perspectives Transfer of engineered genes from crop to wild plants Rikke Bagger Jorgensen, Thure Hauser, Thomas Raundahl Mikkelsen and Hanne Osterg rd The escape of engineered genes - genes inserted using recombinant DNA techniques - from cultivated plants to wild or weedy relatives has raised concern about possible risks to the environment or to health. The media have added considerably to public concern by suggesting that such gene escape is a new and rather unexpected phenomenon. However, transfer of engineered genes between plants is not at all surprising, because it is mediated by exactly the same mechanisms as those responsible for transferring endogenous plant genes: it takes place by sexual crosses, with pollen as the carrier. Such sexual reproduction has been the basis for breeding almost all crops. D ispersal of pollen from one plant to another is only one necessary step in successful gene transfer. For transfer to occur, a number of other requirements must also be met: the distributions of the parents must overlap in space and time; the 'alien' pollen must accomplish successful fertilization; and the hybrid zygote must develop into a viable plant with fertility sufficient to pass the genes on to future generations by backcrosses, selfing or crosses with siblings. Introgression (introgressive hybrid- ization) is the process whereby crosses between plants result in a stable in- corporation of genes from one differen- tiated gene pool into another1,2. Stable integration of new genetic material into the recipient genotype is possible only if the chromosomes of the donor and recipient can exchange genes by recombination. In general, the more distantly related the donor and the recipient are, the more likely it is that meiotic abnormalities will reduce the recombination potential and so also reduce the possibility of donor genes being integrated into the new geno- type. Therefore, intraspecific gene transfer is more common and effective than interspecific transfer. Successful introgression is a complex phenom- enon, and thus analyzing single steps of the process does not justify con- clusions about the extent to which gene transfer takes place. Recombination of new genetic ma- terial into the genetic background of the recipient could occur during mei- osis in the primary hybrid. In hybrids with a complex genomic constitution (e.g. hybrids between species separated by large genomic differences) inte- gration by recombination may occur in later backcross generations. Whereas genetic modification is a purposeful action transferring one or more chosen genes to a target plant, the process of gene transfer through meiotic recombi- nation during sexual reproduction 'drags along' neighbouring genetic material from the donor. The fre- quency with which recombination will break linkage between donor genes may depend on the direction (female to male, or vice versa) of the cross and the position of the engineered gene in the donor genome, because certain chromosomal regions are more active in recombination than others3, 4. Genetic material from the donor may have an impact - either positive or negative - on the vigour of the recipi- ent plant, and hence on the likelihood of transmission of the engineered gene to subsequent generations. As a result of introgression, an engi- neered gene - together with adherent donor genes - is incorporated into a foreign genotype. Here, the expression and inheritance of the engineered gene is controlled by its interaction with other donor genes, the endogenous genes and the environment4. This novel genotype-environment combi- nation may affect natural ecosystems, cultivated land or health. How frequent is introgression? Introgressive hybridization is often considered to be a significant force in plant evolution2, 5. However, Rieseberg and Wendel2 reviewed 165 proposed cases of introgressive hybridization and found that unequivocal verifi- cation of introgression was difficult. Traits shared as a result of introgres- sion are hard to separate from primi- tive traits inherited from a common ancestor or from traits resulting from convergent evolution. When geneti- cally modified plants are released, such character ambiguity is prevented by the introduction to the environment of donors with novel traits - engi- neered genes are the perfect markers of introgression. However, natural introgression has still not been demon- strated with an engineered gene as a marker, probably because until now the cultivation of genetically modified crops has been largely restricted to trials. Crops with close weedy relatives are more prone to gene transfer than crops where an intense domestication has resulted in ecological and reproductive isolation from related wild plants 6. Gene exchange between a crop and a weedy relative may increase the adapt- ability of the weed, making it even more weedy. Gene transfer has added to adaptability in weed beets (intro- gression between the crop Beta vul- garis and related wild species) 7, red rice (Oryza sativa) (intraspecific crosses with cultivated rice) 8 and possibly also in lettuce (introgression between the crop Lactuca sativa and related wild species) 9. We investigated another example of crop-weed gene transfer in field experiments with sown mixtures of oilseed rape (Brassica napus) and its weedy relative B. rapa (synonym B. campestris). Endogenous oilseed rape genetic markers as well as en- gineered genes were used to reveal gene transfer. Oilseed rape - a disperser of genes The first genetically modified oil- seed rape cultivars with engineered herbicide tolerance have been mar- keted lo. This could result in introgres- sion of engineered genes to B. rapa, a common weed of oilseed rape fields in southern Scandinavia and North America (Fig. 1). Oilseed rape (genomic constitution AACC) is an ancient crop plant, but the origin of the species is probably rather recent in evolutionary termslk 356 October 1996, Vol. 1, No, 10 © 1996 Elsevier Science Ltd PII S1360-1385(96)20011-1