Artificial cells: Unique insights into exocytosis using liposomes and lipid nanotubes
- Ann-Sofie Cans*,
- Nathan Wittenberg†,
- Roger Karlsson*,
- Leslie Sombers†,
- Mattias Karlsson‡,
- Owe Orwar‡, and
- Andrew Ewing†,§
- *Department of Chemistry, Göteborg University, S-41296 Göteborg, Sweden; †Department of Chemistry, Pennsylvania State University, University Park, PA 16802; and ‡Department of Physical Chemistry and Microtechnology Center, Chalmers University of Technology, S-41296 Göteborg, Sweden
-
Communicated by Allen J. Bard, University of Texas, Austin, TX (received for review August 22, 2002)
Abstract
Exocytosis is the fundamental process underlying neuronal communication. This process involves fusion of a small neurotransmitter-containing vesicle with the plasma membrane of a cell to release minute amounts of transmitter molecules. Exocytosis is thought to go through an intermediate step involving formation of a small lipid nanotube or fusion pore, followed by expansion of the pore to the final stage of exocytosis. The process of exocytosis has been studied by various methods; however, when living cells are used it is difficult to discriminate between the molecular effects of membrane proteins relative to the mechanics of lipid–membrane-driven processes and to manipulate system parameters (e.g., membrane composition, pH, ion concentration, temperature, etc.). We describe the use of liposome–lipid nanotube networks to create an artificial cell model that undergoes the later stages of exocytosis. This model shows that membrane mechanics, without protein intervention, can drive expansion of the fusion pore to the final stage of exocytosis and can affect the rate of transmitter release through the fusion pore.
Exocytosis involves fusion of a neurotransmitter-containing vesicle, typically ranging from 50 nm to ≈1 μm in diameter with the plasma membrane of a cell that is ≈10 μm in diameter and displays a fascinating mechanical and molecular complexity. However, the extremely small size and high chemical diversity of exocytotic events have made it difficult to piece together all of the molecular and biophysical mechanisms involved during the release process. Nanoscale measurements (1–8) and molecular biology applied to cellular model systems have been used to provide insights into exocytotic release and strong evidence that protein networks are involved in formation of the initial fusion pore (9, 10). In contrast, leakage of transmitter through the fusion pore and expansion of this pore to the final stage of exocytosis is poorly understood, and new models are needed to differentiate mechanisms involving proteins and membrane mechanics in driving the various stages of the exocytotic process.
A great deal of what we understand about membrane fusion has been obtained by using lipid models. A common model has involved vesicles containing channel proteins that are driven by osmotic pressure to fuse with a planar lipid bilayer (11, 12). An adaptation of this model has been used to demonstrate transient opening of fusion pores in protein-free membranes, suggesting that indeed proteins might not be needed for this process (13). Liposomes have been described as artificial cells and have also been used to examine membrane fusion (14–17). Recently, an electroinjection technique has been developed that makes it possible to form lipid nanotubes and networks between liposome reservoirs (18, 19). In this article we describe the use of this technology to develop a protein-free liposome system where a vesicle is formed inside a surface-immobilized liposome as an artificial cell that undergoes the latter stages of exocytosis. Fluorescence microscopy and amperometry were used to detect leakage of transmitter through a nanoscopic fusion pore and quantal release during the final stage of exocytosis.
Materials and Methods
Liposome–Lipid Nanotube Preparations.
Surface-immobilized, unilamellar liposomes and nanotube networks were prepared from soybean lecithin by a dehydration–rehydration method carried out on a borosilicate coverslip that was placed directly on the microscope stage as described (20). Briefly, a small glass micropipette is electrified to transiently disrupt the membrane and assist insertion into a unilamellar liposome. Lipid adhesion to the tip during withdrawal results in a lipid tube. For fluorescence measurements, 5 μM fluoroscein sodium salt (Sigma) in phosphate buffer (5 mM Trizma base/30 mM K3PO4/30 mM KH2PO4/1 mM MgSO4/0.5 mM EDTA adjusted to pH 7.4 with H2SO4) was injected into the lipid tube via the microinjection pipette to inflate each vesicle. For amperometric measurements, 1 mM catechol (Sigma) in phosphate buffer was used.
Bright-Field and Fluorescence Microscopy.
Bright-field imaging was monitored with a Sony Exwave HAD charge-coupled device camera (Sony Medical Systems, Park Ridge, NJ) on an Olympus IX-70 DIC Microscope (Olympus America, Melville, NY) through a ×20 objective. Fluorescence was performed by using an inverted reflected light fluorescence observation attachment, IX-FLA, with a HQ FITC cube (Olympus America) outfitted to the microscope system. Video microscopy imaging was collected with studio dv software (Pinnacle Systems, Mountain View, CA) on a personal computer at 30-ms intervals.
Amperometric Measurements.
Carbon fibers were sealed in pulled glass capillaries with epoxy (Epoxy Technology, Billerica, MA). After beveling on a micropipette beveler (model BV-10, Sutter Instruments, Novato, CA) at an 85° or 45° angle (Fig. 1 e), testing in 0.1 mM dopamine, electrodes were placed against the artificial cell with a micromanipulator (MP-85 Huxley wall-type micromanipulator on a MT-70-9 micromanipulator stand, Sutter Instruments) and current was monitored at 0.70 V vs. a Ag/AgCl reference electrode with an Ensman Instruments (Bloomington, IN) EI-400 potentiostat (filter frequency = 1 kHz). Data were collected at 20 kHz with a Digidata model 1322A interface and recorded with AXOSCOPE 8.1 software (Axon Instruments, Foster City, CA).
Formation and release of vesicles in an artificial cell. (a–d) Schematics of a microinjection pipette electroinserted into the interior of a unilamellar liposome and then through the opposing wall, pulled back in to the interior, followed by spontaneous formation of a lipid nanotube and formation of a vesicle from flow out of the tip of the micropipette. (e) Nomarski image of a unilamellar liposome, with a multilamellar liposome attached as a reservoir of lipid, microinjection pipette (i), electrode for electroinsertion (ii), and 30-μm diameter amperometric electrode beveled to a 45° angle (iii). A small red line depicts the location of the lipid nanotube, which is difficult to observe in the computer image with a ×20 objective, illustrating a vesicle with connecting nanotube inside a liposome. (f–i) Fluid injection at a constant flow rate results in growth of the newly formed vesicle with a simultaneous shortening of the nanotube until the final stage of exocytosis takes place spontaneously and a new vesicle is formed with the attached nanotube. (j–m) Fluorescence microscopy images of fluorescein-filled vesicles showing formation and final stage of exocytosis matching the events in f–i. (Scale bar represents 10 μm.)
Measurements at Pheochromocytoma (PC12) Cells.
Stock PC12 cells from the American Type Tissue Culture were maintained as described (21). Electrodes were placed on cells and held at +0.65 V vs. a locally constructed sodium-saturated calomel reference electrode by using a commercially available patch-clamp instrument (Axopatch 200B, Axon Instruments). The output was digitized at 5 kHz and filtered at 2 kHz by using an internal four-pole low-pass Bessel filter. Data were displayed in real time (AXOSCOPE 1.1.1.14, Axon Instruments) and stored to a computer with no subsequent filtering.
Results and Discussion
Artificial Exocytosis: The Model.
The artificial cell described here mimics the complex dynamic cellular membrane process of exocytosis. A vesicle is formed inside a surface-immobilized liposome via electroinjection (Fig. 1 a–d). The vesicle is connected to the artificial cell membrane by a lipid nanotube. This nanotube initially resembles an elongated fusion pore. Microinjection of fluid into the pulled lipid nanotube is used to inflate a daughter vesicle inside the artificial cell (Fig. 1 d). Inflation of the pulled nanotube leads to a local increase in membrane tension. To reduce this tension difference, lipid material flows from regions of lower tension (outer membrane) along the nanotube toward higher tension, forming the membrane of the small vesicle. This transpore flow of membrane follows the theory of Chizmadzev and colleagues (22, 23) and has been observed experimentally in nanotube-vesicle networks (20). Fig. 1 e shows a Nomarski image of a vesicle formed inside an artificial cell under pressure from the pipette. Because the distance between the injection tip and the outer membrane is fixed, the lipid nanotube shortens when the vesicle is expanded. As the vesicle membrane approaches the artificial cell surface, a transition from a tube of cylindrical geometry to a toroid-shaped fusion pore takes place. At this stage the system directly mimics a cell before undergoing the later stages of exocytosis, and the vesicle diameter for exocytosis is determined by the distance between the pipette tip and the membrane of the artificial cell. Provided that the pore radius is large enough, it will expand in size exponentially driven by total tension in the membrane system (23). Thus, in a system held at high surface tension, the vesicle membrane will be rapidly integrated with the outer membrane, leading to the final stage of exocytosis (Fig. 1 f–i). This artificial exocytosis resembles that for cellular release of large dense core vesicles in many ways. However, differences include the absence of the dense core, the fusion proteins, and perhaps the size of the fusion pore. The generally accepted diameter of the fusion pore in real cells has been estimated from patch-clamp experiments to be 1–3 nm in diameter (24). However, images of chromaffin cells obtained with transmission electron microscopy reveal pore structures between 20 and 300 nm (1, 25), and recent atomic force microscopy measurements on pancreatic acinar cells suggest that some fusion pores could be 150 nm in diameter (26). The nanotube of the artificial cell model has an estimated diameter of 100–300 nm (20).
Measuring Release: Fluorescence Microscopy and Amperometry.
Fig. 1 j–m shows four corresponding frames of video fluorescence microscopy displaying exocytotic release from a vesicle loaded with fluorescein dye. As the vesicle expands to the point of the final stage of exocytosis, dye is released in a hemispherical zone around the point of release. The images demonstrate vesicle expansion (Fig. 1 j), final stage of exocytosis (Fig. 1 k and l), and formation of a new vesicle (Fig. 1 m). Amperometry was combined with fluorescence imaging to define the stages of artificial exocytosis. In the amperometry experiments we used a carbon fiber electrode (shown in Fig. 1 e) placed at the nanotube exit site on the artificial cell to monitor release of vesicles filled with catechol. Four consecutive fluorescence images of vesicles filled with both catechol and fluorescein are directly correlated to the electrochemical trace in Fig. 2. In this image the vesicle is observed undergoing exocytosis opposite the electrode surface, which is outlined to the right of each image. The first two fluorescence images look similar to those in Fig. 1; however, the electrode blocks the outward hemispherical diffusion observed in Fig. 1 l and m.
Fluorescence microscopy correlated to amperometric detection of exocytosis of a 4-μm radius vesicle containing fluorescein and catechol solution. A carbon fiber electrode was placed next to the nanotube connection point of the liposome wall for amperometric detection. A dashed line is used to mark the location of the electrode beveled at an 85° angle. Scattering of the light source off the glass insulating the tip of the electrode has been digitally subtracted for clarity.
Artificial exocytosis with an expanding vesicle attached to a micropipette allows two significant advantages in these experiments. First, the control of solution pressure in the micropipette allows variation of solution flow rate and, therefore, the rate of vesicle expansion. Second, as the lipid nanotube remains after exocytosis, new vesicles are continually formed (Fig. 1 i and m); thus, controlled release of a vesicle can be repeatedly carried out while maintaining membrane integrity.
Dynamics of the Final Stage of Exocytosis.
Repeated release events can be monitored with high temporal resolution and exquisite control of system parameters, and these events can be compared with data obtained for exocytosis at cells in culture. Three consecutive current transients observed for release from a 5.7-μm radius vesicle are compared in Fig. 3 a to a current trace observed for dopamine release at a PC12 cell shown in Fig. 3 b. The release events from the artificial cell model occur at longer time scales, but are qualitatively similar to those observed in a PC12 cell. The events observed at PC12 cells have considerable variation owing to their biological diversity, whereas events observed at the artificial cell model are highly consistent and can be systematically controlled.
Amperometric monitoring of repeated exocytosis events at artificial cells and cells. (a) Amperometric detection of continuous exocytosis of three vesicles from an artificial cell. (Scale bar is 40 pA × 3,000 ms.) (b) Amperometric detection of dopamine exocytosis from a PC12 cell. (Scale bar is 10 pA × 40 ms.) (c) Plot of half-width vs. vesicle radius for vesicles fusing from an artificial cell where the vesicle radius has been the only parameter varied in the experiment.
The temporal characteristics of the electrochemical response to artificial release, measured as the width of the peak at half its amplitude, decreases with decreasing vesicle size (Fig. 3 c). In fact, as the vesicle radius approaches a few micrometers, the half width is between that observed for mast cells [80–480 ms (4); 350-nm average radius (27)] and that for adrenal chromaffin cells [11 ms (8); 170-nm radius (28)]. The plot of half width of the electrochemical response, when plotted relative to vesicle volume, is linear (correlation coefficient = 0.99) with an intercept of 40 ms. These half widths, observed in a protein-free artificial cell system, are consistent with the time for exocytosis observed at cells with large vesicles and are consistent with the theoretical calculations by Chizmadzhev and colleagues (23, 29) for exocytotic systems driven by membrane mechanics. Importantly, this system leaves an intact nanotube before consecutive release events. The residual nanotube apparently gives rise to a small counterforce drag on the fusing vesicle, leading to a nonzero intercept for the exocytosis half width. Lipid composition and system temperature could also be involved in creating the nonzero intercept. Although the conclusions drawn for large dense core vesicles should be accurate, this limitation does not allow us to draw conclusions concerning small synaptic vesicles.
Transport in the Nanotube: A Model of Release via the Fusion Pore.
In addition to being an intermediate to the final stage of exocytosis, the fusion pore is an important structure allowing early release of transmitter. By combining patch-clamp measurements of the dynamics and size of fusion pores with amperometric measurements of release, it has been confirmed that secretion occurs through the fusion pore in adrenal chromaffin and beige mouse mast cells (3, 4). This type of secretion is observed as a prespike increase in oxidation current in the first and second transients shown in Fig. 3 b and obtained from PC12 cells, indicating that transmitter is leaking out of the fusion pore and is oxidized before the final stage of exocytosis. Exocytosis in the artificial cell can be controlled to demonstrate this prespike amperometric “foot” as shown in Fig. 4 a. A schematic correlating the stages of exocytosis to an amperometric trace is shown in Fig. 4 b. Using higher flow rates in the injection pipette during artificial vesicle formation and expansion, a foot is consistently observed, and its duration and amplitude can be characterized from the amperometric trace.
Amperometric monitoring of release via an artificial fusion pore. (a) Amperometric detection of release from a 5-μm radius vesicle showing prespike feet (arrows), indicating catechol transport through the lipid nanotube or fusion pore. (Scale bar is 80 pA × 500 ms.) (b) Time correlation of vesicle growth, transport of transmitter through the lipid nanotube, and the final stage of exocytosis with amperometric detection. (c–e) Plots of foot length (c), foot area (d), and the ratio of foot area over foot length (e) observed with amperometry for vesicles fusing with an artificial cell at three different pressures used to inflate the vesicles. (f) Schematic model of the factors affecting flow in the vesicle and nanotube of the artificial cell.
Data collected from the artificial cell system suggest that the time course for leakage (foot length) through the fusion pore is governed by injection flow rate, as determined by pipette pressure, and is proportional to the size of a vesicle immediately before release (Fig. 4 c). The total amount of material leaking through the lipid nanotube (foot area) is proportional to vesicle size at the stage directly before release but is independent of injection flow rate (Fig. 4 d), indicating that nanotube length is the critical parameter determining total leakage. However, at any given vesicle size the rate of leakage through the lipid nanotube (foot area/foot length) varies with flow rate (pipette pressure) and vesicle size (Fig. 4 e). These observations can be explained by considering the sources of liquid flow inside the nanotube (Fig. 4 f). Lipid is transported from the artificial cell into the vesicle during expansion, resulting in shear flow of the liquid column inside the nanotube and toward the vesicle (20). Shear flow of the internal solution has a linear velocity v l. This is defined by v l ∝ (σb − σa)/2η, where η is the viscous resistance of the system and σa and σb are the tensions of the cell and vesicle membranes, respectively. In contrast, solution pressure from the pipette results in Poiseuille flow from the interior of the vesicle that opposes the direction of the shear flow. Poiseuille flow is is expected to have an average linear velocity, v 0 = (Δp/L t) × (r t 2/8ηw), where rt is the tube radius, Lt is the tube length, ηw is the viscosity of water, and Δp is the pressure difference between the cylinder exit and entrance. As the vesicle volume expands at a constant rate, the growth of surface area slows as the surface area-to-volume ratio decreases. Shear flow is based on lipid movement along the nanotube to accommodate the surface area of the expanding vesicle and its rate will decrease as the vesicle expands and the surface area to volume ratio of the vesicle decreases. Simultaneously, the length (L t) of the lipid nanotube decreases, resulting in increased Poiseuille flow, leading to leakage through the nanotube. Thus, the net flow of fluid inside the lipid nanotube increases in the direction of Poiseuille flow as the vesicle grows, leading to increased leakage of transmitter through the nanotube.
Observation of leakage through the lipid nanotube is equivalent to that through an extended fusion pore. The results presented here support the conclusion that the rate of fluid transport through the nanotube in the model cell depends on the length of the lipid nanotube and the factors affecting the counterbalance of shear and Poiseuille flow in the nanotube. Although this system differs from a cell in that the nanotube is changing length until the final stage of exocytosis, the observations involving lipid flow and pressure are important for understanding these factors in cellular exocytosis. In a cell this rate of transport corresponds to a lipid flow from the plasma membrane to the vesicle to counterbalance the tension difference between the membranes at the moment of exocytosis. Importantly, the leakage of transmitter through the fusion pore is regulated by this flow of lipid through the pore. Interestingly, lipid transfer through the fusion pore before expansion has been hypothesized in cell experiments (22, 30).
Conclusions
The data here are strikingly similar to what is observed during exocytosis in cells. Thus, the artificial cell model can be used to examine fundamental aspects of exocytosis with a great deal of control of many experimental degrees of freedom. These include membrane and solution composition, differential pH values across vesicle membranes, temperature, and vesicle size. Our data clearly suggest that membrane mechanics can drive the expansion of the fusion pore to the final stage of exocytosis at these time scales without the need for protein intervention. Proteins may serve to regulate and control the docking of vesicles and formation of the fusion pore (9, 10, 31) or to create pressure inside the vesicle from an expanding dense core (23, 32, 33). This latter effect could clearly be a driving force for exocytosis at adrenal cells where the dense core appears to fill the entire vesicle. However, looking at the protein-free system of the artificial cell it appears that the minimization of the membrane elastic energy for large vesicles can be the driving force for the final stage of exocytosis. This is an intriguing discovery as it counters the concept that proteins are required to drive this process.
Acknowledgments
A.E. thanks the National Science Foundation and O.O. thanks the Swedish National Science Foundation and the Swedish Foundation for Strategic Research for support of this work.
Footnotes
-
↵ § To whom correspondence should be addressed. E-mail: age{at}psu.edu.
- Copyright © 2003, The National Academy of Sciences
