10853
2007, 111, 10853-10857 Published on Web 08/25/2007
Dynamic Processes in Endocytic Transformation of a Raft-Exhibiting Giant Liposome Tsutomu Hamada,†,£ Yoko Miura,† Ken-ichi Ishii,† Sumiko Araki,‡ Kenichi Yoshikawa,‡,§ Mun’delanji Vestergaard,† and Masahiro Takagi*,† School of Material Science, Japan AdVanced Institute of Science and Technology, 1-1 Asahidai, Nomi, Ishikawa 923-1292, Japan, Department of Physics, Graduate School of Science, Kyoto UniVersity, Kyoto 606-8502, Japan, and Spatio-Temporal Order Project, ICORP, JST, Japan ReceiVed: July 11, 2007; In Final Form: August 7, 2007
The dynamic response of a raft-exhibiting giant liposome to external stimuli, such as the addition of Triton X-100 or osmotic stress, was studied. We observed that daughter vesicles are generated inside of the liposome through endocytic budding. It was found that the budding to generate daughter vesicles is classified into two different routes, simple budding through the invagination of a whole raft and budding from the boundary of a raft accompanied by waving motion. Smaller rafts show a preference for simple budding, whereas large rafts mainly adopt the other process. We discuss the mechanism of this difference in terms of the kinetic pathway of internalization by considering the line energy and bending energy of the membrane.
Introduction It is important to understand the physicochemical mechanisms that govern the morphological changes in cell membrane structure in response to various internal and external stimuli because of the possible implications in membrane trafficking and endosomal systems. Over the past decade, microdomains within a bilayer membrane, such as lipid rafts, have attracted considerable attention as one of the major mechanisms which mediate membrane internalization.1,2 Rafts are dynamic clusters composed largely of cholesterol and sphingolipids, which are rich in highly ordered saturated acyl chains.3 At heterogeneous biological membranes, microdomains are excepted to function as platforms which sort proteins and form endocytic carriers.4 The stability of a curved-out raft domain in a fluid membrane has been predicted theoretically; line tension at the phase boundary drives a spherical cap for each domain connected by a sharp cusp to reduce the interface length.5-7 This indicates that not only embedded and associated proteins but also the mechanical properties of the bilayer itself play an important role in controlling the shape of lipid-raft-derived vacuoles. Along these lines, giant multicomponent liposomes, raft-exhibiting model membranes, are efficient tools for studying the physicochemical properties of microdomains. Using this raft model system, a large number of studies have been carried out (e.g., thermodynamics,8,9 coupling of membrane curvature,10 domain patterns,11 and protein organization12). Recently, it has been suggested that some simple stimuli (osmosis, an enzyme, or detergents) may induce bud formation of raft domains in model membranes.13-15 However, there is still a large gap in our * To whom correspondence should be addressed. E-mail: takagi@ jaist.ac.jp. Phone: +81-761-51-1650. Fax: +81-761-51-1525. † Japan Advanced Institute of Science and Technology. ‡ Kyoto University. § ICORP, JST. £ E-mail:
[email protected].
10.1021/jp075412+ CCC: $37.00
understanding of the intermediate stages of domain internalization that needs to be bridged. In this Letter, we present real-time observations of endocytic transformations in raft-exhibiting model membranes. We studied biphase liposomes formed from a ternary mixture of saturated and unsaturated phospholipids and cholesterol. This ternary system is characterized by phase separation between liquidordered (Lo) and liquid-disordered (Ld) phases, where each phase corresponds to rafts and a surrounding fluid bilayer, respectively.3 As external stimuli, we used the detergent Triton X-100 (TX-100) and a highly osmotic environment (∼10 mM osmolarity glucose solution). To the best of our knowledge, this is the first report on the direct observation of the spatio-temporal evolution of endocytic vesicle formation using raft model membranes. Experimental Section Materials. An unsaturated phospholipid, dioleoyl L-R phosphatidylcholine (DOPC), a saturated phospholipid, dipalmitoyl L-R phosphatidylcholine (DPPC), and cholesterol were purchased from Avanti Polar Lipids. Bovine brain ganglioside GM1 ammonium salt was purchased from Calbiochem. Fluorescent dyes, N-(rhodamine red-X)-1,2-dihexadecanoyl-sn-glycero-3phosphoethanolamine triethylammonium salt (rhodamine red-X DHPE) (λex ) 560 nm, λem ) 580 nm) and Alexa Fluor 488 conjugate cholera toxin subunit B (CtxB-488) (λex ) 495 nm, λem ) 519 nm) were obtained from Invitrogen. All other reagents were purchased from Nacalai Tesque and were of analytical grade. Deionized water obtained from a Millipore Milli Q purification system was used to prepare buffers and reagents. Preparation of Raft-Exhibiting Giant Liposomes. Giant liposomes were prepared using the natural swelling method from a dry lipid film; the lipid mixture dissolved in 1:2 (v/v) chloroform/methanol along with rhodamine red-X DHPE and D(+)-glucose ([lipid]/[glucose] ) 1:3 molar ratio) in a glass © 2007 American Chemical Society
10854 J. Phys. Chem. B, Vol. 111, No. 37, 2007
Letters
Figure 1. Fluorescent microscopic images of the transformation of homogeneous DOPC liposomes after the addition of high osmolarity glucose (A) and TX-100 solutions (B). Time elapsed after treatment.
test tube were dried under vacuum for 3 h to form thin lipid films.16 Next, the films were hydrated with deionized water at 37 °C for an hour before being combined with a CtxB-488 solution. Then, the liposomes were incubated for more than several hours at room temperature (24 °C) before microscopic observations. The final concentrations were 0.2 mM lipids (DOPC/DPPC/cholesterol ) 2:2:1) with 1 mol % GM1 and 0.2 mol % rhodamine red-X DHPE and 25 µg/mL CtxB-488. Later, the dyes, rhodamine red-X DHPE and CtxB-488, were seen to be localized at the Ld and Lo phases, respectively.17 Additionally, homogeneous liposomes without a raft phase were also prepared using the same swelling method. The concentration was 0.5 mM DOPC and 1 mol % rhodamine red-X DHPE, where the entire membrane surface of the single-component liposome consisted of the Ld phase. Microscopic Observation. We constructed an observation chamber with two compartments separated by a porous membrane (0.2 µm diameter pores, Whatman) (see Supporting Information). The liposome solution (5µL) was placed in the lower compartment, and 40 µL of 1/300 (v/v) TX-100 or 10 mM glucose solution was added to the upper compartment. The liposomes were then allowed to interact with the added external stimuli in a gentle manner, where the TX-100 concentration was 3% v/v and the differences of the glucose molar concentrations across the bilayer membrane were 9.4 and 8.5 mM for the multi- and single-component liposomes, respectively. Timedependent changes in membrane morphology were observed using a fluorescent microscope (Axiovert 200, Carl Zeiss), and the images were recorded on HDD at 30 frames/s. Standard filter sets (No.15, Carl Zeiss; ex 540-552 nm, dichroic mirror 580 nm, em 590 nm and No. 17, Carl Zeiss; ex 475-495 nm, dichroic mirror 510 nm, em 515-565 nm) were used to monitor the fluorescence of rhodamine red-X DHPE and CtxB-488, respectively. Results Figure 1 shows examples of the effect of environmental stimuli, glucose osmosis and TX-100, on spherical single-phase liposomes of DOPC without raft domains. As shown in Figure 1A, the spherical liposome began to undulate and to assume an ellipsoid shape by osmosis. It is well-known that the decrease in aqueous volume due to osmotic pressure results in a transformation from spherical to various asymmetric shapes.18 Treatment with TX-100 also enhanced membrane fluctuation followed by a large deformation of its morphology (Figure 1B). Careful observations confirmed that the membrane area gradually increased during TX-100-induced transformations. Thus, in the case of homogeneous liposomes, the external stimuli led to large deformations of the entire membrane surface.
Figure 2. Budding through invagination. (A) Fluorescent images (upper) of raft domains in a biphasic liposome surface with schematic illustrations (lower). In the illustrations, gray regions indicate raft domains on the surface of a mother liposome, and dashed lines represent the surrounding fluid membranes. (B) In-section images (upper) and schematic representations (lower) of the course of the invagination of several raft domains. Time elapsed after the introduction of glucose osmosis to a liposome. The raft domains that are just beginning to bud are indicated by arrows. (C) Time series of the budding process of a raft domain. The elapsed time between snapshots is 66 ms. (D) An example of the development of membrane curvature of a raft domain during budding.
Figure 2 shows fluorescent images of the raft domains in twophase liposomes upon osmotic stress, using a fluorescent dye (CtxB-488) that preferentially partitions into the raft phase.17 The heterogeneous membrane surface was covered by several raft domains (Figure 2A).8 The floating rafts showed negative curvature, that is, the domains were curved toward the center of the liposome. Figure 2B shows the course of domain internalization. Each raft domain that exhibits the invagination is indicated by arrows. The lower illustrations show the surrounding fluid membrane with dashed lines, which was stained by another dye (rhodamine red-X DHPE). During the endocytic transformation of raft domains, the remaining part of the membrane maintained a spherical morphology. Interestingly, smaller domains budded first, followed by mid-sized domains and then the largest domain. In addition, larger buds appeared in the vesicular space during internalization, that is, the endocytic transformation of raft domains tended to proceed from the smaller ones. Figure 2C shows enlarged snapshots of the budding process of a raft domain (the raft on the left is budding). The negative curvature of a spherical cap gradually increased, and then, the edge closed to form an endocytic vesicle. The time development of membrane curvature during the budding process was measured (Figure 2D). The velocity of the change in curvature was maximum during the last step of internalization, indicating that closure of the domain edge is a very fast process. When TX-100 was applied, the vesicles underwent essentially the same budding phenomena. Next, we focused on much larger raft domains, which covered approximately a third of the vesicular surface. We know from our experiments that the domains become larger through collision and fusion during thermal motion to decrease phase
Letters
J. Phys. Chem. B, Vol. 111, No. 37, 2007 10855 after the addition of TX-100 (Figure 3C). These observations at different axes clarify that the emergence of satellite vesicles occurs at the wave-excited edge of the raft domains. Notably, the generated endocytic vesicles were typically within the same size range (Figure 3D; average surface area is 4.6 ( 1.8 µm2). In addition, we observed a wavy interface that produced endocytic vesicles under osmotic stimuli (Figure 3E). Discussion
Figure 3. Wavy budding: endocytic process of a larger raft domain in biphasic liposomes. (A-D) Addition of TX-100 and (E) osmotic stress. In the illustrations, solid lines indicate raft domains, and dashed lines represent surrounding fluid membranes. (A) Fluorescent image sequence (upper) of the production of satellite vesicles from a waveexcited domain interface, with schematic illustrations (middle). Red arrows and squares represent instantaneous emergence of the endocytic vesicle. Framed representations (lower) show the in-section membrane morphology to produce an endocytic vesicle. (B) Typical sectional image of a raft domain during internalization. (C) Top view of a raft domain with the resulting satellite vesicles, where the dashed line indicates the actual size of the liposome. Time lapsed after treatment with TX-100. (D) Area distribution of daughter vesicles generated through budding. (E) Typical fluorescent image (left) and schematic representation (right) of the wavy boundary of a raft domain and budded daughter vesicles upon osmotic shock (164 s after the addition of 10 mM glucose solution).
boundaries, and finally, the entire membrane surface is covered with two different phase regions.19,20 Thus, small and large sizes on the raft domain correspond to the early and late kinetic stages of domain growth. Figure 3A-C illustrates the process of endocytic vesicle formation in biphasic liposomes with one large raft domain under treatment with TX-100. Time-sequence images of the domain interface are presented in Figure 3A. The phase boundary was excited to form waves, and the top of the wave interface spontaneously progressed to a flask-shaped bud, which produced endocytic vesicles. The formation of endocytic vesicles proceeded until the raft phase region disappeared on the mother liposome. Figure 3B shows a typical in-section image of the raft domain with a wavy interface. The edge of the domain was negatively bent, which is different from the invaginated domain cap exhibited by smaller domains, as shown in Figure 2. The time development of cross-sectional schematics corresponding to the process of endocytic transformation is illustrated in Figure 3A (framed pictures).17 Moreover, the top view of a raft indicates that many small vesicles appeared near the edge of the raft domain, where the domain size decreased with time
The present results show that the application of TX-100 or osmotic stress caused phase-separated liposomes to exhibit raft domain internalization, and the entire membrane of the homogeneous liposomes deformed with large fluctuations. The transformation in response to stimuli can be attributed to an increase in excess surface area. If the membrane area just fits the spherical surface of the inner aqueous phase (no excess area), the vesicle has no deformed morphology, such as buds, because of high incompressibility. This situation corresponds to the initial liposomes before stimulation in our experiments. When the liposomes are subjected to high osmolarity, the water efflux across the membranes reduces the inner aqueous volume. Meanwhile, TX-100 leads to an increase in surface area by becoming incorporated into the bilayer membrane.21 The resulting excess membrane area allows transformation into a variety of vesicular morphologies, as we have reported. Additionally, the difference in the response to stimulation between two-phase (Figures 2 and 3) and single-phase liposomes (Figure 1) is derived from the line energy of phase boundaries. In heterogeneous membranes, raft domains show greater effects of line energy and tend to undergo budding deformation to reduce the interface length.5 On the other hand, uniform membranes without raft domains transform their shapes by minimizing the total bending energy over the closed vesicle surface.22 We found that multicomponent liposomes show two different types of internalization process depending on the size of the raft domains. If several small rafts exist in a membrane, simple budding occurs due to the invagination of a whole raft domain area (Figure 2). In contrast, large raft domains tend to produce daughter vesicles from a wavy boundary (Figure 3). We now discuss the mechanism of this difference in the raft domain internalization processes by considering the effects of the line energy of domain boundaries and the bending energy of bilayer membranes. For simplicity, we assume that (i) raft domains are shaped like a spherical cap by defining the variables R (curvature radius) and θ, and (ii) endocytic daughter vesicles at wavy budding are spheres with radius r (Figure 4A). Under the conservation of domain area, 2πR20(1 - cos θi) ) 2πR2(1 cos θ) (for simple budding) and 2πR20(1 - cos θi) ) 2πR20(1 cos θ) + 4πr2n (for wavy budding) are given, where R0, θi, and n are the initial curvature radius, initial angle, and the number of budded daughter vesicles.23 Initially, the domains have a positive R that corresponds to the radius of a spherical liposome. During simple budding, θ changes from θi (positive) to -π, and R becomes negative when θ varies over zero. Meanwhile, the change in θ at wavy budding is from θ0 to 0 with constant R. The free energy for domains undergoing simple budding can be expressed as
Fsimple ) 4πκ(1 - cos θ) + 2πσR sin θ
(1)
where κ and σ denote bending elasticity and line tension, respectively (we assume that the spontaneous curvature is zero).5 The free energy for wavy budding is written as the sum of the energies of the domain within a mother liposome given by eq
10856 J. Phys. Chem. B, Vol. 111, No. 37, 2007
Letters
Figure 4. (A) Schematic representation of raft domains during simple and wavy budding and variables for describing their geometry. (B) Free energies of simple and wavy budding in a small raft domain with an initial θ of π/20. (C) Free energies of simple and wavy budding in a large raft domain with an initial θ of π/4. The free energy profiles are deduced from the model eqs 1 and 2.
a bilayer membrane. If we apply eq 1 to a situation in which two raft domains with a large or small radius show simple budding under a given constant excess area, we should expect smaller domains to undergo budding as opposed to larger ones (see Supporting Information). In vivo, endosomal vesicles have a narrow size distribution, which is typically controlled by coat proteins.26 Intriguingly, our results imply that multicomponent bilayer membranes have the mechanical capability to internalize monodisperse vesicles with no proteins (Figure 3D). Although our model explains that a large raft domain tends to show wavy budding that produces several daughter vesicles, the value of r calculated from our model is larger than that obtained from the experimental results; in our model, for a large raft domain (θi ) π/4) to overcome the energy barrier to undergo wavy budding by thermal fluctuation, r is estimated to be above around 1.3 µm, which is approximately twice that (r ) 0.6 ( 0.1 µm in Figure 3D) in our observation. Several additional factors for decreasing the energy barrier can be considered. The change in the morphology of the raft boundaries was not accurately described by our model. The development of a typical wavelength of the fluctuating interface could make it easier to produce monodisperse daughter buds. Additionally, we should consider the effect of the bending energy of the surrounding fluid membranes. Further theoretical and experimental studies are underway to clarify the mechanism involved in endocytic transformations of heterogeneous membranes with different mechanical properties.
1 and the internalized satellite vesicles
Conclusions
Fwavy ) 4πκ(1 - cos θ) + 2πσR sin θ + 8πκn
(2)
To compare the two budding processes, we introduce a normalized angle that describes the degree of transformation, θ ) (θi - θ)/(θi - θf), where θf is the final angle for the phase boundaries to disappear, that is, θf ) -π for simple budding and θf ) 0 for wavy budding. During internalization, θ changes from 0 to 1. Changes in the free energy ∆F(θ) ) F(θ) - F(0), with θi ) π/20 (small domain) and θi ) π/4 (large domain), are shown in Figure 4B and C, respectively, where ∆F is conveniently converted to a dimensionless energy ∆F/ 2πκ. Each parameter is taken to be as follows: R0 ) 10 µm, r ) 0.5 µm (when θi ) π/20) or 1.3 µm (when θi ) π/4), and σ/κ ) 10 µm-1 (we use typical values of κ ∼10-19 J and σ ∼10-12 N, as reported elsewhere10,24). A comparison of Figure 4B and C shows how the free energies depend on the size of raft domains. While one daughter vesicle generated by simple budding (Fsimple ) 8πκ) is always more stable than several vesicles generated by wavy budding (Fwavy ) 8πκn), the kinetic pathway largely depends on the size of the raft domains. A large domain has a higher energy barrier to undergo simple budding because of a greater effect of line energy for inversion of the membrane curvature from curved-out to curved-in and tends to undergo wavy budding, while a small domain shows simple budding. Thus, our model explains the observed budding pathway, as shown in Figures 2 and 3. In the case of simple budding, we observed that invagination proceeded from smaller domains (Figure 2B). This tendency is in agreement with previous research which reported that outside budding of raft domains is induced by detergents in two-phase liposomes attached to a glass pole.15,25 If a membrane surface is covered by several raft domains, the excess area obtained by external stimuli is distributed among these raft domains within
We have conducted real-time observations of endocytic transformation in raft model membranes after the application of external stimuli. The results clearly demonstrate that there are two distinct pathways for the internalization process, simple budding and wavy budding, which depend on the raft domain size. In the case of simple budding, a raft domain is curved inward with a spherical cap to eventually become an endocytic vesicle. On the other hand, larger rafts produce monodisperse endocytic vesicles from the wavy boundary. The present results could help us to better understand the essential processes in domain-mediated internalization, including cellular endocytosis and membrane trafficking. Acknowledgment. We thank Professor M. Imai and Ms. M. Yanagisawa (Ochanomizu University) for their valuable discussions. This work was supported by KAKENHI (Grantin-Aid Scientific Research) on Priority Area “Soft Matter Physics” and “Life Surveyor” from the Ministry of Education, Culture, Sports, Science and Technology of Japan and by Japan Society for the Promotion of Science (JSPS) under a Grant-inAid for Creative Scientific Research (No. 18GS0421) and for Young Scientists (Start-up) (No. 18840019). Supporting Information Available: Illustration of the observation chamber and a model of the invagination process in simple budding. This material is available free of charge via the Internet at http://pubs.acs.org. References and Notes (1) Hanzal-Bayer, M. F.; Hancock, J. F. FEBS Lett. 2007, 581, 20982104. (2) Kirkham, M.; Parton, R. G. Biochim. Biophys. Acta 2005, 1746, 350-363. (3) Brown, D. A.; London, E. J. Biol. Chem. 2000, 275, 17221-17224. (4) Simons, K.; Ikonen, E. Nature 1997, 387, 569-572.
Letters (5) Lipowsky, R. J. Phys. II 1992, 2, 1825-1840. (6) Ju¨licher, F.; Lipowsky, R. Phys. ReV. E 1996, 53, 2670-2683. (7) Taniguchi, T. Phys. ReV. Lett. 1996, 76, 4444-4447. (8) Veatch, S. L.; Keller, S. L. Biophys. J. 2003, 85, 3074-3083. (9) Cicuta, P.; Keller, S. L.; Veatch, S. L. J. Phys. Chem. B 2007, 111, 3328-3331. (10) Baumgart, T.; Hess, S. T.; Webb, W. W. Nature 2003, 425, 821824. (11) Rozovsky, S.; Kaizuka, Y.; Groves, J. T. J. Am. Chem. Soc. 2005, 127, 36-37. (12) Kahya, N.; Brown, D. A.; Schwille, P. Biochemistry 2005, 44, 7479-7489. (13) Do¨bereiner, H.-G.; Ka¨s, J.; Noppl, D.; Sprenger, I.; Sackmann, E. Biophys. J. 1993, 65, 1396-1403. (14) Staneva, G.; Angelova, M. I.; Koumanov, K. Chem. Phys. Lipids 2004, 129, 53-62. (15) Staneva, G.; Seigneuret, M.; Koumanov, K.; Trugnan, G.; Angelova, M. I. Chem. Phys. Lipids 2005, 136, 55-66. (16) Yamada, N. L.; Hishida, M.; Seto, H.; Tsumoto, K.; Yoshikawa, T. cond-mat/0607672. (17) Bacia, K.; Schwille, P.; Kurzchalia, T. Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 3272-3277. (18) Hotani, H. J. Mol. Biol. 1984, 178, 113-120. (19) Saeki, T.; Hamada, T.; Yoshikawa, K. J. Phys. Soc. Jpn. 2006, 75, 013602/1-013602/3. (20) Yanagisawa, M.; Imai, M.; Masui, T.; Komura, S.; Ohta, T. Biophys. J. 2007, 92, 115-125. (21) With reference to a report on the interaction between detergents and lipid membranes, a two-stage model is proposed; first, the incorporation
J. Phys. Chem. B, Vol. 111, No. 37, 2007 10857 of detergents into membranes leads to an increase in vesicular size, and the saturated membranes then disintegrate because of the formation of lipiddetergent mixed micelles. See Paternostre, M.-T.; Roux, M.; Rigaud, J.-L. Biochemistry 1988, 27, 2668-2677. The transformations in the membrane surface reported here essentially took place in the above first stage since the surrounding fluid membrane started to dissolve within several minutes after such deformations by TX-100. (22) Seifert, U.; Berndl, K.; Lipowsky, R. Phys. ReV. A 1991, 44, 11821202. (23) Rafts are defined as detergent-resistant membranes that are divided from cell membranes by the use of surfactants such as TX-100, which implies that TX-100 tends to largely be incorporated into the surrounding fluid membrane with the Ld phase. We confirmed that the area of raft domains remained nearly constant at TX-100-induced internalization. (24) No specific difference in the mechanical properties (κ and σ) was found even in the presence of CtxB-488. See Semrau, S.; Idema, T.; Holtzer, L.; Schmidt, T.; Storm, C. cond-mat/0612554. (25) In our experiments, budding toward the outside, as reported in ref 15, was rarely observed. The direction of budding seemed to depend on the initial membrane condition (e.g., a membrane attached to glass would be in a more stressed condition than a membrane floating in aqueous solutions; see Sandre, O.; Moreaux, L.; Brochard-Wyart, F. Proc. Natl. Acad. Sci. U.S.A. 1999, 96, 10591-10596.), though at present, we cannot provide any definite conclusions on this matter. We are currently examining this point and will present the findings in a separate report. (26) Rothberg, K. G.; Heuser, J. E.; Donzell, W. C.; Ying, Y.-S.; Glenney, J. R.; Anderson, R. G. W. Cell 1992, 68, 673-682.
