Visualization and Quantification of Transmembrane Ion Transport into Giant Unilamellar Vesicles**

Transmembrane ion transporters (ionophores) are widely investigated as supramolecular agents with potential for biological activity. Tests are usually performed in synthetic membranes that are assembled into large unilamellar vesicles (LUVs). However transport must be followed through bulk properties of the vesicle suspension, because LUVs are too small for individual study. An alternative approach is described whereby ion transport can be revealed and quantified through direct observation. The method employs giant unilamellar vesicles (GUVs), which are 20–60 μm in diameter and readily imaged by light microscopy. This allows characterization of individual GUVs containing transporter molecules, followed by studies of transport through fluorescence emission from encapsulated indicators. The method provides new levels of certainty and relevance, given that the GUVs are similar in size to living cells. It has been demonstrated using a highly active anion carrier, and should aid the development of compounds for treating channelopathies such as cystic fibrosis.

Giant Unilamellar Vesicles (GUVs) were grown in Dex-PEG (1:1 ratio) coated microscope glass slides substrates as described previously. 2 Lipid solution (10 µL) was deposited on a hydrogel coated glass slide, then the lipid solution was dried by evaporating the chloroform under a gentle stream of Nitrogen gas. A liquid chamber was made by placing a 15 mm (OD) glass O-Ring on top of the hydrogel and sealed with high vacuum silicon grease. The lipid film was rehydrated by adding 400 μL of an aqueous solution that contains lucigenin (0.8 mM), NaNO 3 (225 mM) and sucrose (200 mM), into each chamber. GUVs were grown during at least 3 hours at room temperature and subsequently the solution with free floating GUVs was transferred into an eppendorf tube containing 600 μL of NaNO 3 (225 mM) and glucose (200 mM) aqueous solution. The visualization chamber was pre-treated with an aqueous solution of BSA (1 mg/mL, 500 µL) for one hour. The diluted solution with GUVs was transferred into the microscopy visualization chamber (500 µL per well). The GUVs were left to sediment for at least 30 minutes and the excess of non-encapsulated lucigenin was replaced by the NaNO 3 and glucose solution using a peristaltic perfusion pump (Instech P720, ~ 0.4 mL/min, 30 minutes), to produce a clean background for imaging the encapsulated lucigenin dye in the GUVs.
During the imaging of the GUVs in a time lapse experiment, 25 µL 1M NaCl (in NaNO 3 and glucose solution) was added to the well after 30-60 seconds with a microsyringe, giving a final NaCl concentration of ~50 mM.
All imaging was performed on a Leica TCS SPE confocal microscope system. Illumination was provided by a solid state laser using the 488 nm laser line (15% laser power) for scanning lucigenin's fluorescence and 532 nm laser line (10-15% laser power) for scanning lissamine rhodamine's fluorescence. Confocal microscopy was carried out using a 20× dry objective. The analysis of the images was performed in ImageJ software, by measuring the average intensity of an area corresponding to one GUV for the series of time lapsed confocal image frames.
The osmolality of a NaNO 3 (225 mM) solution was determined from the freezing point depression using an Osmometer Roebling Type 13 (calibrated using 100 mOsm/kg NaCl standard solution) and found to be 378 mOsm/kg (standard PBS buffer was 310 mOsm/kg). Addition of 50 mM NaCl to 225 mM NaNO 3 causes the osmolality to increase to 473 mOsm/kg. Figure 3

Fitting of the fluorescence intensities of individual GUVs
According to the Stern-Volmer relationship: the fluorescence in absence of quencher (F 0 ) divided by the fluorescence (F) is proportional to the concentration of quencher [Q], which is chloride in our experiments. In the Stern-Volmer equation k q is the rate constant of the quenching process and τ 0 is the life time of the emitting state of lucigenin, which are both constants. This allows us to take the reciprocal of our normalized fluorescence traces (which are F/F 0 ) to get F 0 /F curves, of which the shape directly represents the concentration of chloride in the vesicles over time.
The first 1000 s of the F 0 /F curves obtained for individual GUVs (0.01% transporter) are fitted with a single exponential decay function: Examples of representative fits are given in Figure S13.

Figure S13
Single exponential fits (red lines) of F 0 /F data of individual GUVs (black dots). Note that the noise level of the measured data increases over time. Due to the decreasing fluorescence intensity, the error in measuring the average fluorescence intensity of the area corresponding to a GUV increases.
We note that the increase of the noise in the curves presented in Figure S13 is caused by the decrease of fluorescence intensity upon quenching. The error in measuring the fluorescence intensity thus increases over time when transport is occurring.
The value obtained for parameter k is the rate constant for the transmembrane transport process. Since the fits are generally good we can assume that the concentration of chloride inside the GUV [Cl -] increases from 0 to 50 mM following the exponential decay function given by equation 2: (2) The rate constant k is obtained from the fit. Differentiation of equation 2 at t=0 gives the initial rate of chloride transport into the GUVs as 50k in mM s -1 . From the measured diameter of the GUVs, the volume and the surface area can be calculated. Subsequently, the initial rate (in mM s -1 ) can be multiplied by the volume of the GUV (V GUV ) and Avogadro's number (N A ) to get the initial rate as number of chloride anions transported into the GUV per second. Dividing this by the calculated number of transporters in the GUVs (two times the surface area k = 0.0023 ± 0.0001 s -1 R 2 = 0.93 k = 0.0018 ± 0.0001 s -1 R 2 = 0.95 S11 A GUV , divided by the surface area per lipid which is assumed to be 46.97 Å 2 for POPC/cholesterol 7:3 4,5 times the ratio of transporter to lipid) gives the initial rate of transport per transporter (I transporter ) in chloride anions per second (equation 3). (3) Data on 56 GUVs from 6 experiments were analyzed this way, giving an average initial rate per transporter of 820 ± 260 Cls -1 .

The half-life of the transport process increases linearly with the diameter of the GUVs
The rate of transport by one transporter molecule is independent of the size of the GUV. However, the firstorder rate constant for the transport of chloride into the GUVs (k), as obtained from the fits, depends on the volume of the GUV (V GUV ) and on the number of transporters present in the membrane and thus on the surface area (A GUV ), according to equation (3). From this it can be shown as follows that, for a given transporter at a given concentration, the half-life (t ½ ) is directly proportional to the diameter of the GUVs: As Itransporter is a constant, must also be constant, indicating that For half-life (t ½ ) one can therefore write: (where is the diameter of the GUVs) Figure S14a gives a plot of the half-lives obtained for the 16 GUVs from one experiment, which show a nearly linear relationship with diameter of the GUVs. The plot in Figure S14b shows the half-lives from all the GUVs of six experiments (all with 0.01% transporter). The latter plot shows more deviations from the linear relationship between the half-life and the diameter, but the main trend is still linear, indicating how the size of the GUVs affects the observed rate of quenching of the fluorescence.

Test for leakage of a fluorescent dye from the GUVs
To exclude the possibility that the decrease of fluorescence of the GUVs is caused by lucigenin leaking from the GUVs via defects formed by the transport and chloride, we repeated the experiment with carboxyfluorescein as the fluorescent dye. 6 As the fluorescence of carboxyfluorescein is not quenched by chloride, we should be able to distinguish between the dye leaking from the GUVs and transport of chloride into the GUVs.

Figure S15
Series of bright-field (a.) and fluorescence microscopy images (b.) of a GUV made of POPC (70%), cholesterol (30%), and transporter 1 (0.1%) containing carboxyfluorescein (0.1 mM, pH 7) and NaNO 3 (225 mM, interior and exterior). The confocal fluorescence imaging is performed with a 488 nm laser at 10% power and a detection filter band between 500-600 nm. Upon the addition of NaCl (25 µL, 1 M) the GUVs shrink producing an increase in the fluorescence, which slightly decreases once the inner content and the exterior reach the equilibrium. c. Comparison of the experiments with carboxyfluorescein and lucigenin. The red line corresponds to the average fluorescence of lucigenin with 0.1 % transporter preincorporated in the membrane (see Figure 3 of the main paper). The green line shows the average fluorescence of carboxyfluorescein of 5 individual GUVs with 0.1 % transporter preincorporated in the membrane. The starting time (t = 0 s) corresponds to the addition of 25 µL NaCl (1M).
The results in Figure S15 show that carboxyfluorescein does not leak from the GUVs. We can therefore conclude that the addition of sodium chloride to the GUVs with transporter does not cause the formation of defects or channels in the membrane. This confirms that the observed quenching of fluorescence of lucigenin is caused by chloride transport into the GUVs. c.