Letter pubs.acs.org/Langmuir
Flow-Driven Rapid Vesicle Fusion via Vortex Trapping Sangwoo Shin, Jesse T. Ault, and Howard A. Stone* Department of Mechanical and Aerospace Engineering, Princeton University, Princeton, New Jersey 08544, United States S Supporting Information *
ABSTRACT: Fusion between suspended lipid vesicles is difficult to achieve without membrane proteins or ions because the vesicles have extremely low equilibrium membrane tension and high poration energy. Nonetheless, vesicle fusion in the absence of mediators can also be achieved by mechanical forcing that is strong enough to induce membrane poration. Here, we employ a strong fluid shear stress to achieve vesicle fusion. By utilizing a unique vortex formation phenomenon in branched channels as a platform for capturing, stressing, and fusing the lipid vesicles, we directly visualize using high-speed imaging the vesicle fusion events, induced solely by shear, on the time scale of submilliseconds. We show that a large vesicle with a size of up to ∼10 μm can be achieved by the fusion of nanoscale vesicles. This technique has the potential to be utilized as a fast and simple way to produce giant unilamellar vesicles and to serve as a platform for visualizing vesicle interactions and fusions in the presence of shear.
■
■
INTRODUCTION
Vesicle fusion plays a key role in many biological processes such as exocytosis, protein translocation, neurotransmission, and fertilization.1 Fusion is initiated by introducing membrane tension that is strong enough to form fusion pores within the membranes.2 These fusion pores serve as nucleation sites for vesicle fusion. In general, vesicle fusion is mediated by proteins,3 but it can also be achieved through the use of multivalent ions,4 polymers,5,6 or surfactants.7 Without the help of proteins or ions, inducing vesicle fusion is difficult because lipid vesicles have extremely low membrane tension at equilibrium8 and large bilayer poration energy, which is on the order of 13kBT.9 However, in the presence of a significant strain or shear, vesicle fusion can also be accomplished without the help of additional mediators.10,11 Vesicle fusion typically occurs on a submillisecond time scale12,13 or shorter.9 The time and length scales make observing this process difficult, thus direct observations of shear-induced vesicle fusion have not been reported to date. Here, we present real-time visualization of vesicle fusion induced by strong shear stress. Our group reported recently that a type of vortex formation can occur in the flow in bifurcating T-junction channels, which can allow stable capturing of particles.14 This effect provides a unique technique for capturing lipid vesicles, subjecting them to a shear stress, and directly visualizing their fusion with a conventional microscope. Below we show that giant unilamellar vesicles (GUVs) having sizes of up to ∼10 μm can be produced from ∼200 nm small unilamellar vesicles (SUVs) by shear stress. We also demonstrate the use of this technique as a platform for visualizing the dynamics of lipid vesicle interactions. © 2015 American Chemical Society
EXPERIMENTAL SECTION
Materials. 1,2-Dioleoyl-sn-glycero-3-phosphocholine (DOPC) and 1,2-dipalmitoyl-sn-glycero-3-phosphoethanolamine-N-(lissamine rhodamine B sulfonyl) (ammonium salt) (Rh-DPPE) were purchased from Avanti Polar Lipids. Sodium chloride and chloroform were purchased from Sigma-Aldrich. Sodium metatungstate solution was purchased from Acros Organics. Lipid Vesicle Solutions. SUVs were prepared via the sonication method.15 One milligram of lipid mixture solution dissolved in chloroform (DOPC and Rh-DPPE mixed in a ratio of 99.5:0.5 mol %) was dried in a glass vial overnight under vacuum. Then, the dried lipids were rehydrated with 0.6 M NaCl solution, followed by sonication using a tip sonicator for 10 min. Platinum pellets that came off of the sonicator tip during sonication were separated from the lipid solution via centrifugation (MiniSpin, Eppendorf). The final lipid solution was mixed with 0.4 M sodium metatungstate solution in a volume ratio of 1:20. Two solutions of 0.4 M sodium metatungstate were prepared, one with lipid vesicles prepared in 0.6 M NaCl and one with lipid vesicles prepared in 0.8 M NaCl. The 0.6 M NaCl concentration was chosen to yield zero net osmotic pressure across the membrane. The 0.8 M NaCl concentration was chosen to result in a small net osmotic pressure, leading to an uptake of water in the lipid vesicles and thus an increase in the tension of their membranes. The osmolarities of the solutions were measured using an osmometer (μOsmette, Precision Systems). The final diameter of the prepared SUVs was measured to be ∼220 nm using dynamic light scattering (Zetasizer Nano-ZS, Malvern Instruments). Flow Experiment. Flow channels were made from poly(dimethylsiloxane) (PDMS, Sylgard 184 elastomer kit, Dow Corning) via soft lithography. The cross-sectional geometry of the channel was 80 × 80 μm2 (Figure 1a). The channel size was chosen by considering the particle size limit for capture to occur14 and the angular velocity of the vortices, which is limited by the performance of our high-speed Received: May 12, 2015 Revised: June 19, 2015 Published: June 22, 2015 7178
DOI: 10.1021/acs.langmuir.5b01752 Langmuir 2015, 31, 7178−7182
Letter
Langmuir
simulations in the inset of Figure 1b. These flow features lead to the capture of low-density particles, such as hollow glass beads or air bubbles.14,16 A numerical simulation of the flow structures in an angled arrow-shaped channel (Figure 1) shows this pair of vortices spanning along the streamwise direction near the stagnation point and forming steady structures with internal recirculation regions. Within the vortex structures, closed-loop streamlines travel upstream along the vortex cores. To visualize these vortices, we plot an isosurface of the Qcriterion (white contours in Figure 1b). The Q-criterion is a numerical technique for visualizing vortices18 and is described in the Supporting Information. As expected, we find that the recirculation regions in our simulations correspond to regions with large values of the Q-criterion. Furthermore, these regions also correspond to the regions where we observe the particle capture experimentally, as described in the next section. We use these vortices as a tool to capture vesicles and induce a strong shear stress on them in order to study vesicle fusion. Capture and Fusion of Vesicles in an Arrow-Shaped Junction. In our initial study of a flow in a T junction, particle capture was found to occur only when the density of the particles is at least ∼30% lower than the density of the bulk fluid.14 To this end, we have used sodium metatungstate solution (0.4 M) to achieve a more dense outer solution (ρs = 1945 kg/m3). The densities of the lipid vesicles containing the 0.6 and 0.8 M NaCl solutions were ρv = 1022 and 1030 kg/m3, respectively, yielding density ratios of ρv/ρs ≈ 0.53, which is below the critical density ratio for capture of ∼0.7. For the vesicles containing 0.6 M NaCl, the concentration of the lipid solution was matched to that of the bulk solution to yield a zero osmotic pressure across the membrane in order to rule out any intrinsic membrane tension that could induce spontaneous vesicle fusion. The trapping of vesicles in the vortices at Re = 230 is presented in Figure 2. The images were taken with a confocal microscope using resonance scanning mode to achieve a high scanning rate (14 Hz). A GUV having a diameter of ∼8 μm is
Figure 1. Vortex trapping in branching channels. (a) Schematic of the channel used in this study. (b) Numerical simulation of the flow streamlines within the channel at Re = 230. Colors represent the velocity magnitude, and the pair of white contours represents an isosurface of the Q-criterion. The inset represents a pocket of fluid that recirculates in the flow, which is responsible for the particle capture. camera (Phantom V9.1, Vision Research). A syringe pump (PHD Ultra, Harvard Apparatus) was used to drive the lipid solutions through the channel at a flow rate of 1.42 mL/min, which corresponds to an average flow speed of U ≈ 3.7 m/s. The solutions have a shear viscosity of μ = 2.5 mPa·s (measured by a rheometer, Physica MCR 301, Anton Paar) and a density of ρ = 1945 kg/m3, which yields a Reynolds number of Re = ρUw/μ = 230, where w is the channel width. Preliminary experiments have demonstrated that this Reynolds number in a 70° angle junction corresponds to a robust and stable capture of particles, so we use this branch angle for the experiments here. A confocal microscope (TCS SP5, Leica) operating in resonance mode was used to image the captured fluorescent large vesicles. A high-speed camera attached to a microscope (DMI4000B, Leica) was used to record at 6000 fps the fusion dynamics of the lipid vesicles near the junction.
■
RESULTS AND DISCUSSION Fluid Mechanics of an Arrow Junction. We have recently identified fluid mechanical features of T junctions and similar bifurcations where three-dimensional effects produce localized circulating fluid regions.14,16 We recognized that such features might allow the assembly and/or coalescence of bubbles and droplets. Here, we apply this principle to vesicle fusion. Steady flows that pass through curved or bent channels are known to develop a secondary flow structure of counterrotating vortices, which are sometimes referred to as Dean vortices.17 Recently, the flow in T junctions has been shown to exhibit these Dean-like vortices. Furthermore, for a certain range of Reynolds numbers, these vortices can form closed recirculation regions in the flow, as shown in the numerical
Figure 2. Confocal imaging of large vesicle trapped inside the vortex. Fused GUVs with a cloud of SUVs are clearly seen from either (a) bright field or (b) fluorescent images. (c, d) Close-up images for (a) and (b), respectively. 7179
DOI: 10.1021/acs.langmuir.5b01752 Langmuir 2015, 31, 7178−7182
Letter
Langmuir
Figure 3. Formation of a GUV by the fusion of SUVs trapped in a vortex. The GUV is stationary whereas the SUVs are constantly orbiting around the vortex core.
the radius of intermediate vesicles (1.5 μm), the fusion time scale is estimated to be ∼1 μs, which is in agreement with predicted values from dissipative particle dynamics simulations.9 This time scale is too fast to be resolved with our highspeed camera. The fast angular speed of the vortex results in a strong shear stress acting on the vesicles, which may lead to the formation of microscopic fusion pores in the membranes that can initiate fusion between the vesicles. Assuming that the vortex has the same angular velocity along the radial direction at a given axial position, the shear rate is estimated to be ∼2.2 × 105 s−1 at Re = 230, which gives an estimated shear stress of ∼550 Pa. A recent study by Kogan et al. reports that the threshold shear rate to induce the fusion of DOPC vesicles is 3100 s−1 with a 40 wt % sucrose solution.11 Assuming their viscosity of the solution is ∼5.5 mPa·s,22 this gives a threshold shear stress of ∼17 Pa, which is more than an order of magnitude smaller than that of our system. Therefore, the strong shear provided by our system is believed to induce sufficient membrane tension that leads to fusion pore formation.2 The morphology of the vesicles in the vortex also implies that the vesicles are experiencing substantial membrane tension due to the strong shear. Viscous stresses from the vortex and the main flow deform the vesicles such that the vesicles are oriented downstream (Figure 4a). Because DOPC has basically zero membrane curvature,23 the deformation is solely from the shear. The total surface area of the deformed vesicle can be obtained from the shape of the vesicle, which is estimated to be ∼212 μm2 (Figure 4c). Because the surface area of a sphere under an equivalent volume condition is ∼207 μm2, this gives an effective area change of ∼2.4%, which is close to the yield strain of DOPC bilayer membranes.10 Such a large vesicle deformation should induce a membrane tension that is strong enough to cause poration and initiate the vesicle fusion process.7,11 Moreover, because of the fact that the vortex captures many SUVs, the effective viscosity of the medium might be locally increased in the capture region, leading to an even greater shear stress acting upon the vesicles.24 The effective viscosity of a dilute suspension of rigid spheres is given by μe = μ(1 + (5/ 2)ϕ), where ϕ is the volume fraction of the suspended particles (here, vesicles). The volume fraction of the captured SUVs cannot be simply estimated from the observed images, even with a high-speed camera, because of the extremely high angular speed of the vortex (ω = 2175 Hz) and the transparency of the vesicles. However, a previous study using a larger system having a lower angular velocity (∼170 Hz) suggests that the volume fraction of the captured particles can be large within the vortex,14 implying that the shear stress
clearly shown to be captured upstream from the vortex, followed by a cloud of SUVs captured downstream. Also, the SUVs are seen to orbit at a certain distance away from the vortex core, whereas the GUV rotates coaxially at the vortex core without any translational movement. The size-dependent positions of the captured vesicles are set by the force balance among the pressure gradient and centrifugal and viscous drag forces,14 which drive larger vesicles closer to the vortex core and further upstream in the main flow. Similar behavior has been observed in water−air bubble systems in T junctions, where larger bubbles were more likely to be captured at the upstream edge of the vortex and close to the core, whereas smaller bubbles were found orbiting around the downstream edge of the vortex.14 To determine whether a captured GUV is a consequence of fusion among the captured SUVs, we observe the capturing events using a high-speed camera. The results of the high-speed imaging shown in Figure 3 reveal that the small vesicles are consecutively merged into one final GUV. (See the Supporting Information for the full movie.) Initially, the SUVs are constantly being captured in the vortices and accumulate over time until the concentration of the captured vesicles becomes high enough to allow direct contact between the nearby vesicles. The captured SUVs are distributed evenly throughout the vortex along the axial direction and orbit a certain distance away from the vortex core. Subsequently, they merge into larger intermediate vesicles having sizes of a few micrometers in diameter. Within a single frame (≤0.17 ms), the singular nature of coalescence leads to rapid fusion of the intermediate vesicles into a single GUV, which is located in the vortex core. The pressure gradient in the radial direction with respect to the vortex core is always positive, indicating that the local pressure inside the vortex is low.14 This pressure gradient along with the centrifugal force drives the particles toward the vortex core and is responsible for the capture and the continuous fusion process. As the vesicles grow, the rate of SUV accumulation will slow because the larger vesicle(s) will occupy more space in the vortex. If several large vesicles (intermediate vesicles) are simultaneously formed or captured within the vortex, then the pressure gradient will drive them into contact with each other, at which time the shear stress will begin to induce fusion between the large vesicles. Finally, once the vesicle reaches a critical size, it will be ejected from the vortex and carried downstream. Because the fusion process is driven by membrane tension, the fusion between the vesicles can be thought of as analogous to droplet coalescence. The time scale for rapid vesicle fusion can then be estimated as t ≈ (ρR3/γ)1/2, where γ is the surface tension and R is the characteristic length scale.19,20 By taking γ as the lysis tension of DOPC vesicles (4.08 mN/m21) and R as 7180
DOI: 10.1021/acs.langmuir.5b01752 Langmuir 2015, 31, 7178−7182
Letter
Langmuir
diluted by 98 vol % by the surrounding bulk fluid. Obviously, such a dilution will not allow the vesicles to be captured because a particle must at least have a density that is 30% less than that of the bulk fluid in order to be captured. Thus, we come to a conclusion that there must be a mechanism for regulating the surface area of the vesicles. One possible mechanism is shear-induced tubulation followed by breakup.26 In this mechanism, the lipid tubulation occurs only in a local area, without much influence on the global morphology of the vesicle.26−28 Unfortunately, as pointed out earlier in the paper, the spatial and temporal resolution of our current high-speed camera limits the observation of the dynamics during the fusion process. Thus, we are unable to confirm this tubulation mechanism as a surface area regulating effect. We are currently developing a way to overcome this problem by introducing multiple stages in the device where the hydraulic diameter of the channel gradually increases so that the angular speed of the vortex can be effectively reduced, which may allow for better visualization. This design is left for future investigation.
■
OUTLOOK By utilizing a vortex structure that occurs during flow in branched channels as a means to capture and fuse lipid vesicles, we have shown that nanoscale SUVs can be fused to generate microscale GUVs. Our study also suggests a fast and simple technique for producing well-defined GUVs that are normally made by the electroformation method, which is timeconsuming and requires additional devices such as an external power supply and electrodes.29 By massively parallelizing the bifurcations, a high throughput can be expected, which would make an attractive method for producing high-quality GUVs with a uniform size distribution.
Figure 4. Shape of GUVs for different osmotic pressures. (a) GUV with an inner solution of 0.6 M NaCl. (b) GUV with an inner solution of 0.8 M NaCl. (c) Comparison of the shapes of the GUVs in (a) (blue) and (b) (red).
acting upon the vesicles might be higher than estimated above, which would lead to increased vesicle fusion. We also investigate the effect of osmotic pressure on the fused GUVs to address the quality of the membrane structure. By changing the concentration of the solution inside the SUVs, we can manipulate the osmotic pressure built up across the membrane, which will contribute to vesicle deformation. Figure 4 presents the effect of osmotic pressure on the morphologies of the captured GUVs with different inner solutions. The vesicle with zero osmotic pressure readily conforms to the local shear stress and thus deforms more (Figure 4a) whereas the vesicle with higher osmolarity retains its shape due to the membrane tension induced by the osmotic pressure, leading to a more spherical shape with only 0.7% area change (Figure 4b). Such an ability to readily adapt the osmotic pressure implies that there is negligible leakage of the inner solution during the fusion process and that the membrane of the fused GUVs are unilamellar.25 We note that there is a further process that remains to be elucidated regarding the surface-to-volume scaling during the fusion process. As the vesicles fuse, we would expect the fused GUVs to be progressively more flaccid because the surface area to volume ratio for a roughly spherical vesicle scales as ∼1/R, where R is the approximate radius of the vesicle. Therefore, when two vesicles fuse, the new, larger vesicle must have more surface area than a sphere containing the same volume as the two initial vesicles. For instance, assuming the initial radius of the small vesicle is Rs = 0.1 μm, the number of SUVs required to produce a single spherical GUV with radius Rg = 5 μm under an equal surface area condition can be estimated to be n = Rg2/ Rs2 = 2500. The volume ratio between 2500 vesicles and the giant vesicle is then given by nRs3/Rg3 = 0.02. This implies that for the surface area to be conserved and for the large vesicle to maintain a spherical shape, the fluid inside the vesicles must be
■
ASSOCIATED CONTENT
S Supporting Information *
Description of the Q-criterion and a high-speed movie of flowinduced vesicle fusion (Figure 3). The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.langmuir.5b01752.
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. Author Contributions
S.S. and J.T.A. contributed equally to this work. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS We acknowledge Jie Feng for help with dynamic light scattering measurements and Daniele Vigolo for valuable comments. We thank the NSF for supporting H.A.S. under grant no. DMS1219366.
■
REFERENCES
(1) Alberts, B.; Johnson, A.; Lewis, J.; Morgan, D.; Raff, M.; Roberts, K.; Walter, P. Molecular Biology of the Cell, 6th ed.; Garland Science: New York, 2015. (2) Shillcock, J. C.; Lipowsky, R. Tension-induced fusion of bilayer membranes and vesicles. Nat. Mater. 2005, 4, 225−228. 7181
DOI: 10.1021/acs.langmuir.5b01752 Langmuir 2015, 31, 7178−7182
Letter
Langmuir (3) Jahn, R.; Lang, T.; Südhof, T. C. Membrane fusion. Cell 2003, 112, 519−533. (4) Chanturiya, A.; Scaria, P.; Woodle, M. C. The role of membrane lateral tension in calcium-induced membrane fusion. J. Membr. Biol. 2000, 176, 67−75. (5) Lee, J.; Lentz, B. R. Secretory and viral fusion may share mechanistic events with fusion between curved lipid bilayers. Proc. Natl. Acad. Sci. U.S.A. 1998, 95, 9274−9279. (6) Safran, S. A.; Kuhl, T. L.; Israelachvili, J. N. Polymer-induced membrane contraction, phase separation, and fusion via Marangoni flow. Biophys. J. 2001, 81, 659−666. (7) Bernard, A.-L.; Guedeau-Boudeville, M. A.; Marchi-Artzner, V.; Gulik-Krzywicki, T.; di Meglio, J.-M.; Jullien, L. Shear-induced permeation and fusion of lipid vesicles. J. Colloid Interface Sci. 2005, 287, 298−306. (8) Jähnig, F. What is the surface tension of a lipid bilayer membrane? Biophys. J. 1996, 71, 1348−1349. (9) Gao, L.; Lipowsky, R.; Shillcock, J. Tension-induced vesicle fusion: pathways and pore dynamics. Soft Matter 2008, 4, 1208−1214. (10) Staykova, M.; Holmes, D. P.; Read, C.; Stone, H. A. Mechanics of surface area regulation in cells examined with confined lipid membranes. Proc. Natl. Acad. Sci. U.S.A. 2011, 108, 9084−9088. (11) Kogan, M.; Feng, B.; Nordén, B.; Rocha, S.; Beke-Somfai, T. Shear-induced membrane fusion in viscous solutions. Langmuir 2014, 30, 4875−4878. (12) Lindau, M.; Alvarez de Toledo, G. The fusion pore. Biochim. Biophys. Acta 2003, 1641, 167−173. (13) Haluska, C. K.; Riske, K. A.; Marchi-Artzner, V.; Lehn, J.-M.; Lipowsky, R.; Dimova, R. Time scales of membrane fusion revealed by direct imaging of vesicle fusion with high temporal resolution. Proc. Natl. Acad. Sci. U.S.A. 2006, 103, 15841−15846. (14) Vigolo, D.; Radl, S.; Stone, H. A. Unexpected trapping of particles at a T junction. Proc. Natl. Acad. Sci. U.S.A. 2014, 111, 4770− 4775. (15) Barenholz, Y.; Gibbes, D.; Litman, B. J.; Goll, J.; Thompson, T. E.; Carlson, F. D. A simple method for the preparation of homogeneous phospholipid vesicles. Biochemistry 1977, 16, 2806− 2810. (16) Chen, K. K.; Rowley, C. W.; Stone, H. A. Vortex dynamics in a pipe T-junction: Recirculation and sensitivity. Phys. Fluid 2015, 27, 034107. (17) Dean, W. R. Note on the motion of fluid in a curved pipe. Philos. Mag. Ser. 1927, 7, 208−223. (18) Hunt, J. C. R.; Wray, A.; Moin, P. Eddies, stream, and convergence zones in turbulent flows. Center for Turbulence Research Report 1988, CTR-S88, pp 193−208. (19) Eggers, J.; Lister, J. R.; Stone, H. A. Coalescence of liquid drops. J. Fluid Mech. 1999, 401, 290−310. (20) Wu, M.; Cubaud, T.; Ho, C.-M. Scaling law in liquid drop coalescence driven by surface tension. Phys. Fluid. 2004, 16, L51−L54. (21) Portet, T.; Dimova, R. A new method for measuring edge tensions and stability of lipid bilayers: Effect of membrane composition. Biophys. J. 2010, 99, 3264−3273. (22) Swindells, J. F.; Snyder, C. F.; Hardy, R. C.; Golden, P. E. Viscosities of Sucrose Solutions at Various Temperatures: Tables of Recalculated Values; National Bureau of Standards Circular, no. 440 supplement; Supt. of Docs., U.S. G.P.O.: Washington, DC, 1958. (23) Szule, J. A.; Fuller, N. L.; Rand, R. P. The effects of acyl chain length and saturation of diacylglycerols and phosphatidylcholines on membrane monolayer curvature. Biophys. J. 2002, 83, 977−984. (24) Frankel, N. A.; Acrivos, A. On the viscosity of a concentrated suspension of solid spheres. Chem. Eng. Sci. 1967, 22, 847−853. (25) Tayeb, L.; Vashaee, D.; Parikh, A. N. Stability of uni- and multillamellar spherical vesicles. ChemPhysChem 2012, 13, 314−322. (26) Sekine, Y.; Abe, K.; Shimizu, A.; Sasaki, Y.; Sawada, S.-i.; Akiyoshi, K. Shear flow-induced nanotubulation of surface-immobilized liposomes. RSC Adv. 2012, 2, 2682−2684.
(27) Hochmuth, R. M.; Mohandas, N.; Blackshear, P. L. Measurement of the elastic modulus for red cell membrane using a fluid mechanical technique. Biophys. J. 1973, 13, 747−762. (28) Galie, P. A.; Nguyen, D.-H. T.; Choi, C. K.; Cohen, D. M.; Janmey, P. A.; Chen, C. S. Fluid shear stress threshold regulates angiogenic sprouting. Proc. Natl. Acad. Sci. U.S.A. 2014, 111, 7968− 7973. (29) Angelova, M.; Dimitrov, D. S. Liposome electroformation. Faraday Discuss., Chem. Soc. 1986, 81, 303−311.
7182
DOI: 10.1021/acs.langmuir.5b01752 Langmuir 2015, 31, 7178−7182
