Chasing the elusive inducer

The Xenopus experiments that demonstrated the default model of neural induction

For more than six decades, scientists around the world searched for the molecular basis of neural-plate induction, without much luck. Back in 1924, Spemann and Mangold had shown that a speck of tissue from the dorsal lip of the early newt gastrula could induce a second neural plate when transplanted into an ectopic site. This group of dorsal lip cells, the 'organizer', developed into mesodermal notochord and induced neural fates in the overlying ectoderm (see Milestone 1).

The standard hypothesis, arising from this and subsequent experiments, postulated that the organizer and resulting notochord, through secreted soluble molecules, instruct neural-plate differentiation in the overlying ectoderm. This straightforward idea took a few knocks over the years as it became apparent that a bewildering array of substances, and even physical manipulation of early embryos, could induce ectopic neural differentiation, and none of these were even remotely plausible candidates for an organizer signal. Nevertheless, this hypothesis reigned into the 1980s.

Then, in 1989, Grunz and Tacke published a surprising observation. It was known that isolated early ectoderm explants differentiated into epidermal tissue in vitro, but what if normal cell–cell interactions were disrupted? If the cells were dissociated then randomly reaggregated immediately, the result was again epidermis. But if reaggregation was delayed for an hour or more, neural markers began to be expressed, and if dissociated ectodermal cells were kept in suspension for 5 hours before reaggregation, they differentiated into neural tissue exclusively. This indicated that the absence, not the presence, of an intercellular signal was necessary for neural differentiation. Neural fate might indeed be the 'default' fate of ectodermal cells.

It was time for molecular biologists to step into the fray. Smith and Harland developed an assay to identify mRNAs that coded for proteins that could induce a neural plate in ventralized Xenopus embryos. They generated and systematically screened a library of mRNAs derived from hyperdorsalized gastrula-stage embryos. In 1992, they isolated noggin, a novel mRNA that could not only induce a neural plate (and complete dorso–ventral axis) without requiring the presence of mesoderm, but also was expressed appropriately in the dorsal lip and in the notochord. However, the mechanism of how noggin caused neural-plate formation, and how it fitted into the new 'default hypothesis', remained obscure.

The default hypothesis was bolstered by Hemmati-Brivanlou and Melton, who reported in 1994 that inhibition of the activin II receptor could, by itself, induce neural differentiation in the absence of the notochord.

In 1995, at last, the concept of a neural inducer and the idea of neural phenotype as the default outcome of ectodermal differentiation came together. De Robertis and colleagues identified chordin, another new dorsal lip protein that could directly induce a neuraxis. They also showed that bone morphogenetic protein-4 (BMP4), a growth factor related to activin, inhibited the neuralizing activity of both chordin and noggin in Xenopus embryos. The same group soon found that chordin directly binds and inactivates BMP4, and Harland and colleagues reported a similar function for noggin.

So, the default hypothesis seemed finally to stand on solid genetic and biochemical ground. It was proposed that neural induction requires the inhibition — by noggin, chordin or other molecules — of the epidermis-inducing, anti-neural BMP4 signal. The search for neural-plate instructive signalling molecules had lasted for seven decades and ended with the recognition that the instructive signal as such did not exist.

This is not the end of the story, however. Recent findings indicate that the neural ectoderm is specified in the blastula, before the Spemann organizer even forms. Fibroblast growth factor (FGF) signalling is required at this stage to enable later neural differentiation. In Xenopus at least, the anterior — but not the posterior — neural ectoderm is specified at the blastula stage, through a mechanism that already involves noggin and chordin. These and other studies are now elucidating the genesis of the 'default' state.