GVs were formed using a standard electroformation chamber with two parallel thin platinum wires as electrodes. Phospholipid (1-palmitoyl-2-oleoylphosphatidylcholine (POPC)) was deposited on the electrodes as thin film. After filling the chamber with ultrapure water, sinusoidal AC voltage was applied between the electrodes. The conditions were similar to that used in microinjection into GVs by others [10].

At 120 min after the application of electric voltage, many spherical GVs of 20–60 μm in diameter were seen on relatively thick lipid deposit (Figure 1A). At the place where the lipid deposit was thin, semispherical “mushrooms” were formed (Figure 1B) [10].

Then, micromanipulation to the membranous objects with a glass microneedle filled with Ficoll solution was attempted. The presence of Ficoll can be detected as change in brightness under the phase contrast observation, and the compound was used as a model substance to monitor encapsulation in vesicles. Figure 2 shows the insertion of a microneedle into a mushroom-like semi-vesicle and microinjection. Successful encapsulation of injected Ficoll was observed as reported by others [10]. It should be noted that the size of the semi-sphere slightly increased upon the microinjection [10]. The extra lipid membrane could have come from the nearby lipid deposit on the electrode.

When the tip of a microneedle was pressed against a semi-spherical membrane, the needle sometimes did not break into the membrane. Instead, that part of the membrane caved in with the needle. Then, in a successful attempt, pulling out the needle left a GV in the interior of the semi-sphere (Figure 3). A small amount of Ficoll that was bleeding from the tip of the needle was encapsulated in the newly formed inner GV as can be seen from its dark interior.

By repeating this “microvesiculation” process, multiple GVs could be produced in the interior of a single semi-sphere (Figure 4). The structure may be of interest as it has topological resemblance to a biological cell that usually possesses multiple organelles.

It should be noted that the inner GVs seemed to be in contact with one another but did not merge, indicating that those GVs were separated by membrane. In Figure 4, the formation of the three inner GVs hardly changed the size of the outer semi-sphere. In microvesiculation of semi-spherical membrane, a part of the membrane of the inner GVs seems to come from lipid deposit on the electrode, as for microinjection.

Breaking the outer membrane by a microneedle after microvesiculation released the inner GVs (Figure 5). This observation indicates that an inner GV is in fact independent of the outer membrane.

The result suggests that the process could also be used as an option for controlled formation of GVs with desired inner contents. With ordinary microinjection, only a small amount of a solution can be introduced into a vesicle because the vesicle has limited capacity and is already almost full. The dilution of the injected solution in the vesicle would be inevitable, and this could complicate the control of the concentration in the vesicle. Meanwhile, preparation of GVs in a solution containing desired substances could encapsulate only a small part of the solution into the GVs, and the efficiency is usually low. In the present case, the whole contents of the formed GV directly came from the solution filled in the microneedle. The concentration should be essentially unchanged, and a minimal amount of the sample would be consumed.

Occasionally, the membrane did not close, and microvesiculation failed. In Figure 6, an inner vesicle with Ficoll seemed to have formed once, but removing the needle left a cave (Figure 6B). Then, the Ficoll was discharged as the membrane restored its spherical shape (Figure 6C).

Sometimes, a thin tether connecting the inner incomplete sphere and the outer membrane was seen after pulling out a microneedle (Figure 6D). After a while, spontaneous disconnection of the tether occurred, and microvesiculation completed. This observation suggests some similarity in the process between microvesiculation and the vesicle inflation technique developed by Karlsson and coworkers for the construction of networks of interconnected vesicles [7,8]. In the latter technique, a lipid nanotube is pulled out from a mother vesicle by the tip of a microneedle, and then a daughter vesicle is formed by injecting solution through the needle to inflate the membrane at the end of the nanotube. In microvesiculation, a microneedle first pushes the membrane of a mother vesicle into the interior to produce a cave, then bleeding Ficoll solution through the needle inflates the membrane at the closed end of the cave, and removal of the needle leaves a daughter vesicle with a tether.

Under the conditions of the present study, approximately 50% of attempted micromanipulation on semi-spheres resulted in successful microvesiculation. Very subtle difference in the conditions seems to decide between microvesiculation and microinjection. In fact, a single semi-sphere could first be the subject of microinjection and then underwent microvesiculation by using the same needle (Figure 7).

This observation suggests that the factor deciding between microvesiculation and microinjection could not simply be such things as the preparation of membrane or a microneedle. Membranes produced by electroformation under various electrical conditions, along with microneedles of different diameters or cut angles of their tips, were tested (data not shown), but no apparent correlation between those factors and the mode of micromanipulation has been seen. We are presently looking further into this point. Meanwhile, this uncertainty in microvesiculation should usually cause little problem in application of the procedure for construction of a plain oligovesicular system. There are many semi-spheres or GVs available, and after repeating the attempt a few times, one may be able to have a satisfactory result.

Unlike semi-spheres that are firmly immobilized, micromanipulation of spherical GVs formed at the top of lipid deposit (Figure 1, left) was difficult because the GVs often escape from the needle tip when they are pressed against it. By detaching a GV from the lipid deposit and holding it on a pipette, microvesiculation to the spherical GV was possible, resulting in an OVV (Figure 8). However, with this procedure, most attempted micromanipulation became microinjection, and approximately only 10% of attempts resulted in successful microvesiculation. Other than possible difference in the characteristics between a semi-sphere on an electrode and a GV on the tip of a glass pipette, lack of the lipid source of an inner GV membrane could be a reason. Similar necessity of a lipid source was also seen in the construction of networks of GVs and lipid tethers by the vesicle inflation technique [7,8].

A more practical way to obtain an OVV is scraping off a semi-sphere after microvesiculation from an electrode by using a microneedle (Figure 9). Upon the detachment, the semi-sphere closed to an OVV. This approach does not require the extra step of holding a GV. More importantly, the protocol allows multiple operations on a single firmly attached semi-sphere before the final detachment and the formation of an OVV.

One of the essential points for an OVV to be an advanced membrane model is that the interior aqueous phase of its inner GV has desired contents. The OVV obtained in this study by microvesiculation and scraping has designated contents in the inner GV, in the present case, Ficoll as a model substance, and fulfills the requirement.

1-Palmitoyl-2-oleoylphosphatidylcholine (POPC) was purchased from Avanti Polar Lipids (Alabaster, AL, USA). The phospholipid was checked with thin layer chromatography on a silica gel plate (Silicagel 70 Plate-Wako from Wako Pure Chemicals (Osaka, Japan)) developed using chloroform/methanol/water (65:25:4, v/v/v), and only a single spot was seen. Platinum wire (purity 99.98%, diameter 0.50 mm) was obtained from Nilaco (Tokyo, Japan). Methanol and diethyl ether were of the analytical grade and products of Wako Pure Chemicals. Ficoll 400 was a product of Pharmacia Biotech (Uppsala, Sweden).

Two parallel thin platinum wires (diameter 0.5 mm, separation 5 mm) were placed at 0.5 mm from the bottom of a round polystyrene well (inner diameter 40 mm, thickness 5 mm) that had a square glass observation window (15 mm × 15 mm) at the bottom. The conditions of electroformation were similar to that used in microinjection into GVs by others [10]. A solution of POPC dissolved in a mixture of diethyl ether and methanol (9:1 by volume) was carefully deposited on the electrode as a tiny droplet (0.50 μL, 0.5 mg lipid/mL). After spontaneous evaporation of the organic solvent, the lipid deposit was further dried under reduced pressure using a water aspirator for 2 h. The formation chamber was then filled with Milli-Q grade ultrapure water, and sinusoidal ac voltage (2.0 Vpp, 10 Hz) was applied between the electrodes from a function generator (Kenwood TMI FG-272, Yokohama, Japan). Formation of GVs was observed on an inverted optical microscope equipped with phase contrast options and a digital image enhancement system (Olympus IX-50, Tokyo, Japan).

A hollow needle was prepared from a borosilicate glass capillary with filament (Narishige GD-1, Tokyo, Japan; the outer and inner diameters were 1.0 mm and 0.6 mm, respectively) using a puller (Narishige PC-10, Tokyo, Japan). The tip of a typical microneedle had the outer diameter of approximately 5 μm and the cut angle of 35 degrees. The needle was attached to a microinjector (Narishige IM-300, Tokyo, Japan) and filled with an aqueous solution of Ficoll 400 (40 mg/mL).

Microvesiculation was attempted 30 times, and the success rate was calculated. Two persons independently tried the procedure, and the rates were similar.