Modulating membrane fusion through the design of fusogenic DNA circuits and bilayer composition

Membrane fusion is a ubiquitous phenomenon linked to many biological processes, and represents a crucial step in liposome-based drug delivery strategies. The ability to control, ever more precisely, membrane fusion pathways would thus be highly valuable for next generation nano-medical solutions and, more generally, the design of advanced biomimetic systems such as synthetic cells. In this article, we present fusogenic nanostructures constructed from synthetic DNA which, different from previous solutions, unlock routes for modulating the rate of fusion and making it conditional to the presence of soluble DNA molecules, thus demonstrating how membrane fusion can be controlled through simple DNA-based molecular circuits. We then systematically explore the relationship between lipid-membrane composition, its biophysical properties, and measured fusion efficiency, linking our observations to the stability of transition states in the fusion pathway. Finally, we observe that specific lipid compositions lead to the emergence of complex bilayer architectures in the fusion products, such as nested morphologies, which are accompanied by alterations in biophysical behaviour. Our findings provide multiple, orthogonal strategies to program lipid-membrane fusion, which leverage the design of either the fusogenic DNA constructs or the physico/chemical properties of the membranes, and could thus be valuable in applications where some design parameters are constrained by other factors such as material cost and biocompatibility, as it is often the case in biotechnological applications.

Tendril annealing: The fusogenic DNA nanostructures (tendrils) were prepared by mixing equimolar amounts of the corresponding oligonucleotides (Table S2) to a final concentration of 10µM (i.e. individual oligo concentration) in 100mM NaCl TE (pH 7.4). The mixture was then annealed by heating to 95°C and then cooling at a rate of -0.5°C/min to a temperature of 15°C.
Large Unilamellar Vesicle (LUV) formation: Lipids dissolved in CHCl3 were mixed to the appropriate molar ratio (Table S3) and, for labelled L liposomes, either Rhodamine-PE and NBD-PE (1%mol each) or Laurdan (1%mol) were added. The organic solvent was then removed using a rotatory evaporator to create a lipid film, which was then dried overnight under vacuum. Afterwards, it was hydrated by adding 100mM NaCl TE buffer (pH 7.4) to a final concentration of 2mg/mL total lipid and vortexed until the solution turned cloudy and the lipid film could no longer be seen. We notice resuspension of DPhPC/DOPE/Chol and DOPC/OA/Chol films required one freeze-thaw cycle. This suspension was than extruded 15 times (above the transition temperature of the lipid with the highest Tm) through a 100nm polycarbonate filter (Avanti Polar Lipids®).
In the case of content mixing assay, vesicles were resuspended in either 15mM TbCl3, 150mM sodium citrate, 5mM NaCl, 10mM Tris or 150mM DPA, 25mM NaCl, 10mM Tris buffer, for L and U liposomes, respectively. The pH of both buffers was adjusted to ~7, and freeze-thawing was performed to improve the encapsulation efficiency. After extrusion, the loaded LUVs were purified using a NAP-10 size exclusion column.
LUV functionalization with DNA tendrils: Vesicle decoration was performed by mixing labelled (L) (5µL, 2mg/mL lipid) and unlabelled (U) 15µL, 2mg/mL) LUVs with the tendrils (T L : 7.62µL, 10µM and T U : 5.49µL, 10µM respectively), which resulted in an estimated tendril:vesicle ratio of 200:1 and 50:1 respectively. These solutions were further diluted in 100mM NaCl TE (pH 7.4) buffer to achieve a final lipid concentration of 0.5mg/mL, and the mixture was incubated overnight at room temperature before use.
Fluorescence spectra and kinetics measurements: In a NUNC-96 well plate, 100µL 100mM NaCl TE (pH 7.4) and 4µL of L vesicles (0.5mg/mL) were mixed with 16µL of U liposomes (0.5mg/mL). The fusion kinetics were then recorded using a BMG CLARIOstar Plus plate reader at 530±10nm and 590±10nm emission after 465±10nm excitation (NBD/Rh FRET) or 435±10nm and 490±10nm emission under 360±10nm nm excitation (Laurdan). In the case of FRET experiments, 4µL of 2% Triton-X detergent were added and the measured NBD and Rh intensities were used as the infinite-dilution values.
The extent of lipid mixing (Emixing) was quantified as: where ( ) is the FRET efficiency at time t, 0 is the FRET efficiency at t=0 and is the FRET efficiency after LUV disruption using detergent Triton-X. In all cases, the FRET efficiency was approximated from the donor ( ) and acceptor ( ) intensities as: We note is not an actual measure of the fusion progress. First, the described FRET assay is sensitive to lipid mixing between the leaflets of different vesicles, but this may not necessarily lead to full fusion (i.e. content mixing). Secondly, previous work has highlighted the effect of membrane structure and detergent use in the extent of fluorophore dequenching upon solubilization of the membrane 1 and, lastly, FRET assays are dependent on the distance between the fluorescent dyes, which in turn will be dictated by the degree of lipid packing and/or the area per lipid. Therefore, proper quantification of lipid mixing would require individual calibrations for each individual L/U combination. However, such approach was deemed unfeasible due to the large number of lipid compositions used in this work.
When Laurdan was used, the lipid ordering of the membrane was evaluated through the General Polarization (GP) function defined as: Kinetic and fixed-time data is presented as mean ± s.d. for n≥3.
For the content mixing assay, the samples were placed in a 50µL low-volume quartz cuvette instead, and Tb(DPA)3 fluorescence intensity was measured using a Horiba Yvon Fluoromax 4 fluorimeter under 260±3nm excitation and 544±3nm emission (n=1).
Giant Unilamellar Vesicle (GUV) formation: 30µL of a 1mg/mL lipid solution with the indicated composition (Table S4) was spread onto an ITO slide. After drying for >1h under vacuum, a PDMS spacer was pressed onto the slide, the chamber was filled with a 400mM sucrose solution and was then closed with a second ITO slide. The electroformation protocol consisted of the application of 1Vpp@10Hz electric field for 90 min, followed by a 30 min detachment phase at 1Vpp@2Hz. The resulting GUVs were recovered by gently tilting the glass slide (to avoid GUV bursting due to shear exerted during pipette aspiration).
GUVs were then diluted ten-fold before imaging (using a custom-made PDMS chamber on top of a BSA-coated glass slide).
Small angle X-Ray Scattering experiments: Dry samples of a given lipid mixture (20mg total mass) were hydrated with DI water to 70% and subjected to 15 freeze-thaw cycles to ensure a proper lipid mixing. The sample was then loaded into a 2mm diameter polymer capillary tube and sealed with a rubber stopper. SAXS measurements were performed at beamline I22 (Diamond Light Source, UK). 2 The diffraction profiles were obtained from the radial integration of the 2D SAXS patterns, and peak position and width was determined by fitting them to pseudo-Voigt functions using a custom-built MATLAB® script. Fig. 3d-f: