Here, we present a protocol to coculture primary cells, tissue models and punch biopsies in a microfluidic multi-organ chip for up to 28 days. Human dermal microvascular endothelial cells, liver aggregates and skin biopsies were successfully combined in a common media circulation.
The ever growing amount of new substances released onto the market and the limited predictability of current in vitro test systems has led to a high need for new solutions for substance testing. Many drugs that have been removed from the market due to drug-induced liver injury released their toxic potential only after several doses of chronic testing in humans. However, a controlled microenvironment is pivotal for long-term multiple dosing experiments, as even minor alterations in extracellular conditions may greatly influence the cell physiology. We focused within our research program on the generation of a microengineered bioreactor, which can be dynamically perfused by an on-chip pump and combines at least two culture spaces for multi-organ applications. This circulatory system mimics the in vivo conditions of primary cell cultures better and assures a steadier, more quantifiable extracellular relay of signals to the cells.
For demonstration purposes, human liver equivalents, generated by aggregating differentiated HepaRG cells with human hepatic stellate cells in hanging drop plates, were cocultured with human skin punch biopsies for up to 28 days inside the microbioreactor. The use of cell culture inserts enables the skin to be cultured at an air-liquid interface, allowing topical substance exposure. The microbioreactor system is capable of supporting these cocultures at near physiologic fluid flow and volume-to-liquid ratios, ensuring stable and organotypic culture conditions. The possibility of long-term cultures enables the repeated exposure to substances. Furthermore, a vascularization of the microfluidic channel circuit using human dermal microvascular endothelial cells yields a physiologically more relevant vascular model.
Current monolayer or suspension cell culture assays for drug development are failing to emulate the human cellular microenvironment and, therefore, lead to a rapid dedifferentiation and loss of function in primary human cell cultures. Tissue models with higher physiological relevance are needed to predict the efficacy and safety of compounds before admitting them to clinical trials. Recently, standard in vitro cell culture techniques have evolved from two-dimensional monolayer cultures toward three-dimensional multi-cellular models, aiming to mimic the in vivo tissue microenvironment. These systems have already shown major improvements towards more accurate prediction of the mode of action of compounds 1,2. Furthermore, adapting the in vitro culture conditions to the highly specialized needs of cells is of particular interest.
Under standard in vitro conditions, a variety of important culture parameters, such as nutrient and oxygen supply, removal of accumulating products, and mechanical force acting on the cells often cannot be controlled thoroughly in most cases. Many organs possess physiologically relevant concentration gradients of substances and dissolved oxygen. However, those highly regulated and optimized conditions are in clear opposition to the uncontrollable diffusion gradients around tissues under in vitro conditions, leading to a highly unstable environment and limiting cellular development 3. Thus, steadier and especially more quantifiable in vitro conditions are required to keep cells viable and differentiated over prolonged periods of time. Perfused systems, where medium components are regularly removed and substituted, are often better characterized and controllable than static cultures concerning the direct surrounding of tissues. Under static conditions, diffusion gradients of cell secretions and culture medium nutrients might surround cultured cells 3. Introducing well-characterized medium flow rates around the tissues allow the cell secretions to mix with the rich medium through perfusion. This enables the generation of defined cellular microenvironments, ensuring a stable cellular phenotype and metabolizing enzyme expression throughout the whole assay duration 4.
Recent developments in multi-organ chip (MOC)-based systems combine the benefits of a controlled medium flow around engineered tissues with the small medium and cell mass requirements of microscale bioreactors, leading to a reduced amount of substance needed during testing. Several microfluidic systems for tissue culture have been described so far 5,6. Tissue-to-fluid ratios within these systems play an especially crucial role in simulating physiologically relevant cellular crosstalk. However, due to technical limitations, such as the use of external pumps and media reservoirs, the overall circulating media volume in most systems is too large compared to the tissue volumes. The group of Shuler et al. were the first to develop a system ensuring proper residence times of substances within the cell culture compartments and in vivo relevant tissue-to-fluid ratios 7,8. This was achieved by scaling the external reservoir down to a 96-well plate, also representing the “other tissues” compartment. In order to minimize the circulating media volume within our MOC platform, we integrated a peristaltic on-chip micropump, eliminating the need for external media circuits. This micropump is able to operate the system at a selectable number of media flow velocities and shear stress rates 9. A microfluidic channel system of 500 µm width and 100 µm height interconnects two standardized tissue culture spaces, each having the size of a single well of a 96-well plate. Adhering to the sizes of industry standard well plates allows the integration of already existing tissue models produced in the transwell format. Furthermore, the vertical position of transwell cell culture inserts is adjustable, enabling the cultivation of tissue models which are not only directly exposed to the fluid flow, but can also be lifted up and shielded from the underlying current. Similarly, air-liquid interface cultures are feasible using this system.
The MOC platform is fabricated from a polydimethylsiloxane (PDMS) layer 2 mm high and a glass microscope slide with a footprint of 75 x 25 mm2, which are permanently bonded by low pressure plasma oxidation to form the fluid-tight microfluidic circuit. The PDMS layer containing the respective channels and cell culture compartments is produced by standard soft lithography and replica molding 9. The microfluidic design of the MOC used during this study consisted of two separate microfluidic circuits per chip, each holding two cell culture compartments interconnected by a channel system 100 µm high. This allowed the performance of two individual two-tissue cocultures using one multi-organ chip. Pumping frequencies were adjusted to yield medium flow rates of 40 µl/min.
This two-tissue MOC design provided the capability to coculture a liver spheroid and a skin punch biopsy in separate culture spaces, albeit in a combined media circuit under physiological flow conditions. Differentiated HepaRG cells were aggregated together with human hepatic stellate cells (HHSteC) at a ratio of 24:1 to form homogenous spheroids. This ratio was found to be optimal, as observed in previous experiments 10, even though, nearly twice the number of hepatocytes were used compared to the in vivo situation. The skin was cultivated at an air-liquid interface inside a transwell cell culture insert, thus enabling topical substance exposure. These tissue models were cocultivated for 28 days in the MOC to demonstrate the comprehensiveness of this system. Furthermore, the microfluidic channel circuit of the chips was fully covered with human dermal microvascular endothelial cells (HDMEC) to more closely simulate the vascular system.
The MOC platform described here represents a stable and powerful tool for cultivating tissues of various origins at dynamic medium flow conditions over prolonged culture periods 10,16. In this example, the platform was used to cultivate primary cells (HDMEC), tissue equivalents generated from a cell line (liver aggregates), and a coculture of the aforementioned with a tissue biopsy. The MOC was able to sustain the three-tissue coculture for up to 28 days in a combined medium circuit. To the best of the authors’ knowledge, this is the first time a multi-tissue coculture including biopsies, primary cells and cell lines has been performed over four weeks.
One of the major drawbacks of microfluidic systems is the affinity of small molecules to adhere to the surface material of the fluidic circuit. As the surface to volume ratio is especially high in microfluidic systems, this effect becomes even more pronounced 17. The stable HDMEC coverage of the channels, introduced here, might act as a biological barrier preventing the adhesion of molecules to the MOC. Furthermore, it may serve as a hemocompatible vessel for whole blood circulation, preventing blood clotting. However, the use of whole blood as a medium substitution is not feasible as a full vascularization of the organ equivalents has not yet been achieved. Existing work on the vascularization of in vitro-generated tissues is promising and guides the way for further studies 18,19.
It is well-known that hepatocytes tend to lose their liver-specific functions over time under static two-dimensional in vitro culture conditions 20. Metabolizing enzymes, such as the cytochrome P450 family, are of especial importance if the metabolism of a certain drug is to be studied. Cytochrome P450 3A4, an enzyme related to the biotransformation of many xenobiotics, and cytochrome P450 7A, which is involved in bile acid synthesis, were expressed in liver aggregates cultured in the MOC over 14 days. This indicates the preservation of a metabolically active phenotype, allowing for drug metabolism studies. The increased albumin production rate of aggregates in the MOC compared to static cultures is an additional indication for adequate culture conditions. The albumin production rates observed during this study were comparable or even higher than previously reported values obtained by microfluidic chips including HepG2 cells 21–23, however, values did not reach those of primary human hepatocyte cultures 24. Furthermore, the MOC system, in its temporary layout, does not allow for a separate segregation of bile. Cells in the aggregate polarized and formed bile canaliculi-like structures, as shown by MRP-2 staining. However, those canaliculi were not connected to a technical channel collecting the bile. This non-physiological mixing of the bile with the blood compartment has to be addressed in a future redesign of the system.
The adjustment of flow characteristics is of high importance 25, especially with regard to tissues sensitive to shear stress, such as the liver. The amount of shear stress perceived by the tissue can be modified in two ways: Firstly, the air pressure used to push down the membranes of the pump can be lowered, decreasing peak shear stress values in the system. Secondly, the tissues can be embedded in an extracellular matrix layering or cultured in transwell culture inserts. The latter shield the tissues from the underlying current with a porous membrane. These adjustments need to be performed on an individual basis for each organ equivalent before starting the MOC experiment. At a pulsatile operation of 2.4 Hz, for example, which corresponds to a high, but still physiological, heart activity of 144 beats/min in humans, the shear stress measured in the channels of the microvascular circuit reaches approximately 25 dyn/cm2. This corresponds to a physiological shear stress at the higher end of the scale in microvasculature and is, therefore, well applicable for experiments including an endothelialization of the channels. However, as the current microfluidic layout of the MOC system presented consists of only one media circuit connecting the two organ compartments, one pumping velocity and shear stress rate has to be chosen for the whole system. Therefore, an exact adjustment of flow characteristics to the needs of each single organ is not always feasible.
Furthermore, care has to be taken in adjusting the cells to the common medium. Cells are cultivated in the MOC in a combined media circuit, therefore, no individual cell culture media can be used for each tissue model, as is the standard for in vitro cell culture. A minimal combined media formulation needs to be defined beforehand and the cells need to be adjusted stepwise to this new media. An adjustment procedure of 80%/20% old to new media for two days, then 50%/50% followed by 20%/80%, and a full exchange always led to a reasonable cell viability and functionality of cultures in our hands.
The current microfluidic layout of the MOC system allows the coculture of up to three tissues. A coculture of at least the ten most important organs of the human body is needed to reach homeostasis. Therefore, the system presented is able to predict specific tissue-tissue interactions, but not the true systemic response to a substance. A further development of the MOC to include more organ cavities is envisaged. Moreover, the validity of the system is to be shown using a set of reference compounds. Preferably, compounds that have failed during clinical trials (such as Troglitazone) are to be tested for their performance in the MOC. Whereas, a true validation of such complex systems is still impeded by the lack of standardization concerning biomarkers and endpoints for functional evaluation, collecting more data on the toxicological performance of this and similar systems will broaden their reliability and area of application.
The authors have nothing to disclose.
The work has been funded by the German Federal Ministry for Education and Research, GO-Bio Grant No. 0315569.
Name of Material/ Equipment | Company | Comments/Description | |
HepaRG cells | Biopredic International | undifferentiated cells | |
HHSteC | ScienCell Research Laboratories | cells and all culture supplements | |
HepaRG Medium | Sigma-Aldrich | William's Medium E 10% FCS 100 U/ml penicillin 100 µg/ml streptomycin 5 µg/ml human insulin 2 mM L-glutamine 5 x 10-5 M hydrocortisone hemisuccinate | |
HDMEC Medium | PromoCell | Endothelial Cell Growth Medium MV2 with Supplement-Pack MV2 and 1% penicillin-streptomycin | |
Dimethyl sulfoxide | Carl Roth | add 2% to HepaRG media | |
Trypsin/EDTA | Biowest | ||
Trypsininhibitor | Carl Roth | ||
MAXYMum Recovery Tips | Corning | 1000 µl Pipet Tips Wide Bore | |
384-Well Hanging Drop Plate | 3D Biomatrix | Perfecta 3D 384-Well Hanging Drop Plate | |
Tissue culture flasks | Corning | 75 cm2 | |
Ultra-low attachment plate | Corning | 24-well | |
Transwell cell culture inserts | Corning | 96-well unit, 0.4 µm pore size | |
Deep well plates | Corning | 96-well, 1 ml | |
Biopsy punch | Stusche | 4.5 mm | |
Glass microscope slide | Menzel | footprint of 75 x 25mm | |
Polydimethylsiloxane | Dow Corning | Sylgard 184 | |
Silicon rubber additive | Wacker Chemie | Wacker Primer G790 | |
Tubes for air pressure | SMC Pneumatik GmbH | Polyurethan-Schlauch, metrisch | |
Alumin ELISA | Bethyl Laboratories | Human Albumin ELISA Quantitation Set | |
Lactate dehydrogenase assay | Stanbio Laboratory | LDH Liqui-UV kit | |
Alexa Fluor 594 acetylated LDL | Invitrogen | 1 mg/ml |