Lateral reorganization of myelin lipid domains by myelin basic protein studied at the air–water interface

Available online at www.sciencedirect.com

Colloids and Surfaces B: Biointerfaces 62 (2008) 22–30

Lateral reorganization of myelin lipid domains by myelin basic protein studied at the air–water interface Yufang Hu a,∗ , Jacob Israelachvili b a

Department of Molecular and Medical Pharmacology and Department of Urology, David Geffen School of Medicine at UCLA, Los Angeles, CA 90095, United States b Department of Chemical Engineering, University of California, Santa Barbara, CA 93106, United States Received 29 June 2006; received in revised form 10 September 2007; accepted 10 September 2007 Available online 29 September 2007

Abstract It has been speculated that adsorption of myelin basic protein (MBP) to the myelin lipid membrane leads to lateral reorganization of the lipid molecules within the myelin membrane. This hypothesis was tested in this study by surface pressure measurement and fluorescent imaging of a monolayer composed of a myelin lipid mixture. The properties of the lipid monolayer before and after addition of MBP into the subphase were monitored. Upon addition of MBP to the monolayer subphase, the surface pressure rose and significant rearrangement of the lipid domains was observed. These results suggest that binding and partial insertion of MBP into the lipid monolayer led to dramatic rearrangement and morphological changes of the lipid domains. A model of adsorption of MBP to the lipid domains and subsequent domain fusion promoted by minimization of electrostatic repulsion between the domains was proposed to account for the experimental observations. The significance of these results in light of the role of MBP in maintaining the myelin structural integrity is discussed. © 2007 Elsevier B.V. All rights reserved. Keywords: Myelin basic protein; Lipid monolayer; Lipid domains; Protein–lipid interactions; Fluorescence microscopy

1. Introduction Myelin basic protein (MBP) is an integral component of the myelin sheath in the central nervous system (CNS). It is the second most abundant protein in the CNS, comprising roughly 30% of the proteins found in CNS myelin. MBP is primarily located in the cytoplasmatic spacing of the myelin sheath and is believed to play an active role in stabilizing the periodic myelin structure via non-specific interactions with the apposing lipid bilayers [1–3]. MBP is antigenic and causes a demyelination condition in animal models of multiple sclerosis (MS) [4]. In solution, MBP alone behaves as a random coil without any apparent secondary structures [5]. Although a great deal of effort has been devoted to study the secondary structure of MBP in the presence of surfactants and lipids (see review by Mendz [5]), the results from different studies are not consistent because the MBP folding is strongly dependent on its surrounding environment. It is gener-

∗

Corresponding author. Tel.: +1 310 825 8511; fax: +1 310 825 3027. E-mail address: [email protected] (Y. Hu).

0927-7765/$ – see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.colsurfb.2007.09.028

ally believed, however, that MBP folds into structures containing a higher degree of secondary structures such as ␣-helices and ␤-sheets in the presence of amphiphiles compared to its folding in the absence of amphiphiles. It has been established by numerous experimental studies that MBP interacts strongly with acidic lipids via electrostatic interactions [6]. Under physiological conditions, the 18.5 kDa MBP carries 21 net positive charges. Investigations of the role of MBP in promoting myelin sheath integrity have largely been focusing on its role in promoting inter-membrane adhesion via its interactions with acidic lipid headgroups in apposing membrane surfaces [1,7,8]. Understood to a lesser degree is the role of MBP in creating lateral intra-membrane heterogeneity. It has been suggested that the lipids in the myelin membrane form microscopic domains around MBP [6,9] and that MBP partially inserts into myelin membrane via hydrophobic interactions with the lipid hydrocarbon tails [10]. These two effects may act in concert to promote strong binding of MBP to the myelin membrane: the insertion of MBP into the membrane could induce a localized rearrangement of the lipid molecules, which then form microdomains around the protein. These microdomains formed

Y. Hu, J. Israelachvili / Colloids and Surfaces B: Biointerfaces 62 (2008) 22–30

by myelin lipids surrounding the MBP are analogous to membrane rafts, which have been suggested to exist in a wide range of biological systems [11,12]. Several Langmuir monolayer studies of the interactions between MBP and various model myelin lipid mixtures have been reported in literature [13–18]. The kinetics of MBP binding to dipalmitoyl phosphatidylcholine and dipalmitoyl phosphatidylserine monolayers have been measured [18]. The insertion of MBP into tissue-extracted myelin lipid mixtures revealed surface pressure dependent insertion behavior [13,15,19]. In an attempt to probe the abnormalities in myelin lipids and MBP in control and MS samples, the phase behaviors of MS myelin lipids and normal myelin lipids at the air–water interface were compared [16]. Most recently, the behavior of MBP in modulating the surface viscosity of myelin lipid monolayer was reported [20]. Although it has long been recognized that the interaction of MBP to myelin lipids contain electrostatic and hydrophobic components, the precise location of MBP within the myelin sheath remains to be determined. It is conceivable that nonspecific interactions between MBP and myelin lipids occur on the length scale of molecular dimensions and are difficult to probe experimentally. Using a monolayer of myelin lipids at the air–water interface to mimic the actual myelin membrane, we attempted to characterize the role of MBP to create lateral lipid compositional heterogeneity within the myelin lipid membrane. Surface pressure and monolayer domain morphology were measured up to a few hours with and without the addition of MBP to the subphase. A simple theoretical model, which is largely based on electrostatic interactions between the dipolar lipid molecules and the charged MBP, was applied to interpret the experimentally observed domain attraction at early times during the monolayer transition. We believe that our measurement of MBP interaction with the myelin lipid monolayer provides a macroscopic proof to the hypothesis of the lateral MBP/myelin lipid interactions within the myelin membrane. The results from the combined surface pressure and fluorescence microscopy measurements presented in this paper demonstrated that the binding and insertion of MBP to the myelin lipid monolayer and subsequent lateral rearrangement of the myelin lipid domains occurred via non-specific interactions of the MBP with the lipids. To our knowledge, this is the first time direct visualization of the lateral rearrangement of myelin lipid monolayer domains due to MBP binding has been reported. 2. Materials and methods 2.1. Materials Details of the preparation and sources of the lipid components making up the myelin lipid mixtures used in the study have been given elsewhere [1] and will not be repeated here. Briefly, seven different classes of tissueextracted lipids were mixed to mimic the experimentally measured lipid compositions in human myelin [21] as shown in Table 1. The fluorescent dye 1,2-dihexadecanoyl-sn-glycero-3phosphoethanolamine, triethylammonium salt (Texas Red-PE® )

23

Table 1 Composition of myelin lipid mixture Lipid mole fraction Phosphatidylcholine (PC) Phosphatidylserine (PS) Phosphatidylethanolamine (PE) Sphingomyelin (SM) Cholesterol (Chol) Cerebrosides (CE) Cerebroside sulfate (CBS)

0.109 0.011 0.156 0.079 0.491 0.108 0.046

was purchased from Molecular Probes (Eugene, OR) and was mixed with the myelin lipid mixture at 1% mole fraction. The 18.5 kDa MBP C1 was kindly provided by Dr. Mario Moscarello’s laboratory at the University of Toronto, which was purified according to previously published protocol [22]. The lyophilized form of the protein was stored at −20 ◦ C until use. For all measurements, MBP was dissolved in MOPS buffer (150 mM sodium nitrate, 10 mM MOPS sodium salt, pH 7.4). The MOPS buffer was made with Milli-Q water, which was purified by a MilliPore Grandiant A10 system (Bedford, MA) and had resistivity ≥18.2 M cm and total organic content ≤5 ppb. All measurements were carried out using MOPS buffer as the subphase. The chemicals were used as received without further purification. The measurements were performed separately on two homemade troughs: one for myelin lipids pressure–area isotherm measurements [23] and one for myelin lipids–MBP interactions [24]. A Nikon Eclipse E800 fluorescence microscope was used for fluorescence microscopy imaging (Nikon Instrument Group, Melville, NY). The fluorescence images from the microscope were fed to a Cohu image intensified CCD camera (San Diego, CA), which was directly connected to the microscope. The black-and-white fluorescence images were recorded using a Panasonic super VHS VCR (Secaucus, NJ). Video images were digitized on a PC installed with ATI 128 PCI 32 MB video card and All-In-Wonder software (Ontario, Canada). 2.2. Probing the MBP–myelin lipid interactions All experiments involving MBP were done in a miniature trough [24], which was small enough to rest on the translation stage of Nikon fluorescent microscope. Due to instrumentation limitations, no simultaneous measurements of monolayer surface pressure and fluorescence microscopy were conducted. Instead, each experiment was divided into two stages: (1) monitoring of surface pressure during MBP insertion into myelin lipid monolayer and (2) fluorescence microscopy of monolayer morphology. The myelin lipid monolayer was first compressed to the desired surface pressure of ∼25 mN/m and allowed to equilibrate while the monolayer surface pressure was measured by a portable Nima pressure sensor (Coventry, UK). Once the surface pressure was stabilized, ∼100 ␮L of MBP solution (concentration 0.6 mg/mL) was slowly injected into the subphase. The bulk concentration of MBP in subphase was 0.1 ␮M. The surface pressure was continuously recorded during protein injection and

24

Y. Hu, J. Israelachvili / Colloids and Surfaces B: Biointerfaces 62 (2008) 22–30

Fig. 1. Schematic of the minitrough and portable pressure sensor setup for measurement of the surface pressure during MBP adsorption to the myelin lipid monolayer.

the subsequent equilibration period as the MBP adsorbed onto the lipid monolayer and interacted with the lipid molecules. The recording was terminated once the surface pressure became stable. The pressure sensor was then removed and the trough was transferred onto the fluorescence microscope translation stage for fluorescence microscopy imaging. A schematic illustration of the experimental setup used for surface pressure measurement and fluorescent imaging is given in Fig. 1; and that of the minitrough with fluorescent microscope was given elsewhere [24]. During the surface pressure measurement, the minitrough was placed on top of a magnetic stirring plate with tunable stirring rate. A small Teflon-coated magnetic stir bar was used to facilitate mixing of the subphase contents upon protein injection. The rate of stirring was kept constant for all experiments. The trough was housed in a home-built Plexiglas enclosure to reduce the disturbance on the pressure sensor by ambient airflow. 3. Results 3.1. Properties of the myelin lipid monolayer The -A isotherm of the myelin lipid monolayer on MOPS buffer is shown in Fig. 2, along with morphology of lipid domains at different surface pressures obtained by fluorescence microscopy. The isotherm appeared to be smooth and was absent of any kinks or steps. At low to intermediate surface pressures, circular lipid domains tens  of microns in size were observed, which melted away at > 30 mN/m. Beyond the transitional surface pressure of 30 mN/m, the monolayer exist in a homogeneous one-phase state. The monolayer collapse occurred at ∼42 mN/m. At intermediate surface pressures circular, non-interacting polydisperse lipid domains could be seen, which were stable for up to 12 h. During this time, the domain shape remained circular and the domains stayed separated from one another without coalescence or fusion. The behavior of these circular myelin lipid domains bears strong resemblance to that observed in the dimyristoyl-phosphatidyl-choline and dihydrocholesterol mixed monolayer in the pseudo phase-coexistence regime [25,26], where the appearance of stable circular domains is due to the balance between attractive line tension and electro-

Fig. 2. Pressure–area isotherm of the myelin lipids and the monolayer morphology at different surface pressure regimes. Circular, non-interacting lipid domains could be seen at intermediate surface pressures. The domains dissolved and the monolayer became homogenous at surface pressures greater than ∼30 mN/m.

static dipolar repulsion [27,28]. Since zwiiterionic lipids such as phosphatidyl-choline were heavily represented in the myelin lipid mixtures (cf. Table 1), we believe the observed circular morphology of the myelin lipid domains was controlled by the same types of line and dipolar forces. 3.2. Interaction of the myelin lipid monolayer and MBP Upon injection of MBP into the monolayer subphase, an immediate increase in surface pressure was observed (Fig. 4). The surface pressure rose from 24.7 to 28.4 mN/m over the course of ∼550 s. An increase in surface pressure due to insertion of soluble proteins into lipid monolayer has been reported for a number of systems such as fribronectin [29], cytochrome C [30], and more significantly, MBP to monolayer of total myelin lipid extract [17]. It is acknowledged that such a surface pressure increase signals the binding or insertion of protein into the monolayer via non-specific interactions such as hydrophobic, depletion, or electrostatic forces.

Y. Hu, J. Israelachvili / Colloids and Surfaces B: Biointerfaces 62 (2008) 22–30

25

Once the surface pressure became stabilized, the pressure sensor was carefully removed from the air–water interface and the minitrough was transferred onto the microscope stage to begin fluorescence microscopy measurements. Measurements of two separate samples with and without presence of the surface pressure sensor showed that disturbance of the monolayer caused by pressure sensor removal did not lead to significant changes in monolayer morphology as the lipid domains appeared to maintain the same shape. In contrast to the monolayer morphology in absence of MBP (cf. Fig. 3), dramatic lipid domain morphological changes were observed shortly after injection of

Fig. 3. The myelin lipid monolayer alone without MBP. (a) Immediately after monolayer spreading. The streaking seen in the image was due to chloroform evaporation during monolayer deposition; (b) 5 h after monolayer spreading. No domain fusion or aggregation was observed. The scale bar is 100 ␮M.

Fig. 4. Surface pressure measured during MBP binding and insertion to myelin lipid monolayer. The monolayer was allowed to equilibrate at 25 mN/m for 1500 s prior to MBP injection. A surface pressure increase of 3.7 mN/m was measured.

Fig. 5. (a–c) Representative images of the myelin lipid domains taken after MBP injection. Significant aggregation and fusion of the lipid domains were observed, accompanied by domain shape transition into irregular, non-circular shapes. The scale bar is 100 ␮M.

MBP to the subphase (Fig. 5). The most striking feature was the emergence of extended and irregularly shaped domains reaching several hundred microns in size (Fig. 5a), the morphology of which were significantly different from the stable, circular lipid-only domains seen in Fig. 3, taken in the absence of MBP. Evidence of early stages of domain shape change such as deviation from circular domain shape and domain aggregation could be seen in some of the images (Fig. 5b). In addition to lipid domain fusion, extended networks of a necklace-like alignment of circular domains were also observed (Fig. 5c). To probe the dynamics of the domain evolution, fluorescence imaging of the domains carried out at different elapsed times after MBP injection are shown in Figs. 6 and 7. At 30 min after MBP injection, alteration in the spatial distribution pattern of the lipid domains was already clearly visible (Fig. 6). In contrast to the stable, repulsive domains seen in the absence of MBP, the domains now appeared attractive. Clusters of domains formed along with occasional fusion of the domains, as highlighted by the arrows. At this time point, the majority of the domains still

26

Y. Hu, J. Israelachvili / Colloids and Surfaces B: Biointerfaces 62 (2008) 22–30

Fig. 7. A model of dipole–dipole and charge–dipole interactions to account for the observed attractions between domains shortly after MBP binding to the monolayer. (a, b) Configuration of the domains and MBP: (a) top view; (b) side view. The lipid molecules are modeled as molecular dipoles oriented perpendicular to the air–aqueous interface. (c) Top view of the linearly aligned domains and MBP.

Fig. 6. Time-lapse fluorescent images of myelin monolayer with absorbed MBP taken at (a and b) 30 min after MBP injection. The domains were no longer repulsive and domain aggregation and fusion could be found throughout the monolayer area, as highlighted by arrows. (c) 2 h after MBP injection. Extensive domain fusion and shape transition have taken place. (d) After the monolayer was fully expanded. The scale bar is 100 ␮M.

appeared circular. However, at 2 h post protein injection, extensive fusion and shape transition of the lipid domains could be seen (Fig. 6c). These measurements showed that compared to the relatively fast MBP binding and insertion into the monolayer, which took place within minutes after the MBP injection (Fig. 4); the sub-

sequent rearrangement of the lipid domains occurred on a much slower time scale (Fig. 6). Expansion of the protein–lipid mixed monolayer to maximum trough area did not reverse the morphology of the fused domains (Fig. 6d), suggesting irreversible binding of MBP to the lipid domains. These fluorescence images clearly demonstrated the lateral rearrangement of myelin lipid monolayer domains caused by MBP adsorption to the monolayer. In the absence of MBP the domains were separated by inter-domain electrostatic repulsion, which was electrostatic in nature due to the presence of anionic and dipolar lipids in the lipid mixture [31,32]. The resulting circular monolayer morphology can be viewed as a balance of this repulsive electrostatic dipolar interaction, which favors domain separation, and attractive line tension, which maintains the circular domain shape [27,28,33,34]. In the following section, the impact of MBP adsorption to the lipid domains on monolayer morphology will be discussed in the context of MBP-domain electrostatic interactions. 4. Discussion and conclusion The surface pressure increase recorded in our study is in good agreement with previously published results by Demel et al. [17]. In a thorough study of the binding and insertion of MBP to different monolayers of lipids found in CNS myelin, these authors found that MBP bound strongly to negatively charged myelin lipids such as cerebroside sulfate and less effectively to neutral lipids such as PC [17]. Further, surface pressure increases at constant total monolayer area as MBP binding and insertion

Y. Hu, J. Israelachvili / Colloids and Surfaces B: Biointerfaces 62 (2008) 22–30

were observed for both anionic and neutral myelin lipid monolayers with the later yielding smaller level of surface pressure rise. At initial surface pressure of ∼25 mN/m, the recorded surface pressure increase due to MBP binding was ∼1 mN/m for monolayer of PC alone, compared to ∼7 mN/m for total myelin lipid extract, and ∼12 mN/m for cerebroside sulfate alone. In our study, an increase in surface pressure of 3.7 mN/m was measured with an initial surface pressure of 24.7 mN/m, which was comparable to that of total myelin lipid extract obtained by Demel et al. The agreement reflected the compositional similarity between the myelin lipid mixture used in our study and the myelin total lipid extract used in the study of Demel et al., who further suggested that the measured increase in surface pressure immediately after MBP injection was indicative of the insertion of the protein into the lipid monolayer. Several mechanisms are likely to account for this initial binding of protein to the monolayer such as hydrophobic interactions between the hydrophobic residues on MBP and lipid hydrocarbon tails; electrostatic attraction between MBP and the anionic lipids; or preferential adsorption of MBP to the domain edge boundary due to depletion interactions. All of these mechanisms have been reported to drive protein binding to lipid monolayers at the air–water interface [29,35,36]. The surface pressure increase was attributed to the insertion of MBP into the lipid monolayer driven mainly by hydrophobic interactions between MBP and the hydrocarbon tails of the lipids. MBP contains 24% hydrophobic amino acids, which are randomly distributed along the protein backbone [6]. Demel et al. demonstrated that no surface pressure increase was observed when polylysine was added to the subphase of a myelin acidic lipid extract monolayer, and that the strength of MBP interaction with lipid monolayer was sensitive to the hydrocarbon tail length. A similar conclusion was reached in a study of MBP adsorption to lipid vesicles [10]. In addition to inducing surface pressure increase, adsorbed MBP mediates morphological changes of the monolayer. As illustrated in Fig. 6, the sequence of events begins with attractions between the previously repulsive domains, and ends with domain fusion and shape transformations. Initially, the domains appear to migrate towards each other under the influence of some attractive interactions. We hypothesize that the adsorbed MBP in- and outside of the domain are the sources of this attractive interactions. Specifically, we believe this MBP-mediated domain attraction is driven by the electrostatic interactions between the MBP and the dipolar lipids in the domains. These myelin lipid domains are enriched with zwitterionic lipids carrying intrinsic dipoles (e.g., posphatidylcholine), which are oriented with their headgroup immersed in the aqueous subphase and the hydrocarbon tail pointing into air. The effective dipole moment of a single, isolated dipole domain is the vectorial sum of the vertical components of the molecular dipoles moments in the domain, which include contributions of the charges in the lipid headgroup region and the electric double layer [36,37]. At separation distance much greater than molecular dimensions, the effective dipole moment of a single domain can be viewed essentially as that of a macrodipole [36]. This macrodipole approximation has been successfully applied in several studies of mixed monolayers [28,34,36]. In this work,

27

we assume the effective dipole moment of the domain is a macrodipole oriented perpendicular to the air–aqueous interface with its dipole moment m = qL, where q is the effective charge and L the charge separation distance. For simplicity, the MBP is modeled as a point charge, Q. One can consider a simple case where two domains and a MBP molecule are configured as shown in Fig. 7a and b, where 0 ≤ θ, β < π. Assume for now there is no MBP absorbed onto the domains—this point will be revisited later. The total interaction energy of the domains and MBP is the sum of two charge–dipole and one dipole–dipole interaction energies: Etotal = EMBP−domain1 + EMBP−domain2 + Edomain1−domain2 . (1) The first two terms in Eq. (1) are essentially charge–dipole interaction energies and the last term is a dipole–dipole interaction energy. Eq. (1) can be written using molecular parameters as: m 1 m2 Qm1 cos θ Qm2 cos γ − − . ETotal = 3 4πεε0 r32 4πεε0 r42 4πεε0 (r1 cos α+r2 cos β) (2) where m1 and m2 are the effective dipole moments of the domains; α, β, γ (or θ) are the angles as defined in Fig. 7a and b; r1 , r2 are the separation distances between the MBP and the domains; r3 , r4 are the distances from the MBP to the mid points of the effective dipole moments m1 and m2 , respectively; and finally, ␧0 and ␧ are the electric permittivity in vacuum and in water. For sake of simplicity, assume that the two circular domains in Fig. 7 are equal in their size and molecular composition. Hence, the equivalent dipole moments carried by the two domains are identical: m1 =m2 =m,

r1 =r2 =r,

θ=γ=η,

α=β=δ.

(3)

2Qm cos η(1 − cos2 η) . 4πεε0 r 2

(4)

and

Eq. (2) can be rewritten as: ETotal =

m2 4πεε0

r 3 (2 cos δ)3

−

ETotal can be minimized by varying the relative positions of the domains and MBP, thus changing the values of the parameters δ, η, r. The first term representing the dipole–dipole interaction energy is minimized when cos δ = 1 or δ = 0 as shown in Fig. 7c, when the two lipid domains and the MBP are aligned along a straight line with MBP sandwiched between the domains. This reflects the tendency of the two parallel dipoles to maximize their separate distance in order to minimize the inter-dipolar repulsion, which scales as 1/r3 . Hence the linear alignment of two charged dipolar domains, which uses MBP as an anchoring point to reduce the dipolar repulsion between the domains, is energetically favorable. This mechanism of the MBP-mediated alignment of lipid domains is analogous to the bridging of charged monolayer domains by oppositely charged microbeads as observed by Nassoy et al. [36]. The microbeads were seen to adsorb onto the domain boundary via a combination of depletion and electrostatic forces, and served as anchoring points for the

28

Y. Hu, J. Israelachvili / Colloids and Surfaces B: Biointerfaces 62 (2008) 22–30

aggregation of two adjacent circular domains. Such linear alignments of repulsive aggregates mediated by attractive forces have been found in several other systems. A “necklace model” has been used to describe the bead-like conformation of hydrophobic polyelectrolyte chains in aqueous medium, which is a consequence of the balance of electrostatic repulsion and attractive hydrophobic interactions [38,39]. Similar necklace-like structure has been observed in the assembly of charged microtubules by attractive microtubule bundle–counter ion interactions [40]. With the two lipid domains and MBP in linear alignment as shown in Fig. 7c, further physical insights into the MBPmediated domain attraction can be obtained when ETotal is plotted as a function of the domain separation distance, r. Assume r  L, where L is the charge separation distance in the lipid headgroup, let L/2 cos η ∼ , = r

(5)

then Eq. (4) becomes ETotal =

m2 LmQ(1 − L2 /4r 2 ) − . 32πεε0 r 3 4πεε0 r 3

(6)

Assume a typical domain of 5 ␮m in radius and area per lipid ˚ 2 , the effective dipole moment of the domain molecule of 60 A m can be calculated as m = NeL, where N is the total number ˚ and e is the electronic of dipolar lipids in the domain, L = 5 A, charge (1e = 1.6 × 10−19 C). With this formulation, the molecular dipole per lipid is 0.3 Debye (D), which is in good agreement with the experimentally measured vertical molecular dipole moment of 0.15 D for carboxylic acid monolayer [37]. The effective dipole moment of the entire domain is 3.93 × 107 D, with the domain containing 1.3 × 108 lipid molecules. The value of Q is taken to be that of a single MBP molecule with 21 net positive charges: Q = +21e. The total interaction energy as a function of r is given as a solid curve in Fig. 8, which shows that ETotal is repulsive at all separation distances because it is dominated by the large dipolar repulsion between the two domains. The single MBP molecule sandwiched between two lipid domains as shown in Fig. 7 is not sufficient to promote the experimentally observed lipid domain attraction. The interaction energy becomes attractive when the dipolar repulsion is weakened and when the charge–dipole attraction is strengthened. As an example, consider the case where m = 4.0 × 103 D and Q = +21 × 106 e (Fig. 8, dashed curve), for which ETotal is negative at small separation distances and the magnitude of the attraction is greater than thermal energy kT (∼4 × 10−21 J). Therefore, the domains now become attractive. The reduced value of m, however, deserves some clarification: we believe the substantial reduction in the effective dipole moment in the domain is brought about by the absorption of MBP to the lipids inside the domains, which increases the tilt angle of the individual molecular dipoles and reduces the vertical dipole moment (Fig. 8, inserts) [41]. Further, the increased value of Q suggests that adsorbed MBP molecules outside of the domains form aggregates. The values of m and Q chosen above are chosen to illustrate the principle of proteinmediated domain attraction. In reality, the formation of large 3 D MBP aggregate in solution via electrostatic interactions alone is

Fig. 8. Interaction energy as a function of domain separation distance. (a) Repulsive interaction energy (m = 4 × 107 D, Q = +21e), (b) attractive interaction energy (m = 4 × 103 D, Q = +21 × 106 e). Inserts show the molecular details of the lipids and proteins inside the domains for each scenario. Altered orientation of the lipid molecular dipoles due to MBP absorption reduces the effective dipole moment of the lipid domains, which together with MBP aggregates formed outside of the domains result in overall attraction between the domains.

very unlikely because of the highly basic nature of the protein, even when the protein–protein attraction is mediated by counter ions. It is possible that the mechanism of formation of large MBP aggregate in our monolayer model is similar to that proposed by Mueller et al. in their AFM study of the adsorption of MBP onto solid supported lipid bilayers [42], where 2 D MBP domains reaching microns in size were observed. It was believed that randomly adsorbed MBP molecules slowly migrate on the lipid bilayer and fuse into microscopic domains due to strong protein–protein lateral interactions. The model of systematic monolayer morphological change due to MBP–lipid interaction proposed above is in good agreement with that reported in literature, such as that by Cristofolini et al. [19]. These authors studied the adsorption of MBP to the dipalmitoyl phosphatidylglycerol (DPPG) monolayer using synchrotron radiation X-ray reflectivity, and discovered that MBP adsorption to the DPPG monolayer caused significant increase in the monolayer roughness, which the authors stated was the result of a “destructuring” effect of the lipid monolayer due to MBP binding and insertion to the lipid monolayer. It is worth emphasizing that the above model applies only to the early stages of the monolayer morphological transition when the domain shapes are still circular. The domain shape transitions that take place at later times are probably a consequence of the reduction of the domain line tension, whose effect is not included in the model. Based on the above analysis, we propose a mechanism whereby MBP first binds and partially inserts into the lipid monolayer; and subsequently causes rearrangement of the lipid domains due to electrostatic charge–dipole interactions, as illustrated in Fig. 9. Upon injection of MBP to the subphase, binding of MBP to lipid monolayer occurs relatively quickly compared to

Y. Hu, J. Israelachvili / Colloids and Surfaces B: Biointerfaces 62 (2008) 22–30

29

Acknowledgements We are grateful to Dr. Mario Moscarello and Ms. Denise Woods at the University of Toronto Hospital for Sick Children for kindly providing the purified 18.5 kDa MBP and for many helpful discussions. We would like to thank Kieche Meleson for technical assistance during some early measurements during her internship at UCSB; and Dr. Alex Levine at UCLA Department of Chemistry & Biochemistry for helpful discussions on electrostatics. This work was partially supported by the Materials Research Laboratory Program of the National Science Foundation under Award DMR00-80034. YFH acknowledges partial support from the UCLA SOMI postdoctoral fellowship (NCI Cancer Education Grant, R25 CA 098010) and the Susan G. Komen Foundation postdoctoral fellowship (PDF0504500). References

Fig. 9. (a–c) Schematic of the proposed mechanism of lateral lipid domain rearrangement induced by MBP adsorption.

the subsequent reorganization of the lipid domains. The binding of MBP to the lipid monolayer leads to attraction and aggregation of the domains (Fig. 9a). The domains aggregate and fuse while MBP continues to adsorb to the edge of the domains (Fig. 9b). The edge adsorption of MBP to the lipid domains reduces the domain line tension, causing the domain shape transition from circular to the observed extended irregular structures (Fig. 9c). Ideally, this MBP/myelin lipid interactions model should be tested via fluorescently labeled MBP, which can permit direct visualization of the binding of the MBP molecules to lipid domains and tracking of its movement during the subsequent reorganization events. However, in reality it may be a challenge to pinpoint the locations and track the movements of the MBP molecules during each stage of the monolayer transition. MBP has long been suspected to cause membrane heterogeneity upon binding to model membrane systems [43]. Our experimental measurements provide direct visualization of the lateral reorganization of the myelin lipid by MBP via nonspecific interactions with lipids. Further, our model suggests that subsequent heterogeneity of the membrane promotes stabilization of the myelin membrane via reduction of electrostatic dipolar repulsion and strengthening of Coulomb attraction. Thus, the role of MBP in maintaining the integrity of the myelin sheath may not be limited to providing the inter-membrane adhesion as it is generally believed, but may also include intramembrane stabilization as well.

[1] Y.F. Hu, I. Doudevski, D. Wood, M. Moscarello, C. Husted, C. Genain, J.A. Zasadzinski, J. Israelachvili, Proc. Natl. Acad. Sci. U.S.A. 101 (2004) 13466–13471. [2] R.P. Rand, N.L. Fuller, L.J. Lis, Nature 279 (1979) 258–260. [3] R.A. Ridsdale, D.R. Beniac, T.A. Tompkins, M.A. Moscarello, G. Harauz, J. Biol. Chem. 272 (1997) 4269–4275. [4] J. Phillips, Arch. Neurol. 58 (2001) 30–32. [5] G.L. Mendz, Structure and molecular interactions of myelin basic protein and its antigenic peptides, in: R.E. Martenson (Ed.), Myelin – Biology and Chemistry, CRC Press, Boca Raton, FL, 1992. [6] J.M. Boggs, M.A. Moscarello, Biochim. Biophys. Acta 515 (1978) 1–21. [7] J.M. Boggs, G. Rangaraj, K.M. Koshy, J.P. Mueller, Biochim. Biophys. Acta-Biomembr. 1463 (2000) 81–87. [8] J.M. Boggs, M.A. Moscarello, D. Papahadjopoulos, Structural organization of myelin. Role of lipid–protein interactions determined in model systems, in: P. Jost, O.H. Griffith (Eds.), Lipid–Protein Interactions, John Wiley and Sons, New York, NY, 1982, pp. 1–51. [9] D.D. Wood, M.A. Moscarello, J. Membr. Biol. 79 (1984) 195–201. [10] D. Papahadjopoulos, M. Moscarello, E.H. Eylar, T. Isac, Biochimica Et Biophysica Acta 401 (1975) 317–335. [11] E. London, Curr. Opin. Struct. Biol. 12 (2002) 480–486. [12] M. Edidin, Trends Cell Biol. 11 (2001) 492–496. [13] R.G. Oliveira, R.O. Calderon, B. Maggio, Biochim. Biophys. Acta 1370 (1998) 127–137. [14] H. Haas, M. Torrielli, R. Steitz, P. Cavatorta, R. Sorbi, A. Fasano, P. Riccio, A. Gliozzi, Thin Solid Films 329 (1998) 627–631. [15] R.G. Oliveira, B. Maggio, Neurochem. Res. 25 (2000) 77–86. [16] L. Ginsberg, N.L. Gershfeld, Neurosci. Lett. 130 (1991) 133–136. [17] R.A. Demel, Y. London, W.S. Geurts van Kessel, F.G. Vossenberg, L.L. van Deenen, Biochim. Biophys. Acta 311 (1973) 507–519. [18] E. Polverini, S. Arisi, P. Cavatorta, T. Berzina, L. Cristofolini, A. Fasano, P. Riccio, M.P. Fontana, Langmuir 19 (2003) 872–877. [19] L. Cristofolini, M. Fontana, F. Serra, A. Fasano, P. Riccio, O. Konovalov, Eur. Biophys. J. 34 (2005) 1041–1048. [20] Z. Khattari, Y. Ruschel, H.Z. Wen, A. Fischer, T.M. Fischer, J. Phys. Chem. B 109 (2005) 3402–3407. [21] B. Ohler, I. Revenko, C. Husted, J. Struct. Biol. 133 (2001) 1–9. [22] S. Cheifetz, M.A. Moscarello, C.M. Deber, Arch. Biochem. Biophys. 233 (1984) 151–160. [23] H.E. Warriner, J. Ding, A.J. Waring, J.A. Zasadzinski, Biophys. J. 82 (2002) 835–842. [24] Y.F. Hu, K.Y.C. Lee, J. Israelachvili, Langmuir 19 (2003) 100–104. [25] M. Seul, N.Y. Morgan, C. Sire, Phys. Rev. Lett. 73 (1994) 2284–2287. [26] D.J. Benvegnu, H.M. McConnell, J. Phys. Chem. 96 (1992) 6820–6824. [27] H.M. McConnell, V.T. Moy, J. Phys. Chem. 92 (1988) 4520–4525. [28] Y.F. Hu, K. Meleson, J. Israelachvili, Biophys. J. 91 (2006) 444–453.

30

Y. Hu, J. Israelachvili / Colloids and Surfaces B: Biointerfaces 62 (2008) 22–30

[29] G. Baneyx, V. Vogel, Proc. Natl. Acad. Sci. U.S.A. 96 (1999) 12518– 12523. [30] P.D. Morse 2nd, D.W. Deamer, Biochim. Biophys. Acta 298 (1973) 769–782. [31] H.M. McConnell, Annu. Rev. Phys. Chem. 42 (1991) 171–195. [32] D. Andelman, F. Brochard, C.M. Knobler, F. Rondelez, Structures and phase transitions in Langmuir monolayers, in: W. Gelbart, A. Ben-Shaul, D. Roux (Eds.), Micelles, Membranes, Microemulsions and Monolayers, Springer–Verlag, New York, 1994. [33] H.M. McConnell, R. Dekoker, J. Phys. Chem. 96 (1992) 7101–7103. [34] H.M. McConnell, R. DeKoker, Langmuir 12 (1996) 4897–4904. [35] H. Haas, H. Mohwald, Thin Solid Films 180 (1989) 101–110.

[36] P. Nassoy, W.R. Birch, D. Andelman, F. Rondelez, Phys. Rev. Lett. 76 (1996) 455–458. [37] V. Vogel, D. Mobius, J. Colloid Interface Sci. 126 (1988) 408–420. [38] A.V. Dobrynin, M. Rubinstein, Macromolecules 32 (1999) 915–922. [39] A.V. Dobrynin, M. Rubinstein, Macromolecules 34 (2001) 1964–1972. [40] D.J. Needleman, M.A. Ojeda-Lopez, U. Raviv, H.P. Miller, L. Wilson, C.R. Safinya, Proc. Natl. Acad. Sci. U.S.A. 101 (2004) 16099–16103. [41] J.N. Israelachvili, Intermolecular and surface forces, Academic Press, London, 1992. [42] H. Mueller, H.J. Butt, E. Bamberg, Biophys. J. 76 (1999) 1072– 1079. [43] E.J. Jo, J.M. Boggs, Biochemistry 34 (1995) 13705–13716.