Letter to the Editor, 17 December 2005
Crop testing
Bill Freese, Washington DC, US
http://www.newscientist.com/article/mg18825300.200;jsessionid=BAFLFMPDKMDE
New Scientist, issue 2530

When expressed in transgenic peas, an innocuous bean protein elicits immune reactions in mice, reviving concerns about the allergenic potential of genetically modified foods (26 November, p 3 and p 5). These "surprising results" from researchers in Australia raise several intriguing questions.

Should regulators require the use of animal models? Allergenicity assessments of transgenic proteins in GM crops are usually limited to in vitro tests of digestive stability, database searches for sequence similarities to known allergens, and in some cases a heat stability test.

While certainly cheap and convenient for GM crop developers, such tests provide no direct immunological information and cannot rule out allergenic proteins. Both the BALB/c mouse strain used in the Australian pea study and the brown Norway rat have shown promise as predictors of human allergic response.

Also, at present, all testing is performed on a bacterial surrogate of the protein, rather than that produced by the plant. GM crop developers complain that it is too inconvenient to extract sufficient quantities of transgenic protein from their plant. But if peas and beans - both legumes - can generate immunologically distinct proteins from the same gene, surely the same is true of bacterium and plant. Thus, results of testing on bacterial surrogates may not reflect the toxic or allergenic profile of the in planta protein people are exposed to.

Other factors also argue against use of bacterial surrogates. For example, allergenic proteins are often glycosylated, and plant glycosylation patterns have been implicated in allergenicity. Bacteria, in contrast, seldom glycosylate proteins.

Finally, perhaps regulators should demand full sequencing of the transgenic proteins in plants. At present, the standard practice is to sequence just 5 to 25 amino acids at the N-terminal as a demonstration of "identity", even if the putative protein is 600-plus residues long. Since the transformation process - the insertion of foreign DNA to a cell - can be sloppy and even point mutations can transform an innocuous protein into an immunogenic/aggregating one, it is unclear why this basic information is not required.

The editor writes:

These are excellent suggestions, and perhaps such a test regime should also be applied to new conventionally bred varieties. These can sometimes generate unexpected allergens or toxins too. For example, a potato variety called Lenape was withdrawn from the US market in the 1960s when it was found to contain dangerously high levels of potato toxins called solanidine glycosides. And in the mid-1980s, American growers abandoned a variety of celery because it contained high levels of psoralens - chemicals which become irritants when activated by sunlight. Workers picking the celery developed skin rashes. Psoralens also occur in parsnips, and have been known to cause rashes through skin contact.