"The terminology that we developed for DNA typing using multi-locus probes has been hijacked and in a very misleading way," says sir Alec Jefferys of the University of Leicester.

Since the discovery of the structure of DNA in 1953, knowledge of the composition and organization of the genetic material has accumulated at an astonishing pace. By the early 1980s it had become clear that most human DNA shows very little variation from one person to another. The small percentage that does vary presents enormous potential for fruitful study.

   Sir Alec Jeffreys's involvement with mammalian molecular genetics began in 1975, when, as a postdoc, he moved from Oxford University to the University of Amsterdam to work with Dick Flavell. There, the two and their colleagues tried to clone a mammalian single-copy gene. They failed, but in the process managed to develop the Southern blot hybridization technique to the point where they could directly detect single-copy genes–and, in so doing, discovered one of the first examples of introns.

   When Jeffreys moved to the University of Leicester in 1977, he chose to change direction completely and study DNA variation and the evolution of gene families. As a result of this work, his laboratory produced one of the first descriptions of RFLPs–restriction fragment length polymorphisms–a common form of variation in human DNA. The aim of the work was to develop a new breed of markers using DNA to track the position of genes. To develop good markers, the researchers needed to find highly variable regions of DNA.

   In 1980, another team made one of the major breakthroughs in the study of DNA polymorphism, with their fortuitous discovery of the first "hypervariable" region of human DNA. These regions were found to consist of short tandem sequences repeated over and over again.

   In 1983, Jeffreys found that these repeat sequences, dubbed "minisatellites," contain certain "core" sequences. This opened the way for the development of probes, containing the core sequences, for detecting many other such regions of variable DNA. One Monday morning in September 1984, Jeffreys and colleague Vicky Wilson successfully tested the effectiveness of such a probe. "The implications for individual identification and kinship analysis were obvious.... It was clear that these hypervariable DNA patterns offered the promise of a truly individual-specific identification system," Jeffreys wrote later (see A.J. Jeffreys, Am. J. Hum. Genet., 53[1]:1-5, 1993). They had stumbled on DNA fingerprinting, and Jeffreys's life was changed.

   Jeffreys, 45, gained a first-class degree in biochemistry from Oxford University in 1972, and his Ph.D., also from Oxford, in 1975. After working in Amsterdam with Flavell between 1975 and 1977, Jeffreys moved to the University of Leicester as a lecturer in genetics and became a full professor in 1987. He was elected a Fellow of the Royal Society (FRS) in 1986.

Science Watch's European correspondent Amir Amirani
spoke with Jeffreys at his laboratory in Leicester.

Your most-cited paper, "Hypervariable minisatellite regions in human DNA," appeared in Nature in 1985. Is the paper highly cited because it's subsequently been used in fingerprinting, or because of the light that the paper shed on the structure of variable DNA?

   Jeffreys: The citations, I think, reflect the fact that at the time this was a novel, very powerful generalized technology that could be applied to a wide range of problems in human and nonhuman genetics. The paper described for the first time a general method for getting at large numbers of highly variable regions of human DNA. Also, almost as an accidental by-product, it suggested approaches for not only developing genetic markers for medical genetic research, but for opening up the whole field of forensic DNA typing. And, from the work in that first paper, we could see immediately the potential applications in individual identification and in establishing family relationships–for example in paternity and immigration disputes.
   Although it wasn't mentioned in the paper for patenting reasons, we also saw the potential for exactly the same technology being applied to nonhuman species as well. This opened up all sorts of interesting possibilities in animal breeding, conservation biology, ecological genetics, and the like.

Has the potential for the animal work been fulfilled?

   Jeffreys: Very much so. That original DNA fingerprinting system, for example, has been used in a fair number of zoos to try and establish family relationships within captive colonies of endangered species of animals and birds, in particular to identify cases of close relationship–those individuals that you do not want to interbreed. The aim, in other words, is to minimize inbreeding and maintain genetic diversity.

Is there a biological function for mini- and microsatellites?

   Jeffreys: That is a very, very tough question. If we look at minisatellites, by and large, there seems to be no obvious biological function. There are a few cases in the human genome, and a fair number of cases outside the human genome, of minisatellites that actually form part of genes. So there are tandem repeated DNA sequences that code for tandem repeated protein sequences. But those are the exception, not the rule. The majority of the minisatellite loci we look at have no obvious function. However, one area that we are very actively examining at the moment is the whole question of how variation arises at these tandem repeat DNA sequences. And that means exploring the mutation processes that go on in sperm and eggs, creating new versions.
   What's come out of that is actually a very surprising result in which the mutation process, rather than just reflecting the instability of tandem repeat DNA, seems to be actively controlled by elements external to the tandem repeats. So it looks as though the tandem repeats themselves are not so unstable, but rather the instability is being directed from a locally acting regulator. We also know that the mutation process is astonishingly complex and operates by a process that is wholly unexpected for minisatellites. We call this process "gene conversion," and it involves chunks of DNA being shifted from one allele to another during the mutation process.
   We also suspect that, in males, the majority of sperm mutations are specific to the male germline and may be meiotic in origin. This suggests a type of recombinational process, controlled by some elements near the minisatellite, and it looks as if it's meiotic as well. And that really does start raising questions–such as, maybe this mutation process isn't just some sort of accidental artifact of having tandem repeat DNA, but rather reflects some basic biological process going on in the DNA. One of our main jobs now is to explore this in a lot more detail.

And is that of purely theoretical interest, or are there going to be practical implications as well?

   Jeffreys: This is basic biology. As to whether there will be practical implications, I don't know. However, in the course of our investigations, we've developed various new strategies for detecting new mutations in human DNA, and this does, in principle, offer practical applications. By mutations, I don't mean, for example, a cystic fibrosis mutation, which is actually not a mutation at all but a variant that's been around for thousands of years. I'm talking about new mutations–actually catching DNA at the point where it has altered its structure. If we can develop methods for measuring mutation rate in an individual undergoing this process–and this is one of my main interests–then we can start asking basic questions about environmental agents, such as ionizing radiation, which might impact upon the mutation rate.

Fingerprinting has been subject to a lot of controversy, something you have alluded to in some of your papers. Do you personally have any reservations about its reliability?

   Jeffreys: Before I answer that, we must clear up a point on semantics, and this is not trivial. The original DNA fingerprinting system we developed, which for technical reasons is not that useful in forensic identification, produces patterns that are idiotypes–they are, for all intents and purposes, completely unique to an individual, except for identical twins. There's no serious dispute about that in my view. Unfortunately, the second generation of DNA typing systems–which is DNA profiling using single-locus probes–do not produce individual-specific patterns test by test. Even with a typical battery of five different tests, they produce patterns where unrelated people are most unlikely, in fact extremely unlikely, to share the same pattern. However, when you come to close relatives, brothers and sisters, there is a real chance, in fact about 1 in 4 to the power of 5 chance, of a brother and sister match, which is 1 in 1,000.
   So, for every 1,000 sibling pairs, over five probes, you find a complete match. So they are not DNA fingerprints, not unique to an individual. However, their variability among unrelated people is pretty spectacular over five tests. Unfortunately–and particularly in the United States–the term "DNA fingerprinting," which we specifically apply to the original multi-locus system in which we look at scores of markers, has been corrupted to be used in almost any DNA typing system. That has created a problem in court, because DNA profiling does not produce DNA fingerprints, but if you call them DNA fingerprints, then the defense lawyer can stand up in court and say, "This is misleading," and that's quite right.
   So this is a semantic problem, but a serious one. Basically, the terminology that we developed for DNA typing using multi-locus probes has been hijacked, and in a misleading way. Now, if we get rid of that semantic part, we can ask how valid is the huge amount of debate that's gone on about the reliability of DNA profiling? In the early days, in particular, there was real cause for concern. Some of the laboratories doing this work were carrying out real forensic analysis with technology that had been very poorly validated and hadn't been standardized.
   I think that this issue has been largely addressed now, through quality controls, the adoption of standard operating conditions, blind proficiency trials, and so on. For DNA profiling, the real source of debate now relates to how one estimates the rarity of a set of DNA profiles out in the population, and how one presents that evidence in court. If you say that a DNA profile of a forensic sample matches a given suspect and is very rare in the population, then that, depending on the context, can be pretty damning evidence.

Let's turn to your current research interests.

   Jeffreys: My current interests are in exploring the basics of mutation of human minisatellites. We now know they are mutating by processes that are totally unexpected. These processes are probably of biological significance and may shed light on another fascinating area of human genetics: the whole area of triplet repeat instability diseases. These are microsatellites that go horribly unstable and cause neurological disease, such as Huntington's chorea, myotonic dystrophy, fragile x syndrome, and so on. These are basically microsatellites, which suddenly become highly unstable, increase their repeat number, become very long, and wreck nearby genes.
   And again, for technical reasons, it's not easy to explore the details of the mutation process going on there, but we can explore in great detail the mutation process going on in minisatellites. We can use a whole battery of techniques that we've developed, which explore these bizarre mutation processes. It's not impossible, though far from guaranteed, that what we discover in minisatellites may actually be applicable to these inherited diseases.
   One sort of science-fiction scenario would be this: let's suppose that what happens in minisatellites also applies to these unstable microsatellites. In other words, instability is conferred upon the array by flanking DNA, which, we suspect, is activating an allele for mutation. It's basically switching an allele on, perhaps by introducing some kind of DNA damage, such as a double strand break into the DNA.
   Now, if that is true for these neurological diseases, and these diseases manifest because of this instability, one could conceivably think of some therapy aimed at blocking that mutation initiation. That's wild fantasy, but who knows? After all, gene therapy was fairly wild fantasy 20 years ago.
   Another area in which we're very much involved is developing new approaches to DNA typing. We've been heavily involved over the last couple of years in an approach called digital DNA typing, where you get a digital readout from the DNA rather than the usual sort of band length measurements in DNA profiling. And that has first of all revealed minisatellites as by far the most variable loci in the human genome. The typical minisatellite has, for example, 100 million different alleles worldwide, and that is astonishingly variable. And that in turn may give us some rather interesting markers for studying recent events in human evolution–by looking at these allele structure and how they've changed over time, how they differ between recently split populations, and so on.

The field of DNA fingerprinting is relatively new. How do you expect this technique to develop, and how do you expect DNA structure studies overall to progress?

   Jeffreys: The field of DNA fingerprinting has diversified to the point of incoherence. It's no longer a single unified field. For example, back in 1987-88, when we had our first congress on DNA fingerprinting, the thing that welded it together was that everybody was playing around with minisatellites, DNA fingerprinting, and DNA profiling.
   What's happened since then, of course, is the advent of DNA amplification by polymerase chain reaction, or PCR. This means, first of all, that there is little doubt that in forensic DNA typing within the next few years all the classic systems of DNA fingerprinting and DNA profiling will be totally replaced by PCR-driven systems. Such systems have their powers and their weaknesses as well–contamination and the like. But the advantage of PCR is that it offers great sensitivity, potential for automation, lower costs, and information that is much less ambiguous in terms of a DNA profiling result.
   Now, what the ultimate DNA forensic typing system will be, I don't know. But to suppose that we've actually arrived there now is naive in the extreme, bearing in mind that information about PCR, or user-friendly PCR, was published only seven years ago. To pretend that we've gone from that to the ultimate DNA typing system is nonsense. There'll be other ones coming along, and that actually creates a major problem for the forensic scientist who is interested in databasing, because once you go in for very large-scale databasing of many thousands of people–you are trapped in that technology. You cannot change that technology because you've got to retype everybody in the database if you do. So the drive towards databasing, I think, is in fundamental conflict with the still rapidly evolving field of forensic DNA typing–the technology itself.
   So I see all kinds of developments on the forensic front. People may actually come up with what everyone is talking about: DNA chips, oligonucleotide chips that will be used to interrogate PCR reactions. At the moment, these are not chips at all in the electronic sense. If one could, however, create a chip in which an oligonucleotide could detect and transduce the detection of a product (such as a PCR product) into an electronic signal, that would open up not just forensic typing, but DNA typing, medical diagnostics, and just about everything else one can think of.