In a field where distinguishing populations of cells can make a career, stem-cell biologists struggle to determine how various cell types respond to chemical cues. Throw in a consideration of physical effects, and the complexity can be almost too much to bear. But for spurring growth and differentiation, researchers are increasingly finding that physical factors can make all the difference.

Mouse embryonic stem cells embedded within a 3D scaffold made of polycaprolactone fibers. The fibers are shown in yellow, green indicates lipid uptake and storage by the cells Credit: Courtesy of Douglas Kniss/Ohio State University

Mesenchymal stem cells can be found in adult bone marrow and are unusual in their ability to differentiate into a variety of cell types. Last year, Dennis Discher and colleagues at the University of Pennsylvania in Philadelphia turned heads when they coaxed mesenchymal stem cells into committing to bone, muscle or nerve-cell lineages simply by varying the elasticity of two-dimensional matrices that mimic the target cells' natural microenvironments, or niches.1 If researchers can direct flat-grown cells so sensitively, how much better might those cells be fine-tuned using three dimensions that more faithfully reflect cells' living arrangements?

Many biologists who are interested in stem-cell differentiation are starting to ask a big new question, says Douglas Kniss, a biomedical engineer at Ohio State University in Columbus: “How does the geometry of a tissue influence the biology of what cells become?”

Interdisciplinary collaborations involving engineers, computer scientists and materials scientists are asking the same thing. And an increasing number of scientists and entrepreneurs are pursuing bio-printing, scaffolding and nanoscale strategies. Even so, the underlying biological question of exactly how such stimuli bring about cell outcomes is only beginning to be explored.

Under construction

Kniss has collaborated with Ohio State's materials science and chemical engineering departments to construct a scaffolding of Dacron fibres (polyethylene terephthalene) that prompts mouse embryonic stem (ES) cells to differentiate into pre-adipocytes (the precursors to fat cells).2 Better known in the textile industry, the fibres are approved by the US Food and Drug Administration for clinical use as a grafting material to reconstruct large blood vessels.

Kniss and his collaborators asked whether the fat-cell precursors grown on the Dacron fibres expressed the same gene markers as cells grown in vivo. They did. The cells also lived for more than two months, long enough to encourage Kniss to evaluate other materials, such as the biodegradable polycaprolactone fibres used in sutures. As the fibres dissolve, cells replace the synthetic scaffolding with their own matrix, a possible boon for regenerating heart and other tissues.

Still, Kniss says he and other researchers do not know enough about the cell biology to evaluate the clinical uses of the technology adequately. Until recently, one barrier has been the limited availability of non-mesenchymal stem cells. Similarly, reagents for three-dimensional environments have been hard to come by, says Linda Griffith, a professor of biological and mechanical engineering at the Massachusetts Institute of Technology (MIT). Researchers can cross-link the proteins fibronectin and collagen in polyacrylamide gels to vary gel stiffness, but the gels are not three-dimensional matrices, which limits the ability of researchers to vary their mechanical properties systematically.

Help may be forthcoming from the Cell Migration Consortium, funded through a so-called 'glue grant' from the US National Institutes of Health — funding that aims to pull together an emerging field's disparate pieces. Part of the grant's focus, Griffith says, is to develop systems that can model stem-cell migration in three dimensions within developing, mature or regenerating tissues.

Three dimensional culture of human osteoblasts on novel polystyrene scaffold Credit: Courtesy of Stefan Przyborski, Chief Scientific Officer, ReInnervate Limited

To learn how progenitor cells behave within the context of a mature tissue, Griffith and collaborators have developed a microfabricated, perfused bioreactor that supports a three-dimensional liver architecture when seeded with primary liver cells.3 But the system is far from plug-and-play. Griffith says her lab has spent the past two years trying to adapt the set-up to a more user-friendly multi-well plate format that cell biologists would recognize.

“There's a demand for a new technology,” agrees Stefan Przyborski, a cell biologist at Durham University, UK. The alginate-based AlgiMatrix from Invitrogen in Carlsbad, California, bears the slogan “as real as it gets”, and became one of the first commercially available three-dimensional scaffolds when it debuted in January this year at US$160 per 96-well plate. Przyborski says that even that culture system is hardly a routine, off-the-shelf option.

Along with Durham University chemists, Przyborski is hoping to push the nascent field forward with a sponge-like scaffolding made of polystyrene. HepG2 liver cells grown within the scaffold secrete albumin much like their in vivo counterparts, whereas flat-grown cells do not. And compared to cells growing in a Petri dish, scaffold-grown liver cells better withstand the toxic effects of the cancer drug methotrexate.4 This isn't just because the drug does not reach deep into cell culture, says Przyborski; even the surface cells are more resistant. Przyborski, who is also chief scientific officer at start-up cell-technology company ReInnervate, based in Durham, says the next step is to be able to manufacture the scaffolding robustly and reproducibly and commercialize it as “a simple piece of plastic".

Early-stage origami

The Centre for Cell Engineering at the University of Glasgow, UK, has nurtured similar interdisciplinary partnerships focusing on biofabrication techniques. Beyond pioneering new lithography methods, researchers there have found that 'classical' technology can be co-opted to the three-dimensional cause. The centre's high-end electron-beam (e-beam) lithographic printer manufactured by Leica can only write on flat surfaces, for instance, but research director Mathis Riehle has found a way around that limitation. “We fabricate a two-dimensional sheet with imparted nano- and microfeatures and then roll it or fold it up, origami-like, in order to create a 'semi-three-dimensional' environment for the cells,” he says of his group's paper now in preparation.

With the technique in hand, the Glasgow microfabricators have created what has been dubbed a 'swiss roll scaffold' by its creators, electronic engineer Kris Seunarine and cell engineer Osian Meredith in collaboration with Riehle. The jam filling is made of aligned fibroblasts and smooth muscle cells; the flexible cake is composed of nano-patterned polycaprolactone films that have been rolled up and glued or sutured. (See Nature's technology feature earlier this year).

Bio-printing specialists are likewise expanding their repertoire from two dimensions to three. Lee Weiss and Phil Campbell at Carnegie Mellon University, Pittsburgh, Pennsylvania, and Julie Phillippi now at the University of Pittsburgh have co-developed a 'bio-printer' that can use proteins as 'ink' to make three-dimensional matrices using collagen, fibrin and other scaffolding proteins. But Phillippi still wants a better understanding of how external stimuli affect cell-culture outcomes. “We don't understand exactly what patterns nature designed for proper tissue formation so that we can recapitulate that with our bio-printer,” she says. “However, we now have the technology to ask that question.”

Phillippi plans to combine her printing technology with real-time imaging to see whether cells respond to differences in the three-dimensional matrix by dividing, differentiating or dying. One of her collaborators, computer-vision expert Takeo Kanade at Carnegie Mellon, is perhaps best known for his technique of videotaping the same event from multiple angles, a method borrowed by US television station CBS and dubbed 'Eye Vision' for the 2001 Super Bowl.

Near the other end of the design spectrum, bioengineer Dan Anderson and others in Robert Langer's lab at MIT have developed a fully automated high-throughput system that uses microarray chips to screen the effects of miniaturized polymers on cell behaviour. One observation is that muscle precursor cells grow fine on some polymers, whereas ES cells do not.5,6 “We don't yet understand how these materials are affecting cell behaviour,” Anderson says, “but believe that materials allowing for the growth of certain cell types [while excluding others] could be valuable in the construction of complex material cell structures.”

Working from the outside in

Theo Palmer at Stanford University's Institute for Stem Cell Biology and Regenerative Medicine is adapting computer circuit technology to produce nanofeatures on printed circuits to analyse what makes neural progenitor cells follow different fates. Although the extracellular nanotopography is largely two-dimensional, Palmer says he's now pondering three-dimensional substrates.

Palmer cautions that growing cells in synthetic environments has severe limitations. “You could find a beautiful artefact for growing stem cells that doesn't reflect the reality of how the brain regulates them,” he says of the neural progenitor cells he studies. And as yet, understanding how physical stimuli spur cell differentiation or division remains largely a guessing game.

Along with other researchers, the University of Glasgow's Matthew Dalby suspects that the first step is mechanical: changing the cells' ability to adhere or spread alters how the cytoskeleton's structural proteins are organized. The cytoskeleton is linked both to the extracellular matrix and to the nucleus via intermediates, and Dalby envisages the cytoskeleton tugging on the nucleus and affecting gene expression as the cell alters its adhesion to external surfaces.

Integrins, the cell adhesion molecules that act as intermediaries between the cytoskeleton and extracellular matrix, are already known to sense varying degrees of stiffness in collagen and to adjust their own conformations in response, transmitting the change across the cytoskeleton. “It's like a spider's web: the fly falls into the web and the spider senses it even though it's far away,” says Gabor Forgacs, a biological physicist at the University of Missouri in Columbia.

Mechanical extruders can help create the proper conditions for “printed” cell aggregates to self-organize Credit: Courtesy of Gabor Forgacs/University of Missouri

But how does the message reach the transcriptional machinery within the cell nucleus? Dalby points out that within the nucleus, chromosomes are associated with lamins, proteins that line the inner surface of the nuclear membrane. If the lamins change in tension, the nuclear territories where the chromosomes reside may also change. That chain of events could alter the balance of active euchromatin and transcriptionally silent heterochromatin within particular chromosomes.

Dalby concedes that any pathway linking integrin signaling with changes in the nucleus is largely conjecture, but he's keen to link the formation of integrin-rich signaling hubs known as focal adhesions to genomic changes that lead to cell differentiation. Recently, he and colleagues showed that nanoplatforms with slightly disordered surface patterns aided stem-cell binding, or adhesion, across nanopits, and that this adhesion affected the cells' cytoskeletal organization.7 Dalby has also demonstrated that genes can be up-regulated or down-regulated, depending on the external topography encountered by fibroblasts.8

Let the cells do the thinking

Researchers now pondering physical cues for cell differentiation are rediscovering what has long both captivated and aggravated their colleagues studying chemical cues. “We create some external conditions,” says Forgacs, “and then we have no control over the system. It does what it wants to do.”

In a new study, he and his co-authors used mechanical extruders to “print” rings of cell aggregates — endothelial cells, smooth muscle cells and fibroblasts — in cylindrical formations.9 Under the right conditions, the aggregates of randomly distributed cells fused and self-organized into blood vessel-like cylinders, so that endothelial cells moved to the lumen, smooth muscle cells moved to the middle and fibroblasts migrated to the outside.

“I'm still fascinated by the fact that they know what to do,” Forgacs says. “We will never be able to reproduce, cell by cell, what nature does.” But as he and other researchers are learning, maybe that won't be necessary. If they can select the right extracellular environment, the biological system appears more than able to guide itself.

The trick, of course, is knowing exactly how — and where — to begin.