Generation of tumor spheroids using a droplet-based microfluidic device for photothermal therapy

Monolayer (2D) cell cultures often overestimate drug efficacy because they lack realistic cell–cell and cell–matrix interactions and physico-biochemical barriers. Three-dimensional (3D) spheroid cultures better recapitulate these features, as illustrated by substantially higher IC50 values (e.g., paclitaxel is two orders of magnitude higher in 3D vs. 2D). Traditional spheroid formation methods (pellet culture, liquid overlay, hanging drop, spinning flasks, magnetic levitation) produce spheroids reliably but suffer from broad size distributions, laborious handling, low throughput, and static conditions that deplete oxygen and nutrients. Microfluidic approaches—microwells, microtraps, and droplet-based systems—address many of these limitations by enabling uniform size control, automation, high throughput, and dynamic media exchange. Droplet-based microfluidics, in particular, allows precise control of spheroid size via droplet volume and cell concentration, and high-frequency generation of homogeneous droplets for cell encapsulation. In parallel, graphene oxide-based nanomaterials are promising for cancer therapy, including photothermal therapy (PTT), especially when modified with polyethylene glycol (PEG) to enhance stability and reduce aggregation, and with branched polyethyleneimine (BPEI) to enable combination therapies. A robust 3D spheroid platform is valuable for evaluating such nanomaterials. This study presents a droplet-based microfluidic system for high-frequency, uniform tumor spheroid generation and demonstrates its use for evaluating rGO-BPEI-PEG nanocomposites in PTT; it also generates uniform NSC-derived neurospheres for neurotoxicity studies.

Device and spheroid generation: A droplet-based microfluidic chip with two inlets (oil phase and aqueous cell phase) and a flow-focusing junction was used to generate aqueous droplets containing U87MG glioblastoma cells via shear forces. Channel diameter was 300 µm; flow rates were set so droplet diameter stayed below channel diameter to avoid wall contact. Droplet size was tuned by varying the fluorinated oil flow rate: increasing oil flow decreased droplet diameter. Measured droplet diameters ranged from 180.8 ± 2.9 µm to 305.9 ± 8.1 µm at a constant aqueous flow of 1 µL/min. Generated droplets were transferred via tubing to well plates and incubated for 2 days to form spheroids. Spheroid classification used diameter (D) and shape index (ShI) metrics to distinguish spheroids (D > 50 µm, ShI > 0.7), aggregates (D > 50 µm, ShI < 0.7), cell units (9 < D < 50 µm), and empty droplets (D < 9 µm). The number of cells per droplet was controlled by (i) droplet volume via continuous phase flow rate and (ii) cell concentration in the dispersed phase. Using 2×10^6 to 6×10^6 cells/mL, spheroid diameters of 98.6 ± 1.0 to 126.4 ± 4.9 µm were obtained. Maximum droplet generation frequency was ~70 Hz at oil 50 µL/min, enabling ~42,000 droplets in 10 min. Immunostaining for E-cadherin confirmed formation of cell–cell junctions indicative of true spheroids. Neurospheres: Uniform NSC-derived neurospheres were also generated in the same device; neurite outgrowth was assessed and shown to be modulated by neurotoxins. Nanocomposite synthesis and characterization: GO-COOH nanosheets were chemically modified: BPEI was conjugated to GO via EDC/NHS chemistry, followed by PEGylation, and chemical reduction (hydrazine monohydrate) to obtain rGO-BPEI-PEG. AFM assessed size/morphology: GO-COOH nanosheets ~100–150 nm; rGO-BPEI-PEG particles ~50–60 nm with round morphology, attributed to folding/reforming during conjugation. FT-IR confirmed functional group changes: GO-COOH showed O–H (~3361 cm^-1) and C=O (1728 cm^-1); GO-BPEI introduced amide (1630–1695 cm^-1), C–H (~2900 cm^-1), N–H (670, 3250 cm^-1); PEG introduced –CH2/–CH3 (1466/1340 cm^-1) and C–O–C (1097, 960 cm^-1). Reduced form showed diminished O–H band, indicating successful reduction. UV–Vis spectroscopy showed increased absorbance for rGO-BPEI-PEG relative to GO-COOH and GO-BPEI-PEG, consistent with restoration of conjugated aromatic domains. Zeta potential measurements: GO-COOH −41 mV; after BPEI and PEG conjugation, positive values (approximately +30 mV and +43 mV, respectively); rGO-BPEI-PEG +28 mV. Photothermal therapy evaluation: U87MG tumor spheroids were incubated with rGO-BPEI-PEG at 20, 40, and 60 µg/mL for 4 h and irradiated with an 808 nm NIR laser. Cell viability was measured by CCK-8; temperature rise under NIR was monitored over 10 min across concentrations. Control groups without nanocomposites and/or without NIR were included.

- High-throughput, uniform droplet and spheroid generation: droplet diameters tunable from 180.8 ± 2.9 µm to 305.9 ± 8.1 µm by oil flow rate; spheroid diameters tunable from 98.6 ± 1.0 µm to 126.4 ± 4.9 µm by adjusting cell concentration (2×10^6 to 6×10^6 cells/mL) and/or droplet volume. - Throughput: ~70 Hz droplet generation at oil 50 µL/min, enabling ~42,000 droplets in 10 min; approximately 20% higher generation yield than previously reported work. - Encapsulation outcome: ~80% of droplets contained exactly one spheroid; no droplets contained more than one spheroid. - Spheroid identity: E-cadherin immunostaining showed strong cell–cell junction expression in tumor spheroids. - Nanocomposite characterization: rGO-BPEI-PEG formed round ~50–60 nm particles (vs. GO-COOH nanosheets ~100–150 nm); FT-IR confirmed conjugation and reduction; UV–Vis absorbance increased after reduction; zeta potentials: GO-COOH −41 mV; GO-BPEI ~+30 mV; GO-BPEI-PEG ~+43 mV; rGO-BPEI-PEG +28 mV. - Photothermal response: NIR irradiation of rGO-BPEI-PEG solutions produced concentration- and time-dependent temperature increases over 10 min; NIR plus rGO-BPEI-PEG reduced U87MG spheroid viability compared to controls without NIR or without nanocomposite (qualitative per reported figures).

The droplet-based microfluidic system addresses key limitations of conventional 3D culture by enabling automated, high-frequency generation of uniform spheroids with minimal manual handling. Tunable droplet and spheroid sizes allow precise control over 3D model dimensions, improving reproducibility in drug testing. The confirmation of E-cadherin-mediated junctions supports that the structures are bona fide spheroids, better modeling in vivo-like cell–cell interactions. Demonstrating PTT on U87MG spheroids with rGO-BPEI-PEG validates the platform for evaluating nanomaterial-based therapies in a 3D context that better reflects in vivo response. The observed photothermal heating and reduced viability under NIR in the presence of rGO-BPEI-PEG indicate effective PTT action. Extending the platform to generate uniform NSC-derived neurospheres and showing neurite outgrowth modulation by neurotoxins highlights the broader applicability for neurotoxicity screening and neuroscience research.

This work presents a droplet-based microfluidic device capable of automated, high-throughput generation of uniform 3D tumor spheroids and NSC-derived neurospheres. Spheroid size is readily controlled via flow conditions and cell concentration, and the system achieves high generation frequencies (~70 Hz). The platform’s utility for therapeutic evaluation is demonstrated by characterizing rGO-BPEI-PEG nanocomposites and validating their photothermal effects on brain tumor spheroids under NIR irradiation. These results support the device as a powerful tool for photothermal therapy assessment and broader drug screening in physiologically relevant 3D models. Future work could integrate on-chip culture and analysis modules for fully continuous workflows, expand to co-culture and core–shell spheroids, and systematically compare therapeutic responses across multiple cancer types and nanomaterial formulations.