A microfluidic platform for the co-culturing of microtissues with continuously recirculating suspension cells

Abstract

In vitro evaluation of novel therapeutic approaches often fails to reliably predict efficacy and toxicity, especially when recapitulating conditions involving recirculating cells. Current testing strategies are often based on static co-culturing of cells in suspension and 3D tissue models, where cell sedimentation on the target tissue can occur. The observed effects may then mostly be a consequence of sedimentation and of the corresponding forced cell-tissue interactions. The realization of continuous medium flow helps to better recapitulate physiological conditions and cell-tissue interactions. To tackle current limitations of perfused organ-on-chip approaches, we developed a microfluidic chip and operation concept, which prevents undesired sedimentation and accumulation of suspended cells during multiple days by relying on gravity-driven perfusion. Our platform, which we termed "human immune flow (hiFlow) chip", enables to co-culture cells in suspension with up to 7 preformed microtissue models. Here, we present the design principle and operation of the platform, and we validate its performance by culturing cells and microtissues of a variety of different origins. Cells and tissues could be monitored on chip via high-resolution microscopy, while cell suspensions and microtissues could be easily retrieved for off-chip analysis. Our results demonstrate that primary immune cells and a range of different spheroid models of healthy and diseased tissues can be maintained for over 6 days on chip. As proof-of-concept cell-tissue interaction assay, we used an antibody treatment against diffuse midline glioma, a highly aggressive pediatric tumor. We are confident that our platform will help to increase the prediction power of in vitro preclinical testing of novel therapeutics that rely on the interaction of circulating cells with organ tissues.

Fig. 1

The hiFlow in vitro platform. a Schematic of the hiFlow chip (backside view), with the components of the microfluidic network highlighted in the design. Each chip hosts two microfluidic networks and features standard microscopy-slide dimensions. b Overview of the cell-interaction chambers, in which flowing suspended cells can interact with microtissue spheroids. The chamber features 7 microtissue compartments (in the image, 4 microtissues in the corresponding compartments are shown), while suspended cells flow across the chamber. c Close-up view of a cross-section of a microtissue compartment. A funnel structure with a top hydrophobic rim enables the loading of microtissues into the compartment. The compartment is surrounded by a 100 µm-high microfluidic channel. The shallow height of the channel prevents microtissues from escaping from the compartment. d Each polystyrene-based hiFlow chip was sealed with a pressure-sensitive-adhesive film to close the microfluidic channels. Four hiFlow chips (top view) can be loaded into a handling frame, which was designed so that microtissue loading ports and medium reservoirs were located at SLAS/ANSI standard positions. Tilting of the whole assembly generates a hydrostatic-pressure difference between the reservoirs and induces fluid flow in the microfluidic networks. e Up to four plates can be operated with a single tilter, resulting in the use of 32 units in parallel

Fig. 2

Flow rate and cell distribution in the hiFlow chip. a Measured flow rate in the microfluidic network when tilted at 85^o. The flow rate changes while the device is kept at the quasi-vertical tilting position due to a reduction of the hydrostatic pressure between the medium reservoirs, caused by medium flowing from the top to the bottom reservoir. N = 7 per time point, 4 independent chips. b Numerical simulation of the flow rate in the chip at steady state. The tilting angle of the platform is indicated in blue, the corresponding induced flow rate in red. c Finite-element-method simulation of the flow distribution in the interaction chamber. Microtissue compartments (shown as two concentric rings) induce a local reduction in the fluidic resistance, which promotes flow focusing into the compartments. d Cell distribution in the compartment at different time points. On the left, the picture shows fluorescent cells in the interaction chamber after 1 h of tilting. The 7 MT compartments are highlighted. On the right, the percentage of cells in the interaction chamber in each compartment after 1 h and 48 h of tilting is depicted. N = 4 per time point. e Chip operation. During perfusion, the hiFlow chip is tilted around the short axis, which results in a height difference of 27 mm between the reservoirs. For medium exchange, the chip is placed vertically, by tilting it around its long axis. This configuration results in a height difference of only 9 mm. The large width promotes cell sedimentation in the meandering part of the channel, so that cells are not removed during medium exchange from the lower reservoir. The flow rates (Q) at the tilted positions are reported in the figure. * indicates P ≤ 0.05, ** P ≤ 0.01, *** P ≤ 0.001, **** P ≤ 0.0001, calculated performing a Kruskal-Wallis test followed by multiple comparison analysis. For D), ε² = 0.56

Fig. 3

PBMC viability and population composition under static and perfused conditions. a PBMC viability in the hiFlow chip was measured under standard static conditions over 6 days. PBMCs from 3 different donors (illustrated by different colors) were tested (N = 3–18 hiFlow channels and N = 3–18 wells under static conditions for each donor). b t-SNE representation and clustering into main cell populations. Data from 12 individual experiments from 3 different donors were combined. All major cell populations were detected. c Ratio of cell populations, which were determined by t-SNE clustering after 0, 3, and 6 days of culturing. d Absolute cell densities of the most prevalent cell types (T_H cells, T_C cells, B cells, and others), assessed by flow cytometry. Cells were kept unstimulated or were stimulated with anti-CD3/CD28 antibodies (N = 2–3 hiFlow channels or 3–6 wells). e Relative frequency of CD69- and CD25-expressing cells in the hiFlow chip and on well plates under unstimulated or anti-CD3/CD28-stimulated conditions (N = 2–3 hiFlow channels or 3–6 wells)

Fig. 4

Microtissue viability and functionality on the hiFlow chip. a Bright-field micrographs of HCT-116 spheroids in the hiFlow chip (top) and in well plates (bottom) over six days of culturing. The pictures show microtissues that were formed by 600 cells/well at D-3. b Quantification of microtissue diameters on chip and under static conditions. N = 28 MTs for D0 to D3, N = 14 MTs for D3 to D6 in 3 hiFlow channels; N = 12 MTs for D0-D3, N = 6 MTs for D3-D6 under static conditions. c ATP quantification of the microtissues at D3 and D6. N = 14 MTs in 2 hiFlow channels, N = 6 MTs for static conditions. d Albumin secretion of the hLiMts under perfusion (hiFlow) and static conditions. N = 3 channels for hiFlow conditions, N = 5 MTs for static conditions. e LDH release for control and drug-treated microtissues in the hiFlow platform. N = 3–4 channels per condition. f ATP quantification at the end of the drug-treatment assay on chip. Each point indicates the ATP concentration for a single hLiMT. N = 20-28 MTs in 3-4 hiFlow channels. g Bright-field micrograph of hLiMTs on chips for control and drug-treated conditions. Scale bars = 200 µm. * indicates P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001, calculated by performing a multiple comparisons analysis test following an ANOVA. For (c) R² = 0.67, (d) R² = 0.58, for (e) R² = 0.98, for (f) R² = 0.98

Fig. 5

Interactions between brain cancer microtissues (DMG-MTs) and PBMCs in the hiFlow chip. a Experimental timeline of the experiment with handling steps for DMG MTs in red, PBMCs in blue, and chip preparation in gray. All components were combined on day 0, and endpoints were taken at day 3. b Expression of the surface protein disialoganglioside GD2 in the DMG cell line KISPIDMG-105, measured by flow cytometry. c Relative size change of MTs from day 0 to day 3 in well plates (static) and in the hiFlow chip (perfused), with and without unstimulated PBMCs (N = 19 static wells and N = 11–18 MTs in 2–3 hiFlow channels per condition). d Representative images of DMG MTs after 3 days in well plates (static) or in hiFlow chips (perfused), with and without unstimulated PBMCs. DMG cells were stained with CellTracker Green prior to MT formation, and PBMCs were stained with CellTrace Violet prior to loading into the chip. e Representative maximum intensity projection images of DMG MTs and stimulated PBMCs after 3 days of interaction in the hiFlow chip. Dinutuximab was administered as an ADCC alone or in combination with IL-2. Caspase 3/7 staining was added directly before confocal imaging. f Individual Z-sections at different heights at 20x magnification of a DMG MT that was treated with stimulated PBMCs, Dinutuximab, and IL-2. g Number of PBMCs attached to the MTs in the imaged volume (N = 20–21 MTs in 3 hiFlow channels per condition). h Number of tumor cells forming the DMG MT in the imaged volume (N = 20–21 MTs in 3 hiFlow channels per condition). i ATP measurements of individual MTs after 3 days in the hiFlow chip (N = 11–19 MTs from 3 hiFlow channels per condition). * indicates P ≤ 0.05, ** P ≤ 0.01, *** P ≤ 0.001, **** P ≤ 0.0001, calculated by performing an ANOVA test, followed by a multiple comparisons analysis. For (g) R² = 0.60, for (h) R² = 0.59, for (i) R² = 0.72