This is exceedingly cool:

SCIENTISTS have created "miracle mice" that can regenerate amputated limbs or damaged vital organs, making them able to recover from injuries that would kill or permanently disable normal animals.

The experimental animals are unique among mammals in their ability to regrow their heart, toes, joints and tail.

And when [fetal liver] cells from the test mouse are injected into ordinary mice, they too acquire the ability to regenerate [...]

As usual, though, there's a lot of important caveats missing from the "miracle mouse!" version of the story (and whenever you hear "miracle", especially in science, think of David Hume). The article to read is The scarless heart and the MRL mouse by Ellen Heber-Katz (who runs the lab responsible for most of these discoveries) et al.. It's a review article from about a year ago, so it doesn't cover the stem cell work, but it gives a good background. It's open source (but pdf, since the html version seems not to be working) and written for a fairly general audience.

The mouse strain in question is an inbred strain called MRL, and has been around since 1979. It was originally selected for large size and has a lymphocyte proliferative disorder which gives rise to a variety of immune problems, including autoimmune symptoms. For instance, the MRL mouse is a common model for systemic lupus erythematosus. The regenerative abilities of this mouse were discovered by researchers marking mice by punching small holes in their ears; within 30 days, the MRL mice healed the ear holes closed whereas other mice retain the holes for their lifetimes (mice live about two years).

Further investigation revealed that the MRL mice can regenerate almost all tissues except brain. This regenerative healing is fundamentally different from normal mammalian wound healing, and takes place without scar formation (which is of particular interest to cardiologists, since scars formed in response to heart injuries, including infarcts, are probably the primary cause of subsequent chronic heart disease and failure). Such healing is known in mammals, but only very early in development -- interestingly, prior to the development of certain immune, especially inflammatory, responses. Heber-Katz et al. report that T-cells from nonhealer mice do inhibit the ear wound closure response. It doesn't seem, however, that their immune dysfunction is the only mediator of the regenerative response in MRL mice. For instance, matrix metalloproteases 2 and 9 and their specific inhibitors have been shown to be differentially activated in healer vs. non-healer mice (MMPs and MMP inhibitors are primary players in tissue remodelling, including wound healing). In fact, at least 20 genetic loci (chromosome regions) have been shown to be involved in the MRL regenerative phenotype. Importantly, many of these show no overlap with the loci mapped to the autoimmune disorder. (In very plain English: it is not likely that the primary cause of the regenerative capacity is also the cause of the immune disorder, although there may be some overlap; this means that we may be able to replicate the regenerative ability without causing immune dysfunction.)

It is also not clear exactly which cells are doing the healing. In bone marrow transplant/transfer experiments, healing in both heart and ear tissue followed the recipient not the donor phenotype, meaning that bone marrow derived stem cells are not likely to be driving the healing response (although some involvement of donor cells was observed). Moreover, in these model systems recipient hematopoiesis is destroyed by X-ray exposure, so the cells responsible for the healing must be resistant to such treatment. It's also possible to reconstitute irradiated hematopoiesis using fetal liver cells, which contain a population of hematopoietic stem cells. Heber-Katz' group has tried that too (another open source article). The results were somewhat surprising: in the heart, healing followed the donor phenotype (i.e. the fetal liver cells transferred the regenerative capacity or lack thereof), whereas in ear injuries healing followed the recipient phenotype (as seen with bone marrow transplant/transfer). Once again, donor cells are seen in the healed heart but the mechanism of their involvment is not clear, nor is it clear why cardiac but not ear tissue could regenerate in this model.

Here's the thing that jumped out at me: because non-healer liver cells transferred that phenotype, it appears that scarring inhibits regeneration in mammals. In the MRL animals, something is holding back the formation of scar tissue, and (therefore??) regeneration is taking place. In non-healer mice which received healer fetal liver cells, high degrees of chimerism (~60-80%) were seen, whereas non-healer into healer transfers showed an average of only 12% chimerism. Why was 12% non-healer enough to cause normal healing and scarring in that transfer, but 20-40% non-healer was not enough to stop MRL-type healing without scarring in the reciprocal model? The authors offer one clue: "We do not know which cell population is responsible for [scarring with only 12% chimerism] and it may be different than the population that allows for a regenerative response in the reciprocal chimeras."

This much at least is already clear: the MRL mouse model will provide profound insights into mechanisms of wound healing (including the possibility of regenerative medicine) and the functions of hematopoietic stem cells.