This is an unedited version of the article by David Schubert, a Professor in the Cellular Neurobiology Laboratory at The Salk Institute, published as 'Regulatory Regimes for Transgenic Crops' in the journal Nature Biotechnology (23, 785 - 787; July 2005)

Schubert is responding to Bradford et al's, 'Regulating transgenic crops sensibly: lessons from plant breeding, biotechnology and genomics' which was published in Nature Biotechnology in April 2005 (23(4):439-44).

After analysing their arguments, Schubert concludes, "Because of the high mutagenicity of the transformation procedures used in GE, the assumptions made by Bradford et al. and also the FDA about the precision and specificity of plant GE are incorrect. Nonetheless, it appears that the positions of Bradford et al. and the biotech industry, as well as the current regulatory framework [in the U.S.] for the labeling and safety testing of GE food crops, is to maintain the status quo and hope for the best.

"The problem is that there are no mandatory safety testing requirements for unintended effects and that it may take many years before any symptoms of a GE-caused disease appear. In the absence of strong epidemiology or clinical trials, any health problem associated with an illness caused by a GE food is going to be very difficult, if not impossible, to detect unless it is a disease that is unique or normally very rare."

By David Schubert

In a recent article Bradford and colleagues argued that the methods used to produce food crops should not be the focus of regulatory oversight, only the phenotypic traits of the resultant plants as defined in terms of standard agricultural practice1. They propose that any risk and safety assessments of crops produced by genetic engineering (GE) should be based only upon the nature of the introduced genes. They also claim that transgenic crops face a "daunting" array of regulatory requirements.

However, safety testing requirements in the United States are largely voluntary and in my view inadequate. These regulations have been reviewed elsewhere2 and will not be discussed further. Safety concerns related to the GE process itself as well as its unintended consequences are set aside by Bradford et al as irrelevant, for they claim that the products of genetic events that occur naturally and with standard plant breeding techniques are fundamentally the same as those that occur with GE. Are these arguments a valid reflection of what is known about the precision and consequences of the GE process as compared with naturally occurring genomic variation?

The basic assumption underlying the concept of a one-to-one relationship between the transgene and the resultant phenotype is that the GE process is relatively precise. However, none of the current transgene insertion techniques permit control over the location of the insertion site or the number and orientation of the genes inserted. Indeed, over one-third of all Agrobacterium-mediated insertion events disrupt functional DNA3,4. These and related transformation and cell culture-induced changes in chromosomal structure have been recently documented in great detail5. For example, translocations of up to 40 Kb6, scrambling of transgene and genomic DNA7, large scale deletions of over a dozen genes8 and frequent random insertions of plasmid DNA9 can all be caused by the procedures used to make GE plants. In fact, the most commonly used transformation procedure is sometimes itself used as a mutagen10, and can activate dormant retrotransposons that are highly mutagenic11. Moreover, mutations linked to the transgene insertion site cannot be removed by additional breeding as long as there is selection for the transgene itself. Collectively these data indicate that the GE process itself is highly mutagenic.

Some modern breeding technologies introduce new traits into plants via chemical or radiation mutagenesis or by wide cross hybridizations that overcome natural species barriers. Mutagenesis was used in the United States during the middle part of the last century, but food crops made by this technique now constitute less than a few percent of US production, with sunflowers being the major representative12. However, plants produced by wide crosses, such as those between quackgrass and bread wheat to yield a widely planted grain that has all of the chromosomes of wheat and an extra half genome of the quackgrass, while unique, are fundamentally different from those produced by either mutagenesis or GE. In wide crosses and other forms of ploidy manipulation there are clearly changes in gene dosage, and proteins unique to only one parent can be produced in the hybrid, but there is no a priori reason to assume that mutations are going to occur simply because there is a change in chromosome or gene number. While the extent and suddenness of all of these modern breeding technologies are unlike anything known to occur during the course of evolution or with traditional breeding, only GE and mutagenesis introduce large numbers of mutations. Any new cultivars derived by the latter two methods should be subjected to similar regulatory requirements.

Bradford et al. correctly state that plants normally contain the same Agrobacterium and viral DNA sequences that are used to create GE transfection constructs, but fail to point out that with GE these pieces of DNA are part of a cassette of genes for drug resistance along with strong constitutive viral promoters that are used to express foreign proteins at high levels in all parts of the plant, hardly a natural event. They incorrectly imply that changes in ploidy, gene copy number, recombination, and high genomic densities of transposable elements in normal plants continually lead to mutations and changes in gene expression similar to those caused by GE.

Ploidy is notoriously unstable in plants, but changes involve moving around large blocks of intact genes while maintaining their regulated expression pattern. It should also be remembered that recombination is not the same as random mutagenesis, for there has been tremendous selective pressure for alleles to express functionally similar proteins. The statement that "retrotransposons continuously insert themselves between genes" is incorrect, for these high copy number elements are transpositionally inactive in normal modern food plants13, have evolved and rearranged in the distant past14, but can be activated by tissue culture or by mutagenesis11. In fact their discovery by Barbara McClintock was facilitated by the use of mutagenized corn13.

While Bradford et al. propose that regulatory efforts should be focused upon the expression of the transgene, I believe that the major hazards of the highly mutagenic plant transformation techniques are the potentials for a decrease in nutritional content or an increase in dangerous metabolites. While it is widely recognized that the breeding of some crops can produce varieties with harmful characteristics, millennia of experience have identified these crops, and breeders test new cultivars for known harmful compounds, such as alkaloids in potatoes15,16. In contrast, unintended consequences arising from the random and extensive mutagenesis caused by GE techniques opens far wider possibilities of producing novel, toxic, or mutagenic compounds in all sorts of crops. Unlike animals, plants accumulate thousands of nonessential small molecules that provide adaptive benefits under conditions of environmental or predator-based stress17. Estimates are that they can make between 90,000 and 200,000 phytochemicals with up to 5000 in one species18. These compounds are frequently made by enzymes with low substrate specificity19 in which mutations can readily alter substrate preference20,21

There are many examples of unpredictable alterations in small molecule metabolism in GE organisms. In yeast genetically engineered to increase glucose metabolism, the GE event caused the unintended accumulation of a highly toxic and mutagenic 2-oxoaldehyde called methylglyoxal22. In a study of just 88 metabolites in four lines of potatoes transformed for altered sucrose metabolism, Roessner et al. found that the amounts of the majority of these metabolites were significantly altered relative to controls18. In addition, nine of the metabolites in GE potatoes were not detected in conventional potatoes. Given the enormous pool of plant metabolites, the observation that 10% of those assayed are new in one set of transfections strongly suggests that undesirable or harmful metabolites may be produced and accumulate23. Contrary to the suggestions of Bradford et al., Kuiper and his colleagues strongly recommend that each transformation event should be assayed for these types of unintended events by metabolic profiling24.

A well documented horticultural example of unintended effects is the alteration in the shikimic acid pathway in Bt corn hybrids derived from Monsanto's MON810 and Syngenta's Bt11 plants as well as glyphosate-tolerant soybeans. Stem tissue of both groups of plants has elevated levels of lignin, an abundant non-digestible woody component that makes the plants less nutritious for animal feed25,26. Components of this same biochemical pathway also produce both flavonoids and isoflavonoids that have a high nutritional value, and rotenone, a plant-produced insecticide that may cause Parkinson's disease27. Isoflavonoids are abundant in legumes like soy beans, and rotenone is synthesized directly from isoflavones in many legume species28. Because of the promiscuity of many plant enzymes and the large and varied substrate pools of phytochemical intermediates, it is impossible to predict the products of enzymes or regulatory genes mutated during the GE event23. While I are not aware of any testing of GE soybeans for rotenone, it has been shown that glyphosate-tolerant soybeans sprayed with glyphosate have a reduced flavonoid content29.

The safety testing of GE crops need not be as extensive as that done with drugs, food additives or cosmetics. Many suggestions have been put forward (see, for example 30,2,5,24) including those by the World Health Organization31. I believe that the most important safety tests include metabolic profiling to detect unexpected changes in small molecule metabolism24 and the Ames test to detect mutagens32. Molecular analysis of the gene insertion sites and transformation-induced mutations5 should also be performed along with both multigenerational feeding trials in rodents to assay for teratogenic effects and developmental problems, and allergenicity testing performed according to a single rigorous protocol31 The animal studies are of particular importance for crops engineered to produce precursors to highly biologically active compounds such as Vitamin A and retinoic acid, molecules that can act as teratogens at high doses33.

In summary, Bradford et al. state that there is a low risk from the consumption of GE plants "where no novel biochemical or enzymatic functions are imparted". The question is, of course, how can one know if a novel and potentially harmful molecule has been created unless the testing has been done? How can one predict the risk in the absence of an assay? Because of the high mutagenicity of the transformation procedures used in GE, the assumptions made by Bradford et al. and also the FDA 34 about the precision and specificity of plant GE are incorrect. Nonetheless, it appears that the positions of Bradford et al. and the biotech industry, as well as the current regulatory framework for the labeling and safety testing of GE food crops, is to maintain the status quo and hope for the best.

The problem is that there are no mandatory safety testing requirements for unintended effects2 and that it may take many years before any symptoms of a GE-caused disease appear. In the absence of strong epidemiology or clinical trials, any health problem associated with an illness caused by a GE food is going to be very difficult, if not impossible, to detect unless it is a disease that is unique or normally very rare. Therefore, while GE may be able to enhance world health and food crop production , its full potential is likely to remain unfulfilled until rigorous pre-release safety testing can provide some assurance to consumers that the products of this new technology are safe to eat.


1. Bradford, K. J., Van Deynze, A., Gutterson, N., Parrott, W. & Strauss, S. H. Regulating transgenic crops sensibly: lessons from plant breeding, biotechnology and genomics. Nat Biotechnol 23, 439-44 (2005).

2. Freese, W. & Schubert, D. Safety testing of genetically engineered food. 21 Biotechnology and Genetic Engineering Reviews, 299-325 (2004).

3. Szabados, L. et al. Distribution of 1000 sequenced T-DNA tags in the Arabidopsis genome. Plant J 32, 233-42 (2002).

4. Forsbach, A., Schubert, D., Lechtenberg, B., Gils, M. & Schmidt, R. A comprehensive characterization of single-copy T-DNA insertions in the Arabidopsis thaliana genome. Plant Mol Biol 52, 161-76 (2003).

5. Wilson, A., Latham, J. & Steinbrecher, R. 35 (EcoNexus, Brighton, UK, 2004).

6. Tax, F. E. & Vernon, D. M. T-DNA-associated duplication/translocations in Arabidopsis. Implications for mutant analysis and functional genomics. Plant Physiol 126, 1527-38 (2001).

7. Makarevitch, I., Svitashev, S. K. & Somers, D. A. Complete sequence analysis of transgene loci from plants transformed via microprojectile bombardment. Plant Mol Biol 52, 421-32 (2003).

8. Kaya, H. et al. Hosoba toge toge, a syndrome caused by a large chromosomal deletion associated with a T-DNA insertion in Arabidopsis. Plant Cell Physiol 41, 1055-66 (2000).

9. Kim, S. R. et al. Transgene structures in T-DNA-inserted rice plants. Plant Mol Biol 52, 761-73 (2003).

10. Weigel, D. et al. Activation tagging in Arabidopsis. Plant Physiol 122, 1003-13 (2000).

11. Hirochika, H., Sugimoto, K., Otsuki, Y., Tsugawa, H. & Kanda, M. Retrotransposons of rice involved in mutations induced by tissue culture. Proc Natl Acad Sci U S A 93, 7783-8 (1996).

12. Ahloowalia, B. S., Maluszynski, M. & Nichterlein, K. Global impact of mutation-derived varieties. Euphytica 135, 187-204 (2004).

13. Feschotte, C., Jiang, N. & Wessler, S. R. Plant transposable elements: where genetics meets genomics. Nat Rev Genet 3, 329-41 (2002).

14. Brunner, S., Fengler, K., Morgante, M., Tingey, S. & Rafalski, A. Evolution of DNA Sequence Nonhomologies among Maize Inbreds. Plant Cell 17, 343-60 (2005).

15. Korpan, Y. I. et al. Potato glycoalkaloids: true safety or false sense of security? Trends Biotechnol 22, 147-51 (2004).

16. Ewen, S. W. & Pusztai, A. Effect of diets containing genetically modified potatoes expressing Galanthus nivalis lectin on rat small intestine. Lancet 354, 1353-4 (1999).

17. Verpoorte, R. in Metabolic Engineering of Plant Secondary Metabolism (eds. Verpoorte, R. & Alfermann, A. W.) 1-29 (Kluwer Academic Publishers, Dordrecht , The Netherlands, 2000).

18. Roessner, U. et al. Metabolic profiling allows comprehensive phenotyping of genetically or environmentally modified plant systems. Plant Cell 13, 11-29 (2001).

19. Schwab, W. Metabolome diversity: too few genes, too many metabolites? Phytochemistry 62, 837-49 (2003).

20. Zubieta, C. et al. Structural basis for substrate recognition in the salicylic acid carboxyl methyltransferase family. Plant Cell 15, 1704-16 (2003).

21. Johnson, E. T. et al. Alteration of a single amino acid changes the substrate specificity of dihydroflavonol 4-reductase. Plant J 25, 325-33 (2001).

22. Inose, T. & Murata, K. Enhanced accumulation of toxic compound in yeast cells having high glycolytic activity: A case study on the safety of genetically engineered yeast. Intl J Food Sci Tech 30, 141-6 (1995).

23. Grotewold, E. Plant metabolic diversity: a regulatory perspective. Trends Plant Sci 10, 57-62 (2005).

24. Kuiper, H. A., Kleter, G. A., Noteborn, H. P. & Kok, E. J. Assessment of the food safety issues related to genetically modified foods. Plant J 27, 503-28 (2001).

25. Saxena, D. & Stotzky, G. Bt corn has a higher lignin content than non-Bt corn. Amer J Botany 88, 1704-6 (2001).

26. Gertz, J. M., Vencill, W. K. & Hill, N. S. in Proceedings of the 1999 Brighton Crop Protection Conference: Weeds 835-840 (British Crop Protection Council, Farnham, UK, 1999).

27. Betarbet, R. et al. Chronic systemic pesticide exposure reproduces features of Parkinson's disease. Nature Neurosci. 3, 1301-1306 (2000).

28. Morgan, E. D. & Wilson, I. D. in Comprehensive Natural Products Chemistry (ed. Mori, K.) 363-375 (Pergamon Press/Elsevier Science, Oxford, 1999).

29. Lappe, M. A., Bailey, E. B., Childress, C. & Setchell, K. D. R. Alterations in clinically important phytoestrogens in genetically modified, herbicide-tolerant soybeans. J Med Foods 1, 241-245 (1999).

30. Edmonds_Institute. Manual for assessing ecological and human health effects of genetically engineered organisms. (Edmonds Institute, 1998).

31. FAO-WHO. Evaluation of Allergenicity of genetically modified foods. Report of a Joint FAO/WHO expert consultation on allergenicity of foods derived from biotechnology. January 22-25, 2001. (2001).

32. Maron, D. M. & Ames, B. N. Revised methods for the Salmonella mutagenicity test. Mutat Res 113, 173-215 (1983).

33. McCaffery, P. J., Adams, J., Maden, M. & Rosa-Molinar, E. Too much of a good thing: retinoic acid as an endogenous regulator of neural differentiation and exogenous teratogen. Eur J Neurosci 18, 457-72 (2003).

34. Kessler, D. A., Taylor, M. R., Maryanski, J. H., Flamm, E. L. & Kahl, L. S. The safety of foods developed by biotechnology. Science 256, 1747-9, 1832 (1992).