New Scientist

Unpalatable Truths

Demanding proof that genetically modified foods are safe is all very well, but without a rational system for testing conventional foods, we may never get it.

Earlier this year, Britain was rocked by claims that genetically modified foods are dangerous. Arpad Pusztai, a biochemist who used to work at the Rowett Research Institute in Scotland, said he had shown that GM potatoes were harmful to rats because of their genetic modification alone.

Were the GM potatoes toxic ? On the basis of Pusztai's evidence, it's impossible to say. In fact, his results support only one obvious conclusion: rats hate potatoes.

Pusztai fed separate groups of rats on normal or GM potatoes to see if the GM food had different effects. That's good, basic toxicology. Unfortunately he couldn't make the animals eat enough potato, so they were malnourished no matter which kind they were eating.

According to toxicologists who examined the data, changes in their organ weights and immune reactivity showed no unambiguous association with genetic modification (This Week, 6 March, p 13). Starvation or known toxins in raw potato were the most likely culprits for any changes seen in the rats.

These experiments reveal a serious problem that is only now being grasped by the biotechnology industry: standard toxicology tests don't work for food. It is often difficult to feed lab animals enough GM fodder, whether or not they find it palatable, to see if it has undesirable effects compared with unmodified food. Essentially, animal models are not sensitive enough to reveal small differences between modified and unmodified foods.

Nonstarter

Even if you manage to get animals to eat enough test food, you risk changing their diet so profoundly that even those eating unmodified food will be abnormal. For all but the most blatantly toxic GM foods, this may make it impossible to draw meaningful conclusions from such experiments.

Politicians, taken aback by huge public mistrust of "Frankenfoods", are also realising that safety testing of these foods is not straightforward. In Britain, the Cabinet's biotechnology committee has commissioned a report on the human health implications of GM foods from the government's Chief Medical Officer and Chief Scientific Adviser, due to be published this month. A Cabinet Office memo, leaked by Friends of the Earth, asks: "Why don't we require a pharmaceutical-type analysis of the safety of these foods, with proper trials?" But as the problems to date have shown, the proposition is a nonstarter.

So how can we check the safety of GM food ? Scientists from the 29 industrialised countries of the OECD concluded at a meeting in Paris in December that a whole new approach is needed. In September, they will meet again to start drawing up ways of carrying out such checks.

They are up against some serious logistical problems. Harry Kuiper of the State Institute for Quality Control of Agricultural Products in Wageningen, Netherlands, tested a GM tomato by freeze-drying it and feeding so much to rats that each got the equivalent of 13 of fresh tomatoes a day. Any more, and they would have been poisoned by the basic nutrients, such as potassium, in the tomato powder.

"But toxicologists still said we hadn't fed them enough to get a meaningful result," says Kuiper. The usual approach for testing a new food additive, for instance, is to feed it to a rat until a toxic effect is observed. That way, you get an idea of the nature and threshold of any toxicity. But with tomatoes, the researchers never managed to reach that threshold. In standard toxicological terms, says Kuiper, they have not been adequately tested. Others would argue that if such large amounts are harmless, the food cannot reasonably be called toxic.

Nonetheless, these difficulties mean that GM food developers usually avoid testing whole foods. Instead, they try to isolate the changed portion and test that. As an example, Roy Fuchs, head of scientific affairs at Monsanto, one of the world's biggest developers of GM food, quotes potatoes carrying a gene for the Bt toxin, an insecticide normally produced by "Bacillus thuringensis". Monsanto sells its Bt potatoes in the US and is applying for a European licence. Fuchs says that the potatoes, like all genetically engineered plants so far, do not produce enough of the product of the novel gene for it to be isolated from the plants themselves and tested. "So we put the novel genes in bacteria, produce the gene product and test it by conventional methods." However, the protein made by the bacteria may not be the same as that made by the plant, especially in its potential to cause allergy.

The production of a novel protein is only one of the potentially harmful changes that occur in when a foreign gene is inserted into a plant. Because the positioning of the novel gene within the plant's DNA is essentially random, it may alter the plant's expression of its own genes - with unpredictable effects. It is this kind of change that stymies conventional toxicology. Food is a complex mixture of substances that occur in different quantities in different varieties of crops and in the same variety grown under even slightly different conditions. When is a change in one or several of those substances a problem?

Unfortunately, says Kearns, no one has ever tested conventional food for toxicity, so no one quite knows where to start. One exception is potatoes. Conventional plant breeders in the US and the Netherlands test new potato varieties for elevated levels of known toxins such as solenines. French breeders do not - and there are no legal requirements in any country to do so. And that still leaves toxins in GM foods that we may not yet know about. "We have to think through these things case by case," says Kearns, starting with a better understanding of what is in normal crops.

Kuiper's institute is working on a screening test that detects differences in the pattern of messenger RNA molecules produced by normal and transgenic tomatoes. The hope it that this will provide a fast way to see if there have been large changes in gene expression. The method can reliably detect differences between red and green tomatoes - which is encouraging, says Kuiper, because green ones produce more toxins.

Key differences

The team has also compared the chemicals synthesised by normal and transgenic plants by looking at their nuclear magnetic resonance (NMR) spectra. Nearly every chemical compound in the plant produces a characteristic "fingerprint" of peaks. The screening test revealed that there were up to eightfold differences in concentrations of sugars, amino acids and various unidentified compounds. Impressive as this sounds, it may not be significant: Kuiper notes that there were greater differences between unaltered tomatoes grown in different conditions than there were between GM and normal tomatoes grown in identical conditions.

A better way of exploiting NMR might be to use it to find substances that differ in transgenic foods and then to test these substances in, for example, cell cultures, to see if the changes could be harmful. The need for such tests may be soon be pressing. But when crops are engineered to produce a number of desired nutrients or "nutraceuticals", changes in the plant's own gene expression could become much more complex and their potentially toxic effects harder to test.

However, proponents of GM foods point out that whichever direction food testing goes, the subtly altered products on our plates will have been tested far more thoroughly than any conventional food. After all, even ordinary kidney beans are poisonous if undercooked. Dozens of people die each year from cyanide from peach seeds. Manioc, the staple diet of millions, had to be grated, squeezed and cooked to drive off the cyanide before improved varieties became available. And some of the most notorious food-linked poisons, such as aflatoxins in grain, do not come from the food but moulds that infect it. In the comparison between modified and unmodified foods, nothing is clear cut. And testing is never simple.