Authors for correspondence: Bao-Rong Lu and Feng Wang Tel: +86 21 65643668 (B-RL) Fax: +86 21 65643668 (B-RL) Email: email@example.com (B-RL); firstname.lastname@example.org (FW)
• Crop-to-crop transgene flow will affect seed purity of non-GM rice varieties, leading to unwanted consequences. To assess the maximum probability of transgene outflow in rice (Oryza sativa), gene flow experiments were conducted with three cultivation patterns with different mixed-planting proportions of adjacent GM and non-GM rice at two sites in Fujian and Hainan Provinces of China.
• Three GM rice lines containing two insect-resistance genes (Bt/CpTI) and their non-GM counterparts were used in the experiments to allow natural hybridization to occur. A hygromycin resistance gene was used as a selective marker for identifying hybrids.
• Based on the examination of > 645 700 geminated seeds, the result showed low frequencies (0.05–0.79%) of transgene flow from GM to non-GM rice at close spacing, although with significant variation among mixed-planting proportions.
• It is concluded that rice transgene flow will occur at a very low frequency (< 1.0%), even if the GM rice is planted at close spacing with non-GM rice, and high densities of GM rice cultivated in the neighborhood of non-GM rice will increase the probability of outcrossing with the non-GM rice.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
With the rapid development of transgenic biotechnology, a number of genetically modified (GM) crops have been produced, many of which have entered commercial markets or been released into environment for field-testing. Undoubtedly, transgenic technology and the resulting GM crops will offer greater opportunities for the global food security, particularly in the developing world (Huang et al., 2002). However, the release and commercialization of GM crops also arouses enormous concerns about biosafety worldwide (Ellstrand, 2001; Snow, 2002). The potential ecological consequences associated with transgene outflow to non-GM crop counterparts (crop-to-crop) and wild/weedy relative species (crop-to-wild) are foremost among these concerns (Ellstrand et al., 1999; Stewart et al., 2003). It is therefore important to assess transgene flow and its potential consequences to guarantee the safe use of transgenic crops. The determination of frequencies of gene flow between GM and non-GM crops, as well as between GM crops and their weedy/wild relatives, is the first step in assessing the potential magnitude of possible consequences of transgene escape. The measurement of gene flow is also necessary for monitoring and managing possible consequences of the large-scale release of GM crops (Ellstrand et al., 1999; Lu & Snow, 2005).
In the case of crop-to-crop gene flow, if transgene movement occurs from a GM crop to its non-GM counterpart in fields, it will affect seed purity of a non-GM crop variety. Consequently, this will most likely alter the deployment of non-GM and GM crop varieties in a region, probably affecting international trade of non-GM crops and influencing local markets if the crops are destined for GM free or ‘organic’ products (Messeguer, 2003). Information about gene flow from a GM crop variety to its non-GM counterpart is therefore valuable to estimate the likelihood of adventitious GM seeds in the non-GM crop. Pollen-mediated gene flow through cross-pollination is a natural process that can significantly cause transgene outflow. Therefore, estimation of the pollen-mediated gene flow will facilitate our assessment of potential consequences, provided that the seed-dispersed gene flow caused by human intervening can be controlled at a certain level. The study of crop-to-crop transgene flow can also provide useful information for crop-to-weed and crop-to-wild transgene flow assessment and help to establish the corresponding control measures (Lu et al., 2003; Lu & Snow, 2005).
Rice is a major world crop to which transgenic biotechnology has been effectively applied for genetic improvement (Tyagi & Mohanty, 2000; Huang et al., 2002; Jia, 2004). Many GM rice varieties have been produced (Huang et al., 2002; Jia & Peng, 2002) and released into the environment for field-testing (Messeguer et al., 2001, 2004; Chen et al., 2004). It is apparent that, as an important world crop, transgenic rice varieties will be released for commercial production on a large scale, probably in the near future (Chen et al., 2004; Jia, 2004). In China, GM rice varieties with insect resistance (for example Bt and CpTI genes), disease (Xa21) resistance, and herbicide tolerance (bar, EPSPs) have been developed, and these products are now in the pipeline for commercialization pending approval by the biosafety regulatory agency (Jia, 2002). Significant progress has also been made with drought- and salt-tolerant GM rice varieties (Jia, 2004). It is therefore an urgent task to accurately measure transgene flow from GM to non-GM rice, particularly for those GM rice varieties/lines that are nearly ready for commercialization, as a part of routine biosafety assessment.
Very few studies on crop-to-crop transgene flow in rice are reported (Messeguer et al., 2001, 2004; Jia, 2002), although there are a number of estimates of outcrossing rates between different rice varieties (Diao et al., 1996; Rong et al., 2004). In general, a low frequency (0–2%) of pollen-mediated gene flow between different rice varieties at various distances has been observed using different methods such as dominant morphological traits and molecular markers (Bashir et al., 2004). However, those experiments did not provide general ideas of maximum probability of gene flow in each case. Maximum transgene flow via pollen will take place when there is a high density of transgenic rice varieties planted adjacent to nontransgenic rice with similar flowering times. The estimate of the ‘zero distance’ crop-to-crop transgene flow can provide the maximum likelihood (extreme cases) of pollen-mediated transgene flow from GM rice varieties to their non-GM counterparts. In China, it is very common for farmers (households) to manage small field plots or small pieces of farmland (< 0.1 hectares). As a consequence, complex mixed-planting of rice varieties either within or between field plots is a widespread practice in many rice-growing areas. Mixed-planting patterns of GM and non-GM rice varieties with various proportions will be expected when GM rice is approved to grow in China. Therefore, the ‘extreme cases’ in which GM and non-GM rice are cultivated adjacently should be taken into consideration. Knowledge of the maximum probability of transgene flow from GM rice to its non-GM counterparts planted at close spacing will address these considerations.
Accurate measurement of frequency of transgene flow can be obtained by using introduced marker gene(s), such as antibiotic or herbicide resistance genes (Paul et al., 1995; Messeguer et al., 2001; Chen et al., 2004). The hygromycin resistance gene is a commonly used marker gene in rice genetic engineering. Plants carrying a hygromycin resistant gene will survive in a selective culture medium containing hygromycin B, which is an aminoglycosidic antibiotic produced by Streptomyces hygroscopicus that kills bacteria, fungi and higher eukaryotic cells by inhibiting protein synthesis (http://www.roche-applied-science.com/pack-insert/0843555a.pdf). It therefore provides a powerful method for detecting transgene flow from GM rice that contains the hygromycin resistance marker gene to non-GM rice varieties on a massive scale, by identifying hybrids between the GM and non-GM rice varieties.
The objectives of this study were (1) to measure the maximum frequencies of transgene flow from GM rice lines containing a hygromycin resistance gene to their nontransgenic counterparts (isogenic lines) planted adjacently to each other and (2) to determine variation of transgene flow frequencies in different GM/non-GM mixed-planting patterns.
Materials and Methods
Three Bt/CpTI rice lines (i.e. KeFeng6 (restorer), IIYouKeFeng6 (hybrid) and 21SKeFeng6 (hybrid)) bred from the same transgenic event (used as the pollen donor) and their non-GM isogenic counterparts (pollen receivers) were used in the transgene flow experiments (Table 1). Bt (Bacillus thuringiesis) is an insect resistance gene under the control of the maize ubiquitin (Ubi) promoter producing Cry1A(c) protein that kills lepidopteran insects, and CpTI (cowpea trypsin inhibitor) is also an insect resistance gene under the control of P-act promoter. These two transgenes were tightly linked with the selectable marker gene hpt (hygromycin resistant gene) under the control of the CaMV 35S promoter. The GM rice line KeFeng6 containing both Bt and CpTI transgenes (Bt/CpTI) was produced using the Agrobacterium-mediated transformation technology and had been bred through selfing and selection of individuals with stable insect resistance beyond the T7 generation. Therefore, KeFeng6 showed a stable genetic inheritance. The IIYouKeFeng6 and 21SKeFeng6 were hybrid rice crossed with KeFeng6 (Table 1).
Table 1. Nongenetically modified (non-GM) rice varieties and GM (Bt/CpTI) rice lines used in the pollen-mediated transgene flow experiments
Non-GM rice variety
GM rice line
Labeled as: MSR–Minghui-86 (a male sterility restore line)
Labeled as: MSR+KeFeng6 (a male sterility restore line Minghui-86 containing transgene Bt/CpTI), homozygous
Labeled as: HY1–IIYouming-86 (a hybrid: male-sterile line II-32 A × Minghui-86)
Labeled as: HY1+IIYouKeFeng6 (a hybrid obtained from male-sterile line II-32 A × KeFeng6), heterozygous
Labeled as: HY2–2186 (a hybrid: male-sterile line SE21S × Minghui-86)
Labeled as: HY2+21SKeFeng6 (a hybrid obtained from male sterile line SE21S × KeFeng6), heterozygous
The GM rice lines and non-GM rice varieties used in the experiments have the same flowering time and duration. MSR– is one of the most widely used rice varieties (MS restorer) in China, and its transgenic counterpart MSR+ is a genetically stable line (homozygous) that has received permission from the Chinese Biosafety Office for release to controlled field-testing. HY1– and HY2– are hybrid rice varieties with MSR– as their paternal parent, and HY1+ and HY2+ were the corresponding GM hybrid rice lines (heterozygous) with MSR+ as their paternal parent (Table 1). All of the experimental materials were provided by Fujian Province Key Laboratory of Genetic Engineering for Agriculture, Fujian Academy of Agricultural Sciences, Fujian Province, China.
Experimental design and sampling strategies
The transgene flow experiments were conducted in the confined GM rice experimental fields in 2003 at two sites in China. One site was in Fuzhou of Fujian Province (26.1° N, 119.3° E) and the other in Sanya of Hainan Province (18.33° N, 109.52° E). To determine differences in gene flow frequencies associated with pollen density variation, three cultivation patterns (experiments) for each of the three GM and non-GM rice pairs with three special proportions of GM rice lines and non-GM rice varieties were used. Consequently, an experiment designed for a three-way anova (three rice pairs × three mixed-planting proportions × two sites) was conducted. Experiments were arranged in 8 × 8 m2 plots. In all the plots, the adjacent GM and non-GM rice individuals were planted with a distance of 20 cm. The detailed layout of the three experiments (A–C) is described in the following text.
Experiment A: GM majority In Experiment A, rice was planted in a unit of nine individuals (see the boxed area in Fig. 1a). In each unit, one non-GM rice seedling was surrounded adjacently by eight GM rice seedlings (8 GM : 1 non-GM). It was the extreme case for crop-to-crop transgene flow. Consequently, 169 units were planted in a plot (Fig. 1a). At seed maturity, c. 40 of the units were sampled separately at random from each plot. About 150–200 seeds from each non-GM individual (the middle one in a unit) of the units were harvested separately for the identification of hybrids.
Experiment B: non-GM majority In Experiment B, rice was also planted in a unit of nine individuals (see the boxed area in Fig. 1b). In each unit, one GM rice seedling was adjacently surrounded by eight non-GM rice seedlings (1 GM : 8 non-GM). As a result, 169 units were planted in a plot (Fig. 1b). At seed maturity, c. 40 of the units were sampled separately at random from each plot. About 150–200 seeds from each of the eight non-GM individuals (the marginal ones in a unit) of the units were harvested separately for the identification of hybrids.
Experiment C: random In Experiment-C, equal numbers of GM and non-GM rice seedlings were mixed before transplanting into the plots. Therefore, the mixed-planting proportion was 1 GM : 1 non-GM, in which the mixed-planting proportion was between Experiment A and Experiment B. At maturity, rice seeds were harvest separately from nine individuals in a sample unit as indicated in Experiments A and B. About 40 such units were sampled randomly from the plot. About 150–200 seeds from each of the individuals of the units were examined for the identification of hybrids. Individuals from each unit with all seeds (homozygous MSR line) or most seeds (heterozygous hybrid lines) resistant to hygromycin were considered as GM rice and discarded from further testing. Only those with few (or no) seeds resistant to hygromycin were considered as non-GM rice and included in testing.
Identification of hybrids between non-GM and GM rice
Seed samples collected from all the experiments were soaked in fresh water for c. 2 d after being stored for > 3 months to break seed dormancy. The well-soaked seeds were germinated at 37°C for c. 1–3 d. Germination rates were recorded as the number of seeds germinated divided by the total number of seeds soaked in each sample.
The germinated seeds were transferred into Petri dishes moisturized with 0.5× MS liquid culture medium (Murashige & Skoog, 1962) containing only macroelements and microelements and 50 µg ml−1 hygromycin B (Roche Diagnostics (Shanghai) Ltd., Shanghai, China), for c. 5 d in an illuminated growth chamber at 25–27°C. The surviving individuals were determined as hybrids because of their resistance to hygromycin B (Fig. 2).
To confirm the hybrids identified by hygromycin B method, 10% of the surviving seedlings from the hygromycin B treatment were randomly selected for polymerase chain reaction (PCR) identification with specific primer pairs designed for Bt gene (forward ACACCCTGACCTAGTTGAGC; reverse TGCAGAGAGCTTCAGAGAGTG), CpTI gene (forward AAAATGAAGAGCACCATCTTC; reverse TCTAGAGTTCATCTTTCTCATC), and the hygromycin resistance gene (forward TACACAGCCATCGGTCCAGA; reverse: TAGGAGGGCGTGGATATGTC). Total genomic DNA was extracted from leaf samples of individual seedlings following the method described by Doyle and Doyle (1987). The PCR reactions were performed in a PTC 10096v thermocycler (MJ Research Inc., Watertown, MA, USA). A denaturation period of 5 min at 94°C was followed by 35 cycles of 1 min at 94°C, 1 min at 56°C and 1 min at 72°C, and then 10 min at 72°C for final extension. Reactions were performed in a volume of 20 µl containing 1 × buffer, 1.675 mmol l−1 MgCl2, 0.5 µmol l−1 primer (TaKaRa Biotechnology (Dalian) Co., Ltd., Dalian, China), 200 µmol l−1 dNTP, c. 50 ng genomic DNA, and 1 U Taq polymerase (TaKaRa Inc.). The PCR products were distinguished by electrophoresis using 1.5% agarose gel.
Determination of transgene flow
Transgene flow frequency of a sample unit was determined as the ratio of number of hybrids detected against the number of seeds examined in the unit. (Fgf = number of hybrids/number of seeds examined). The transgene flow frequency of a plot was determined as the average of the units in the plot.
To test the effects of cultivation patterns (different GM rice proportions), rice pairs, planting sites, and their interactions on gene flow, the General Linear Models procedure (GLM, for anova with unbalanced data) was used and Student–Newman–Keuls (SNK) multiple comparisons tests were applied to compare the level of statistical significance. Before analysis, the percentage transgene flow data of sample units from Experiments A–C were log-transformed to satisfy the assumption of homogeneity of variance. The statistical analyses were all performed using the software package of SAS version 8.02 (SAS Institute Inc., Cary, NC, USA), and the significant level (α) of 0.05 was taken into consideration.
The average germination rates of seeds collected from all GM rice lines and non-GM rice varieties were very high (≥ 90%) after 3 months storage, which ensured us obtaining reasonable data of transgene flow with sufficient seedlings in this study. The transgene-specific PCR examination confirmed that hybrids identified by the hygromycin B treatment were true hybrids generated from transgene flow (Fig. 3). This indicates that the method of the hygromycin B screening for hybrids applied in the present study was reliable.
Frequencies of transgene flow from the three Bt/CpTI transgenic rice lines to their adjacent nontransgenic counterparts were generally very low. This result was based on the screening of a total of 645 775 seeds collected from two experimental sites in Fuzhou (Fujian Province) and Sanya (Hainan Province). Data indicated that frequencies of transgene flow from the Bt/CpTI GM rice lines to their adjacently planted non-GM counterparts never exceeded 1.0% in any of the experimental plots with different mixed-planting proportions in this study. Detailed results of the transgene flow experiments for the three GM and non-GM rice pairs at the two sites were summarized in Table 2.
Table 2. Results of transgene flow obtained from three genetically modified (GM) rice lines to their non-GM counterparts in experiments at two different sites (Fuzhou, Fujian Province and Sanya, Hainan Province)
GM and non-GM rice pair
Cultivation pattern with mixed-planting proportion
Average frequency (%) of transgene flow ± standard error1
Numbers in parentheses indicate total number of seeds examined.
HY1+ vs HY1–
Experiment A: GM majority
0.295 ± 0.063 (7800)
0.788 ± 0.179 (8000)
Experiment B: non-GM majority
0.049 ± 0.015 (59761)
0.100 ± 0.016 (62957)
Experiment C: random
0.170 ± 0.040 (27050)
0.541 ± 0.065 (33144)
HY2+ vs HY2–
Experiment A: GM majority
0.680 ± 0.108 (7800)
0.538 ± 0.108 (7800)
Experiment B: non-GM majority
0.050 ± 0.011 (62542)
0.046 ± 0.016 (62697)
Experiment C: random
0.265 ± 0.053 (16238)
0.444 ± 0.074 (42600)
MSR+ vs MSR–
Experiment A: GM majority
0.603 ± 0.103 (5822)
0.204 ± 0.051 (7603)
Experiment B: non-GM majority
0.071 ± 0.022 (50079)
0.070 ± 0.018 (59758)
Experiment C: random
0.482 ± 0.172 (20852)
0.159 ± 0.089 (33216)
The results showed that the cultivation patterns (mixed-planting proportion) had an obvious effect on transgene flow frequencies (Fig. 4). The anova test confirmed significant differences in frequencies of transgene flow among the three cultivation patterns (Experiments A–C) with different GM and non-GM rice mixed-planting proportions (Table 3). However, no significant differences in frequencies of transgene flow were detected among the rice pairs and between the experimental sites. The SNK multiple comparison tests revealed three significant groupings of transgene flow frequencies: (1) GM majority (Experiment A), (2) random (Experiment C) and (3) non-GM majority (Experiment B). In addition, the anova test also indicated that there was no significant interaction between the mixed-planting proportions and experimental sites, although significant interaction between the rice pairs and experimental sites was detected (Table 3).
Table 3. Results of anova presenting degrees of freedom (df), mean square (MS), F-values and probability values (P) for effects of genetically modified (GM) and non-GM rice mixed-planting proportions, GM and non-GM rice pairs, experimental sites, and their interactions on transgene flow (%, log-transformed) calculated by the general linear models procedure (GLM)
There is a considerable pressure to promote commercialization of GM rice in China, because of the steady decrease in domestic rice production over the past 5 yr and a fear that this country will be unable to produce enough rice to support its growing population (Jia, 2004). All three GM rice lines used in this study are targeted for commercial production in the near future, once they have received permission from the Chinese Biosafety Office. The GM rice MSR+ already has permission from the Office to be released into the environment for field-testing, and this GM rice bred from a widely used rice male-sterility-restoring line (MSR–) will also be extensively used in transgenic hybrid rice production. The GM hybrid rice lines HY1+ and HY2+ are technically ready for commercialization and now have permission for pre-release field production tests. Therefore, it is timely to conduct the experiment to determine the possibility of crop-to-crop transgene flow from these GM rice lines to non-GM rice varieties, including their corresponding non-GM counterparts, as a part of their biosafety assessment. This is because the pollen-mediated ‘contamination’ of non-GM rice varieties by GM rice grown at the vicinity is usually of great concern for biosafety, particularly when international trade of rice is involved and organic market is targeted. In China, experiments are greatly needed to measure the magnitudes of transgene flow in fields with various GM and non-GM rice mixed-planting patterns because of the nature of small farm owners in China who choose to grow their own rice varieties in the small patches of farmland where GM and non-GM rice varieties may grow at close spacing in a mosaic pattern.
Our study indicated that transgene movement from GM to non-GM rice did occur if they were sympatrically grown. However, results from this experiment revealed very low frequencies of transgene flow from the three Bt/CpTI GM rice lines to their non-GM counterparts at extremely close spacing, although the frequencies varied significantly in different plots (0.05–0.79%). Even in the ‘extreme scenario’ of the GM majority experiment, the transgene flow frequencies were always below 1.0%– the most strict threshold that is adopted to determine ‘transgene contamination’ in the international trade of cereals. This result is matchable with the reported frequencies of crop-to-crop and crop-to-weed rice gene flow by different authors (0.01–0.53% in Messeguer et al., 2001, 2004; 0.04–0.20% in Rong et al., 2004; c. 0.18% in Bashir et al., 2004; 0.011–0.046% in Chen et al., 2004). The observed values of rice gene flow based on molecular or transgene markers were somehow slightly lower than the traditionally expected outcrossing rate (1–2%) (Oka & Morishima, 1967; Diao et al., 1996). Furthermore, the detected crop-to-crop rice gene flow was significantly lower than the reported crop-to-wild (Oryza rufipogon) rice gene flow (c. 2.94% in Song et al., 2003), which is obviously attributed to the differences in mating systems between cultivated and wild rice species. The wild rice O. rufipogon usually demonstrates a much higher outcrossing rate than cultivated rice (Oka & Morishima, 1967; Bajaj & Mohanty, 2005).
It is important to emphasize that the frequency of hybrids from our experiments was determined based on the observation of > 645 700 geminated seeds collected from the transgene flow experimental fields at two sites using a hygromycin resistance gene as the selection marker (confirmed by PCR of the transgenes). The data can well represent the actual transgene outflow from the three GM rice lines to their non-GM counterparts with a relatively high confidence. This study also provides a good example of using an efficient screening method such as hygromycin resistance to estimate transgene flow from GM to non-GM rice at a large scale. This method will facilitate the massive scale of experiments that can provide a convincing assessment of likelihood of transgene outflow in rice fields. It would be impossible to conduct a gene flow experiment at such scales using molecular markers, even using some morphological markers, to detect hybrids.
The frequencies of transgene flow determined from this experiment were all based on the GM and non-GM rice individuals cultivated adjacently with a distance of 20 cm between the hills in a row and between the rows. This means that the transgene flow frequencies measured from GM to non-GM rice were nearly at the ‘zero distance’, because the canopy size of a rice plant is usually about the same as or larger than an area of 20 × 20 cm2 and rice panicles of different individuals were connected to each other in the experiment. Therefore, the frequencies of transgene flow from GM to non-GM rice in this study are the extreme (or maximum) situation. Song et al. (2004) observed a considerable decrease in rice pollen density with the increase of horizontal and vertical distances in a rice-pollen-flow study. We expect a drastic decrease in transgene flow frequencies with increasing distances between GM and non-GM rice varieties. We are conducting another experiment to measure the decrease in transgene flow frequencies at different distances from pollen donors. This will illustrate the relationships between transgene flow and distances, and guide the establishment of effective physical isolation to minimize the transgene escape from GM to non-GM rice.
Among the cultivation patterns used in this study, the GM majority pattern (Experiment A) showed the highest values of transgene flow. This result can be explained by the high densities of transgene pollen around the non-GM rice plants. The frequencies of transgene flow reduced significantly in the random and non-GM majority patterns (Experiments C and B), which, in contrast, indicated the effect of low densities of transgene pollen in transgene outflow. Insignificant interaction detected between the mixed-planting proportions and experimental sites suggests that mixed-planting patterns of GM and non-GM rice varieties may consistently affect the magnitude of transgene flow in different rice planting areas. This result further indicates that the magnitude of transgene flow is closely associated with the relative pollen density of the pollen donors (GM rice), irrespective of the rice planting areas.
The GM and non-GM rice pairs showed a significant interaction with the experimental sites on frequencies of transgene flow. This result suggests that no valid conclusions for rice transgene flow everywhere can be drawn based only on data from a single experiment at only one site. Therefore, for an accurate assessment of pollen-mediated transgene flow from GM to non-GM rice varieties, we should take all possible factors, such as cultivation patterns, amount of pollen from GM varieties, outcrossing rates of the non-GM varieties, experimental sites, and distances between GM and non-GM rice varieties into consideration. This will facilitate decision-making on GM rice biosafety assessment of gene flow based on the ‘case-by-case-principle’.
We thank the Nature Science Foundation of China for Distinguished Young Scholars (Grant no. 30125029) and the Science and Technology Commission of Shanghai (Grant no. 02JC14022 and 03dz19309) to B. R. Lu, and support from National High Science and Technology Program (863) (2001AA212031, 2001AA212041) and the Important Science and the Technology Program of Fujian Province (99-Z-3, 2003 N002) to F. Wang. We appreciate the valuable comments on the manuscript by Andrew Watkinson of University of East Anglia, UK, and Allison Snow of Ohio State University, USA.