Experimental Embryology

The next question was how cells become committed. In Freiburg Hans Spemann developed microsurgery with fine glass needles and loops of baby's hair. For her doctorate, Hilde Mangold took a small piece of tissue from the dorsal lip, the edge of the pore where cells stream into the early newt embryo, and grafted it into the belly of an embryo of a darker species. The host grew a second back, with a second nerve cord and muscle blocks, built mostly from its own dark cells. The graft had not made the new body. It had organised it.

Mangold died in 1924, before the paper appeared. Spemann received the Nobel prize in 1935. Induction, one tissue instructing another, became the central idea of the field, and embryologists spent decades searching for the organiser's chemical signal without finding it. The tools to identify it arrived only with developmental genetics.

Two ideas from outside the laboratory shaped the next phase. In 1952 Alan Turing showed that two chemicals reacting and diffusing at different rates can create spots and stripes from nothing, a mathematical result that belongs as much to differential equations as to biology. In 1969 Lewis Wolpert proposed positional information: a cell reads its position from a signal that fades with distance, and responds according to its own history. His French flag, a row of cells split into blue, white and red thirds by two thresholds, became the standard picture.

Meanwhile the old question of the nucleus was settled. If cells become different by losing genes, a specialised nucleus should fail to support development. Robert Briggs and Thomas King learned to transplant a nucleus into an egg whose own nucleus had been removed. Nuclei from early embryos made normal tadpoles, those from older embryos rarely did. In 1962 John Gurdon got tadpoles, and later fertile frogs, from the nuclei of specialised gut cells. Specialised cells keep their genes. They silence most of them.

Nuclear transfer almost always fails, which made the successes hard to interpret. Here are two landmark experiments:

Dolly was made at the Roslin Institute by a team led by Ian Wilmut and Keith Campbell. Of the 277 fused eggs, 29 grew into early embryos that could be transferred to surrogate ewes, about 10 per cent, and one of those 29, about 3 per cent, was born.

How many attempts should a laboratory expect to need? If each attempt succeeds independently with chance $p$, the chance that $n$ attempts all fail is $(1-p)^n$. Setting this to one half and solving,

For Gurdon's rate, $p = 10/726$, this gives $n \approx 50$ attempts for an even chance of a tadpole. For Dolly's, $p = 1/277$, it gives $n \approx 192$. To be 95 per cent sure of at least one lamb would need about 830 attempts. The Roslin team had been lucky, or determined, and many doubted that the result could be repeated. It was, in mice, in 1998.

The low rates had a sharper consequence. A population of "specialised" cells may contain a few unspecialised ones, and if only one transfer in a hundred works, those few could account for every success. So a rare success does not by itself prove that a fully specialised nucleus can make an animal. The proof came in 2002, when mice were cloned from mature immune cells. Such cells permanently rearrange their antibody genes as they specialise, and the cloned mice carried the rearrangement in every cell. Their nucleus had certainly come from a specialised cell.

By 1970 experimental embryology had framed the right questions. Cells are told what to become by signals from their neighbours, they read their position, and they keep every gene while using only some. What the signals were, and which genes read them, could not be answered by surgery. The answers came from mutant fruit flies in developmental genetics. Gurdon's result also had a second life. If an egg can reset a specialised nucleus, perhaps a specialised cell could be reset without an egg at all. That question led to stem cell biology.