Artificial tissues

The homily that the human body is more than the sum of its parts becomes all too apparent when you have to assemble its components from scratch.

Figure 1

The liver, for instance, is a highly structured organ, and its function is not reproduced by a culture of hepatocytes (liver cells) alone. Liver-like function (producing enzymes, filtering out toxins and so on) is retained when hepatocytes are instead co-cultured with tissue-forming fibroblast cells — but a random co-culture is still a poor substitute for real liver tissue. At a meeting last month¹, Mehmet Toner (Harvard Med. School) explained how a marriage of cell biology with microfabrication can do better.

The artificial liver — a device for supporting liver-failure patients awaiting a transplant — is not a new idea. The Minnesota-based Regenerex company, for example, already produces a machine in which hepatocytes sustained in hollow fibres by a perfusion of nutrients do their job on the patient's blood. But such devices are prone to cell leakage, cell death and inefficiency.

A synthetic liver-like tissue is a more attractive option. Unfortunately, fibroblasts and hepatocytes cultured together do not automatically adopt the arrangement found in the liver; and because the two then communicate poorly, the hepatocytes are inefficient. So Toner and colleagues have used patterned substrates to grow the two cell types in more controlled geometric relationships.

They use the standard photolithographic techniques of microelectronic technology to imprint the surface of a borosilicate wafer with patterned films of collagen, which promotes cell adhesion. When hepatocytes are cultured on the surface, they adhere only to the collagen-coated regions, and the rest of the culture can be washed away (as shown in these images of stained hepatocytes). Fibroblasts can then be introduced to the regions of bare surface, producing an intimate mixture of the two cell types in a periodic pattern.

In this way, an arbitrary ratio of cell types can be prepared on the substrate, so adjusting the ratio to its physiological value is straightforward. Patches of hepatocytes perform best (as measured by their level of albumin or urea secretion) when they are small, because they can only communicate well with the surrounding fibroblast cells up to a certain distance from the interface — a fact revealed by stains selective to albumin or urea production. So the best choice of pattern element size, shape and number density can be deduced and engineered to order in this artificial tissue.