Toward the Growth and Division of Prebiotically Plausible Protocells

To determine if the observed growth and division of model protocells depended on the presence of LUVs, we next mixed fluorescently labeled, crude MLVs with MLVs extruded to 8 μm in the absence of LUVs. Here, fluorescently labeled, crude MLVs were mixed with MLVs extruded to 8 μm. Upon mixing, the C18:1 MLVs grew and divided when fed with 50 or 100 equiv of C14:1 crude MLVs (Figure ​Figure33). Subsequently, the resulting C14:1-rich mixture was subjected to another round of feeding by the addition of 4 equiv of C18:1 crude MLVs. Again, the vesicles grew and divided. Finally, C18:1 MLVs were added to the output from the second round of growth and division. Fluorescence microscopy indicated that the MLVs grew and divided (Figure ​Figure33). Therefore, iterative rounds of growth and division were possible with C18:1-rich and C14:1-rich vesicles.

Open in a separate window

Figure 3

Growth and division of protocells with mixed compositions of bilayers. Scale bars indicate 100 μm.

Prebiotic chemistry likely produced mixtures of lipids of less than 14 carbons.²⁹ To test if shorter, more prebiotically plausible lipids could give rise to the same behavior, laurate (C12:0) and caprylate (C8:0) vesicles were investigated. As expected, pure C12:0 MLVs (ca. 40 or 20 mM) grew when fed with 9 or 19 equiv of pure C8:0 vesicles (ca. 360 or 380 mM) (Figure ​Figure44a). To determine whether iterative growth and division was possible, the vesicles produced from mixing C12:0 MLVs with C8:0 LUVs (C8:0-rich vesicles) were further subjected to another round of growth and division by the addition of C12:0-rich crude vesicles. More specifically, the 19:1 caprylate/laurate MLVs (short-chain-rich MLVs) produced from the first round of growth and division were fed with 1 equiv of 1:19 caprylate/laurate LUVs (long-chain-rich) or 4 equiv of laurate vesicles. A nearly 2-fold increase in the number of vesicles was detected after this second round of growth and division, as judged by fluorescence microscopy and vesicle counting (Figures ​Figures44b,c and S6). The data were consistent with that of the C18:1 and C14:1 mixtures (Figures ​Figures11–3), as the long-chain MLVs exhibited filamentous growth and the short-chain MLVs showed rapid division. The increase in the number of vesicles was of similar magnitude to the ~3-fold increase observed when C14:1 MLVs were fed with 1 equiv of C18:1 LUVs (Figure ​Figure22c). Overall, the observed behavior was dependent on a disparity in lipid composition between different vesicles and was not due to the presence of specific lipids.

Open in a separate window

Figure 4

Growth and division with prebiotically plausible fatty acids. (a) Growth of C12:0 MLVs fed with C8:0 vesicles. Upper images were with 19 equiv of C12:0 vesicles, while lower images were with 9 equiv of C12:0 vesicles. (b) Growth and division of C8:0-rich (19:1 caprylate/laurate) MLVs fed with 1 equiv of C12:0-rich (1:19 caprylate/laurate) vesicles. (c) Quantification of vesicle abundance. Scale bars indicate 100 μm. Images were taken with identical exposure times. Imaging flares are due to a long exposure time.

Retention of Entrapped Material

Growing and dividing protocells would likely have aided the emergence of functional nucleic acids if the protocells were capable of retaining material. To probe the likelihood of such a scenario, crude mixtures of MLVs and LUVs were prepared in the presence of either calcein or fluorescently labeled DNA. After purification by size-exclusion chromatography, fresh LUVs were added to the purified mixture of MLVs and LUVs, and retention of encapsulant was quantified by an additional round of size-exclusion chromatography (Figure S7a,b). The starting population of C18:1 MLVs and LUVs retained ~90% of the encapsulated calcein after feeding with 50 equiv of C14:1 LUVs (Figure ​Figure55a). The same reaction with calcein loaded C18:1 LUVs in place of a mixture of C18:1 MLVs and LUVs gave similar results, ~80% (Figure ​Figure55b). Mixtures of C14:1 MLVs and LUVs were less capable of retaining encapsulant when fed with C18:1 LUVs. The addition of 2.5 equiv of C18:1 LUVs led to an ~50% loss of encapsulated DNA from a population of C14:1 MLVs and LUVs (Figure ​Figure55a). Retention was the same if the C14:1 MLVs were substituted with C14:1 LUVs. Receiver C14:1 vesicles lost 50% of the encapsulant in all cases, e.g., ~60–70% was retained without feeding and 10–20% was retained after feeding with C18:1 LUVs (Figures ​Figures55b and S7c,d). The reasons for the large loss of entrapped material were unclear and may have reflected rupture of a subset of the population, formation of large, transient pores,³⁹ or other remodeling processes during division.

Open in a separate window

Figure 5

Retention of encapsulant during growth and division. (a) Size exclusion chromatography (SEC) of mixtures of MLVs and LUVs after the addition of LUVs. The encapsulant was either calcein or fluorescently labeled DNA. (b) SEC of fluorophore-containing LUVs after addition of unlabeled LUV. SEC was run 10 min after the addition of LUVs.

Fatty Acid Dynamics

Past work showed that disparities in lipid composition between fatty-acid-based membranes could drive the growth of vesicles.²⁶ In these systems, kinetically trapped phospholipids could not equilibrate between bilayers.³⁵ For the systems described herein, all of the lipids could dynamically equilibrate between all of the aggregates present.⁴⁰ There were no kinetically trapped species, and a flux of lipids between membranes was always present. Nevertheless, kinetic differences must have existed, as the energetic cost of desorption and interleaflet flip-flop are impacted by the surface area of the hydrophobic chain. To confirm that differences in desorption rates were present, LUVs made from C14:1, C16:1, and C18:1 were mixed with 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC) vesicles containing a pH-sensitive fluorophore (HPTS). In this assay, the rate of decrease in pH reflected the rate-limiting step, which was the desorption of the lipids from the LUVs (Figure S8a).⁴¹ As expected, the desorption rate was greatest for C14:1 followed by C16:1 and C18:1 (Figure ​Figure66a). The data were corroborated with POPC vesicles that contained FRET-labeled lipids in place of encapsulated HPTS (Figures ​Figures66b and S8b). Rates of lipid flip-flop⁴² followed the same trend (Figures ​Figures66c and S8c). Finally, the ability of longer fatty acids to provide a more ordered intramembrane environment was confirmed by measuring the anisotropy of 1,6-diphenyl-1,3,5-hexatriene (DPH), a fluorophore embedded within the membrane (Figures ​Figures66d and S8d). Greater order within membranes is associated with decreased monomer off-rates of lipids.^43,26

Open in a separate window

Figure 6

Lipid dynamics. Desorption rates determined by changes in pH (a) or FRET (b) of receiver POPC vesicles. (c) Rate of fatty acid flip-flop. (d) Fluorescence anisotropy of DPH³⁴ within different membranes as an indicator of fluidity. All vesicles were LUVs. Data are mean with ±SD; n ≥ 4.

Mechanism of Protocell Growth

Although differences in lipid dynamics were present, these differences alone could not explain the observed behavior. Decreased rates of desorption of C18:1 in comparison to C14:1 may have favored the growth of C18:1 vesicles at the expense of C14:1 vesicles. However, C14:1 vesicles were also able to grow at the expense of C18:1 vesicles. Instead, growth seemed to be primarily dependent on the greater displacement from equilibrium of the receiver MLVs upon mixing in comparison to the donor LUVs. In other words, donor LUVs were typically provided in excess, meaning that the concentration at equilibrium more closely resembled the starting concentration of lipids of the donor LUVs rather than the receiver MLVs. Therefore, the net direction of growth was not reliant on a specific composition of lipids as long as a disequilibrium existed. However, the different lipids must be sufficiently dissimilar in terms of their dynamics to mediate the behavior observed here.

Greater desorption kinetics suggest more facile equilibration with micellar aggregates, which exist in solutions containing fatty acid vesicles. As micelles are less thermodynamically favorable than vesicles above a critical concentration of lipid under specific solution conditions, the addition of micelles to vesicles results in growth.^{20,22,26,37,38} However, the system described herein did not show a dependence on a micellar phase, and LUVs and MLVs are equally stable. Large fatty acid vesicles of the same lipid composition but different diameters coexist without fusion and do not give rise to growth.³⁵ For example, the concentration of lipid was well above the critical vesicle concentration (CVC) (Table S2), indicating that the fraction of micelles in solution was minor. Calculations⁴⁴ indicated that the concentration of micelles increased upon the addition of C14:1 LUVs to C18:1 MLVs (CVC increased by more than 25-fold) and decreased upon the addition of C18:1 LUVs to C14:1 MLVs (decreased more than 25-fold).

Micelle-dependent growth was also not supported by kinetic measurements. Assuming that growth of acceptor membranes was first order with respect to feeder vesicles, we extracted the effective rate constants of growth from the FRET data in Figures ​Figures11 and ​and22 (Figure S9a,b). The growth rate of C14:1 MLVs upon feeding with C18:1 LUVs (k = 0.014/s) was one order of magnitude greater than for C18:1 MLVs upon the addition of C14:1 LUVs (k = 0.0014/s). Both growth rates were much slower than the rates observed by micelle addition.³⁷ Taken together, these observations further bolstered the hypothesis that membrane growth in our system occurred via pathways other than simple micelle-driven monomer adsorption.

We then sought to understand the morphological changes following growth. Others have observed that a rapid increase in surface area to volume ratio can lead to nonequilibrium structures consisting of lobes and protrusions.^45,46 The resulting lobes and protrusions are intrinsically unstable, since these structures deviate from the thermodynamically favorable spheroid form, ultimately leading to division.^47,48 When the buffer (HEPES), which crosses the membrane slowly, was replaced with the much more permeable ammonium acetate,²⁰ we still observed shape changes, albeit to a lesser extent (Figure S10). These results suggest that the rate of membrane growth in our system could still outpace volume growth when the lipid tails differed by four carbon units, highlighting how small, subtle differences in molecular structure can lead to large changes at the population level.

Preparation and Purification of Protocells

All vesicles were prepared by either the hydration of a thin film or the dispersion of a neat fatty acid oil. Organic solvents used to dissolve the lipids were either chloroform for phospholipids or 9:1 methanol/chloroform for fluorescently labeled phospholipids. Briefly, desired amounts of lipids from stock solutions were pipetted into glass vials and evaporated under a gentle flow of N₂. Then, mixtures were dried by house vacuum overnight (<16 h) to remove residual solvent. The next day, the vesicles were hydrated with buffers or buffers containing fluorophore, vortexed briefly multiple times (4–5 s), and then left to equilibrate with tumbling for <24 h. For C14:1, C16:1, and C18:1 vesicles, the buffer was 0.2 M Na⁺-HEPES, pH 8.0, with the pH adjusted with 5 M NaOH. After tumbling, vesicles were extruded with 100 nm or 8 μm track-etched polycarbonate membranes using an Avanti miniextruder (at least 11 passes). The extruded vesicles were left undisturbed in the dark overnight (16–24 h). For C8:0 and C12:0 vesicles, the buffer was 1 M Na⁺-HEPES, pH 7.0, with the pH adjusted with 5 M NaOH. The vesicles of C8:0 and C12:0 were constantly kept above 55 °C with a conductive metal heat block, and these experiments were completed within 2 h after preparation of the vesicles. If needed, the vesicles were purified by size exclusion chromatography, with lipid concentrations kept above the CVC in the running buffer. Fractions were collected with a Gilson FC204B multichannel fraction collector, analyzed with a Tecan M200 (BioTek) or a Varioskan (ThermoFisher) plate reader. All purified vesicles were used within 24–72 h of the final extrusion.

Growth and Division of Protocells

Multilamellar vesicles were prepared as described above and were tagged with 0.15 mol % N-(lissamine rhodamine B sulfonyl)-1,2-dihexadecanoyl-sn-glycero-3-phosphoethanolamine (LR-DHPE). Following tumbling and extrusion, the small vesicles with <5 μm size were removed by multiple cycles of brief centrifugation with spin column filters (Ultrafree–MC–Durapore 5 μm with PVDF membrane, Millipore) while keeping the overall lipid concentration above the CVC with vesicles of the same composition. Feeder vesicles were also prepared as above and used after equilibration for 24 h except for the C8:0 and C12:0 lipids, which were used within 2 h of preparation. For iterative growth, the vesicles were mixed and equilibrated for 10 min and then imaged.

FRET Assay for Vesicle Surface Area Change

Pure and mixed membrane growth kinetics were measured taking previous reports as a guide.^{19,20,22,25−28} The surface area changes were monitored at λ_ex = 430 nm and λ_em = 580 nm with FRET donor/acceptor pairs, 1,2-dipalmitoyl-sn-glycero-3-phosphoethanolamine-N-(7-nitro-2-1,3-benzoxadiazol-4-yl) (NBD-PE) and N-(lissamine rhodamine B sulfonyl)-1,2-dihexadecanoyl-sn-glycero-3-phosphoethanolamine (LR-DHPE). The total mol % of FRET-labeled lipids was between 0.2 to 0.4 mol %, and the relative surface area changes were extrapolated using a calibration curve for fluorescence readouts of known fluorophore concentrations. Triton X-100 (1% (v/v, final)) was used to disrupt the vesicles to assess background levels.

Determination of Interleaflet Flip-Flop Kinetics

Single-chain amphiphile vesicles were prepared as above to encapsulate 5 mM 8-hydroxypyrene-1,3,6-trisulfonic acid (HPTS) in 0.2 M Na⁺-HEPES, pH 8.0, and mixed in a stopped-flow device with 1:1 (v/v) 0.2 M Na⁺-HEPES, 7.0. Fluorescence was monitored continuously with λ_ex = 454 nm and λ_em = 515 nm for 10–200 points/s.

Determination of Desorption Rates

The pH-sensitive fluorophore HPTS was used to monitor the decay of the pH gradient across the membrane of reporter unilamellar POPC vesicles, as previously established.^41,42 Fatty acid vesicles were prepared as above, and POPC vesicles encapsulated 5 mM HPTS in 0.2 M Na⁺-HEPES, pH 8.0. The two vesicle populations were mixed in a stopped-flow device, and the fluorescence measurements were taken with λ_ex = 454 nm and λ_em = 515 nm and 100–200 points/s. Alternatively, reporter unilamellar POPC vesicles contained 0.2 mol % of FRET-pair lipids. The raw data was fit to a single-phase decay F(t) = F(∞) + F(0)exp(−t/τ) to determine the experimental time constant τ = t_obs. Then the t_1/2 was calculated as ln(2)(t_obs), and k_off was calculated as 1/t_obs.

Funding from the Simons Foundation (290358FY18 and 290358FY19) is acknowledged. This project received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 824060. We thank J. W. Szostak and R. Krishnamurthy for valuable comments on the manuscript.