The potential for ecological release following introgression of virus-resistance transgenes into natural populations of wild Brassica species 

Proceedings of the 6th International Symposium on 
The Biosafety of Genetically Modified Organisms Jul00

http://www.ag.usask.ca/isbr/Symposium/Proceedings/Section0.htm

A.F. Raybould,1 A.E. Jones,1 M. Alexander,1 D. Pallett,2 M.I. Thurston,2 J.I. Cooper,2 M.J. Wilkinson3 and A.J. Gray1

1Centre for Ecology and Hydrology, Winfrith Technology Centre, Dorchester, Dorset, United Kingdom 2Centre for Ecology and Hydrology, Mansfield Road, Oxford, United Kingdom 3Department of Agricultural Botany, School of Plant Sciences, The University of Reading, Whiteknights, Reading, United Kingdom

Introduction

The introgression of transgenes for virus resistance from crops into wild plants has the potential to change the composition of natural plant communities. If the gene codes for viral RNA, there is the possibility of recombination between the transgene message and RNA from other viruses infecting the transgenic plant, which could produce viruses with broader host range or greater virulence. New viruses could also be produced if the transgene codes for a viral coat protein that can package nucleic acid from other viruses (transcapsidation) (Falk and Bruening 1994). Plant communities may also change if the abundance of a wild plant is controlled by virus infection. In this situation, acquisition of virus resistance could cause the population growth rate of the wild relative to increase, leading to the invasion of habitats occupied by other species (Cooper and Raybould 1997). Unlike recombination and transcapsidation, such "ecological release" is independent of the nature of the transgene conferring resistance.

Assessment of the likelihood and extent of ecological release following the cultivation of virus-resistant crops requires data on several aspects of the biology of the wild relative and the GM crop. First, we need to know whether gene flow between the crop and wild relative is possible and to what extent it will occur after commercial release of the crop. We also need to know the amount of gene flow among populations of the wild relative. Finally, we need to assess whether population size of the wild relative is suppressed by virus infection. Here, we describe methods to estimate the extent of these processes following the release of GM virus-resistant oilseed rape (Brassica napus) in the United Kingdom.

Hybridization with wild relatives

Hybrids between rape and many wild crucifer species have been produced under laboratory conditions, often using male-sterile rape as the female parent (Scheffler and Dale 1994). "Spontaneous" gene flow by fertilization of a fully fertile wild relative with rape pollen has been recorded for Brassica rapa, B. oleracea, B. juncea, Hirschfeldia incana and Raphanus raphanistrum. Only B. rapa and (possibly) B. oleracea are natives of the U.K. The proportion of hybrid seed set is strongly dependent on the experimental design. Over 90% of seed on isolated B. rapa plants in rape fields can be hybrids (Jørgensen and Andersen 1994), whereas 0.4%_1.5% of seeds are hybrids in B. rapa growing on the margins of rape fields (Scott and Wilkinson 1998).

Clearly, predicting the amount of hybridization after a commercial scale release by extrapolation from small-scale observations is not ideal. Therefore, we devised a method to measure hybridization at the regional scale, based on knowledge of the biology of the wild relatives (Wilkinson et al., in press). Hybridization between B. napus and its parental species will be almost entirely restricted to wild populations adjacent to B. napus fields because of the rapid decline in airborne pollen density with distance from B. napus fields and pollen competition from conspecifics (Scott and Wilkinson 1994). We also know that in the south of England, B. rapa grows mainly on riverbanks and B. oleracea is confined to coastal cliffs.

Using remote sensing data from spring 1998, we located rape fields in a 15,000 km2 area of southeast England and identified those within 50 m of a river (potential sites of sympatry with B. rapa) or coastal cliffs known to have B. oleracea. In spring/early summer 1999, we visited these sites of potential sympatry and took leaf samples of all B. rapa plants or B. oleracea seedlings. The samples were tested by flow cytometry to identify triploid "potential hybrids." We used microsatellite analysis to determine triploids as hybrids or autotriploids. We also sampled plants from 21 populations of ruderal Brassica populations to test whether B. rapa is restricted to riverbanks in the study area.

Lanes 1-3 = B. napus: AACC genome

Lanes 4-7 = known (synthetic) hybrid: AAC genome

Lanes 8-9 = `putative' hybrid: 3n

Lanes 10-13 = B. rapa: AA genome

Lanes 14-21 = B. oleracea: CC genome

Figure 1 Confirmation that the triploid plant from Cockmarsh is a hybrid between B. rapa and B. napus because it contains A and C genome-specific microsatellite bands.

We identified nearly 100 sites of potential sympatry between rape and B. rapa. However, only two sites had B. rapa populations, with 505 plants in total. One plant from Cockmarsh, Berkshire (SU 885 874) was triploid and contained both A and C genome-specific microsatellites, confirming it as a hybrid (Figure 1). At the single site of sympatry between B. oleracea and rape, all 9 of the newly recruited seedlings were diploid. All ruderal Brassicas proved to be tetraploid B. napus. Therefore hybridisation between rape and its parental species in south east England is very rare, largely because there are very few sites of sympatry (Wilkinson et al., in press).

Gene flow among populations of wild Brassica species

An important factor determining the impact of a transgene in natural populations is the extent of gene flow from an initial population of transgenic hybrids. We estimated allele frequencies at 7 microsatellite loci in 7 populations of B. oleracea in Dorset, UK (Raybould et al 1999a). From these data, we calculated unbiased estimates of q between all pairs of populations. The parameter q is a measure of allele frequency differences among populations and, given certain assumptions, is related to the amount of gene flow (Nm) such that Nm = (1/4q ) _ 0.25

We found a statistically significant regression between log10 Nm and log10 distance between populations (Figure 2). From the regression slope, we estimated the distance at which (on average) a pair of populations exchanges less than one migrant per generation _ a rough measure of isolation distance. The estimate is 6.1km, with 95% confidence intervals of 1.5_10.4 km. The confidence intervals are underestimates because they assume the 21 pair-wise Nm values are independent data points, whereas the independent data points are the population allele frequency estimates.

Preliminary data from four microsatellite loci and four populations suggest that B. nigra in Dorset has a similar distribution of genetic variation to B. olercaea. Mean q over all populations is 0.245 for black mustard (A.E. Jones, in prep.) and 0.234 for wild cabbage (Raybould et al 1999a).

Figure 2 Relationship between gene flow (Nm) and distance in B. oleracea populations.

Do viruses affect fitness and control population dynamics in wild Brassicas?

Several viruses are common in populations of B. oleracea in Dorset. In 1996 and 1999, we found plants infected with beet western yellows luteovirus (BWYV), cauliflower mosaic caulimovirus (CaMV), turnip mosaic potyvirus (TuMV) and turnip yellow mosaic tymovirus (TYMV) (Raybould et al. 1999b and D. Pallett, in prep.). Although there were significant differences in the proportion of plants infected among populations within years and between the same population in different years, all populations contained all viruses, apart from one population in 1996 that had no plants infected with TYMV. Many plants had two or more different viruses. In 1999 we tested for another three viruses: turnip crinkle carmovirus (TCV), turnip rosette sobemovirus (TRosV) and cucumber mosaic cucumovirus (CMV). These viruses were not found in any population.

We tested populations of B. nigra for all of the above viruses in 1998 and 1999 (apart from CMV in 1998). In general, a lower proportion of plants was infected compared with B. oleracea. Only TRosV was found in all populations in both years, and TCV was found in all populations, apart from one population in 1998, from which only three plants were tested. Some viruses (e.g., TuMV) were absent from B. nigra populations even though sympatric B. oleracea populations had a large proportion of infected plants.

We have carried out common garden experiments to test the effects of single viruses on plant performance. B .oleracea plants inoculated with TYMV had lower survival and lower dry weight and produced fewer seeds than controls inoculated with water. B. oleracea with TuMV also had lower survival than controls and produced fewer seed, however dry weight was not significantly different (Maskell et al. 1999). We detected no significant differences between B. oleracea plants inoculated with CaMV and controls (M. Alexander, in prep.), although in this experiment plants were in pots rather than planted in a fallow field.

We are currently assessing the effects of viruses in B. nigra (M. Thurston, in prep.). Plants inoculated with TuMV or CaMV almost always die before they flower and set seed (Figure 3). TCV and TRosV produce similar symptoms in B. nigra, with TRosV usually being more severe than TCV. Plants tend to have crinkled leaves and very short stems. In the glasshouse, the viruses are not usually fatal, but early results suggest that these viruses kill very young seedlings in the field (see below). Plants inoculated with TYMV can develop severe yellow patches that coalesce to turn whole leaves yellow (Figure 3) and we are currently assessing the effect of this virus on plant fitness.

We are studying seedling cohorts of B. nigra in the field at several sites to determine when they become infected with viruses and whether infected plants are less likely to survive. At "Time 1" we count the number of seedlings of a particular age in a 20-cm x 20-cm "demography quadrat," and every two weeks we count the number of seedlings of the cohort that are still alive. In a neighbouring "test quadrat," we sample seedlings of the same age to test for the presence of viruses and assume that equal proportions of seedlings are infected in the test and demography quadrats. Unfortunately, we cannot test the plants in the demography quadrat because sampling is destructive when seedlings are very small.

Figure 4 shows typical results from a quadrat at in Dorset, on the south coast of England. A flush of seedlings was produced in mid-September (Time 1), some of which became infected with TCV and TRosV, probably by mechanical inoculation into abrasions. Over the next four weeks, about three-quarters of the seedlings died, including all of those infected with TCV and TrosV, and similar patterns were found in subsequent cohorts (M. Alexander & D. Pallett, in prep.). Therefore, it appears that virus infection increased the likelihood of seedling death, although of course this makes the big assumption that there is a uniform spatial distribution of virus infection.

Figure 3 Effects of TuMV (above) and TYMV (below) on Brassica nigra.

Figure 4 Changes in the number of plants in a seedling cohort of Brassica nigra in a 20-cm x 20-cm quadrat in consecutive two-week periods. The number of seedlings infected with viruses is inferred from a neighbouring quadrat (see text).

Conclusions

Although we do not have a complete picture for any single wild relative_virus resistance gene combination, we are in a position to make tentative predictions. The transfer of transgenes for virus resistance from rape to its wild relatives will be a sporadic process, probably limited to isolated plants or populations adjacent to rape fields. If the transgene becomes established in a fertile hybrid, there is sufficient movement of pollen and/or seeds to transfer it to other populations within a few kilometres of the initial hybridization event. The fitness advantage of a transgene will be strongly dependent on the virus and the crop. A resistance gene to TuMV might have a higher selective advantage in B. nigra (in which the virus has killed all plants tested so far) than in B. oleracea (in which at least some genotypes are tolerant of the virus and can reproduce). The effect of TuMV on B. nigra also indicates that the failure to detect a virus does not mean that a gene for resistance to that virus will have no impact on populations of the wild relative. In summary, we can suggest that it is possible for virus resistance transgenes to enter and spread within and among populations of wild Brassica species. However, we cannot yet predict whether the genes will change the population dynamics of wild Brassicas because we do not yet know the role of density-dependent mortality in determining population size.

Acknowledgements

We thank DETR, MAFF and NERC for financial support.

References

Cooper, J.I. & Raybould, A.F., 1997. Transgenes for stress tolerance: consequences for weed evolution. Proceedings of the Brighton Crop Protection Conference - Weeds, 265-272.

Falk, B.W. & Bruening, G., 1994. Will transgenic crops generate new viruses and new diseases? Science 263, 1395-1396.

Jørgensen, R.B. & Andersen, B., 1994. Spontaneous hybridization between oilseed rape (Brassica napus) and weedy B. campestris (Brassicaceae): a risk of growing genetically modified oilseed rape. American Journal of Botany 81, 1620-1626.

Maskell, L.C., Raybould, A.F., Cooper, J.I., Edwards, M-L. & Gray, A.J., 1999. Effects of turnip mosaic virus and turnip yellow mosaic virus on the survival, growth and reproduction of wild cabbage (Brassica oleracea). Annals of Applied Biology 135, 401-407.

Raybould, A.F., Mogg, R.J. Clarke, R.T., Gliddon, C.J. & Gray, A.J., 1999a. Variation and population structure at microsatellite and isozyme loci in wild cabbage (Brassica oleracea L.) in Dorset (UK). Genetic Resources and Crop Evolution 46, 351-360.

Raybould, A.F., Maskell, L.C., Edwards, M-L., Cooper, J.I. & Gray, A.J., 1999b. The prevalence and spatial distribution of virus infections in natural populations of Brassica oleracea. New Phytologist 141, 265-275.

Scheffler, J.A. & Dale, P.J., 1994. Opportunities for gene transfer from trangenic oilseed rape (Brassica napus) to related species. Transgenic Research 3, 263-278.

Scott, S.E. & Wilkinson, M.J., 1998. Transgene risk is low. Nature 393, 320.

Wilkinson, M.J., Davenport, I.J., Charters, Y.M., Jones, A.E., Allainguillaume, J., Butler, H. T., Mason, D.C. & Raybould, A.F. A direct regional scale estimate of transgene movement from GM oilseed rape to its wild progenitors. Molecular Ecology (in press).