Neural induction requires BMP inhibition only as a late step, and involves signals other than FGF and Wnt antagonists.

A dominant molecular explanation for neural induction is the 'default model', which proposes that the ectoderm is pre-programmed towards a neural fate, but is normally inhibited by endogenous BMPs. Although there is strong evidence favouring this in Xenopus, data from other organisms suggest more complexity, including an involvement of FGF and modulation of Wnt. However, it is generally believed that these additional signals also act by inhibiting BMPs. We have investigated whether BMP inhibition is necessary and/or sufficient for neural induction. In the chick, misexpression of BMP4 in the prospective neural plate inhibits the expression of definitive neural markers (Sox2 and late Sox3), but does not affect the early expression of Sox3, suggesting that BMP inhibition is required only as a late step during neural induction. Inhibition of BMP signalling by the potent antagonist Smad6, either alone or together with a dominant-negative BMP receptor, Chordin and/or Noggin in competent epiblast is not sufficient to induce expression of Sox2 directly, even in combination with FGF2, FGF3, FGF4 or FGF8 and/or antagonists of Wnt signalling. These results strongly suggest that BMP inhibition is not sufficient for neural induction in the chick embryo. To test this in Xenopus, Smad6 mRNA was injected into the A4 blastomere (which reliably contributes to epidermis but not to neural plate or its border) at the 32-cell stage: expression of neural markers (Sox3 and NCAM) is not induced. We propose that neural induction involves additional signalling events that remain to be identified.