Dna: The Secret of Life

by Watson, James

  In the months following the Alta meeting, Botstein, Davis, and Skolnick, together with Ray White, then at the University of Massachusetts, pursued the RFLP concept. In 1980, a landmark paper that grew out of this collaboration heralded the new age of molecular human genetics. They laid out a clear plan showing how RFLPs could be used, and they worked out the math concerning how many would be needed to ensure that every point in the human genome was within reasonable proximity of at least one RFLP marker – conditions that would in principle permit the mapping of the entire genome. It would be like having enough U.S. cities fixed on a map of North America to allow any unmarked place to be located with respectable accuracy-using only information about how close it was to the labeled cities. But, for the genetic map, what was "reasonable" proximity? Botstein and his colleagues calculated that 150 RFLPs spread uniformly across the entire human genome would be enough. The most immediate benefit of the system was a new strategy for identifying genes that cause disease. Using families in which a disorder spanned several generations, they would take DNA samples from both affected and unaffected individuals. Then they would use recombinant methods to test RFLPs one after another, looking for ones that tracked the disease through the families (see Plate 54).

  In 1979, before the publication of the paper, White presented these ideas at a Cold Spring Harbor Laboratory conference. He noted that "among the more kosher molecular biologists, there was a lot of bitching and grumbling." What he was hearing was great skepticism as to whether the method would work at all; even those who thought it would couldn't agree on the best way to go about using it. These disagreements came into the open during a later meeting to discuss how RFLP linkage analysis could be used to find the gene involved in Huntington disease.

  Nancy Wexler wanted her Lake Maracaibo genealogy to be considered immediately for linkage studies, but Botstein and White thought it was far too early to use RFLP linkage analysis to look for the Huntington gene or any other. They argued that much groundwork needed to be done first – the markers themselves had to be found and mapped – before the technique could be applied for such a specific purpose. In the end, Wexler's determination resulted in a parting of ways: while the Hereditary Disease Foundation pressed on with the hunt for the Huntington gene, Botstein and White pushed for a complete map of the human genome.

  The latter goal required finding RFLP markers on every chromosome, and finding enough of them to ensure that at least one was close to every point in the genome. It was soon necessary to make an upward revision of the initial estimate of 150. But, undeterred, academic laboratories like White's began to isolate RFLPs, and soon commercial biotechnology was getting in on the action as well.

  In 1983 Helen Donis-Keller, an experienced molecular biologist and in those days David Botstein's wife, established the Human Genetics Department at Collaborative Research, Inc., a Boston-area company. She aimed to produce an RFLP linkage map of the whole human genome, with sufficient markers to locate disease genes on any chromosome. The fruits of the effort were published four years later, in a paper aptly entitled "A Genetic Linkage Map of the Human Genome." The map included 403 loci – many more than Botstein's original estimate – and calculations showed that 95 percent of the genome was within reasonable proximity (or "linked") to a marker. It was a great day for genome mapping, but by 1987 rifts and rivalries were appearing once more among the researchers.

  For one thing, there was resentment in academic quarters that Collaborative had incorporated freely available data from university labs while disclosing none of its own. (In this respect, Collaborative was pioneering the best-of-both-worlds strategy that Craig Venter and other would-be genome profiteers were soon to follow in the sequencing sweepstakes.) The French immunologist Jean Dausset, for instance, had been following a somewhat different course. His 1980 Nobel Prize in Physiology or Medicine attracted a generous benefactor, whose substantial gift allowed Dausset to pursue his own strategy for preparing a human linkage map. He realized the task would be much easier if all researchers worldwide were working with a standard set of pedigrees – DNA samples from the same families. So, Dausset created the Centre d'Etude Polymorphisme Humain (CEPH) in Paris to collect pedigrees optimal for genetic analysis: large families with three living generations from which to draw samples. The CEPH collection eventually contained DNA from sixty-one families, including many of the Mormons studied by Ray White, Nancy Wexler's Lake Maracaibo families, and Amish families catalogued by Victor McKusick of the Johns Hopkins Medical School. CEPH made DNA samples from all these families freely available to researchers, with the sole proviso that recipients give their analyses to CEPH for integration into the worldwide database. Collaborative Research took full and fair advantage of this resource.

  By far the most serious criticism of the Collaborative map, however, was the patchiness of the distribution of its markers. Chromosome 7 – linked to cystic fibrosis, one of Collaborative's targets – had 63 markers, but on chromosome 14 only 6 were identified. The distance between markers on the marker-poor chromosomes was very much greater than the average for the genome as a whole. Ray White was particularly upset by Collaborative's claims. He himself had found over 470 markers but had been publishing his data chromosome by chromosome as each was filled in with the required density of RFLPs. "We would never have dreamed of making such a publication with our data set, which is substantially larger than theirs, because we still have significant gaps," he remarked, rejecting Collaborative's grandiose claim. Whether the claims were grandiose or not, though, Collaborative's map had proved the feasibility of genome-wide mapping and was a significant advance.

  But as we have noted, some, like Nancy Wexler, had seen another path opening up in the wake of the breakthrough 1980 RFLP paper. As efforts to produce a comprehensive map gathered steam, David Housman at MIT was gearing up for what David Botstein had declared to be mission impossible at this stage of the game: to discover the location of the Huntington disease gene. He placed this tall order in the hands of Jim Gusella, who had just completed his Ph.D. in Housman's lab. Now the mapping work would surge ahead on another front.

  Botstein's initial pessimism stemmed from the lack of markers: RFLPs looked good on paper, but the work of actually collecting them had only just begun. Indeed, it would take years of effort on the part of White, Donis-Keller, and others for the number of known markers to creep up into the hundreds. Starting out in the dawn of the RFLP era, Gusella had his work cut out for him. By 1982, he had a total of only twelve markers, five he had found himself and seven supplied by others. Wexler meanwhile was back at Lake Maracaibo, trying to fine-tune her genealogy: working out who was married to whom, what children they had, who was whose cousin. Local custom was sometimes a hindrance: some names were quite common, and many individuals were known by more than one. The tree Wexler managed to construct for one family nevertheless wound up with seventeen thousand names on it! Periodically she and her colleagues would set aside a whole day just to collect blood; samples had to be dispatched to Boston all together lest the tropical heat of Lake Maracaibo accelerate the degradation of the DNA.

  As for Gusella, he wasn't waiting for the Lake Maracaibo samples. I remember a meeting at Cold Spring Harbor in October 1982 at which he presented his earliest data. With a small Huntington-afflicted family from Iowa as his sample, he had tested just five of his twelve RFLPs, checking each to see whether it correlated with the disease. None did, and I couldn't help thinking that having set out to find a needle in a haystack, he was making rather much of having lifted out a few straws. Only with careful analysis of the whole haystack – the vast genome in its entirety – or, alternatively, with a lot of luck could anyone hope to find what Gusella was looking for. And so when he closed his talk by saying that the "localization of the HD gene is now just a matter of time," I said to myself, "Yes, a very long time."

  But fortune favors the brave. Gusella returned to his laboratory and tried more RFLP markers. To his astonishment, the twelfth, called
G8, seemed to show linkage with Huntington disease in the Iowa family. But the statistical correlation wasn't very strong. And so he eagerly awaited samples from Lake Maracaibo, testing them for G8 as soon as he received them. Now excitement was irrepressible: G8 indeed tracked with Huntington disease. By the summer of 1983, against all the odds, Gusella had discovered a linkage after trying only twelve RFLPs. But this was no ordinary stroke of luck: for the first time, the gene for a human disorder had been located on a chromosome without the helping hand of sex linkage and without any prior knowledge of the illness's biochemical basis. Suddenly a new scientific vista was opening up: it seemed we would finally be able to analyze rigorously all those genetic defects that have plagued our species for as long as it has existed. RFLPs had proved they were indeed an effective tool. And having traced the Huntington disease gene to a manageable portion of the human genome, it was surely just a matter of time before our powerful gene cloning techniques would lead to the isolation of the gene itself.

  Huntington disease strikes its terrible blow in adulthood. But genetic disorders that strike in childhood have an added awful-ness, afflicting those who have hardly had a chance to live. Following a diagnosis, it is often possible to predict with grim certainty the course of the child's life. Such is the case with Duchenne muscular dystrophy (DMD), a progressive muscle-wasting disease. DMD is a sex-linked disorder: the mutation responsible occurs in a gene carried on the X chromosome. Women may carry the mutation on one of their two X chromosomes, but they are usually protected by the presence of a normal version of the gene on their other X chromosome. It's highly unlikely a female will receive two defective copies since males carrying the mutation almost never survive to have children. If, however, the chromosome with the mutated gene is passed to a son, the boy will develop DMD because he has no other X chromosome to supply a normal copy of the gene. When he is about five years old, his parents will notice he has difficulty getting up from the floor or climbing stairs. By about ten he will need a wheelchair. He will probably die in his late teens or early twenties. DMD is not rare: it affects 1 in every 5,000 male children.

  The hunt for genes involved in human disorders is a story dominated less by great research institutions and plucky entrepreneurs than by groups like the Hereditary Disease Foundation, organizations founded by those with firsthand experience of the devastation a particular genetic illness can bring. Led by people with something very precious at stake, these groups are by nature more willing to back risky or novel research, going where universities or biotech companies may fear to tread.

  The Muscular Dystrophy Association of America and its counterparts in Europe had long supported laboratory research directed at understanding the basic biology of Duchenne muscular dystrophy. In the late seventies cytogeneticists (who study chromosomes microscopically) provided the first genetic clue. Among the very small number of girls who do develop DMD an abnormality was found on the short arm of one of their X chromosomes, at a location called Xp21. Could this be the location of the DMD gene?

  Not long thereafter, Bob Williamson at St. Mary's Hospital Medical School in London initiated RFLP-based searches for both the gene causing cystic fibrosis and the one involved in DMD. His colleague Kay Davies hunted up RFLPs on the X chromosome and tested them for linkage to Duchenne. She was successful, and the clincher was their location: they were in the Xp21 region, just as would have been expected given those strange X chromosomes in the women with DMD.

  While the gene hunters pushed ahead trying to isolate the genes involved in Huntington disease and DMD, a revolution of a quieter kind was taking place in the offices of clinical geneticists. From the first, Nancy Wexler and David Housman realized that RFLPs linked to a disease gene could be used not only to localize the gene itself but also as a diagnostic test to determine which members of a particular family were carrying the mutation. They could be used to test even an unborn child. Consider the case of a hypothetical family with DMD. At least one boy will be diagnosed – the "index case" that first reveals the presence of a DMD mutation in the family. His mother, the carrier of a mutated gene, also has one normal copy. Her sisters may also be carriers, so any sons they may have are at risk. Now suppose the mother becomes pregnant once again with a male fetus; the chances are 50-50 that the second son will be affected. But with RFLPs her physician can tell her what fate awaits that fetus if carried to term.

  First, the affected son's X chromosome is analyzed to identify the particular RFLPs linked to the DMD gene in this family. Next, DNA is taken from the fetus, either a sample of the placenta or of the amniotic fluid, which contains fetal cells. If the fetus's RFLPs match those of the affected boy, then we can be pretty certain that the unborn fetus will also be affected. Why only pretty certain? As we saw in chapter 1, when egg cells are produced, the chromosome pairs undergo recombination, exchanging DNA: the two copies of chromosome 1 trade with each other, as do the two copies of chromosome 2, the two copies of the X chromosome, and so on. If this swap should occur at a point on the X chromosome between the RFLP markers and the DMD gene, the RFLPs we have found to be associated with the normal version could possibly wind up associated with the mutated (DMD) copy. Experience taught us that with the first RFLPs for DMD this happens about 5 percent of the time, and so RFLP-based diagnosis has only a 95 percent chance of being accurate. This degree of imprecision is an unavoidable consequence of recombination. So while such diagnosis represented a tremendous advance, absolute certainty depended on identifying the gene itself, not merely the markers associated with it.

  The key to isolating the DMD gene was a young boy named Bruce Bryer, whose X chromosome was missing a very large piece from the short arm. The piece was so large that Bruce suffered from three other genetic disorders in addition to DMD. In 1985 Lou Kunkel at Harvard Medical School reasoned he could use Bruce's DNA to "fish out" a normal gene from the DNA of an unaffected boy. Bruce's case was special because the disease was caused not by a defective copy of the gene but by its complete absence. Kunkel realized that all of Bruce's DNA should be present in a normal boy's, but whatever sequences the latter had and Bruce lacked would hold the key. Using recombinant methods, Kunkel subtracted Bryer's DNA from the normal DNA and kept the difference – the DNA that should contain the DMD gene. The subtraction didn't work perfectly, but it did work well enough that he could find the DNA pieces he wanted by using genetic markers associated with the Xp21 region.

  Tony Monaco, a graduate student with Kunkel, took on the job of determining which, if any, of these Xp21 pieces of DNA might constitute part of the DMD gene itself. The only way to do this was to test each piece against DNA from several unrelated patients with DMD. Monaco hit the jackpot with the eighth try: a sequence called pERT87 was found to be absent in five of his DMD boys. This meant almost certainly that pERT87 was very close to the gene and perhaps even a part of it. Monaco began to isolate other sequences close to pERT87, and these too proved to be missing in the DNA of DMD patients. By 1987, Kunkel's group had isolated the complete gene. Now it could be given a proper name: dystrophin. Even with the genome sequence completed, it still holds the record for largest gene in the human genome, owing mainly to its many large introns.

  Immediately the new knowledge was applied to produce foolproof prenatal diagnosis for DMD. And soon scientists discovered that a range of different mutations could impair dystrophin and cause the disease. But it remained unclear what the gene actually did. Would its function give us clues to developing effective therapies for DMD?

  The first step was to locate the protein produced by the gene in muscle cells. Eric Hoffman in Lou Kunkel's laboratory found that the dystrophin protein was typically located in muscle cells just below the membrane that encloses the muscle fiber. Further studies have revealed dystrophin's critical role in connecting proteins that make up the muscle cell's interior architecture to a set of molecules that span the cell membrane and interact with other proteins outside the cell. The linking of the interior molecules to those
in the membrane somehow secures the cell membrane when muscles contract and relax. Without dystrophin, the membrane suffers damage and the muscle dies cell by cell. Given our new and detailed knowledge of dystrophin and its function, it may seem remarkable that there is still no cure for DMD. This is the central frustration inherent in the current state of the art: genetics has made it possible to identify and understand disease, without yet permitting us in most cases to right the genetic wrong.

  Kunkel's approach typifies the modern mapping-based approach to dissecting a disorder. Though now common practice, when Kunkel applied it the method was far enough beyond the bounds of research orthodoxy that the Muscular Dystrophy Association was taking something of a gamble in supporting four years of his efforts – a gamble that paid off handsomely. In the old days you tried to use biochemical analyses of a disease's symptoms to identify the disease gene; these days, following Kunkel, you map the gene, and then interpret the symptoms in the light of the gene's function.