Mitigation Using a Tandem Construct Containing a Selectively Unfit Gene

Plant Biotechnology Journal (2005) 3, pp. 000–000 doi: 10.1111/j.1467-7652.2005.00153.x

MitigationHaniBlackwellOxford,PBIPlant1467-7644©?2Original 2005 2005 BiotechnologyAl-Ahmad UKArticleBlackwell Publishing, of transgene Publishing and Journal Ltd. Jonathan Ltdestablishment Gressel in hybrid progeny using a tandem construct containing a selectively unfit gene precludes establishment of Brassica napus transgenes in hybrids and backcrosses with weedy Brassica rapa

Hani Al-Ahmad and Jonathan Gressel*

Plant Sciences, Weizmann Institute of Science, Rehovot 76100, Israel

Received 30 March 2005; Summary revised 9 June 2005; Transgenic oilseed rape (Brassica napus) plants can interbreed with nearby weedy Brassica accepted 18 June 2005. rapa, potentially enhancing the weediness and/or invasiveness of subsequent hybrid *Correspondence (fax +972 (8)934-4181; e-mail [email protected]) offspring. We have previously demonstrated that transgenic mitigation effectively reduces the fitness of the transgenic dwarf and herbicide-resistant B. napus volunteers. We now report the efficacy of such a tandem construct, including a primary herbicide-resistant gene and a dwarfing mitigator gene, to preclude the risks of gene establishment in the related weed B. rapa and its backcross progeny. The transgenically mitigated and non-transgenic × B. rapa B. napus interspecific hybrids and the backcrosses (BC1) with B. rapa were grown alone and in competition with B. rapa weed. The reproductive fitness of hybrid offspring progressively decreased with increased B. rapa genes in the offspring, illustrating the efficacy

of the concept. The fitness of F2 interspecific non-transgenic hybrids was between 50% and

80% of the competing weedy B. rapa, whereas the fitness of the comparable T2 interspecific

transgenic hybrids was never more than 2%. The reproductive fitness of the transgenic T2 BC mixed with B. rapa was further severely suppressed to 0.9% of that of the competing Keywords: dwarfism, ecological 1 competition, gene flow, interspecific weed due to dwarfism. Clearly, the mitigation technology works efficiently in a rapeseed hybridization, oilseed rape, transgenic crop–weed system under biocontainment-controlled environments, but field studies should mitigation. further validate its utility for minimizing the risks of gene flow.

and the backcross progeny with B. rapa have varying Introduction numbers of C chromosomes (AA + 0–9C, 2n = 20–29) (U, Outcrossing of oilseed rape (Brassica napus = canola) to 1935). In contrast, gene flow between B. napus and the weedy relatives is problematic because oilseed rape is one of genetically distant wild relative Raphanus raphanistrum the few crops that may have interfertile weeds occurring (wild radish, RrRr, 2n = 18) is extremely rare (Chevre et al., within the cultivated crop. Several studies have demonstrated 2000; Rieger et al., 2001; Warwick et al., 2003), and gene the field transfer of genes between B. napus and the genet- introgression into stabilized backcrosses could not be ically close weedy B. rapa (= B. campestris = wild turnip) achieved (Chevre et al., 1998). (Jorgensen and Andersen, 1994; Brown and Brown, 1996; Many factors influence the success of spontaneous Mikkelsen et al., 1996a; Halfhill et al., 2002; Warwick et al., hybridization between B. napus and related weeds under 2003; Hall et al., 2005). B. napus is an allotetraploid (AACC, field conditions, including the physical and genetic distances 2n = 38) with the AA genome from B. rapa (2n = 20) and the between the parents, synchrony of flowering, whether pollen

CC genome from B. oleracea (2n = 18) (U, 1935). F1 hybrids dispersal is mainly by wind or by insect activity, direction of between B. napus and B. rapa are triploid (AAC, 2n = 29), the cross and environmental conditions (Scheffler and Dale,

2 Hani Al-Ahmad and Jonathan Gressel

1994). Genetic studies of B. napus-specific allozyme alleles, Transgene integration and expression in the B. rapa chromosome counts, isozymes, randomly amplified × TM B. napus interspecific hybrids and their polymorphic DNA (RAPD) and amplified fragment length recurrent backcrosses with weedy B. rapa polymorphism (AFLP) markers have shown that genes of volunteer and feral B. napus can stably introgress in The integration and expression of the tandem ahasR (aceto- weedy B. rapa in a few generations (Jorgensen and Andersen, hydroxy acid synthase) and ∆gai (gibberellic acid-insensitive)

1994; Mikkelsen et al., 1996b; Hansen et al., 2001, 2003). inserts in the dwarf TM T1 and T2 interspecific hybrids and BC1

Although F1 hybrids and most of the first backcross gener- progeny was confirmed by polymerase chain reaction

ation (BC1) with B. rapa have a reduced productivity (Lee (PCR) (Figure S2, available as Supplementary material), AHAS and Namai, 1992), some B. rapa-like plants with high enzyme assay (Figure S3, available as Supplementary material) fitness have been found in the first hybrid and backcross and seed selection on imazapyr and/or kanamycin agar generations (Mikkelsen et al., 1996a; Hauser et al., 1998b, medium, as described in Al-Ahmad et al. (2005a). The pattern 2003). and magnitude of resistance of the interspecific hybrids and

We have previously demonstrated the effectiveness of BC1 progenies (Figure S3, available as Supplementary material) transgenic mitigation (TM) technology in tobacco (Nicotiana were similar to those of the TM B. napus plants (Figure S2a,b tabacum) as a model (Al-Ahmad et al., 2004, 2005b). In in Al-Ahmad et al. 2005a). B. napus, we demonstrated the effectiveness of the transgenic mitigation system against the crop as a volunteer Productivity of the B. rapa × TM B. napus weed (Al-Ahmad et al., 2005a) using a dwarfing gene that interspecific hybrids and their backcross progeny increased the yield of the crop when grown without com- grown in the glasshouse at wide spacing without petition from other varieties or species. These results strongly competition suggested that the Brassica crop–weed pair should be tested to ascertain how well the tandem construct will mitigate The productivity (growth without competition) of hybrids and transgene establishment in crop–weed hybrid offspring. The backcross progeny was first tested to ascertain how it com- herbicide-resistant and dwarf genes engineered in tandem pared with that of the wild-type weedy B. rapa plants (from in the same TM construct (Al-Ahmad et al., 2004) were a natural population in Quebec, Canada). Plants of the above manually crossed from TM B. napus into the closely related genotypes, as well as B. napus, were grown separately in 1- B. rapa weed. The utility of mitigation technology in preclud- L pots in the glasshouse at wide spacing between the plants. ing transgene introgression and establishment in B. rapa × B. rapa plants were taller (P ≤ 0.01) because they bolted ≤ B. napus hybrids and their recurrent backcrosses (BC1) to earlier, formed more leaves (P 0.01), grew faster and B. rapa was tested in a biocontainment screen-house thus completed their life cycle earlier than the non-transgenic ecological competition experiment, as reported below. B. napus crop and the non-transgenic interspecific hybrid (B. rapa × B. napus ) plants (Figure S4 and Table S1, available as Supplementary material). The F interspecific hybrids Results and discussion 1 grew faster and were taller than the crop (P ≤ 0.01), and ≤ Non-transgenic and hemizygous TM T1 lines J7, J9 and J16 showed earlier leaf senescence than the crop (P 0.01; B. napus cv. Westar (Al-Ahmad et al., 2005a) were manually Figure S4b, available as Supplementary material). The yield crossed as pollen donors with individual non-transgenic per plant and harvest index of B. rapa and the non-transgenic

weedy B. rapa plants. This is the main direction of crossing F1 and F2 interspecific hybrids were significantly lower than expected in the field, as B. napus is mostly self-fertile with those of the non-transgenic B. napus (P ≤ 0.01), demon- about 70%−80% of the seed normally arising from self- strating the results of domestication (Table S1, available as pollination (Becker et al., 1992), whereas B. rapa is highly Supplementary material). In addition, the seed quality of the self-incompatible and relies on cross-pollination, mainly by triploid hybrids was decreased, with 60% fully developed insects and wind. Thus, reciprocal hybrids were not made in seeds, compared with > 90% fully developed seeds for both this study. The TM and non-transgenic interspecific hybrids the non-transgenic crop and weed parents (Table S1, available were selfed, as well as used as pollen donors in backcrosses as Supplementary material). × (BC1) to B. rapa. The phenotypic appearance of the hybrid The B. rapa TM T1 B. napus line J9 transgenic inter- and backcross progeny is described in the Supplementary specific hybrids were dwarfed at the rosette stage (P ≤ 0.01; material available online (Figure S1). Figure 1a), with shorter internodes (P ≤ 0.01; Figure 1a) and

darker and had more leaves (P ≤ 0.01; Figure S1a, available as Supplementary material), which had shorter petioles (P ≤ 0.01; Figure S5a, available as Supplementary material). There was no significant difference in stem thickness compared with the non-transgenic interspecific hybrid plants (Figure 1a). Growth at the rosette stage was extended, delay- ing flowering (P ≤ 0.01) in most of the tested TM interspecific dwarf hybrid plants (Figure S5b, available as Supplementary material), while all the non-transgenic interspecific hybrid plants grew synchronously until ripening. The backcross progeny of the interspecific TM hybrids to × × B. rapa [B. rapa (B. rapa TM T1 B. napus line J9 )] were dwarfed at the rosette stage (P ≤ 0.01) as expected,

with shorter internodes than the non-transgenic F1 BC1

plants (P = 0.01; Figure 1b). The TM T1 BC1 plants had darker leaves and thicker stems (P ≤ 0.01; Figure 1b), and growth and flower initiation were significantly delayed compared with the non-transgenic backcrosses cultivated separately (Figure S5c, available as Supplementary material). Similar phenotypes were also observed amongst the progeny of TM × T1 BC2 (B. rapa TM T1 BC1 ) plants.

All the non-transgenic F1 BC1 control plants (n = 25), measured in two independent experiments, grew synchronously in a uniform growth pattern through development and ripening (stage 5; Figure S5c, available as Supplementary material).

However, 30% of the TM BC1 plants (n = 23 in two independent experiments) were stunted at the rosette stage with severely compact canopies, and only one-half of the trans-

genic BC1 plants completed their life cycle (Figure S5c, available as Supplementary material). This variation between TM

transgenic BC1 plants could be a result of the variation in chromosome number, with one or more copies of the dwarfing ∆gai gene in the different plants. These experiments sug-

gest that the fitness of the TM BC1 plants vs. non-transgenic populations should be low in a competitive environment, and therefore transgene persistence and spread would be limited. The height of both the non-transgenic and bolted TM Figure 1 Transgenically mitigated (TM), dwarfed (Brassica rapa × TM interspecific hybrid, as well as the BC1 genotypes, increased J9 B. napus ), T1 interspecific hybrids (a) and their backcross to B. rapa rapidly when flowering started (stage 4). Thus, the difference (T1 BC1) (b) had shorter internodes and delayed leaf senescence compared with the non-transgenics (NT) when grown without competition in the in the shoot dry biomass between the non-transgenic and glasshouse. The growth parameters were measured at intervals on TM hybrids and BC1 genotypes was insignificant by harvest independent non-transgenic plants and on preselected TM T hemizygous 1 time (Tables S2b and S2c, available as Supplementary interspecific hybrid genotypes. The points are the mean (± SE) of 13–15 interspecific hybrids and 10–23 BC1 plants. No error bars are shown material). when the standard errors are too small to be seen. Different letters within × The weed crop F1 interspecific hybrids and BC1 progeny ≤ ≤ a panel indicate significantly different values at *P 0.05 or **P 0.01 had about 50% pollen viability, whether transgenic or not, (least significant difference test). compared with > 98% for crop and weed species (Table S2, available as Supplementary material), as previously found by other groups (Jorgensen and Andersen, 1994; Pertl et al., 2002), probably due to the aneuploid genome (Lu and Kato,

2001). Aneuploidy may also account for the infertility in cohorts had a much higher seed yield than the non-transgenic the crop–weed hybrids. The reproductive productivity of TM B. napus plants grown at similar planting distances.

and non-transgenic F1 interspecific hybrids with B. rapa The weedy B. rapa plants cultivated alone at close spacing ≤ and the BC1 progeny grown alone in pots at wide spacing grew faster and taller than TM B. napus (P 0.05), the T2 without competition was lower than that of the TM and interspecific hybrids (B. rapa × TM B. napus ) (P ≤ 0.01) × non-transgenic B. napus plants (Table S2, available as and the backcross T2 BC1 (B. rapa TM T1 hybrids ) Supplementary material). The hemizygous TM and non- (P ≤ 0.01) plants cultivated alone at similar spacing transgenic crops formed more than 25-fold more seeds (Figure S6a, available as Supplementary material). B. rapa per silique (P ≤ 0.01) and per plant (P ≤ 0.01), yielded more plants grown alone at 3- and 6-cm spacing formed more dry weight per plant (P ≤ 0.01) and thus had a higher harvest leaves than non-transgenic B. napus grown at similar spacing index (P ≤ 0.01) than both TM and non-transgenic inter- (P ≤ 0.05) and TM B. napus grown at 6 cm (P ≤ 0.05;

specific hybrids with B. rapa and BC1 progeny. There were no Figure S6b, available as Supplementary material). In addition, significant differences between the TM and non-transgenic B. rapa plants grown alone formed more branches (P ≤ 0.01)

BC1 plants measured by most productivity criteria (Tables S2b than all the other TM and non-transgenic genotypes grown and S2c, available as Supplementary material). There were also alone at both spacings (Table S3, available as Supplementary

no differences between the F1 interspecific hybrid plants grown material), but had 45% and 23% less dry biomass than separately in pots at wide spacing. The only exception was the non-transgenic hybrid and backcross plants, respec- the higher number of seeds formed per non-transgenic inter- tively, when each genotype was grown separately (P ≤ 0.05; specific hybrid plant (P ≤ 0.05; Tables S2b and S2c, available Figure 2c,e). as Supplementary material). This could have been caused by The survival of self-competing B. rapa cultivated at 3-cm a slight yield drag exerted by the herbicide resistance gene spacing was 7% less than that of non-transgenic B. napus (Al-Ahmad et al., 2005b). grown alone at similar spacing (P ≤ 0.05). However, the weed self-thinning was less than that of TM and non-transgenic interspecific hybrids and BC plants cultivated separately at Efficacy of the TM system in preventing transgene flow 1 3-cm spacing (P ≤ 0.05; Figure S6d, available as Supplementary and establishment in hybrid offspring material). Mitigation of transgene flow is measured in competition at As expected, B. rapa plants grown alone at close spacing close planting distances, typical of weed seed rain in agroeco- began to flower and to set seed earlier than TM B. napus ≤ systems, where typically the hybrids and backcrosses make (P 0.05) and T1 interspecific hybrids and BC1 plants up a very small proportion of the population. Mutual inter- (P ≤ 0.05). Although male fertility was high (99% viable ference due to relative fitness can only be described ade- pollen), the weedy B. rapa plants set approximately 70% quately if plant species are grown in mixtures and compared less (P ≤ 0.05, Figure 3a,b) and approximately 50% smaller with their monocultures (Haldane, 1960; De Wit and Van (P ≤ 0.05) seeds, and had approximately 67% lower harvest Den Bergh, 1965). To accentuate the utility of the system, index (P ≤ 0.05), than the B. napus plants (Table S4, available 50% of the competing plants were hybrids or backcrosses as Supplementary material). This may be due to a native lack competing with the weedy parent, a much higher percent- of productivity or, more probably, to an insufficient cross- age than expected in nature. pollination of the outcrossing B. rapa.

Productivity of TM and non-transgenic Brassica Productivity of the T2 interspecific hybrids (B. rapa × genotypes cultivated separately at close spacing in the TM B. napus ) and the backcross T2 BC1 (B. rapa × biocontainment screen-house T1 hybrids )

We first measured the productivity of the parents at close The TM T2 interspecific hybrids and their backcrosses to

planting distances. The productivity of the parent non- B. rapa (BC1) grown alone at 3-cm spacing were shorter ≤ transgenic and TM B. napus plants grown alone has been (P 0.05) than the non-transgenic F2 interspecific hybrid and

previously described (Al-Ahmad et al., 2005a). There BC1 plants grown under similar conditions. However, there were (Figure 5 and Table S5 in Al-Ahmad et al., 2005a), it was no significant differences in the number of leaves or branching observed that the homozygous dwarf TM transgenic formed per plant, or in the shoot fresh biomass measured B. napus plants grown alone at 3- and 6-cm spacing between at various times, between the self-competing TM and the

Figure 2 Vegetative growth suppression of unfit transgenically mitigated (TM) Brassica genotypes when in competition with weedy B. Figure 3 Strong suppression of the reproductive fitness of unfit rapa. Br, B. rapa; Bn, B. napus; F2 (non-transgenic) and T2 (TM transgenic) transgenically mitigated (TM) Brassica genotypes when competing with selfed interspecific (B. rapa × B. napus ) hybrids; BC , selfed 1 weedy B. rapa. Br, B. rapa; Bn, B. napus; F2 (non-transgenic) and T2 (TM backcrosses (B. rapa × F or T interspecific hybrids ). The plants were × 1 1 transgenic) selfed interspecific (B. rapa B. napus ) hybrids; BC1, grown 3 cm from each other in the screen-house, without using × selfed backcrosses (B. rapa F1 or T1 interspecific hybrids ). The plants herbicide. No error bars are shown when the standard errors are too small were grown and the data were analysed as in Figure 2. to be seen. Different letters within a panel indicate significantly different values at P ≤ 0.05 (least significant difference test). Arrows within the right panels highlight the progressively reduced vegetative fitness of the ∆ TM transgenic crop–weed hybrid progeny and the recurrent backcross Supplementary material). In conclusion, the mitigator gai gene into the weed. had vegetative and reproductive advantages for the TM crop grown alone (Al-Ahmad et al., 2005a), but had no vegetative non-transgenic interspecific hybrid and BC1 genotypes advantage and conferred strong negative reproductive (Figure S6a–c and Table S3, available as Supplementary effects on both the TM hybrid and backcross progeny when material). each was grown alone at close spacing. This disadvantage The transgenic (B. rapa × TM B. napus ) interspecific was even greater when transgenic progeny were grown in hybrid plants had 36% less pollen viability than the non- competition with weedy B. rapa, as described below. transgenic interspecific hybrids (P = 0.05; Table S4, available as Supplementary material). The TM interspecific hybrids grown Progressively lower competitive fitness of TM hybrid alone set 65% less seed per plant (P = 0.01; Table S4, available and backcross progeny grown in competition with as Supplementary material) and per unit area (P = 0.01; weedy B. rapa at close spacing Figure 3c,d) than the comparable non-transgenic interspecific hybrid plants. The fertility and yield of the BC1 plants were The competitive fitness of TM B. napus vs. non-transgenic much lower than those of the other Brassica genotypes. B. napus has previously been described in a companion paper

The TM BC1 plants had 39% less viable pollen than the (Al-Ahmad et al., 2005a). There (Table 1 and Figure 4), it (B. rapa × TM B. napus) interspecific hybrids, and > 60% less was observed that the dwarf TM B. napus plants were than the B. napus genotypes and the weedy B. rapa cultivated exceedingly unfit when grown in competition with non- alone under similar conditions (P = 0.05; Table S4, available as transgenic tall B. napus cohorts.

available as Supplementary material). The surviving TM T Fitness of TM B. napus competing with weedy B. rapa 2 hybrid plants were highly unfit when in competition with weedy When B. napus plants were grown in competition with B. rapa at close 3-cm spacing. They were shorter (P ≤ 0.01), weedy B. rapa, 1.5% of the non-transgenic crop and 13% formed fewer leaves (P ≤ 0.01) and had less fresh shoot biomass of the weedy plants (P ≤ 0.01) died, and 6% of the homo- (P ≤ 0.01) than the competing weed (Figure S6a–c, available zygous TM B. napus vs. 6.4% of the weed (not significant) as Supplementary material). At harvest, the non-transgenic

died (Figure S6d, available as Supplementary material). F2 hybrid progeny grown alone or mixed with the competing The non-transgenic B. napus and the competing weedy weed accumulated similar dry shoot biomass (not significant;

B. rapa plants grew to about the same height, had the same Figure 2c). Conversely, the TM T2 hybrids competing with the number of leaves per plant, the same shoot fresh biomass weed accumulated 58% less dry biomass than when grown and the same number of branches (Figure S6a–c and alone at similar spacing (P ≤ 0.01; Figure 2d).

Table S3, available as Supplementary material). In contrast, The TM T2 hybrids competing with B. rapa plants were TM B. napus plants competing with B. rapa were shorter grossly less reproductive: the transgenic hybrids produced (P ≤ 0.01), formed fewer leaves (P ≤ 0.01), had less shoot 90% less seed per plant (P ≤ 0.01), only 5% the yield per unit biomass (P ≤ 0.01) and had fewer branches (P ≤ 0.05) than area (P ≤ 0.01; Figure 3d) and 18% the harvest index ≤ the weedy plants (Figure 2b; also Figure S6a–c and Table S3, (P 0.01) of parallel TM T2 hybrid plants grown by them- available as Supplementary material). selves without competition (Table S4, available as Supple-

The pollen viability of non-transgenic and TM B. napus mentary material). The TM T2 hybrids co-cultivated with genotypes in competition with weedy B. rapa was high B. rapa were also significantly less reproductive (P ≤ 0.05)

(> 99%), just as when the crop genotypes were cultivated than the non-transgenic F2 hybrids grown in competition with

alone (Table S4, available as Supplementary material). Non- B. rapa. The competing TM T2 hybrids produced 96% less transgenic B. napus competing with the weed had a 21% seed number and seed weight per plant, only 5% the yield higher harvest index (P ≤ 0.05) than when B. napus plants per unit area (Figure 3c,d) and 13% the harvest index of

were grown alone (Table S4, available as Supplementary the non-transgenic F2 hybrid progeny competing with B. rapa material). This is probably because of the way in which the (Table S4, available as Supplementary material).

weed was crowded out (Figure S6d, available as Supplemen- The reproductive fitness of the TM T2 hybrids was much tary material) and left more space between B. napus plants, lower than that of the competing weedy B. rapa, as expected in contrast with the dense B. napus plants when grown alone (P ≤ 0.05; Table 1). The relative reproductive fitness of the at 3-cm spacing, which had a high survival (Figure S6d, avail- TM hybrids was always 98% less than the weedy B. rapa. The

able as Supplementary material), but low productivity under non-transgenic F2 hybrids had between 50% and 80% rela- intense monoculture stands. Conversely, the competition tive fitness of the competing weed, depending on the repro- from B. rapa reduced the yield of the TM crop, which had ductive parameter measured (Table 1). Thus, most of the loss 40% less seed per plant (P ≤ 0.05) (measured as seed of fitness was not a result of hybrid incompatibility, but of number and seed dry weight; Table S4, available as Supple- the mitigation gene. Indeed, when non-transgenic B. rapa × ≤ mentary material), 35% less shoot dry biomass (P 0.05; B. napus F1 hybrid offspring were competed with their non- Figure 2b) and 56% less yield per unit area (P ≤ 0.05; transgenic weedy parent at different densities under field Figure 3b) than the TM crop grown alone at 3-cm spacing. conditions, the hybrids were significantly more fit than These data demonstrate the mitigation effect observed B. rapa (Hauser et al., 1998b, 2003), producing, on average, amongst dwarfed TM B. napus plants, which were less fewer seeds per fruit but many more fruits per plant, so that competitive against the weed than the taller, fitter, non- the total number of seeds produced per hybrid plant was transgenic B. napus (Figures 2a,b and 3a,b). 2.5-fold higher than that of the weedy B. rapa (Hauser

et al., 1998b). Conversely, F2 interspecific non-transgenic B. rapa × B. napus hybrids were, on average, less fit than Lowest fitness of (B. rapa × TM B. napus ) T 2 their weedy parent under field conditions, but some of the interspecific hybrids competing with weedy B. rapa individuals were as fit as their parents (Hauser et al., 1998a). × None of the non-transgenic B. rapa B. napus F2 hybrid plants Such introgression was also observed by Hansen et al. (2001) died as a result of competition with B. rapa. In contrast, 25% under natural conditions. Thus, the introgression of B. napus × of the far weaker B. rapa TM B. napus T2 hybrids were lost alleles into weedy B. rapa, interspecific hybrids and backcrosses when competing with the weed (P ≤ 0.01; Figure S6d, would be slowed, but not stopped.

Table 1 Very low relative fitness* of transgenically mitigated (TM) Brassica napus × B. rapa hybrids and backcrosses grown in competition with weedy B. rapa in the biocontainment screen-house

Ratio relative to B. rapa

Shoot dry No. of seeds Weight of Weight of seeds Genotype competing with B. rapa Spacing (cm) biomass per plant seeds per plant per unit area

Non-transgenic (NT) B. napus 3 1.15 2.35 4.32 4.55 TM B. napus 3 0.86 1.55 2.81 3.40 × F2 hybrids (B. rapa NT B. napus )3 1.32 0.53 0.86 0.69 × T2 hybrids (B. rapa TM B. napus )3 0.31 0.02 0.02 0.02 × F2 BC1 [B. rapa (F1 NT hybrids )] 3 0.85 0.17 0.21 0.13 × T2 BC1 [B. rapa (T1 TM hybrids )] 3 0.12 < 0.01 < 0.01 < 0.01

*Calculated from data in Figures 2 and 3 and Table S4 in the Supplementary material as the ratio of the data for each parameter of the genotype divided by the data obtained for the competing weedy B. rapa.

very low reproductive fitness: their seed set was much less Lowest fitness of the backcross T BC (B. rapa × T 2 1 1 than that of the competing B. rapa (P ≤ 0.01). The TM T BC hybrids ) competing with weedy B. rapa 2 1 plants had 1% the seed number and weight per plant Experiments were performed to ascertain whether the con- (P ≤ 0.01), and per unit area (P ≤ 0.01; Figure 3f), with a tinuation of backcrossing of weed × TM crop hybrids to the much lower harvest index (P ≤ 0.01; Table S4, available as weed would improve the fitness of weedy plants bearing the Supplementary material). They also had a minuscule 0.9% TM construct, and thus facilitate their establishment and reproductive fitness (based on seed set) and 12% vegetative persistence within an agroecosystem. The opposite was fitness (based on shoot dry biomass) relative to the weed found, as described below. (Table 1).

The vegetative fitness of both F2 non-transgenic and TM The reproductive fitness of the TM T2 BC1 plants mixed

T2 backcrosses to B. rapa plants (BC1) was less than the com- with B. rapa was significantly less than that of the non- peting weedy B. rapa. Both F2 and T2 BC1 genotypes were transgenic F2 BC1 plants competing with B. rapa, as well as ≤ ≤ shorter (P 0.01), formed fewer leaves (P 0.01) and had that of TM T2 BC1 plants grown alone. The dwarf TM T2 BC1 less fresh shoot biomass (P ≤ 0.01) than the competing weed plants competing with the weed had > 85% fewer seeds and (Figure S6a–c, available as Supplementary material). The veg- yield per plant (P ≤ 0.05), and yield per unit area (Figure 3f), etative growth of the TM T2 BC1 plants grown in competition than TM T2 BC1 plants grown alone (Table S4, available as with B. rapa was significantly less than that of TM T2 BC1 Supplementary material). In addition, TM T2 BC1 plants plants grown alone (P ≤ 0.05; Figure S6a–c, available as Sup- grown in a mixture with B. rapa had > 90% less yield than the ≤ plementary material). At maturity, the TM T2 BC1 plants had non-transgenic F2 BC1 plants mixed with B. rapa (P 0.05; 70% fewer branches (P ≤ 0.01; Table S3, available as Supple- Figure 3e,f; also Table S4, available as Supplementary material). mentary material) and accumulated 88% less dry biomass than the competing B. rapa (P ≤ 0.01; Figure 2f), and 63% Concluding remarks less dry biomass than the self-competing TM T2 BC1 plants (P ≤ 0.01; Figure 2f). Conversely, there was no significant The results from this study demonstrate that the mitigation difference between the vegetative productivity of the non- system previously verified to be effective with tobacco as a transgenic F2 BC1 plants grown alone or mixed with B. rapa model (Al-Ahmad et al., 2004, 2005b) is also applicable to for most growth parameters measured. oilseed rape, a crop with a totally different growth pattern. Competition from the weedy B. rapa hardly affected the The mitigation system has also been demonstrated against survival of the TM transgenic T2 BC1 plants compared to those oilseed rape hybrids with the competitive related B. rapa grown alone (Figure S6d, available as Supplementary mater- weed, providing information about the ‘real world’. ial). However, the surviving TM T2 BC1 plants had less than The spacing between plants is typically closer in many one-half the pollen viability of the competing weed (P ≤ 0.05; agroecosystems than in this study, as weeds typically shed Table S4, available as Supplementary material). They had a thousands of seeds in a small area. Competition between

With replacement series experiments using tobacco, we have previously demonstrated that, even with an exceedingly unlikely mixture (in nature) of 90% TM vs. 10% non-transgenics, the TM transgenics would still be eliminated (Al-Ahmad et al., 2005b). Our results with TM tobacco, and now with TM oilseed rape, clearly indicate that the simulations of Hay- good et al. (2003) were based on inaccurate assumptions about both plant progeny number and competition for replacement. Even when 75% of the B. napus plants were transgenic, they were completely out-competed by the minority of fitter (but lower yielding by themselves), taller, non-transgenic B. napus plants (Figure 3a,b in Al-Ahmad et al. 2005a). In all cases but one, the TM B. napus out- yielded the non-transgenic B. napus plants when cultivated in pure stands, by at least 15% and up to 2.5-fold (Figure 4a); yet, when in competition with non-transgenic B. napus, the TM plants were almost without yield (Figure 4b). Figure 4 Transgenically mitigated (TM) Brassica napus is (a) highly These data are meant to be the basis for field studies with reproductive when cultivated alone, but (b) severely unfit when in competition with tall, non-transgenic (NT) B. napus (meta-summary of the B. napus/B. rapa system to further evaluate the positive various experiments). Data points are calculated as the ratio of seed set implications of the transgenic mitigation strategy. Sufficient per TM plant to that of the non-transgenics from glasshouse (filled seed stocks have been generated during this investigation for symbols) and screen-house (open symbols) experiments at various confirmatory field studies that could include: evaluation of spacing between the plants. The glasshouse plants measured were: () TM and non-transgenics grown in separate 1-L pots at wide spacing TM oilseed rape productivity; synthesis of advanced back- (from Table S2a, available as Supplementary material); (, ) 2 : 1 cross populations and assessment of their competitiveness as hemizygous/homozygous TM segregants vs. non-transgenics (from volunteer weeds in cereal rotations, e.g. with wheat. Addition Table S4 in Al-Ahmad et al. 2005a); (, , ) pure homozygous TM vs. non-transgenics (from Table S4 in Al-Ahmad et al. 2005a). of a second TM trait to the tandem construct, such as anti- The screen-house plants measured were: (, ) pure homozygous TM vs. shattering genes; and stacking of TM traits with other trans- non-transgenics (from Table S4, available as Supplementary material). gene containment mechanisms to further decrease the risks of transgene introgression.

Experimental procedures competitive weeds and the rare, less fit TM hybrid and backcross plants would be even more severe in the field, and the Crossing of TM B. napus transgenes into non- frequency of the remaining TM progeny within plant popu- transgenic weedy B. rapa lations would be very low, unless the selective herbicide was

used. This is clearly shown in Figures 2 and 3 by following the The primary hemizygous TM plants were designated T0, and

arrows within the right panels of both figures. It can be seen subsequent selfed generations followed the pattern T1 (prog-

that the TM B. napus grown without interspecific competi- eny of T0) and T2 (progeny of T1). The dwarf/herbicide resist- tion had a higher yield than the non-transgenics grown in ance TM 1 tandem construct, described in Al-Ahmad et al. parallel. Conversely, under competitive pressure from weedy (2004), was used to transform B. napus L. cv. Westar. Sixteen B. rapa, TM B. napus, its interspecific hybrid progeny and TM transgenic B. napus lines were regenerated (Al-Ahmad their backcross with the weed showed progressively reduced et al., 2005a) and representative independent lines J7, J9 and vegetative and reproductive fitness. This would clearly preclude J16 were used in this study. A weedy accession of B. rapa transgene establishment, as improvement, not deterioration, (#2974) was collected from a naturally occurring population is expected with backcrossing. in Quebec, Canada, and was kindly provided by Dr Suzanne These crop–weed data clearly do not fit the model predic- Warwick, Agriculture and Agri-Food Canada, Eastern Cereal tions of Haygood et al. (2003), who concluded that unfit crop and Oilseed Research Centre, Ottawa, Ontario, Canada. B. genes could ‘swamp’ wild relatives of the crop and slowly rapa (female parent = ) × B. napus (male parent = ) were

establish the unfit genes in the progeny over generations. used to produce the F1 interspecific hybrid generation, and

B. rapa × F interspecific hybrids were crossed to produce the 1 Competition of TM vs. non-transgenic Brassica F BC generation. The weedy B. rapa was used as the female 1 1 genotypes and self-competition of each alone under parent in all crosses to simulate transgene flow under field biocontainment screen-house conditions conditions, as B. napus is self-compatible and more likely to be self-pollinated, whereas B. rapa is self-incompatible and The competition between B. rapa weed and non-transgenic pollination by a massive crop of B. napus is likely. or homozygous TM progeny of each T2 B. napus of line J7, T2

interspecific hybrid of line J9 or T2 BC1 of line J9 was inde- pendently evaluated in a internally replicated experiment in Manual pollination the biocontainment screen-house, as described previously for

Hemizygous TM T1 B. napus plants (lines J7, J9 and J16) were the B. napus experiment (Al-Ahmad et al., 2005a). As a result crossed manually with weedy B. rapa. Maternal flowers were of self-incompatibility of the weedy B. rapa, bumblebees emasculated prior to anthesis, pollinated by rubbing two to (Bombus terrestris L.) were introduced to enhance cross- three male parent anthers on the stigma surface, and the pollination. One beehive was placed close to the containers flowers were then covered with glassine pollination bags during the flowering period. The paternal fertility was (PBS International, Scarborough, UK) for 10 days to prevent determined from the percentage pollen viability. uncontrolled cross-pollination. Surplus flowers on maternal plants were removed. Mature siliques were harvested, and Statistical analyses the fully developed seeds were selected for herbicide resistance on agar medium containing 0.5 µM imazapyr. The resist- The plant growth parameters, productivity and fitness data ant/dwarf hybrid plants were cultivated in the glasshouse, were analysed using the JMP® program (version 4.0.1; SAS and were analysed by PCR for the presence of the ahasR Institute 2000; Cary, NC) by one-way analysis of variance and ∆gai inserts, and by the AHAS enzymatic assay for (ANOVA) and by comparing the least significant differences herbicide resistance, as described previously in Al-Ahmad (LSD). Probability levels were considered to be statistically et al. (2005a). significant at P ≤ 0.05 and highly significant at P ≤ 0.01. Dif- ferences were not considered to be statistically significant at P > 0.05. Backcross of TM T1 interspecific hybrids to the parent weedy B. rapa: B. rapa × (B. rapa × TM B. napus )

Acknowledgements TM T1 interspecific hybrids (lines J7 and J9) ( ) were separately backcrossed to the weedy B. rapa parent () by hand, We thank Alexandra Savitsky, Judith Karmi and Shiri Gerson as described above. The seeds were harvested from the for their excellent technical assistance, Dr Roy Chaleff, Amer- crossed plants, sown in soil in 1-L pots and the AHAS resist- ican Cyanamid Company, for kindly providing plasmid ant/dwarf TM T1 BC1 plants were tested by PCR as above; pAC456 and Drs Donald Richards and Nick Harberd, Depart- they were allowed to self-pollinate and set seed (i.e. produce ment of Molecular Genetics, John Innes Centre, UK, for pro- ∆ T2 BC1 progeny). The T1 BC1 plants ( ) were manually back- viding the gai gene. This research was supported by the crossed again to weedy B. rapa ( ) to obtain the TM T1 BC2 Levin Foundation, by INCO–DC contract no. ERB IC18 CT 98 progeny. Crosses between non-transgenic B. napus () and 0391 and by a bequest from Israel and Diana Safer. B. rapa (), and backcrossing of their hybrid () with B. rapa (), were also conducted as controls. References

Al-Ahmad, H., Galili, S. and Gressel, J. (2004) Tandem constructs to Productivity of the F1 interspecific hybrids and mitigate transgene persistence: tobacco as a model. Mol. Ecol. backcrosses grown without competition 13, 697–710. Al-Ahmad, H., Dwyer, J., Moloney, M. and Gressel, J. (2005a) Miti-

TM and non-transgenic F1 interspecific hybrids and plants gation of establishment of Brassica napus transgenes in volun- backcrossed with weedy B. rapa were grown separately in teers using a tandem construct containing a selectively unfit gene. 13-cm diameter, 1-L pots under the same growth conditions Plant Biotechnol. J. in press. Al-Ahmad, H., Galili, S. and Gressel, J. (2005b) Poor competitive fit- and with the same productivity measurements as described ness of transgenically mitigated tobacco in competition with the previously in Al-Ahmad et al. (2005a) for B. napus grown wild type in a replacement series. Planta (http://dx.doi.org/ alone at wide spacing without competition. 10.1007/s00425-005-1540-6).

Becker, H.C., Damgaard, C. and Karlsson, B. (1992) Environmental Mikkelsen, T.R., Andersen, B. and Jorgensen, R.B. (1996a) The risk variation for outcrossing rate in rapeseed. Theor. Appl. Genet. 84, of crop transgene spread. Nature, 380, 31. 303–306. Mikkelsen, T.R., Jensen, J. and Jorgensen, R.B. (1996b) Inheritance Brown, J. and Brown, A.P. (1996) Gene transfer between canola of oilseed rape (Brassica napus) RAPD markers in a backcross (Brassica napus L. and B. campestris L.) and related weed species. progeny with Brassica campestris. Theor. Appl. Genet. 92, 492– Ann. Appl. Biol. 129, 513–522. 497. Chevre, A.M., Eber, F., Baranger, A., Hureau, G., Barret, P., Picault, H. Pertl, M., Hauser, T.P., Damgaard, C. and Jorgensen, R.B. (2002) and Renard, M. (1998) Characterization of backcross generations Male fitness of oilseed rape (Brassica napus), weedy B. rapa and

obtained under field conditions from oilseed rape–wild radish their F1 hybrids when pollinating B. rapa seeds. Heredity, 89, 212–

F1 interspecific hybrids: an assessment of transgene dispersal. 218. Theor. Appl. Genet. 97, 90–98. Rieger, M.A., Potter, T.D., Preston, C. and Powles, S.B. (2001) Chevre, A.M., Eber, F., Darmency, H., Fleury, A., Picault, H., Letan- Hybridisation between Brassica napus L. & Raphanus raphanis- neur, J.C. and Renard, M. (2000) Assessment of interspecific trum L. under agronomic field conditions. Theor. Appl. Genet. hybridization between transgenic oilseed rape and wild radish 103, 555–560. under normal agronomic conditions. Theor. Appl. Genet. 100, Scheffler, J.A. and Dale, P.J. (1994) Opportunities for gene-transfer 1233–1239. from transgenic oilseed rape (Brassica napus) to related species. De Wit, C.T. and Van Den Bergh, J.P. (1965) Competition between Transgenic Res. 3, 263–278. herbage plants. Neth. J. Agric. Sci. 13, 212–221. U, N. (1935) Genome analysis in Brassica with special reference Haldane, J.B.S. (1960) More precise expressions for the cost of to the experimental formation of B. napus and peculiar mode of natural selection. J. Genet. 57, 351–360. fertilization. Jpn. J. Bot. 7, 389–452. Halfhill, M.D., Millwood, R.J., Raymer, P.L. and Stewart, C.N. (2002) Warwick, S.I., Simard, M.J., Legere, A., Beckie, H.J., Braun, L., Zhu, Bt transgenic canola hybridization with its weedy relative, Brassica B., Mason, P., Seguin-Swartz, G. and Stewart, C.N. (2003) Hybrid- rapa. Environ. Biosafety Res. 1, 19–28. ization between transgenic Brassica napus L. and its wild relatives: Hall, L.M., Rahman, M.H., Gulden, R.H. and Thomas, A.G. (2005) Brassica rapa L., Raphanus raphanistrum L., Sinapis arvensis L. and Volunteer oilseed rape: will herbicide-resistance traits assist feral- Erucastrum gallicum (Willd.) OE Schulz. Theor. Appl. Genet. 107, ity?. In Crop Ferality and Volunteerism (Gressel, J., ed.), pp. 59– 528–539. 79. Boca-Raton, FL: CRC Press. Hansen, L.B., Siegismund, H.R. and Jorgensen, R.B. (2001) Intro- gression between oilseed rape (Brassica napus L.) and its weedy Supplementary material relative B. rapa L. in a natural population. Genet. Resour. Crop The authors have supplied additional data that are available Evol. 48, 621–627. Hansen, L.B., Siegismund, H.R. and Jorgensen, R.B. (2003) Progres- online, including a supplementary results section (phenotypic sive introgression between Brassica napus (oilseed rape) and B. appearance of the B. rapa × TM B. napus interspecific rapa. Heredity, 91, 276–283. hybrids and their backcross progeny grown in the glasshouse at Hauser, T.P., Damgaard, C. and Jorgensen, R.B. (2003) Frequency- wide spacing without competition). dependent fitness of hybrids between oilseed rape (Brassica napus) and weedy B. rapa (Brassicaceae). Am. J. Bot. 90, 571– Figure S1 Morphological similarities of the non-transgenic 578. Brassica rapa × B. napus interspecific hybrids and back- Hauser, T.P., Jorgensen, R.B. and Ostergard, H. (1998a) Fitness of crosses with their parents. (a) Leaves of the F1 interspecific backcross and F2 hybrids between weedy Brassica rapa and hybrid plants resemble those of B. napus, but the F1 BC1 oilseed rape (B. napus). Heredity, 81, 436–443. leaves are similar to those of weedy B. rapa, which has a Hauser, T.P., Shaw, R.G. and Ostergard, H. (1998b) Fitness of F1 hybrids between weedy Brassica rapa and oilseed rape (B. napus). distinguishable small leaf that typically clasps the branches

Heredity, 81, 429–435. after bolting. (b) The flowers of the F1 interspecific hybrid Haygood, R., Ives, A.R. and Andow, D.A. (2003) Consequences plants are similar to those of B. napus in the pattern in which of recurrent gene flow from crops to wild relatives. Proc. R. Soc. they are borne below uppermost still closed buds, while the London Ser. B-Biol. Sci. 270, 1879–1886. Jorgensen, R.B. and Andersen, B. (1994) Spontaneous hybridization flowers of the weed and F1 BC1 are held above the unopened

between oilseed rape (Brassica napus) and weedy B. campestris buds. (c) The F1 interspecific hybrid plants have shorter (Brassicaceae): a risk of growing genetically modified oil seed siliques with fewer viable seeds than their parents. rape. Am. J. Bot. 81, 1620–1626. Figure S2 Detection of the transformed transgenically mit- Lee, K.H. and Namai, H. (1992) Pollen fertility and seed set percentage after backcrossing of sesquidiploids (AAC genomes) derived igated (TM) genes in the interspecific hybrids and back- from interspecific hybrid between Brassica campestris L. (AA) and crosses with Brassica rapa. (a, b) Electrophoresis of 15 µL of B. oleracea L. (CC) and frequency distribution of aneuploids in the polymerase chain reaction (PCR) product through a 1% progenies. Jpn. J. Breed. 42, 43–53. (w/v) agarose gel at 50 V for 45 min. M, molecular marker Lu, C.M. and Kato, M. (2001) Fertilization fitness and relation to (1-kb DNA ladder); NT, non-transgenic; (–), control without chromosome number in interspecific progeny between Brassica napus and B. rapa: a comparative study using natural and resyn- template DNA; P, TM 1 plasmid control; numbers below thesized B. napus. Breed. Sci. 51, 73–81. lanes denote representative independent TM lines. A–D

denote the sites chosen for PCR amplification: A, over the panel indicate significantly different values at P ≤ 0.01 (least right border of T-DNA and the 5′ region of the ahasR pro- significant difference test). moter; B, over the 3′ terminal region of ahasR, the linker, and Figure S5 Short petioles and delay in growth development the 5′ region of ∆gai; D, over the kanamycin resistance gene of transgenically mitigated (TM) hybrid genotypes. (a, b)

(kan) located at the left border of the T-DNA (see Figure S3a Petiole length and time of flower initiation in the F1 hybrids in Al-Ahmad et al., 2005a). grown in the glasshouse. Different letters indicate different Figure S3 Magnitude of imazapyr resistance in trans- values at **P = 0.01 (least significant difference test); n = 15. genically mitigated (TM) Brassica genotypes. The AHAS (c) The percentage of non-transgenic and hemizygous T1 BC1 × (acetohydroxy acid synthase) enzyme activity was assayed (B. rapa TM T1 J9 interspecific hybrids ) plants at each spectrophotometrically in leaf discs as described in Al- growth stage (n = 23). The TM plants were preselected Ahmad et al. (2004), and presented as the percentage for herbicide resistance via the AHAS (acetohydroxy acid activity of controls tested in the absence of the herbicide. synthase) enzymatic assay.

(a) F1 interspecific hybrids. The individuals assayed were Figure S6 Growth suppression and poor survival of unfit non-transgenic (B. rapa × B. napus ) hybrids () and transgenically mitigated (TM) Brassica genotypes when in the preselected hemizygous (B. rapa × TM J9 B. napus ) competition with weedy B. rapa. B. rapa (grey symbols), transgenic progeny (). The enzyme activities of the non- non-transgenic (open symbols) and TM dwarf/herbicide- transgenic and TM controls were 0.044 and 0.057 µmol ace- resistant (filled symbols) plants were planted alone or co- 2 tolactate/cm leaf disc/h, respectively. (b) F1 BC1. The individ- cultivated at 3 and 6 cm from each other in the screen-house × uals assayed were non-transgenic (B. rapa F1 hybrids ) without using herbicide. NT, non-transgenic; hybrid, selfed × backcrosses ( ) and the preselected hemizygous (B. rapa (B. rapa B. napus ) interspecific hybrids; BC1, selfed × × TM J9 hybrids ) transgenic progeny ( ). The enzyme (B. rapa F1 interspecific hybrids ) BC1. The points are activities of the non-transgenic and TM controls were 0.092 the mean (± SE) of n = 4–10. No error bars are shown when and 0.11 µmol acetolactate/cm2 leaf disc/h, respectively. the standard errors are too small to be seen. Different letters × Figure S4 Non-transgenic (Brassica rapa B. napus ) F1 within a panel indicate significantly different values at interspecific hybrids have an intermediate rate of growth *P ≤ 0.05 or **P ≤ 0.01 (least significant difference test). between the slower crop and the faster weedy parent. The Table S1 Vegetative and reproductive productivity of non- three genotypes were grown at the same time without com- transgenic Brassica genotypes grown in the glasshouse with- petition at wide spacing in the glasshouse. (a) The appear- out competition ance of the three genotypes after 2 months of cultivation. Table S2 Vegetative and reproductive productivity of trans-

B. napus was at early rosette stage 2, the F1 interspecific hybrid genically mitigated (TM) Brassica genotypes grown in the was at early stage 3 (bolting) and the B. rapa weed was at glasshouse at wide spacing without competition the middle of flowering stage 4. (b) Vegetative characters of Table S3 Number of branches formed by Brassica plants the three genotypes. The points are the mean (± SE) of measured at maturity in the biocontainment screen-house n = 10. No error bars are shown when the standard errors Table S4 Productivity and fitness of Brassica genotypes are smaller than the data points. Different letters within a grown under biocontainment screen-house conditions