¿ìè¶ÌÊÓƵ

(see Graphic) The signs were barely obvious when ‘Jonathan’ was born: not much more than a poor appetite and unusually small hands and feet. But by his fourth birthday he had begun to eat compulsively and was clearly suffering from mild mental retardation. Jonathan had Prader-Willi syndrome, one of the most common forms of genetic obesity.FIG-mg18744301.jpg

Most of the 1 in 10 000 children affected by this disorder carry a genetic defect which can be seen under a powerful microscope: the deletion of a small segment of one of their two copies of chromosome 15. But in a few children, such as Jonathan, the causes are more subtle. None of their 22 pairs of autosomal chromosomes or two sex chromosomes harbour any obvious defect.

Until a few years ago researchers assumed that these children must carry a genetic mutation which is simply too small to see under a microscope. But in 1989 it became clear that some carry no defects whatsoever. The causes of their illness lie instead with a mysterious phenomenon called genomic imprinting, in which the eggs and sperm of parents attempt to stamp the authority of their sex on the DNA of future generations.

TELL-TALE FINGERPRINTS

Genomic imprinting had previously been detected in the mouse. Its discovery in humans came from DNA fingerprinting studies. In the mid-1980s, Harvard researchers studying Prader-Willi syndrome had begun to use DNA fingerprinting techniques to compare the children’s chromosomes with those of their parents. One of their aims was to trace exactly which chromosomes had been inherited from each parent. To their astonishment, the researchers discovered that in Jonathan’s case, his two copies of chromosome 15 both came from his mother. His other chromosomes had been inherited in the normal way, one from each parent. But with chromosome 15 some kind of error had evidently occurred – either when the chromosomes were being packaged into eggs and sperm, or when the newly fertilised egg was undergoing its first few rounds of cell division.

Either way, Robert Nicholls and his team at Harvard Medical School, along with their collaborators at universities in Tennessee and Michigan, had stumbled on an entirely new class of genetic disorder.

Previously, the causes of inherited diseases had been traced to genetic defects, or to the presence of one too many (or too few) chromosomes. But now it was clear that genetic disorders could also afflict children unlucky enough to inherit an unbalanced mix of maternal and paternal DNA. The corollary was nothing short of a revelation: the copies of chromosomes we inherit from our fathers must, in some obscure way, be different from the corresponding copies we inherit from our mothers. Human chromosomes – and, by implication, some of the tens of thousands of genes etched into their DNA – must carry some kind of physical imprint of the sex of the parent from whom they were passed on. In short, parents deal their offspring genes from a marked deck.

Today, genomic imprinting is coming under unprecedented scrutiny as molecular biologists search for clues to its physical origins. The stakes are high. In the four years since the discovery about Jonathan, researchers have uncovered the hidden hand of imprinting in other diseases, ranging from developmental disorders to childhood cancers. Teams working on Huntington’s disease and a condition known as Fragile X syndrome, for example, believe that imprinting could help to explain why the symptoms occasionally become more severe as they are passed from one generation to the next .

Studies of genetically engineered mice reveal that at least some cases of imprinting in mammals occur when DNA is chemically ‘marked’ by methylation inside sperm and egg cells. But researchers agree that, overall, imprinting involves more than one molecular process, and is likely to differ from one organism to the next (it occurs not only in mammals but in yeast, some insects and certain plants).

ELUSIVE IMPRINTS

Despite these breakthroughs, geneticists still have a long way to go in unravelling all the causes and consequences of imprinting. One obstacle is that the number of human genes affected by imprinting is likely to be extremely small, making them hard to trace. In mice, for instance, researchers have so far identified only a handful of imprinted genes. ‘For the vast majority of our genes . . . there is no difference between the gene from Mom and the gene from Dad,’ says Daniel Driscoll, a molecular geneticist at the University of Florida, Gainsville.

Unfortunately, the small number of human genes which are imprinted are likely to be major players in disease. Normally we inherit two, equipotent copies of each gene, one from each parent. If one of these copies is disabled, the second can usually compensate for the loss. But with imprinted genes, the maternal and paternal copies are not equally active in cells and so cannot substitute for each other. Geneticists believe this effect might explain a host of long-standing medical puzzles: why, for example, children are more likely to develop diabetes mellitus if their father, rather than their mother, suffered from the disease; and why people are more likely to suffer from asthma or hay fever if their mother, as opposed to their father, suffered from them.

It will be some time before researchers work out how imprinting affects diseases such as these, but few doubt that the phenomenon has far-reaching implications. With many genetic disorders, ranging from manic-depressive syndrome to hereditary ovarian disease, the patterns of inheritance are confusing, a finding that may be due to imprinting. It is one more factor for parents and doctors to worry over as they attempt to predict whether children in affected families will suffer from the disorders. ‘I think the whole imprinting field is going to make genetics tougher and make us more hesitant about what’s going on and how we do genetic counselling,’ says Driscoll.

In the meantime, researchers are concentrating on the handful of diseases with a proven link to imprinting. Some of the clearest examples are developmental disorders, explains Judith Hall, a biologist studying imprinting at the University of British Columbia in Vancouver. She cites the fact that children who inherit both copies of chromosomes 7 or 16 from their mother tend to suffer retarded growth – the implication being that there are certain paternal genes on these chromosomes which are important to growth. But it is chromosomes 15 and 11 that have so far grabbed most of the attention.

MISPLACED LAUGHTER

A worldwide study found that one in five children with Prader-Willi syndrome carry two maternal copies of chromosome 15. In 1991, researchers in Britain discovered that inheriting two paternal copies of chromosome 15 also causes problems. In this case, patients suffer from Angelman’s syndrome, which leaves children mentally retarded, prone to jerky movements and liable to burst into misplaced laughter. Once called the ‘happy puppet’ syndrome, it is caused by an absence of maternal genes belonging to chromosome 15 – in direct contrast to Prader-Willi syndrome. Angelman’s syndrome can also occur if a damaged copy of chromosome 15 is inherited from the mother: an unbalanced mixture of maternal and paternal genes can arise in several different ways but cause the same disease.

This message is driven home by research into Beckwith-Wiedemann syndrome, where imprinting leads to childhood cancer as well as developmental abnormalities. Affected babies grow unusually rapidly in the womb, and are born with physical abnormalities such as a large tongue. In some cases, the disorder occurs because the patient has inherited two paternal copies of chromosome 11. But sometimes it strikes children with equal numbers of maternal and paternal chromosomes. And these cases involve a different kind of genetic blunder: the erroneous duplication of a small segment of the paternal copy of chromosome 11. Either way, the final outcome is the same – children burdened with too many paternal copies of chromosome-11 genes.

The dream of any researcher studying imprinting is to identify the genes affected in specific diseases. In the case of Beckwith-Wiedemann syndrome, a vital clue has emerged. The segment of chromosome 11 that is duplicated in some patients was already familiar to biologists because it carries a gene for a growth hormone called ‘insulin-like growth factor 2’. In mice, only the pa-ternal copy of the IGF2 gene is active in most cells: the copy inherited from the mother is not expressed. If the same is true of the human IGF2 gene, reason the researchers, then inheriting two paternal copies of the gene could lead cells to overproduce IGF2. This, in turn, might explain one of the symptoms of Beckwith-Wiedemann syndrome – the unusually rapid growth of embryos. Sometimes Beckwith-Wiedemann syndrome results from a mutation on the maternal copy of chromosome 11. Perhaps this mutation also leads to an excess of the growth hormone.

Jack Tarleton runs a DNA diagnostic laboratory at the Greenwood Genetic Center in Greenwood, South Carolina, and is committed to discovering the different molecular mechanisms that cause Beckwith-Wiedemann syndrome. Although most cases of the disorder result from chance chromosomal errors, made during the formation of eggs and sperm for example, some parents carry genetic defects that cause the syndrome and so can transmit it to more than one child. Tarleton wants to identify the molecular defect responsible for this transmissible form of the disorder. ‘We could then do DNA-based testing,’ he explains, to advise parents who want to know whether their future offspring are likely to inherit the disorder.

Another reason researchers are so excited about imprinting is its connection with childhood cancers. Children with Beckwith-Wiedemann syndrome are strongly predisposed to developing a cancer of the kidney called Wilms’ tumour, as are children who inherit certain kinds of genetic mutation on chromosome 11. Over the past six years it has become clear that imprinting probably contributes to these and other childhood cancers, including tumours of the bone, retina and muscle. The common ingredients are so-called ‘tumour-suppressor’ genes, which act to check the uncontrolled cell growth that leads to tumours.

Most people inherit two working copies of each tumour-suppressor gene, one from each parent. Only if both copies become disabled in the same cell does a tumour develop. Studies of tumour cells isolated from children with cancer reveal a two-step mechanism. The first tumour-suppressor gene to be ‘hit’ is usually disabled by a mutation in its DNA sequence. This leaves only the second copy of the gene to prevent the cell from becoming cancerous. What happens next is quite strange: the second copy of the gene simply vanishes from the cell. In some cases, the entire chromosome carrying the gene disappears due to chance mistakes made during the complex sequence of events leading to cell division. In others, the gene is ousted from the chromosome by its sibling – and by now dysfunctional – gene, in a process known as gene conversion. Either way, the cell is left free to proliferate and form a tumour.

All things being equal, there is no reason to suppose that a tumour-suppressor gene inherited from your father is any more or less likely to be ‘hit’ than one inherited from your mother. But the evidence from children with Wilms’ tumours, which is linked to a gene known as WT1, suggests otherwise. In the vast majority of Wilms’ tumour cells that lack WT1 activity, it is the paternal copy of the gene which is first disabled and the maternal copy that subsequently vanishes from the cell. The inescapable conclusion is that tumour-suppressor genes are imprinted: the reason some children are predisposed to Wilms’ tumour is that the copy of WT1 they inherit from their father is either disabled at birth or predisposed to accumulating disabling mutations later in life. Both kinds of imprinting may play a part in other childhood cancers, says Judith Hall.

A completely different kind of inborn molecular problem has been traced to imprinting in leukaemia. Recently, biologists at the Children’s Cancer Research Institute at St Anna Children’s Hospital in Vienna discovered a role for imprinting in chronic myeloid leukaemia, a form of the disease where the leukaemic cells contain an unusual hybrid chromosome. This so-called Philadelphia chromosome consists of virtually the whole of chromosome 22 fused to a small section of chromosome 9. At the intersection, two genes, one from each of the original chromosomes, dovetail to form a single ‘fusion gene’. It is this hybrid gene which, for reasons that remain obscure, seems to cause the problems. It causes leukaemia if implanted into mice.

Imprinting plays a vital part in the fusion. The Vienna team found that each parent makes a different contribution to the Philadelphia chromosome: the chromosome 22 section is always inherited from the mother, the chromosome 9 section from the father. One explanation is that only this combination gives a functional fusion gene, perhaps because each of the two genes involved is imprinted in a particular manner. But nobody knows for sure: detecting the hidden hand of imprinting in disease is easier than uncovering its molecular origins.

Only by tracing these origins, however, will geneticists be able to answer some of the most fundamental questions about imprinting. Why should two genes with exactly the same DNA sequence behave differently simply because they were passed on from different sexes? Why does imprinting affect some genes and not others? And what evolutionary function does imprinting serve? For clues researchers have turned to transgenic mice.

One practical problem in investigating the molecular basis of imprinting is how to be certain of a gene’s parentage. With mice, molecular biologists can implant artificial ‘reporter’ genes into embryos and follow the fates of these genes as they are passed on from one generation to the next. Once implanted, the genes fit into chromosomes at random positions, and whether they subsequently become imprinted seems to be a hit-or-miss affair. In some cases, where transgenes do become imprinted in mice, the genes are active only when passed on from females; in other cases, the genes function only when passed on from males. Perhaps, says Driscoll, transgenes only become imprinted if they end up in particular regions of chromosomes.

The same kind of research has provided powerful clues to the physical origins of imprinting. It has been known for more than 30 years that cells contain enzymes which in certain circumstances act to modify the chemical structure of DNA, decorating stretches of its long helical chain with methyl groups. Bacteria, for example, methylate their DNA to distinguish it from DNA belonging to viruses. For over a decade researchers have known, too, that in cells with nuclei, methylating DNA has the effect of muzzling any genes it encodes. Methylation is reversible, giving cells an ‘on-off’ switch for gene activity.

Yet the link between DNA methylation and imprinting is a recent discovery. The main evidence comes from two observations. The first is that when imprinted transgenes are passed from one generation of mice to the next, the genes are switched either on or off, depending on the sex of the parent carrying the gene. Secondly, whenever a transgene is switched off, it is next to a stretch of DNA that is more heavily methylated.

It is not yet clear if methylation is the main way parents mark their offspring’s genes, or if it is a secondary response to a different and as yet unidentified mark. But whatever the eventual answer, one thing looks certain: the genes acquire their marks as they are packaged into eggs and sperm. Concrete evidence for this has come from Richard Chaillet and his colleagues at the Harvard Medical School, who have looked for signs of DNA methylation in the germ-line cells (the cells from which eggs and sperm develop) in mice. Their findings show that the methylation patterns around imprinted transgenes are ‘wiped clean’ in germ-line cells, only to be re-established later, in the final steps of egg and sperm development.

It is impossible to say whether biologists already have the measure of imprinting’s role in human diseases, or whether what they have found is merely the tip of the iceberg. Most of the diseases and disorders with a proven link to imprinting involve relatively simple genetics, so they can be traced to mutations in just one or two genes. But the majority of common diseases involve many different genes – a complication which will make the role of imprinting much harder to unravel.

* * *

Origins of imprinting

Why did imprinting evolve in mammals? One suggestion is that it restrains embryo growth so that female mammals can survive pregnancy (see Science, ¿ìè¶ÌÊÓƵ, 8 June 1991). Another theory links imprinting to the development of that uniquely mammalian feature, the placenta. Here, vital clues have come from observations of the fates of eggs that mistakenly inherit all their chromosomes from only one parent.

Rare biological ‘errors’ made during the first division of a fertilised egg sometimes produce a cell that contains 46 paternal chromosomes and no maternal chromosomes. Such cells never develop into normal embryos. Instead, they grow into hydatidiform moles, which consist of mainly placental tissue and little or no embryonic tissue. Paternal genes alone can make the placenta develop but not the embryo.

The converse case, inheritance of only maternal chromosomes, occurs when chance errors cause unfertilised eggs to form with 46 chromosomes, instead of the usual 23. Such eggs begin to divide while still in the ovary, where, in a bizarre approximation of normal development, they form benign tumours called teratomas. Teratomas contain many types of tissue – connective tissue, nervous tissue, bone and tooth cells – but no placental tissue. By themselves, it seems, maternal genes can trigger the development of embryonic tissues but not placental tissue.

A similar division of labour between maternal and paternal genes occurs in the development of the mouse embryo and placenta. It might well be a general barrier to unfertilised mammalian eggs producing viable embryos. Mammals are the only vertebrates in which the birth of young from unfertilised eggs has never been recorded – implying that mammals are the only vertebrates in which genomic imprinting occurs.

* * *

Anticipating disease

Huntington’s disease usually begins to affect its victims between the ages of 30 and 50. But about 1 in 10 people who inherit the disease develop symptoms before the age of 20. In 90 per cent of such cases, the genetic defect that causes the disease is inherited from the father. This tendency of certain genetic diseases to occasionally become more severe as they are passed on from one generation to the next is known as anticipation.

Much of the evidence linking imprinting to anticipation has emerged from studies of Fragile X syndrome, the most common form of inherited mental retardation. Anticipation increases the chance that in successive generations of families with FXS, children will develop symptoms rather than remain asymptomatic carriers. Since anticipation only happens when the mother is the carrier, there is every reason to suspect imprinting.

In Fragile X syndrome the X chromosomes often break when patients’ cells are cultured in a particular way. At the breaking point lies a gene called FMR1, embedded in a stretch of DNA that contains a ‘three-letter’ code repeated like a molecular stutter.

Researchers believe that the length of this ‘trinucleotide repeat’ holds the key to the disease. In normal people the three-letter code is repeated 6 to 58 times, in FXS carriers 60 to 230 times, and in people with the symptoms of FXS well over 230 times. As the repeat grows, so the nearby FMR1 gene becomes less and less functional.

The connection with anticipation is not hard to divine. Any genetic mutations which increase the length of the repeat will increase the likelihood of a child developing the symptoms. In FXS carriers, such mutations are far more likely in egg cells than sperm – no one knows why – and any that do occur in sperm never seem to increase the length of the repeat past the critical point where disease symptoms appear. As a result, anticipation occurs only when the mother is the carrier.

Nobody yet knows why the repeat is more likely to mutate in women carrying FXS than in normal people. But one idea is that the probability of mutations is controlled by the length of the repeat itself: the longer it is, the greater the chance of a mutation, and the more the repeat grows from one generation to the next. This new under-standing of FXS should help in diagnoses. By measuring the approximate length of the molecular stutter that causes FXS, researchers should in future be able to tell who has the syndrome and who is a carrier.

Unfortunately, however, the expansion of the repeat is too erratic for genetic counsellors to tell women carrying FXS exactly how likely it is that their children will suffer FXS symptoms. And to add to the confusion, researchers in Belgium and the Netherlands last year discovered that the repeat can even contract as the gene passes from one generation to the next.

A similar picture is now being pieced together for Huntington’s disease, where anticipation usually occurs when the disease is passed from father to child. Earlier this year, a 10-year quest for the Huntington’s gene ended when researchers discovered that the disorder is caused by a trinucleo-tide repeat disrupting the function of a vital gene (see ‘To catch a killer gene’, ¿ìè¶ÌÊÓƵ, 24 April). Discovering why the disease begins so early in some cases might help researchers to discover how to delay the onset of the disease, says Michael Conneally at the School of Medicine of Purdue University in Indianapolis, a member of the group that first documented anticipation in Huntington’s disease.