Bottom-up tissue engineering

Assembly of cells and cell aggregates

An exciting development is the DNA-directed self-assembly of multiple cell types. Bertozzi and colleagues have previously developed alkyne derivatives that undergo [3+2] cycloadditions in the absence copper [9–11]. Strained alkynes react readily with an azide, with no detectable cross-reactions with moieties found in cells and tissues. Saccharides such as N-azidoacetylmannosamine added to cell culture medium are taken up by cells and used to produce the extracellular glycocalyx. In a recent study, cell surfaces decorated with azide sugars were reacted with single-stranded oligonucleotides modified with a cyclooctyne derivative. Two batches of Jurkat cells were separately labeled with complementary oligonucleotides. Upon mixing in a 1:1 ratio, the cells assembled into large aggregates. To control aggregation, the ratio of the two types of cells was adjusted to 1:50. In this case, the less common cell type existed as single cells surrounded by rosettes of the other cell type, termed microtissues (Fig. 2). DNA-directed cell assembly was then used to produce a paracrine signaling network between IL-3 expressing CHO cells and an IL-3 dependent untransformed hematopoietic progenitor cell line (FL5.12). FL5.12 cells that were assembled into microtissues with the CHO cells and embedded in agarose gels were able to proliferate even after IL-3 removal. Single FL5.12 cells did not proliferate under these conditions [12].

Formation of microtissues may also be performed on larger scales by co-culturing pre-formed cell aggregates. Forgacs and colleagues have demonstrated that spheroids may be formed by cutting cylinders of a cell slurry followed by overnight culture in solution. The spheroids that form are of uniform size and are used as bioink for cell printing [13]. Spheroids of different cell types may be printed into tissue-like structures using a custom 3D printer. This highly promising technique has been very recently reviewed [14]. Morgan and colleagues have demonstrated that if a poorly adhesive surface contained small posts, the cells surrounded the posts, eventually forming a multilayer of cell aggregate with holes after removal from the substrate [15]. More recently, these honeycomb-like structures were formed from ovarian theca cells. Inside the holes, ovarian granulosa cells were added, creating a structure similar to that found in the ovary. Such techniques may allow in vitro maturation of oocytes, potentially greatly improving in vitro fertilization [16]. The same group also demonstrated the formation of toroids on poorly-adhesive posts. As the cells contracted around tapered posts, they actually pushed themselves up and off. The toroids proved to be useful building blocks to assemble highly porous tissue-like structures [17]. The group also studied the kinetics of assembly of spheroids in linear troughs. Linear arrays of spheroids tended to fuse into rods and contract over time. The kinetics of fusion and contraction could be manipulated by the length of time the spheroids were allowed to mature, or by blocking cell-cell adhesion [18, 19].

Arrays of spheroids may also be produced. Morgan and colleagues used poorly adhesive multi-wells made from agarose and seeded them with cells to form spheroids. An alginate solution was then poured over the multi-wells and gelled. When removed from the agarose, the alginate gel was decorated with spheroid-laden posts [20]. In a similar spheroid-patterning strategy, Fukada and colleagues filled microwells in a microfluidics device with HepG2 liver cells (the other surfaces of the device were coated with albumin, such that the HepG2 spheroids could only form in the microwells). Then, a solution of polylysine was flowed over the surface, adsorbing to the albumin coating and rendering the surface cell adhesive. Fibroblasts were seeded on the substrate and these cells surrounded the spheroids but did not coat them [21].

Assembly of micron-scale hydrogels

Sefton and colleagues introduced an intriguing approach to modular scaffold formation in 2006 [22]. The approach essentially inverts the architecture of the native vasculature. Cylinders (modules) are produced by cutting a tube collagen gel that is loaded with cells. The modules are then coated with endothelial cells. The modules may be packed into a chamber and subjected to blood flow. Although the thrombogenicity of the construct is low in flowing blood, adhesion of leukocytes was present [23]. This may have resulted from the flow geometry, as a endothelial-lined microfluidic chamber with similar geometry also led to greater adhesion of leukocytes [24]. Collagen/Matrigel modules were formed with embedded rat pup cardiomyocytes. The modules were seeded with endothelial cells and embedded in an alginate gel. The modules contracted in sync with electrical stimulation. However, the endothelial coating did attenuate the response [25]. The modules may be used as vehicles for cell transplantation in vivo. Human endothelial-coated collagen modules were implanted in athymic rats. The modules formed hollow channels at their interfaces and the human endothelial cells formed primitive vessels in the channels [26]. Transplanting modules coated with allogeneic rat endothelial cells into immune compromised rats resulted in the outgrowth of the transplanted cells to form functioning blood vessels that persisted for at least 60 days [27]. The modules were also implanted in immune-compromised mice by injecting through a needle. Endothelial apoptosis was noted to be present, which was decreased by coating the collagen modules with fibronectin. The fibronectin-coated modules yielded greater numbers of mature vessels lined with the transplanted endothelial cells [28]. Synthetic materials are also being developed as mechanically stronger substitutes for collagen in the cylinders [29]. A video of the methods for modular scaffold fabrication has been published [30]. The Whitesides lab has also introduced micromolding methods to produce similar, yet smaller, cylinders [31, 32].

Khademhossenei and colleagues have intensively investigated the assembly of cell-laden microgels to produce scaffolds (Fig. 3). The microgels have been assembled primarily by manipulating the hydrophilicity of surfaces and interfaces. For example, a PDMS stamp was used to pattern a hydrophobic silane onto a glass slide. Cell-laden microgels were added to the glass substrate in PBS. After pouring off the excess solution, the buffer solution remained in the hydrophilic regions of the slide, trapping some of the microgels in the remaining liquid [33]. A mathematical model of microgel assembly in oil-like media has been developed [34].

Hexagonal microgels were recently produced by the photopolymerization of PEG or the calcium-induced gelation of alginate. The microgels were assembled in two-dimensional arrays and pores were generated by chelating calcium to dissolve the alginate microgels, with little impact on cell viability within the PEG microgels. Multiple layers could be stacked to produce three-dimensional structures [35]. Microgels of various shapes were also assembled while floating on the solvent perfluorodecalin. Cell-laden microgels aggregated and were crosslinked by a second photopolymerization with little loss of cell viability [36]. PEG microgel rings were also synthesized, which stacked into tubes in mineral oil. By varying the number of holes in the rings, branching structures could be introduced. Concentric rings were also produced by sequential photopolymerization with different photomasks. Endothelial cells were polymerized in the inner ring, while smooth muscle cells were polymerized in the outer ring, which could be assembled into blood vessel-like structures. [37]. The group also introduced the concept of micro-masonry . PEG microgels containing HepG2 cells were formed in various shapes (usually bricks). The microgels were suspended in a solution of PEG-methacrylate with photoinitiator and added to a PDMS template (e.g., a cylinder). When excess solution was removed, capillary forces brought the bricks together around the template. Photopolymerization was used to crosslink the solution-phase PEG-methacrylate, serving as the mortar that bound the bricks together in the shape of the template [38].

In another strategy from the same group, mouse embryoid bodies were formed and deposited into microwells on a substrate. Photopolymerizable PEG was added and photopolymerized in roughly half of the microwell. Another mask was used to photopolymerize gelatin in the other half of the microwell. Thus, half of each embryoid body was encapsulated in a block of PEG gel, and the other half was encapsulated in a block of crosslinked gelatin. Vasculogenic cells sprouted from the embryoid body into the gelatin matrix, introducing polarity into the construct [39].

Harada et al. recently used the differential affinities of cyclodextrins to selectively assemble hydrogels. Adamantane or tert-butyl groups bind strongly to β-cyclodextrin, while n-butyl groups bind strongly to α-cyclodextrin. When the various chemical groups were added to poly(acrylamide) gels, the gels self-segregated based on the specific interactions (Fig. 4A) [40]. Although the gels were in the millimeter size regime, the approach may prove to be useful for the assembly of microscale hydrogels.

In the author s lab, we have recently produced macroporous scaffolds by the assembly and chemical reaction of hydrogel microspheres. Microspheres with different functions (structural, drug delivery and porogenic) were synthesized from reactive PEG derivatives and proteins. The method that was developed for microsphere fabrication relied a thermally induced phase separation of PEG in the presence of sodium sulfate [41]. The microspheres retained reactive groups, allowing crosslinking upon mixing the microspheres and centrifugation, forming a scaffold. The crosslinking was mild enough to be performed in the presence of living cells. The porogenic microspheres dissolved with two days in culture to produce porous scaffolds without affecting cell viability [42]. The density of the crosslinks in the microspheres affected their buoyancy relatively independently of their size, allowing the formation of gradients in microspheres upon centrifugation (Fig. 4B) [43]. Microspheres used for scaffold assembly may also be produced by the precipitation photopolymerization of PEG-diacrylate [44].

A number of groups are investigating aggregates consisting of cells and drug-delivering hydrogel microspheres. The overall hypothesis is that growth factor diffusion into spheroids is a limiting factor in the differentiation of stem cells. Hyaluronic acid microspheres containing perlecan for controlled release of BMP-2 were added to mesenchymal stem cells in a micromass to promote chondrogenesis [45]. Others have developed chondroitin sulfate, agarose and gelatin microspheres were developed to release growth factors to affect cell differentiation. The microspheres were incorporated within mouse embryoid bodies, with good viability and some effects on stem cell differentiation [46, 47]. Gelatin/chitosan microspheres have been similarly used to promote the differentiation of adipose derived stem cells [48].

These strategies are intermediate between scaffold-based and scaffold-free approaches, and evolved from techniques using non-hydrogel microspheres made from degradable poly(esters) [49–54].

Conclusions

The self-assembly strategies discussed here may evolve to allow the formation of spatially organized tissues. The ability to control the architecture of 3D cell/material hybrid structures may prove to be critical for the development of highly advanced scaffolds for cell transplantation that are better mimics of natural ECM.