|Home | About | Journals | Submit | Contact Us | Français|
Here, we describe a simple micromolding method to construct three-dimensional arrays of organotypic epithelial tissue structures that approximate in vivo histology. An elastomeric stamp containing an array of posts of defined geometry and spacing is used to mold microscale cavities into the surface of type I collagen gels. Epithelial cells are seeded into the cavities and covered with a second layer of collagen. The cells reorganize into hollow tissues corresponding to the geometry of the cavities. Patterned tissue arrays can be produced in 3–4 h and will undergo morphogenesis over the following 1–3 d. The protocol can easily be adapted to study a variety of tissues and aspects of normal and neoplastic development.
The ability to recapitulate normal and diseased tissue histology faithfully and reproducibly in culture would revolutionize science and medicine. Engineered tissues could be used by cell and developmental biologists to investigate the basic processes underlying normal morphogenesis, by cancer biologists to study how those control processes are coopted or circumvented during neo-plastic progression and by clinicians as therapeutic replacements for diseased organs. Indeed, a few relatively simple models used extensively over the past 30 years have yielded insight into the normal and diseased development of mammary gland acini1,2, renal cysts3,4 and microvascular endothelial cords5,6. In these now-traditional assays, cells are embedded in gels of extracellular matrix (ECM), usually reconstituted type I collagen or an extract of basement membrane. Although they produce tissue structures with some similarity to their in vivo counterparts, the methods rely primarily on cell-driven self-assembly and are poorly controlled either spatially or temporally. The resulting tissues are therefore heterogeneous in size, geometry and composition and are difficult to analyze quantitatively.
Reproducing in vivo tissue structure requires building three-dimensional (3D) systems with micrometer-scale resolution and control. A plethora of techniques have been developed to create patterns of proteins and cells in two dimensions (2D)7,8. Most rely on variations of photolithography (light-based patterning) or soft lithography (contact-based patterning using elastomeric stamps to transfer pattern). Only recently have investigators succeeded in adapting these techniques for 3D systems. Several groups have focused on creating synthetic hydrogels containing specialized chemical moieties that can be polymerized into complex microscale topologies using patterns of light9,10. Combined with optical or electrophoretic methods to direct the location of cells, this approach can be used to define the geometry and position of microscale colonies of cells11,12. However, in studies published to date, although the cells achieve differentiated function, they fail to cohere into a tissue or to faithfully recapitulate in vivo structure, likely due, in part, to the artificial nature of the synthetic hydrogels and the resulting lack of appropriate biochemical signals.
To build microscale topologies using native ECM proteins, we and others have developed contact-based techniques using elastomeric stamps of polydimethylsiloxane (PDMS) to mold microscale features into ECM gels13. Tien and colleagues14 identified surface treatments for PDMS that would allow the molded ECM gels to detach easily from the stamps without distorting the patterned features. Defined cavities can be created within monolithic gels by using sacrificial elements such as paraffin or gelatin15–17, and stacking multiple gels can generate more complicated multilayered structures14,18. These techniques have been used successfully to construct simple endothelial tubes with correct histology and physiology that are capable of being perfused with blood or other solutions19.
Here, we describe a technique that uses replica micromolding and layer-by-layer assembly to generate geometrically precise arrays of multicellular epithelial tissues in 3D ECM gels (Fig. 1). In brief, an elastomeric PDMS stamp containing a relief of the desired tissue architecture is used as a releasable mold. The stamp is treated with a solution of inert protein to render the surface nonadhesive to the ECM gel. Modified stamps are placed on a drop of liquid-neutralized collagen or Matrigel under conditions that favor gelling of the ECM polymers. Removal of the stamp reveals microscale indentations within the gel that correspond to the bas-relief pattern on the stamp. A concentrated suspension of cells or primary organoids is allowed to settle within the micromolded gel cavities. Excess cells are removed by gentle washing, and the cavities containing cells are sealed by placing a slab of unpatterned gel on the surface.
The 3D patterning technique is flexible—it can be used to mold microscale features into a wide range of natural and synthetic polymers. Sharply defined features down to <1-µm resolution can be introduced into the gels14. We have successfully used this procedure to pattern multicellular tubules of human and murine mammary epithelial cells20, kidney epithelial cells as well as microvascular endothelial cells. The geometry of the tissue is dictated by the geometry of the molded cavities, which is determined by the features on the surface of the stamp, which are determined a priori by the investigator (Fig. 2). The assay is readily quantifiable with a high level of statistical confidence because each sample consists of an array of hundreds of multicellular tissues, each having the same initial geometry. We have used this principle to analyze quantitatively the spatial and temporal dynamics of gene expression changes and alterations in cell positions during branching morphogenesis of mammary epithelial tubules20.
The potential applications of the 3D patterning protocol are diverse. Basic studies of cell–cell interactions can be performed by simultaneously patterning two cell types (e.g., luminal epithelial and myoepithelial) within the molded cavities. The heterotypic mixture of cells reorganizes to form a bilayered structure that approximates in vivo histology20. Epithelial/mesenchymal interactions can be studied by patterning epithelial cells in the cavities and interspersing fibroblasts or other mesenchymal cells in the bulk ECM gel. Patterning cancer-derived cells or cells with oncogenic mutations can be used to analyze aspects of neoplastic progression such as loss of polarity, luminal filling and uncontrolled cellular invasion. The tissue arrays could also conceivably be used to screen libraries to identify potential drug targets or therapeutic agents. The final tissues produced are only limited by the properties of the ECM gel (softer gels, in general, fail to retain pattern), the supply of cells and the resolution of the lithographic techniques used to create the initial mold. Highly compliant ECM gels that lose their structure when patterned with the technique described here may still be successfully employed as scaffolds using laser-guided approaches21.
Troubleshooting advice can be found in Table 1.
The patterned tissues should precisely match the size, geometry and spacing of the pattern etched into the silicon master. Tissues down to one-cell diameter (~ 10 µm) can be reproducibly constructed. Stained tissues can be photographed under the microscope and quantified as represented in Figure 2g,h. We have found that for tissues constructed of phenotypically normal cells (mammary, kidney and endothelial), the invasion of the cells into the surrounding ECM is controlled by the initial geometry of the tissue20. We expect qualitatively similar results for epithelial cells derived from other branched organs, although the governing relationship is likely to vary from organ to organ. Over the time frame of mammary epithelial morphogenesis, we have observed no changes in cell viability as determined by tracking cell number per tubule in real time, although this can be determined explicitly by staining for apoptosis or necrosis in situ. Other aspects of morphogenesis and differentiation (cellular movements, gene expression changes, etc.) can also be easily quantified by measuring differences in fluorescence intensity at different time points or via live imaging.
This work was supported in part by grants from the Department of Energy (DE-AC03-76SF00098 and a Distinguished Fellow Award to M.J.B.), the NIH (CA64786 and CA57621 to M.J.B.) and the Department of Defense (an Innovator Award to M.J.B. and W81XWH-04-1-0582 to C.M.N.). C.M.N. holds a Career Award at the Scientific Interface from the Burroughs Wellcome Fund.
Reprints and permissions information is available online at http://npg.nature.com/reprintsandpermissions