$$\rightleftharpoonup{xx}$$
$$\longleftharp{xx}$$,
$$\longrightharp{xx}$$,
Historically, scientists have relied on preclinical animal testing for drug discovery, but a growing number of these methods have been questioned because of poor correlation with human outcome1. The implementation of the "3Rs" principles to Replace, Reduce, and Refine animal experimentation urges scientists to find new in vitro alternative methods to support preclinical drug and chemical toxicology risk assessment2. However, many in vitro models developed to date lack the biological architecture, cellular complexity, and mechanical environment necessary to recapitulate the dynamic nature of human living organs3,4.
Conventional in vitro preclinical systems typically employ 2D monocultures of human cells grown on a rigid plastic surface. These methods provide a tool for conducting simple mechanistic studies and enable rapid screening of drug candidates. Owing to their relatively low cost and high robustness, 2D models are often paired with automatic high-throughput systems and used for the rapid identification of potential drug candidates during the early stage of the drug development process5,6. However, such 2D models do not provide a translational approach for modeling tissue-level, organ-level, or systemic responses to therapeutic candidates, which is needed for accurate predictions of drug safety and efficacy during the preclinical stage of their development. Flat cell cultures do not recapitulate the native tissue microenvironment, including the complex multicellular interplay, biomechanical properties, and three-dimensional (3D) architecture of human tissues7. Cells growing on a flat surface often do not acquire a mature phenotype and, therefore, cannot respond to pharmacological stimuli as they would in the native tissue. For example, primary human alveolar epithelial cells grown in vitro exhibit a squamous phenotype and lose key phenotypic markers, including surfactant proteins C and B (SP-C and SP-B)8. In addition to insufficient differentiation, primary cells frequently become insensitive to biological stressors in vitro, as certain biochemical pathways associated with tissue inflammation become non-functional9. Such loss of cell function seems to be primarily associated with the use of stiff substrates as well as the lack of soluble factors naturally released by tissue-specific stromal cells such as lung fibroblasts and smooth muscle cells10,11.
Understanding that the lack of chemo-physical and biological complexity limits the physiological behavior of cells in vitro has fostered the development of more sophisticated multicellular models, which have proven to better capture the complexity of human tissues outside the body12,13. Since the creation of the first co-culture models in the early 1970s14, the introduction of synthetic and natural hydrogels has significantly improved the ability to mimic native tissue microenvironments and has become an invaluable tool for driving cellular differentiation, guiding the self-organization of cells into tissue-like structures, and restoration of native tissue functions15,16. For instance, when grown in the appropriate 3D scaffold, human cells can self-arrange into functional structures such as spheroids or organoids, expressing stem cell markers, and are capable of self-renewal17. In contrast, human cells (including stem cells), when grown on traditional 2D substrates, rapidly age and undergo senescence after a few passages18. In addition, hydrogels can be "tailored" to match specific tissue properties such as porosity, pore size, fiber thickness, viscoelasticity, topography, and stiffness or further engineered with tissue-derived cellular components and/or bioactive molecules enabling emulation of the physiological or pathological conditions19,20. Despite their enormous potential for drug testing, 3D hydrogel-based models used in pharmaceutical research do not fully recapitulate the complex cytoarchitecture of the in vivo tissues and lack important hemodynamic and mechanical stimuli normally present in the human body, including hydrostatic pressure, cyclic stretch, and fluid shear21.
Microphysiological systems (MPSs) such as Organs-on-chips (OOCs) have recently emerged as tools that are capable of capturing complex physiological responses in vitro22,23. These models often employ the use of microfluidic platforms, which enable the modelling of the dynamic microenvironment of living organs.
We have combined the principles of 3D tissue bioengineering and mechanobiology to create an Open-Top Chip model of complex human epithelial tissue. This allowed us to closely recapitulate the multicellular and dynamic microenvironment of epithelial tissues. This includes tissue-specific biochemical and biomechanical cues naturally present in living organs but often neglected by traditional in vitro models24. The Open-Top Chip incorporates two compartments: a vascular compartment (Figure 1A) and a stromal compartment (Figure 1B) separated by a porous membrane, allowing for the diffusion of nutrients between the two chambers (Figure 1C). The vascular compartment is exposed to continuous fluid flow to recapitulate physiological shear stress, while the stretchable design of the stromal chamber allows for the modeling of the mechanical strain associated with breathing motions or intestinal peristalsis. The stromal compartment houses the tunable 3D hydrogel scaffold designed to support the physiological growth of tissue-specific fibroblasts. It possesses a removable lid that facilitates the establishment of an air-liquid interface, a condition that allows greater emulation of human physiology of mucosal tissues as well as direct access to the tissue for administrating drugs directly onto the epithelial layer. Supplementary Figure 1 captures some of the key components of the Open-Top Chip design including dimensions and biological compartments (Supplementary Figure 1A-D) as well as the main technical steps described in this protocol (Supplementary Figure 1E).
Perfusion of the Open-Top Chip is achieved with a programmable peristaltic pump (Figure 1D). The peristaltic pump setup allows 12 Open-Top Chips to be perfused simultaneously. Most incubators can house two setups enabling the culture of up to 24 chips per incubator. Mechanical stretching is achieved using a custom-made programmable vacuum pressure regulator (Figure 1E). It consists of an electro-pneumatic vacuum regulator controlled electronically by a digital-to-analog converter. In other words, the electro-pneumatic vacuum regulator generates a sinusoidal vacuum profile with an amplitude and frequency that is determined by the user. Cyclic strain ranging from 0% to 15% is generated by applying negative pressure to the vacuum channel of the Open-Top Chip at an amplitude ranging from 0 to -90 kPa and a frequency of 0.2 Hz. It is a custom-made system equivalent to the commercially available Flexcell Strain Unit previously adopted and described in other papers25. To mimic the mechanical tissue deformation associated, for example, with the breathing motion of the lung or the peristalsis of the intestine, the pneumatic actuator applies sinusoidal vacuum/strain waves whose magnitude and amplitude can be adjusted to match the physiological level of strain and frequency that human cells experience in their native tissue.
Here, we describe an efficient and reproducible method for engineering and culturing organotypic epithelium equivalents on a prototype Open-Top Chip platform. It allows the generation of complex organ models such as skin, alveolus, airway, and colon while integrating a vascular fluid flow and mechanical stretching. We will outline key technical aspects that must be considered while implementing principles of tissue engineering for generating complex epithelial models. We will discuss the advantages and possible limitations of the current design.
An overview of the main steps used to achieve tissue and organ maturation, including flow and stretch parameters, is reported in: Figure 2 for the skin, Figure 3 for the alveolus, Figure 4 for the airway, and Figure 5 for the intestine. Additional information concerning media composition and reagents used for culturing the different organ models are included in the supplementary tables (Supplementary Table 1 for the skin; Supplementary Table 2 for the alveolus; Supplementary Table 3 for the airway, and Supplementary Table 4 for the intestine).