Construction of multiple concentration gradients for single-cell level drug screening

The rapid development of drug screening techniques presents a viable solution to preventing infections and treating human diseases1. Due to patient participation in preclinical validation and clinical trials, new drug development has become a costly, risky, and time-consuming project2-4. The pharmaceutical industry is confronted with many difficulties, including escalating costs and protracted timelines for the development of new drugs5-7. Additionally, in genetically identical populations, the heterogeneity of individual cells is crucial for cell growth and development. Phenotypic heterogeneity among genetically identical cells plays a crucial role in tumor metastasis, drug resistance, and stem cell differentiation8-10. Single-cell isolation and manipulation strategies are of considerable importance in revealing cell heterogeneity, disease diagnosis, drug delivery, and cancer biology11-13. Therefore, it is essential to thoroughly analyze single cells, from their survival state to their lysis state14. To fully understand the heterogeneity of cells, it is necessary to use traditional biological tools, such as Petri dishes and porous plates, to perform diversified operations and comprehensive analysis of cells at the single-cell level. However, there exist many difficulties and challenges in the process and evaluation of single cells of small size15-17. Multicomponent, high-sensitivity detection and high-throughput analysis of a large number of individual cells remain key challenges in achieving this goal.

As one of the most representative microanalysis platforms in this century, microfluidic chip technology has become a popular research topic because of its advantages of low reagent consumption, integrality, easy control, and good biocompatibility11,18. Among microfluidic platforms, microfluid-based systems for single-cell studies offer a powerful approach as the study of cell population heterogeneity progresses19,20. Compared with traditional petri dish or orifice plate experiments, studies on microfluidic single cells offer many advantages, such as high throughput, small sample size, automatic sample processing, and low contamination risk, making microfluidics an ideal technology for single-cell analysis to reveal key information about cellular heterogeneity that is often obscured in traditional ensemble measurements11,17.

Microfluidic chips are a powerful tool for gradient generation due to their advantages of high throughput and low consumption21-23. These systems have been developed for reagent mixing and drug screening by way of producing different species concentrations without manual pipetting24-26. Drug screening and treatment optimization otherwise can require the study of dose-dependent cellular responses at different drug concentrations, so concentration-gradient microfluidic chips have become a powerful tool in this field27-29. Their miniature size allows for parallelization with a minimal sample requirement, which is critical for high-throughput drug screening30-32. In addition, concentration-gradient microfluidic chips can be used for quantitative and large-scale assessments of toxicity and optimal concentrations of different drugs, which can not only improve the throughput and reduce the experimental cost but also rapidly and accurately control and interact with gradients at a higher resolution33,34. At present, microfluidic chips mostly use double or multiple-concentration gradients to study the entire cell population, conditions which cannot simultaneously study the single effect and interaction of two drugs on tumor cell heterogeneity at the single-cell level35,36. This paradigm may lead to a misunderstanding of tumor cell heterogeneity and the loss of key information. The establishment of multifunctional, single-cell drug screening integrated microfluidics systems based on multiconcentration gradients to predict drug synergies and optimal dosages professionally remains an urgent and challenging task. It is of great significance to provide guidance for determining rational drug combinations in clinical applications.

Our previous work has proven that the unique inertial microfluidic method can effectively reduce flow dependence and quickly build a stable and controllable multiconcentration gradient microfluidic device37. After, a device was also established for multistage microfluidics that was used for the isolation and capture of single cells based on cell size and deformability38,39. Based on this, we optimized the microfluidic device to achieve the formation of a higher flux concentration gradient, and by combining it with a single-cell capture array, we successfully constructed a single-cell microfluidic drug screening platform. In this study, the chemical gradients produced by a concentration-gradient generator were theoretically processed, calculated, and verified in a fluorescence experiment. Subsequently, cisplatin and 5-fluorouracil were used as model drugs to perform single or multidrug combination chemotherapy on human breast carcinoma (MCF-7) cells and human hepatoma (HepG2) cells at the single-cell level. Then, the applications of the system in gradient construction, single-cell capture, cell culture, and single-cell analysis were demonstrated. The results show that the developed device can explore the heterogeneity of tumor cells under multiple drug gradients with the required stability and high-throughput capability. We propose that this system provides a flexible and controllable platform for the study of pharmacological functions and other fields involving concentration gradients and single-cell analytical operations.

Materials and methods described an integrated microfluidic platform fabrication, simulation, setup, cell handling, and imaging protocols.

Device fabrication:

• Microfluidic devices were fabricated via soft lithography using an AZ 50XT master mold on silicon. Device structures were designed in AutoCAD, printed to form a chrome mask, and patterned under UV using AZ 50XT photoresist on a BG401A mask aligner (7 mW cm−2).
• Molds were silanized with trimethylchlorosilane vapor for 3 min before casting.
• PDMS (RTV 615) base and curing agent mixed 10:1 (w/w) were poured onto the mold to yield ~3 mm thick PDMS replicas, degassed, and cured at 80 °C for 50 min, then peeled and inlet/outlet holes punched.
• For bonding, a thin PDMS prepolymer layer [RTV 615 A:B = 20:1 (w/w)] was spin-coated on a clean glass slide (3000 rpm, 60 s, 15 s ramp), the PDMS replica was placed on it, and cured at 80 °C for 20 min; final bake at 80 °C for 48 h readied the device.

Numerical simulation:

• CFD simulations were run in ESI-CFD (CFD-ACE+) using FLOW and CHEM modules with finite volume method.
• Inlets had specified flow rates; outlet had fixed pressure; no-slip at walls. Water properties: density ρ = 1000 kg m−3, dynamic viscosity μ = 10−3 kg m−1 s−1.
• Species diffusion coefficient D = 1×10−10 m2 s−1; two fluids (water A=0, B=1) used for mixing simulations. Second-order limiting scheme for species diffusion; mass fraction convergence 10−6; ~2000 time steps to reach outlet.

Experimental setup:

• 1 mL disposable syringes on three Longer LSP04-4A syringe pumps controlled flows via 25 cm Tygon tubing (ID 0.42 mm).
• Chips were UV-irradiated 2 h, rinsed with 70% ethanol (2–3 min, 10 µL/min), then ultrapure water and PBS before use; identical flow rates were applied to the three inlets during tests.

Cell culture, capture, and treatment:

• Human cancer cell lines MCF-7 and HepG2 were cultured in DMEM supplemented with 10% PBS (as stated), 100 U/mL penicillin, and 100 µg/mL streptomycin; maintained at 37 °C, 5% CO2; passaged 1:3 every 3 days; harvested by 0.25% trypsin in Ca2+/Mg2+-free HBSS at 37 °C; trypsinization stopped with fresh DMEM; cells centrifuged (1000 rpm, 2 min) and diluted to required concentrations.
• For single-cell capture devices, chips were sterilized by sequential flushing with 75% ethanol, ultrapure water, PBS (0.01 M, pH 7.4), and fresh DMEM, then coated with fibronectin (FN, 50 µg/mL) for 2 h at 37 °C to promote adherence. Chambers were soaked with serum-free DMEM for 1 h to remove excess protein.
• Cells were seeded at 1.0×10^5 cells/mL at 200 µL/min. Drug solutions: DMEM with 5-FU (100 µM), DMEM with cisplatin/DDP (10 µM), and drug-free DMEM were continuously injected through the inlets. Single-cell viability was assessed after 2 h of stimulation.

Microscopy and image analysis:

• Imaging was performed on an Olympus CKX41 inverted fluorescence microscope with a DP72 CCD camera; image processing with Image-Pro Plus 6.0; data analysis with Origin 9.

Device design specifics:

• Integrated platform composed of a concentration-gradient generator and 24 single-cell capture devices.
• Gradient generator: fluid layer + thin PDMS layer + glass slide; flow layer contained 42 Tai Chi-spiral mixers (50 µm width, 50 µm height), 24 liquid storage chambers (1200 µm width × 2200 µm length × 50 µm height), three sample inlets, and 24 outlets. Three inlets allow formation of single-drug or two-drug combination gradients via spread-mixing of the spiral mixers.
• Single-cell capture device: five sets of capture matrix columns in a 2D array; each matrix has 180–205 capture units (25 µm height). Each unit comprises two adjacent H-shaped microstructures forming a single-cell capture unit with two minimum pores, the first pore being 2 µm wider than the second; array spacing decreases along the matrix. Separate portals for cell suspension and reagents enable sequential capture and stimulation.

Gradient formation validation:

• Numerical simulations and fluorescence-based experiments (injecting luciferin and PBS via three entrances) assessed concentration distribution across 24 chambers. Fluorescence images were quantified to extract concentration data; means and standard deviations computed.
• Dean flow behavior in the Tai Chi-spiral mixers was analyzed via simulation to explain mixing performance across flow rates, including alternating clockwise/anticlockwise vortices at S-shaped junctions that enhance mixing.

• The integrated microfluidic device successfully combines a three-inlet concentration-gradient generator with 24 downstream single-cell capture devices, enabling high-throughput, single-cell-level drug screening.
• The Tai Chi-spiral mixer design rapidly accelerates mixing and yields linear, stable concentration gradients that are insensitive to flow rate within the tested range.
• Numerical simulations showed that three sets of identical concentration gradients can be established simultaneously across 24 liquid storage chambers, with drug distributions mapped to specific chamber ranges (e.g., drug A across chambers 1–8 and 18–24; drug B across 2–16; drug C across 10–24).
• Fluorescence experiments (luciferin/PBS) confirmed simulation predictions: concentration profiles across chambers matched simulations and showed a strong linear relationship between measured fluorescence intensity and expected concentration; gradient profiles remained consistent across different flow rates, indicating robust mixing.
• Dean flow within the spiral mixers strengthens with increasing flow rate; alternating vortices at S-shaped junctions further improve mixing by promoting rapid transverse transport and interfacial contact, supporting uniform fluid splitting and efficient mixing over a wide range of flow rates.
• The single-cell capture array (H-shaped microstructures; 180–205 units per matrix) effectively isolates single cells based on size and deformability and enables controlled exposure to generated gradients.
• Drug testing on MCF-7 and HepG2 single cells using 5-fluorouracil (initial 100 µM) and cisplatin (initial 10 µM) demonstrated: both agents inhibit cancer cell growth; combination therapy was more effective for human hepatoma (HepG2) cells at the single-cell level.
• Analysis of biomechanical heterogeneity revealed that small and/or more deformable tumor cells exhibited greater drug resistance than larger and/or less deformable cells.
• The platform supports simultaneous evaluation of single-agent and combination dosing in defined percentage steps (e.g., 0–100% in 12.5% increments for 5-FU across chamber series), facilitating dose–response assessment at single-cell resolution.

The study addresses the challenge of accurately predicting drug performance by enabling simultaneous generation of multiple, well-defined concentration gradients and coupling them to a high-throughput single-cell capture array. By leveraging inertial microfluidics and Dean flow in Tai Chi-spiral mixers, the device achieves rapid, robust, and flow-rate-insensitive mixing, producing linear gradients across 24 chambers. This capacity allows parallel assessment of single-drug and two-drug combination effects on individual cells, directly tackling cellular heterogeneity that is often masked in bulk assays.

The experimental validation demonstrates close agreement between simulations and fluorescence measurements, confirming the reliability of gradient formation. Application to MCF-7 and HepG2 cells shows that both 5-FU and DDP inhibit growth, with enhanced efficacy of the combination for HepG2 cells. Importantly, correlating cell biomechanical properties (size and deformability) with drug response at single-cell resolution highlights subpopulations with higher resistance (small and/or deformable cells), providing mechanistic insight into heterogeneity-driven treatment outcomes.

Overall, the platform’s integrated design enables high-throughput, automated, and low-consumption single-cell pharmacological testing, informing optimal dosing and potential synergies, and offering a route to more precise screening of chemotherapy regimens.

This work presents a flexible, controllable microfluidic platform that integrates a three-inlet Tai Chi-spiral mixer-based gradient generator with 24 single-cell capture devices to conduct single-cell drug screening across multiple concentration gradients. The system produces stable, linear, and flow-rate-insensitive gradients validated by simulation and fluorescence experiments. Applied to MCF-7 and HepG2 cells, both 5-FU and cisplatin were effective, with combination treatment exhibiting superior efficacy in HepG2 cells. Single-cell analyses further linked biomechanical heterogeneity to drug resistance, identifying small and/or deformable cells as more resistant.

The platform provides a simple and reliable means to explore optimal dosing and drug combinations at single-cell resolution and can be extended to broader pharmacological studies involving concentration gradients and single-cell analyses. Future work could apply this system to additional drug candidates and complex combination regimens to guide rational chemotherapy strategies and further dissect heterogeneity-driven responses.