A Digital Microfluidic Electroporation Platform for Low-input CRISPR Genome Editing and mRNA Transfection In Suspension T Cells and 3D Cell Models

Summary

This protocol describes miniaturized CRISPR–Cas9 genome editing of TCRα/β receptor in primary human T cells using ‘DMF-ection’, a digital microfluidic (DMF) electroporation platform. The method enables parallel, low-input gene editing with high viability and is further adaptable for mRNA delivery into multicellular 3D spheroids.

Abstract

Digital microfluidic (DMF) electroporation enables precise, low-volume genetic manipulation of mammalian cells while minimizing cellular input by up to 100x and preserving viability. This study presents a high-throughput DMF-based transfection workflow for CRISPR-mediated knockout of the TRAC locus in primary human suspension T cells and for mRNA transfection of three-dimensional HEK293T spheroids. Using spatially deposited CRISPR guide RNAs and on- cartridge ribonucleoprotein (RNP) assembly, efficient TRAC locus disruption was achieved in both CD4⁺ and CD8⁺ T-cell populations using only 10,000 cells per condition, with post-editing viabilities exceeding 85%. Biophysical characterization using flow-induced and Taylor dispersion analyses revealed that polymer additives stabilize Cas9–sgRNA complexes under electroporation buffer conditions, supporting reproducible editing at sub microliter volumes. The workflow was further adapted for 3D applications by delivering EGFP mRNA into intact HEK293T spheroids, resulting in robust and spatially uniform fluorescence without impairing spheroid growth or morphology. Together, these results demonstrate that DMF electroporation enables efficient genome editing and mRNA delivery across both suspension immune cells and multicellular spheroids. This platform provides a scalable and low-input solution for applications in CAR-T cell therapy, functional genomics, and advanced 3D cellular models.

Introduction

Gene editing in primary human cells is increasingly central to both therapeutic development and basic research, including applications in immuno-engineering, rare disease modeling, and functional genomics¹. Although electroporation enables highly efficient genome editing, conventional cuvette-based systems typically require up to millions of cells per condition, restricting their use for patient-derived samples and rare immune populations². In addition, these systems are poorly suited for large-scale screening due to limitations in throughput, automation, and reagent consumption³.

Microfluidic electroporation has emerged as a promising approach for low-input genome editing⁴. However, most existing platforms rely on continuous flow architectures that involve complex fluid handling, limited flexibility in reagent mixing, and relatively high per-condition cell requirements. In contrast, digital microfluidics (DMF) enables active manipulation of discrete nanoliter-scale droplets on a planar electrode array, allowing precise control over reaction composition, timing, and spatial localization. This plug-and-play droplet-based format is well-suited for low-volume, high-throughput workflows where conservation of primary cells and reagents is essential⁵.

Recent work has demonstrated the feasibility of DMF-based intracellular delivery using tri-droplet electroporation architectures that generate localized, low-current electric fields for efficient RNP and mRNA delivery while minimizing thermal and electrochemical stress^6,7,8. Building on this foundation, a next-generation DMF electroporation platform, termed ‘DMF-ection’ featuring 48 independently addressable reaction sites, full automation compatibility, and scalable cartridge manufacturing⁹ (Figure 1) was developed. The DMF cartridge consists of a bottom PCB plate and a plastic top plate separated by a gasket, assembled via alignment snaps in a single correct orientation (Figure 1A). The 48-plex layout is organized into eight identical families (Figure 1A–H), each containing six independently addressable electroporation sites, supporting up to 48 parallel editing conditions per cartridge (Figure 1B). Guide RNA libraries are deposited onto the substrate surface using an acoustic dispenser prior to cartridge assembly. Positional accuracy of deposition was confirmed across all 48 sites, with drop positions clustering within sub-millimeter deviation from the electrode midpoint (Figure 1C), supporting reliable on-cartridge RNP assembly. Following loading, cells and liquid electrode droplets are actuated on-cartridge to form the tri-droplet electroporation geometry (Figure 1D). After electroporation, cells are offloaded into culture plates for recovery and downstream analysis, including flow cytometry, sequencing, and microscopy (Figure 1E). This system enables high-efficiency genome editing using up to 100-fold fewer cells than standard cuvette-based methods, significantly expanding access to genome engineering in low-input and high-throughput applications.

This study assesses the platform’s performance by delivering multiple biomolecular cargoes, including mRNA and CRISPR–Cas9 ribonucleoprotein (RNP) complexes, achieving efficient transfection and gene disruption. In addition to primary T cells, DMF-ection provides a generalizable framework for delivering nucleic acids into complex multicellular structures. Three-dimensional (3D) spheroids and organoid-type models impose significant barriers to reagent penetration due to their architecture and fragility, and conventional electroporation approaches often disrupt morphology or generate heterogeneous delivery. To assess whether the same miniaturized electroporation principles used for T-cell editing extend to structured tissues, the DMF workflow was applied to 3D HEK293T spheroids. These experiments demonstrated that DMF-based electrowetting can gently position intact spheroids within tri-droplet geometries and support uniform mRNA delivery while preserving structure and growth. Although these results are outside the primary T-cell protocol described here, they illustrate the broader applicability of programmable DMF electroporation for emerging 3D cellular systems relevant to drug discovery and disease modeling.

Protocol

Primary human T cells used in this protocol were obtained from commercial suppliers under informed donor consent and in compliance with applicable institutional, ethical, and regulatory guidelines. All handling of human-derived biological materials was conducted in accordance with institutional biosafety protocols. Researchers implementing this protocol are responsible for ensuring that their use of primary human cells complies with local institutional review board (IRB) or ethics committee requirements, applicable biosafety regulations, and any relevant national or regional legislation governing the use of human biological material. No patient-identifiable information was associated with the cell material used in this study.

1. Preparation and activation of primary human T cells

1. Thaw cryopreserved primary human CD3⁺ T cells rapidly in a 37 °C water bath. Transfer the cells to prewarmed basal T-cell medium and centrifuge at 180 × g for 10 min. Discard the supernatant and resuspend the pellet in fresh medium.
2. Determine cell concentration and viability using an automated cell counter.
3. Resuspend cells at 1 × 10⁶ cells/mL in T-cell culture medium supplemented with a soluble tetrameric antibody CD3/CD28 T-cell activator stimulus (according to vendor instructions) and 200 IU/mL IL-2. Incubate cells for 72 h at 37 °C, 5% CO₂.
   NOTE: Maintain cell density between 0.5–2 × 10⁶ cells/mL during activation. Perform activation in U-bottom 96-well plates to optimize cell-to-activator contact and activation efficiency.

2. Preparation of guide RNA spots on the DMF substrate (acoustic dispenser spotting)

1. Reconstitute lyophilized sgRNA in nuclease-free water to a stock concentration of 100 µM. In a separate tube, prepare a 100 mg/mL stock of poly-L-glutamic acid (PGA, sodium salt, MW 15,000–50,000 Da) in nuclease-free water. Combine sgRNA and PGA stock at a 5:4 volume ratio (sgRNA: PGA) that yields a final sgRNA concentration of 55.6 pmol/µL and a final PGA concentration of 44.4 mg/mL in nuclease-free water. Vortex briefly and centrifuge at 1,000 × g for 30 s and aliquot into single-use volumes and store unused aliquots at −80 °C for up to 6 months.
   NOTE: Avoid repeated freeze-thaw cycles.
2. Add the guide-PGA mixture to a low-dead-volume acoustic source plate in wells corresponding to the desired reaction site layout.
3. Transfer 40 nL (equivalent to 2.2 pmol sgRNA) of the guide-PGA mixture to each target spotting region of the DMF substrate using a calibrated acoustic dispenser.
   NOTE: At 55.6 pmol/µL, 40 nL delivers 2.2 pmol sgRNA per site (55.6 × 0.040 = 2.22 pmol). Dried guides are reconstituted during droplet merging on-cartridge; complete resuspension is confirmed by consistent editing efficiency across sites.
4. In a biological safety cabinet, allow the spotted droplets to air-dry completely at room temperature until no visible liquid remains (approximately 5 min).
   NOTE: Do not use forced airflow or heat, as this may dislodge or degrade the guide RNA.
5. Assemble the DMF cartridge.
   1. In a biological safety cabinet, align the top plate over the spotted substrate by matching the asymmetrical plastic snapping features from the top plate with the notches on the bottom plate. Apply even, simultaneous pressure along the short edges until a distinct mechanical snap is felt, confirming full engagement of the locking features.
      NOTE: The cartridge is designed with a single correct orientation; incorrect assembly will not produce the snap. Hold substrates by the edges at all times to avoid contamination or damage to functional surfaces and avoid sliding the substrates against each other as this may scratch the surface coatings.
   2. Checkpoint: Confirm correct assembly by the mechanical snap.
      NOTE: There is no prerun electrical indicator of coating integrity; any functional failures resulting from improper assembly are detected automatically and reported in the post-run QC report.
      Pause point: Store unused assembled cartridges in a sealed package with desiccant (silica gel packets) at room temperature for up to 48 h.

3. Cell preparation

1. Prepare 1 mL of Complete Transfection Buffer freshly before each use by combining DMF-compatible electroporation buffer with the provided biocompatible non-ionic surfactant at a 20:1 (v/v) ratio (electroporation buffer: surfactant), yielding a final surfactant concentration of 0.05% (v/v). Mix gently and keep on ice.
   NOTE: The DMF-compatible transfection buffer used here is a low-conductivity, isotonic buffer (conductivity: 3.5 mS/cm, osmolality: 305 mOsm/L). Researchers using alternative buffers should verify compatibility with DMF actuation and cell viability. The biocompatible non-ionic surfactant serves to reduce droplet pinning; if the provided surfactant is not used, a suitable substitute.
2. Wash activated T cells once with Transfection Buffer (containing no surfactant). Centrifuge at 180 × g for 5 min, aspirate the supernatant, and resuspend the cell pellet in Complete Transfection Buffer (containing surfactant) at 1.0 × 10⁷ cells/mL in a total volume of 100 µL per cartridge. Keep on ice.
3. Immediately before loading onto the cartridge, add Cas9 nuclease directly to the 100 µL of T-cell suspension and mix by gentle pipetting; keep on ice.
   NOTE: For primary T cells, Cas9 is loaded at 42 pmol per 100 µL cartridge volume (0.42 pmol/µL). Each electroporation site processes approximately 1 µL of cell suspension, delivering 0.42 pmol Cas9 per edit. Optimize this dosage when adapting to other immune cell types (see Table 1).

Component	Final concentration per edit
Cas9	0.42 pmol in ~1 µL droplet
sgRNA	2.1 pmol (from 40 nL spot)
PGA	44.4 mg/mL stock in spotting mix
Surfactant	0.05% (v/v)
Cell input	~10,000 cells

Table 1: Summary table of final concentrations in each transfection droplet.

4. Transfection on the DMF-ection platform

1. Using a multichannel pipette, dispense 10 µL of the T-cell/Cas9 mixture into each of the 8 cell loading ports, and 10 µL of complete Transfection Buffer into the 16 liquid electrode loading ports (as shown in Figure 1).
2. Set the voltage to 500 V, pulse duration to 3 ms, and number of pulses to 2. Initiate the electroporation sequence according to the platform instructions.
   1. Checkpoint: Review the postrun QC report immediately after the run. Sporadic failures (<10% of sites) require no intervention. If failures are systematic, replace the cartridge with the provided QC board, run the System Health Check (~2 min), and contact technical support if issues persist.

5. Offloading and recovery of edited T cells

1. Using a liquid handler, transfer electroporated cells from each reaction site into a 96-well U-bottom plate containing 150 µL of prewarmed basal T-cell medium supplemented with 200 IU/mL IL-2.
2. Incubate cells for 72 h at 37 °C, 5% CO₂ to permit protein turnover and surface receptor expression changes.

6. Flow cytometry analysis of TRAC knockout

1. Transfer cells to a sterile 96-well conical bottom plate to reduce cell loss between washes. Wash cells twice with Flow Cytometry Buffer (1× PBS, 2% FBS, 2 mM EDTA) by centrifuging at 400 × g for 5 min and removing the supernatant by gently inverting the conical bottom plate. Cells will remain pelleted in the plate.
2. Prepare a surface staining master mix containing anti-CD4, anti-CD8, and anti-TCRα/β antibodies in Flow Cytometry Buffer at manufacturer-recommended concentrations. Add 25 µL of the master mix per well and incubate for 20 min at 4 °C protected from light. Wash twice with 1× PBS (without FBS) by centrifuging at 400 × g for 5 min. Resuspend cells in 100 µL of 1× PBS containing a viability dye at 1:1,000 (v/v) and incubate for 10 min at room temperature protected from light.
   NOTE: Conical bottom plates are effective at pelleting cells; check that cells are well resuspended in all steps.
3. Wash cells twice with Flow Cytometry Buffer and resuspend in 150–200 µL per sample for acquisition.
4. Configure the cytometer with appropriate laser lines and filter sets for the fluorochromes used (e.g., 488 nm excitation for viability dye; 405 nm or 488 nm for surface markers as appropriate). Set voltages using single-stain compensation controls and an unstained sample. Collect a minimum of 5,000 live singlet events per sample. Gate sequentially on: scatter singlets → live cells (viability dye negative) → CD4⁺ or CD8⁺ → TCRα/β expression.
   NOTE: Under standard conditions, total live events acquired can range from 5,000 to over 10,000 per sample following losses during washing and staining, given that the cells have had time to proliferate.
5. Determine TRAC gene disruption by measuring the frequency of TCRα/β-negative cells within the CD4⁺ and CD8⁺ populations. Report knockout efficiency as the percentage of TCRα/β-negative cells relative to the live singlet gate for each subset.

Results

Expected outcomes
Electroporation of 10,000 activated T cells per condition typically results in >85% TRAC knockout efficiency with >95% viability 72 h post-transfection. Flow cytometry reveals a clear loss of TCRα/β staining in the TRAC-targeting condition relative to non-targeting sgRNA controls (NTC) and controls not exposed to any electric fields. Unless otherwise stated, n represents independent DMF-ection runs performed on separate days. Technical replicates from the same cartridge run were averaged prior to statistical analysis and do not contribute independently to n. All T cell experiments reported here were conducted using cryopreserved cells sourced from a single healthy donor. Donor-to-donor variability using the same platform is characterized further⁹.

Improved Cas9–sgRNA ribonucleoprotein stability under electroporation conditions due to anionic polymer additives
A core design requirement of digital microfluidics is a highly hydrophobic surface, which allows droplets to glide freely during electrowetting. While this property is essential for droplet motion and precise actuation, it creates challenges for reagent loading. For arrayed screening applications, it is critical that user-defined payload libraries can be deposited directly onto the cartridge surface. This enables reagent layouts to mirror the user’s library design and ensures that the digital microfluidic workflow can seamlessly execute arrayed experiments without additional transfer or manual intervention. Nano dispensing with an acoustic dispenser is well suited for this task, as it can deliver nanoliter volumes with high positional accuracy across large arrays onto various surfaces. However, when droplets strike a hydrophobic cartridge surface, they often bounce, bead, or fail to spread uniformly, leading to inconsistent deposition and recovery. This variability is most evident at intermediate droplet volumes ranging from 40 nL to 120 nL (Figure 2A, n = 3 per condition, *** p value < 0.01), where it is hypothesized that the kinetic energy of the droplet is high enough to cause rebound, but surface adhesion is insufficient to anchor the liquid. As a result, direct on-cartridge deposition at higher payload dosages can be unreliable without further modification. To mitigate this effect, reagents were supplemented with the biosafe polymer poly-L-glutamic acid (PGA). PGA is a biodegradable, negatively charged polypeptide widely used as a stabilizer in pharmaceutical formulations, with excellent compatibility for live-cell workflows. The charged backbone is hypothesized to lower the droplet contact angle and promote more uniform spreading, reducing rebound and improving surface adhesion. With PGA present, deposited fluorescein volumes scaled linearly with absorbance, and variance was markedly reduced across cartridges. Statistical analysis confirmed that PGA significantly improved reproducibility at 40, 90, and 120 nL spotting volumes (***p < 0.001, ****p < 0.0001), while very small droplets (10–20 nL) showed little difference, as rebound is less pronounced at low kinetic energies.

Next, the addition of polymer additives used in the DMF workflow were evaluated for their influence on Cas9–sgRNA ribonucleoprotein (RNP) behavior under the low-volume buffer conditions used for electroporation. To this end, Flow-induced dispersion analysis (FIDA) technology was used. FIDA was selected as it provides an accurate, solution-phase measurement of hydrodynamic radius and binding affinity, allowing protein-nucleic acid interactions to be quantified under the same buffer conditions used in an experimental setting. FIDA demonstrated that free sgRNA exhibited a hydrodynamic radius (Rh) of 3.45 ± 0.26 nm, consistent with a small, single-stranded RNA species. Titration with Cas9 produced a stable RNP complex with an Rh of 15.32 ± 0.46 nm and a sub-nanomolar binding affinity (KD = 0.70 ± 0.14 nM) (Figure 2).

To assess the effect of polymer stabilization, poly-L-glutamic acid (PGA) was added to the guide-RNA prior to RNP formation. The presence of PGA reduced the apparent affinity between Cas9 and sgRNA (K_D = 67.7 ± 9.5 nM) and decreased complex size to approximately 10 nm (Figure 2A right panel). Although the reduced affinity suggests less tightly associated complexes, the smaller and more uniform R_h indicated that PGA prevents the formation of oversized RNP assemblies (Figure 2B).

Next, RNP stability was assessed under electroporation conditions using Taylor dispersion analysis (Figure 2C). Without polymer, Cas9–sgRNA mixtures produced broad, poorly fitted Taylorgrams indicative of aggregation (Figure 2B). In contrast, addition of 0.0025 mg/mL PGA yielded well-defined, monodisperse dispersion profiles (R² = 0.999) with R_h ~ 9.6 nm. These results demonstrate that PGA prevents aggregation and improves RNP uniformity, thereby supporting reproducible nanoliter-scale electroporation. This stabilization provides a mechanistic rationale for the high editing efficiencies observed in downstream T-cell experiments.

Dose-dependent editing efficiency in primary T cells

Although FIDA experiments indicated that PGA reduces the efficiency of RNP assembly, functional editing outcomes were hypothesized to be less affected, since electroporation is typically performed with excess guide RNA relative to Cas9. Therefore PGA–RNP input volume was evaluated for impact on knockout efficiency in primary human T cells at the TRAC locus (Figure 2D). Across four cartridges, editing efficiency was consistently high (>75–80%) at RNP inputs of 40 nL (2.1 pmol), 20 nL (1.05 pmol), and 10 nL (0.53 pmol), with minimal variability between CD4⁺ and CD8⁺ subsets. Even at 5 nL (0.26 pmol), editing remained robust, though a modest reduction was observed compared to higher doses. At the lowest dose tested (2.5 nL, 0.13 pmol), editing efficiency decreased substantially, plateauing near ~60% TRAC KO. Controls confirmed the specificity of editing: no-electroporation wells (0V) and off-cartridge controls (OCC) exhibited only background knockout. The elevated TCR-negative frequency observed in the off-cartridge control (OCC) CD8⁺ population likely reflects donor-specific baseline TCR downmodulation following activation, independent of electroporation or gene editing as these cells were not exposed to the cartridge. Together, these results demonstrate that DMF-ection enables efficient CRISPR-mediated editing at nanoliter-scale RNP inputs. Importantly, editing was comparable between CD4⁺ and CD8⁺ T cells, highlighting the platform’s applicability across T cell subsets and suggesting that RNP usage can be reduced by more than an order of magnitude without compromising knockout efficiency.

Efficient gene disruption in primary CD4⁺ and CD8⁺ T cells using miniaturized DMF-ection
To validate the performance of the DMF-ection workflow in more depth, CRISPR–Cas9-mediated knockout of the TRAC locus was assessed in primary human CD4⁺ and CD8⁺ T cells using only 10,000 activated cells per condition. Three experimental groups were included: the TRAC-targeting guide condition (TRAC KO), a non-targeting control (NTC) to account for non-specific effects of RNP delivery, and an off-cartridge control (OCC) to establish baseline TCRα/β expression in untreated, unmanipulated cells. A standard gating strategy was applied to identify live singlets, followed by CD4⁺ and CD8⁺ subsets and TCRα/β expression (Figure 3A). High-efficiency gene disruption of the TRAC locus was achieved in both T-cell subsets across 40 independent electroporation runs. In CD4⁺ T cells, TRAC knockout efficiency reached 88.4 ± 4.9%, compared to 3.2 ± 4.7% in NTC and 9.8 ± 5.1% in OCC conditions. A comparable result was observed in CD8⁺ T cells, where TRAC KO reached 89.8 ± 4.2%, versus 4.2 ± 7.4% (NTC) and 20.2 ± 4.9% (OCC) (Figure 3B, left). Importantly, electroporation produced no substantial cytotoxicity across conditions. Viability exceeded 96% in all groups, with no significant difference between TRAC KO (97.1 ± 1%), NTC (96.8 ± 1%), and OCC (97 ± 1%) conditions (Figure 3B, right). Together, these data demonstrate that DMF electroporation enables robust, high-efficiency TRAC disruption while preserving cellular viability across multiple independent runs. Primary human T cells were obtained from commercial sources with informed donor consent and in accordance with institutional and vendor ethical guidelines.

Handling of HEK293T spheroids with digital microfluidics
Next, the digital microfluidic electroporation workflow was evaluated for applications with multicellular three-dimensional structures. Hollow HEK293T spheroids were formed by seeding approximately 100 cells per well in low-attachment microplates and culturing for 24 h to permit spheroid formation (Figure 4A). Formed spheroids were subsequently collected, resuspended, and loaded onto the digital microfluidic cartridge for electroporation. Following treatment, spheroids were offloaded into a 96-well plate for downstream imaging and growth analysis.

To achieve an average recovery of one spheroid per well after electroporation, an excess number of spheroids was intentionally loaded onto the cartridge to compensate for material loss during handling and transfer. Approximately 192 spheroids were resuspended in 100 µL of electroporation buffer and loaded onto the cartridge. Each reaction family received ~9.3 µL of the suspension, corresponding to ~18 spheroids, with individual droplets containing ~1 µL (1–2 spheroids per droplet). Because approximately half of the spheroids remained in the loading port during cartridge operation, this loading strategy resulted in an average of ~1 spheroid per tri-droplet electroporation site.

Analysis of spheroid recovery indicated that 0–1 spheroids were typically offloaded per well after electroporation, corresponding to approximately 4–6 spheroids recovered per reaction family (Figure 4B left panel). This recovery is lower than the theoretical maximum and reflects losses occurring during washing, aspiration of supernatant, and retention of spheroids in the loading port. These losses are attributed to the small spheroid diameter (~100 µm) and the inherent difficulty of maintaining the uniform distribution of non-single-cell structures within suspension. As a result, some spheroids remain unevenly distributed or fail to enter the tri-droplet region. Loading an excess number of spheroids ensured that at least one spheroid was present in most electroporation sites, enabling consistent downstream analysis as demonstrated by on-microscope imaging of DAPI-stained spheroid in the microfluidics cartridge (Figure 4B, right panel). Further process optimization may reduce the number of spheroids required to achieve equivalent recovery. Despite these handling losses, the workflow enabled gentle positioning of intact spheroids within the tri-droplet electroporation geometry.

EGFP mRNA delivery and viability in HEK293T spheroids
Efficient delivery of EGFP mRNA into HEK293T spheroids was observed following electroporation (50 ng of GFP mRNA per spheroid). Fluorescence microscopy performed 24 h post electroporation revealed robust and spatially uniform GFP expression throughout electroporated spheroids, whereas no detectable fluorescence was observed in 0 V or off-cartridge control conditions (Figure 4C, left panel). Quantitative analysis of normalized mean fluorescence intensity (MFI) per spheroid area confirmed successful transfection under both electroporation parameter sets tested (Figure 4C, middle panel). Specifically, spheroids electroporated at 500 V, 2 ms, 2 pulses exhibited an MFI of 6.96 × 10^-5 ± 3.50 × 10^-5, while those electroporated at 375 V, 6 ms, 1 pulse showed an MFI of 5.05 × 10^-5 ± 3.29 × 10^-5. No measurable fluorescence was detected in control conditions (0 ± 0). Both electroporation conditions produced significantly higher fluorescence compared to controls (p < 0.05), with no statistically significant difference between pulse parameters. To assess whether electroporation adversely affected spheroid growth or structural integrity, spheroid diameters were measured daily over 5 days following treatment. Electroporated spheroids exhibited growth kinetics comparable to untreated controls, indicating that mRNA transfection did not impair spheroid viability or proliferative capacity (Figure 4C, right panel). Together, these results demonstrate that digital microfluidic electroporation enables efficient mRNA delivery into intact multicellular spheroids while preserving structural integrity and growth, supporting the feasibility of extending this platform to three-dimensional cellular models.

Data are presented as mean ± standard deviation. Statistical significance was determined using one-way ANOVA with Tukey's post hoc correction for multiple comparisons, with a significance threshold of p < 0.05. N values represent independent biological replicates (separate donors or independent cartridge runs) as specified in the figure.

Figure 1. DMF electroporation platform architecture and workflow. (A) Photograph of the DMF cartridge. (B) Annotated schematic of the assembled 48-plex cartridge layout (center image), showing the (A–H) organization of eight identical families, each containing six independently addressable electroporation sites, with liquid electrode loading ports, cell loading ports, anode, cathode, and nano spot regions labeled (left inset image). Key features, including loading ports, offloading ports, and assembly snaps, are labeled. The cartridge is assembled by snap-fitting the two plates in a single correct orientation until a distinct mechanical engagement is felt. Right image: liquid handler dispense map showing color-coded positions for loading ports (red), offloading ports (pink), and spotting target regions (blue). (C) Left: image showing the positional accuracy of acoustic dispensing across all 48 sites (n = 48), with drop positions plotted relative to the electrode midpoint (mm). The tight clustering confirms sub-millimeter deposition accuracy. Right: representative pictures showing a correctly centered spot before rehydration by the cell droplet (top) and during the rehydration, where the droplet is shuttled back and forth where the guide is in order to resuspend (bottom, red arrow). (D) Top-view picture of discrete electroporation regions, showing the stepwise movement of droplets from (i) rehydration of the payload by shuttling the droplet back and forth over the deposited guide RNA, (ii) movement to the electroporation site, between the anode and cathode, where the cell droplet bridges with the two flanking liquid electrode droplets and gets pulsed, and (iii) the complete tri-drop moving to its offloading port prior to retrieval by the liquid handler. (E) End-to-end workflow schematic. Guide RNA is deposited onto the cartridge substrate using an acoustic dispenser. Cells and payload are prepared and loaded onto the cartridge using an automated liquid handler. Electroporation is performed on the DMF platform. Cells are offloaded into a 96-well plate and transferred to an incubator for recovery. Downstream analysis is performed using flow cytometry, microscopy, next-generation sequencing, or other appropriate readouts. Please click here to view a larger version of this figure.

Figure 2. Impact of polymer supplementation on deposition and RNP assembly. (A) Polymer supplementation improves acoustic deposition fidelity. Fluorescein absorbance following deposition and reconstitution across four independent cartridges, with or without poly-L-glutamic acid. Variance was reduced at intermediate volumes (40–120 nL) in the presence of PGA. 
(B) Impact of PGA on Cas9–sgRNA assembly. FIDA affinity binding curves measuring hydrodynamic radius (Rh) in the absence (left panel) or presence (right panel) of PGA. Free sgRNA exhibited Rh of 3.45 ± 0.26 nm, while Cas9 titration yielded RNP complex size of 15.32 ± 0.46 nm (Kd = 0.70 ± 0.14 nM). Addition of PGA reduced complex size and weakened binding affinity in a concentration-dependent manner. With the addition of 0.005 mg/mL PGA, Free sgRNA exhibited Rh of 4.31 ± 0.07 nm, while Cas9 titration yielded RNP complex size of 10.21 ± 0.38 nm (K_d = 67.7 ± 9.5 nM). Experiments were performed in a low-conductivity electroporation assay buffer with 0.05% Pluronic F-68. Fluorescence was detected at 480 nm using the integrated LED module. A permanently coated capillary (Ø 75 µm, L = 100 cm) was used for all experiments. To test the effect of polyanionic polymers on RNP assembly, PGA was added to assay mixtures at concentrations of 0.0025 mg/mL and 0.005 mg/mL. These conditions were selected to model the range used in digital microfluidics electrowetting and electroporation buffers. For direct binding analysis, 10 nM FAM-labeled sgRNA targeting the TRAC locus was used as the fluorescent indicator. Increasing concentrations of Cas9 nuclease (up to 500 nM) were titrated against the indicator in a CapMix assay to generate dispersion profiles for complex formation. All experiments were performed at 25 °C, with each datapoint consuming 40 nL of Indicator and ~12 µL of analyte (Cas9). Experiments were run with the following sequence: (1) flushing with 0.1% Phosphoric Acid at 3,500 mbar for 30 s; (2) flushing with assay buffer at 3,500 mbar for 30 s; (3) filling the capillary with analyte at 3,500 mbar for 20 s; (4) injecting the indicator at 50 mbar for 10 s; (5) mobilizing and measuring with analyte at 400 mbar for 180 s. Hydrodynamic radii were calculated by using the extracted peak width at one half max peak height, and dissociation constants (K_D) were determined by nonlinear regression fitting to a 1:1 binding model. (C) Effect of PGA on RNP stability. Taylor dispersion profiles of Cas9–sgRNA complexes with and without PGA, performed with 500 nM Cas9 nuclease and 50 nM FAM-tagged sgRNA. In the absence of PGA, no stable complex was observed, as indicated by the increasing formation of large aggregates along with a decreasing signal from the complex over time (left). Addition of 0.0025 mg/mL PGA yielded a well-defined, monodisperse Taylorgram with minimal aggregation measuring a R_h ~ 9.6 nm (R² = 0.999, right). Complex stability was assessed under standard electroporation conditions (500 nM Cas9 + 50 nM sgRNA) with or without 0.0025 mg/mL PGA. Samples were premixed and maintained at 4 °C in the inlet sample tray while injections were conducted in 10 min intervals over a period of several hours. Experiments were run with the following sequence: (1) flushing with 0.1% Phosphoric Acid at 3,500 mbar for 30 s; (2) flushing with assay buffer at 3,500 mbar for 30 s; (3) filling the capillary with analyte (assay buffer) at 3,500 mbar for 20 s; (4) injecting the indicator at 50 mbar for 10 s; (5) mobilizing and measuring with analyte at 400 mbar for 180 s. Unstable complex formation was shown through the formation of large aggregates (spikes in the Taylorgram) and decreasing signal. Stable complexes were identified by well-fitted, monodisperse Taylorgrams (R² > 0.99), and R_h values were extracted from fitted profiles. Please click here to view a larger version of this figure.

Figure 3. TCR alpha/beta knockout efficiency and viability in primary human T cells. (A) Representative flow cytometry gating strategy for live singlets, CD4⁺ and CD8⁺ T-cell subsets, and TCRα/β expression in control (non-targeting sgRNA) and experimental (TRAC sgRNA) conditions. Cells were gated sequentially on FSC/SSC, singlets, viability, and CD4⁺ or CD8⁺ populations prior to assessment of TCRα/β surface expression. (B) Left: TCRα/β knockout efficiency in CD4⁺ and CD8⁺ T-cell subsets across three conditions: TRAC KO, non-targeting control (NTC), and off-cartridge control (OCC). CD4⁺: TRAC KO = 88.4 ± 4.9% (n = 40), NTC = 3.2 ± 4.7% (n = 24), OCC = 9.8 ± 5.1% (n = 18). CD8⁺: TRAC KO = 89.8 ± 4.2% (n = 40), NTC = 4.2 ± 7.4% (n = 24), OCC = 20.2 ± 4.9% (n = 18). Right: Post-electroporation viability across all conditions, gated on singlets. Viability: TRAC KO = 97.1 ± 1% (n = 40), NTC = 96.8 ± 1% (n = 24), OCC = 97 ± 1% (n = 18). Individual data points are shown with mean and SD. n represents independent electroporation runs across multiple cartridges and days. Please click here to view a larger version of this figure.

Figure 4. Efficient mRNA delivery into HEK293T spheroids using DMF-ection. (A) Workflow diagram describing the process of forming spheroids and performing spheroid transfection. A hundred HEK293T cells are loaded per well in specialized spheroid microplate and cultured for 24 h to form spheroids of 100 µm diameter. Spheroids (192) are offloaded and resuspended into 100 µL of transfection buffer. This suspension, with appropriate payload, is loaded onto the guide spotted DMF-ection cartridge at 9.3 µL per family. Once electroporation occurs, samples are offloaded into a 96 well plate for downstream analysis. (B) Left: Number of spheroids offloaded per family on DMF-ection cartridge as compared to hand-pipetted off-cartridge controls. Middle: Fluorescent image of tri-drop, where larger red square outlines the tri-drop, and the red oval circles the spheroid. Right: Fluorescent image showing loading port with loaded spheroids (blue circles). (C) Left panel: Brightfield images of intact HEK293T spheroids before electroporation and Fluorescence images of spheroids 24 h after EGFP mRNA electroporation, compared with 0 V and off-cartridge controls. (D) Top panel: Quantification of mean fluorescence intensity normalized to spheroid area for two electroporation conditions (500 V, 2 ms, 2 pulses; 375 V, 6 ms, 1 pulse). No significant difference between electroporated conditions, but all electroporated conditions are significantly different from 0 V and off-cartridge controls (500 V 2 ms 2 pulse: 6.955 × 10^-5 ± 3.502 × 10^-5, 375 V 6 ms 1 pulse: 5.045 × 10^-5 ± 3.294 × 10^-5, 0 V: 0 ± 0, Off-cartridge control : 0 ± 0  (*p < 0.05). Images were processed using FIJI image processing package¹⁰. (500 V, 2 ms, 2 pulses: n = 6; 375 V, 6 ms, 1 pulse: n = 4; Off-cartridge control: n = 3; 0 V control: n = 3). (D) Bottom panel: Spheroid diameter post electroporation over 5 days across conditions. Diameter was measured using FIJI from phase contract images taken at 20x magnification with an inverted microscope. (500 V, 2 ms, 2 pulses: n = 6; 375 V, 6 ms, 1 pulse: n = 4; Off cartridge control: n = 3; 0 V control: n = 3). Please click here to view a larger version of this figure.

Discussion

This protocol describes a miniaturized digital microfluidic electroporation workflow that enables efficient intracellular delivery into primary human T cells using nanoliter-scale droplets. A central advantage of this method is the ability to perform CRISPR–Cas9 genome editing using only 10,000 cells per reaction, dramatically reducing material requirements compared to conventional electroporation. The droplet-based architecture supports automation, ensuring reproducible reagent handling and reaction assembly across multiple parallel sites.

In previous studies, direct benchmarking of the DMF-ection platform against a widely used system demonstrated that conventional nucleofection produced less than 2% GFP-positive T cells at 10,000 cells per edit, while the DMF-ection platform achieved >90% transfection efficiency under the same low-input conditions. For an arrayed screen of 192 conditions, the DMF platform required 25-fold fewer cells and approximately 50-fold less Cas9 per condition than conventional methods⁹. These comparisons highlight the material advantages of miniaturized DMF electroporation for low-input and screening applications

Critical steps and troubleshooting
Several steps are critical for success. Accurate calibration of the acoustic liquid handler ensures precise deposition of guide RNA onto the cartridge substrate, which is essential for consistent RNP assembly. If spotting is inconsistent or droplets fail to adhere to the substrate, verify that PGA is included in the guide-RNA mixture at the correct concentration (5 mg/mL final) and that the acoustic dispenser is calibrated for the specific source plate geometry. Inconsistent spotting at intermediate volumes (40–120 nL) is most commonly caused by droplet rebound on the hydrophobic surface; increasing PGA concentration slightly or reducing dispense height can resolve this.

Maintaining the correct cell concentration prior to loading is important, as deviations in density can impair droplet formation and reduce editing efficiency. If editing efficiency falls below 75%, first verify that T-cell activation was sufficient (confirm CD25 and CD69 upregulation by flow cytometry prior to electroporation), and confirm that Cas9 and sgRNA were combined immediately before loading rather than in advance.

Polymer additives play an important role in stabilizing Cas9–sgRNA complexes under electroporation buffer conditions¹¹. Although PGA reduces nominal binding affinity, its stabilizing effect improves reproducibility and prevents aggregation at nanoliter volumes. This represents a tunable parameter for researchers adapting the workflow to other protein–RNA complexes.

Troubleshooting
Failed droplet bridging
Review the instrumental post-run QC report after each experiment. Absence of detected current indicates failed bridging, either due to insufficient droplets or failure to fuse, and is automatically flagged by the system. Affected sites cannot be recovered within the same run. If failures are sporadic (<10%), no action is required; distribute replicates across multiple cartridges to minimize impact. If failures are systematic, run the QC board and System Health Check and contact technical support if issues persist.

Inconsistent sgRNA spotting
After dispensing, visually confirm that all expected droplets are present, centered within electrode boundaries, and free of satellite droplets. Include PGA in the guide RNA mixture to prevent polydispersal. For missing droplets, check the dispenser run report and re-dispense if needed. For poor meniscus formation, maintain a minimum of 5 µL per well, spin the plate briefly, and tap gently on a flat surface. Work on ice and seal plates when not in use to prevent evaporation.

Reduced editing efficiency
If reduced editing efficiency is observed, troubleshoot in the following order: (1) validate delivery conditions using EGFP mRNA before proceeding with RNP; (2) confirm measured QC current matches expected values; (3) titrate cell density, guide RNA, and Cas9 concentration; (4) confirm high cell viability immediately before loading; (5) verify reagent integrity and storage conditions; (6) confirm the readout method is appropriate for the cell density and target of interest.

Adapting the platform to alternative cell types and models
For other primary immune cell types, such as NK cells or regulatory T cells, electroporation parameters will require optimization. Cell size is an important determinant of the electric field threshold required for membrane permeabilization: smaller cells require higher field strengths, while larger cells are permeabilized at lower voltages—a well-documented relationship in electroporation biophysics. As a starting point for cell types smaller than activated CD3⁺ T cells, maintain or slightly increase voltage (500–550 V) while keeping pulse duration short (2–3 ms) to avoid excess thermal stress. For larger cells, reduce voltage to 350–400 V, and modulate pulse duration without exceeding 10 ms. Additionally, micro-optimization at the donor level is common practice in electroporation, and parameters validated in healthy donor cells may not translate directly to diseased or patient-derived material. Cells from donors with hematological malignancies, chronic infection, or immunodeficiency frequently exhibit altered biophysical properties, and editing efficiency and viability should be re-evaluated for each new donor, particularly when working with non-healthy or clinical samples. Cell viability should be assessed at 24 h and 72 h post-electroporation and parameters adjusted iteratively. Cas9 dosage may also need to be increased by 1.5–2× relative to the T-cell protocol if RNP delivery efficiency is reduced.

For larger or denser 3D models such as organoids exceeding 200 µm in diameter, the gasket in the cartridge may require modification to accommodate increased structure size and diffusion barriers. Users attempting such applications should consider pre-dissociating outer cell layers to improve reagent penetration or applying multiple pulse conditions in sequence. Contact the corresponding author for guidance on adapting droplet geometry and pulse parameters for structures beyond the 100 µm spheroid diameter validated here.

Translational relevance
The translational potential of this workflow extends beyond simple knockout. The platform has been applied to homology-directed repair knock-in of a 4 kb GFP construct and a functional anti-HER2 CAR at the TRAC locus in primary T cells, as well as to a 45-target arrayed CRISPR screen identifying novel regulators of CD4⁺ T cell exhaustion⁹. These applications illustrate a path from protocol-level optimization to functional discovery in clinically relevant primary cell models. Nonetheless, important limitations remain: functional outcomes such as allelic disruption rates by sequencing, cytokine secretion, and in vivo engraftment were not assessed in the current protocol and represent necessary next steps for therapeutic translation. Additionally, while the platform supports 48 parallel reactions per cartridge, multi-cartridge parallelization will be required for larger screens.

Limitations
As with any miniaturized platform, certain practical constraints should be acknowledged. Downstream phenotyping must be adapted to the small reaction scale; flow cytometry acquisition may require additional concentration steps or pooling of replicate wells to achieve sufficient event counts. The platform's electroporation settings may also require optimization for cell types with high sensitivity to electric fields, and users should treat published parameters as starting points rather than universally fixed values. Finally, formal viability staining at extended post-electroporation time points was not performed for the spheroid workflow presented here; diameter-based growth tracking was used as a surrogate measure of structural integrity and represents an important direction for future characterization.

Overall, DMF-ection, a non-viral transfection method, provides a powerful and flexible approach for genome editing and mRNA delivery in low-input and high-throughput settings. Its compatibility with primary T cells, structured 3D models, and automated liquid handling expands opportunities in cell therapy engineering, multi-condition CRISPR screening, and miniaturized functional genomics.

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
12 Channel VOYAGER Adjustable Tip Spacing Pipette	Integra	4732	Instrumentation
1x PBS	Gibco		Flow Buffer
96 Well Plate, Sphera Low-Attachment Surface	Thermo Fisher	174927	Spheroid protocol
96-well conical-bottom plate	Sarstedt	82.1583.001	Spheroid protocol
Alexa Fluor 647 anti-human CD4 Antibody	Biolegend	357421	1 in 200, Clone A161A1
Attune Flow Cytometer	ThermoFisher		Instrumentation
DMEM Media	Gibco	10564011	Media Supplement- Spheroid culture
DMF compatible buffer	DropGenie		Transfection
DropGenie Transfection System	DropGenie		Instrumentation
Echo Acoustic Dispenser 650 Series	Beckman Coulter Life Scienes
Echo Qualified 384-well Low Dead Volume	Beckman Coulter Life Scienes
EDTA	Invitrogen		Flow Buffer
EGFP mRNA	Trilink	L-7201
EVOS Fluorescent Microscope	ThermoFisher		Instrumentation
Fetal Bovine Serum	Gibco	16000044	Media Supplement- Spheroid culture
Fida 1 instrument and Fida Neo 480 nm detector	Fida Biosystems ApS		Instrumentation
Ghost Dye Violet 510	Cytek Biosciences	13-0870
HEK293T Cells	ATCC		Spheroid protocol
Human IL-2 Recombinant Protein,	Peprotech	200-02 50ug	T cell media supplement
Human Primary Pan CD3+ T Cells	All Cells		Peripheral Blood, Cryopreserved, pan CD3+ Helper T Cells, Negatively Selected
Immunocult CD3/CD28 Activator	StemCell Technologies	10970	T Cell Activation media
Immunocult Expansion Media	StemCell Technologies	10981	T Cell Activation media
Incucyte Live Cell Analysis System	Sartorius		Instrumentation
INTEGRA ASSIST PLUS pipetting robot	Integra	4505, 4-Position Portrait Deck (PN 4521),	Instrumentation
non targeting synthetic guide RNA	Synthego		5’ GCACTACCAGAGCTAACTCA 3'
NucleoCounter NC-202	Chemometec
PE anti-human TCR α/β Recombinant Antibody	Biolegend	380805	1 in 200, Clone QA20B12
PE-Cy 7 Anti-Human CD8	BD Pharmingen	557750	1 in 200, Clone RPA-T8
Penicillin-Streptomycin (10,000 U/mL)	Gibco	15140122	Media Supplement- Spheroid culture
Poly-L-glutamic Acid (PGA)	Sigma Aldrich	P4761-100MG	Use a at 100mg/ml
Prism version 8.0.0	GraphPad Software, Inc.		Software
sNLS-SpCas9-sNLS Nuclease	IDT	10017687
Spheroid Microplate	Corning	3830	Spheroid protocol
Surfactant F	DropGenie		Transfection- use at 1:20 in DMF compatible buffer to make compete Transfection buffer
TRAC synthetic guide RNA	Synthego		5’ AGAGTCTCTCAGCTGGTACA 3'
Trypsin-EDTA (0.05%), phenol red	Gibco	25-300-062	Spheroid protocol