by Bryan Sykes
The two exceptions to this theoretical scenario are the direct matrilineal and patrilineal ancestries that are by now so familiar. They coalesce into universal ancestors a lot further back, around 65,000 years for the “Y-chromosome Adam” and 170,000 for “Mitochondrial Eve.” Why the difference? It all has to do with the behavior of men who have more than their fair share of children, something I explore in depth in Adam’s Curse.
Even without the complications of the fairly recent universal ancestors, our genomic DNA ancestry is a snarled tangle compared with the linear simplicity traced by mitochondrial and Y-chromosome DNA. We know exactly which ancestral path they have taken from the deep past to the present. They are the two clear voices above the tumult of the genome, but despite their clarity they cannot tell us the complete story of our ancestry, and a lot of it remains hidden. I used to think this was a blessing in disguise because it is almost impossible to grasp the concept and complexity of our complete DNA ancestry in a way that means anything. All narrative is lost as the numbers of our ancestors grow into the thousands and beyond. For a long time I could see no way of narrating our genetic past that went beyond mitochondria and Y chromosomes. I was happy listening to the crystal-clear voices as the Callas and Domingo of genetics sang their duets. I wasn’t interested in the cacophony of the orchestra and chorus.
Then, by chance, I did catch something in the air: the faintest possibility of a melody rising above the swirl of countless different instruments. It came in late 2008, when a former colleague, John Loughlin, was explaining to me how a new DNA technology was making his work on the genetics of osteoarthritis so much easier. He had been looking for genes involved in the cascade of biochemical events that lead to severe arthritis, requiring a major joint replacement. This was a very reasonable quest because he had already shown that this particular form of osteoarthritis had a high hereditable component, ergo there must be genes involved. But as so many prospectors of the genome have discovered, it is hard work, and the rewards rarely match up to the promise first imagined—more like panning laboriously for specks of gold than striking a mother lode. I knew that he had spent many years teasing apart the genomes of hundreds of patients who had undergone joint-replacement surgery. He and his team were looking for differences between these patients and the hundreds of other people of similar age and background who did not suffer from osteoarthritis. Like other scientists on a similar quest, he used the enormous range of genetic markers that had been discovered en route to decoding the entire human gene sequence. The theoretical basis was that if his joint-replacement patients inherited particular versions of any of these markers, compared with the control group, then this might indicate the chromosomal location of an “osteoarthritis gene.” It didn’t mean that the genetic marker itself was that gene, but that it might lie nearby. The work involved was enormous, with each marker being either analyzed alone or with a few more. With thousands of them to get through, the vast majority of which would be duds, it was a massive effort, and John spent most of his time either raising the money to pay for the work or cheering on his research team.
The technical breakthrough that made the difference was arrived at independently by scientists in Britain and America: They developed ways of fixing DNA to glass. Since I had known one of the English pioneers, Ed Southern, who was working in Oxford, I had seen the early versions using sheets of window glass about ten inches square, which he covered with a matrix of small drops of DNA solution, each containing a different synthetic segment of DNA that had been made to match exactly its equivalent in the human genome. These glass sheets were early prototypes, and by the time John Loughlin started to use the new technology for his osteoarthritis research, the whole system had been miniaturized so that half a million markers now fitted onto a silicon “DNA chip” about one inch square. The matrix of synthetic DNA markers was now far too small to see the individual spots, so the reactions were observed under a microscope.
For John and the other gene prospectors, this advance meant that instead of examining the thousands of markers in his patients and controls, either individually or in small groups, he could analyze half a million at once with a single DNA chip. No wonder he was pleased. The chips were and still are expensive, and the machinery to read them is beyond the budget of most university laboratories, so it made sense for this work to be contracted out to commercial labs that could benefit from the economies of scale. So now all John needed to do was get hold of the DNA, send it off to one of these labs, get the results back, and interpret them—and spend even more time raising the money to keep going.
While I could not fail to be impressed with the technical achievement of the DNA chips and the sheer slog it was saving John and his team, I did not immediately see how this would help unravel the tangled ancestry of our genome. It was only when John referred me to 23andMe, a Californian company that was offering chip-based DNA tests to the public, that I began to understand their potential. On its Web site were examples of what the company called “chromosome paintings.” The moment I saw them I caught the first melody from our genomic ancestors. What had been until then a formless noise, audible only to the oscilloscope of computation, suddenly resolved into woodwinds, strings, and brass. Within a week I was on my way to San Francisco.
Company headquarters are in the broad, winding avenues of Mountain View, right at the heart of Silicon Valley twenty miles south of San Francisco. As I found my way, I passed neat yet unremarkable two-story buildings set back from the road and half hidden by trees. The buildings may have been unremarkable, but the signs outside were certainly not, for here and in nearby Cupertino were the research headquarters of some of the best known global companies in electronics and computing: Google, Apple Computers, Cisco Systems, Siemens, and more. I had managed to arrange a visit at such short notice because the company’s director of research was former Stanford geneticist Joanna Mountain, whom I had met on the academic conference circuit and whose work on mitochondrial DNA I knew well. The place was buzzing, because 23andMe had recently won the 2008 Time magazine Invention of the Year award.
The central theme of the business was to use the DNA-chip technology to provide customers with information about their risks of developing a range of genetic diseases. Some diseases, like sickle-cell anemia and Tay-Sachs, have a simple one-to-one correspondence between identifiable mutations in known genes and developing the disease. However, for most diseases with an inherited component, like the osteoarthritis that John Loughlin was researching, the links to specific genes are a great deal more tenuous. One of the great research efforts of the past decade has been to identify these genes, hoping that what was true for Tay-Sachs and sickle-cell would also be true for diabetes, hypertension, Parkinson’s, and the rest. The initial optimism that drove the furiously competitive search for these genes, fueled by the prospects of patenting them and making a fortune, was soon tempered by reality. They proved to be at first elusive and then quite impossible to tie down. As the British geneticist (and wit) Steve Jones once remarked, trying to find them resembles T. S. Eliot’s description of the hunt for Macavity the Cat, the “Napoleon of Crime,” in his Old Possum’s Book of Practical Cats.
He’s the bafflement of Scotland Yard, the Flying Squad’s despair:
For when they reach the scene of crime—Macavity’s not there.2
That is not to say that genes are not involved at all in these common diseases, just that the prediction that they would be small in number and big in effect turned out to be wrong. The reality is that the genes involved are many in number and individually weak in their effect. Although the search for the “Napoleons of Crime” may have been abandoned, there are plenty of minor accomplices that have been found “loitering with intent” and taken in for questioning, and the outcome of these enquires has been to use the DNA results to adjust an individual’s risk of developing a disease.
People’s perceptions of risk are notoriously wayward, particularly when there are numbers attached, and bea
r only scant relation to anxiety levels. For example, I am much more worried about being crushed by a herd of stampeding cattle than I am of being killed in a car accident, even though the statistics show that I have it completely the wrong way around. In the United States 105 people were killed by cattle between 2003 and 2007 while 192,256 died on the roads. Not strictly comparable, I know, but you see what I mean. I could modify my personal risks downward by never going into a field full of cows, or avoiding traveling in a car. One of the presumptions of personal genetic risk analysis is that we will modify our lifestyles accordingly: If we have a higher than normal genetic risk of obesity then we will go on a diet, or if we are told we have an elevated risk of developing diabetes then we will avoid sugar. I have always thought this a very tenuous piece of reasoning. After all, millions of people smoke though they know very well that it might kill them. But I could not have put it better than a journalist, from The New Yorker, I think, who once interviewed me about the results of some tests he had done on himself through another company. After he had finished his questions, I asked him what he was going to do about this new knowledge about himself. “Eat more broccoli,” came his sardonic reply.
As you can imagine, there has been a great deal of debate about the value of these results, and even whether the tests should be offered to the public at all. A lot of this has been among professional medical geneticists who are fearful that people will discover they are at high risk of developing a grave genetic disease. There are good reasons for taking this seriously, and during my time in medical genetics, I have been impressed with the arrangements for counseling people who are contemplating a DNA test because of a family history of a genetic disorder. None illustrates the dilemma better than Huntington’s disease. This insidious and invariably fatal affliction seems deliberately designed to maximize cruelty to its unfortunate sufferers and their relatives. The symptoms of neurological and personality collapse do not show until around the age of thirty, after which there is a steady decline toward dementia and death. The pattern of its inheritance means that children who see one of their parents succumb have a 50 percent chance of inheriting the mutant gene and developing the disease themselves. Unlike Tay-Sachs and other recessive disorders, one mutant copy of the gene is enough to give the symptoms.
Finding the Huntington’s gene, in 1993, was one of the triumphs of the early years of genetic exploration and immediately offered the prospect of a genetic diagnosis before the onset of symptoms. Not that anything could be done about stopping the development of the disease, but there were circumstances when the DNA test was requested, most commonly when someone who was at risk but too young to show the symptoms was contemplating starting a family. Often this was someone who had already witnessed the suffering of a parent but did not know whether they carried the same death sentence in their DNA. There are so many factors that need to be considered, even before having a DNA test, that professional advice is essential. How will you respond to a positive result? Or even a negative one, which you would have thought would bring unrestrained relief but is often met with a deep feeling of guilt. How about identical twins? Since their DNA is exactly the same, the result of a test would apply equally to both, but what to do when one wants the test but the other does not? It is no surprise that suicide has been the response of some to finding out that they have the mutant gene, and under these circumstances it is easy to see that it would be catastrophic to offer the Huntington’s DNA test directly to members of the public without the backup of professional genetic counseling. I think considerations of this kind have made the medical genetics profession extremely wary about direct-to-consumer genetic testing for less acute disease susceptibilities, which is why on the whole, it is not in favor.
I have certainly had vigorous arguments with my colleagues about this, and I think they are wrong. First of all I think they underestimate the sophistication and common sense of customers. Second, their response is both arrogant and hypocritical in the sense that the same medical genetic community that has trumpeted the benefits of genetic research now wants to restrict public access. By all means root out the charlatans, but instead of sniffily looking the other way, help companies that have the resources and the motivation to do a difficult job well. And, by the way, I am not being paid to say this.
Although their primary objective is in the health-care aspects of modern genetic analysis, 23andMe was also well aware that the same genetic information could be used for personal ancestry research. Organizations like Oxford Ancestors and Family Tree DNA had proved that there was a market, while the appetite for personalized genetic health-risk evaluation by members of the public had never really been tested when they launched in 2007. No one knew how much people would be prepared to pay and how many would want it. But for ancestry the figures were there, and since it required no additional genetic analysis, only interpretation and presentation, it was sensible to offer ancestry testing as a sideline. And it was this sideline that brought me to Mountain View.
I was met at the door by Joanna Mountain and one of the cofounders, Linda Avey, whose background is in marketing. I had prepared a short presentation, mainly about the narrative qualities of DNA, as I recall, after which Joanna did the same, explaining how the company was adapting its DNA-chip system for ancestry applications. The others in the audience were mainly young, mostly scientists or software engineers. After a short tour and some individual meetings I left to rejoin the hell that is Highway 101 going north to San Francisco. Except this time I hardly noticed. I was very pleased with how things had gone. Although this was only an intial contact after all, I came away with a very positive impression of the company and the people and, more important, an offer of help with my research for DNA USA.
What had intrigued me from the start and had seemed to offer a way into the complexity of our genomic ancestry was the way in which the ancestral origins of human chromosomes were portrayed. Each of our twenty-two pairs of autosomes—that is to say all of our chromosomes, except the X and Y—were laid out in horizontal rank and in numerical order from the largest, number 1, at the top to the smallest, number 22, at the bottom. Each chromosome was sliced lengthwise along the middle so that the top and bottom slices represented the two copies of each chromosome that we possess. In examples of people with a mixed ancestry, different colors picked out the segments of their DNA that had come from one of three continental origins. Dark blue for European, green for African, and orange for Asian—which in the United States is a proxy for Native American.
This was not the first system to estimate the continental components in an individual’s genome. An earlier method had been developed that gave a quantitative estimate of African, European, and Asian DNA, but it did not break this down into chromosome segments. Rick Kittles, the cofounder of African Ancestry, had used a system, called AIMs for “ancestry informative markers,” with some interesting results that we will look at later, but something about the numerical brutality of AIMs made me wary of its use in individuals. Chromosome painting, on the other hand, seemed to overcome my misgivings and come much closer to the real situation for individuals with ancestors from different continents, and the visual representation made it much harder to misinterpret.
So how do you go about painting someone’s chromosomes? The underlying science depends, as always in genetics, on the variations between one individual’s DNA and the next. Without these there would be no genetics. One of the triumphs of the Human Genome Project, aside from reading the entire human DNA sequence, has been to discover millions of tiny differences between human genomes, known by their acronym SNPs, which we have already encountered in the Y chromosome. The initials stand for “single nucleotide polymorphism,” which means a difference only in the DNA sequence at a particular location on a particular chromosome. For example, where the sequence might be GGATTA on one chromosome and GGATCA on another, this is a SNP. Millions of SNPs have been discovered throughout the human genome, which once found can be identified by the DNA se
quences on either side. So the SNP we introduced as GGATTA/GGATCA is flanked by unchanging DNA sequences that are known from the human genome sequence. You need only a sequence of around twenty bases on each side to uniquely identify any SNP. These short DNA segments are easy to synthesize, and easy to immobilize on a chip. Once on the chip they are able to detect which of the two sequences is present at the SNP in any DNA they are asked to test—or “interrogate” in the lingo. After some clever chemistry the spot on the chip where the synthetic SNP sequences are attached glows a fluorescent red for one version and green for the other, with these tiny signals being picked up by a powerful automated microscope. As there are half a million SNPs on a typical chip, and the microscope can scan all of them within a few minutes, very soon you know the sequence at all half million SNPs in the DNA being interrogated.
However, these chips are analyzing DNA from individuals who have two copies of each chromosome. This means that there are not two but three possible results for each SNP. If, in our example, GGATTA glows red and GGACTA glows green, when both chromosomes have the GGATTA version the spot will glow red. On the other hand, when both chromosomes have the alternative sequence GGACTA at the SNP, the spot will be green. But there will be times when both versions are present and one of the chromosomes has the sequence GGATTA while the other has GGACTA at the SNP. Under these circumstances the spot on the chip glows both red and green. Fluorescence filters on the microscope can deal with this and record both versions of the SNP. Even though the chip has analyzed half a million bases, this is only a fraction of the total of three thousand million. However, these bases have been chosen as the ones that are known to vary between chromosomes. We also know the precise chromosomal position of all half million of them.
At each generation our chromosomes shuffle their DNA sequences. Most of the time the chromosomes we received from our mothers and the ones we inherited from our fathers don’t have much to do with each other. They lead physically separate lives in the cell nucleus and carry on barking their instructions to our cells quite independently from one another. In most of our body cells they live apart throughout our lives, but in our germ-line cells there is a final embrace. Just before our chromosomes become packaged into eggs or sperm, the pairs line up with each other and swap DNA. Then they move apart and go their separate ways into the germ cells. There is a very sound evolutionary reason for this tender parting, as it creates an enormous amount of genetic diversity in the next generation that protects the offspring from parasites and pathogens, again something I explored in Adam’s Curse.
