Effect of electrical parameters and cholesterol concentration on giant unilamellar vesicles electroformation

Abstract

Giant unilamellar vesicles (GUVs) are used extensively as models that mimic cell membranes. The cholesterol (Chol) content in the fiber cell plasma membranes of the eye lens is extremely high, exceeding the solubility threshold in the lenses of old humans. Thus, a methodological paper pertaining to preparations of model lipid bilayer membranes with high Chol content would significantly help the study of properties of these membranes. Lipid solutions containing 1-palmitoyl-2-oleoyl-glycero-3-phosphocholine (POPC) and Chol were fluorescently labeled with phospholipid analog 1,1’-dioctadecyl-3,3,3’3’-tetramethylin docarbocyanine perchlorate (DiI) and spin-coated to produce thin lipid films. GUVs were formed from these films using the electroformation method and the results were obtained using fluorescent microscopy. Electroformation outcomes were examined for different electrical parameters and different Chol concentrations. A wide range of field frequency-field strength (ff-fs) combinations was explored: 10 - 10000 Hz and 0.625 - 9.375 V/mm peak-to-peak. Optimal values for GUVs preparation were found to be 10 - 100 Hz and 1.25 – 6.25 V/mm, with largest vesicles occurring for 10 Hz and 3.75 V/mm. Chol:POPC mixing ratios (expressed as a molar ratio) ranged from 0 to 3.5. We show that increasing the Chol concentration decreases the GUVs size, but this effect can be reduced by choosing the appropriate ff-fs combination.

1. Introduction

Giant unilamellar vesicles (GUVs) have been an important topic of research lately due to their similarity to biological membranes [1–6] and their potential as drug carriers [7–9]. Preparation methods for GUVs are still being perfected in order to obtain the best result, which in turn depends on the selected application of lipid vesicles [10]. One of the first methods used for GUVs formation was controlled hydration (known also as gentle hydration, natural swelling or spontaneous swelling), developed by Reeves and Dowben, which includes spontaneous swelling of phospholipids in aqueous solution of nonelectrolytes [11]. An advancement in preparation of GUVs came from Angelova and Dimitrov [12] who introduced liposome electroformation as an alternative to the traditional swelling method. The initial procedure was further improved by involving alternating current (AC) instead of direct current (DC) electric fields [13]. Electroformation features that made it the most commonly used method [14] are: high percentage of formed unilamellar vesicles [15,16], homogeneous size distribution [17], simple manipulation and short preparation time [16].

Since then, effects of many parameters have been examined in order to improve the yield, size and homogeneity of the obtained GUVs. Most notable parameters include the AC-field frequency and strength, lipid composition, electroformation duration, temperature and thickness of the lipid films. However, the interplay of these parameters has not yet been fully explored. For instance, most studies used frequencies of about 10 Hz and electric fields of about 1 V/mm [13,18–21]. Conversely, when exploring the field frequency-field strength (ff-fs) combinations effect on electroformation efficiency, some groups reported higher frequencies and AC-field strengths than those used before [22–24]. Li et al. found the optimal combination to be 11 kHz and 5 V/mm. For 10 V/mm best results were also obtained using a 11 kHz frequency, but some irregular vesicles appeared [23]. Similar results were obtained by Wang et al. [24]. Politano et al. [22] showed that electroformation can be successful at frequencies as high as 10 kHz (at field values of 0.212 - 2.12 V/mm) and field strengths of up to 20 V/mm (for frequencies 1 - 100 Hz).

One of the reasons for optimization of electroformation parameters is their strong dependance on the lipid composition in preparation mixtures [22]. Because of its biological significance [25,26], cholesterol (Chol) was often added to different phospholipid mixtures. However, because the Chol proportion in most mammalian cell membranes is lower than 50% [27], Chol:Phospholipid (PL) ratio used in these studies was usually lower than 1:1 [13,14,18,22–24,28,29]. Examples of systems with high Chol content are fiber cell plasma membranes of the eye-lenses [30–33] and atherosclerotic smooth muscle cells [25]. Within the human eye lens, the Chol:PL molar ratio of the fiber cell membrane ranges from 1 to 2 in the lens cortex and from 3 to 4 in the lens nucleus [30,31]. Chol is hypothesized to be crucial for maintaining lens transparency by creating Chol bilayer domains (CBDs) which ensure that the surrounding phospholipid bilayer is saturated with Chol. This keeps the bulk physical properties of the membrane consistent and unaffected by phospholipid composition changes [34–43]. In contrast, cholesterol bilayer enrichment is related to negative effects during the development of atherosclerosis.

In order to cover the range of Chol concentrations mentioned above, here we will compare samples with Chol:POPC mixing ratios ranging from 0 to 3.5. Throughout the text, we will use the term mixing ratio to define the molar concentration ratio before evaporating the solvent. The need for differentiation between mixing ratios and molar ratios in investigated membranes of GUVs arises due to possible demixing of Chol during sample preparation. This process can lead to a decreased Chol:POPC molar ratio in the bilayer when compared to the mixing ratio. The highest molar ratio was determined from the maximum solubility limit observed previously to be around 66 mol% [44,45].

This issue could be addressed by using preparation methods that avoid those stages, such as rapid solvent exchange (RSE) [46]. A study using RSE and electroformation to create vesicles with phospholipids and Chol was already conducted [29]. However, the lipid composition was chosen to achieve a coexisting liquid ordered-liquid disordered phase in the membrane (the highest Chol concentration was 46 mol%), while we are aiming for lipid compositions with Chol content higher than 50 mol%. At this high Chol content CBDs are embedded into the surrounding PL bilayer saturated with Chol forming a structured (or dispersed) liquid ordered phase. Consequently, the Chol:PL ratio used was too low to provide insight into systems such as the eye lens cortex and nucleus. Also, the electroformation field frequency and strength used in the study were kept constant at around 0.5 V/mm and 10 Hz, and in our experiment we explored a much broader range of electrical parameters.

Other electroformation studies mainly focused on providing the best electroformation parameters for different phospholipid compositions while keeping the Chol level constant. Here, however, we determined the optimal electroformation parameters for lipid membranes with Chol:PL ratios suitable for modeling the cortical and nuclear fiber cell plasma membranes of the human eye lens.

2. Materials and methods

2.1. Materials

One-palmitoyl-2-oleoyl-glycero-3-phosphocholine (POPC) and cholesterol-3β-cholest-5-en-3-ol (Chol) were obtained from Avanti Polar Lipids Inc. (Alabaster, AL). Fluorescent dye 1,1’-dioctadecyl-3,3,3’3’-tetramethylin docarbocyanine perchlorate (DiI) was purchased from Invitrogen, Thermo Fisher Scientific (Waltham, MA). When not used, the lipids were stored at −20 °C. Chemical structures and their approximate locations in the membrane bilayer are presented in Fig. 1. Other chemicals of at least reagent grade were obtained from Sigma-Aldrich (St. Louis, MO). Indium-tin oxide coated glass (ITO, CG-90 INS 115) was purchased from Delta Technologies (Loveland, LO). Glass dimensions were 25 × 75 × 1.1 mm. New ITO glass was used for each preparation in order to prevent coating deterioration. Mili-Q deionized water was used as a chamber solution.

Fig. 1.

Chemical structures of lipids and fluorescent dye and their approximate locations in the lipid bilayer membrane

2.2. Deposition of the lipid film

A common concern in electroformation protocols is uniform lipid film deposition on the electrode surface. Deposition is traditionally done by dropping the lipid solution on the ITO electrode and letting it dry [13]. This method can be further improved by spreading the film using a small object (e.g. a glass pipette or rod) before the solvent has a chance to evaporate [18,47,48]. Although this does increase the film thickness homogeneity, the method isn’ precise enough to guarantee reproducibility since film thickness can’t be accurately controlled. This issue was addressed using the spin-coating method as described by Estes and Mayer, where a lipid film is spread over the electrode surface by spinning the electrode at very large angular velocities [19]. Here, prior to spin-coating, the glass was immersed in deionized water for at least 45 min before being wiped four times with 70% ethanol moistened wipes. Properties of samples with Chol:POPC mixing ratios ranging from 0 to 3.5 were compared. The POPC:DiI molar ratio was always kept at 1:0.002. Consequently, for different Chol:POPC ratio used here, the dye:lipid concentration ratios ranged from 0:002:1 (for sample without Chol) to 0.0004:1 (for highest Chol content). A mixture containing 3.75 mg/mL of lipids in 95% chloroform, 5% acetonitrile solution was prepared [19]. The solution (350 μL was deposited onto the ITO surface and a thin lipid layer was created using a spin-coater (Sawatec SM-150). The glass was spun for 4 minutes at 600 rpm. The final velocity was reached in 1 second. After coating, the lipid film was placed under vacuum for 30 minutes to evaporate any remaining solvent.

2.3. Electroformation chamber

The electroformation chamber consisted of two 25 mm × 37.5 mm ITO coated glass electrodes separated by a 1.6 mm thick teflon spacer. Electrodes were made by cutting a 25 mm × 75 mm ITO glass slide using a diamond pen cutter. After one of the electrodes was lipid-coated, the chamber was assembled by attaching the spacer to the electrodes using vacuum grease (Fig. 2A). The opening inside the spacer is square shaped with rounded corners so air bubbles are less likely to be trapped. The entrance to the chamber was blocked using a teflon stopper (Fig. 2B). The stopper is rectangularly shaped and a bit narrower than the neck of the gap in the spacer. This small difference in width allowed for ejection of remaining air bubbles along with the surplus of fluid from the chamber. Upon insertion, the stopper was also sealed with vacuum grease. This way, there was no contact between the grease and the lipid solution inside, so the probability of contamination was decreased. After sealing, the chamber was attached to a voltage source (UNI-T UTG9005C pulse generator) and placed inside an incubator at a temperature of 60 °C. In order to assure good contact between conductor wires and the electrodes, the outer edges of the chamber were covered with copper tape (Fig. 2A). After two hours, the voltage was turned off and the chamber was kept in the incubator for another hour.

Fig. 2:

Experimental setup. A Illustration of the electroformation chamber. B Positions on the chamber surface at which microscope images were taken. All numeric values are expressed in mm

2.4. Image and data analysis

In order to search the entire volume of the chamber, we scanned 13 points on the chamber surface at 3 different depths as indicated in Fig. 2B. Images were obtained using a fluorescence microscope (Zeiss Axio - imager M1 or Olympus BX51). Vesicle diameters were measured using the line tool in Fiji software [49] and data analysis was performed using R programming language [50].

Sample distribution normality was tested through qq-plots and Shapiro - Wilk test. If the normality assumption was violated, appropriate nonparametric tests were used. Goodness of linear fit was estimated using the coefficient of determination R².

3. Results and discussion

3.1. GUVs formation for different field frequency-field strength combinations

Electroformation successfulness averaged over the whole sample depends on a variety of parameters as listed in the introduction, one of which is film thickness [12,19,24]. Even though the spin-coating method creates macroscopically homogeneous lipid films, there are microscopic differences in thickness over the sample [19]. Additionally, the ITO layer isn’t perfectly smooth, so it can contribute to variations in thickness as well (Supplementary Fig. S1).

Depending on those differences, we can have varying degrees of electroformation successfulness throughout the sample. Consequentially, averaging over the whole sample can decrease our ability to identify the best possible ff-fs combination. Therefore, we compared only the best frames in each sample. To reduce the influence of local variations in film thickness, we used a small magnification objective (10x) to make the quantification area as large as possible.

Electroformation outcomes were described using the mean and standard deviation (sd) of GUVs diameters and the number of GUVs. Even though the vesicle number is indicative of electroformation successfulness, it can be misleading when interpreted by itself because the maximum number of vesicles that can fit inside a limited area is dependent on their diameter (Supplementary Fig. S2).

This is why, alongside the GUVs number, we also calculated the coverage N/N_max for each sample, where N_max is the maximum number of vesicles that could fit inside the observed region. N_max is derived assuming all vesicles are spherical and have a diameter equal to the average diameter measured in that sample. Therefore, it is calculated solely from geometrical considerations as explained in the Supplementary. Obviously, coverage should be a positive value ranging from zero to one. An additional benefit of using coverage is the ability to compare results when using different microscope magnifications, which can’t be said for comparing the number of vesicles. Fig. 3A shows that electroformation seems to be most successful for ff-fs combinations of 10 - 100 Hz and 1.25 - 6.25 V/mm. Increasing the frequency further leads to a decrease in successfulness for both Chol:POPC ratios (1:2, 2:1), so after setting the frequency at 10000 Hz, electroformation efficacy dropped to zero. Comparing the means between all pairs for two different Chol:POPC ratios confirms the assumption that very high Chol contents reduce the electroformed GUVs diameter (mean ± standard error (sem) = 14 ± 5 μm, p = 0.01, Studen’s paired t-test) (Fig. 3B). Although not statistically significant, increasing Chol concentration seems to also decrease the coverage of the observed region (mean ± sem = 0.12 ± 0.06, p = 0.09, Wilcoxon signed-rank test) (Supplementary Fig. S3). We have also observed vesicle diameter distribution becoming narrower with decreasing average vesicle size (Fig. 4). Although such a behavior has been noted before [22], here we show that there seems to be a linear dependence between the two parameters.

Fig. 3.

GUVs electroformation successfulness. A ff-fs diagram for two different Chol:POPC ratios. In cases where vesicles were successfully formed (X points), the number above is the vesicle diameter denoted as mean ± sd, expressed in μm. Numbers given below are the number and coverage of tracked vesicles, respectively. All numbers pertain to the best frame observed for a particular sample. B Mean, sd and coverage value differences between 1:2 and 2:1 Chol:POPC ratios. ΔMean and ΔSd are expressed in μm. × symbols denote individual sample parameter differences. Black dot and error bars denote the mean difference and standard error of the differences.

Fig. 4.

GUVs standard deviation vs mean diameter, y = 0.36× +0.70 μm, R² =0.65.

3.2. Effect of cholesterol concentration on GUVs formation

Here we show how Chol concentration impacts GUVs formation at 10 and 100 Hz for field strengths covering the optimal electroformation region – 1.25, 3.75 and 6.25 V/mm (Fig. 5). Increasing Chol concentration shifted vesicle diameter distributions toward lower values. The phenomenon is probably due to Chol changing the elasticity of the lipid bilayer, thus making the bending required for GUVs formation much harder. Interestingly though, the change of Chol concentration impacts GUVs formation differently depending on the electric field parameters.

Fig. 5.

Effect of Chol concentration on electroformation successfulness. A Change of vesicle diameter distribution with change in Chol:POPC ratio. B Vesicle diameter as a function of Chol:POPC ratio. Points and bars represent means and their standard errors.

As mentioned earlier, the shift in size is accompanied by narrower distributions with an almost identical regression line slope (Fig. 4, Fig. 5A, Supplementary Fig. S4). Also, GUVs formed at higher frequencies have smaller or similar average diameters than their lower frequency counterparts (Fig. 5). This is especially visible at low voltages. The effect could possibly be explained from the lipid vibration amplitude formula derived by Dimitrov and Angelova [15]. The general form of the expression is A = $\frac{U}{af(1+bf)}$, where A, U and f are the vibration amplitude, voltage and frequency, and constants describing other system properties. The formula predicts smaller amplitudes for lower voltages and higher frequencies. Therefore, it is possible that higher frequencies in combination with lower voltages prevent the formation of very large GUVs by decreasing the vibration amplitude too much. Although the same formula predicts a higher amplitude (and therefore possibly a larger GUVs diameter), the GUVs diameter seems to grow as we increase the voltage and then drops as the voltage gets too high. The decrease might be caused by an overly high mechanical stress or by excessive oxidation of lipid molecules [51].

One of the authors (Witold Karol Subczynski), using the RSE method to prepare POPC multilamellar liposomes, shows that the formation of pure CBDs in POPC membranes starts at a Chol:POPC molar ratio of 1:1 [52]. However, as we demonstrated in our recent paper, in GUVs CBDs started to be observed only at Chol:POPC mixing ratios greater than 2:1 [53]. In that investigation GUVs were prepared using the electroformation method from the lipid film formed using the film deposition method. This indicates that the demixing of Chol in the form of Chol crystals during the preparation using the film deposition method significantly decreases the true Chol:POPC molar ratio in membranes of GUVs. Such high cholesterol concentrations affect electroformed GUVs diameters the least for the field strength of 3.75 V/mm (Fig. 5). However, between the two explored frequencies (10 and 100 Hz), the 10 Hz frequency led to a more stable mean diameter – Chol concentration dependence. Consequently, the 10 Hz – 3.75 V/mm ff-fs combination seems to be the best when creating model membranes containing CBDs.

4. Summary

Due to their role as biological membrane models, GUVs have been extensively studied lately. Nowadays, the most popular method for their creation is electroformation. Consequently, optimization of electroformation parameters for GUVs formation is very important. In our experiment, electroformation of GUVs containing varying Chol:POPC mixing ratios was conducted (from 0 to 3.5). Electroformation was characterized by calculating the vesicles diameter mean value and sd, vesicle number and coverage. Coverage was added to the characterization because the number of vesicles by itself can lead to effectiveness misinterpretation due to not taking into account the diameter dependance. A wide range of ff-fs combinations was explored: 10 - 10000 Hz and 0.625 - 9.375 V/mm peak-to-peak, with average vesicle diameters ranging approximately from 5 to 60 μm. The optimal ff-fs combinations for investigated lipid compositions are around 10 - 100 Hz and 1.25 - 6.25 V/mm. The upper electroformation limit is around 1000 Hz for frequency and around 9.375 V/mm peak-to-peak for field strength. Comparing pairwise differences in mean diameters between two Chol:POPC ratios (1:2 and 2:1) for multiple ff-fs combinations, we determined that there is a significant difference between the two Chol concentrations (p = 0.01, Student’s t-test). Although not enough to be statistically significant, vesicle coverage also seems to be negatively affected by increasing Chol concentration (p = 0.09, Wilcoxon signed-rank test).

In order to further inspect the impact of Chol concentration, we observed the change in GUVs formation for five different Chol concentrations and three different field strengths covering the optimal electroformation region (1.25, 3.75 and 6.25 V/mm). The conclusions were in line with our initial hypothesis. High Chol concentrations decreased the average GUVs diameters, although the effect strength varied depending on the ff-fs combination. High Chol concentrations affected electroformed GUVs diameters the least in the case of 10 Hz - 3.75 V/mm ff-fs combination. Consequently, that combination seems to be the best one when dealing with model membranes containing very large amounts of Chol.

Because of the large number of variables included in the electroformation process, there is still room for improvement. For instance, incubation temperature and duration, chamber dimensions and lipid film thickness could be altered in order to get better results. Also, this study used deionized water as a chamber solution. The effects of using physiological solutions under these Chol concentrations remain to be tested. When using film deposition methods, another problem to be tackled is the potential artifactual solid-state demixing of Chol during sample prepa ration. The problem could be solved by using preparation methods that avoid those stages, such as RSE.