Whereas 2D approaches allow well-controlled analysis of the impact on stem cells of individual components of the niche, 3D approaches should allow reconstruction, and realization of the complexity, of the natural tissue (). In epithelial tissues (for example skin and gut), stem cells adhere to 2D, sheet-like basement membranes, but most stem-cell niches (for example those in bone marrow, brain and muscle) are 3D microenvironments composed of hydrated, crosslinked networks of ECM proteins and sugars. In three dimensions, stem cells can be exposed to a solid microenvironment that fully ensheathes them (), in contrast to 2D platforms, in which cells are typically exposed to a solid, flat surface on the basal side and to liquid at the apical surface. However, although conceptually appealing, the construction of 3D artificial microenvironments is not simple53
. There are chemical challenges in the production process, considerations of appropriate elasticity, and the need to overcome the physical constraints that impede migration or morphogenesis. First and foremost, in most cases, cell viability remains a problem; second, understanding the read-outs from such complex multicomponent systems is not straightforward. As a result, high-throughput analyses are currently not possible, and few of the many possible variables can be systematically explored. Nonetheless, progress is being made.
Engineering 3D in vitro models of stem-cell niches
Several impediments to 3D culture must be overcome. First, to expose stem cells to an accurate 3D artificial environment, chemical approaches that allow the embedding of stem cells must be used. This is ideally performed in situ
(that is, during the formation of the 3D material), which requires a mild and highly specific crosslinking chemistry so as not to compromise cell viability as a result of adverse side reactions. Several methods of forming synthetic or semi-synthetic hydrogel matrices under physiological conditions have been developed for this purpose and are reviewed in, for example, refs 54
. Some of these approaches explore not only highly specific chemical or enzymatic reactions but also physical mechanisms of crosslinking, such as the molecular self-assembly of small-molecule building blocks (including peptides, peptide amphiphiles and oligonucleotides). Each of these approaches has been demonstrated to yield viable encapsulated cells after crosslinking; the strategies differ primarily in the hydrogel-network structures that are produced and in how cells respond to these different network structures (of which some are porous and others are dense meshworks).
Second, the biophysical characteristics of the 3D environment are important. Cells embedded in a 3D environment can suffer from a lack of gases and nutrients. This problem is overcome by using scaffolds made of solids such as polymers with interconnected porosity and by using hydrogel networks with microscopic meshes, as such structures readily allow the diffusion of macromolecules. Third, substrate elasticity and materials with mechanical properties closely approximating those of natural stem-cell niches are desirable28
, as described above. Last, physical constraints that impede cell proliferation, migration and morphogenesis should be avoided. To avoid the potential problems of having physical barriers in three dimensions, materials that have matrix porosity on the scale of cellular processes can be designed. For example, nanofibrillar hydrogels that contain microscopic pores large enough to facilitate cell growth have been developed56
. An attractive alternative approach uses polymer gels that can be synthesized to contain chemically crosslinked substrates for proteases naturally secreted by cells, for example during cell invasion. This feature allows a dynamic interplay between the cells and their microenvironment such that the cells locally degrade and then ‘remodel’ the matrix. For example, PEG-based hydrogels have been rendered chemically degradable through hydrolytic breakdown of ester bonds57
and have been developed with cleavage sites for cell-secreted matrix metalloproteinases or plasmin4
. This cell-regulatable breakdown of the matrix allows cell migration and proliferation in a manner determined by the cells.
Probing stem-cell–matrix interactions in three dimensions
A long-standing question in stem-cell biology and tissue engineering is that of how the numerous components of the stem-cell niche govern stem-cell fate in three dimensions. This question is difficult to address in vivo
or using any existing 2D in vitro
approaches. A 3D stem-cell niche is extremely complex (), and the number of its physical, chemical and mechanical effectors is too great to define in practice. Even if the specific nature of its components were known, testing them systematically would not be possible. Thus, developing new approaches aimed at high-throughput screening of combinations of 3D microenvironmental variables, in a manner analogous to 2D ECM protein microarrays or other cellular arrays described above, is a major goal58–60
The production of high-throughput microarrays of 3D matrices could be possible using robotic liquid-dispensing and printing approaches in combination with biomaterial-crosslinking chemistries. Combinatorial mixtures of liquid precursors of hydrogel networks can be deposited in minute volumes and at high density onto solid substrates61
. In one of the first examples of this emerging strategy62
, 3D PEG-hydrogel arrays were produced to screen for the individual and combinatorial effects of gel degradability, cell-adhesion-ligand type and cell-adhesion-ligand density on the viability of human mesenchymal stem cells. Increased PEG-network degradability and greater cell-adhesion-ligand density were both found to increase the viability of the stem cells in a dose-dependent manner.
Measures of cell viability constitute a minimal first step. It is necessary to design more-sophisticated ways of measuring stem-cell proliferation, asymmetrical and symmetrical division, self-renewal and differentiation into selected lineages that can be assessed in three dimensions. One challenge in this endeavour will be to analyse cellular responses in three dimensions, for which one focal plane for microscopic read-out is not sufficient. Ultimately, it would be desirable to investigate the role of the 3D microenvironment in controlling stem-cell fate on a more comprehensive (‘systems’) level, integrating the complete set of relevant variables. Importantly, when promising candidate microenvironments are identified through such studies, selected materials need to be further evaluated using in vivo approaches, for example by transplantation of cell–matrix constructs into mice.
Probing cell–cell interactions in three dimensions
Important components of stem-cell niches are the cells that abut stem cells, which are sometimes referred to as support cells or niche cells. These can include vascular cells, neural cells, and stromal cells such as fibroblasts. They not only provide instructive secreted signalling cues but also send signals through transmembrane proteins or bound matrix proteins. Although this type of cellular crosstalk is conceptually appreciated as being highly significant to stem-cell behaviour (to quiescence, activation and proliferation), the study in three dimensions of which factors have a critical role and how they act together is a nascent field.
Nonetheless, progress is being made in technologies that would allow the investigation of such cell–cell signalling interactions in near-physiological 3D microenvironments (). One approach is based on the electropatterning of mammalian cells within hydrogels50
. Electropatterning localizes live cells (possibly of any type) within hydrogels, such as photopolymerized PEG gels, by using dielectrophoretic forces. Large numbers of multicellular clusters of precise size and shape have been formed in three dimensions on one focal plane. By modulating cell–cell interactions in 3D clusters of various sizes, this microscale tissue organization was, for example, shown to influence the biosynthesis of bovine articular chondrocytes, with larger clusters producing smaller amounts of sulphated glycosaminoglycan per cell.
Other work has combined gel patterning with microfluidic technology to analyse angiogenesis in 3D co-cultures63
. Primary liver and vascular endothelial cells were cultured on each side wall of a collagen gel between two microfluidic channels. Morphogenesis of 3D hepatic cultures was found to depend on fluid flow across the nascent tissues. Vascular cells formed 3D capillary-like structures that extended across an intervening gel to the hepatocytes’ tissue-like structures. This is a remarkable advance, as microvascular networks are considered to be important components of several stem-cell niches6
. Thus, these approaches could prove useful in addressing fundamental questions in stem-cell biology.
3D biomolecule gradients in stem-cell biology
Morphogen gradients have long been known to regulate cell fate and tissue or organ development64
. Biomolecule gradients are crucial regulatory components of dynamic tissue processes, not only during development but also during homeostasis and regeneration. Therefore, the creation of biomolecule gradients in 3D biomaterials systems has received increasing attention in stem-cell bioengineering (). Such gradients could be shallow, such that a given cell experiences one concentration along its whole length, or steep, such that the cell experiences a different concentration at each end. Cells may migrate away from or towards a particular biomolecule concentration. Alternatively, when gradients are steep, cell polarity and asymmetry may be induced, just as in a stem-cell niche.
Arguably the most precise and robust way of generating a biomolecule gradient is through microfluidic technology65
, because microfluidics allows the well-controlled manipulation of very small amounts of fluid. Microfluidic gradient platforms have already been applied to stem-cell biology, albeit in two dimensions (see, for example, ref. 66
). However, 3D gradient systems are rapidly being developed67,68
. One example is a microfluidics-based approach whereby cells within alginate gels could be exposed to desirable soluble gradients in 3D microenvironments67
. Applied to adult stem-cell culture, such intricate control over the biochemical microenvironment in three dimensions is an important step towards the in vitro
recapitulation of stem-cell microenvironments that are more complex. The advantages of combining biomaterials engineering with microfluidics for stem-cell applications are clear69
: this combination offers the potential for arrays of individually addressable cell-culture chambers70,71
in which artificial microenvironments are exposed to spatially and temporally controlled biomolecule gradients (temporal control allowing delivery at any time during an experiment). Because proteins can be tethered to gel networks, it should be possible to combine tethering and soluble gradients of protein morphogens to mimic the exposure of cells to both ECM-bound protein gradients and soluble gradients, to recreate a stem-cell niche in three dimensions more accurately.
Mimicking the spatial 3D niche heterogeneity
Stem cells sense and respond to the spatial heterogeneity of 3D microenvironments. Many in vivo stem-cell microenvironments are ‘polarized’ structures, in that they expose individual stem cells to differential niche components. An example is the niche of the satellite cell (the canonical muscle stem cell), which is located between the muscle-fibre membrane and the surrounding basement membrane (). An ideal 3D in vitro model of a stem-cell niche would allow recapitulation of this type of complex architecture and manipulation at a desired time during an experiment, for instance to address the question of whether microenvironmental polarity dictates when a cell is quiescent and when it is activated.
Application of hydrogel engineering using photochemistry suggests that the construction of such complex microenvironments in three dimensions will be possible and will allow impressive precision and control over the dynamics72–74
(). For example, in photopolymerized PEG hydrogels, photolabile building blocks have been synthesized74
: these can be cleaved by a controlled light beam to modulate biophysical and biochemical gel properties locally at a given time. Mesenchymal stem cells were shown to respond to locally induced network changes in stiffness and cell-adhesion properties; in a densely crosslinked gel, the decrease in crosslinking density obtained through cleavage of the backbone of the photolabile chain induced a significant morphological change in the encapsulated stem cells (initially round in shape, they became more spread out). Moreover, the controlled manipulation of the concentration of cell-adhesive peptide ligands in the PEG gel led to inducible changes in chondrocyte differentiation. Differentiation into chondrocytes increased when an RGD peptide, which binds to integrins, was removed using light at a later time during 3D cell culture.
Microfluidic technology could also be used to mimic to some extent the spatial heterogeneity of stem-cell microenvironments75
. Several 3D matrices (such as type I collagen, Matrigel or fibrin) containing cells were micropatterned within a single microfluidic channel, stably interfacing each other. Cell culture was performed over several weeks and led to spatially restricted development of multicellular structures within designed patterns. These new methods will be of use in studying a great number of questions in stem-cell biology.
From artificial niches to 3D in vitro ‘tissues’
The 3D approaches discussed above serve as powerful model systems to elucidate extrinsic stem-cell regulation, but they would not form an appropriate basis on which to reconstruct large-scale tissue models76
using stem cells and biomaterials as building blocks, because they do not facilitate the modular and spatially well-controlled combination and positioning of these building blocks and they do not extend to scales of millimetres to centimetres. A technology known as bioprinting may be the method of choice in this endeavour, because theoretically it is feasible to combine layers of ECM and cell mixtures in modules of varying composition on a micrometre scale and in three dimensions. In bioprinting, custom-designed inkjet printers deposit, in a controlled layer-by-layer fashion, cells and biomaterials in almost picolitre-sized droplets at a rate of tens of thousands per second (see, for example, ref. 77
). On deposition on a substrate, these droplets can be polymerized to form a solid gel that could encapsulate stem cells or contain biomolecules with locally modular composition. Although the bioprinting field has arguably had little impact on stem-cell biology as yet, the results obtained so far with other cell types look promising. For example, viable 3D composites of embryonic neurons and astrocytes have been patterned in multilayered collagen78
. Currently, bioprinting is cumbersome, mainly because a suitable ‘bio-ink’ (that is, a hydrogel system that can be rapidly crosslinked, with high spatial precision, and is simultaneously highly biologically active and permissive) is lacking. However, if this obstacle could be overcome, bioprinting could be a significant step towards achieving the long-standing goal of tissue engineers, namely the formation of functional tissues outside the human body.