If it wasn't for the generation of an in-built barrier, the phrase 'spiralling out of control' could easily be applied to the positive-feedback loop that drives limb outgrowth in vertebrates. Such a barrier has now been discovered by Scherz et al. and their findings are published in Science.

Usually, Sonic hedgehog (Shh) in the zone of polarizing activity (ZPA) of the posterior mesenchyme in the limb bud maintains the expression of several fibroblast growth factors (Fgfs), including Fgf4, in the apical ectodermal ridge (AER) by upregulating Gremlin in the adjoining mesenchyme. Gremlin antagonizes bone morphogenetic proteins (Bmps), which normally downregulate Fgf4 — so Fgf4 expression is maintained. Fgfs, in turn, maintain Shh expression. That is, until embryonic day 6 (E6), when Shh is downregulated and Fgf4 and Gremlin are no longer expressed. At this point, cells proliferate more slowly in the limb — so this breakdown of the Shh–Fgf loop is essential for regulating limb size.

To figure out how the breakdown occurs, the authors ectopically expressed each of the individual components of the loop in chick embryos just before E6 to see if they could maintain expression of the other genes that would normally be downregulated. Overexpressing either Fgf4 or Gremlin could maintain expression of the other genes, which indicated that cells had not lost responsiveness to either of these signals. However, when a bead soaked in Shh was implanted in the posterior limb, neither Gremlin nor Fgf4 expression was maintained. This ties in with the known occurrence of a zone of exclusion of Gremlin expression in the posterior limb — the cells in this zone cannot express Gremlin in response to Shh.

When the authors marked the descendants of cells that expressed Shh, they found them to expand anteriorly into a domain that resembled the Gremlin-free domain. Indeed, the cells that formerly expressed Shh did not express Gremlin. So Scherz et al. proposed that, as limb outgrowth progresses, the proliferating Shh descendants form a barrier between the Shh source and the cells that respond to Shh by expressing Gremlin. Initially, Shh can diffuse across a small barrier, but as this barrier enlarges and is filled with Gremlin-negative Shh descendants, such diffusion is no longer possible. Failure to express Gremlin results in the failure to antagonize Bmps, and so Fgf4 is downregulated.

To test their hypothesis, the authors physically removed the 'barrier' cells and rejoined the remaining posterior cells to the remainder of the anterior limb at E5. The feedback loop then operated for longer than in wild-type organisms and the limbs grew until they resembled those of the wild type in both structure and size. The proposed explanation is simple: removal of the barrier removes the obstruction to Gremlin induction and the feedback loop is restored — until the barrier forms again. One question to ask is why the Shh descendants cannot induce Gremlin. Another is whether the downregulation of other Fgf genes, such as Fgf9 and Fgf17, occurs through the same mechanism, although the chances are that it does.