Matt Ridley

Somehow, imprinted genes escape this process. They resist the demethylation. There are intriguing hints about how this is achieved, but nothing definitive.9

That imprinted genes escape demethylation is, we now know, all that stood between science and the cloning of mammals for many years. Toads were fairly easily cloned by putting genes from a body cell into a fertilised egg, but it just didn't work with mammals, because the genome of a female's body cells had certain critical genes switched off by methylation and the genome of a male's body cells had other genes switched off - the imprinted genes. So, confidently, scientists followed the discovery of imprinting with the announcement that cloning a mammal was impossible. A cloned mammal would be born with all its imprinted genes either on or off on both chromosomes, upsetting the doses required by the cells of the animal and causing development to fail. 'A logical conse-S E X 2 1 3

quence', wrote the scientists who discovered imprinting,10 'is the unlikelihood of successful cloning of mammals using somatic cell nuclei.'

Then, suddenly, along came Dolly the cloned Scottish sheep in early 1997. Quite how she and those that came after evaded the imprinting problem remains a mystery, even to her creators, but it seems that a certain part of the treatment meted out to her cells during the procedure erased all genetic imprints.11

The imprinted region of chromosome 15 contains about eight genes. One of these is responsible, when broken, for Angelman syndrome: a gene called UBE3A. Immediately beside this gene are two genes that are candidates for causing Prader-Willi syndrome when broken, one called SNRPN and the other called IPW. There could be others, but let us assume for the moment that SNRPN

is the culprit.

The diseases, though, do not always result from a mutation in one of these genes but from an accident of a different kind. When an egg is formed inside a woman's ovary, it usually receives one copy of each chromosome, but in rare cases where a pair of parental chromosomes fails to separate, the egg ends up with two copies.

After fertilisation with a sperm, the embryo now has three copies of that chromosome, two from the mother and one from the father.

This is especially likely in elder mothers, and is generally fatal to the egg. The embryo can go on to develop into a viable foetus and survive more than a few days after birth only if the triplicate chromosome is number 21, the smallest of the chromosomes - the result being Down syndrome. In other cases the extra chromosome would so upset the biochemistry of the cells that development would fail.

However, in most cases, before that stage is reached, the body has a way of dealing with this triplet problem. It 'deletes' one chromosome altogether, leaving two, as intended. The difficulty is that it does so at random. It cannot be sure that it is deleting one of the two maternal chromosomes, or the single paternal one. Random deletion has a sixty-six per cent chance of getting one of the maternal ones, but accidents do happen. If, by mistake, it deletes the paternal 2 1 4 G E N O M E

one, then the embryo goes merrily on its way with two maternal chromosomes. In most cases this could not matter less, but if the tripled chromosome is number 15, you can see immediately what will ensue. Two copies of UBE3A, the maternally imprinted gene, are expressed, and no copies of SNRPN, the paternally imprinted gene. The result is Prader-Willi syndrome.12

Superficially, UBE3A does not look a very interesting gene. Its protein product is a type of 'E3 ubiquitin ligase', members of an obscure proteinaceous middle management within certain skin and lymph cells. Then in the middle of 1997, three different groups of scientists suddenly discovered that, in both mice and human beings, UBE3A is switched on in the brain. This is dynamite. The symptoms of both Prader-Willi and Angelman indicate something unusual about the brains of their victims. What is even more striking is that there is good evidence that other imprinted genes are active in the brain. In particular, it seems that in mice much of the forebrain is built by maternally imprinted genes, while much of the hypothalamus, at the base of the brain, is built by paternally imprinted genes.13

This imbalance was discovered by an ingenious piece of scientific work: the creation of mouse 'chimeras'. Chimeras are fused bodies of two genetically distinct individuals. They occur naturally — you may have met some or even be one yourself, though you will not know it without a detailed study of the chromosomes. Two genetically distinct embryos happen to fuse together and grow as if they were one. Think of them as the opposite of identical twins: two different genomes in one body, instead of two different bodies with the same genome.

It is comparatively easy to make mouse chimeras in the laboratory by gently fusing the cells from two early embryos. But what the ingenious Cambridge team did in this case was to fuse a normal mouse embryo with an embryo that was made by 'fertilising' an egg with another egg's nucleus, so that it had purely maternal genes and no contribution from the father. The result was a mouse with an unusually large head. When these scientists made a chimera between a normal embryo and an embryo derived only from the father (i.e., S E X 2 1 5

grown from an egg whose nucleus had been replaced by two sperm nuclei), the result was the opposite: a mouse with a big body and a small head. By equipping the maternal cells with the biochemical equivalent of special radio transmitters to send out signals of their presence, they were able to make the remarkable discovery that most of the striatum, cortex and hippocampus of the mouse brain are consistently made by these maternal cells, but that such cells are excluded from the hypothalamus. The cortex is the place where sensory information is processed and behaviour is produced.

Paternal cells, by contrast, are comparatively scarce in the brain, but much commoner in the muscles. Where they do appear in the brain, however, they contribute to the development of the hypothalamus, amygdala and preoptic area. These areas comprise part of the 'limbic system' and are responsible for the control of emotions. In the opinion of one scientist, Robert Trivers, this difference reflects the fact that the cortex has the job of co- operating with maternal relatives while the hypothalamus is an egotistical organ.14

In other words, if we are to believe that the placenta is an organ that the father's genes do not trust the mother's genes to make, then the cerebral cortex is an organ that the mother's genes do not trust the father's genes to make. If we are like mice, we may be walking around with our mothers' thinking and our fathers' moods (to the extent that thoughts and moods are inherited at all). In 1998 another imprinted gene came to light in mice, which had the remarkable property of determining a female mouse's maternal behaviour. Mice with this Mest gene intact are good, caring mothers to their pups. Female mice who lack a working copy of the gene are also normal except that they make terrible mothers. They fail to build decent nests, they fail to haul their pups back to the nest when they wander, they do not keep the pups clean and they generally seem not to care. Their pups usually die. Inexplicably, the gene is paternally inherited. Only the version inherited from the father functions; the mother's version remains silent.15

The Haig theory of conflict over embryonic growth does not easily explain these facts. But the Japanese biologist Yoh Iwasa has 2 l 6 G E N O M E