation of the anchors, toward the upper phase, and good control of their lateral density. A density of 5 mol% anchors is sufficient to maintain attachment of a lipid bilayer.
![]() | ||
Fig. 2 Schematic sketch of the adsorption of a vesicle and formation of tethered bilayers on a cushion of amphiphilic polymer. Reprinted with permission from ref. 79, copyright 2000, American Chemical Society. |
Low surface densities of the hydrophobes plunged in the membrane do not hinder the diffusivity of lipids (up to ∼20 mol% anchors relative to lipids). Low density of anchors are also compatible with high diffusivity of integral membrane proteins in tethered bilayers.85,86 However, higher density may freeze the Brownian motion in membranes.80,87,88 Naumann et al.89 have studied and quantified the obstructed diffusion of both lipids and membrane proteins by particle-tracking experiments. They show that the diffusivity decreases almost linearly with the density of anchors, down to zero mobility at a threshold that depends on the size of the mobile species (Fig. 3). This threshold and the complex signature of lateral diffusion at intermediate densities match well with hard–core repulsion with obstacles (the anchors) that are randomly distributed at low density (typically <10 mol% of lipid anchors) but gradually form small aggregates reaching percolation at higher densities. Note that the length of the polymer, which ultimately controls the swelling, the thickness, and roughness86 of the tether also appears critical to avoid the aggregation of protein inclusions in the membranes and maintain fluidity.77 The method of deposition (vesicle adsorption/fusion, langmuir–blodget, etc.) is in addition a major determinant of the long-range diffusion in tethered membranes, because it affects the continuity or the “patchy” morphology of the layer.30
![]() | ||
Fig. 3 Lateral diffusivity of a labelled lipid or of the transmembrane protein Bacteriorhodopsin, as a function of the molar fraction of polymer-attached lipid anchors in a polymer supported bilayer. Zero-diffusivity matches with the percolation threshold of obstacles and hard core interactions with the probes, assuming that the anchors are immobile. Reprinted with permission from ref. 89, copyright 2005, Biophysical Society. |
In bulk solution, similar attraction between long macromolecules and liposomes may produce aggregates or stacks of membranes,39,90 and finally gelation of the mixtures. Liposome-loaded gels are tentative candidates for the encapsulation and the (slow) release of drugs.91,92 The formation of a network of polymer-bridged membranes is generally accepted to explain the sharp increase in the viscosity upon supplementation of a solution of vesicles with macromolecules (Fig. 4).93–96 In the more dilute solutions investigated, the distance between the vesicles may, however, be larger than the typical radius of the polymers. there is no doubt that the exact origin of viscosification is more complicated and could involve partial disruption of the initial liposomes, and contribution of a polymer–micelle mixed network.93,95 The capacity for drug entrapment and release in such gels would accordingly depend on membrane barriers against free diffusion, rather than encapsulation in liposomes. Alternatively, the conjugation of intact liposomes to a gel phase such as Sepharose is investigated for chromatography across modified phases that contain native integral membrane proteins as functional receptors.97,98 The function of the macromolecules in this case is to immobilize the vesicles by covalent attachment onto the stationary phase (e.g.disulfide bridges). Polymers also act as versatile spacers that prevent the adsorption of lipids in the phase (repulsion of the polymer corona) and preserve the accessibility to small substrates. Hydrophobic anchoring of the polymer is expected to be compatible with the functionality of proteins embedded in immobilized proteoliposomes.97
![]() | ||
Fig. 4 Sketch of the physical gels and polymer–membrane bridges that may form in concentrated mixtures of amphiphilic macromolecules and liposomes. Reprinted with permission from ref. 96, copyright 2005, American Chemical Society. |
We discuss first the non-specific effect brought by repulsions between hydrophilic moieties of anchored chains. It is important to determine the magnitude of free energy added by PEO layers, and whether this could contribute to membrane disruption or not. To this aim, the energetics of polymer coupling with a soft membrane has been extensively examined by Marsh et al.11,32,99 A minimal surface density is needed to bring into “contact” the hydrophilic moieties of the anchored chains and develop significant lateral perturbation. The PEO coils have a typical radius of gyration of 4–11 nm (Rg∼a×N0.6, in the “mushroom” regime with the monomer size a = 0.4 nm and N the number of monomers in the range 50–250).11,32,100 Assuming that the surface per lipid head is Alipid = 70 Å2, the PEO chains are brought into contact (overlap) when their density reaches 0.2–1.5 mol% of anchor (relative to the lipids, Fig. 5). Note that even well below the overlap density, isolated hydrophilic chains repel the membrane. The perturbation by isolated chains has been treated theoretically by many authors.9,10,32,101 In particular, the bending stiffness and spontaneous curvature of the membrane are modified to an extent that does not afford the disruption of membrane (free-energy change < kT per lipid), but would considerably affect the membrane's curvature9,10 and thermal fluctuations.102 Changes in the shape properties are discussed in section 4. Experimentally, Ligoure et al. showed that anchored polymers cause an effective rigidification and a decrease in thickness of surfactant/co-surfactant membranes (see ref. 103 and references therein).
![]() | ||
Fig. 5 The mushroom regime of non-interacting end-grafted polymers (upper panel) and brush regime at higher molar fraction Xp of polymer anchors (lower panel). Reprinted with permission from ref. 32, copyright 2003, Elsevier. |
Most preparations usually contain more than 5 mol% anchors and therefore the PEO chains are close to or above their overlap density and are stretched in the direction perpendicular to the membrane. It is often assumed that the PEO layers have reached the brush regime. Parsegian et al. however noted that for short chains, typically of Mw below 5000 g mol−1, this regime is reached well above the overlap density, above 10–20 mol% of anchors in the membrane.104 Energy terms as given by scaling laws for PEO brushes provide nevertheless a useful order of magnitude.9,11,32 The lateral pressure exerted by polymers on the lipids is written:
Πbrush∼mkTNa2m(X/Alipid)m+1 | (1) |
With X being the molar fraction of anchors in the lipids, N is the average number of monomers per chain, m = 5/6 (scaling theory) or 2/3 (mean field theory).9,11 At a molar fraction of anchors X = 0.2, and with N = 100–200 monomers, a considerable lateral pressure Πbrush (up to 20 mN m−1) contributes to the stretching of the membrane. The free energy evolved upon increasing the surface of the membrane by a few percent (ΠbrushΔAlipid per lipid molecule) is of the order of the free energy of hydrophobic association (ΓΔA2lipid/Alipid with Γ being the oil–water interfacial tension = 50 mJ m−2).40 Therefore, the self-assembly as a flat lamellar phase can be unstable and the membrane would break at high densities of PEO. To decrease its free energy, the membrane may however offer more space to the PEO corona by forming buds (local high positive curvature), or being partially solubilized in mixed micelles. Experimental data match well with the calculated impact of PEO on lamellae–micelle coexistence.99 Upon increasing the amount of PEO in samples, a critical composition is reached, above which the lamellar phase with a low polymer–lipid ratio has the same free energy as a micellar structure with a high fraction of polymer. The destabilization of PEO-decorated bilayers has been reported both with hydrophobically anchored chains99 and electrostatically bound ones.105 The excess free energy due to PEO repulsion also decreases the temperature of fluid–gel transition by 1–2 °C.32,99 Inter-PEO repulsion makes the transition less favourable because of either the lower area of the membrane in the gel phase (increasing density of chain upon phase transition) or the expulsion of the anchors from the gel phase that would increase the density of PEO chains in fluid domains (in equilibrium with a pure gel phase). Experimentally, the temperature of the fluid–gel transition shifts by typically −1 to −2 °C, a result in relative agreement with the model developed by Marsh et al.
Although thermodynamically destabilized, the PEO-coated lipids are kinetically protected against attack from proteins or micelles. Penetration in the brush layer of a micelle of radius R would push away the polymer chains, which requires additional free energy of the order of Πbrush.R2. Needham et al. have estimated the effect of this energy cost on the kinetics of membrane disruption by micelles of a lysolipid.100 The drop in the frequency of contacts between micelles and PEO-coated membranes protects the surface of the lipids. Only the monomers of the lysolipids can spread across the polymer brush, which in practice increases the time required to saturate the membrane with the surfactant, and in turn increases the lifetime of the membrane. Altogether, the results above show that conventional anchor densities of highly hydrophilic chains do not usually break the membrane, but make it more fragile.
In contrast, obvious destabilization and permeabilization have been achieved with stimuli-responsive polymers. These macromolecules contain hydrophobic anchors to bind to the membranes and most importantly are close to bad solvent conditions. A wide variety of stimuli-responsive macromolecular structures have been tested in liposome formulations under the form of amphiphilic copolymers. 8,106,107 The membrane breakage is generally obtained near critical pH or temperature conditions that make the polymer insoluble in bulk water, and would trigger coil–globule transition of the conformation of the chains in solution.8,108 Most typically, pH-responsive macromolecules are hydrophilic at high pH (>7) in a polyanion form (Fig. 1(e,f), as salts of carboxylic acid side groups. Below a critical pH value, or in vivo at the pH of late endosomes, those polymers become abruptly insoluble because they are partially or totally neutralized.8,41,109 On the other hand, temperature-responsive systems contain N-isopropylacrylamide39,110,111 units, or propylene oxide units,73,112 or organophosphazene.113 These monomers become less hydrated at temperatures above 35–40 °C (i.e. the low critical solubility temperature of the corresponding polymer chains, LCST). The principle of permeabilization at critical solubility conditions appears remarkably independent on the chemical structure. Biocompatibility issues have, however, motivated the development of diverse compounds, to provide stability in serum (low adsorption of protein in the “hydrophilic” conditions of high pH or low T),114,115 and efficient haemolytic activities or permeabilization of mammalian cells.116,117,118,119,120 But the exact origin of membrane disruption in vivo is largely conjectural. Within the limits of weak interactions, an analogy should exist with the above mentioned PEO-induced lateral pressure. The switch from the repulsive regime toward a (weak) attractive regime is accordingly expected to stabilize the planar geometry and to “compress” the membrane, making it less permeable.121 The validity of the above theory however fails in poor solvent conditions, when strong attraction between monomers leads to chain collapse and the formation of globular aggregates. It is likely that insoluble segments of the chains penetrate inside the bilayers and introduce defects in their organization by a deep inclusion of ethoxy or carboxy groups in the apolar lipid core.122,123 This mechanism obviously differs from the (weak) destabilization of the membrane achieved by hydrophilic coronae of chains. In practice, it has been observed that flat membrane fragments, with disk-like shapes and radii below 40 nm, can be obtained by disruption of small vesicles.124,125Polymers forming micelles in water are presumably capable of stabilizing the rim of fragments of lipid bilayers with local curvatures below 3–4 nm. Compared to surfactant-induced disruptions, slow kinetics of reorganization (often days long) can be observed close to the critical conditions for disruption. The slow rates may be due to a slow adsorption and reorganization of the chains.31 (the slow kinetics of adsorption due to repulsions between polymers is now well documented).126,127 Even after several days incubation, the equilibrium sorption of charged amphiphilic polyelectrolytes is not reached on some hydrophobic surfaces.
Transient local budding and tubulation have been observed by Tsafrir et al. in floppy oblate lipid vesicles prepared by the swelling of a lipid reservoir to obtain membrane with zero tension.131 Injection of a weakly hydrophobic dextran in the vicinity of the vesicle triggers the growth of long tubes (Fig. 6(a)). Shrinkage and re-absorption of the tubes immediately follow the end of supplementation with polymers, which is consistent with an adsorption-induced spontaneous curvature followed by diffusive homogenization of the polymer concentration in the reservoir of lipids. In the case of isolated giant vesicles close to spherical shapes, amphiphilic macromolecules have also been shown to trigger the formation of persistent buds and tubes, even in the absence of a concentration gradient in the outer medium (Fig. 6(b)).132 The injection of chains that adsorb onto the vesicle do not afford a good control of the polymer density and eventually homogeneity. To overcome this experimental limit, Nikolov et al. have recently proposed a three-stage construct of membrane associates with long DNA molecules, via tight but non-covalent associations on biotinylated lipid heads.9 Their results confirmed the polymer-induced budding, even at rather low surface density, in the mushroom regime with essentially isolated chains on the membrane. The shape instabilities as triggered by polymer adsorption depend on the initial geometry of the vesicles. Hollow tubular vesicles relax the trend to form curved structures by the so called “pearling” instability, which results in a string of smaller spherical vesicles connected by extremely narrow necks (Fig. 6(d)).133 Multilamellar tubes of stacked membranes slowly form coils (single strand or double strands) upon swelling in the presence of amphiphilic polymers (Fig. 6(c)).134 The relaxation toward these geometries can be accounted for within a framework in which polymer chains affect the curvature of the bilayers. However, the obviously large range of curvatures present in some geometries (e.g. heterogeneous pearl radii) suggests the additional requirement for large gradients of the polymer surface density. The coiling of multilamellar tubes can also be explained by a transversal gradient of polymer concentration across the stack of membranes.134
![]() | ||
Fig. 6 Typical shapes formed upon adsorption of amphiphilic polymers on lipid membranes: (a) transient tube growth/resorption from the rim of an oblate vesicle connected to a reservoir of lipids, scale bar = 10 μm (reprinted with permission from ref. 131, copyright, APS 2003). (b) Day-long stable buds and fluctuating tubes protruding from a giant unilamellar vesicle (reprinted with permission from ref. 132, copyright 2002, American Chemical Society, vesicle diameter = 40 μm). (c) Coiling of multilamellar stacks of membranes swollen in a solution of polymer, scale bar = 20 μm (with permission from ref. 134, copyright 2003, APS). (d) Pearling instability of a tube connected to a reservoir of lipids, the left-hand “pearl” diameter is = 20 μm (with permission from ref. 133, copyright 2001, American Physical Society). |
At low polymer coverage, the curvature can be induced by two mechanisms:135–137 i) an effective repulsion between the polymer segments and the bilayer over distances of the order of the polymer radius, the bending energy of the membrane balances the gain in free energy of a polymer chain that is allowed more conformational freedom when the impermeable membrane is pushed away; ii) an increase in the area of the anchor(s)-containing leaflet with respect to the other. Both effects i) and ii) result in an increase of the spontaneous curvature of the leaflet on which the polymer is stuck (Fig. 7). Theoretical treatment of the polymer as a perturbation of the membrane mechanical properties predicts in addition that effect i) increases the bending stiffness.10,101,102 This would lead in practice to increase of the membrane curvature, while decreasing the thermal fluctuations (more rigid membranes) as observed and quantified by Nikolov et al.9 The magnitude of variation of the bending rigidity can be surprisingly high and clearly beyond small perturbations effects (increase of the modulus by more than 10 kT)9,103 Theories capture nevertheless correct trends with concentration and polymer size. At high coverage, the lateral pressure due to interpolymer interactions dominates, and an increase in the outer surface of the membrane with increasing the curvature relaxes the excess free energy brought by the effective interchain repulsions.138 Long range lateral interactions between different polymers are also mediated by the membrane deformation. The pair potential due to local membrane distortion close to the anchors is attractive for chains grafted on the same side of the membrane,139 and may favor the lateral segregation of the chains.140 Accounting for thermal modes of fluctuation of both polymer conformations and membrane bends, theoretical approaches (sp. by Lipowsky et al. and Auth and Gompper) predict the order of magnitude of the observed curvatures and the possible shapes of vesicles.10,101,141,142 Finally, the phenomena are mostly driven by repulsions between the hydrophilic segments of the chains and an impenetrable soft membrane, which in turn triggers the membrane's protrusions toward the side of higher polymer density.
![]() | ||
Fig. 7 Schematic representation of the origin of polymer-coupled curvature of a membrane: (a) increase of the spontaneous curvature due to the gain in conformation freedom of the hydrophilic corona; (b) swelling of the outer leaflet by hydrophobic inclusions, which is balanced by the bending of both leaflets in order to match their surface of contact. |
Transverse density gradients of the polymer coat has accordingly a marked impact on curvature and it has been shown that when liposomes are prepared from a mixture of polymer and lipids at a particular ratio, a spontaneous vesiculation occurs with a well-defined radius and a lower internal polymer density.143,144 The contribution of hydrophobic anchors plays a marginal role by (slightly) increasing the volume of a leaflet, but the local structure, and homogeneity of the bilayer is essentially not modified provided that the macromolecules bear a low density of hydrophobic side groups. In contrast, if attractive forces drive the adsorption of most segments of the polymer chains, one would expect membrane invaginations. Angelova et al. showed that coulombic binding of DNA and other synthetic polyanions onto positively charged giant vesicles (sphingosine-containing membranes) induces endocytosis-like phenomena55,145 In this case, the mechanism involves the formation of lateral domains whose composition differs from the average one, and surface and charge asymmetries between the two leaflets.
![]() | ||
Fig. 8 Well-defined channels formed in the presence of amphiphilic polyacrylate (a) schematic drawing of transversal pores whose size depends on ionic strength conditions and polymer density (from ref. 21, copyright 2007, Royal Society of Chemistry ); (b) discrete current steps recorded by patch-clamp experiments in the presence of poly(2-ethylacrylic acid) at low pH (reprinted with permission from ref. 173, copyright 1996, American Chemical Society) |
The origin of the stability of macromolecule-containing pores is still debated, specifically the stabilization of a well-defined radius. A common feature of all pore-forming molecules is the presence of a significant fraction of lipo-soluble segments and their capacity to promote a curvature strain at the rim of pores (i.e. an effective decrease of the line tension, which stabilizes the pores).148 The free energy of a lipid pore combines the gain of relaxing the tension of the membrane over the area of the pore with the energy needed to create the rim of a pore of radius R:
Fpore = 2ΠRγ−ΠR2σ. | (2) |
Where γ is the line tension (of the order of 10−11 J m−1 in pure lipid bilayers, see ref. 40,149 and references therein), and σ the tension of the membrane (ranging from zero to almost 10 mN m−1 in an experimentally stretched membrane close to rupture).150 If both of the tensions are invariant with R, Fpore reaches a maximum at a critical nucleation radius, Rc= γ/σ, and decreases beyond Rc. In other words, small pores are expected to rapidly close, and large ones should grow with no size limit. The presence of inclusions other than lipids contributes to a dependence of the tension parameters on the size of the pores. The additional (internal) tension due to lateral repulsions between inclusions on the membrane favours the formation of a pore (stretching effect) and can be partially decreased by the additional space offered in the pore. For instance, repulsions between charged surfactants is proposed to increase the membrane tension, until the surfactants are gathered in the pore and form a structure with high average curvature and with more accessible space for the cloud of their counter-ions.151,152 In the case of peptides, Lee et al have accounted for inter-peptide repulsions by assuming a linear variation of σ with the radius of the pore, i.e. in proportion to the amount of peptides transferred in the rim of the pore.153 The line tension should similarly reach an optimal value if the average curvature in the pore is matched with the spontaneous curvature of the additives covering the pore edges.151 Amphiphilic polyelectrolytes such as polyacrylates presumably affect the tension parameters at a similar magnitude as do amphiphilic polypeptides. The surface charge density due to such polymers can be as high as that created by charged peptides (assuming a typical density of 1 mg m−2 of macromolecules)132 and they share with biocide peptides the ability to decrease the line tension and to self-assemble in paucimolecular particles with high positive curvature.154 Although the origin of pore stability is not elucidated, it appears clear that non-specific parameters as used in eqn (2) should capture the main features of synthetic pores, irrespective of the molecular arrangements and possible polydispersity of the macromolecular inclusions.
We focus now on the case of tight attractions occurring in the polar part of the membranes, on the hydrophilic heads of the lipids. As pointed in section 1, coulombic attraction with charged lipid heads is a common feature of such a case, both with synthetic polymers and proteins or peptides. The electrostatic affinity of peptide and proteins for oppositely charged lipids have been analyzed in great details showing that multicharged peptides may trigger sequestration of lipids and lateral domain formation (e.g. by McLaughlin et al.).59,155,156 The organization in the apolar region though not directly affected may betray the modification in polar headgroups, since a relative increase of head–head attraction compared to the thermal energy favours the gel phase of the lipid layer, and more generally triggers phase transitions in membranes. The neutralization of ionic lipids with polymers of opposite charge induces accordingly a significant rise in the temperature of fluid–gel transition. Experimental signatures of the conversion of a fluid state to a patchy solid state have been found since 1978 in poly(lysine)-coated small vesicles.157 Shifts in the melting temperature of faceted catanionic vesicles (mixture of anionic and cationic single-tailed surfactants) have been reported in the presence of an amphiphilic polyanion, but not with an hydrophilic polyanion.93,158,159 When anionic phospholipids membranes are mixed with a polycation, calorimetric measurements clearly demonstrate that the binding causes domains (so called “rafts”) of anionic lipids to form in the outer leaflet, and it is these domains that electrostatically bind the polymer (Fig. 9).50,60,160 It is possible to regulate the trend for clustering of the ionized lipids by using polyanions attached to a voluminous neutral block. An increase in the lateral pressure (cf. supra) due to the bulky neutral block balances the trend for anion–cation complexation.161 In other cases, the adsorption of a polyelectrolyte increases the local density of lipids with the opposite charge, which eventually generates complexes that are not in a lamellar phase. For example, hexagonal phases of DNA and cationic lipids have been extensively characterized162,163 but similar phases were also obtained with more flexible polyanions. Vivares and Ramos164 have shown that the formation of such dense complexes stretches the membrane of a multilamellar vesicle (made of didodecyl dimethyl ammonium cations) beyond the critical tension that can open a pore. Further complexation then rapidly peels off the stack of membranes, one by one.
![]() | ||
Fig. 9 Rigid domains formed by coulombic association of a polyelectrolyte on a fluid membrane. Ionized lipids are concentrated in the vicinity of the adsorbed chain(s) by lateral diffusion, and in some cases translocation. |
Other interactions in the polar region that are not of a coulombic nature also trigger similar formation of domains and phase transitions. Hydrogen bonds with the phosphorylcholine heads and/or dehydration are presumably involved in the shift of fluid–gel temperature achieved in the presence of poly(methacrylic acid).165Poly(2-ethylacrylate) adsorbed on small lipid vesicles upon a decrease of pH (i.e. under a partially acid form) triggers the formation of domains with a shifted melting temperature as revealed by the presence of two sharp peaks in differential scanning calorimetry.166 We have recently demonstrated that a rise in the melting temperature by almost 20 °C can be achieved by association with hydrophobically modified polyacrylic acid. The domains consisting of fluorescent and non-fluorescent regions were observed by optical microscopy in giant vesicles.132 Their melting above 25 °C, and several hours of reconstitution at 8 °C was also confirmed by the same technique with various fluorophores-labelled lipids (Fig. 10).167
![]() | ||
Fig. 10 Polymer-induced domains on egg-PC giant unilamellar vesicles (a) upper view of a faceted vesicle with fluorescent lipid probes segregated in the fluid regions; (b) diametral view showing the faceted shape. (reprinted with permission from ref. 132, copyright 2002, American Chemical Society) |
Lateral head–head interactions play an essential role in the homogeneity of mixed membranes that contain different lipids. In the simple case of binary mixtures of lipids, segregation effects and clustering have been triggered by coupling lipid head groups with bulky hydrophilic moieties.18 The lateral diffusivity of lipids enables rapid variation of the local composition, which in turn switches cooperative transition and formation of “raft-like” domains. For instance, the formation of highly charged patches efficiently strengthen the binding of polyelectrolytes; at low average charge densities, it has been accordingly reported that only fluid liposomes were able to form an association with polymers of the opposite charge, whereas below the temperature of fluid–gel transition, no association took place.168 The growth of a polymer-induced domain has also been proposed as a generic driving force for fusion between two vesicles brought into contact.169 When depletion forces have expelled the polymer out of the contact, the adsorption and domain formation at the opposite pole of a vesicle would attract lipids of the outer leaflets (marangoni flow) and weaken lipid–lipid cohesion in the contact zone. Profound change in membrane features are expected as a consequence of patch and domain formation and shape modification. Lipowsky and Dimova have formalized the geometric and energetic constraints applied to a vesicle whose membrane becomes laterally heterogeneous.170 For instance, when a giant vesicle develops facets, the fluid part of the membrane develops tension in order to accommodate the shape variation at constant inner volume. In vesicles initially close to a spherical shape, the increase of tension would stop the phase separation (frustration) and favour multiple faceting. On polymer-coated vesicles, the formation of 4–6 facets was the most common feature (Fig. 10).
An intriguing effect observed in polymer-controlled domains was the strong inter-leaflet coupling of lipid composition.132 Fluorescent-labelled lipids are similarly and simultaneously depleted from both the inner and outer leaflet by the growth of label-enriched facets, despite the fact that the polymer did not cross the membrane.31 Transbilayer coupling drives the co-location of domains formed in both leaflets of a giant vesicle.171 Similar coupling, so-called “domain registration”, is observed on polymer-tethered membranes, as well as in fluid membranes (liquid ordered–liquid disordered phase separation in the presence of cholesterol).172 Through building asymmetric bilayers, Naumann et al. have shown in this system the absence of flip-flop and domain induction in the upper leaflet by domains present in the lower leaflet. The transversal coupling occurs even if the upper composition should not be thermodynamically biphasic. The only requirement for transversal coupling was the possibility for domains to diffuse freely in both leaflets as suggested by the loss of coupling when the lipids interact with a glassy substrate (across a thin polymer tether).172 As a possible origin of strong inter-leaflet coupling, even in pure lipid layers, Lipowsky and Dimova suggest that a transverse bi-phasic organization would display a transverse gradient of the packing density of lipids, which creates tension and a driving force for co-localization.170
This journal is © The Royal Society of Chemistry 2008 |