Chemical Analysis of Lipid Boundaries after Consecutive Growth and Division of Supported Giant Vesicles

Summary

The reproduction of the shape of giant vesicles usually results in the increase of their “population” size. This may be achieved on giant vesicles by appropriately supplying “mother” vesicles with membranogenic amphiphiles. The next “generation” of “daughter” vesicles obtained from this “feeding” is inherently difficult to distinguish from the original mothers. Here we report on a method for the consecutive feeding with different fatty acids that each provoke membrane growth and detachment of daughter vesicles from glass microsphere-supported phospholipidic mother vesicles. We discovered that a saturated fatty acid was carried over to the next generation of mothers better than two unsaturated congeners. This has an important bearing on the growth and replication of primitive compartments at the early stages of life. Microsphere-supported vesicles are also a precise analytical tool.

Graphical Abstract

For a Figure360 author presentation of this figure, see https://doi.org/10.1016/j.isci.2020.101677.

Highlights

• •Protocellular model membranes
• •Growth & Division
• •Self-evolving molecular systems

Soft Matter; Materials Science; Colloids

Introduction

Studying the self-reproduction (Stano and Luisi, 2010; Szostak, 2017) of lipidic giant vesicles (GVs) (Walde et al., 2010) is crucial for understanding the replication of prebiotic compartments in autopoietic systems (Varela et al., 1974). This replication of shapes and objects can result from a growth and division (G&D) process: feeding the lipid boundary of a mother vesicle with amphiphilic compounds induces its growth in size and eventually its division into daughter vesicles (Figure 1) (Stano and Luisi, 2010; Szostak, 2017). In pioneering studies, G&D of oleic acid vesicles was carried out by feeding them with oleic anhydride that rapidly inserted into the pre-existing membranes and hydrolyzed to oleic acid and oleate at basic pH (Figure 1A). The size of these vesicles and the G&D steps have been analyzed by microscopy (Walde et al., 1994; Wick et al., 1995). In more recent investigations, a wider variety of synthetic protocells have been used to describe the phenomenon (Lopez and Fiore, 2019; Matsuo et al., 2020, 2019; Ruiz-Mirazo et al., 2014; Schwille, 2019). For instance, the observation of mother vesicles encapsulating a ferritin cargo (Berclaz et al, 2001a, 2001b) or fluorescent probes (Hanczyc et al., 2003; Zhu and Szostak, 2009) assessed their division through the distribution of their content in daughter vesicles (Figure 1B). Furthermore, GVs composed of membranes that contained two different fluorescent probes allowed to monitor their growth in size by measuring the Förster Resonance Energy Transfer (FRET) effect (Figure 1C) that took place when the membranes budded (Chen, 2004; Hanczyc et al., 2003). Other techniques such as free-flow electrophoresis enabled the chemical characterization of the vesicles obtained after a single reproduction and sorting according to their charges (Pereira de Souza et al., 2015). However, one of the major issues in studying the G&D of GVs is the impossibility to distinguish and to separate mother from daughter vesicles after the replication step. Indeed, despite the fact that the different steps of the G&D process are known, for instance, budding, evagination, tubing, pearling, dumbbell formation, and division (Figure S1), most of the lipid exchange, which actually occurs between the vesicles and the medium during the phenomenon, remains uncertain. A clear separation of mother and daughter vesicles would allow one to independently characterize their lipid compositions in order to describe more precisely and reliably the lipid movements. The use of surface-mediated vesicle replication allows for the distinction of mother from daughter vesicles after a G&D process. However, only a few examples were reported in the literature. Vesicles can be anchored to a surface by either a specific integral membrane-bound linkage or through adsorption (Rebaud et al., 2014). Vesicles may be tethered to an avidin-coated surface via biotinylated phospholipids (Pignataro et al., 2000). Surface-attached vesicles would grow through the uptake of additional membrane components such as fatty acids in the form of free molecules or micelles, or through the fusion of added phospholipid vesicles. Vesicles adsorbed to a glass surface coated with hydrocarbons have also been shown to fuse with additional vesicles that were provided from a fluid flowing above. In that case, a microfluidic device delivered phospholipidic vesicles of 30–100 nm diameter that were adsorbed to and fused with the tethered lipidic quartz surface. Two-color fluorescence signatures were used to monitor the process (Johnson et al., 2002). This process was also observed by Morigaki and Walde when fatty acid micelles were delivered instead of phospholipid vesicles (Morigaki and Walde, 2002).

Figure 1.

Reported Experiments on Vesicles Formed from Membrane Growth and/or Self-Reproduction Processes

(A) Growth and division of fatty acid (orange amphiphiles) vesicles (Walde et al., 1994; Wick et al., 1995).

(B) Cargo distribution after growth and division (Berclaz et al, 2001a, 2001b; Hanczyc et al., 2003; Zhu and Szostak, 2009).

(C) Vesicle growth observed by FRET (Chen, 2004; Hanczyc et al., 2003). Examples (A)–(C) do not allow for any distinction between mother and daughter vesicles.

(D) Tethered membranes and daughter vesicles formed by feeding glass or silica surface with fluorescent vesicles (Johnson et al., 2002; Morigaki and Walde, 2002).

(E) Preparation of glass microsphere-supported giant phospholipid (blue amphiphiles) vesicles (Gopalakrishnan et al., 2009).

(F) Growth and detachment of membranes using supported giant vesicles allow distinction and separation of mothers from daughters (Albertsen et al., 2014; Fiore et al., 2018).

We have found that glass microsphere-supported giant vesicles (g-MSGVs), being GVs filled with a functionalized glass bead of a defined size (5.0 μm) bearing a tethered membrane around this glass core (schematized in Figure 2A), can generally serve to monitor the lipids during several generations of the G&D process. We have observed that G&D arises from the budding and eventually detachment of lipid membranes that aggregate to a significant part into daughter vesicles, alike the corresponding process occurring from non-supported vesicles, commonly studied as models for protocells (Albertsen et al., 2014; Chen, 2004; Hanczyc et al., 2003; Hardy et al., 2015; Lopez and Fiore, 2019; Terasawa et al., 2012; Tomita et al., 2011; Walde et al., 1994; Wick et al., 1995; Zhu and Szostak, 2009). Such supported membrane boundaries can be submitted to several distinct feeding processes, just as any phospholipidic vesicles can, yet with the advantage of being able to separate unambiguously the daughter vesicles formed after a G&D process from the left-over surface-supported mother vesicles by centrifugation (Fiore et al., 2018). The first preparation of this construct was reported by Gopalakrishnan et al. (2009) by means of a mixture of phospholipids anchored to a substrate thanks, again, to an avidin/biotin interaction (Pignataro et al., 2000). G&D was performed by Monnard and co-workers who used decanoic acid/decanoate membranes supported on the same monodisperse glass bead microspheres of 5.0 μm diameter that we are presenting here (Albertsen et al., 2014). We reasoned that we could combine these separate studies to answer the question: how can the compositional dynamics during consecutive G&D processes of phospholipid boundaries be followed experimentally? Here we provide an optimization of our previously described method (Fiore et al., 2018), that is, a general tool for chemically monitoring the lipid exchange that occurs during several consecutive G&D processes.

Figure 2.

Process for Coating g-M₀SGVs with Different Phospholipids Using Our Method and Structure of Used Amphiphiles

(A) Glass microspheres were first tethered with avidin/biotinylated phospholipid, then coated with phospholipids (Fiore et al., 2018).

(B) Structure of the phospholipids 1–6 (without counterions for clarity) used for preparation of g-M₀SGVs and fatty acids 7–10 used for feeding supported vesicles. Color code serves to distinguish phospholipids from fatty acids in the following figures.

One key feature of the g-MSGVs is their higher density, owing to the glass bead inside, allowing for their separation by centrifugation from all lipid material floating in the host solution. Fatty acids, when supplied to g-MSGVs, partitioned into their phospholipidic boundary. This made their boundary grow in size until a fragment eventually detached to generate, among others, giant daughter vesicles (DGVs), whereas regularly sized supported mother vesicles with altered membrane composition were left behind. The heavier mother vesicles could be thus isolated from all lipids that detached from the mothers. The initial preparation of the g-MSGVs covered with phospholipids (DOPC) was recently reported (Fiore et al., 2018). Phospholipids with a low critical vesicle concentration (cvc) (Zhou et al., 1999) (phospholipids 1–4, Figure 2B) were anchored to glass microspheres using a chemical architecture made of avidin and biotin-DSPE (5, Figure 2B). The g-MSGVs were classified based on their phospholipidic composition; the list of the supported vesicles used in this study is reported in Tables 1 and S1. POPC, DOPC, POPA, and DOPA were blended in different ratios. Pure POPC membranes were named A type, a blended binary mixture of POPC-POPA (4:1 molar ratio) B type, and quaternary mixtures from POPC-POPA-DOPC-DOPA (4:1:4:1 molar ratio) C type (Tables 1 and S1). Type A⁺ g-MSGVs (POPC, Table S1) containing additional 0.2 mol % DOPE-Rh (6, Figure 2B) served for their observation under the confocal microscope to evidence G&D (cf. Figure 3B). Supported vesicles containing phosphocholine extracts from soybean and egg yolk (g-MSGVs D and E, Table S1) were prepared as well. However, in order to limit the complexity of the system, these two populations of g-MSGVs bearing a tethered “natural” membrane were not used for consecutive feeding experiments as described below. The visual aspect of the supported vesicle samples was the same whatever were their composition (Table S1, Figure S2). Noteworthy, after the final washing step (phase 3, Figure 2B) no floating vesicles were observed in the hosting solution (HS) (Figure S2). Four fatty acids (7–10, Figure 2B) were selected as feeding molecules. Our choice was based on the results reported by Luisi and Szostak (Stano and Luisi, 2010; Szostak, 2017). In addition, two synthetic lipophilic and fluorescent probes ω-(dansyl)laurinyl derivative 11 and δ-(isopropanylidene)tryptophanyl derivative 12 (Schemes S1 and S2 and Figures S16–S21) were used. The growth and partial detachment of lipid boundaries from g-MSGVs of all membrane types A–E were followed by microscopy. Owing to precipitation issues with the amphiphiles 10–12 (Table 1, Entries 6–9), only oleic, myristic, and myristoleic acids (7–9 Figure 2B) were kept for consecutive feeding in G&D experiments. We optimized the feeding of g-MSGVs by using ethanolic solutions of 7–9 at 10, 25, and 50 mM concentrations. Myristic acid (8) also precipitated but only when a 50 mM feeding solution was used. A 10 mM initial feeding concentration was chosen as it produced in preliminary test feedings an approximative 1:1 partitioning of the phospholipids between the first-generation g-MSGVs and the lipids in the hosting solution (Figure S3). This 1:1 ratio was not reproduced later in more systematic tests (cf. Figures 4 and 5 and Table 1). During each feeding period ethanol evaporated only gradually, thus, it kept the fatty acids as free compounds, which avoided for some time the formation of micelles or other aggregates (Walde et al., 2010). Control experiments when g-MSGVs where fed with pure ethanol showed no detachment of phospholipids in solution by microscopy (Fiore et al., 2018). The feeding ratio was tuned to 33 μL/h and the total feeding time 3 h. The selected parameters minimized the early de novo formation of aggregates, thus permitted to reach a total concentration of 1 mM of 7–9 in the HS.

Table 1.

Microscopic and Chemical Analysis of Lipid Boundaries Content after Two or Three Consecutive Feeding Experiments on D₁(GV + LA)s, D₂(GV + LA)s, D₃(GV + LA)s, g-M₂SGVs, and g-M₃SGVs by Using g-M₀SGVs Type A–C

Entry	g-M0SGVsa	Microscopic Observationsb	Vesicles Product	PL Concentrationc,d (mM)	PLfound/PLinitial	1st Feeding	2nd Feeding	3rd Feeding	Final Composition
						7found/7fedd,e	8found/8fedd,e	9found/9fedd,e
1	A (1)	Budding, Growth, Pearling, and Division	D1(GV + LA)s	0 18 ± 0.01	0.33 ± 0.01	0.90 ± 0.01	–	–	1 + 7
			D2(GV + LA)s	0.17 ± 0.01	0.31 ± 0.01	0.06 ± 0.02	0.68 ± 0.01	–	1 + 7+8
			g-M2SGVs	0.20 ± 0.01	0.36 ± 0.01	0.04 ± 0.01	0.32 ± 0.01	–	1 + 7+8
2	A (1)	Budding, Growth, Pearling, and Division	D1(GV + LA)s	0.14 ± 0.01	0.25 ± 0.01	0.91 ± 0.03	–	–	1 + 7
			D2(GV + LA)s	0.15 ± 0.01	0.27 ± 0.02	0.03 ± 0.01	0.54 ± 0.03	–	1 + 7+ 8
			D3(GV + LA)s	0.13 ± 0.01	0.24 ± 0.01	0.03 ± 0.01	0.33 ± 0.01	0.76 ± 0.05	1+7-9
			g-M3SGVs	0.13 ± 0.01	0.24 ± 0.01	0.03 ± 0.01	0.13 ± 0.02	0.24 ± 0.05	1+7-9
3	B (1 + 2)	Budding, Growth, Pearling, and Division	D1(GV + LA)s	0.20 ± 0.01	0.37 ± 0.01	0.89 ± 0.03	–	-	1 + 2+7
			D2(GV + LA)s	0.16 ± 0.02	0.28 ± 0.03	0.08 ± 0.02	0.65 ± 0.03	-	1 + 2+7 + 8
			g-M2SGVs	0.19 ± 0.01	0.35 ± 0.02	0.03 ± 0.01	0.35 ± 0.03	-	1 + 2+7 + 8
4	C (1–4)f	Budding, Growth, Pearling, and Division	D1(GV + LA)s	0.11 ± 0.01	0.20 ± 0.01	0.99 ± 0.03	–	-	1-4+7
			D2(GV + LA)s	0.34 ± 0.02	0.62 ± 0.03	0.008 ± 0.0001	0.57 ± 0.03	-	1-4+7 + 8
			g-M2SGVs	0.10 ± 0.01	0.18 ± 0.02	0.002 ± 0.001	0.43 ± 0.03	-	1-4+7+8

Inverted Feeding Order			Vesicles Product	PL Concentrationc,d	PLfound/PLinitial	1st Feeding	2nd Feeding		Final Composition
Entry	g-M0SGVsa	Microscopic Observationsb				8found/8fedd,e	7found/7fedd,e
5	A+ (1)	Budding, Growth, Pearling, and Division	D1(GV + LA)s	0.20 ± 0.02	0.37 ± 0.04	0.43 ± 0.01	–	-	1 + 8
			D2(GV + LA)s	0.18 ± 0.02	0.33 ± 0.04	0.45 ± 0.02	0.95 ± 0.01	-	1 + 8+7
			g-M2SGVs	0.17 ± 0.02	0.30 ± 0.03	0.12 ± 0.01	0.05 ± 0.01	-	1 + 8+7

Other Feedings		Fed with	Vesicles Product					Final Composition
Entry	g-M0SGVsa
6	A (1)	10	None/aggregates					–
7	A (1)	11	None/aggregates					–
8	A (1)	12	None/aggregates					–

^a1=POPC, 2=POPA; 3=DOPC, 4=DOPA, 7=oleic acid, 8=myristic acid, 9=myristoleic acid.

^bDescriptions and images of the phenomena are available in the Supplementary Information.

^cConcentrations of phospholipids 1, 2, 3, and 4 measured by Stewart assay and normalized to 0.55 mM.

^dMean values and standard errors were determined from experiments performed in triplicates (n=3).

^eConcentration of fatty acids 7, 8, and 9 measured by HPLC and normalized to 1 mM.

^fAnalysis of the lipid boundaries containing soybean and egg-yolk extracts (Table S1, entries 5 and 6, D and E types, respectively) was not performed. C type MSGVs can be considered as a simplified model of D and E type coatings.

Figure 3.

Experimental Setup and Representative Microscopic Results

(A) Workflow of monitoring of three consecutive feeding processes on phospholipid-coated glass beads, that is, glass microsphere-supported giant vesicles (g-M₀SGVs, A-C type coatings) fed with fatty acids.

(B) Micrographs of periodic microscopic confocal observations during feeding of POPC g-M₀SGVs bearing 0.2% red fluorescent DOPE-Rh (see MPB in Transparent Methods). The confocal microscope images do not correspond to a time-lapse monitoring of a single g-MSGV. They show different objects that are thought to represent the most prominent stages of a G&D model process as shown schematically below: (a) A⁺ type g-M₀SGVs (Table S1, entry 3) before feeding; (b) budding (60 min); (c) membrane growth (60 min); (d) detachment with apparition of a daughter vesicle (D₁GV, 60 min). Blue arrows indicate g-M₁SGVs, white arrows indicate the G&D phenomena, and orange arrow indicates recently formed daughter vesicle. Pictures were accompanied by a simplified draw, below the confocal images, that resume the most important phases observed during feeding experiments: The scale bar is 5 μm for all images.

Figure 4.

Graphical Representation of the Data Obtained from the Analysis of g-M_nSGVs and D_n(GVs+LA)s after Consecutive Growth and Division of Supported Giant Vesicles

(A–E) Concentrations (mM) of phospholipids and fatty acids in g-M₀SGVs, D_1-3(GVs + LA)s and g-M_2-3SGVs after two or three feeding periods. Composition of g-M₀SGVs: (A), (D), (E) POPC (1, A type coating); (B) POPC/POPA 4:1 (1 + 2, B type coating); (C) POPC/POPA/DOPC/DOPA 4:1:4:1 (1 + 2+3 + 4, C type coating). Feeding order: (A–C) first oleic acid (7) then myristic acid (8); (D) first 8 then 7; (E) first 7 then 8 then myristoleic acid (9).

(F–I) Concentrations (mM) of fatty acids incorporated into g-MSGVs (F and H); phospholipids released from g-MSGVs into HS (G and I), after respective feeding periods. Amounts and standard deviations obtained from Table 1: A and B type g-MSGVs grouped (sample size = 12) (F and G); C type g-MSGVs only (sample size = 3) (H and I). Concentration of incorporated fatty acids (F and H) corresponds to (i) amount found in g-M_nSGVs if corresponding fatty acid was last supplied during feeding n, (ii) sum of amounts found in D_n+1(GV + LA)s and g-M_n+1SGVs when corresponding fatty acid was second to last supplied during feeding n, (iii) sum of amounts found in D_n+1(GV + LA)s, D_n+2(GV + LA)s and g-M_n+2SGVs when corresponding fatty acid was third to last supplied during feeding n. Concentration of released phospholipids (G and I) corresponds to amount found in D_n(GV+LA)s after supply with corresponding fatty acid during feeding n.

Figure 5.

Exemplary Pseudo-Molecular 2D Representation of Triple-Feeding Process Resulting in Measured and Deduced Phospholipid and Fatty Acid Concentrations and Partition Values

Percentages in square brackets refer to concentration of phospholipids and fatty acids of g-M_0-3SGVs and D_1-3(GV + LA)s partitioned over A type mothers and daughters after each feeding step: (% = concentration of one amphiphile type/total concentration of this amphiphile type). Total concentration of PL (0.55 mM, 215 blue objects) corresponds to amount on g-M₀SGVs. Total concentration of each fatty acid (1 mM, 390 yellow, green, or red objects) corresponds to supply during each respective feeding period. Object sizes not scaled, number of tethered molecules not proportional, multilamellarity not depicted.

To stay unambiguous and clear at the same time, we renamed the g-MSGVs as g-M_nSGVs, where g stands for glass, M stands for “microsphere” or “mother,” and the subscript n is an increasing positive integer for M₁, M₂, M₃ to distinguish each mother population generated from each feeding process (Figure 3A). The initial population of g-MSGVs was named g-M₀SGVs. With DGVs, where D stands for “daughter,” we designated the floating vesicles generated from a first, second, or third G&D process, so, accordingly, an increasing subscript integer for D₁, D_2, D₃ appointed each generation of the population of daughters. With time passing after each feeding step, we observed to a varying degree the appearance of floating aggregates of lipids that accompanied the daughter vesicles. Therefore, we used the abbreviation D(GV + LA) to designate all floating lipid material in the HS, where LA stands for other lipid aggregates. After each G&D process, the supported mother vesicles g-M_1-3SGVs were isolated by sequential cycles of centrifugation and washings from all floating lipids D_1-3(GV + LA)s (Figure 3A). In this study, we took advantage of these model giant vesicles to independently characterize the lipid composition of the g-M_nSGVs and the D_n(GV+LA)s after every successive feeding experiment (results in Figure 4 and Table 1). The model diagram below the images in Figure 3B is meant as a guide that puts images taken at about the same time (Figure 3B, images a-d) into an assumed process-dependent context. It does not imply an actual time lapse following of the same g-MSGV in a, b, c, and d. The phospholipids and fatty acids were extracted with chloroform (Bligh and Dyer, 1959). RP-HPLC (Heron et al., 2007) served to quantify the composition of fatty acids after one, two, or three consecutive feeding processes (Table 1). The Stewart assay (Stewart, 1980), even if less precise than an HPLC analysis, was used to determine, independently of the fatty acid quantification, the initial (Table S2) and final (Table 1) concentrations of phospholipids (PLs) in g-M₀SGVs, g-M_2-3SGVs, and D_1-3(GV + LA)s, respectively. All experiments were performed in triplicate, and these data were used to calculate the mean concentrations and the associated standard deviations (Table 1).

Results

Prior to any chemical analysis, we monitored that G&D was indeed occurring. The main types of phenomena described during the growth of a lipid boundary upon feeding were budding, growth (in size), pearling, and division/detachment (Figure S1) (Stano and Luisi, 2010; Szostak, 2017). They all happened upon addition of oleic, myristic, and myristoleic acids (7–9 Figures 2, 3B, and S8–S13). Of note, there were no remarkable differences between the different coating types of g-MSGVs (A–E, Table S1) during our observations by confocal (Figure 3B) and epifluorescence microscopy (Figures S8–S13). However, it appeared that oleic acid (7) induced more easily the G&D from g-MSGVs. The critical micelle concentration (CMC) of pure 7 is 1 mM, which is lower than the CMC of shorter fatty acids––the CMC of 8 is 15 mM––and in the range of the final concentrations at the end of the feeding (0–1 mM) (Budin et al., 2012). The critical aggregation concentrations (CACs) of pure 7 and 8 are 0.1 and 2 mM, respectively. In our opinion, owing to the relatively high expected critical vesicle concentration (CVC) of the fatty acids, and their relatively narrow pH and salt tolerance, only fatty acid micelles and aggregates were likely to form during the feeding process before they partitioned between the HS and the g-MSGV's membrane (Cape et al., 2011). Our choice for the feeding concentration and amount of added fatty acid generated final concentrations of at most 1 mM. This meant that the fraction of 7–9 that was not bound to the phospholipid membranes (the leftover fatty acids) could only be present in concentrations much below their respective CMC. However, the effective CAC, CMC, and CVC of fatty acid mixtures may become even lower in the presence of phospholipids. Therefore, some aggregates, micelles, and even vesicles could be present after the feeding period. Feeding supported giant vesicles with higher concentrations of fatty acid solutions (25 and 50 mM, Figure S3) expectedly produced more rapidly the same previously described phenomena (Figures S10–S13), although evagination was barely seen here (Figure S1), in contrast to what we described in our previous communication (Fiore et al., 2018).

We performed a set of experiments directly above the (inverted) lens of a confocal laser scanning microscope. The prepared POPC g-M₀SGVs type A⁺ membranes (Table S1) were initially expectedly spherical (Figure 3B, image a). The formation of buds was observed after feeding g-M₀SGVs with oleic acid (25 mM ethanolic solution 7) (Figure 3B, image b). Such buds could increase in size during the growth step (Figure 3B, image c). Budding and membrane growth were observed at the same moment. We noticed that, in some cases, the fast local growth also induced the formation of lipid tubes (pearling) from g-M₁SGVs, which were similar to those observed in the pearling phenomenon prior to the division of oleic acid vesicles (Figures S1 and S13) (Szostak, 2017; Zhu et al., 2012; Zhu and Szostak, 2009). Finally, we observed the formation of new vesicles (D₁GVs, Figure 3B, image d) due to feeding with fatty acids. It was possible to ascertain that these new D₁GVs were freshly formed from the g-M₀SGVs, since they contained the fluorescent phospholipid probe 6 that was only added to the mother vesicles and shown not to leak even after prolonged time periods (Figure 3B, image d). Similar observations were also made when G&D processes were performed by feeding g-M₀SGVs bearing phospholipid membranes of more complex B and C type compositions (Table S1, Figures S8–12).

After the confirmation that G&D was indeed occurring on our mother vesicles, we set out to monitor the lipid exchange between the g-M_nSGVs and the medium during consecutive G&D. The phospholipid concentration of the g-M₀SGVs was determined by the Stewart assay using B type g-M₀SGVs (Table S1). This type of supported giant vesicles of binary compositional complexity was used as a reference for all the others, since the phospholipids (POPC and DOPC, POPA and DOPA) were fairly similar to one another (Figure 2B). An average concentration of 0.55 mM was measured for samples containing 9 × 10⁵ g-M₀SGVs. This amount corresponded to 3.7 × 10¹¹ phospholipid molecules coated on one single g-M₀SGV and compared with a maximal coating of 2 × 10⁸ biotinylated tether molecules; hence, an at least 2,000-fold molar excess untethered over tethered phospholipids (Table S2). Surprisingly, the amount of coated phospholipid molecules was much higher than the one expected for unilamellar vesicles (Walde et al., 2010) and can be explained by the fact that the g-MSGVs formed could be at least partly multilamellar (Albertsen et al., 2014). Besides, the multilamellarity more readily explains the pearling phenomenon (Szostak, 2017; Zhu et al., 2012; Zhu and Szostak, 2009) that was observed during the G&D from g-MSGVs (Figure S12). The measured lipid concentrations were in general lower than expected. This was probably due to losses occurring during the extraction step. In order to compare the different conditions, the measured phospholipid concentrations were normalized to a total amount of 0.55 mM. This concentration corresponds to the initial amount present on the g-M₀SGVs in the initial sample volume before feeding started, whereas the measured fatty acid concentrations were normalized to a total concentration of 1 mM corresponding to the amount supplied during one feeding period (Figure 4, Table 1).

The feeding of three different types of g-M₀SGV membrane coatings (A, B, and C, Tables 1 and S1) was tested by performing two consecutive feedings, first with oleic acid (7) then myristic acid (8) (Figures 4A–4C, Table 1). We assumed that changing the nature of the membrane would impact the stability properties and the exchanges occurring between the g-MSGVs and the HS. Our results instead showed that the composition of the phospholipidic membranes did not seem to have any significant impact on the feeding with fatty acids (Figures 4A–4C and 5). Of note, when the first feeding was performed using C type g-MSGVs (Figure 4C), a higher amount of oleic acid (7) was found in the HS. Besides, the feeding with this fatty acid induced less detachment of phospholipids for C type g-MSGVs compared with A and B type g-MSGVs (Figures 4A and 4B). The second feeding with 8 released markedly more phospholipids from C type membranes, whereas the fatty acid concentrations were much the same for all types (Table 1, entries 1 and 3).

In the end, the three different populations of vesicles and floating lipid aggregates obtained, D₁(GV + LA)s, D₂(GV + LA)s, and g-M₂SGVs, bore as membrane constituents a mixture of all the amphiphiles but in different molar ratios (Figures 4A–4C and 5, and Table 1). However, the amount of 7 found in the most complex C type membranes (Table 1, entry 4) was the smallest when compared with both A and B type membranes (Table 1, entries 1 and 3). We assume that this observation is related to the fact that 7 hardly exchanged with the phospholipid membrane of the initial M₀SGVs of any membrane type A–C. In a situation where the kind of polar headgroups was constant in all types of phospholipid membranes, the difference in their acyl chains (1–4) apparently did not strongly influence the fatty acid uptake properties of the membranes.

As mentioned above, when A type g-MSGVs were first fed with oleic acid (7) then with myristic acid (8). 7 was much more present in the host solution containing D₂(GV + LA)s than 8 (Figure 4A). This result could be due to the fact that, during the second feeding, more room was available on the A type g-M₁SGVs because their phospholipids progressively detached and were released to the host solution. To test this, the order of feeding was inverted. Curiously, we observed that the concentration of myristic acid (8) in the HS containing D₂(GV + LA)s was still lower than that of oleic acid (7) (Figure 4D). Thus, we could confirm that the saturated fatty acid 8 tended to stay more in the lipid membrane of the mother vesicles after a G&D when compared with the unsaturated fatty acid 7 (Figures 4A–4D). With these experiments it was also possible to observe that the compositional complexity of a membrane could increase owing to a continued feeding process because, rather than completely replacing one another, all different types of amphiphiles were present. The PC/PA headgroups (1–4), oleic and myristic acids (7, 8), were found in the membranes of the second-generation mother vesicles g-M₂SGVs at the end of the experiments, that is, after being separated from their D₂(GV + LA)s (Table 1).

In order to observe whether the compositional complexity could reach yet another level and if unsaturated fatty acids were indeed more released in the host solution than saturated fatty acids, a third feeding was performed with myristoleic acid (9, Figure 4E). To this end, g-M₂SGVs were generated from g-M₀SGVs upon two consequent feedings with 7 and, respectively, 8 and those were supplied with 9. After this third feeding all amphiphiles (PL = 1, FA = 7–9) were detected as constituents of, both, mothers g-M₃SGVs and daughters D₃(GV + LA)s in their respective membranes (Table 1, entry 2). Interestingly, we observed that, despite the same number of carbons present in the fatty acid chains, myristoleic acid (9) was more present in the host solution than myristic acid (8) after their respective feeding periods. Besides, after a G&D the incorporation of 9 was similar to that of 7 (oleic acid) and lower than that of 8 (Table 1). Thus, the unsaturated fatty acids 7 and 9 partitioned consistently less into the membranes of g-M₀SGVs and g-M₁SGVs after a G&D compared with the only tested saturated fatty acid 8.

Discussion

The use of giant vesicles filled with, and supported by, a monodisperse microspherical glass bead (g-MSGVs) allowed us to better understand the lipid exchange that occurred upon feeding the supported phospholipid membranes with fatty acids. For the quantification of lipid exchange we chose feeding conditions that in preliminary experiments (Figure S3) led to the detachment of at most one-half of the phospholipids from the zeroth-generation mother vesicles (g-M₀SGVs) to generate membranes of daughter vesicles and some structurally less-defined aggregates. We did not separately quantify the lipids that were present in giant daughter vesicles (DGVs) from those that had formed other lipid aggregates (LA), but we could identify the microscopic difference between daughter vesicles and other aggregates present in the host solution only for objects larger than about 0.5 μm. The low final feeding concentrations of the fatty acids favored their coexistence in the available phospholipid membranes and aggregates, which is consistent with the finding by others that fatty acids remain more firmly bound in phospholipid containing membranes than they would in pure fatty acid bilayer membranes (Jin et al., 2018). This paves the way for an assumed spontaneous rigidification and enrichment of prebiotic membranes bearing higher and higher amounts of phospholipids. Pure POPC membranes and those containing 20% POPA (coating types A and B, respectively) showed similar properties in incorporating fatty acids and detaching phospholipids (data grouped in Figures 4F and 4G). Those that contained 50% DOPC/DOPA 4:1 (C type coating) were particularly prone to be detached by myristic acid (8) but much less so by oleic acid (7) (Figures 4H and 4I). In Figure 5 the distribution of phospholipids and fatty acids in the triple-feeding experiment on A type membranes is depicted in a generation-dependent fashion, where the power of phospholipid detachment from mothers to daughters was related to the feeding material (for a more schematic graphical depiction of all distributions listed in and derived from Table 1 see Figures S14 and S15).

We focused our attention on the mother vesicles because of their obvious superior properties as evolvable micro-compartments in later generations. The growth in size, detachment, and closure of the grown membranes into giant daughter vesicles reminds very strongly of what happens when “natural” unsupported giant vesicles grow in size then divide (G&D) into smaller daughter vesicles, thus, of vesicles that have grown in number after each feeding round and G&D process. Despite the fact that the type of membranes used for the supported vesicles did not really influence the fatty acid incorporation, we have shown that the type of fatty acid used as feeding material mattered indeed. Irrespective of the feeding order, a saturated, short fatty acid (myristic acid) was integrated in the lipid membrane of the g-MSGVs better than the unsaturated oleic and myristoleic acids.

Since both kinds, saturated and unsaturated fatty acids, were able to induce G&D, it is probable that their release from the g-M₀SGVs was different. Lately, a study using free-flow electrophoresis took advantage of the charge on deprotonated oleic acid (7), oleate, to show that feeding small unilamellar POPC vesicles (diameter 70 nm) with this acid led to the formation of two populations of vesicles, oleate-poor vesicles and oleate-rich vesicles (Pereira de Souza et al., 2015). Considering the fact that the same phenomena typical for G&D (budding, pearling, evagination, etc.) have been observed on our A type first-generation g-M₀SGVs (Figures 3B and S8–S13) and regarding our results from first, second, and third generation g-M_1–3SGVs, it is highly likely that a similar lipid exchange was occurring on our supported membranes. We have observed that more 8 was incorporated in the g-MSGV membranes than 7, whereas the amount of phospholipid released in the HS by the feeding was very similar (Figures 4F, 4G, and 5). Thus, g-M₀SGVs made of POPC and fed with 7 would produce oleate-poor vesicles, viz. the g-M₁SGVs, and oleate-rich vesicles and aggregates D₁(GV + LA). Furthermore, this phenomenon would also occur when A type g-M₀SGVs were fed with myristoleic acid (9) (Figures 4G and 5 and Table 1). However, using a saturated fatty acid such as 8 would generate membranes that were more symmetrically blended over mothers and daughters (Figures 4E and 5). Interestingly, “hybrid protocells” made of these two types of amphiphiles (phospholipids and fatty acids) showed good properties of stability and permeability (Jin et al., 2018) but they also grew faster than pure fatty acid vesicles (Budin and Szostak, 2011). Our results suggest that the lipid exchange occurring on free vesicles during a G&D process is similar to the one occurring on g-MSGVs and that saturated fatty acids are the privileged amphiphilic feeding material to generate from phospholipid vesicles these more complex, more blended, thus potentially more competitive protocells (Lancet et al, 2018, 2019).

Conclusion and Perspectives

This optimized methodology for the feeding of glass microsphere-supported giant vesicles is a convenient general tool for the observation and the understanding of the growth and division process over several feeding rounds (generations) in a more controlled analytical way than what was possible before. The method has the potential of providing experimental data for simulation studies on compositional replication in a hypothetical “lipid world” (Kahana and Lancet, 2019; Lancet et al, 2018, 2019; Segre et al., 2000). It could also be applied to other amphiphiles and possibly tested on more artificial coatings than natural phospholipids. In the context of self-evolving compartments, the possibility of generating supported vesicles being composed of mixtures of amphiphiles opens strong perspectives for the study of vesicles that are obtained under so-called prebiotic conditions (Alberstsen et al., Albertsen et al., 2014; Altamura et al., 2020; Fayolle et al., 2017; Fiore et al., 2017; Fiore and Buchet, 2020; Monnard and Deamer, 2003). In this perspective, we can now go on and test how the newly acquired fatty acids could be chemically fixed so as to enrich the compositional complexity of future generations in a more persistent way (Szostak et al., 2001).

Limitations of the Study

Glass microsphere supported giant vesicles (g-MSGVs, mothers) were used as model protocells to monitor and study growth and division processes of phospholipidic membrane bilayers upon consecutive feedings with fatty acids. The use of g-MSGVs has shown a few limitations and more experiments are required to fully understand their potentiality in origin of life studies. Saturated fatty acids longer than fourteen carbon atoms cannot be used as feeding material, however short saturated chains are more plausibly prebiotic. Time-lapse monitoring of g-MSGVs or DGVs (daughters) was not possible probably due to the technical limitation set by our equipment.

Resource Availability

Lead Contact

Further information and requests for resources should be directed to and will be fulfilled by the Lead Contact, Michele Fiore (michele.fiore@univ-lyon1.fr).

Materials Availability

This study did generate two new unique compounds ω-(dansyl)laurinyl derivative 11 and δ-(isopropanylidene)tryptophanyl derivative 12 (Figures S16–S22). Their chemical synthesis and characterization are described in Supplementary Information, and there is no restriction for availability.

Data and Code Availability

In this study, we have not used any unpublished custom code, software, or algorithm that is central to supporting the main claims of the paper.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.

Published: November 20, 2020