COST OF TRANSGENIC AND NONTRANSGENIC HERBICIDE RESISTANCE GENES IN ARABIDOPSIS THALIANA

Colin B. Purrington and Joy Bergelson

Department of Ecology and Evolution, University of Chicago, 1101 E. 57th Street, Chicago, IL 60637

SUMMARY

Two common assumptions in risk assessments of transgenic plants are that expression of extra genes will be costly and that mutations induced by the insertion of a transgene can disrupt the function of genes important in reproductive success. Together, these mechanisms are often cited as guarantees that the products of transgenic manipulation will never attain fitness values higher than that of the nontransformed parent and, therefore, such plants will never become invasive. To investigate these issues, we produced transgenic lines of Arabidopsis thaliana that were resistant to the herbicide chlorsulfuron. We then compared seed production of these lines with a nontransgenic mutant that was also resistant to chlorsulfuron. In this way we could assess the cost of the same resistance gene in both a transgenic and nontransgenic context. In addition, by growing plants at high and low levels of fertilizer, we were able to test whether costs become higher under resource limitation, a common prediction made by researchers studying costs of resistance. The results of this experiment, conducted under greenhouse conditions, indicated a significant cost of the herbicide resistance gene in the nontransgenic accessions and also highlighted the importance of variability among transgenic lines. The performance differences among transgenic and nontransgenic lines are interpreted in the context of current risk assessment procedures.

Key words: Cost of resistance, herbicide resistance, transformation, risk assessment, Arabidopsis thaliana

INTRODUCTION

Currently, the focus of risk assessment of genetically modified plants is on the product itself, with companies wishing to commercialize a transgenic plant charged with answering a simple question: Is the transgenic crop weedier than a nontransformed variety? Although it is possible to answer this question with a series of simple experiments (Purrington and Bergelson, 1995), more often it is addressed by merely stating that crops, already dependent upon human care, are likely to be further handicapped by the numerous mutations that are produced during the procedures of genetic alteration (Brears, 1989; Feldman et al., 1994). Moreover, it is hypothesized that the expression of an extra gene will always entail a metabolic cost that would prevent the transgenic crop from ever attaining a reproductive potential and competitiveness equivalent to its untransformed parent (Tiedje et al., 1989). Although this latter argument is persuasive, the prevalence of costs for resistance traits has only been demonstrated in nontransgenic plants (reviewed in Bergelson and Purrington, 1994). As a consequence, persons involved in the evaluation of transgenic plants are placing considerable weight on the belief that the cost of transgenic resistance is comparable to costs of natural resistance. Given the potentially harmful outcomes that may occur if this assumption is incorrect, there is a strong need to determine if resistance traits display similar costs both in their original species and in the species to which they are artificially transferred.

Another concern in assessment of risk is whether costs are detectable under all environmental conditions. Many authors have argued that metabolic costs will only be detectable under conditions of resource limitation (Coley et al., 1985; Herms and Mattson, 1992; Bergelson, 1994a; 1994b). If true, this would mean that the crops expressing extra genes may show reduced seed production relative to the untransformed parent only under non-ideal conditions. Because it is likely that most experimental sites of transgenic crops are weeded, watered, and fertilized, it possible that our current perception of risk is applicable only to ideal conditions. If we are concerned about the escape of transgenic crops into nearby uncultivated areas where drought, nutrient scarcity, and pathogen attack are more commonplace, we must be careful to perform trials that include environmental conditions more representative of these sites.

Perhaps the greatest difference between a natural and transgenic resistance gene is that the former usually has a single location within a genome, whereas the position of a transgene is essentially random. Variability in expression levels among transgenic lines is often attributed to the effect of insertion site, and it is therefore likely that the cost of an transgene will be similarly affected. Therefore, in order to fully understand the cost of a transgene it is necessary to follow multiple lines of a given transgenic product.

This paper discusses a greenhouse experiment in which transgenic and nontransgenic individuals of Arabidopsis thaliana with resistance to the herbicide, chlorsulfuron, were grown along with a susceptible control at two levels of nutrient availability. This design allowed us to answer the following questions: i. Is there a cost to expressing a gene that confers resistance to the herbicide chlorsulfuron? ii. Is there an identical cost to expressing a transgenic form of the resistance gene? iii. Is there significant variability among lines from different transformations involving the same transgene? and, iv. Are costs different at different levels of fertilizer addition? These questions will be answered in the context of current procedures of testing transgenic crops prior to commercialization.

MATERIALS AND METHODS

Four genotypes of Arabidopsis thaliana, two transgenic and two nontransgenic, were used in the experiment. The transgenic lines, 'pGH8-A' and 'pGH8-B', were produced through Argrobacterium tumefaciens-mediated transformation. These lines are products of separate transformation events using a plasmid (donated by Dr. G. Haughn, University of Saskatchewan) containing a gene conferring resistance to the herbicide chlorsulfuron and the selectable marker NPT-II (kanamycin resistance). The seeds used in the present experiment are progeny of the original transformants regenerated through tissue culture. The nontransgenic genotypes are the 'Columbia' ecotype of A. thaliana, which was used as the parent for the above transgenic lines, and 'GH50', which is an EMS-generated mutant of 'Columbia' that is resistant to the herbicide chlorsulfuron and has been backcrossed to 'Columbia' for six generations. Resistance to chlorsulfuron in 'GH50' is attributed to a single mutation which results in the production of an mutant form of acetolactate synthase (ALS), an enzyme that is involved in the synthesis of branched-chain amino acids (Haughn and Somerville, 1986). The mutant enzyme has altered binding specificity which renders plants containing the gene tolerant to the highly-active sulfonylurea herbicide, chlorsulfuron (Ray, 1984).

On 4 July 1994, single seeds were placed on the surfaces of 7.5 cm square plastic pots filled with Terra-Lite vermiculite (Grace Sierra) and then lightly dusted with Redi-Earth Peat Lite (Grace Sierra). Each pot was provisioned with either a small (0.04 g) or large (0.4 g) amount of a controlled-release fertilizer (Sierra Chemical Company) with a 17-6-12 (N-P-K) formulation. For the 'Columbia' and 'GH50' genotypes, there were 28 replicates at each nutrient level, whereas 112 replicates were made for the two transgenic genotypes. (This difference in replication was because the transgenic plants are from a seed source segregating for the transgene, and therefore one-quarter, approximately, of the 112 are actually homozygous transgenic. For the current analysis, the null segregants, hemizygous transgenic, and homozygous transgenic individuals will be analyzed together.) The positions of the 560 pots were randomized on benches in a greenhouse at Washington University (St. Louis, MO). Pots were equipped with cylinders constructed of plastic sheets that collected dispersing seeds yet allowed the plants to be watered at the base.

Seed production of plants was estimated by counting the number of seeds in weighed subsamples of the total seeds, and then dividing the total weight of all seeds by the average seed weight. Analysis of variance of seed production and comparisons of means were performed in SuperAnova (Abacus Concepts, 1989).

RESULTS

An analysis of variance (model not shown) on seed production showed a strong effect of nutrient level (df = 1, F = 106.01, p = 0.0001), with plants in the high nutrient treatment achieving more than twice the fecundity of plants provided with lower amounts of nutrients (Figure 1). The three genotypes, however, did not differ from each other (df = 2, F = 2.26, p = 0.106), and, furthermore, the interaction of genotype and nutrient level was not significant (df = 2, F = 2.27, p = 0.105). Although there appeared to be more variation among genotypes within the high nutrient treatment, no significant differences were detected in a planned comparison treating 'Columbia' as the control group (df = 1, F = 0.55, p = 0.461).

A separate analysis of variance (not shown) found no significant differences between the two lines of 'pGH8' (df = 1, F = 1.97, p = 0.161). Nutrient level, as above, had a large effect on the seed production of each line (df = 1, F = 116.10, p = 0.0001), but the two lines had the same relative performance at both nutrient levels, as indicated by the interaction term (df = 1, F = 2.16, p = 0.143). The trend (also nonsignificant) for differences to be stronger at the high nutrient level was again observed (Figure 2).

DISCUSSION

The results have several implications. First is that a cost of resistance may not always be present for herbicide resistance. Potentially, even the treatment involving low amounts of fertilizer may have provided adequate amounts of nutrients. Experimental conditions that reduce the seed production of genotypes by more than 50% (the current reduction) may be needed to reveal an underlying cost. Bergelson (1994a) found that 'GH50' had reduced performance relative to 'Columbia' only when the two genotypes were grown with high numbers of interspecific competitors. Average seed number in those high-density treatments was approximately 40, which is considerably lower than the seed production in the higher stress treatment in this experiment (approximately 2400). Therefore, it is likely that an experiment with a greater range of nutrient levels is needed to more accurately diagnose the effect of nutrient limitation on costs of herbicide resistance. We are currently undertaking such an experiment. In addition, we are exploring the role that amino acid starvation may play in the origin of costs of expressing a mutant ALS gene.

We also found in this experiment that the transgenic lines containing the mutant ALS genes did not produce fewer seeds than the untransformed parent, 'Columbia'. Although this may indicate a lack of a cost of transformation as well as a lack of cost of expression of the mutant ALS gene, there is a more likely explanation. Because we utilized seed from segregating lines, only a fraction of the plants we treat as "transgenic" actually possess two copies (i.e., are homozygous) for the transgene. The remaining plants, hemizygotes and null segregates, are likely to dilute the magnitude of any underlying differences inherent in plants that are producing high levels of the mutant enzyme. To investigate this possibility, we are using the subjecting the progeny of the study plants to tests that will identify which plants are homozygous transgenic. Upon completion of these assays it will be possible to reanalyze the data with more accuracy, even though the overall sample size will be necessarily reduced. We note that this complication is present in many studies of transgenic crops (Purrington and Bergelson, 1995), and emphasize that similar designs are inadequate for detecting differences between homozygous transgenics and the parental variety unless further tests are employed to identify the genotype of every individual. Given that biotechnology companies are required to produce such comparisons in their petitions for nonregulated status (USDA APHIS, 1993), it is important that persons evaluating these petitions are informed whether or not the tested genotypes are, in fact, true breeding, homozygous lines.

We are also concluding a season of field experiments in which backcrossed, homozygous transgenic lines are evaluated relative to the performance of 'Columbia' individuals. These experiments have also involved nontransgenic lines that were derived from the transgenic lines, and will therefore contain mutations that are not closely linked to the transgene itself. The inclusion of these extra lines will allow us to make more accurate conclusions about the effects of transformation versus the effects of possessing transgenes, factors that are confounded in most trials of transgenic organisms (Elkind et al., 1995). Because risk assessment procedures are largely tailored to the product (plants with foreign genes) rather than the procedure (transformation and plant regeneration), these experiments will provide biopractitioners with a model case on which to base similar assessments of risk. Our experiments also utilize transgenic lines (and their null segregates) that contain a control plasmid, pBIN19, that contains the selectable marker conferring resistance to the antibiotic kanamycin. This additional control is needed because most transgenes are delivered into plant DNA on fragments also containing marker genes. These genes, although not usually important to the final phenotype, may also be contribute to the cost attributed to the transgene and therefore to the assessed risk of causing weediness.

The majority of our recent research has been on yield trials of the replicate transgenic lines and the various control lines. The results of these trials, which involve alteration of levels of nutrients, competition, and herbicide application, will correspond to work already done in which the emphasis is on the use of individual performance measures to predict risk. However, if we are indeed worried about escape of populations of transgenic plants (Crawley et al., 1993; Bergelson, 1994b), experiments must be done that examine the risk of population increases following escape of plants from containment. With this in mind we have begun an experiment that measures the rate of population increases within plots containing transgenic plants or mixtures of transgenic and nontransgenic genotypes.

The results of these experiments will provide a rare opportunity to examine the effects of a resistance gene in both a natural and a transgenic context. In addition, the inclusion of multiple transgenic lines of the same construct, several different types control genotypes, and large numbers of replicates of each line make the current research program one which will yield data on a wide variety of concerns that have been raised about the planned introduction of transgenic plants into the environment.

ACKNOWLEDGMENTS

We thank G. Haughn for the plasmid containing herbicide resistance and the 'GH50' mutant, and L. Singer for greenhouse assistance. Funding from the USDA Biotechnology Risk Assessment Program is gratefully acknowledged.

REFERENCES

Abacus Concepts, SuperAnova (1989) Abacus Concepts, Inc., Berkeley, CA.

Bergelson J (1994a) Changes in fecundity do not predict invasiveness: A model study of transgenic plants. Eco. 75:249-252.

Bergelson J (1994b) The effects of genotype and the environment on costs of resistance in lettuce. The Am. Nat. 143: 349-359.

Bergelson J, Purrington CB 1994 Costs of resistance in plants and their relevance to biotechnology risk assessment. In: Proceedings of the 6th Investigators' Metting for EPA's Environmental Releases of Biotechnology Products Research Program.

Brears T, Curtis GJ, Lonsdale DM (1989) A specific rearrangement of mitochondrial DNA induced by tissue culture. Theo. and Appl. Gene. 77: 620-624.

Coley PD, Bryant JP, Chapin FS, III (1985) Resource availability and plant antiherbivore defense. Science 230: 895-899.

Crawley MJ, Hails RS, Rees M, Kohn D, Buxton J (1993) Ecology of transgenic oilseed rape in natural habitats. Nature 363: 620-623.

Elkind Y, Binyamin N, Nadler-Hassar T (1995) Quantitative analysis of the transgene variability among primary tobacco transformants. Trans. Res. 4: 30-38.

Feldman KA, Malmberg RL, Dean C (1994) Mutagenesis in Arabidopsis. In: E.M. Meyerowitz EM, C.R. Somerville (eds) Arabidopsis. Cold Spring Harbor Laboratory Press, NY pp.137-172.

Haughn G, and Somerville C (1986) Sulfonyluria-resistant mutants of Arabidopsis thaliana. Molec. and Gener. Genet. 204: 430-434.

Herms DA, Mattson, WJ (1992) The dilemma of plants: To grow or defend. The Quarterly Review of Biology 67: 283-335.

Purrington CB, Bergelson J (1995) Assessing weediness of transgenic crops: Industry plays plant ecologist. Trends in Ecology and Evolution 10: 340-342.

Ray TB (1984) Site of action of chlorsulfuron. Plant Physiology 75: 827-831.

Tiedje, JM, Colwell RK, Grossman YL, Hodson RE, Lenski RE, Mack RN, Regal PJ (1989) The planned introduction of genetically engineered organisms: Ecological considerations and recommendations. Ecology 70: 298-315.

USDA APHIS (1993) 7 CFR Part 340. Federal Register 58: 17044-17059.

Figure 1. Seed production (+ S.E.M.) of 'Columbia', 'GH50', and pGH8 at low and high nutrient levels.

Figure 2. Seed production (+ S.E.M.) of pGH8-A and pGH8-B at low and high nutrient levels.