mice were exposed to each dose; in no case was significant mortality, weight loss, or a general adverse clinical symptom found.
There is a limit to the amount of plant material that can be consumed by mice, and that limits testing to acute tests with purified material. In a novel approach (Jackson et al. 1995), Bt-producing corn (finely ground seed) was incorporated into the diet of catfish (100 fish per treatment) over a period of 66 days. The ground corn made up 35% by weight of the catfish diet. Non-Bt corn was used as a control. No significant effects on weight gain or rates of feed conversion were found. Although this is not a traditional toxicity test for human health assessment, the duration of the test, the sample size per treatment, and the amount of plant material consumed provide useful information. Catfish are not a close physiological model for humans, but forage- and grain-consuming mammals might be more appropriate models (see section 2.5).
The committee concludes that
Although a number of the experiments performed in support of registration for transgenic pest-protected plants containing Bt proteins could be improved by modifications suggested above, the total weight of evidence from combined studies presented and previous knowledge about Bt proteins, provides reasonable support for the toxicological safety of crops containing the tested Bt proteins (that is, Cry1Ab, Cry1Ac, and Cry3A).
Similar tests of other Bt proteins would be appropriate in the future, but a question remains about the strength of data that should be required when a class of plant-defensive substances has not been previously characterized as well as Bt proteins have been. As stated above, biochemical and functional equivalence is difficult to prove. Functional activity can be affected by single amino acid substitutions that would be difficult to detect with current methods. Although the sequence of the cloned gene is usually known and will largely remain intact during plant expression, modifications such as amino acid substitutions, proteolytic processing, and glycosylation are all possible. Therefore, it would be helpful to use plant-produced defensive substances in as many tests as is feasible. However, because it may often be difficult to extract and purify pest-protective