Genetically modified crops and the food derived from them have had a bad press, but a rare piece of good news is provided by Daniell et al.(1) in this month's Nature Biotechnology. They report a development that has profound implications for the risk assessment of genetically modified crops. Most crops are modified by inserting genes into the nucleus, and the genes can therefore spread to other crops or wild relatives by movement of pollen. By engineering tolerance to the herbicide glyphosate into the tobacco chloroplast genome, however, the authors have not only obtained high levels of transgene expression, but, because chloroplasts are inherited maternally in many species, they have also prevented transmission of the gene by pollen — closing a potential escape route for transgenes into the environment.

Glyphosate is the most widely used herbicide in the world. It interferes with 5-enol-pyruvyl shikimate-3-phosphate synthase (EPSPS), an enzyme that is encoded by a nuclear gene and catalyses a step in the biosynthesis of aromatic amino acids in the chloroplasts. Conventional strategies for producing glyphosate-tolerant plants are to insert, into the nucleus, an EPSPS gene from a plant or a glyphosate-tolerant bacterium (the bacterial gene is modified so that the enzyme is correctly targeted to the chloroplasts), or a gene that inactivates the herbicide.

But Daniell et al. have now achieved maternally inherited glyphosate tolerance in tobacco by particle bombardment of leaves with a vector containing the EPSPS gene of petunia. The gene was modified with border sequences that allow integration into the chloroplast genome by homologous recombination and, using the polymerase chain reaction and Southern blot analysis, the authors confirmed that the gene had inserted into chloroplast DNA of regenerated plants. All of the chloroplasts were transgenic, and the very high copy number of the gene (up to 10,000) resulted in a tenfold increase in glyphosate tolerance compared with transgenic plants that carry the same gene in the nuclear genome (2).

Genes that confer tolerance to herbicides have been the target of genetic engineers for more than a decade. Up to January this year, there were nearly 2,300 experimental releases of herbicide-tolerant crops in OECD countries (about 35% of all genetically modified crop releases up to then) (3). As well as its agronomic objective, herbicide tolerance has been widely used as a selectable marker, enabling the easy detection of successful transformations.

Herbicide-tolerant crops such as the glyphosate-tolerant soybean Glycine max were among the first genetically modified species to be grown commercially in North America. Such crops are also near to market in Europe — for example, both glyphosate- and glufosinate ammonium-tolerant oilseed rape, Brassica napus, are undergoing variety trials in the United Kingdom. Tolerance to glyphosate has featured in almost one-third of all herbicide-tolerance field trials and in 20 different crops, ranging from lettuce and tomato to eucalyptus, chestnut and poplar, although the major crop targets have been maize, soybean, oilseed rape, sugar beet and cotton.

The environmental and agricultural implications of the large-scale cultivation of herbicide-resistant crops have been hotly debated (4). Although disagreeing about the possible effects on the use of herbicides, environmentalists and industry both recognize the potential problems created by gene flow to wild relatives and, specifically, the creation of herbicide-tolerant weeds. These problems are addressed by regulators such as the UK Advisory Committee on Releases to the Environment in their risk assessment of genetically modified crops.

Hybridization with wild relatives may cause problems in a number of ways: by the creation of 'superweeds' (5); by genetic erosion (particularly in centres of diversity); and — a somewhat purist concept — through genetic pollution of natural gene pools. Crops from which gene flow to wild relatives is known to have occurred include quinoa, squash, carrot, maize, sorghum, sunflower, strawberries and sugar beet (5,6) (Fig. 1). Particular problems with gene flow are anticipated in outbreeding crops such as forage grasses, which have extensive dispersal of pollen and many sexually compatible wild relatives. In the case of glyphosate tolerance, preventing gene flow to weedy relatives is especially worthwhile because tolerance in weeds does not seem to have evolved, despite extensive use of the herbicide for over 20 years (7). By contrast, resistance to more than 15 herbicide groups has been observed in well over 100 species of plant worldwide (8).

As well as preventing escape to wild relatives, maternal inheritance of transgenes would prevent (or reduce) gene 'stacking' by cross-contamination of crops — an advantage that is not mentioned by Daniell et al.(1). For example, oilseed rape plants tolerant to both glyphosate and glufosinate ammonium could appear where genetically modified crops are grown adjacently. The ability to prevent the contamination of food crops with genes involved in the production of industrial and pharmaceutical products is also of great potential benefit.

Of course, maternal inheritance will not completely prevent the escape of transgenes. Paternal or biparental inheritance of chloroplast DNA is common in gymnosperms (plants in which the seed is not protected by an ovary), and this also occurs in several genera of angiosperm (flowering plants) (9). Furthermore, like conventional crops, transgenic plants will be able to form volunteer populations — plants that arise in fields after harvest. Such populations can cause economic losses because of direct competition with the intended crops for resources such as light and water, and because they can act as 'green bridges' (hosts for pests and pathogens while crop plants are absent). Nor will maternal inheritance stop transgenic crops from establishing 'feral' populations. In the United Kingdom, for example, oilseed rape is notorious for producing such populations on roadsides because of seed spillage during transport to processing factories (10). Moreover, the seed can persist in the soil for several years and germinate after the ground is disturbed (11).

Although one of the escape routes for genes from genetically modified crops has been cut off with the work of Daniell et al., not all of the exits are blocked. Nevertheless, the option to prevent transgene movement via pollen offers scope for simplifying the risk assessment of transgenic crops with strict maternal inheritance of chloroplasts — particularly those species in which pollen dispersal is extensive and male-sterility systems are absent or unreliable.

References

1. Daniell, H., Datta, R., Varma, S., Gray, S. & Lee, S.-B. Nature Biotechnol. 16, 345-348 (1998).
2. Shah, D. M.et al. Science 233, 478-481 (1986).
3. http://www.olis.oecd.org/biotrack.nsf
4. Hill, J. E. Brighton Crop Prot. Conf. -- Weeds 1027-1034 (1995).
5. Kling, J. Science 274, 180-181 (1996).
6. Raybould, A. F. & Gray, A. J. Trends Ecol. Evol. 9, 85-89 (1994).
7. Bradshaw, L. D., Padgette, S. R., Kimball, S. L. & Wells, B. H. Weed Technol. 11, 189-198 (1997).
8. Cousens, R. & Mortimer, M. Dynamics of Weed Populations (Cambridge Univ. Press, 1995).
9. Reboud, X. & Zeyl, C. Heredity 72, 132-140 (1994).
10. Crawley, M. J. & Brown, S. L. Proc. R. Soc. Lond. B 259, 49-54 (1995).
11. Charters, Y. M., Robertson, A., Wilkinson, M. J. & Ramsay, G. Theor. Appl. Genet. 92, 442-447 (1996).
12. Boudry, P., Mrchen, M., Saumitou-Laprade, P., Vernet, P. & van Dijk, H. Theor. Appl. Genet. 87, 471-478 (1993).