pH-Triggered Assembly of Endomembrane Multicompartments in Synthetic Cells

By using electrostatic interactions as driving force to assemble vesicles, the droplet-stabilized method was recently applied to reconstitute and encapsulate proteins, or compartments, inside giant unilamellar vesicles (GUVs) to act as minimal synthetic cells. However, the droplet-stabilized approach exhibits low production efficiency associated with the troublesome release of the GUVs from the stabilized droplets, corresponding to a major hurdle for the droplet-stabilized approach. Herein, we report the use of pH as a potential trigger to self-assemble droplet-stabilized GUVs (dsGUVs) by either bulk or droplet-based microfluidics. Moreover, pH enables the generation of compartmentalized GUVs with flexibility and robustness. By co-encapsulating pH-sensitive small unilamellar vesicles (SUVs), negatively charged SUVs, and/or proteins, we show that acidification of the droplets efficiently produces dsGUVs while sequestrating the co-encapsulated material. Most importantly, the pH-mediated assembly of dsGUVs significantly improves the production efficiency of free-standing GUVs (i.e., released from the stabilizing-droplets) compared to its previous implementation.

Surprisingly, the TNS response showed a decrease in fluorescence intensity in acidic environment, even though an increase intensity was expected based on previously published experiments. 1-4 The molecular structure of the pH sensitive moieties was shown by Walsh and coworkers to impact its interaction with TNS, mitigating, to some extent, the expected rise in fluorescence from TNS. 5 Compared to most potent pH sensitive lipids, typically possessing tertiary amines 2, 6 , DOBAQ possesses a permanent positively charged quaternary amine sterically hindered by a carboxylic acid function located at the forefront of the lipid headgroup (Fig. S1A). As a result, DOBAQ possesses a zwitterionic state in physiological condition, while acidic environment leads to the protonation of the carboxylic acid group rendering DOBAQ positively charged ( Fig S1A). We hypothesize that the decrease in fluorescence is associated to the improved capability of DOBAQ to act as an hydrogen bond donor toward the negatively charged TNS at low pH, leading to an apparent increase of the polarity of the local lipophilic environment thus explaining the shape the of TNS assay (Fig S1B).                In all cases, a ratio of 1:2 water: oil was respected, with the typical working volume of 50 µL of aqueous:100 µL oil phase. All aqueous solutions were prepared in Milli-Q water, and all oil-surfactant mixture were prepared in HFE-7500. A 30 mM stock Krytox solution was prepared in HFE-7500.
Self-assembly of compartmentalized GUVs in presence of low pH citrate buffer: In all cases, the oilsurfactant mixture was composed of 1.4% w/w PEG-based fluorosurfactant, and 10 mM Krytox.  By assuming 3D Brownian motion, the mean square displacement, r 2 (t), of a spherical particle correspond to Where D correspond to the diffusion coefficient of the particle. Thus, the average time t required for a particle to travel a distance of r 2 (t) 1/2 is By applying the Stoke-Einstein equation of diffusion Where kB is the constant of Boltzmann, T the temperature,  the viscosity of the solution and R to the radius the particle in solution. Therefore, the average time t may be expressed as By assuming a typical viscosity of 0.904 x 10 -3 Pa  s for a phosphate buffer saline (PBS), a temperature of 298 K, and particle radius of 50 nm, the time require by a SUVs travelling from the center of the W/O droplet possessing a mean diameter of 15 µm to reach the periphery (r 2 (t) 1/2 = 7.5 µm; r 2 (t) = 56.25 µm 2 ) will be roughly 400 milliseconds.
Alternatively, an SUV of 100 nm diameter would have a D of 24.13 µm 2 s approximated by the Stoke-Einstein equation (3).

Supplementary note 4: FRAP analysis
For the FRAP analysis, the normalized fluorescence intensity values were calculated as fellow: Where IBleached correspond to the fluorescence intensity of the bleached spot, Ireference is the fluorescence intensity on a unbleached/reference spot. IPre Bleached and IPre Reference were calculated by averaging the 10 measured fluorescence intensity values before the bleaching of the Bleached, and unbleached area, respectively. A non-linear least-square function was fitted to the normalized intensities from the recovery phase. The fit-function was: Where A and  are fit parameters, and x o correspond to the time point after bleaching, also referred as the start of the recovery phase. Then, by applying the protocol reported by Axelrod 16 Where  is the half-recovery time, and r is the radius of the bleaching area.

Supplementary note 5: SUVs-to-GUV conversion efficiency calculation
The number of required lipid molecules to assemble a single giant unilamellar vesicle (GUV), NLip per GUV, can be written as: Where ASUV is the area of a GUV possessing a radius rGUV, and AHead is the area occupied by a single lipid head group.
The total number of lipids provided within the system, NLip, may be deduced from the total concentration of lipid used to produce dsGUVs: Where cLip is the lipid concentration encapsulated during dsGUV formation, NA is the Avogadro number and Vprod is the total volume of aqueous phase used to generate the W/O emulsion.
Therefore, the theoretical number of GUVs possible to produce with the provided lipids can be described by: Similarly, the experimental number of GUVs generated and released into physiological condition can be approximated by evaluating the lipid contain, and hence number of free-standing GUVs, of the released aqueous phase by: Where f is a dilution factor associated to the dilution of the lipids, cGUV is the lipid concentration associated to the free-standing GUVs measured by a calibration curve generated with the precursor SUVs, and Vwell is the well's volume used to assess the fluorescence intensity by a plate reader measurement.
Then, the percentage of conversion efficiency, %Conversion, may be described as: With an average cGUV of 7.75 µM, a well's volume of 100 µL, an initial lipid concentration cLip of 1.5 mM lipids and an aqueous volume of production VProd of 50 µL, the SUVs-toGUV conversion efficiency is roughly 20%.
Note* The factor f was evaluated as follow: Typically, 10 µL of free-standing GUVs were diluted to 100 µL (1:10 dilution), additionally, the initial 50 µL of lipid mixture used to assemble dsGUVs were released into a final 100 µL (thus, 1:2 dilution from initial lipid concentration). The resulting f factor corresponded to 20 in our experiment.
Video S1: pH sensitive SUVs entrapped in W/O droplets in presence of citrate buffer pH 5 using a two inlets microfluidics.
Video S2: Mechanical splitter module with a single inlet for production of multicompartment GUVs by microfluidics Video S3: F-actin cytoskeleton inside a free-standing GUV