ÐÓ°ÉÔ­´´

MANY ORGANISMS that cause disease are difficult to detect. Standard tests to detect Listeria, for instance, a bacterium that causes food poisoning, take between five and 12 days. They work by distinguishing it from other bacteria, by attempting to grow it under conditions in which it, but not others, will flourish. Microbiologists culture the suspect sample for several days, and then spread any healthy bacteria growing on it onto several media containing specific combinations of nutrients and antibiotics. Its growth under some but not other conditions, coupled with examination under the microscope, uniquely identifies the species. Laboratory robots can reduce the time to two or three days, but this is still too slow. It may be good enough for testing cheeses that keep, but it is far too late for deciding how to treat a patient. A hospital test should produce the result at the same time as the test.

To make matters worse, viruses and many organisms that are sexually transmitted do not show up under standard microbiological tests. Copulating mammals transfer living cells to each other, providing a perfect channel for a pathogen to exploit. The pathogen may be incapable of life outside a human being, making it hard to culture and identify them in the laboratory. Current tests rely on antibodies that bind specifically to the pathogen; this enables technicians to test clinical samples directly without the need for culturing. These tests are rarely as sensitive as one would like, however; a millilitre of blood taken from someone suffering from AIDS may contain only one virus particle. The newly developed DNA probes promise to provide selective and sensitive tests for these obstinate organisms. Their development has L depended not on new biological methods but on applying established biochemical tests with hitherto L unimagined sensitivity and ease.

The challenge has been to design a test that overcomes a fundamental stumbling block: the fact that most disease-causing organisms are biochemically very similar to their target. Any test must distinguish between two similar, or sometimes virtually identical, types of cells. Diagnostic tests based on DNA have the edge, because any pathogen has its own unique genetic material. So the genetic material – the DNA, or in the case of some viruses, the RNA – makes an ideal target for a diagnostic test.

To be useful in a hospital, a test must be quick, easy and safe to perform, be sensitive enough to detect fewer organisms than cause a disease, and must give the right answer. The last point might seem obvious, but the difficulty of avoiding ‘false positives’ is a major reason that tests based on antibodies are not more widely used. A DNA test, based on detecting the binding of a DNA probe to its match in its pathogen, can spot a very small number of DNA molecules, and can be extremely accurate even when very few virus particles are present . This approach, known as DNA hybridisation, is more sensitive than other analytical methods, because biologists can engineer the probe to be as specific or as general as they wish. At one end of the spectrum, a probe to a piece of RNA crucial to the synthesis of proteins and shared by all bacteria, can detect any bacterium. At the other end, a probe to the gene responsible for a protein on the surface of an influenza virus detects only the viruses descended from a specific influenza pandemic.

This ability to adjust the specificity of a test is unique to DNA probes, and is one of their strong points. But until a few years ago, DNA-based tests were far from safe and easy to perform. They took days of a graduate student’s time to do, and used radioactive labels and X-ray film to detect the result. The advances that put commercial tests on the market have tackled both these problems.

ÐÓ°ÉÔ­´´s can accelerate the hybridisation process by increasing the concentration of some reagents or by carrying out the hybridisation reaction in a thick soup of a polymer such as polyethylene glycol. The molecules of the polymer take up nearly all the space in the solution, pushing the DNA together. These ways of speeding up the test have been known since the late 1970s, and researchers can now test samples of blood serum for viruses in 30 minutes. The test also works on ‘crude’ clinical samples – blood, urine or swabs – as the chemical reaction between the two DNA strands is not disrupted by low concentrations of other organic molecules. This gives DNA-based tests a second advantage over ones based on enzymes or antibodies.

Another, more recent, advance has been the replacement of the clumsy laboratory tests by a ‘sandwich assay’ format. Here, as elsewhere, molecular biologists took a leaf out of the immunologists’ instruction manual. Sandwich assays use two probes to hybridise to adjacent sections of the target DNA and attach it to a solid support . This approach is standard in many immunological tests, but was slow to catch on in genetics. The format is well suited to DNA probes, however, with one probe acting as a capture probe and another acting as a labelling probe. Because both probes stick to the target DNA by hybridisation, they can bind under the same conditions and at the same time. If the capture probe is fastened to a magnetic bead, a magnet can pull the hybridised complex away from the unhybridised probes, making it easy to measure the amount of complex present. So a complicated series of manipulations is reduced to one reaction and a wave of a magnet.

Achieving a sensitive test that does not use radioactive labels was more difficult. You need something attached to the hybridised probe, a ‘reporter group’, that you can detect easily in very small amounts. If you want to detect 10 target DNA molecules, the method must detect 10 labelled probe molecules. The detector also has to be cheap and safe, and radioactive labels clearly fulfil neither criterion. So far, researchers have come up with five ways to achieve a sensitivity equivalent to a radioactive label without using radioactivity.

The first is to start off with more molecules. The polymerase chain reaction, or PCR, uses enzymes to multiply the number of starting molecules by a billion, so making them far easier to detect . The drawback is that it uses enzymes, whose functions may be blocked by the many other chemicals in crude clinical samples. So the first step is to purify the DNA, a complex and time-consuming procedure. But this stumbling block may soon be overcome, and the American company Biotechnica Diagnostica claims to have sidestepped it already.

The second approach is to tie more than one reporter group onto the labelled probe. You can do this by making the probe very long, as the company Enzo in New York did with its Bioprobe; or you can make the probe branched, the solution pursued by Chiron and Imclone in different ways. Either strategy makes room for several labelling groups on each probe.

Molecular biologists can then adjust the label. A fluorescent molecule is most popular, but it is difficult to obtain sufficient sensitivity, as many other molecules in biological samples are fluorescent, creating a background ‘noise’ that can swamp a weak signal. Researchers have tried hooking an enzyme onto a probe: a thorough washing needed as part of the hybridisation reaction removes impurities in clinical samples that would otherwise inhibit the enzyme. This enzyme generates a large number of coloured molecules which are then detected. Enzo Diagnostics and Gen-probe use variations on this system. If the vitamin biotin is joined to the probe, the protein avidin will bind to it extremely tightly. If you then add an enzyme with more biotin chemically linked to it, this also binds to the avidin, making a complex of many enzyme molecules clustering around each probe molecule. Orion, the Scandinavian biotechnology company, has commercialised this approach. Variations on the theme, using luminescent reactions or detection systems involving a complex of enzymes, are also possible.

These techniques push the level of detection down to 10 000 target molecules, which is good enough for many applications. Such labels could readily pick out infections with rotavirus, which cause severe diarrhoea, for example, as each sample contains billions of virus particles. Salmonella in food, however, is present at levels on the borderline of detectability, and traces in chicken feed would probably stay hidden. The existing DNA-based test for Salmonella enteriditis requires that the suspect sample be incubated in a growth medium first to multiply the number of target bacteria. This step takes much longer than the hybridisation phase. But in many cases, such as infections of hepatitis and HIV, we need to detect the small numbers of organisms present before a major infection gets under way and irreversible damage is done. So label technology itself it not yet ideal.

The fourth approach is to automate the test, which gives two advantages. It makes sure that each test is performed under ideal conditions, producing the maximum sensitivity. While research laboratories attain such an ideal, hard-pressed hospital technicians cannot match this performance. Moreover, very complicated assays are not feasible in clinical laboratories. Automation can overcome such problems. A test ‘hidden’ inside a machine can become a ‘one-step’, 5-minute test for the operator. Gene-Trak in Massachusetts in the US is developing such a device for clinical laboratories, as is PA Technology in Britain.

The final approach is the most radical. A few pioneers have called in the physicists. The techniques of physical chemistry can detect a handful of molecules using spectroscopy, but they often rely on lasers as intense light sources, and on complex procedures and sophisticated electronics to analyse the result. This approach has been taken out of the research laboratory and into prototypes of clinical tests, however, by the advent of sophisticated robots in the manufacturing industry, cheap photomultipliers from military and space applications, $10-lasers from CD players, and the ubiquitous solid-state chip. These put such techniques as raman spectroscopy and time-resolved fluorimetry at the fingertips of biologists with a good engineering workshop behind them. Such novel detection systems will probably overtake the clever but complex biological detection methods in the next decade, producing a test as easy to use, if not as fast, as a pH meter.

In the late 1970s, researchers suspected that they could use gene probes to diagnose infectious disease, but only since 1983 have they been able to carry out the large-scale research needed to realise this potential. Most work is now done in the US. Various companies have collectively put about $100 million into R&D to create DNA probe tests. The companies are aiming at a market estimated to be worth at least $500 million by 1992, by selling the tests to hospitals that need to diagnose ‘difficult’ infectious diseases. For each of these biotechnology companies there is a pharmaceuticals giant also conducting research, and several smaller biotechnology companies with a clever idea and a hunger for venture capital.

Curiously, the key elements of the test – the DNA probes – are not a limiting factor. Academic researchers are the major source of probe sequences; sometimes, the scientists create them for use in diagnosis but more often they are part of research into the mechanisms of infection, into epidemiology or into the fundamental molecular biology of viruses and bacteria. The results are freely available in technical journals or computer databases such as that run by the European Molecular Biology Laboratory at Heidelberg in West Germany. The more important the disease, the more eager academics are to clone and sequence the genes of its causative agent, and so the more data about DNA sequences become available. The best example of this is HIV, the virus responsible for AIDS. It went from being a hypothetical agent causing a ‘fringe’ disease in 1980 to the basis of five DNA- based test kits by 1988.

Some companies turn to the test itself to secure the coveted ‘intellectual property rights’, the patent protection for their test that ensures that their R&D effort will not be swamped by cheap copies. Collectively known as the test ‘format’ – the chemical conditions, physical layout, detection and readout systems – this is the area in which big advances have been made in the 1980s. The technology to generate probes was in place long before the Nobel prize in 1982 went to Fred Sanger, Allan Maxam and Walter Gilbert for methods of sequencing DNA. Since then, the work has been edging into the doctor’s surgery.

Compared to many applications of biotechnology, DNA probe tests are easy to produce. A single run on an automated DNA synthesiser makes milligrams of DNA, enough for millions of tests. The fragments of DNA, and the other reagents, cannot replicate even inside a bacterial cell, and so there are none of the regulatory fears that accompany the release of genetically engineered organisms into general use. And there is a wealth of experience in DNA hybridisation to show that the test will perform properly.

By 1992, many pathogens will be detected in hospitals by DNA-based tests. These will determine not only the organisms in a sample, but also whether they are pathogenic strains and whether they are likely to be resistant to antibiotics, as all these characteristics are the result of particular genes. For bacteria such as E. coli this is an important point: while some strains cause diarrhoea in babies, others are harmless colonists of our guts. Of 40 known strains of adenovirus, only three cause disease. Differences between strains are not obvious, because the genes that enable them to adhere to human cells or make toxins are frequently producing their protein products only on infection. In the laboratory, pathogenic and harmless strains can look almost identical. For example, strains of Salmonella responsible for typhus produce the proteins on their surface that enable them to invade human cells only when they encounter those cells.

The tests could also be edging into clinics for sexually transmitted diseases and larger GP group surgeries, where computer-based expert systems analyse the mass of information they produce. And here is the rub. DNA probe tests can generate an enormous amount of information, and some clinical microbiologists worry that these tests will mislead people who do not realise their limitations: high sensitivity could be a drawback. Little is known about the low levels of pathogenic bacteria that may normally live in our bodies or their immediate environment. Edwina Currie’s observation that most egg-producing flocks of chickens are ‘infected’ with Salmonella means just that – Salmonella enteriditis can be found in the housing for those flocks. But what does this mean for human health? Estimates vary by a thousand-fold. A DNA-based test, more rapid and sensitive than existing culture methods, could raise the estimates of Salmonella infection even further, while getting no nearer to its impact on people. In fact, existing methods test ventilator fluff and droppings, not chickens. The British Working Party on Salmonella in Eggs has sensibly suggested blood tests for laying hens, which would get round this point at least. Similar tests on people may be needed, as no one knows how many minor or benign infections of Salmonella occur for every acute hospital case of food poisoning, and what the relationship between the two may be. Legionella is also said to be widespread: a DNA-based test could confirm this, detecting it in a range of sites outside stagnant water and air-conditioning plants. But we need a clearer understanding of the steps between an apparently benign bacterium in the soil and a fatal case of Legionnaire’s disease. We are awash in a sea of microorganisms, but rarely fall ill. Making the sea more visible may help to reduce the rate of drowning in it, but will not eliminate it. So DNA probe tests will be of enormous help in diagnosing infectious disease, especially viral and sexually transmitted diseases, and in guiding their treatment. But doctors will have to sue them with care, realising their limitations. The tests could turn out to be misleading in environmental monitoring, until epidemiology has caught up with their enormous potential.

Dr William Bains is a consultant biotechnologist with the PA Consulting Group.

* * *

HOW DIAGNOSTIC DNA MEETS ITS MATCH

DNA probes set DNA to detect other DNA, and the basis of doing that is the hybridisation reaction. The two strands of the DNA double helix are held together by weak interactions, called hydrogen bonds. If the two strands of the helix are separated, they will spontaneously come back together. DNA strands from different sources can form double helices in this way if the bases that make up the two strands fit into each other.

To form a stable helix, the four bases that are repeated in various patterns in the DNA must fit together correctly in space, so that the distance between the outer edges of the helix is constant and a regular helix can form. This means that, of the 10 possible pairwise combinations of bases with each other, only A with T and G with C are allowed. So if one strand has the base sequence ACGTCCG, the other must have the sequence TGCAGGC opposite to it to form a double helix.

Whether a helix forms depends on what percentage of the bases are ‘matched’ in this way – the smaller the fraction of matched bases, the less likely the helix is to be stable, because the fewer hydrogen bonds can be formed. Similarly, reducing the length of the single strands reduces the stability of the helix, as does altering a variety of reaction conditions – increasing the temperature, for example, ‘melts’ the helix apart. So a synthetic DNA can be tailored such that, under particular L hybridisation conditions, it will hybridise with only an exactly matched DNA, or with one that is within 5 per cent of exact matching, or within 10 per cent. Combined with knowledge of the likely variation in the DNA sequences of different genes in viruses and bacteria, this allows scientists to construct a probe that will hybridise to the DNA from a specific group of bacteria – one strain, one species, one genus – but not to any other.

* * *

MAKING A SANDWICH OUT OF DNA

The technique known as sandwich hybridisation uses two probes that hybridise to adjacent sections of the target DNA. One, the ‘capture probe’, is fastened to a solid support so that when the target hybridises to it, it is also linked to the support. The other, the ‘labelled probe’, is linked to a reporter group – a fluorescent molecule, an enzyme or a radioactive atom. This probe will not hybridise to the capture probe, and so in the absence of target molecules the labelled probe will not stick to the support. If the target is present, how ever, it links capture probe and labelled probe, holding the labelled probe onto the solid support. If this support is a tube, washing the tube out will remove all the unbound labelled probe, so the amount left in it is a measure of the amount of target in the original sample. If the support consists of tiny plastic particles, microbeads, these can be separated by centrifugation or magnetic attraction. from the unbound labelled probe. The amount of the labelled probe remaining stuck to them tells how much target DNA there is.

* * *

SETTING OFF A CHAIN REACTION AMONG THE GENES

DEVELOPED by Cetus Corporation, one of the earliest genetic engineering companies, based in California, the polymerase chain reaction, or PCR, uses enzymes to amplify a rare DNA molecule into billions of copies. The originally double-helical DNA is ‘melted’ by heating, separating the two strands. Two short DNA strands, the primers, are hybridised to it on either side of the target DNA. An bacterial enzyme, a DNA polymerase that synthesises a new DNA strand, is then added. This makes new DNA starting from the primer molecules. These enzymes can only extend existing single strands of DNA, and so make new DNA only over the target region. This produces two double helices where only one existed before. The double helices are then melted, more primer hybridised, more DNA synthesised and so on.

In practice, researchers use a DNA polymerase, called Taq polymerase, from a heat-tolerant bacterium, so the enzyme can stand being heated repeatedly to 90 Degree C for short periods. Once the enzyme, primer and other ingredients are mixed together, the researcher needs only to cycle the reaction through the temperature changes to do the PCR. A typical reaction will take 70 minutes, including 40 cycles and produce 10 million DNA molecules for each starting target. Cetus and Perkin Elmer have produced a machine to perform this cyclic reaction automatically, as have three academic groups at much lower cost. The only drawback is that, like many DNA-synthesising enzymes, Taq polymerase is strongly inhibited by traces of other macromolecules such as those found in bacterial cell walls.