Automated microfluidic platform for dynamic and combinatorial drug screening of tumor organoids

Abstract

Three-dimensional (3D) cell culture technologies, such as organoids, are physiologically relevant models for basic and clinical applications. Automated microfluidics offers advantages in high-throughput and precision analysis of cells but is not yet compatible with organoids. Here, we present an automated, high-throughput, microfluidic 3D organoid culture and analysis system to facilitate preclinical research and personalized therapies. Our system provides combinatorial and dynamic drug treatments to hundreds of cultures and enables real-time analysis of organoids. We validate our system by performing individual, combinatorial, and sequential drug screens on human-derived pancreatic tumor organoids. We observe significant differences in the response of individual patient-based organoids to drug treatments and find that temporally-modified drug treatments can be more effective than constant-dose monotherapy or combination therapy in vitro. This integrated platform advances organoids models to screen and mirror real patient treatment courses with potential to facilitate treatment decisions for personalized therapy.

Fig. 1. Automated microfluidic 3D cellular and organoid culture platform for dynamical drug perturbations.

a A programmable membrane-valve-based microfluidic chip (multiplexer control device) provides automated stimulation profiles to various chambers of a separate 3D culture platform (b) to produce many parallel and dynamical culture experiments. b, c The 3D culture chamber platform contains 200 individual chambers that are compatible with temperature-sensitive gels (i.e., Matrigel), and an overlaying channel layer enables 20 independent fluidic conditions (scale bar 100 μm). The channel layer is reversibly clamped on top of the chamber layer to provide media and other chemical stimulation without leakage. c A cross-section of the two-layer multichambered PDMS-based 3D culture chamber device. d 30 chemical inputs and 30 outlets of the multiplexer control device (a) are preprogrammed to provide combinatorial and time-varying stimulations to the 3D culture chamber device (b). e, f Organoids or 3D cellular structures are continuously observed through time-lapse imaging for quantification; fluidic culture conditions can be changed on demand. The 3D culture chamber device can also be disassembled for cell harvesting and further cellular assays.

Fig. 2. Human tumor organoid culture and growth on microfluidic platform.

a On-platform growth of organoids: three separate patient-derived pancreatic ductal adenocarcinoma (PDAC) organoids in Matrigel from single cells to formation of differentiated morphology of complex 3D structures (scale bar 100 μm). b Organoids from two patients were grown in parallel in a traditional 24-well plate and on our microfluidic platform (scale bar 100 μm). After mature organoid formation, organoids were harvested, H&E stained, and their morphologies compared and analyzed. In both platforms, organoids from patient 1 exhibited back-to-back glands with a high degree of nuclear atypia and pleomorphism with an accumulation of apoptotic luminal necrotic cells. Organoids from patient 4 demonstrated a well-differentiated morphology with simple spherical organoids and uniform nuclear and cytoplasmic features with little or no accumulation of necrotic luminal cells. c Organoid growth curves of PDAC organoid samples derived from three patients grown from single cells for 7 days on the platform. Each dot represents the cross-sectional area of an individual organoid. Patient 1 (blue), Patient 2 (red), Patient 3 (green). d Long-term culture, growth, and fluorescent staining of fixed PDAC organoids on the platform. Nuclei staining (DAPI) and F-actin (Phalloidin) are demonstrated (scale bar 100 μm). e Multiple Z image slices or stacks of a group of fixed and fluorescently stained organoids with DAPI and phalloidin (scale bar 100 μm).

Fig. 3. Combinatorial drug treatment of human tumor organoids on microfluidic platform.

a On-platform drug treatment and stimulation with continuous fluorescence and phase imaging of organoids for the treatment duration. Each color represents a different drug formulation. Drug treatments on each channel can be changed on demand, creating time-varying drug treatments. Organoids can be analyzed for growth, morphology changes, or death. b Representative images (10×) of gemcitabine (100 nM) treated organoids for a 4-h drug pulse followed by normal growth media, continuous treatment of paclitaxel (10 nM) for 72-h, continuous treatment of gemcitabine (100 nM) for 72-h, a combination dose of gemcitabine (100 nM) + paclitaxel (10 nM) for 72-h, and negative and positive controls (staurosporine 10 mM). Caspase 3/7 reagent (green) used for apoptosis detection and propidium iodide (red) for dead cells along with phase contrast images (scale bar 100 μm). c Average caspase 3/7 signal over 72-h period of continuous single drug treatments for patient 1. d Average caspase 3/7 signal over 72-h period for a 4-h pulse of a single drug treatment followed by normal growth media for patient 1. e, f 72-h (e) and 4-h (f) drug treatments similarly examined for multiple known combinations of drugs. c–f All data presented as mean values ± SEM, n = 3, and normalized to positive control. Overall, combination chemotherapy treatment resulted in significantly increased apoptosis in tumor organoids compared to single drug treatments as expected. Source data for panels e, f are available.

Fig. 4. Sequential and temporal drug treatment on the microfluidic platform reveals the efficacy of dynamic temporal drug treatment for personalized therapy.

a Schematic of sequential drug delivery schedules of single drugs delivered temporally in pulses to recapitulate dynamic combination chemotherapy with the platform. Colors in each row represent a different drug formulation, which can be changed on-demand. b–h Comparison of the temporal delivery for five combination chemotherapies using average caspase 3/7 signal to detect apoptosis for patient 1. All data presented as mean values ± SEM, n = 3, and normalized to positive control. c–g Comparison of temporal delivery for each of the five combination chemotherapies to their 72-h and 4-h constant delivery counterparts (i.e., all drugs in the sequence at once). Details of the drugs used in each therapy regimen are shown below the graph and described in more detail in Table 1. Time course of each drug on the x-axis is to scale. h Comparison of all investigated therapies at the end of the 72-h drug treatment period (asterisk denotes significant differences from temporal treatment, two-way ANOVA, p values from left to right; FOLFIRINOX: 8.3E-07, 7E-07; FOLFIRI: 1.7E-06, 7E-07; FOLFOX: 0.03, 0.01; Gem+5-FU: 3.9E-05, 8.3E-05; Gem+Pac: 9.6E-10, 6.16E-06). Sequentially-administered combination therapy is more efficient in inducing tumor cell death. Source data for panels c–g are available.

Fig. 5. High-throughput drug testing of multiple patients on chip.

a Heatmaps of organoids from three patients grown and stimulated in a simultaneous experiment. Cellular apoptosis (caspase 3/7) and death (propidium iodide) were recorded over the 72-h drug treatment period to assess drug sensitivity. b End-point analysis of average organoid cellular apoptosis and death for each patient and combination treatment group. All data presented as mean values ± SEM, n = 3, and normalized to positive control. Response comparisons between the different patients revealed distinct sensitivities to specific drug regimens across both independent cell viability assays. Source data for panels a, b are available.