Enhancing cellular transfection via microfluidic mechanoporation and carriers

Abstract

Disclosed is a method that includes encouraging a cell having a diameter through at least a first gap smaller than the diameter of the cell so as to perturb the membrane of the cell and promote fusion to the membrane by a carrier having a nucleic acid disposed therein. Also disclosed is a system that includes a mechanoporation region in fluid communication with an inlet, the mechanoporation region defining a direction of flow, the mechanoporation region defining n sets of gaps arranged in the direction of flow, wherein n is from 6 to 12 and wherein an (n+1)th gap is downstream relative to an nth gap along the direction of flow, a gap being sized so as to effect perturbation of the membrane of a cell encouraged through the gap, the perturbation being effective to promote fusion of a carrier to the membrane, a gap being defined between protrusions.

Description

103241.007498 / 25-10960 - 25-11237ENHANCING CELLULAR TRANSFECTION VIA MICROFLUIDIC MECHANOPORATION AND CARRIERSTECHNICAL FIELD

[0001] The present application claims priority to and the benefit of United States patent application no. 63 / 711,253, “Enhancing Cellular Transfection Via Microfluidic Mechanoporation And Carriers” (filed October 24, 2024) and United States patent application no. 63 / 838,146, “Droplet Squeeze Microfluidic Platform For Generating Extracellular Vesicle Hybrids For Drug Delivery” (filed July 3, 2025). All foregoing applications are incorporated herein by reference in their entireties for any and all purposes.TECHNICAL FIELD

[0002] The present disclosure relates to the field of microfluidics and also to the field of cellular transfection, particularly to the field of cellular transfection using lipid nanoparticles.BACKGROUND

[0003] Chimeric antigen receptor (CAR)-T cell therapy has revolutionized cancer treatment by engineering patients’ T cells to specifically target cancer cells. Traditional CAR-T cell manufacturing methods use viral transduction to integrate CAR genes into T cells, but this approach can cause severe side effects, immune reactions, and is costly. Accordingly, there is a long-felt need for improved approaches to integrate genes into cells, in particular integrating CAR genes into T cells.SUMMARY

[0004] In meeting the described long-felt needs, the present disclosure provides a system, comprising: a mechanoporation region in fluid communication with an inlet configured to receive at least one of a supply of cells and a supply of carriers having nucleic acid disposed therein, the mechanoporation region defining a direction103241.007498 / 25-10960 - 25-11237of flow, the mechanoporation region defining a set of first gaps and a set of second gaps, the second gaps being arranged downstream relative to the first gaps along the direction of flow, a gap being sized so as to effect perturbation of the membrane of a cell encouraged through the gap, the perturbation being effective to enhance passage of a carrier through the membrane, a gap being defined between two protrusions, a protrusion having length measured in the direction of flow and a protrusion having a cross-sectional dimension measured along the length of the protrusion and perpendicular to the direction of flow, the cross-sectional dimension being free of reduction along the direction of flow.

[0005] Also provided is a system, comprising: a mechanoporation region in fluid communication with an inlet configured to receive at least one of a supply of cells and a supply of carriers having nucleic acid disposed there, the mechanoporation region defining a direction of flow, the mechanoporation region defining n sets of gaps arranged in the direction of flow, wherein n is from 6 to 12 and wherein an (n+l)th gap is downstream relative to an nth gap along the direction of flow, a gap being sized so as to effect perturbation of the membrane of a cell encouraged through the gap, the perturbation being effective to promote fusion of a carrier to the membrane, a gap being defined between two protrusions.

[0006] Additionally provided is a method, comprising: encouraging a cell having a diameter through at least a first gap having a cross-sectional dimension smaller than the diameter of the cell, the encouraging mechanically perturbing the membrane of a cell so as to promote fusion to the membrane by a carrier having a nucleic acid disposed therein, and the first gap optionally having a width in the range of from about 15% to about 30% of diameter of the cell.BRIEF DESCRIPTION OF THE DRAWINGS

[0007] In the drawings, which are not necessarily drawn to scale, like numerals may describe similar components in different views. Like numerals having different letter suffixes may represent different instances of similar components. The drawings illustrate generally, by way of example, but not by way of limitation, various aspects discussed in the present document. In the drawings:103241.007498 / 25-10960 - 25-11237

[0008] FIG. 1. LNP + Squeeze platform design and workflow. (A) Schematic of LNP synthesis using a herringbone mixer microfluidic device. (B) Schematic of the microfluidic squeeze device showing the flow of LNPs and cells. (C) Schematic illustrating the intracellular delivery mechanism of pDNA-loaded LNPs through mechanoporation.

[0009] FIG. 2. pDNA-loaded LNP screening and characterization. (A) Transfection yield and representative histograms of GFP expression in Jurkat cells 3 days after transfection with GFP pDNA-loaded LNPs containing different ionizable / cationic lipids. (B) Plots showing the diameter and poly dispersity index (PDI) of different LNP formulations. (C) Plots showing the zeta potential and encapsulation efficiency (EE) of different LNP formulations. (D) Transfection yield and representative histograms of GFP expression in Jurkat cells 3 days after transfection with different GFP pDNA-loaded LNP formulations. (E) Fluorescent images displaying GFP expression (green) in CD45+ Jurkat cells (red) posttransfection. (F) Cell viability of Jurkat cells following transfection. (G) TEM image of pDNA-loaded LNP formulation 4. Error bars represent the mean ± standard deviation.

[0010] FIG. 3. Characterization and optimization of the microfluidic squeeze device. (A) Schematic of the microfluidic squeeze device with brightfield microscope images showing cell deformation as they pass through individual squeeze gaps.Loading efficiency and cell viability of Jurkat cells loaded with 4 kDa FITC-Dextran with (B) varying flow rates during the squeezing process with 2 pm squeeze gap width and 10 squeeze rows, (C) different squeeze gap width at 25 mL / hr with 10 squeeze rows, and (D) varying numbers of squeeze rows at 25 mL / hr with 2 pm squeeze gap width. The control is a condition where cells were incubated with FITC-Dextran without squeeze. (E) Fluorescent image of CD45+ Jurkat cells (red) loaded with 4 kDa FITC-Dextran (green) with and without squeeze (control). (F) Gene expression analysis of housekeeping and innate immunity genes in Jurkat cells 24 hours postsqueeze compared to a unsqueezed negative control. Fold changes were calculated using 18s rRNA as the reference gene. Error bars represent the mean ± standard deviation.103241.007498 / 25-10960 - 25-11237

[0011] FIG. 4. Intracellular delivery of GFP-pDNA into Jurkat cells using the LNP + Squeeze intracellular delivery platform. (A) Transfection yield and representative histograms of GFP expression in Jurkat cells 3 days post-transfection with GFP pDNA-loaded LNPs using the LNP + Squeeze method with the optimal LNP formulation 4. (B) Transfection yield and cell viability of Jurkat cells following incubation with GFP pDNA-loaded LNP formulation 4 at varying dosages, corresponding to different concentrations of encapsulated GFP pDNA. (C) Transfection yield, cell viability, and representative histograms of GFP expression in Jurkat cells 3 days post-transfection with GFP pDNA, comparing different transfection methods. (D) Time-course analysis of GFP expression in Jurkat cells using various transfection methods. (E) Fluorescent images of CD45+ Jurkat cells (red) posttransfection with GFP pDNA (green), comparing LNP + Squeeze versus LNP-only methods. (F) Radar chart evaluating the transfection performance of different transfection methods.

[0012] FIG. 5. Characterization of primary human CAR-T cells engineered using LNP + Squeeze intracellular delivery platform. (A) A schematic of the transfection of primary human T cells with CAR-pDNA, followed by the activation of effector T cells in the presence of target cancer cells. (B) Transfection yield and (C) viability of primary human T cells evaluated using different transfection methods 3 days after transfecting with CAR pDNA. (D) Time-course analysis of CAR expression in primary human T cells using the LNP + Squeeze intracellular delivery platform. Killing efficiency of CAR-T cells compared to non-transfected T cells after 48 hours of co-culture with target cells: (E) A375, (F) WM9, and (G) WM35, at various effector-to-target ratios. Error bars represent the mean ± standard deviation. Statistical significance is indicated by *** (P < 0.001) and **** (P < 0.0001), while "ns" denotes no significant difference.

[0013] FIG. 6. Jurkat cells uptake on different ionizable / cationic lipid LNP. (A) Fluorescent images of Jurkat cells expressing GFP 3 days after incubation with GFP pDNA-loaded LNPs. (B) Cell viability of Jurkat cells following uptake of LNPs. Error bars represent the mean ± standard deviation.103241.007498 / 25-10960 - 25-11237

[0014] FIG. 7. LNP + Squeeze platform design and workflow. (A) Schematic of the herringbone mixer microfluidic device used for LNP synthesis. (B) Schematic of the microfluidic squeeze device.

[0015] FIG. 8. Jurkat transfection flow rate study for LNP + Squeeze intracellular delivery platform. (A) Fluorescent images of Jurkat cells expressing GFP 3 days post-transfection using the LNP + Squeeze method at various flow rates. (B) Variations in squeeze gap width across different flow rates.

[0016] FIG. 9. Jurkat transfection LNP dosage study. Fluorescent images and representative histograms of Jurkat cells expressing GFP 3 days post-transfection with the optimal GFP pDNAloaded LNP formulation 4, incubated at varying dosages corresponding to different concentrations of encapsulated GFP pDNA.

[0017] FIG. 10. Jurkat cells transfected with GFP pDNA using various transfection methods. (A) Differential ct values (Act) calculated from qPCR using pDNA-specific primers between cellular DNA isolated from the cytoplasm and the nucleus. (B) Time-course analysis of Jurkat cell viability following transfection with GFP pDNA using various transfection methods. (C) Fluorescent images of Jurkat cells 3 days post-transfection with GFP pDNA using different transfection methods. Error bars represent the mean ± standard deviation. Statistical significance is indicated by ** (P < 0.01) and **** (P < 0.0001).

[0018] FIG. 11 provides a non-limiting depiction of the disclosed technology.

[0019] FIG. 12. Schematic illustrating the generation of EV hybrids for targeted drug delivery.

[0020] FIG. 13. Synthesis and characterization of EV hybrids, a) Dynamic Light Scattering (DLS) measurement of the size of EVs, LNPs, and EV hybrids (bars) and poly dispersity index (PDI) labeled as dark squares, b) Zeta potential measurements of EVs, LNPs, and EV hybrids indicating their surface charge, c) Nanoparticle Tracking Analysis (NTA) data showing the size distribution of EVs, LNPs, and EV hybrids, d) EV hybrids were imaged using a fluorescent microscope with NHS-PEG-AF555 labeling of EVs seen in yellow, Cy5 labeling of LNPs seen in red, and colocalized orange spots showing EV hybrids. Scale bar 10 pm. e) Stability study showing DLS data of LNPs and EV hybrids at day 1 and day 30 along with PDI represented by inverted (LNPs) and upright (EV hybrids) arrows, f) Schematic of a103241.007498 / 25-10960 - 25-11237droplet squeezer illustrating droplet generation with inputs for oil, EVs, and LNPs, along with droplet squeezing geometries featuring smooth, hill-shaped structures and a microscopic screenshot of droplets being squeezed in action. (n=3, mean with standard deviation (SD) was analyzed for all samples)

[0021] FIG. 14. COMSOL simulation of flow in the microfluidic device, a) The model device design shows velocity profiles in red lines; boundary conditions with input velocity are equivalent to 3.3 mL / hr. matching experimental condition, and pressure is zero at both input and output, b) Shear rate across the squeeze gaps, with a zoomed-in FIG. illustrating the increased shear rate at the walls of the squeeze gaps. COMSOL Multiphysics software was used to generate the plots.

[0022] FIG. 15. Optimization and characterization of droplet squeezing device for LNP-EV fusion, a) This schematic illustrates the Fluorescence Resonance Energy Transfer (FRET) dynamics wherein NBD functions as the donor and RhB serves as the acceptor molecule. LNPs incorporating this FRET pair demonstrate enhanced RhB emission, depicted in red. Upon fusion, the spatial separation of the donor and acceptor molecules leads to a significant reduction in RhB fluorescence, primarily due to the increased distance between the NBD donor, which is essential for facilitating the emission from RhB. b) Spectrophotometer reading showing NBD and RhB emission of LNPs and EV hybrids upon excitation at 460nm (NBD excitation wavelength). Assessment of fusion activity (NBD / RhB ratio) across c) different numbers of squeezing rows, d) squeeze gaps, and e) flow ratios, f) Fusion activity with droplet squeezing was compared with bulk mixing (37° C for Ihr), only bulk incubation, and other conventional methods, including extrusion and freeze-thawing. g) Assessment of fusion activity between droplets containing EVs and LNPs within diverse microfluidic mixing architectures, including serpentine channel mixers (SERP), staggered herringbone mixers (SHM), and droplet squeezing techniques, alongside appropriate control conditions for droplet mixing and incubation (no-mix). h) Evaluation of fusion activity upon loading of various cargo, including random siRNA (13kDa), EGFP plasmid (5kb length or 3300 kDa), and small molecule (433 Da). A One-Way ANOVA followed by Tukey’s post-hoc test was utilized to determine significance values. n=3 samples with SD were analyzed for each group.103241.007498 / 25-10960 - 25-11237

[0023] FIG. 16. Cryo-electron tomography of LNP-EV hybrids. a. Representative cryo-ET images of LNPs reveal their predominantly monolayered (and occasional bilayered) and spherical morphology. Many LNPs exhibit electron- dense cores, indicative of successful encapsulation of the hydrophobic drug. b. EVs display heterogeneous morphologies (shapes ranging from spherical to ovoid and sizes ranging from 100 to 500 nm) with structural diversity including unilamellar and multilamellar vesicles. EVs are characterized by electron-lucent aqueous cores and variably dense luminal inclusions, c. LNP-EV hybrid samples exhibit hallmark features of membrane fusion and hybridization, including LNPs adhered or fused with EV membranes. In some instances (last two images in c), the complex morphologies observed reflect potentially complex fusion mechanisms and reorganizations. Scalebar 100 nm.

[0024] FIG. 17. EV hybrid characterization, a) The flow cytometry analysis of small particles displays dot plots illustrating the positional shifts of NHS-PEG- AF555 stained EVs, Cy5 labeled LNPs, and the double-positive population indicative of EV hybrids, b) TEM analysis revealed gold nanoparticles conjugated Abs staining of DR5 scFv on EVs and EV hybrids. In contrast, LNPs appeared as distinct dark spherical entities, facilitating differentiation from the lighter stained EV structures, scale bar is 200 nm. c) DR5 monoclonal Ab staining of A375 melanoma cells with various degrees of DR5 receptor expression. A clear shift is seen from low expression (KO) to increased expression of DR5 in WT and Oexp cells, d) fluorescent images showing varied cell uptake of EV hybrids in DR5 KO to higher DR5 expressing WT and Oexp cells. Nuclei are stained with DAPI for visualization. Scale bar is 20 pm. e) Quantitative analysis of EV hybrid uptake per cell in diverse melanoma cell lines expressing DR5, facilitated by CellProfiler software. Paired t-test with *p<0.05 and n = 3 with SD.

[0025] FIG. 18. Characterization of drug-loaded LNPs. a) Schematic of microfluidic device with Staggered Herring bone Mixer (SHM) used to synthesize drug-loaded LNPs. b) LCMS-based measurement of drug loading efficiency in lysed LNPs. c) Cell viability of A375 melanoma cells after treatment with 100 nM ulixertinib as free drug or drug loaded LNPs (LNP-Drug), measured from luciferase activity, d) Cell viability of A375 melanoma cells after treatment with lOOnM103241.007498 / 25-10960 - 25-11237Ravoxertinib as free drug or drug-loaded LNPs (LNP-Drug), measured from luciferase activity e) IC50 calculation of ulixertinib drug in A375 melanoma cells (10-2000 nM), measured from luciferase activity, f) IC50 calculation of ulixertinib drug in WM3000 NRAS mutant metastatic melanoma cells (10-1000 nM), measured with MTS reagent for cell viability, g) Cell viability of human fibroblast BJ cells upon treatment with ulixertinib drug (10- 2000 nM), measured with MTS reagent, h) DLS measurement of size, zeta potential, and PDI of blank LNPs and ulixertinib-loaded LNPs and EV hybrids made from fusion with NK92-derived DR5 expressing EVs. n = 3 with SD.

[0026] FIG. 19. Therapeutic potential of EV hybrids as a drug delivery system, a) Schematic of in vitro treatment of EV hybrids in a 2D and 3D cell culture, b) Cell viability of luciferase A375 melanoma cells cultured in 96-well plate treated with ulixertinib loaded EV hybrids (ulix EV hybrids) for 3 days. 500 nM ulixertinib was used across all groups, c) Gene expression of apoptotic genes generated from qPCR experiments with RNA extracted from EV hybrid treated A375 melanoma cells. d) Cell viability of luciferase A375 melanoma cells in 3D spheroids after treatment with 2.5 pM ulixertinib as a free drug (ulix drug) or loaded in LNPs and EV hybrids. e) LIVE / DEAD staining of 3D spheroids shows the reduction in GFP expressing A375 melanoma cells and increase in dead cells after 3 days of treatment with EV hybrids. BJ fibroblast cells in the spheroids do not have GFP expression. Dead cells were visualized with propidium iodide staining Scale bar - 200 pm. Ulixertinib drug is represented by ulix. A one-Way ANOVA followed by Tukey’s post-hoc test was utilized to determine significance values. n=3 samples with SD were analyzed for each group. *p<0.05, **p<0.01.

[0027] FIG. 20. Characterization of LNP lipid composition, a) Lipid composition utilized for generating LNPs that fuse with EVs: NBD and RhB fluorescent probes tagged with DSPE were added for FRET studies. DLS measurements were conducted with varying concentrations ranging from 0-10% DSPE-PEG-2K to assess their b) size and c) zeta potential.

[0028] FIG. 21. Western Blot of CAR-NK92 sEVs. a) Western blot of CD9 and CD81 expression in sEVs derived from CAR-NK92 cells, along with activated T cell control and NK cell whole cell lysates (WCL). EVs were lysed using RIPA buffer supplemented with protease inhibitor cocktail, and 10-20 pg of total protein was103241.007498 / 25-10960 - 25-11237loaded per lane. CD9 detection was performed under non-reducing conditions, while CD81 detection was carried out under reducing conditions, b) Calnexin expression in CAR-NK92-derived sEVs and corresponding whole cell lysate was assessed under reducing conditions, c) TSG101 expression in CAR-NK92-derived sEVs and corresponding whole cell lysate was assessed under reducing conditions.

[0029] FIG. 22. a) A microfluidic device is configured to generate EV hybrids through droplet squeezing, b) Illustration depicting droplet squeezing through the specialized microfluidic geometry designed for the DASH method.

[0030] FIG. 23. Lyophilization of EV hybrids, a. DLS measured size and PDI of EV hybrids stored at 4°C, resuspended after lyophilization with and without 8.5% w / v sucrose, b. DLS measured zeta potential of resuspended lyophilized EV hybrids with and without 8.5% sucrose along with 4°C stored hybrid control. FD: Freeze Drying

[0031] FIG. 24. a) Small particle flow cytometry dot plot data with EV hybrids (Pl quadrant) generated from NHS-PEG-AF488-stained EVs and Cy5-labeled LNPs at EV: LNP ratios of 10:1, 1 : 1, and 1 : 10. b) %EV hybrids calculated for various ratios using the equation below.co localized population in Pl quadrant% EV hybridsEVs (P2) + LNPs (P4)

[0032] FIG. 25. EV hybrid generation utilizing adherent (A431 cells) and suspension (NK92 cell EVs) was achieved. Various techniques were employed, including bulk mixing, droplet mixing, and droplet squeezing. The droplet squeezing method showed higher fusion activity, while suspension cells EVs have comparatively higher fusion activity.

[0033] FIG. 26. a) Illustration demonstrating DR5 scFv staining of EV hybrids formed from the fusion of Cy5-labeled LNP and NHS-PEG-AF555 labeled DR5 sEVs, using FITC Fab Ab. b) The contour plot from the small particle flow cytometer indicates a double positive population of EV hybrids (purple), c) Isolated double-positive populations were gated based on their FITC fluorescence, along with control groups of LNPs, EVs, and EV hybrids that did not undergo any FITC Fab Ab staining.103241.007498 / 25-10960 - 25-11237

[0034] FIG. 27. Fluorescent images demonstrate the uptake of DR5 scFv EVs in yellow (first column) and Cy5 labeled LNP in red (second column) across various DR5 receptor-expressing melanoma cells A375 (knockout, wild type, and overexpressing cells).

[0035] FIG. 28. The uptake of EV hybrids in cells was quantified using CellProfiler software. A detailed pipeline for quantifying EV hybrid uptake in each cell was established.

[0036] FIG. 29. DLS measurements indicate the size, zeta potential, and PDI of ravoxertinib-loaded LNPs and EV hybrids created from the fusion of these LNPs with DR5 EVs.

[0037] FIG. 30. a) Schematic illustration of maleimide-thiol click chemistry demonstrating the process of antibody conjugation. Thiolated antibodies were efficiently conjugated to maleimide-functionalized LNPs. b) To confirm the successful conjugation of human DR5 antibodies on LNPs, the nanoparticles were stained with secondary antibodies: mouse IgG and human IgG. Staining results indicate the presence of functional DR5 antibodies on the LNPs. c) A range of molar ratios of antibodies to maleimide (in LNP) was evaluated to determine the optimal conditions for antibody conjugation. Fluorescently labeled secondary human IgG was utilized for staining to validate the conjugation of DR5 antibodies.

[0038] FIG. 31. a) Cell viability was evaluated in luciferase-positive (Luc+) A375 melanoma cells subjected to treatment with ulixertinib, administered as a free drug or encapsulated within EVs, LNPs, DR5 antibody-conjugated LNPs, and hybrid EV formulations at a concentration of 500 nM. b) A heat map illustrates the gene expression profiles of six apoptosis-related genes, contrasting the effects of ulixertinib treatment as a free drug, LNPs, and hybrid EV formulations. Gene expression levels were normalized to the reference gene RPL13A. c) The evaluation also included cell viability from spheroids comprising Luc+ A375 melanoma cells and luciferasenegative BJ human fibroblasts, treated with 2.5 pM ulixertinib, either as a free drug or delivered through EVs, LNPs, DR5 Ab-conjugated LNPs, and hybrid EVs.

[0039] FIG. 32. Baseline toxicity of free ulixertinib and DR5 Ab treatments. A375 melanoma cells were treated with DR5 Ab molecules at a molar concentration equivalent to EV hybrids loaded with 100 nM of ulixertinib (ulix). Treatment groups103241.007498 / 25-10960 - 25-11237included DR5 Ab alone, free ulix (100 nM), a combination of ulix at 100 nM with DR5 Ab, and EV hybrids loaded with ulix (100 nM). Cells were incubated for 72 hours, and cell viability was assessed using a luciferase-based assay.

[0040] FIG. 33. Example image of droplets flowing through rows of squeeze channels.DETAILED DESCRIPTION OF ILLUSTRATIVE EMBODIMENTS

[0041] The present disclosure may be understood more readily by reference to the following detailed description of desired embodiments and the examples included therein.

[0042] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art. In case of conflict, the present document, including definitions, will control. Preferred methods and materials are described below, although methods and materials similar or equivalent to those described herein can be used in practice or testing. All publications, patent applications, patents and other references mentioned herein are incorporated by reference in their entirety. The materials, methods, and examples disclosed herein are illustrative only and not intended to be limiting.

[0043] The singular forms “a,” “an,” and “the” include plural referents unless the context clearly dictates otherwise.

[0044] As used in the specification and in the claims, the term "comprising" can include the embodiments "consisting of' and "consisting essentially of.” The terms “comprise(s),” “include(s),” “having,” “has,” “can,” “contain(s),” and variants thereof, as used herein, are intended to be open-ended transitional phrases, terms, or words that require the presence of the named ingredients / steps and permit the presence of other ingredients / steps. However, such description should be construed as also describing compositions or processes as "consisting of' and "consisting essentially of' the enumerated ingredients / steps, which allows the presence of only the named ingredients / steps, along with any impurities that might result therefrom, and excludes other ingredients / steps.

[0045] As used herein, the terms “about” and “at or about” mean that the amount or value in question can be the value designated some other value103241.007498 / 25-10960 - 25-11237approximately or about the same. It is generally understood, as used herein, that it is the nominal value indicated ±10% variation unless otherwise indicated or inferred. The term is intended to convey that similar values promote equivalent results or effects recited in the claims. That is, it is understood that amounts, sizes, formulations, parameters, and other quantities and characteristics are not and need not be exact, but can be approximate and / or larger or smaller, as desired, reflecting tolerances, conversion factors, rounding off, measurement error and the like, and other factors known to those of skill in the art. In general, an amount, size, formulation, parameter or other quantity or characteristic is “about” or “approximate” whether or not expressly stated to be such. It is understood that where “about” is used before a quantitative value, the parameter also includes the specific quantitative value itself, unless specifically stated otherwise.

[0046] Unless indicated to the contrary, the numerical values should be understood to include numerical values which are the same when reduced to the same number of significant figures and numerical values which differ from the stated value by less than the experimental error of conventional measurement technique of the type described in the present application to determine the value.

[0047] All ranges disclosed herein are inclusive of the recited endpoint and independently of the endpoints. The endpoints of the ranges and any values disclosed herein are not limited to the precise range or value; they are sufficiently imprecise to include values approximating these ranges and / or values.

[0048] As used herein, approximating language can be applied to modify any quantitative representation that can vary without resulting in a change in the basic function to which it is related. Accordingly, a value modified by a term or terms, such as “about” and “substantially,” may not be limited to the precise value specified, in some cases. In at least some instances, the approximating language can correspond to the precision of an instrument for measuring the value. The modifier “about” should also be considered as disclosing the range defined by the absolute values of the two endpoints. For example, the expression “from about 2 to about 4” also discloses the range “from 2 to 4.” The term “about” can refer to plus or minus 10% of the indicated number. For example, “about 10%” can indicate a range of 9% to 11%, and “about 1”103241.007498 / 25-10960 - 25-11237can mean from 0.9-1.1. Other meanings of “about” can be apparent from the context, such as rounding off, so, for example “about 1” can also mean from 0.5 to 1.4.

[0049] Further, the term “comprising” should be understood as having its open-ended meaning of “including,” but the term also includes the closed meaning of the term “consisting.” For example, a composition that comprises components A and B can be a composition that includes A, B, and other components, but can also be a composition made of A and B only. Any documents cited herein are incorporated by reference in their entireties for any and all purposes.

[0050] Any embodiment or aspect provided herein is illustrative only and does not limit the scope of the present disclosure or the appended claims. Any part or parts of any one or more embodiments or aspects can be combined with any part or parts of any one or more other embodiments or aspects.

[0051] Chimeric antigen receptor (CAR)-T cell therapy has transformed cancer treatment by engineering patients' T cells to precisely target and eradicate cancer cells. Conventional methods for generating CAR-T cells rely on viral transduction to integrate specific CAR genes into the patient's T cells’ genome. As of 2024, all six FDA-approved CAR-T cell therapies utilize lentiviral and gammaretroviral vectors to deliver CAR genes into T cells. Despite exhibiting a high transduction efficiency of up to 70%, viral-based transduction can trigger severe adverse effects and provoke unwanted immune responses such as cytokine release syndrome, neurologic toxicities, prolonged cytopenias, and even cancer development. Recently, the FDA issued a boxed warning for "T cell malignancies following treatment with BCMA-directed or CD19-directed autologous CAR-T cell immunotherapies." This highlights the need for alternative methods of manufacturing CAR-T cells without using viral vectors.

[0052] To address the challenges of host genome integration from viral vector-based delivery, alternative non-viral transient transfection methods, such as the intracellular delivery of mRNA or plasmid DNA (pDNA), have been explored. These methods mitigate the risk of genotoxicity associated with integrating vectors into the host genome, which can trigger permanent gene expression in T cells and potentially lead to long-term autoimmune toxicities.11While mRNA-based transfection offers higher efficiency than plasmid-based transfection by circumventing the need for103241.007498 / 25-10960 - 25-11237nuclear translocation and transcription, its elevated cost poses a challenge for cost- effective strategies.

[0053] Intracellular pDNA delivery methods can be broadly categorized into membrane disruption-mediated and carrier-mediated strategies. Electroporation, the gold standard for hard-to-transfect primary cells, achieves high transfection efficiencies but often causes significant cell death due to the high electrical currents required to generate pores in both cell and nuclear membranes. Microfluidic mechanoporation offers an alternative by squeezing cells through narrow channels to create transient pores, maintaining good viability but with lower pDNA transfection efficiency, especially in primary human T cells. Carrier-mediated methods like polyethylenimine (PEI) and lipofectamine use positively charged nanocarriers to facilitate pDNA delivery but are limited by high cytotoxicity, making them less suitable for clinical use. Furthermore, using PEI and lipofectamine for transfection in primary cells also results in low transfection efficiency. Lipid nanoparticles (LNPs) are gaining attention for CAR-T cell manufacturing due to their high transfection efficiency and cell viability. LNP -based pDNA transfection remains less common than mRNA approaches due to the relatively low efficiency in primary human T cells, underscoring the need for improved methods to reduce the wastage of patient-collected cells.

[0054] Here, we present a rapid and highly efficient intracellular delivery system that combines microfluidic-controlled mechanoporation with LNP-based delivery mechanisms into a single intracellular delivery platform (LNP + Squeeze), ensuring high cell viability (FIG. 1). We performed a screening of a pDNA-loaded LNP library to identify the optimal lipid composition for our LNP formulation for downstream transfection studies. Our cell squeezing microfluidic platform was designed and optimized for high throughput to maximize transfection yields and significantly reduce the risk of cell clogging. We achieve this by utilizing highly parallelized squeezing regions with gaps instead of channels, featuring multiple gaps per row and multiple rows within a compact microfluidic device. Using the LNP + Squeeze platform, we transfected primary human T cells with CAR-DR5-4-1BB expressing pDNA (CAR pDNA) that encodes for the CAR gene, alongside death receptor 5 (DR5) and TNF receptor superfamily 9 (4- IBB, commonly known as103241.007498 / 25-10960 - 25-11237CD 137). DR5 receptors are commonly found overexpressed in cancer cells, such as melanoma cells, while 4-1BB (CD137) is critical in CARs as it provides powerful costimulatory signals upon ligation. With the disclosed technology, we successfully generated CAR-T cells with high transfection yield and potent cytotoxicity when cocultured with various melanoma cell lines.

[0055] Methods

[0056] LNP Synthesis and Characterization

[0057] Plasmid-encapsulated lipid nanoparticles (LNPs) are synthesized by mixing an ethanolic phase containing lipid formulations with an aqueous phase of pDNA using a herringbone mixer microfluidic device at a 1 :2.5 volume ratio with syringe pumps (Harvard Apparatus) as previously reported. The ionizable lipids, including C12-200 (Cayman Chemical), l,2-dioleoyl-3-dimethylammonium-propane (DODAP, Avanti Polar Lipids), and DLin-MC3-DMA (Cayman Chemical), as well as the cationic lipid l,2-dioleoyl-3 -trimethylammonium -propane (DOTAP, Avanti Polar Lipids), are prepared in ethanol at 10 mg / mL starting concentrations. Stock lipid formulations are made by combining these ionizable and cationic lipids with 1,2- dioleoyl-sn-glycero-3-phosphoethanolamine (DOPE, Avanti Polar Lipids), cholesterol (Avanti Polar Lipids), and l,2-dimyristoyl-sn-glycero-3-phosphoethanolamine-N- [methoxy(polyethylene glycol)-2000] (ammonium) (C14-PEG 2000, Avanti Polar Lipids) in ethanol at 10 mg / mL individual concentrations, at predetermined molar ratios (Table 1 and 2). The aqueous phase containing pDNA is prepared in 10 mM citrate buffer, pH 3.0 (Thermo Scientific Chemicals). The plasmid DNAs used are pCMV-GFP (Altogen Biosystems) and CAR-DR5-4-1BB. Post-synthesis, the LNPs undergo a two-hour dialysis period against PBS in Spectrum Spectra / Por Float- A- Lyzer G2300 kDa MWCO (Fisher Scientific). The dialyzed LNPs are sterilized by passing through 0.22 pm Pall Acrodisk sterile filters with RC membrane (VMR International LLC) and stored at 4°C.

[0058] The hydrodynamic diameter, poly dispersity index (PDI), and zeta potential of the LNPs are characterized using the Zetasizer Nano ZS-90 machine (Malvern Instrument) by dynamic light scattering. The encapsulation efficiency (EE) is quantified using a PicoGreen DNA quantification assay after lysing the LNPs in 2% Triton-X for 10 minutes, with fluorescence measurements obtained using a Spark®103241.007498 / 25-10960 - 25-11237multimode microplate reader (Tecan Life Sciences). Transmission electron microscopy (TEM) is performed on the optimized LNP to analyze the structure using the JEOL 1010 electron microscope operated at 80 kV at the Electron Microscopy Resource Lab (EMRL), Perelman School of Medicine, University of Pennsylvania.

[0059] Microfluidic Device Fabrication

[0060] Both the microfluidic squeeze device and the herringbone mixer microfluidic device are fabricated using standard photolithography and soft lithography techniques at the Singh Center for Nanotechnology at the University of Pennsylvania. The microfluidic design patterns are created in AutoCAD (Autodesk, 2024) and patterned on chrome photomasks using the Heidelberg DWL 66+ laser lithography tool (Heidelberg Instruments). For the microfluidic squeeze device, silicon molds with a height of 25 pm were fabricated using the negative photoresist SU-8 2015 (Kayaku Advanced Materials). In contrast, the herringbone mixer microfluidic device required a two-layer photolithography process due to differing channel and herringbone structure heights (FIG. 7A). Initially, the SU-83050 negative photoresist (Kayaku Advanced Materials) was used to achieve a channel height of 80 pm. This was followed by a second photolithography step using the same SU-8 3050 photoresist to create the 50 pm high herringbone structures. Polydimethyl siloxane (PDMS, Dow Corning) at 10:1 base: crosslinker ratio is cast on the silicon mold, degassed, and baked for at least 2 hours in a 65 °C oven before being peeled and hole-punched for inlets and outlets. The PDMS device is then bonded to glass slides using a plasma cleaner (Sigma Aldrich) and baked in a 65 °C oven overnight. Prior to usage, the microfluidic device is flushed with 70% ethanol, followed by cell media specific to the cells to be transfected.

[0061] Cell Culture

[0062] Jurkat, A375, WM9, and WM35 cells were cultured using standard protocols in RPMI 1640 medium (Invitrogen), supplemented with 10% fetal bovine serum (FBS, Corning) and 1% penicillin-streptomycin (Pen-Strep, Invitrogen). All cells were maintained at 37°C in a humidified atmosphere with 5% CO2.

[0063] Intracellular Delivery Procedure

[0064] For the optimization of the microfluidic mechanoporation squeeze device operating parameters, cells are prepared at a concentration of 1 x 106cells / mL103241.007498 / 25-10960 - 25-11237and mixed with 300 pg / mL of 4kDa FITC-Dextran (Sigma-Aldrich) in culture media. This mixture is delivered into the microfluidic device at desired flow rates using syringe pumps. After squeezing, the cells are diluted in PBS to a total volume of 25 mL, spun down, and the supernatant is removed to ensure complete washing away of untransfected 4kDa FITC-Dextran. For positive controls of transfection, cells are treated with 5 pg of pDNA per 1 x 106cells using different methods: electroporation, lipofection, and polyethylenimine (PEI) treatment. Electroporation is performed using the Neon NxT Electroporation System (Invitrogen) with 3 pulses at 1150 V. For lipofectamine transfection, pDNA is mixed with Lipofectamine 3000 (Invitrogen) and added to the cells according to the manufacturer’s instructions. For PEI transfection, PEI MAX (Polysciences) reagent is added to pDNA at a ratio of 6.5 pL of 1 mg / mL PEI MAX reagent per 1 pg of pDNA and incubated for 5 minutes before being added to the cells. For LNP-based transfection, pDNA-loaded LNPs are added to the cells at the desired pDNA concentration. For LNP + Squeeze intracellular delivery, cells mixed with LNPs are delivered into the microfluidic device as described above. After processing, the cells are collected and added into fresh culture media.

[0065] Housekeeping and Innate Immunity Gene Assessment

[0066] Jurkat cells were squeezed and cultured for 24 hours, with unsqueezed cells serving as a control. After 24 hours, RNA was extracted from both conditions using TRIzol Reagent (ThermoFisher) and labeled with GlycoBlue Coprecipitant (ThermoFisher). Genomic DNA was removed with a DNase treatment, using 2U of Turbo DNase (ThermoFisher) at 37 °C for 30 minutes, followed by the addition of DNase inactivation reagent for 5 minutes at room temperature. The iScript cDNA Synthesis Kit (Bio-Rad) was used for reverse transcription PCR (RT-PCR) following the manufacturer's protocol, and reactions were run in a SimpliAmp Thermal Cycler (Applied Biosystems). qPCR was performed using PowerTrack SYBR Green Master Mix (Applied Biosystems) and customized primers (Integrated DNA Technologies). A total of 40 qPCR cycles were conducted using the QuantStudio 3 Real-Time PCR machine (Applied Biosystems). All ct values were normalized against the Ct value of the 18s rRNA housekeeping gene prior to the calculation of fold change using the formulaFoldchange = 2~AACt103241.007498 / 25-10960 - 25-11237where AAct is the Act between squeezed and unsqueezed conditions after normalization.

[0067] Plasmid Localization Study

[0068] To determine the localization of intracellularly delivered pDNA, nuclei and cytosolic DNA from squeezed and unsqueezed cells were isolated and analyzed for the presence of transfected pDNA. Jurkat cells were squeezed with pDNA-loaded LNPs and cultured for 3 days, with unsqueezed cells serving as controls. After 3 days, the Jurkat cells were harvested and washed before DNA isolation. For the isolation of cytosolic DNA, methods described by Mosley et al. were followed. Briefly, cells were treated with a Cytosolic Extraction Buffer to collect the cytosolic fraction. DNA was then extracted using phenol-chloroform, followed by proteinase K and RNase A treatment to remove cytosolic proteins and RNA, respectively. For the isolation of cell nuclei, methods described by Nadelmann et al. were used. Cells were treated with Nuclei Isolation Buffer and Homogenization Buffer, followed by centrifugation to pellet the nuclei. TRIzol extraction was used to isolate DNA from the nuclei pellet, followed by the removal of RNA and protein. qPCR was performed on both nuclear and cytosolic DNA using methods similar to those previously mentioned, with customized primers targeting sequence-specific regions of the pDNA.

[0069] Primary Human T Cell Activation

[0070] Primary T cells are activated from fresh peripheral blood mononuclear cells (PBMCs). PBMCs are purchased and isolated from HLA-typed leukapheresis donors at the Human Immunology Core, Perelman School of Medicine at the University of Pennsylvania. Primary T cells are activated using a 1 pg / mL antihuman monoclonal antibody cocktail of CD3 (OKT3 clone) / CD28 (CD28.2 clone) (eBioscience) and stimulated for 48 hours. Stimulated primary T cells are harvested and cultured in RPMI 1640 medium, supplemented with 10% FBS, 1% Pen-Strep, and 100 lU / mL of recombinant human IL-2 (Peprotech) for in vitro expansion. Primary T cells are proliferated for up to 14 days post-activation. All experiments were performed using primary T cells before 14 days of activation.

[0071] CAR-T Cell Generation

[0072] For the generation of primary CAR-T cells using the LNP + Squeeze intracellular delivery platform, primary T cells are prepared at 1 x 106cells / mL with103241.007498 / 25-10960 - 25-1123735 pg of pDNA-loaded LNPs in culture media. The cell mixture is pumped at 2 mL / hr flow rate through the microfluidic mechanoporation squeeze device and collected at the outlet. Transfected cells are added into fresh culture media and cultured for 3 days.

[0073] Flow Cytometry

[0074] Before flow cytometry, harvested cells are washed and resuspended in cell staining buffer (BioLegend). CAR expression is measured by staining with AlexaFluor 488 Anti-Human IgG, F(ab')2 dye (Jackson ImmunoResearch Laboratories) at a 1:50 dilution. Cell viability is assessed using 7-AAD Viability Staining Solution (BioLegend). Cells are also stained with APC anti-human CD45 [Clone: HI30] (BioLegend) stain as a positive control. The stained cell suspension is analyzed using the BD FACSymphony A3 Lite flow cytometer (BD Biosciences) at the Penn Cytomics and Cell Sorting Shared Resource Laboratory at the University of Pennsylvania. Flow cytometry data is analyzed and plotted using FlowJo VI 0 Software (BD Biosciences). During flow analysis, cells are first gated to exclude cell debris and aggregates using forward scatter (FSC) area vs. side scatter (SSC) area plots. Single cells are then gated using FSC-area and FSC-height plots, followed by positive gating of APC fluorescence to identify CD45+cells. To measure viability, cells are gated to determine the percentage of 7-AAD negative (live) and positive (dead) populations. For measuring FITC-Dextran loading efficiency, GFP, or CAR expression, viable cells are gated to determine the percentage of cells with green fluorescence relative to the untransfected negative control cells.

[0075] Killing Assays using Engineered Primary Human CAR-T cells

[0076] All killing assays are conducted with Luciferase Assay System (Promega Corporation). Target luciferase-exhibiting melanoma cancer cells are seeded in 96-well plates at 2 x 104cells per well in 50 pL of RPMI 1640 medium without FBS growth serum and incubated for 4 hours to allow for complete adhesion of the target cells. Effector T cells are then seeded on the target cells at predetermined ratios in 50 pL of primary human T cell culture medium and cultured for up to 48 hours. To measure cell-associated luciferase activity, the cells are gently spun down at 1000 rpm for 5 mins to remove the supernatant, followed by a Dulbecco’s phosphate-buffered saline (DPBS, Invitrogen) wash and the subsequent removal of the wash supernatant.45 pL of IX Reporter Lysis Buffer (Promega Corporation), diluted in DI water, is103241.007498 / 25-10960 - 25-11237added to each well. This is followed by one complete freeze-thaw cycle by freezing at -80°C and thawing at room temperature to ensure complete cell lysis. The lysed mixture is then transferred into a Nunc White 96-well microplate (Thermo Scientific). Then, 100 pL of Luciferase Assay System reagent is directly added into each well, and luminescence measurements are performed immediately using a Spark® multimode microplate reader (Tecan Life Sciences). Luminescence readings from wells with blank media are used to set the 100% killing efficiency (maximum cell death) reference, and luminescence readings from wells with only target cells are used to set the baseline 0% killing efficiency (target cell spontaneous death) reference. The percentage killing efficiency was calculated as

[0077] Killingefficiency =Experimentaldata-targetcellspontaneousdeath -mnn / X 1 U U / oBlankmediamaximumcelldeath-targetcellspontaneousdeath

[0078] Statistical Analysis

[0079] All data are represented as mean ± standard deviation. All analyses and plots were performed using GraphPad Prism 10 (GraphPad Software Inc).Unpaired student's t-test was used for analysis between two groups. For all tests, p values <0.05 indicate statistical significance.

[0080] Results

[0081] LNP Screening and Characterization

[0082] We aimed to identify the optimal lipid composition for achieving the highest transfection yield while maintaining cell viability. Jurkat cells were selected as our model system as they are a human T cell leukemia line, closely mimicking our ultimate goal of transfecting primary T cells. To achieve this, an initial library of LNPs encapsulating GFP-expressing pDNA (Altogen Biosystems) was synthesized using different ionizable / cationic lipids (C12-200, DODAP, DLin-MC3-DMA, and DOTAP), with fixed lipid compositions of a PEG-lipid conjugate (C14-PEG2000), phospholipid (DOPE), and cholesterol (Table 1). These LNPs were characterized by hydrodynamic size, poly dispersity index (PDI), and zeta potential. All four LNP formulations exhibited similar sizes ranging from 134 nm to 186 nm, low PDI values (below 0.21), encapsulated pDNA concentrations between 22 to 60 ng / pL, and high encapsulation efficiencies (above 88%). C12-200 and DOTAP -based LNPs demonstrated larger positive zeta potentials (13.8 mV and 12.7 mV, respectively),103241.007498 / 25-10960 - 25-11237while DLin-MC3-DMA and DODAP -based LNPs exhibited smaller negative zeta potentials (-3.9 mV and -7.5 mV, respectively), consistent with previous findings.Transfection yield and cell viability were assessed in Jurkat cells 3 days postincubation with the LNPs using flow cytometry. Among the formulations, only C12- 200 lipid-based LNPs induced strong GFP expression (58.4%) while maintaining good cell viability (95.6%) compared to the non-transfected control, outperforming the other ionizable / cationic LNPs (FIG. 2A, FIG. 6). Previous studies have shown that using DOTAP -based LNPs to deliver pDNA at our DOTAP / cholesterol molar ratio typically achieve transfection efficiencies between 5% to 10%, with the optimal molar ratio only reaching up to 20% efficiency. Furthermore, DODAP and DLin-MC3-DMA are more commonly used for siRNA delivery rather than for pDNA due to their poor endosomal escape and DNA release characteristics. As expected, C 12-200-based LNPs performed the best, consistent with recent efforts demonstrating their efficacy for pDNA delivery using LNPs. Consequently, C12-200 was selected as the ionizable lipid for downstream LNP formulation studies.

[0083] Table 1. Ionizable / cationic lipid selection tableMolar Ratio (%) CharacterizationLNP Ionizable C14- pDNA DOP Cholester Diamete Zeta E.E. / Cationic PEG PDI (ng / pLE ol r (nm) (mV)ipid 2000 (%))0.120178.03 13.77 ± 59.71 ± 90.19 Cl 2-200 35 42.5 20 2.5 ±± 3.06 0.31 1.75 ± 0.180.029DLin- 0.055185.93 -3.88 ± 57.91 ± 97.19 MC3- 35 42.5 20 2.5 ±± 2.46 0.11 6.26 ± 0.31 DMA 0.0090.208175.83 -7.45 ± 22.66 ± 87.87 DODAP 35 42.5 20 2.5 ±± 3.58 0.07 2.80 ± 1.370.0200.139133.90 12.70 ± 21.87 ± 95.95 DOTAP 35 42.5 20 2.5 ±± 1.85 0.46 0.40 ± 0.040.019

[0084] Subsequently, we screened four different DOPE and cholesterol lipid molar ratios, using C12-200 as the ionizable lipid (Table 2). The resulting LNPs ranged in size from 169.5 nm to 214.6 nm, with PDI values between 0.120 and 0.175,103241.007498 / 25-10960 - 25-11237and high encapsulation efficiencies ranging from 90.19% to 94.70% (FIG. 2B, C). Notably, increasing the DOPE molar ratio above 42.5% resulted in increased zeta potentials, consistent with previous findings. LNP formulations 3 and 4 exhibited significantly larger positive zeta potentials (13.8 mV and 12.1 mV, respectively) compared to the smaller positive zeta potentials of formulations 1 and 2 (5.1 mV and 3.5 mV, respectively) (FIG. 2C). In Jurkat cells, the increase in DOPE molar ratio positively correlated with transfection yield, again consistent with previous reports. Specifically, LNP formulations 2, 3, and 4 induced strong GFP expression, with formulation 4 showing the highest transfection yield (70.1%) while maintaining excellent cell viability (96.2%) compared to the non-transfected control (FIG. 2D-F). Therefore, LNP formulation 4 was identified as the lead LNP formulation for further studies. TEM analysis revealed that LNP formulation 4 displayed a spherical morphology with an electron-dense core (FIG. 2G).

[0085] Table 2. C 12-200 lipid composition selection tableMoiar RatioLNPC12-2S0 DOPE Cholesterol C14-PEG 2000 LNP 1 35 32.5 30 2.5 LNP 2 35 37.5 25 2.5 LNP 3 35 42.5 20 2.5 LNP 4 35 47.5 15 2.5

[0086] Microfluidic Squeeze Device Optimization

[0087] The intracellular delivery of payloads into cells via microfluidic mechanoporation is largely influenced by flow conditions and the design of the microfluidic channels and structures. However, most designs for microfluidic mechanoporation devices are prone to cell clogging due to the use of single squeeze channels. To mitigate this, we highly parallelized our squeezing regions by designing gaps instead of channels, with 20 gaps per row and multiple rows packed into a compact microfluidic device (FIG. 3 A, FIG. 7B). The squeeze gap length was kept constant at 50 pm. To optimize gap designs and operating conditions, we loaded 4 kDa FITC-Dextran into Jurkat cells. By varying flow rates from 2 to 25 mL / hr, we observed that the loading efficiency increased with higher flow rates compared to the unsqueezed negative control (FIG. 3B). Increasing flow rates generate higher shear103241.007498 / 25-10960 - 25-11237stress, leading to greater cell head expansion and nucleus deformation, potentially resulting in larger transient cell pores that facilitate the exchange of cytosolic and external fluids. The loading efficiency began to plateau at flow rates above 25 mL / hr, which higher flow rates were also avoided due to the excessive internal microfluidic channel pressure that led to tubing bursts and cell loss. Given that Jurkat cells range from 10 pm to 16 pm, we optimized the squeeze gap widths by varying them between 2 pm and 10 pm, as previous reports indicate that the optimal gap widths should be between 15% to 30% of the cell diameter. We observed that devices with 2 pm squeeze gap widths achieved the best loading efficiency without compromising cell viability (FIG. 3C). When varying the number of parallel squeeze rows, we found that 10 squeeze rows exhibited the best loading efficiency (FIG. 3D). As cell viability did not significantly decrease even with up to 10 rows, we selected ten squeeze rows to accommodate harder-to-transfect cells and larger payloads.

[0088] Jurkat cells loaded with 4 kDa FITC-Dextran under optimized conditions (2 pm gap, 25 mL / hr, and 10 rows) exhibited excellent loading efficiency (78.4%) comparable to previous studies, retained great cell viability (89.5%), and maintained similar phenotypic characteristics to unsqueezed cells (FIG. 3E). To ensure negligible effects of cell squeeze on cell perturbation, we compared the expression of 7 housekeeping genes and 4 innate immunity genes between squeezed and unsqueezed cells 24 hours post-squeezing. qPCR analysis revealed no significant differences in gene expression (unpaired t-test; p = 0.998). Using 18s rRNA as a housekeeping gene to calculate fold change, each gene remained close to 1, indicating negligible differences between squeezed and unsqueezed cells (FIG. 3F).

[0089] LNP + Squeeze Enabled Intracellular Delivery

[0090] Next, we aimed to test the hypothesis that combining LNP uptake with microfluidic mechanoporation would enhance transfection efficiencies by accelerating LNP fusion through temporary cell deformation. To the inventors’ knowledge, this is the first reported instance of integrating a membrane disruption- mediated method with a carrier-mediated one. Previous efforts have typically focused on combining different membrane-disruption methods, such as mechanoelectroporation and electro-sonoporation, rather than merging membrane disruption with carrier-mediated delivery. We hypothesized that the combined LNP + Squeeze103241.007498 / 25-10960 - 25-11237approach required re-optimization of flow conditions to lower and gentler flow rates compared to squeeze-only methods. This was necessary because while the delivery of FITC-Dextran needed to be entirely intracellular, LNP-loaded pDNA delivery required gentler flow rates to promote fusion of the LNP to the cell membrane. As anticipated, a flow rate of 2 mL / hr, unlike the previously optimized flow conditions for squeeze-only methods, provided a slightly better transfection yield (69.47%) without substantially decreasing cell viability (FIG. 4A, FIG. 8A). This improvement may also be attributed to the increase in effective channel diameter at higher flow rates, which diminishes the squeeze effects on the cells (FIG. 8B). We also optimized the LNP dosage for maximum transfection yield. Transfection yield increased as we raised the initial dosage from 7 pg to 35 pg of pDNA in LNP per 106cells, beyond which no further increase in transfection yield was observed (FIG. 4B, FIG. 9).

[0091] Using the optimized flow condition and LNP dosage, we tested the GFP pDNA transfection efficiency of the LNP + Squeeze method on Jurkat cells compared to conventional methods such as electroporation, lipofection, and PEI transfection, as well as pDNA endocytosis-only, squeeze, and LNP-only methods. Three days post-transfection, we observed that only electroporation, LNP-only, and LNP + Squeeze methods elicited strong transfection yields, with the LNP + Squeeze method showing a significant increase compared to the other two methods (LNP + Squeeze: 75.23%, LNP: 44.53%, electroporation: 42.97%) (FIG. 4C). We hypothesize that the squeezing mechanism facilitates LNP adhesion to the cell membrane by promoting the spontaneous formation of unique protein coronas, distinct from those formed in a static fluidic environment, thereby enhancing cell uptake and overall transfection efficiency. Furthermore, we confirmed that the LNP + Squeeze method allows more pDNA to localize in the cell nucleus rather than the cytoplasm, compared to LNP incubation alone, following endosomal escape (FIG. 10A). Transfection yields plateaued three days post-transfection, while cell viability remained high for at least seven days (FIG. 4D, FIG. 10B). Additionally, the LNP + Squeeze platform not only transfected more cells but also delivered more payload per cell, as evidenced by the higher frequency of cells with stronger GFP expression compared to other methods (FIG. 4E, FIG. 10C). Compared to other transfection methods, the LNP + Squeeze method exhibited stronger transfection yield while maintaining good cell viability103241.007498 / 25-10960 - 25-11237(91.5%). It also demonstrated high transfection throughput and uniformity, compatibility with different cell types, and allowed for denser payload packaging through LNP encapsulation (FIG. 4F).

[0092] Primary Human CAR-T Cell Engineering

[0093] To evaluate the potential clinical relevance of the disclosed LNP + Squeeze technology, we engineered primary human CAR-T cells by delivering CAR pDNA using the disclosed platform, followed by the co-incubation of the CAR-T cells with target cancer cells (FIG. 5A). CAR pDNA-loaded LNPs were synthesized using the optimized LNP formulation 4 (Table 3). We compared the disclosed technology against various transfection methods on primary human T cells. Three days posttransfection, we found that the LNP + Squeeze method achieved the highest CAR expression (57.44 ± 17.75%), as measured by flow cytometry, significantly outperforming other methods, which did not show a clear increase in CAR expression compared to the non-transfected control (FIG. 5B). However, the LNP + Squeeze method also resulted in a substantial reduction in cell viability (65.99 ± 18.71%) (FIG.5C). Interestingly, the squeeze-only and LNP-only transfection methods did not cause such a marked decrease in cell viability, suggesting that the viability reduction could be due to an excess of CAR pDNA translocating into the nucleus. Moreover, we observed considerable variability in both transfection yield and cell viability, which can be attributed to donor-to-donor variability in primary human T cells which is also a common occurrence in clinical settings. For an optimal incubation time after transfection, we determined it to be 3 days, consistent with previous findings in Jurkat cells (FIG. 5D).

[0094] Table 3. CAR pDNA-loaded LNP Characterization CharacterizationDiameter (nm) PDI Zeta (mV) pDNA (ng / pL) E.E. (%) 176.97 ± 3.45 0.139 ± 0.044 4.86 ± 0.16 110.08 ± 8.94 93.86 ± 0.54

[0095] Specific Killing of Target Cells by Engineered CAR-T Cells

[0096] To evaluate the functional efficacy of the engineered primary human CAR-T cells, we co-cultured them with various melanoma cancer cell lines at different effector-to-target (E:T) ratios for 48 hours. When incubated with A375 cells, the CAR-103241.007498 / 25-10960 - 25-11237T cells showed high killing efficiencies across E:T ratios from 5:1 to 1:1.6 (FIG. 5E). However, the untransfected control also displayed a similar trend, albeit with slightly lower efficiencies, with the difference in killing efficiency being only significant down to an E:T ratio of 1.25:1. This sensitivity could be due to A375 cells being more susceptible to molecules like perforin and granzymes secreted by activated T cells. To further emphasize the improved killing efficiencies of the CAR-T cells over untransfected T cells, we tested them against WM9 and WM35 cell lines, which exhibited lower killing efficiencies for the untransfected T cells. For WM9 cells, the CAR-T cells demonstrated significantly greater killing efficiencies compared to the untransfected T cells, even at low E:T ratios where the untransfected T cells showed almost no killing (FIG. 5F). At these ratios (1.25:1 and 1:1.6), the CAR-T cells maintained substantial killing efficiencies (79.89 ± 4.01% and 47.94 ± 7.69%, respectively). Similarly, for WM35 cells, the CAR-T cells exhibited improved killing efficiencies down to an E:T ratio of 1:3.2 (54.64 ± 3.65%) (FIG. 5G). Overall, these killing assays on various melanoma cancer cells demonstrate a successful engineering of CAR-T cells using the disclosed LNP + Squeeze platform.

[0097] Discussion

[0098] Non-viral intracellular delivery methods are revolutionizing the next generation of CAR-T cell manufacturing, offering a safer alternative to current viralbased transduction methods. In this study, we introduce a novel approach that combines the dual actions of membrane-disruption mechanoporation and carrier- mediated LNP delivery to efficiently engineer CAR-T cells. We demonstrated that these two methods have synergistic effects on transfection efficiency when used together. By utilizing LNPs to package payloads at relatively high concentrations, we achieved higher delivery efficiency compared to mechanoporation alone. The versatility of LNP encapsulation allows for the delivery of various types of payloads, including mRNA and siRNA, highlighting the multifaceted nature of the LNP + Squeeze platform. The disclosed high-throughput design, featuring parallelized rows of multiple squeeze gaps within a single microfluidic device, mitigated cell loss due to clogging, an issue that is common in squeeze-based methods.

[0099] Conclusions103241.007498 / 25-10960 - 25-11237

[0100] In summary, provided here is an innovative platform capable of achieving high-throughput transfection in human primary T cells while preserving cell viability and the expression of innate immunity genes. This is accomplished through a synergistic approach that combines LNP -mediated endocytosis with mechanoporation. To the inventors’ knowledge, this is the first instance of coupling a membranedisruption method with a carrier-mediated delivery system to significantly enhance transfection efficiency. Our platform has successfully demonstrated the efficient delivery of sizable plasmid payloads to hard-to-transfect primary human T cells for generating CAR-T cells without relying on traditional viral vector-based intracellular delivery methods. This approach provides cost-effective and scalable solutions in CAR-T cell manufacturing and can be applied to the treatment of other diseases.

[0101] Additional Disclosure

[0102] Extracellular vesicles (EVs) are emerging as versatile drug delivery systems due to their intrinsic biocompatibility and targeting capabilities. However, EV integrity and efficient drug loading challenges hinder their clinical translation. To address these limitations, hybrid systems integrating lipid nanoparticles (LNPs) with EVs have gained attention for their potential in targeted and combinatorial drug delivery. This study presents a robust microfluidic approach for the scalable generation of drug-loaded EV-LNP hybrids (EV hybrids). Our method facilitates controlled fusion between EVs and LNPs by utilizing a droplet-mediated squeezing mechanism.

[0103] Extracellular vesicles (EVs) are emerging as versatile drug delivery systems due to their intrinsic biocompatibility and targeting capabilities. However, EV integrity and efficient drug loading challenges hinder their clinical translation. To address these limitations, hybrid systems integrating lipid nanoparticles (LNPs) with EVs have gained attention for their potential in targeted and combinatorial drug delivery. This study presents a robust microfluidic approach for the scalable generation of drug-loaded EV-LNP hybrids (EV hybrids). Our method facilitates controlled fusion between EVs and LNPs by utilizing a droplet-mediated squeezing mechanism.

[0104] We optimized lipid composition and microfluidic parameters for the fusion of EVs and LNPs and determined physicochemical and functional characterizations of the EV hybrids. In vitro studies demonstrated that EV hybrids exhibit enhanced targeting efficiency. Moreover, we successfully encapsulated small-103241.007498 / 25-10960 - 25-11237molecule therapeutics within EV hybrids, significantly improving cytotoxic efficacy against melanoma in 2D and 3D culture models compared to drug-loaded EVs or LNPs alone. Our work introduces a scalable, minimally disruptive microfluidic platform for engineering EV hybrids, offering a promising strategy to advance precision nanomedicine.

[0105] Introduction

[0106] Lipid nanoparticles (LNPs) have shown immeasurable success as drug delivery vehicles. Small molecules and nucleic acids (e.g., siRNA, mRNA, and plasmids) are susceptible to enzymatic degradation and have low bioavailability and targeting, leading to various side effects. LNPs provide several benefits to improve the bioavailability of these drugs by encapsulating them in materials such as lipids, controlling drug release and offering engineering for targeting abilities. Despite their promising potential, lipid nanoparticles (LNPs) still face several limitations in drug delivery applications to specific targets. Even with sophisticated LNP systems, such as the use of ionizable lipids and targeting moieties, they still face toxicity issues, adverse clinical effects, and low therapeutic outcomes due to their clearance.Furthermore, achieving precise targeting and controlled release of the therapeutic agent remains challenging, necessitating ongoing research to optimize these delivery systems.

[0107] Extracellular vesicles (EVs) are naturally occurring, membrane-bound particles released by cells that have emerged as promising vehicles for drug delivery. Due to their biocompatibility, ability to evade the immune system, and inherent capacity to deliver therapeutic cargo such as proteins, RNA, and small molecules directly to target cells, EVs offer significant advantages over synthetic material-based drug delivery systems and cell membrane-coated nanoparticles. Similarly, cell membrane-coated nanoparticles have been explored for targeted drug delivery due to their potential for immune evasion and tissue-specific targeting. However, their application is constrained by inconsistencies in membrane coating techniques and the absence of intrinsic bioactive components characteristic of EVs. While EVs face challenges such as heterogeneous populations and limited drug loading capacity, their integration with lipid nanoparticles offers a synergistic platform that overcomes these limitations and provides superior therapeutic potential compared to cell membrane103241.007498 / 25-10960 - 25-11237coating alone. Various methods are explored to improve drug loading, with potential disruption of EV integrity during loading processes. Various methods have been explored to load cargo into EVs, with very few reaching clinical trials, which exhibits the low translational efficacy of EV-loaded cargo. New methods of loading small molecule drugs without interfering with EV integrity are needed to exploit the therapeutic effects of EV targeting and immunomodulation.

[0108] EV hybrids combined with LNPs and EVs offer several advantages. These include enhanced targeting specificity and reduced off-target effects due to their natural ability to home in on specific tissues or cells. These hybrids combine the biocompatibility and low immunogenicity of natural EVs with the customizable functionality (e.g., cargo, surface membrane) of synthetic LNPs. Therefore, a robust and scalable approach to making EV hybrids can improve their utility and clinical translation. Currently, various physical and chemical methods are employed to synthesize EV hybrids. Physical methods include freeze-thawing, sonication, and extrusion of liposomes with EVs of interest. Chemical methods involve changes in lipid composition, transfection agents, pH gradients, and complementary affinity molecules. These methods are limited to small molecules with low drug loading and potentially interfere with the integrity of the EV membrane and leakage of bioactive components in the lumen. Overall, various strategies to synthesize EV hybrids have been reported, involving diverse compositions and complex and time-consuming methods. They have poor characterization and standardized protocols, eventually curbing their clinical translation.

[0109] Few technologies have been applied to generate hybrid vesicles combining EVs with lipid-based nanoparticles or other synthetic materials. These methods face limitations such as disruption of EV membranes arising from sonication and the loss of potent EV cargo (nucleic acids, proteins, and other bioactive molecules), which reduce the therapeutic effect. Hence, a more robust microfluidic approach to reduce EV membrane disruption and fusion with lipid nanoparticles can improve drug delivery outcomes.

[0110] To enable the highly efficient generation of EV hybrids with the least amount of EV modifications, we provide a microfluidic platform that enables Droplet- Assisted Squeezing for EV Hybrids (DASH). Compared to bulk fusion or continuous-103241.007498 / 25-10960 - 25-11237flow methods, DASH improves vesicle uniformity and scalability. In this disclosure, we provide a droplet microfluidic-based method to enable a robust and straightforward approach to generating EV hybrids with high efficiency and precise control. This was achieved by utilizing charged lipids and polyethylene glycol (PEG) in LNPs and an on-chip droplet squeezing to provide high shear rates, facilitating fusion between EVs and LNPs to form EV hybrids. Our approach enables robust fusion of EVs with LNPs carrying various cargo, without manipulating EVs maintaining their intrinsic features. To show the feasibility of our platform for drug delivery applications, we formulated EV hybrids to target death receptor 5 (DR5) receptors overexpressed on melanoma cells and deliver small molecule therapeutics (FIG. 12). Here we employed DR5 scFv expressing EVs isolated from NK92 natural killer cells. These cells have been well-characterized and used in clinical trials, establishing their safety profile. In a previous study, we also showed excellent safety profiles of NK92 EVs in vivo. In this proof-of-concept study, we showed that the EV hybrids generated by our platform enhanced cancer killing in 2D and 3D melanoma cell cultures.[OHl] Results

[0112] Generation and characterization of LNPs

[0113] Cationic lipids have been used for LNP fusion with negatively charged EV membranes. We employed DOTAP cationic lipid and neutral lipids of DPPC and cholesterol in our LNPs to facilitate fusion between LNPs and EVs (FIG.20a). Additionally, the PEG molecule, which has been shown to enhance cell-cell membrane fusion, was included in the lipid composition. Unlike most of the previous fusion studies using PEG in the buffer for mixing EVs and LNPs, PEG was directly embedded in the LNP membrane via l,2-distearoyl-sn-glycero-3- phosphoethanolamine-N-[square(polyethyleneglycol)-2000] (DSPE-PEG-2000). We screened DSPE-PEG-2000 concentrations (0-10% molar ratio) in our lipid composition for LNPs and found that an increase in PEG concentration helps reduce the size of LNPs. Here, we employed a 10% molar ratio of PEG concentration, which exhibited an optimal LNP size (~100nm) and zeta potential (positive charge) for our fusion purposes (FIG. 20b-c). A microfluidic chaotic mixer using staggered herringbone geometry facilitated the synthesis of positively charged LNPs with a size103241.007498 / 25-10960 - 25-11237of 90 ± 7 nm, a poly dispersity index (PDI) of 0.09 (FIG. 13a), and zeta potential of 5 ± 1 mV (FIG. 13b), measured by Dynamic Light Scattering (DLS). Nanoparticle Tracking Analysis (NTA) showed a uniform distribution of LNP sizes, ranging from 95 ± 20 nm with a narrow peak, in agreement with the DLS data (FIG. 13c).

[0114] Droplet squeezing to generate EV hybrids

[0115] Small EVs expressing Death Receptor 5 (DR5) agonist (DR5 sEVs) were engineered from CAR-NK cells. NK-92 cells, which naturally produce sEVs containing cytotoxic proteins with anti-tumor activity, were genetically modified to secrete sEVs displaying chimeric antigen receptors of DR5 agonistic single-chain variable fragments (scFvs) on their surface. Our Western blot analysis demonstrates strong expression of EV markers CD9, CD81, and TSG101 in CAR-NK92-derived sEVs (FIG. 21). CD9, low in whole cell lysates of NK-92 cells, was enriched in EVs, suggesting selective sorting during vesicle formation. CD81 was robustly expressed in both cells and EVs, consistent with its known role as a broadly conserved tetraspanin marker. TSG101, part of the ESCRT machinery, was detectable in EVs, confirming vesicle identity and isolation quality. Additionally, calnexin marker was absent in EVs compared to cell lysate, indicating that EVs were free from cell debris contamination. These findings align with prior reports of immune cell EV profiles and follow MISEV2018 guidelines. These DR5 sEVs (referred to as EVs unless specified) were subsequently fused with LNPs using a specialized microfluidic device featuring a droplet-squeezing geometry (FIG. 3f).

[0116] The microfluidic device setup with EV and LNP inlets and collection outlet are shown in FIG. 22a. Smooth, hill-shaped structures were arranged in 10 per row across 40 rows, designed to facilitate droplet squeezing at high throughput between the structures with 4 pm squeezing gaps (FIG. 22b). Droplets sized 40 ± 5 pm were generated using a flow-focusing method with equal flow rates of EVs and LNPs with premixing geometry before droplet formation. A flow rate ratio of 1:5 (EV / LNP to oil) was selected for the droplet generation process. Droplet squeezing was enabled when the droplets (40 pm) traveled through the squeezing gap (4 pm). Droplets generated contained ~34 picoliter of EVs and LNPs at an optimal ratio of 1 : 1 with an input concentration of 1 — 2 x 1010LNP or EV / mL. Higher concentrations of EVs or LNPs were employed to enhance their interaction during droplet squeezing.103241.007498 / 25-10960 - 25-11237The droplet-generating microfluidic device design was integrated with a dropletsqueezing region, simplifying the process into a single device. The droplets were packed with EVs and LNPs, as seen in FIG. 13f and FIG. 33. After squeezing, the droplets were disrupted, and EV hybrids were collected by phase separation. As expected, the size of EV hybrids increased over 100 nm compared to LNP and over 45 nm compared to EV populations (FIG. 13a). Zeta potential of EV hybrids was less negative (-6 mV) compared to negatively charged EVs (-20 mV), indicating the mixing of EVs and LNPs surface membranes (FIG. 13b). Fusion was also validated from fluorescent microscopy images showing co-localization of AF555 labeled EVs (yellow) and Cy5 labeled LNPs (red) in EV hybrid samples (orange) (FIG. 13d). DLS data reveal that EV hybrids and LNPs had similar average sizes over a month when stored at 4° C, indicating the stability of the EV hybrids generated through our DASH platform (FIG. 13e). However, the PDI of both LNPs and EV hybrids doubled, indicating the presence of other subpopulations of particles.

[0117] We further assessed the stability of EV hybrids via freeze-drying. EV hybrids mixed with 8.5% w / v sucrose were lyophilized overnight and measured for size and surface charge. EV hybrids stored at 4°C showed similar size and PDI compared to hybrids lyophilized and resuspended using 8.5% w / v sucrose (FIG. 23a). In contrast, hybrids without sucrose as cryoprotectant had a 3 -fold increase in size and higher PDI, indicating aggregation. Also, the zeta potential of EV hybrids with sucrose showed a similar surface charge of EV hybrids compared to those stored at 4 °C (FIG.23b). Overall, EV hybrids produced in this method can be lyophilized using cryoprotectants for downstream clinical applications.

[0118] COMSOL simulation of DASH device flow dynamics

[0119] Shear has been shown to induce fusion between lipid bilayers.Mechanical force-induced stress can increase the membrane curvature stress, promoting faster and more efficient mixing and lipid membrane fusion.

[0120] Here, we simulated the flow conditions in our droplet squeezing geometry with COMSOL Multiphysics software to assess the shear rate generated by droplet squeezing. Velocity profiles showed a reduced velocity in the microfluidic chamber compared to the inlet due to flow distribution in a widely opened chamber for squeeze parallelization (FIG. 14a). The shear rate was very high at the squeeze gaps, as103241.007498 / 25-10960 - 25-11237expected in constricted spaces. Most critically, the shear rate increased to6.5 x 105s-1at the squeeze gaps compared to 0.5 x 105s-1outside of the gaps (FIG. 14b). Compared to conventional extrusion shear rate ranging from 100 s-1to a few 1000 s-1, shear rate in microfluidic devices especially at unique geometries like the squeeze gaps employed in this work, is very high. This high shear rate facilitates mechanical perturbations in the EV and LNP particles, including membrane permeability, membrane pore formation, and faster and more efficient mixing that encourage particle fusion. Our simulation shows a higher shear rate of our system compared to other techniques employed for membrane fusion, supporting the hypothesis of enhanced particle fusion phenomenon via droplet squeezing (Table. 1). To note, droplets generated with EVs and LNPs were intact under the high shear rates generated from flow rates used in the experiments.

[0121] Table 1. Shear rates of techniques used for lipid membrane fusion Device type Shear rate generated Effects of shear Droplet squeezing (This >6.2x 105s'1LNP / EV fusion, drug loading work)Couette flow cell >3000 s'1Membrane fusion and deformation Rotational Viscometer 2700 s'1Increase membrane permeabilityExtrusion 100-1000’s s’1Membrane pores, Liposome EV fusionUltra Turrax 400 s'1Increase elastic constant

[0122] Optimization of fusion using droplet squeezing

[0123] To optimize microfluidic device parameters, LNPs were prepared with a fluorescence resonance energy transfer (FRET) pair of 2,l,3-benzoxadiazole-4- yl (NBD) egg liss PE as donor and egg Liss Rhodamine B (RhB) PE as acceptor molecules (FIG. 20a). An increase in the NBD / RhB ratio indicates reduced energy transfer between the FRET pair due to their increased distance from the fusion of the LNP membrane with EVs (FIG. 15a). Compared to LNP, EV hybrids showed a reduction in RhB emission and overall NBD / RhB ratio increase when excited for NBD103241.007498 / 25-10960 - 25-11237fluorescence, confirming successful fusion (FIG. 15b). The fusion activity (NBD / RhB ratio) increased with the number of squeezes and plateaued after 30 squeezes (FIG. 15c). Additionally, the gap length between squeeze units influenced fusion activity, with a 4 pm squeeze gap width showing optimal fusion (FIG. 15d). Interestingly, a lower 2 pm squeeze length did not improve the fusion activity. Different flow rates ranging from 1-8 mL / hr had similar effects on the fusion process (FIG. 15e).

[0124] Using the optimal microfluidic parameters, we compared droplet squeezing to various conventional methods such as extrusion, freeze-thaw, bulk mixing, and other microfluidic geometries. Among different methods, our DASH method showed significant improvement in fusion activity compared to other conventional methods owing to the higher shear rates generated via droplet squeezing (FIG. 15f). Microfluidic chaotic mixing of droplets increased fusion activity compared to plain droplet mixing, indicating the effect of shear rate in droplet mixing (FIG. 15g). Small and large nucleic acid loading into LNPs had a minimal impact on their fusion activity with EVs. But there was a reduction of fusion activity with LNPs loaded with hydrophobic small molecule cargo, possibly due to the interference of small molecules loaded within the hydrophobic lipid membrane (FIG. 15h). The EV:LNP ratio significantly influenced the synthesis of EV hybrids, with an increased proportion of LNPs enhancing the percentage of EV hybrids at EV: LNP ratios of 1 : 1 and 1:10. While small particle flow cytometry showed %EV hybrids plateaued beyond a 1 : 1 ratio, the highest degree of EV colocalization with LNPs was observed at a 1:10 ratio (FIG. 24a-b). Finally, we compared the fusion activity with different EV types to understand the effect of EV surface on fusion using various methods. The DASH method showed higher fusion over other methods for both suspension NK cell-derived sEVs and adherent A431 cell-derived sEVs (FIG. 25). The fusion activity in suspension NK cell-derived sEVs was comparatively higher than adherent A431 cell EVs. Although both EVs have negatively charged surfaces, the lipid composition of suspension cell EVs contains higher amounts of PS and PE lipids,

[0125] *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001

[0126] which can contribute to membrane flexibility, favoring fusion.Meanwhile, adherent cell-derived EVs have higher cholesterol and sphingomyelin content, rendering their membranes rigid.103241.007498 / 25-10960 - 25-11237

[0127] Cryo-Electron Tomography of EV hybrids

[0128] To further assess the EV and LNP fusion, we performed cryo-electron tomography (cryo-ET) on drug-loaded LNPs and NK92 cell-derived EVs individually and after mixing them (to generate LNP -EV hybrids). As part of our analysis, we extensively surveyed our samples using high magnification (33,000X) projection images to capture the distribution of various morphologies and subsequently analyzed a representative subset by cryo-ET. LNPs predominantly displayed classical morphology characterized by uniform, electron-dense spherical structures with sizes ranging from 25 to 100 nm (mostly <50 nm; FIG. 16a). They predominantly exhibited a bounding monolayer with a relatively homogeneous darker central contrast that likely indicates drug encapsulation. It is however worth noting that we occasionally observed a complete or incomplete double-layered bounding membrane around these entities (arrow pairs). In contrast, EVs were predominantly bounded by a distinct membrane bilayer and varied in shape (from spherical to slightly ovoid) and size (mostly 100-300 nm) (FIG. 16b). Their interiors contained a heterogeneous mixture of electron-lucent regions, electron-dense granular inclusions and internal vesicles, consistent with known features of native EVs.

[0129] Cryo-ET imaging of hybrid vesicles revealed several features suggestive of fusion between EVs and LNPs (FIG. 16c). Across multiple tomograms, vesicle populations exhibited fusion interfaces and bilayer continuity between adjacent vesicles (arrowheads). We found numerous instances of a unique encapsulated particle that possessed a dense core, interspersed with smaller electron-lucent regions (last two panels of purple overlays). They often exhibited a double-layered bounding membrane (arrow pair) but at other times, possessed only a single discernible bounding layer (single arrow). Importantly, these particles were not found in either of the individual samples (LNP only or EV only). Together, our observations suggest that we have captured several potential examples of LNP -EV hybrids; their varied morphology likely indicates different kinds or states of fusion. Our results are consistent with fusion-driven integration between synthetic and biologically derived vesicles, likely facilitated by microfluidic mixing or membrane destabilization during droplet squeezing.

[0130] DR5 EV hybrid characterization103241.007498 / 25-10960 - 25-11237

[0131] To characterize DR5 EV hybrids, we evaluated the EV hybrids using small particle flow cytometry and electron microscopy. EV membrane proteins were labeled with NHS-PEG4-AF555 and fused with Cy5-labeled LNPs using our DASH device. Flow cytometry analysis revealed that more than 65% of the detected particles were double positive (AF555+ Cy5+), indicating successful fusion of EVs and LNPs (FIG. 17a). Additionally, DR5 scFv antigens on EVs and EV hybrids were immunostained with DR5 antibody (Ab) and subsequently labeled using secondary Ab conjugated with 10 nm gold nanoparticles for electron microscopy. Transmission electron microscopy (TEM) images of the immunostained EVs demonstrated spherical structures with an average diameter of 150-200 nm. The EVs exhibited a lighter contrast with high-contrast gold nanoparticle staining, confirming the presence of DR5 scFv molecules on their surface (FIG. 17b). In comparison, LNPs appeared as dark, spherical structures with an average size of 80-100 nm. TEM analysis of the EV hybrids showed the fusion of dark LNPs with lighter-shaded EVs displaying gold nanoparticle labeling, further corroborating the successful fusion of EVs and LNPs.

[0132] To assess DR5 scFv functionality in EV hybrids, we stained them with Fab Ab targeting DR5 scFv (FIG. 26a). Double-positive population flow data of EV hybrids generated from labeled individual EVs and LNPs were isolated and plotted against their FITC fluorescence of Fab Ab staining of DR5 scFv molecules (FIG. 26b). Flow data shows significant fluorescence peak shifts of DR5 Ab staining in EVs, and EV hybrids compared to LNPs alone, indicating the presence of functional DR5 scFv (FIG. 26c). Although DR5 scFv presence was seen on EVs and EV hybrids, EVs stained with DR5 Ab showed higher fluorescence than EV hybrids. This may be due to fusion activity reducing the accessibility of some DR5 scFv molecules to the Fab Ab. Small particle flow cytometry and TEM results reveal active DR5 targeting scFv molecules on EV hybrids generated from our DASH device. These results indicate that our platform does not affect the integrity of the EV membrane during fusion and retains functional protein molecules on the EV hybrids which are vital for targeting.

[0133] Target binding affinity of EV hybrids

[0134] DR5 is overexpressed in various cancers, including liver cancer, melanoma, and pancreatic cancer, while its presence in normal tissues is significantly lower.47'49DR5 scFv agonists containing EV hybrid’s binding affinity to DR5103241.007498 / 25-10960 - 25-11237receptors were assessed using A375 melanoma cell lines with DR5 knockout (KO), wild type (WT), and DR5 overexpression (OExp). Labeling with DR5 monoclonal Ab showed various levels of DR5 expression from least (DR5 knockout) to highest (DR5 overexpression) (FIG. 17c). Cell uptake study shows a significantly increased uptake of EVs in WT and OExp compared to KO cells (FIG. 27). In contrast, LNPs showed similar uptake activity in both the KO and WT cells, indicating the affinity of DR5 scFv EVs to target DR5 receptors (FIG. 26). EV hybrids showed higher uptake in WT and OExp compared with KO cells displaying the affinity of the active DR5 scFv targeting molecules on the EV hybrids (FIG. 17d). Z-stacked images were acquired and analyzed using CellProfiler 4 software to assess the colocalization and uptake of EV hybrids (FIG. 28). The custom pipeline analysis considered key factors, including EV and LNP colocalization within each cell, exclusion of large aggregates due to increased frame thickness, and removal of small pixels to minimize false positives. Quantitative data compiled from Z-stacked images of the mid-cell region revealed a significant uptake of EV hybrids in DR5 WT and OExp A375 melanoma cells compared to DR5 KO melanoma cells. (FIG. 17e).

[0135] EV hybrids as a drug delivery system against melanoma

[0136] Given the enhanced affinity of EV hybrids for DR5 -expressing cells, we studied their potential as drug delivery vehicles, specifically for targeting DR5- expressing melanoma cells and delivery of cancer therapeutics. ERK inhibitors are currently used in clinical trials to treat various MAPK pathway-driven cancers such as melanoma, colorectal, thyroid, and lung cancer. Although these drugs are highly potent against cancer cells, they still present toxi cities in clinical studies.Dermatologic adverse effects and other adverse events observed in the clinical trials can be explained by the involvement of ERK 1 / 2 activity across various physiological processes, including those in the gastrointestinal tract and skin. We employed two ERK 1 / 2 inhibitors, ulixertinib and ravoxertinib, as our cargo loaded into the LNPs to enable combinational targeted delivery with EV hybrids. Each hydrophobic drug was mixed with lipids in ethanol and later mixed with PBS (buffer) in a microfluidic staggered herringbone (SHM) chaotic mixer to generate homogeneous drug-loaded LNPs (FIG. 7a). Loading efficiencies for ulixertinib and ravoxertinib were 6.3% and 8%, respectively, with both drug concentrations at ~40 pg / mL in the LNPs, measured103241.007498 / 25-10960 - 25-11237from lysed samples via LC-MS (FIG. 7b). Drug release kinetics of similar hydrophobic drugs from liposomes show release of over 80% of drugs within 2 days. Both drugs have high partition coefficient (ulixertinib = 3.6, ravoxertinib = 2.7) towards octanol, complicating the use of perfluorooctanol (PFO) for droplet disruption and EV hybrid collection. To overcome this challenge, we employed a static gun to disrupt the droplets, effectively preventing drug loss or leakage into PFO.

[0137] To evaluate drug bioavailability following incorporation into LNPs, we assessed the cytotoxicity of drug-loaded LNPs in WT A375 melanoma cells. At a drug concentration of 100 nM, ulixertinib-loaded LNPs exhibited cytotoxic effects comparable to the free drug, whereas ravoxertinib-loaded LNPs showed no significant difference from control LNPs, suggesting that at similar concentrations, ravoxertinib has poor bioavailability when delivered via LNPs (FIG. 7c-d). Based on these cytotoxicity results, ulixertinib-loaded LNPs were selected to prepare EV hybrids for subsequent drug delivery studies. Concerning drug release, previous studies indicate a sustained release of hydrophobic drugs from both LNPs and the hybrids generated from them. Based on the reported data, EV hybrids have a slightly faster drug release compared to LNPs. Systematic exploration of lipid compositions in LNPs and EV membrane composition is needed to understand drug release mechanisms in EV hybrids. In particular, the DASH platform, by enabling controlled fusion of EVs, offers a robust framework for tuning drug release profiles in EV hybrids.

[0138] To determine the optimal dosage for drug delivery and assess biocompatibility, free ulixertinib was given to control BJ fibroblast cells, WT A375 cells, and metastatic melanoma cells with NRAS mutation (WM3000) at varying concentrations. Ulixertinib demonstrated significant cytotoxicity in melanoma cells, reducing cell viability by 50% at 90 nM in WT A375 cells (FIG. 7e) and at 5.8 nM in WM3000 cells (FIG. 7f), while exhibiting minimal cytotoxic effects in BJ cells up to 2000 nM (FIG. 7g). Based on these findings, a concentration range of 100-500 nM ulixertinib was selected for subsequent evaluations of EV hybrids in melanoma- targeted drug delivery. To make drug loaded EV hybrids, ulixertinib-loaded LNPs were fused with DR5 EVs at 1 : 1 ratio using our DASH technique. DLS measurements showed a two-fold increase in the size of LNPs and a change in charge from +8mV to -12m V, indicating a fusion of EVs and drug-loaded LNPs (FIG. 7h). Similarly,103241.007498 / 25-10960 - 25-11237ravoxertinib-loaded LNPs when fused with DR5 EVs showed a change in charge from +6.7mV to -15mV and a three-fold increase in size of LNPs was observed after fusion with EVs, indicating the utility of DASH technique to work with various drugs (FIG.18).

[0139] Therapeutic potential of EV hybrids in 2D and 3D melanoma cell cultures

[0140] ERK inhibitors show promise in melanoma therapy but face toxicity challenges in clinical trials. To mitigate this, we explored EV hybrids for targeted ulixertinib delivery. The treatment efficacy of EV hybrids was studied in 2D and 3D melanoma cell line models. As a critical control to mimic EV hybrids, DR5 Ab conjugated LNPs were prepared using maleimide-thiol chemistry (FIG. 19a). Staining of LNPs with anti-human IgG secondary Ab showed successful DR5 Ab conjugation onto LNPs with an optimal molar ratio of 2.5:1 (DR5 Ab: mal eimide in LNP) (FIG. 19b-c). To evaluate their therapeutic efficacy, ulixertinib-loaded EV hybrids and LNPs were given to both 2D and 3D melanoma cultures with luciferase DR5 overexpressing A375 cells (FIG. 8a). In 2D cultures treated with an equivalent drug concentration of 500 nM across all groups, ulixertinib-loaded EV hybrids demonstrated significantly lower cell viability (10%) compared to drug-loaded LNPs (16%), free drug (22%), and native EVs (77%) (FIG. 8b). Notably, while NK cell- derived sEVs inherently exert an inhibitory effect on melanoma cells, the combination EV hybrids exhibited increased cytotoxicity compared to DR5 Ab coated ulixertinib- loaded LNPs and ulixertinib-loaded EVs (FIG. 20a). This enhanced therapeutic efficacy is likely attributable to a combinatorial approach involving direct ulixertinib- mediated cytotoxicity, selective engagement of DR5 receptors by the DR5 agonist scFv to trigger apoptotic signaling, and the presence of cytotoxic components within the EV lumen — an advantage absent in DR5 Ab LNPs. Supporting this data, prior studies have demonstrated thatNK cell sEVs harbor cytotoxic proteins and intrinsic antitumor properties, potentiating melanoma cell death and modulating the tumor microenvironment.

[0141] Further RNA analysis of EV hybrid-treated 2D cell cultures revealed differential expression of key apoptosis-related genes, including Caspase-8, Caspase-3, STAT-3, BIRC5, Bax and Cyclin DI (FIG. 8c). Gene expression levels were103241.007498 / 25-10960 - 25-11237normalized to the reference gene of RPL13A. Notably, the anti-apoptotic genes such as Cyclin DI and BIRC5 were downregulated following EV hybrid treatment, suggesting an effective apoptosis. This observation aligns with previous studies demonstrating Cyclin DI downregulation upon ulixertinib treatment, indicating ERK pathway inhibition as a contributing mechanism. Consistently, our analysis revealed a significant downregulation of BIRC5 and Cyclin DI across all drug-loaded groups, confirming efficient intracellular drug release from LNPs and EV hybrids (FIG. 20b). Additionally, pro-apoptotic genes, including Caspase-8, Caspase-3, and Bax, were upregulated following EV hybrid treatment, supporting an apoptotic driven mechanism of action. This result is consistent with previous findings where ulixertinib treatment led to increased expression of Caspase-3 and Bax. Note that the STAT3 gene was upregulated across all drug-related groups. A comparable apoptotic gene expression profile was observed across free ulixertinib, drug-loaded LNPs, and EV hybrid treatment, further validating the robustness of our DASH method to generate EV hybrid drug delivery systems (FIG. 20b).

[0142] We further studied the ability of EV hybrids to deliver cargo in a 3D spheroid model composed of WT A375 melanoma cells and BJ fibroblasts. EV hybrids showed significantly lower cell viability (22.8 %) of melanoma cells in the spheroids compared with drug-loaded LNPs (36.08%) or EVs alone (71.3%) (Fig. 8d). Moreover, EV hybrid treatment exhibited significant cytotoxicity against DR5- expressing melanoma cells compared to DR5 Ab-conjugated and ulixertinib-loaded LNPs at a drug concentration of 2.5 pM (FIG. 20c). Reduction in GFP+A375 melanoma cells in spheroid cultures further confirmed cytotoxic effects within the central layers, suggesting effective penetration of EV hybrids into the spheroid structure (FIG. 8e). Propidium Iodide staining of spheroids shows an increased number of dead cells in EV hybrid treatment compared to other groups. These findings highlight the potential of the DASH technique as a powerful platform for generating customizable EV hybrids with superior therapeutic capabilities.

[0143] EV hybrids loaded with ulixertinib drug exhibited enhanced cytotoxicity against A375 melanoma cells compared to LNPs or EVs alone. To delineate the source of this increased efficacy, we compared EV hybrids to a combination treatment of free ulixertinib and DR5 Ab, which engages with TRAIL103241.007498 / 25-10960 - 25-11237pathway to induce apoptosis. Our results demonstrate that EV hybrids outperform this combination, indicating that the therapeutic effect is not solely due to DR5-mediated targeting or drug synergy, but also influenced by the intrinsic bioactive cargo of NK cell-derived sEVs employed (FIG. 21).

[0144] Discussion

[0145] To overcome the challenges of preparing EV hybrids, we provide an approach for utilizing droplet squeezing to facilitate LNP and EV fusion. Our device successfully generated EV hybrids using different EV sources and encapsulated various cargos within the EV hybrids, including plasmids, small RNAs, and drugs. Unlike most strategies reported for EV fusion, the DASH technique eliminates the requirement for pre-processing of EVs with tags or affinity -binding molecules, preserving the integrity of the EV membrane and its internal bioactive components critical for therapeutic applications. Maintaining the pristine nature of EVs, the DASH technique can be extended as a minimally disruptive method for profiling EV cargo. This platform can also be applied to generate various barcoded EV hybrids for EV profiling and screening of EV targets for organ-specific delivery and blood-brain barrier crossing.

[0146] Material and Methods

[0147] Cell Culture

[0148] A431 cell line was purchased from the American Type Culture Collection and CAR-NK cell producing DR5 scFv expressing EVs were previously developed in the lab.29A431 cells were maintained in Dulbecco’s modified Eagle’s medium, supplemented with 10% fetal bovine serum (FBS), 100 IU penicillin, and streptomycin (100 pg / ml) (Pen-Strep). CAR-NK92 cells were maintained in a modified RPMI media containing 10% FBS, 10% horse serum, 0.2 mM myo-inositol, 0.1 mM 2-mercaptoethanol, 0.02 mM folic acid, recombinant human IL-2 (200 U / ml), and 1% penicillin-streptomycin at 37°C and 5%CC>2. A375, WM3000, and BJ cells were cultured in DMEM supplemented with 10% FBS and 1% penicillin.

[0149] EV isolation and characterization

[0150] CAR-NK92 and A431 cells were cultured in the medium as described in previous section, but FBS and horse serum were replaced with 5% EV-depleted serum once cells reached a confluency of 70-80% of cells. After 48 hrs., supernatants103241.007498 / 25-10960 - 25-11237were centrifuged at 2,000g for 15 min to remove cells and debris, followed by centrifugation at 10,000g for 30 min to remove microvesicles. Later, the supernatant was centrifuged at 120,000g for 2 hours at 4°C. Pelleted sEVs were resuspended in PBS and stored at -80°C in aliquots until further experiments.

[0151] After isolating EVs, the samples were analyzed using two methods. The protein concentration was determined with the Qubit system (Thermo Fisher), following the manufacturer’s protocol with the Qubit protein assay kit. Additionally, the particle count was assessed using nanoparticle tracking analysis (NTA). The NTA measurements were performed using the ZetaView PMX220 Twin instrument (Particle Metrix) at the Extracellular Vesicle Core of the University of Pennsylvania School of Veterinary Medicine. The analysis was conducted with a sensitivity setting of 70.

[0152] Western blot

[0153] NK92 cells or CAR NK92 sEVs were lysed in radioimmunoprecipitation assay (RIP A) buffer with Protease inhibitor cocktail. A total of 10 to 20 pg of protein was processed for SDS-PAGE (polyacrylamide gel electrophoresis) and electroblotted onto poly vinylidene difluoride (PVDF) membranes (Invitrogen, USA). Blots were blocked with 5% nonfat milk and incubated with corresponding primary antibodies (CD9, CD81, Calnexin, and TSG101) and horseradish peroxidase (HRP)-conjugated secondary antibodies (BioLegend, USA). Membranes were developed using enhanced chemiluminescence (ECL) detection reagents (Pierce, Thermo Scientific, MA, USA).

[0154] Antibodies

[0155] DR5 monoclonal antibodies (tigatuzumab) were manufactured by Biointron Biological USA Inc. (Metuchen, NJ, USA). The molar ratios and number of antibody molecules needed for LNP conjugation was calculated using the surface area of the LNP and Avogadro’s number. Western blot antibodies of CD9 (BioLegend, Cat #31202), CD81 (BioLegend, Cat #349502) Calnexin (BioLegend, Cat #699401), and TSG101 (BioLegend, Cat #934302)

[0156] 3D spheroid formation

[0157] 50 pl of 1.5% agarose solution was used in coating 96-well plates before seeding cells. After Ihr incubation, WT A375 melanoma cells (20,000 / well)103241.007498 / 25-10960 - 25-11237and BJ cells (40,000 / well) were mixed and seeded. This cell mixture was incubated for 48 hours for spheroids to form.

[0158] LNP synthesis

[0159] Lipids for LNP synthesis were purchased from Avanti polar lipid (USA). 1,2, -dioleoyl-3-trimethylammonium-propane (chloride salt) (DOTAP), 1,2- dipalmitoyl-sn-glycero-3-phosphocholine (DPPC), Cholesterol and 1,2-distearoyl-sn- glycero-3-phosphoethanolamine-N-[amino(polyethyleneglycol)-2000] (DSPE- PEG2000) lipids (in molar ratio 40:40:10:10) were dissolved at 10 mg / mL in ethanol. For FRET LNPs, Egg liss NBD and Rhodamine B PE lipids were added at a molar ratio of 0.5 each. Cy5 labeled PC lipids at 0.05% molar ratio were added to lipid mixture for fluorescent labeling of LNPs. PBS buffer was used as an aqueous phase for the formulation. A staggered herringbone micromixer generated LNPs with a flow rate ratio of 1 :3 (Lipid in ethanol: aqueous PBS).72Later, the LNPs were dialyzed against PBS to remove ethanol. LNPs were then characterized for their size and zeta potential using DLS (Dynamic Light Scattering) and Nanoparticle Tracking Analysis.

[0160] Fabrication of droplet squeezer

[0161] The device was initially designed using AutoCAD (USA) software, and a 50 pm height SU-8 master mold was fabricated through conventional photolithography. A droplet generator was created with two sample inlets and one oil inlet, followed by short mixing unit and 40 rows of droplet squeezing structures, each row had 10 squeeze units with a gap of 4 pm. To construct the droplet squeezer, a Polydimethyl siloxane (PDMS; Coming, USA) mixture, prepared at a 10:1 base-to- curing agent ratio, was poured over the master mold and cured at 65°C overnight. Once cured, the PDMS slab was removed, cut to the desired dimensions, and perforated with 1 mm holes using a biopsy punch (Miltex, USA) to create fluidic ports. The PDMS device was then plasma bonded to glass slides and cured at 65°C overnight to ensure a leak-proof seal.

[0162] Droplet generation and optimization of droplet squeezing device

[0163] We generated droplets to optimize the droplet generator device concerning squeeze number. EVs and LNPs at 1 x 1010partici e / mL each were pumped in with both at a flow rate of 0.4 mL / hr and oil at 2.5 mL / hr forming droplets. Sequentially, we squeezed the droplets with EV and LNPs encapsulated through103241.007498 / 25-10960 - 25-11237devices with 10, 20, 30, and 40 squeezing rows with a total flow rate of 3.3 mL / hr. Similarly, we generated various devices with a squeezing gap of 2, 4, 8 & 10 pm and ran droplets across these devices to assess fusion activity between EVs and LNPs. Using 40 squeezes and a squeeze channel length of 4 pm, we evaluated the effect of flow using the total flow rates ranging from 1-8 mL / hr.

[0164] COMSOL simulations

[0165] The droplet generator’s AutoCAD design file was imported into COMSOL Multiphysics software and extruded 50 pm to mimic the device. Both inlet and outlet boundaries are defined with pressure at zero and initial velocity equal to the experimental flow rate velocity of 4.37 X 10-11m3 / s (3.3 mL / hr). Study solutions were computed with a finer mesh setting. The most critical shear rate parameter was calculated using the data plots generated from the study solution of velocity and pressure plots.

[0166] Fluorescent imaging of EV hybrids

[0167] To determine the fusion via fluorescence, EVs, and LNPs were stained with different fluorophore molecules. EV surface proteins were stained with NHS-PEG-AF555 dye, and LNPs were stained by incorporating Cy5-DPPC lipid during formulation. Excess NHS-PEG-AF555 was removed using a 40 kDa Zeba column (Thermo Fischer Scientific). EV hybrids generated from droplet squeezing were imaged in AF555 (TRITC) and Cy5 channels using an 1X83 inverted fluorescence microscope with a 40x objective (Olympus, Japan).

[0168] Cryo-electron tomography (Cryo-ET)

[0169] Sample preparation for Cryo-ET : Samples were plunge frozen using a Leica EM GP2 (Leica Microsystems, Wetzlar, US) using liquid ethane-propane- mixture as the cryogen. The cryogen was in turn cooled by liquid nitrogen and held at - 180 C. 4 pL of each sample was added to EM grids (Quantifoil R2 / 2 carbon membrane on Cu 200 mesh), blotted for 3-5 sec from the front and quickly plunged into the cryogen. Sample concentrations: LNP —3 X 1010particles / mL; EV —3 X 1010particles / mL; and hybrids: 5 X 1010particles / mL at LNP:EV ratio of 1:1 particles.

[0170] Cryo-ET Image Acquisition: Cryo-ET was performed on a ThermoFisher Krios G3i 300keV field emission gun cryo-TEM as mentioned previously73. Images were collected using the SerialEM software on a K3 direct103241.007498 / 25-10960 - 25-11237electron detector (Gatan Inc., Pleasanton, CA, USA) that operated in electron-counted mode. After initially assessing the samples at lower magnifications for suitability of ice thickness and lipid membrane integrity, tilt series were collected with a span of 120° (-60° to +60°; dose-symmetric scheme) with 2° increments at a magnification of 33,000X (with a corresponding pixel size of 2.67 A) and a defocus value of -5 pm. The cumulative dose of each tilt series was close to 100 e“ / A2. The Gatan Imaging Filter (Gatan Inc., Pleasanton, CA, USA) operated with a slit width of 20 eV helped to increase contrast by removing inelastically scattered electrons (Krivanek et al). Once acquired, tilt series were aligned and reconstructed into tomograms using the IMOD software package. Slices through tomograms were used to analyze LNP fusion with EVs and to generate the representative images included in this manuscript.

[0171] FRET studies for fusion activity

[0172] Fusion activity was studied using Fluorescence Resonance Energy transfer (FRET). Briefly, FRET pair of 0.5 mol% each of NBD-PE (donor) and Rhodamine B-PE (acceptor) lipids were incorporated in the lipid mixture. LNPs formulated with these lipids were employed to perform FRET studies. FRET pairincorporated LNPs were fused with NK or A431 cell EVs using our droplet squeezing technique. The fluorescence signal from the fused vesicles was measured using a spectrophotometer (TEC AN Spark Control, Switzerland). The fusion activity was assessed from the ratio of NBD / RhB fluorescence. For this measurement, the samples were excited at 460 nm and emissions were recorded at 530 nm and 585 nm. The increase in NBD / RhB ratio directly correlates with an increased fusion of LNPs and EVs.

[0173] Conventional techniques used for the synthesis of EV hybrids

[0174] Various conventional EV hybrid generation techniques were employed for comparing fusion against droplet squeezing. Bulk mixing included mixing 1 X 1010particles / mL equal ratios of EVs and LNPs and incubating the sample at 37°C for 1 hour at 700rpm mixing. Similarly, avanti mini extruder was employed to fuse EVs and LNPs with a 0.2 pm Nuclepore Track-Etch membrane (Cytiva). A minimum of 15 rounds of extrusion was applied to generate EV hybrids from 1:1 ratio of EVUNPs at 1 x 1010particles / ml each. For the freeze-thaw method, we mixed EVs and LNPs in equal ratios (1 X 1010particles / ml) in a falcon103241.007498 / 25-10960 - 25-11237tube and plunged into liquid nitrogen for 30 seconds, followed by a 15-minute thaw at room temperature. At least five cycles of freeze-thaw were performed.

[0175] Cargo loading of EV hybrids

[0176] For nucleic acid loading, EGFP plasmid or random siRNA were mixed with 50mM citrate buffer and lipid ethanol solution in a microfluidic mixer with staggered herring bone geometry. To remove ethanol, LNPs loaded with nucleic acids were dialyzed overnight with a membrane molecular weight cutoff of 300kD. For small molecule loading, chemo drugs were mixed in ethanol while preparing a lipid mixture. Later, the drug / lipid solution and PBS were mixed in a microfluidic mixer to generate drug-loaded LNPs. LNPs were dialyzed against the 300kD molecular weight cutoff membrane to remove ethanol and unloaded drug.

[0177] Nano-flow cytometry

[0178] NHS-PEG-AF555-stained EVs and cy5-labeled LNPs were used to generate EV hybrids through droplet squeezing. EV hybrids, along with single stain and unstained controls, were analyzed using nano-flow cytometry (NanoFCM, Inc., Xiamen, China) from the University of Pennsylvania School of Veterinary Medicine Extracellular Vesicle Core. A similar BD cytoFLEX nano flow cytometer was employed to assess EV hybrids with dual staining from EV and LNPs. All samples were diluted to achieve a particle count with the range between 2000-12000 / min to remove any swarming effects. Total of 50,000 events were collected, and dot plots were generated using NanoFCM software (NanoFCm Profession VI.0) and Flow Jo software (USA).

[0179] Cell immunostaining

[0180] To determine the DR5 receptor expression in A375 melanoma cell lines, immunostaining with DR5 Ab and secondary fluorescent Ab was employed. Briefly, DR5 knockout, wild type, and DR5 overexpressing A375 melanoma cells were stained with DR5 monoclonal antibody after blocking with bovine serum albumin. Later cells were washed and secondary goat anti-human IgG (Invitrogen) tagged with fluorophore was used for staining the cells. Then, cells were washed 3X PBS and ran in BD LSR II (Penn cytomics and cell sorting laboratory, University of Pennsylvania). Flow cytometry data was analyzed using Flowjo software (Flowjo LLC, USA)103241.007498 / 25-10960 - 25-11237

[0181] Immunoelectron microscopy of EV hybrids

[0182] For immunoelectron microscopy, EVs or EV hybrids suspended in PBS were deposited onto formvar carbon-coated nickel grids, then blocked and incubated with F(ab')2 Fragment Goat Anti-Human IgG. This was followed by incubation with gold-conjugated goat anti-biotin (10 nm, electron microscopy grade) from Electron Microscopy Sciences (PA, USA). After each staining step, the samples underwent five PBS washes before being contrast-stained with 2% uranyl acetate. A similar staining protocol was performed for LNPs, excluding the F(ab')2 antibody staining. The EVs, LNPs and EV hybrids were then imaged using a JEM-1011 transmission electron microscope.

[0183] EV hybrid uptake studies

[0184] EVs, LNPs, and EV hybrids were given to A375 melanoma cell lines with varying DR5 expression levels, including knockout, wild-type, and overexpression variants. EVs were labeled with NHS-PEG-AF555, while LNPs were labeled with Cy5. For this experiment, cells were seeded at confluency in an 8- chambered cover glass system (Cellvis, USA). The cells were treated with 1 x 1010particles / mL of either EVs, LNPs, or EV hybrids. After a 4-hour incubation, cells were washed three times with PBS, stained with DAPI (Invitrogen, USA) for nuclear localization and CellTracker Green (Invitrogen, USA) for cell boundary visualization. Fluorescent images were captured using an 1X83 inverted fluorescence microscope (Olympus, Japan), focusing on the mid-section of cells with nuclei in focus to analyze internalized EV hybrids.

[0185] Cellular uptake images were processed using a custom-designed CellProfiler pipeline. Briefly, image intensities were rescaled, and cell boundaries were identified based on CellTracker Green staining. EVs and LNPs were segmented based on pixel size and granularity. Colocalization of EVs and LNPs was determined using the RelateObjects module, which pairs identified particles based on spatial proximity. Once colocalized EVs and LNPs were identified, these particles were overlaid onto the cell mask derived from boundary identification via cell tracker green stain. The number of internalized particles per cell was quantified, and the average particle count per cell across three images was calculated and plotted.

[0186] LCMS protocol for LNP drug concentration measurement103241.007498 / 25-10960 - 25-11237

[0187] Ulixertinib concentrations were determined by LCMS analysis of ethanol-lysed samples. Ulixertinib calibration curves were generated by running samples containing 10 ng / ml to 100 pg / ml and then integrating across the 254nm absorbance ulixertinib peak. The calibration curve was measured every 6 months to account for drift. Samples were analyzed using a 1260 Agilent Infinity II HPLC system equipped with an SB Cl 8 2.1 X 50 mm, 1.8 mm column with a mobile phase consisting of H2O (solvent A), acetonitrile (solvent B) and a gradient composed of 5% B at 0 min to 95% B at 10 min, then to 5% B at 12 min with a flow rate of 0.4 mL / min. Experimental samples were run under identical conditions; the 254nm UV absorbance was integrated across the ulixertinib peak and compared to calibration curves to determine the experimental concentration. Known concentrations of LNPs were lysed with 100% ethanol and measured by LCMS to find drug concentrations of ulixertinib loaded inside the LNPs. Similar protocol was employed to measure ravoxertinib concentrations. Loading efficiencies were calculated using the below equation.Drug in lysed mg of LNPsLoading efficiency = — - — - — - - — ; — X 100T otal drug used for formulation

[0188] Lyophilization studies of EV hybrids

[0189] EV hybrids generated from DASH techniques were stored at 4°C. EV hybrids from the stored stock solution were divided into two groups, one without any cryoprotectant and the other mixed with 8.5% w / v sucrose. Both groups were lyophilized using a Labconco FreeZone 2.5 freeze-drying system. Samples were collected after overnight lyophilization. Later, similar initial volumes of PBS were used to resuspend lyophilized sample powders. The resuspended lyophilized samples’ size and zeta potential were measured using a DLS.

[0190] Drug dosage studies

[0191] Two ERK inhibitor drugs, Ulixertinib (BVD-523) and Ravoxertinib (GDC-0944), which have shown potency against melanoma, were employed to load into LNPs. Both hydrophobic drugs were mixed with lipids in ethanol and were injected into microfluidic chaotic mixer along with PBS to generate drug-loaded LNPs. LNPs were dialyzed against lOkD molecular weight cutoff dialysis membrane overnight. Luciferase positive A375 cells at 20,000 cells / well were seeded without FBS until attached. Later LNPs and free drug of ulixertinib and ravoxertinib were103241.007498 / 25-10960 - 25-11237given to A375 melanoma cells with a DMEM media concentration of lOOnM in 1% FBS and 1% pen-strep. After 3 days, cells were washed with PBS and lysed with a lysis reagent of 45 pL. Luciferase assay reagent (100 pL per well) was added to each well and read using a TEC AN SPARK spectrophotometer (Luciferase Assay system, Promega).

[0192] Antibody conjugated LNP preparation

[0193] As previously described, DR5 monoclonal antibodies were conjugated to the surface of lipid nanoparticles (LNPs) using maleimide-thiol click chemistry. Briefly, ulixertinib-loaded LNPs were synthesized, incorporating DSPE- PEG5k-Mal (0.5 mol%) into the lipids. DR5 monoclonal Abs were functionalized by reacting with SATA (N-succinimidyl S-acetylthioacetate) in DMSO (15 mg / mL). This reaction targeted primary amines in the antibody. This reaction was carried out in PBS containing 1 mM EDTA at a protein concentration of 10 mg / mL. The reaction was incubated at room temperature for 1 hour, after which unreacted SATA was removed using Zeba desalting columns (7 kDa molecular weight cutoff, Thermo Fisher).

[0194] To deprotect the sulfhydryl (-SH) groups, a 10% (v / v) hydroxylamine solution was added, containing 50 mM sodium phosphate, 25 mM EDTA, and 0.5 M hydroxylamine HC1. The mixture was incubated at room temperature for 2 hours, followed by another desalting step to remove excess reagents. The thiolated Abs were mixed with maleimide-functionalized LNPs at a 2.5-fold molar excess (Ab: maleimide) and incubated at room temperature for 2 hours. Conjugation was completed by overnight dialysis against PBS using a 300 kDa molecular weight cutoff dialysis membrane to remove unbound Abs and reaction byproducts.

[0195] In vitro cancer killing assay

[0196] LNPs loaded with ulixertinib were fused with DR5 scFv expressing EVs derived from CAR-NK92 cells to generate ulixertinib-loaded EV-LNP hybrids. EVs were loaded with ulixertinib via sonication at 37°C for 30 minutes. Luciferase WT A375 melanoma cells were seeded at a density of 40,000 cells per well in a 96- well plate and maintained in serum-free conditions until attachment. Both 2D cell cultures (in 96-well plates) and 3D spheroid models were treated with ulixertinib- loaded EV hybrids with control groups including EVs, ulixertinib-loaded EVs, ulixertinib-loaded LNPs, and DR5 Ab-conjugated ulixertinib-loaded LNPs. An103241.007498 / 25-10960 - 25-11237equivalent drug concentration of 500 nM was maintained across all groups. The drug concentration in LNPs, EVs, and EV hybrids was quantified by lysing the particles with 2% Triton X-100, followed by LCMS analysis. After 72 hours, 2D and 3D cultures were lysed using a lysis reagent and mixed with a luciferase reagent to measure melanoma cell luciferase activity using a TECAN SPARK plate reader (Luciferase Assay System, Promega). For 3D spheroid assays, spheroids were carefully transferred to 96-well plates via pipetting to minimize contamination from the agarose gel layer. For the LIVE / DEAD imaging of spheroids, after 3 days, GFP+melanoma cells within the spheroid structures were subjected to imaging across various experimental groups to assess live cell populations. To differentiate dead cells, propidium iodide was incorporated into the culture medium at a concentration of 10 pg / mL, facilitating the visualization of necrotic and apoptotic cells.

[0197] For determining baseline toxicity of free ulixertinib and DR5 Ab, we performed a cytotoxicity study in A375 melanoma cell lines. DR5 Ab, free ulixertinib, free ulixertinib + DR5 Ab, ulixertinib loaded EV hybrids with an equivalent drug concentration of 100 nM ulixertinib were given to A375 cells (40,000 cells / well seeded in 1% FBS). DR5 Ab dose was calculated based on the average number of DR5 scFV from TEM images of EVs. After 3 days, cells were washed, lysed and treated with luciferase reagent. Later, luminescence was captured using a plate reader (TECAN).

[0198] RNA was extracted using TRIzol reagent (Invitrogen) to evaluate gene expression following treatment in 2D cell experiments. Complementary DNA (cDNA) was synthesized using the i Script cDNA synthesis kit (Bio-Rad). Apoptosis- related genes, including Caspase-8, Caspase-3, STAT-3, Bax, BIRC5, and Cyclin DI, were analyzed to assess the effects of EV hybrid treatment (primers listed in Table 2). A housekeeping gene of RPL13A was employed to calculate the fold change expression of various genes. cDNA and gene-specific primers were combined with PowerTrack SYBR Green Master Mix (Applied Biosystems; A46109), nuclease-free water (Integrated DNA Technologies) for gene expression studies. Quantitative PCR (qPCR) was performed in 10 pL reactions, running for 40 cycles on a QuantStudio 3 Real-Time PCR system (Applied Biosystems).

[0199] Table 2. Primers for qPCR103241.007498 / 25-10960 - 25-11237Genes Forward primer (5 ’-3’) Reverse primer (5 ’-3’) RPL13A CTCAAGGTGTTTGACGGCATCC TACTTCCAGCCAACCTCGTGAG Caspase-8 AGAAGAGGGTCATCCTGGGAGA TCAGGACTTCCTTCAAGGCTGC Caspase-3 G G AAG CG AATCAATG G ACTCTG G GCATCGACATCTGTACCAGACC STAT-3 CTTTGAGACCGAGGTGTATCACC GGTCAGCATGTTGTACCACAGG BIRC5 CCACTGAGAACGAGCCAGACTT GTATTACAGGCGTAAGCCACCG Bax TCAGGATGCGTCCACCAAGAAG TGTGTCCACGGCGGCAATCATC Cyclin DI GATGCCAACCTCCTCAACGA ACTTCTGTTCCTCGCAGACC

[0200] Aspects

[0201] The following Aspects are illustrative only and do not limit the scope of the present disclosure or the appended claims. Any part or parts of any one or more Aspects can be combined with any part or parts of any one or more other Aspects.

[0202] Aspect 1. A system, comprising: a mechanoporation region in fluid communication with an inlet configured to receive at least one of a supply of cells and a supply of carriers having nucleic acid disposed therein, the mechanoporation region defining a direction of flow, the mechanoporation region defining a set of first gaps and a set of second gaps, the second gaps being arranged downstream relative to the first gaps along the direction of flow, a gap being sized so as to effect perturbation of the membrane of a cell encouraged through the gap, the perturbation being effective to enhance passage of a carrier through the membrane, a gap being defined between two protrusions, a protrusion having length measured in the direction of flow and a protrusion having a cross-sectional dimension measured along the length of the protrusion and perpendicular to the direction of flow, the cross-sectional dimension being free of reduction along the direction of flow. As shown in, for example, FIG.IB, a gap can be sized so as to effect perturbation of the membrane (i.e., the cell membrane) of a cell encouraged through a gap.

[0203] Aspect 2. The system of Aspect 1, wherein the carrier comprises a lipid nanoparticle (LNP). Example, non-limiting LNPs are provided herein.

[0204] Aspect 3. The system of any one of Aspects 1-2, wherein the nucleic acid comprises any one or more of siRNA, mRNA, or pDNA. pDNA is considered103241.007498 / 25-10960 - 25-11237particularly suitable, but the disclosed systems can be used with other nucleic acids and are not limited to use with pDNA.

[0205] Aspect 4. The system of Aspect 3, wherein the nucleic acid comprises pDNA.

[0206] Aspect 5. The system of any one of Aspects 1-4, wherein a protrusion is polygonal in cross-section.

[0207] Aspect 6. The system of any one of Aspects 1-5, wherein a protrusion is asymmetric about a line drawn perpendicular to the direction of flow. One such example is the protrusion shown in FIG. 11, which depicts a protrusion that is roughly triangular in shape and broadens as one moves in the direction of flow from the rounded apex of the protrusion. As shown, the protrusion is asymmetric about a line drawn perpendicular to the direction of flow, i.e., the protrusion includes a region that broadens as one moves in the direction of flow from the rounded apex of the protrusion but the protrusion does not also include a corresponding region that narrows in the direction of flow, beginning from the broadest part of the protrusion.

[0208] Aspect 7. The system of any one of Aspects 1-6, where a protrusion comprises a region of increasing cross-sectional dimension along the direction of flow and a region of constant cross-sectional dimension along the direction of flow. Such a protrusion is shown in FIG. 11; as shown, the protrusion includes a region of increasing cross-sectional dimension along the direction of flow and a region of constant cross-sectional dimension along the direction of flow.

[0209] Aspect 8. The system of any one of Aspects 1-7, wherein a first gap is in at least partial register with a second gap along the direction of flow. This is shown in FIG. 11, in which the labeled first gap is in register with the labeled second gap, in the direction of flow.

[0210] Aspect 9. The system of any one of Aspects 1-7, wherein a first gap is not in at least partial register with a second gap along the direction of flow.

[0211] Aspect 10. The system of any one of Aspects 1-9, wherein an inlet configured to receive the supply of cells and an inlet configured to receive the supply of carriers having nucleic acid disposed therein. Such an arrangement is shown in FIG. 11, which depicts an inlet that receives a supply of cells and also depicts an inlet that receives a supply of carriers.103241.007498 / 25-10960 - 25-11237

[0212] Aspect 11. The system of any one of Embodiments 1-10, wherein the first set of gaps is arranged in a row and the second set of gaps is arranged in a row. This is shown in FIG. 11, which depicts first gaps arranged in a row and second gaps arranged in a row. As shown, rows of gaps can be parallel to one another.

[0213] Aspect 12. A system, comprising: a mechanoporation region in fluid communication with an inlet configured to receive at least one of a supply of cells and a supply of carriers having nucleic acid disposed there, the mechanoporation region defining a direction of flow, the mechanoporation region defining n sets of gaps arranged in the direction of flow, wherein n is from 6 to 12 and wherein an (n+l)th gap is downstream relative to an nth gap along the direction of flow, a gap being sized so as to effect perturbation of the membrane of a cell encouraged through the gap, the perturbation being effective to promote fusion of a carrier to the membrane, a gap being defined between two protrusions.

[0214] Aspect 13. The system of Aspect 12, wherein, each of the n sets of gaps is arranged in a row, and wherein an (n+l)th row of gaps is arranged parallel to an nth row of gaps. This is shown in FIG. 11, in which the second set of gaps is arranged in a row (i.e., the row of second gaps), and the first set of gaps is arranged in a row (i.e., the row of first gaps), with the second row of gaps being downstream relative to the first row of gaps along the direction of flow.

[0215] Aspect 14. A method, comprising: encouraging a cell having a diameter through at least a first gap having a cross-sectional dimension smaller than the diameter of the cell, the encouraging mechanically perturbing the membrane of a cell so as to promote fusion to the membrane by a carrier having a nucleic acid disposed therein, and the first gap optionally having a width in the range of from about 15% to about 30% of diameter of the cell.

[0216] The encouraging can be accomplished by, for example, flowing the cell through the gap. An example, non-limiting depiction is provided in FIG. IB, which depicts flowing cells through gaps, which gaps are defined between protrusions in a flow channel. FIG. 1C provides an example, non-limiting depiction of the described technology; as shown, a cell can be deformed by being squeezed through a cap, which in turn perturbs the cell membrane in a way that promotes fusion by a103241.007498 / 25-10960 - 25-11237carrier - in this instance, a LNP - to the cell membrane. The carrier is then delivered to the interior of the cell by endocytosis.

[0217] Aspect 15. The method of Aspect 14, wherein the carrier is delivered into the cell by endocytosis.

[0218] Aspect 16. The method of any one of Aspects 14-15, wherein the carrier comprises a lipid nanoparticle. Example, non-limiting LNPs are provided elsewhere herein. In some Aspects, the LNP comprises cholesterol, 1,2-dimyristoyl- sn-glycero-3-phosphoethanolamine-N-[methoxy(polyethylene glycol)-2000] (ammonium) (C14-PEG 2000) and any one or more of C12-200, l,2-dioleoyl-3- dimethylammonium-propane (DODAP), DLin-MC3-DMA, l,2-dioleoyl-3- trimethylammonium-propane (DOTAP), l,2-dioleoyl-sn-glycero-3- phosphoethanolamine (DOPE). The foregoing species are illustrative only and should not be understood as limiting the scope of the present disclosure or the appended claims.

[0219] Aspect 17. The method of any one of Aspects 14-16, wherein the nucleic acid disposed within the carrier can be, for example, any one or more of mRNA, siRNA, or pDNA. pDNA is considered particularly suitable.

[0220] Aspect 18. The method of any one of Aspects 14-17, wherein the cell is a T cell. Without being bound to any particular theory or Aspect, T cells are considered particularly suitable for use in the disclosed methods.

[0221] Aspect 19. The method of any one of Aspects 14-18, wherein the gap is defined between two protrusions. Example gaps and protrusions are provided in non-limiting FIG. IB.

[0222] Aspect 20. The method of any one of Aspects 14-19, wherein the lipid nanoparticle comprises cholesterol, l,2-dimyristoyl-sn-glycero-3- phosphoethanolamine-N-[methoxy(polyethylene glycol)-2000] (ammonium) (C14- PEG 2000), and any one or more of C12-200, l,2-dioleoyl-3-dimethylammonium- propane (DODAP), DLin-MC3-DMA, l,2-dioleoyl-3-trimethylammonium-propane (DOTAP), l,2-dioleoyl-sn-glycero-3 -phosphoethanolamine (DOPE).

[0223] Aspect 21. A method, comprising: subjecting a droplet comprising an extracellular vesicle (EV) and a lipid nanoparticle (LNP) to shearing sufficient to effect fusion between the EV and LNP and form an EV hybrid.103241.007498 / 25-10960 - 25-11237

[0224] An LNP can have a cross-sectional dimension - such as a diameter - of, for example, up to about 1 pm.

[0225] Without being bound to any particular theory or embodiment, one can use PEGylated cationic lipids (DOTAP, DPPC, DSPE-PEG) to modulate LNP charge and fusion compatibility. One can tune the input ratios of EVs and LNPs, thereby enabling control over composition.

[0226] An EV can have a cross-sectional dimension - such as a diameter - of, for example, up to about 250 nm. So-called small EVs of sEVs are considered particularly suitable; such sEVs can have a size of from about 50 nm up to about 250 nm, such as about 150 nm. EVs from plants, mammals, or other sources are all considered suitable.

[0227] Without being bound to any particular theory or embodiment, EV hybrids formed using the disclosed technology exhibited favorable stability, for example, stability for at least one month at 4°C without significant size change. The disclosed technology can be used to form lyophilization-compatible EV hybrids, which can in turn enhance their clinical potential. As described herein, fusion can be confirmed by zeta potential shifts, increase in particle diameter, and electron microscopy.

[0228] Aspect 22. The method of Aspect 21, wherein the subjecting is performed in a fluid channel that defines a direction of flow, the fluid channel optionally defined between features that project from a substrate. A fluid channel can include a portion that has a width of from about 5% to about 99% of the width of the droplet. In some embodiments, the width is about 10% of the width of the droplet.

[0229] Aspect 23. The method of Aspect 22, wherein the fluid channel comprises a constriction region having a cross-sectional area that varies along the direction of flow, the cross-sectional area optionally diminishing along the direction of flow.

[0230] Aspect 24. The method of any one of Aspects 22-23, wherein the fluid channel comprises a region having a cross-sectional area that remains constant long the direction of flow.

[0231] Aspect 25. The method of Aspect 24, wherein the region having a cross-sectional area that remains constant along the direction of flow has a width that103241.007498 / 25-10960 - 25-11237is from 5% to 99% of a diameter of the droplet, optionally from 10% to 25% of a diameter of the droplet.

[0232] Aspect 26. The method of any one of Aspects 22-25, wherein the fluid channel defines a region having a width of from about 0.1 to about 10 pm, optionally from about 2 to about 8 pm.

[0233] Aspect 27. The method of any one of Aspects 21-26, wherein the shearing comprises a shear rate of from about 1 x 105s'1to about 10 x 105s’1, optionally from about 1 x 105s’1to about 9 x 105s’1.

[0234] Aspect 28. The method of any one of Aspects 21-27, wherein the shearing comprises subjecting the extracellular vesicle (EV) and the lipid nanoparticle (LNP) to a plurality of shearing cycles.

[0235] Aspect 29. The method of Aspect 8, wherein the plurality of shearing cycles comprises from 2 to 50 cycles, optionally from 10 to 40 cycles.

[0236] Aspect 30. The method of any one of Aspects 21-29, wherein at least one of the EV and the LNP comprises a targeting moiety, the targeting moiety optionally comprises any one or more of a receptor, a receptor fragment, a ligand, a ligand fragment, an antibody, and an antibody fragment.

[0237] Aspect 31. The method of any one of Aspects 21-30, wherein at least one of the EV and the LNP comprises a drug or a nucleic acid disposed therein or thereon, the drug optionally comprising a kinase inhibitor.

[0238] Aspect 32. The method of any one of Aspects 21-31, further comprising placing the EV and the LNP in the droplet, optionally wherein the placing and the subjecting are performed within a single system.

[0239] Aspect 33. A method, comprising: subjecting a droplet comprising an extracellular vesicle (EV) and a lipid nanoparticle (LNP) to shearing sufficient to effect fusion between the EV and LNP and form an EV hybrid, the subjecting comprising encouraging the droplet through a fluid channel defined between features that project from a substrate, the fluid channel defining a direction of flow.

[0240] Aspect 34. The method of Aspect 33, wherein the shearing comprises a shear rate of from about 1 x 105s’1to about 10 x 105s’1, optionally from about 1 x 105s’1to about 9 x 105s’1.103241.007498 / 25-10960 - 25-11237

[0241] Aspect 35. The method of any one of Aspects 33-34, wherein the shearing comprises subjecting the extracellular vesicle (EV) and the lipid nanoparticle (LNP) to a plurality of shearing cycles.

[0242] Aspect 36. The method of Aspect 35, wherein the plurality of shearing cycles comprises from 2 to 50 cycles, optionally from 10 to 40 cycles

[0243] Aspect 37. The method of any one of Aspects 33-36, wherein the fluid channel comprises a region having a cross-sectional area that remains constant along the direction of flow, the region having a width that is from 5% to 99% of a diameter of the droplet, optionally from 10% to 25% of a diameter of the droplet.

[0244] Aspect 38. A method, comprising: subjecting a droplet comprising an extracellular vesicle (EV) and a lipid nanoparticle (LNP) to shearing sufficient to effect fusion between the EV and LNP and form an EV hybrid, the subjecting comprising encouraging the droplet through a fluid channel defined between features that project from a substrate.

[0245] Aspect 39. A system, comprising: a droplet generation stage, the droplet generation stage configured to form droplets having EVs and LNPs disposed therein; and a droplet squeezing stage, the droplet squeezing stage configured to (1) receive droplets from the droplet generation stage, and (2) subject received droplets to shearing sufficient to effect fusion between the EV and the LNP of a droplet so as to form an EV hybrid.

[0246] As described herein, a system can include a microfluidic chip that includes a flow-focusing region to create droplets and a downstream section; the downstream section can include one or more rows of structured squeeze units, with each row comprising constriction gaps. As the droplet traverses these gaps, the encapsulated EVs and LNPs are brought into proximity under shear, enabling membrane fusion.

[0247] Aspect 40. The system of Aspect 39, wherein the droplet generation stage comprises at least one of (1) an inlet configured to receive EVs and (2) an inlet configured to receive LNPs.

[0248] Aspect 41. The system of Aspect 39, wherein the droplet generation stage comprises at least one of (1) a stage configured to generate EVs and (2) a stage configured to generate LNPs.103241.007498 / 25-10960 - 25-11237

[0249] Aspect 42. The system of any one of Aspects 39-41, wherein the droplet squeezing stage comprises least one fluid channel that defines a direction of flow.

[0250] Aspect 43. The system of Aspect 42, wherein the fluid channel comprises a constriction region having a cross-sectional area that varies along the direction of flow, the cross-sectional area optionally diminishing along the direction of flow.

[0251] Aspect 44. The system of any one of Aspects 42-43, wherein the fluid channel comprises a region having a cross-sectional area that remains constant long the direction of flow.

[0252] Aspect 45. The system of any one of Aspects 42-44, wherein the fluid channel defines a region having a width of from about 0.1 to about 10 pm, optionally from about 2 to about 8 pm.

[0253] Aspect 46. The system of any one of Aspects 41-45, wherein the shearing comprises a shear rate of from about 1 x 104s'1to about 10 x 105s’1, optionally from about 1 x 105s’1to about 9 x 105s’1.

[0254] Aspect 47. The system of any one of Aspects 41-46, wherein the droplet squeezing stage comprises a row of fluid channels, each fluid channel defining a direction of flow, and a fluid channel of a row optionally parallel to at least one other fluid channel of that row.

[0255] Aspect 48. The system of Aspect 47, wherein the droplet squeezing stage comprises a plurality of rows of fluid channels, the rows of fluid channels being optionally arranged parallel to one another, and optionally perpendicular to the direction of flow defined by a fluid channel of a row of fluid channels.

[0256] Aspect 49. The system of Aspect 48, wherein the droplet squeezing stage comprises from 2 to 50 rows of fluid channels.

[0257] Aspect 50. The system of any one of Aspects 48-49, wherein each fluid channel is configured to effect the same shearing.

[0258] Aspect 51. An EV hybrid, the EV hybrid formed according to any one of Aspects 1-18 or formed by a system according to any one of Aspects 39-50.103241.007498 / 25-10960 - 25-11237

[0259] Aspect 52. An EV hybrid, the EV hybrid comprising an EV fused to an LNP, the EV hybrid (1) retaining one or more surface proteins of the EV and the LNP, (2) comprising a therapeutic disposed therein, or both (1) and (2).

Claims

103241.007498 / 25-10960 - 25-11237What is Claimed:

1. A system, comprising:a mechanoporation region in fluid communication with an inlet configured to receive at least one of a supply of cells and a supply of carriers having nucleic acid disposed there,the mechanoporation region defining a direction of flow,the mechanoporation region defining a set of first gaps and a set of second gaps,the second gaps being arranged downstream relative to the first gaps along the direction of flow,a gap being sized so as to effect perturbation of the membrane of a cell encouraged through the gap, the perturbation being effective to enhance passage of a carrier through the membrane,a gap being defined between two protrusions,a protrusion having length measured in the direction of flow and a protrusion having a cross-sectional dimension measured along the length of the protrusion and perpendicular to the direction of flow, the cross-sectional dimension being free of reduction along the direction of flow.

2. The system of claim 1, wherein a carrier comprises a lipid nanoparticle.

3. The system of any one of claims 1-2, wherein the nucleic acid comprises any one or more of siRNA, mRNA, or pDNA.

4. The system of claim 3, wherein the nucleic acid comprises pDNA.

5. The system of any one of claims 1-2, wherein a protrusion is polygonal in crosssection.103241.007498 / 25-10960 - 25-112376. The system of any one of claims 1-2, wherein a protrusion is asymmetric about a line drawn perpendicular to the direction of flow.

7. The system of any one of claims 1-2, wherein a protrusion comprises a region of increasing cross-sectional dimension along the direction of flow and a region of constant cross-sectional dimension along the direction of flow.

8. The system of any one of claims 1-2, wherein a first gap is in at least partial register with a second gap along the direction of flow.

9. The system of any one of claims 1-2, wherein a first gap is not in at least partial register with a second gap along the direction of flow.

10. The system of any one of claims 1-2, comprising an inlet configured to receive the supply of cells and an inlet configured to receive the supply of carriers having nucleic acid disposed therein.

11. The system of any one of claims 1-2, wherein the first set of gaps is arranged in a row and wherein the second set of gaps is arranged in a row.

12. A system, comprising:a mechanoporation region in fluid communication with an inlet configured to receive at least one of a supply of cells and a supply of carriers having nucleic acid disposed there,the mechanoporation region defining a direction of flow,the mechanoporation region defining n sets of gaps arranged in the direction of flow,wherein n is from 6 to 12 and wherein an (n+l)th gap is downstream relative to an nth gap along the direction of flow,a gap being sized so as to effect perturbation of the membrane of a cell encouraged through the gap, the perturbation being effective to promote fusion of a carrier to the membrane,103241.007498 / 25-10960 - 25-11237a gap being defined between two protrusions.

13. The system of claim 12, wherein each of the n sets of gaps is arranged in a row, and wherein an (n+l)th row of gaps is arranged parallel to an nth row of gaps.

14. A method, comprising:encouraging a cell having a diameter through at least a first gap having a cross- sectional dimension smaller than the diameter of the cell,the encouraging mechanically perturbing the membrane of a cell so as to promote fusion to the membrane by a carrier having a nucleic acid disposed therein, andthe first gap optionally having a width in the range of from about 15% to about 30% of diameter of the cell.

15. The method of claim 14, wherein the carrier is delivered into the cell by endocytosis.

16. The method of any one of claims 14-15, wherein the carrier comprises a lipid nanoparticle.

17. The method of claim 14-15, wherein the nucleic acid comprises any one or more of mRNA, siRNA, or pDNA.

18. The method of any one of claims 14-15, wherein the cell is a T cell.

19. The method of any one of claims 14-15, wherein a gap is defined between two protrusions.

20. The method of any one of claims 14-15, wherein the lipid nanoparticle comprises any one or more of C12-200, l,2-dioleoyl-3-dimethylammonium-propane (DODAP), DLin-MC3-DMA, l,2-dioleoyl-3-trimethylammonium-propane (DOTAP), l,2-dioleoyl-sn-glycero-3-phosphoethanolamine (DOPE), cholesterol, or l,2-dimyristoyl-sn-glycero-3-phosphoethanolamine-N-[methoxy(polyethylene glycol)-2000] (ammonium) (C14-PEG 2000).103241.007498 / 25-10960 - 25-1123721. A method, comprising:subjecting a droplet comprising an extracellular vesicle (EV) and a lipid nanoparticle (LNP) to shearing sufficient to effect fusion between the EV and LNP and form an EV hybrid.

22. The method of claim 21, wherein the subjecting is performed in a fluid channel that defines a direction of flow, the fluid channel optionally defined between features that project from a substrate.

23. The method of claim 22, wherein the fluid channel comprises a constriction region having a cross-sectional area that varies along the direction of flow, the cross- sectional area optionally diminishing along the direction of flow.

24. The method of any one of claims 22-23, wherein the fluid channel comprises a region having a cross-sectional area that remains constant long the direction of flow.

25. The method of claim 24, wherein the region having a cross-sectional area that remains constant along the direction of flow has a width that is from 5% to 99% of a diameter of the droplet, optionally from 10% to 25% of a diameter of the droplet.

26. The method of any one of claims 22-23, wherein the fluid channel defines a region having a width of from about 0.1 to about 10 pm, optionally from about 2 to about 8 pm.

27. The method of any one of claims 21-23, wherein the shearing comprises a shear rate of from about 1 x 105s'1to about 10 x 105s’1, optionally from about 1 x 105s’1to about 9 x 105s’1.

28. The method of any one of claims 21-23, wherein the shearing comprises subjecting the extracellular vesicle (EV) and the lipid nanoparticle (LNP) to a plurality of shearing cycles.

29. The method of claim 28, wherein the plurality of shearing cycles comprises from 2 to 50 cycles, optionally from 10 to 40 cycles.103241.007498 / 25-10960 - 25-1123730. The method of any one of claims 21-23, wherein at least one of the EV and the LNP comprises a targeting moiety, the targeting moiety optionally comprises any one or more of a receptor, a receptor fragment, a ligand, a ligand fragment, an antibody, and an antibody fragment.

31. The method of any one of claims 21-23, wherein at least one of the EV and the LNP comprises a drug or a nucleic acid disposed therein or thereon, the drug optionally comprising a kinase inhibitor.

32. The method of any one of claims 21-23, further comprising placing the EV and the LNP in the droplet, optionally wherein the placing and the subjecting are performed within a single system.

33. A method, comprising:subjecting a droplet comprising an extracellular vesicle (EV) and a lipid nanoparticle (LNP) to shearing sufficient to effect fusion between the EV and LNP and form an EV hybrid, the subjecting comprising encouraging the droplet through a fluid channel defined between features that project from a substrate, the fluid channel defining a direction of flow.

34. The method of claim 33, wherein the shearing comprises a shear rate of from about 1 x 105s'1to about 10 x 105s’1, optionally from about 1 x 105s’1to about 9 x 105s’ i35. The method of any one of claims 33-34, wherein the shearing comprises subjecting the extracellular vesicle (EV) and the lipid nanoparticle (LNP) to a plurality of shearing cycles.

36. The method of claim 35, wherein the plurality of shearing cycles comprises from 2 to 50 cycles, optionally from 10 to 40 cycles37. The method of any one of claims 33-34, wherein the fluid channel comprises a region having a cross-sectional area that remains constant along the direction of flow, the region having a width that is from 5% to 99% of a diameter of the droplet, optionally from 10% to 25% of a diameter of the droplet.103241.007498 / 25-10960 - 25-1123738. A method, comprising:subjecting a droplet comprising an extracellular vesicle (EV) and a lipid nanoparticle (LNP) to shearing sufficient to effect fusion between the EV and LNP and form an EV hybrid, the subjecting comprising encouraging the droplet through a fluid channel defined between features that project from a substrate.

39. A system, comprising:a droplet generation stage,the droplet generation stage configured to form droplets having EVs and LNPs disposed therein; anda droplet squeezing stage,the droplet squeezing stage configured to (1) receive droplets from the droplet generation stage, and (2) subject received droplets to shearing sufficient to effect fusion between an EV and an LNP of a droplet so as to form an EV hybrid.

40. The system of claim 39, wherein the droplet generation stage comprises at least one of (1) an inlet configured to receive EVs and (2) an inlet configured to receive LNPs.

41. The system of claim 39, wherein the droplet generation stage comprises at least one of (1) a stage configured to generate EVs and (2) a stage configured to generate LNPs.

42. The system of any one of claims 39-41, wherein the droplet squeezing stage comprises least one fluid channel that defines a direction of flow.

43. The system of claim 42, wherein the fluid channel comprises a constriction region having a cross-sectional area that varies along the direction of flow, the cross- sectional area optionally diminishing along the direction of flow.103241.007498 / 25-10960 - 25-1123744. The system of claim 42, wherein the fluid channel comprises a region having a cross-sectional area that remains constant long the direction of flow.

45. The system of any one claim 42, wherein the fluid channel defines a region having a width of from about 0.1 to about 10 pm, optionally from about 2 to about 8 pm.

46. The system of any one of claims 41-43, wherein the shearing comprises a shear rate of from about 1 x 104s'1to about 10 x 105s’1, optionally from about 1 x 105s’1to about 9 x 105s’1.

47. The system of any one of claims 41-43, wherein the droplet squeezing stage comprises a row of fluid channels, each fluid channel defining a direction of flow, and a fluid channel of a row optionally parallel to at least one other fluid channel of that row.

48. The system of claim 47, wherein the droplet squeezing stage comprises a plurality of rows of fluid channels, the rows of fluid channels being optionally arranged parallel to one another, and optionally perpendicular to the direction of flow defined by a fluid channel of a row of fluid channels.

49. The system of claim 48, wherein the droplet squeezing stage comprises from 2 to 50 rows of fluid channels.

50. The system of claim 48, wherein each fluid channel is configured to effect the same shearing.

51. An EV hybrid, the EV hybrid formed according to a system according to any one of claims 1-2.

52. An EV hybrid, the EV hybrid formed according to a system according to any one of claims 39-41.

53. An EV hybrid, the EV hybrid comprising an EV fused to an LNP, the EV hybrid (1) retaining one or more surface proteins of the EV and the LNP, (2) comprising a therapeutic disposed therein, or both (1) and (2).

Images