杏吧原创

FOR a handful of people, it has made all the difference. Around 20 boys, for instance, are now leading nomral lives. They no lon 鈥� Around 20 boys, for instance, are now leading normal lives 鈥� ger have to live sealed in a bubble because their immune systems cannot fight off diseases. And it鈥檚 all down to gene therapy.

Confused? Imagine if you had to rectify every spelling mistake in an article by adding a copy of the full corrected sentence at a random point. Well, that鈥檚 pretty much how gene therapy and genetic engineering work right now. Instead of editing individual letters of the DNA code, you have to add a whole new piece of DNA, and there鈥檚 no way to control where it ends up.

Just as a stray sentence can cause confusion in a magazine article, so a stray chunk of DNA in a cell鈥檚 genome can create havoc if it lands in the wrong place. In the worst-case scenario the result is cancer, as may have happened to three of 11 boys in a gene therapy trial that started in 1999 at the Necker Hospital for Sick Children in Paris, France.

What is needed, say genetic engineers, is a method for altering DNA in much the same way that a word processor can correct a mistake such as 鈥渘omral鈥�. Then they could fix mutations while leaving the rest of the genome untouched, replace one version of a gene with another or insert new sequences at a precise spot.

Researchers have been dreaming of this for 30 years, and now the necessary tools are starting to appear. 鈥淲e are moving into a new era, into the second generation of transgenics,鈥� says Paul Eggleston of Keele University in the UK.

The consequences will be profound. For starters, it will become easier and cheaper to create genetically modified plants and animals. In medicine, the precision approach promises to deliver everything from powerful new gene therapies to new treatments for diseases such as HIV. More controversially, this is the technology that could open the door to genetically engineered human beings.

It is just 30 years since researchers discovered how to 鈥渃ut and paste鈥� short pieces of DNA from one strand to another in a test tube. Methods have advanced in leaps and bounds since then. In 2002, biologists in the US assembled the 7500-base-pair polio virus genome from scratch, and there are plans to do the same for a bacterial genome, hundreds of thousands of base pairs long.

Altering the DNA inside cells, however, is harder. For starters, the enzymes used to splice DNA in a test tube are far too large to enter cells. Despite this, biologists now have tricks that allow them to make pretty much any change they please to bacteria such as E. coli. The more complex cells of animals and plants are a different matter. Here, DNA is packaged up with proteins and tucked away behind the double membrane of the cell nucleus. There is also a massive amount of it: 6 billion letters in 46 separate pieces (the chromosomes) in the case of humans. All this means that making precise changes is a huge challenge.

As a result, most genetic engineering still relies on the discovery made in the 1970s that if you can just get a piece of DNA into plant or animal cells, it will occasionally be incorporated into the genome. Since then engineers have developed dozens of ways to deliver DNA to cells, from bombarding them with DNA-coated projectiles to hijacking the DNA delivery systems of viruses. Yet in almost all cases the final, crucial step 鈥� incorporating the DNA into the cell鈥檚 genome 鈥� is down to chance. There is no way to control where the new DNA will end up.

Unsurprisingly, this approach creates endless problems. For starters, random integration is a very rare event 鈥� one cell in a few thousand, if you鈥檙e lucky. Integration itself brings all kinds of difficulties. Sometimes the extra DNA integrates slap bang in the middle of another gene, rendering it useless. In the worst case this can cause cancer.

Even if the DNA lands in a safe location, there is no guarantee it will behave in the way you want it to. Genomes contain numerous 鈥渟witches鈥�, or regulatory elements, that control the activity of nearby genes, turning them on or off at specific times or only in specific tissues. The result is that the same piece of DNA can behave in 10 different ways in 10 different locations, depending on which regulatory elements it ends up next to. 鈥淭he same gene can have very different activity,鈥� says Eggleston. That鈥檚 bad news, whether you are trying create GM crops or cure genetic diseases. Indeed, some researchers think the three boys in the French gene therapy trial got cancer not because the added gene landed in the wrong place, but because it was too active.

For all these reasons, genetic engineers have long sought ways to control precisely where DNA integrates into an organism鈥檚 genome. Perfecting 鈥済ene targeting鈥� would make gene therapy far safer and genetic engineering vastly more powerful.

鈥淲e are now moving into a new era, into the second generation of transgenics鈥�

False dawn

Back in the 1980s, molecular biologists thought they had cracked it. They found that if they took mouse embryonic stem cells (ESCs) and added a new DNA sequence flanked by mouse DNA, overeager repair enzymes would sometimes cut the matching sequence out of the mouse genome and splice in the new DNA instead, a phenomenon called homologous recombination. Exactly why this happens remains a mystery, and it only works in around 1 in 100,000 cells. But it works.

Thanks to homologous recombination, genetic engineers now have exquisite control over the mouse genome. The procedure is so routine that the European Union has set up a project to study 20,000 mouse genes 鈥� two-thirds of the total 鈥� by creating 20,000 modified strains in just three years. No wonder mice are so popular with geneticists.

However, when biologists tried to make homologous recombination work in other animals, they got a nasty shock. For reasons that nobody fully understands, the phenomenon is exceedingly rare in just about every cell type other than mouse ESCs. That makes it impractical for genetic engineering and a complete non-starter for gene therapy. So while geneticists perform ever more spectacular tricks with mice, the 鈥渃huck it in and hope鈥� approach has remained the only way to modify most animals and plants.

Now this is changing. Genetic engineers are developing several methods for making precise changes to the DNA of living cells. These methods are also far more efficient, transforming up to 1 in 5 cells.

One relies on the tricks evolved by viruses that infect bacteria. Many of these can precision-engineer their hosts鈥� DNA, integrating their genes at specific target sites with the help of enzymes called recombinases.

Recombinases work by recognising two distinct sequences: one on the viral DNA to be spliced into the host, and a second at the target site on the host鈥檚 genome (see Diagram). The enzyme cuts the DNA at both sites and stitches the whole lot together. This ability gives recombinases enormous potential for genetic engineering. Simply add the first target sequence to any piece of circular DNA and the recombinase will integrate it into any other piece of DNA containing the second target.

The catch is that because recombinases evolved to target bacteria, most plant and animal genomes do not contain the second target. But geneticists have realised this need not stop them: if an organism does not have a target sequence, they can add one using old-fashioned random integration. That might sound like a return to square one, because you cannot determine where the target sequence lands, but it鈥檚 not. All you have to do is create one animal with the target in a good location, and you can then breed it and add any gene to that location in the offspring. 鈥淵ou know exactly what鈥檚 happening, where it鈥檚 going,鈥� says Eggleston. 鈥淚t鈥檚 like a cassette deck.鈥�

Recombinases are becoming increasingly popular with genetic engineers. Eggleston has been working with a particularly promising one called phi C31. His team鈥檚 overall goal is to engineer mosquitoes to make them unable to transmit the malaria parasite. Last year, his team created several strains of mosquito with the phi C31 target sequence. Using the recombinase, they can now add different genes to the same spot time and again, greatly improving the reliability of their results. This method could be used to modify a wide range of animals, and possibly plants too.

鈥淕ene targeting will make gene therapy far safer and genetic engineering vastly more powerful鈥�

When it comes to gene therapy, however, adding a 鈥渃assette deck鈥� in advance is not an option. Fortunately Michele Calos of Stanford University in California, who pioneered the use of phi C31, has shown that it is not necessary. She found that phi C31 can add DNA to 鈥減seudo sites鈥� with only a 30 per cent match to the target. The integration process is less efficient, but chance alone dictates that the genomes of many animals will include at least one pseudo site.

In 2002, her team proved that phi C31 could be used for gene therapy. They used the recombinase method to add a gene for a clotting agent called Factor IX 鈥� the lack of which causes one kind of haemophilia 鈥� to the livers of mice. Soon many of the mice鈥檚 liver cells were pumping out factor IX, enough to cure the disorder (Nature Biotechnology, vol 20, p 1124). 鈥淚t was the first study we did and it worked first time,鈥� Calos says. 鈥淲e have had a lot of successful trials in animals.鈥�

The challenge is proving it will be safe in people. Phi C31 inserts DNA at two main sites in the mouse genome, but in human cells it is less discriminating. Calos has identified 19 main integration sites and another 82 where it occurs from time to time. 鈥淭hat is wider than we would like,鈥� she admits. Even so, having 101 known integration sites is a massive improvement on random insertion. Calos also plans to 鈥渆volve鈥� recombinases that are more specific by mutating the gene for phi C31.

Another cause for concern is a recent study showing that very high levels of phi C31 can trigger chromosomal rearrangements in human cells. Random rearrangements of this kind can cause cancer. Calos points out, however, that the phi C31 rearrangements are not random: they occur at the integration sites, none of which are near cancer-related genes. 鈥淲e may never be able to eliminate off-target activity completely,鈥� she says. 鈥淏ut the bottom line is that we are not seeing any adverse events or tumours in animals.鈥�

Calos hopes clinical trials will begin within two years, with haemophilia likely to be the first disease targeted. There have also been successful tests in mice or human cells of treatments for a hereditary skin disorder, a liver disease called tyrosinemia and X-SCID, the 鈥渂ubble boy鈥� immune disease.

The ultimate goal, though, is to target any sequence, rather than merely deliver DNA to a handful of predetermined sites. Why? Because many genetic disorders are caused not just by the absence of a normal gene, but by the presence of a faulty one. In these cases, you have to fix the spelling mistake; it is not enough to add a corrected 鈥渟entence鈥�.

Enter biotech company Sangamo Biosciences of Richmond, California. Sangamo specialises in designing 鈥渮inc fingers鈥� 鈥� naturally occurring substructures found within most DNA-binding proteins that lock onto specific DNA sequences. In theory, by using novel combinations of zinc fingers, you can make custom proteins that will bind to any sequence you choose.

Gene scissors

Adding zinc fingers to DNA-cutting enzymes called nucleases creates 鈥渕olecular scissors鈥� capable of cutting a strand of DNA at a specific point (see Diagram). Why would you want to do that? The answer lies with the mysterious process of homologous recombination, which is carried out by repair enzymes. Several teams have shown that if you cut the DNA you want to alter at the same time as adding the replacement version, you force the repair enzymes into action and enormously increase the rate of homologous recombination.

In June last year, Sangamo reported that it had used custom-designed scissors to correct the mutation that causes X-SCID in 18 per cent of bone marrow cells taken from patients (Nature, vol 435, p 646). The paper created quite a stir. 鈥淭he Sangamo group has achieved truly remarkable efficiencies,鈥� Dana Carroll of the University of Utah in Salt Lake City told New 杏吧原创 at the time.

Sangamo has since proved its result was not a one-off by taking human immune cells and using molecular scissors to make them resistant to HIV (New 杏吧原创, 2 July 2005, p 14). Repopulating a patient鈥檚 immune system with these cells might keep AIDS at bay indefinitely. 鈥淚 think the potential is enormous,鈥� says Matthew Porteus of the University of Texas, Dallas, one of the authors of the X-SCID paper.

Sangamo is focusing on gene therapy, but molecular scissors also look set to become a key tool in genetic engineering. In 2003 Carroll鈥檚 team showed that they could be used to genetically engineer fruit flies, and has since done the same in plants.

Promising as they are, however, molecular scissors are not the full answer. One problem is that they can cut DNA at sites other than the target, causing double-strand breaks. The breaks can kill cells but this won鈥檛 be a huge issue as long as the scissors are used to treat cells outside the body prior to reimplantation, as Porteus envisages.

More of a worry is that if the breaks are repaired incorrectly, cells might turn cancerous. As no cells modified by scissors have yet been reimplanted in animals, it is not yet clear how big the risk is. 鈥淢y view is that the problem is solvable,鈥� says Porteus. There are other limitations though. For instance, the bigger the piece of DNA you want to insert, the less efficient molecular scissors become.

The ideal solution would be to combine the virtues of both approaches 鈥� the ability to target any desired sequence, as Sangamo can, and the ability to add or remove very large chunks of DNA, as can be done with recombinases. To achieve that means finding a way to alter recombinases to bind to the specific sequence you wish to target. That is not easy, says biochemist Marshall Stark of the University of Glasgow in the UK, because recombinases are far more complex than the nucleases in Sangamo鈥檚 scissors. Nevertheless, his group has already succeeded in altering one recombinase to target a sequence it would not normally recognise. Getting a recombinase such as phi C31 to target any sequence is just a question of time, he says.

The efforts of researchers like Stark and Calos are slowly ushering in a new era in genetic engineering. Soon the extraordinary feats biologists take for granted in mice will become possible in many other species, including our own. Twenty or so lives have already been transformed; millions more are awaiting their turn.