The assessment of transgenic herbicide tolerance
A herbicide is any substance that eliminates unwanted plants. For example, in lowland rice cultivation, water is used as an efficient herbicide for weed control. In agro-systems where weeds are perceived to seriously limit crop productivity, herbicides are used to replace manual labour or other mechanical methods of weed control. Key issues in herbicide development and use are:

• environmental impact; each herbicide needs to offer an acceptable balance between its potential harm to the environment, for instance by its accumulation in ecosystems or food chains, and its desired efficacy in weed control.
• selectivity; the selectivity of a herbicide allows for discrimination between sufficient damage to the weed and no damage to the crop. The desired effect of a herbicide generally depends on the amount used.

(1) Will the presence of the bar gene and the PAT protein transform the crop into an uncontrollable weed? Enhancement of fitness of the crop caused by genetic transformation is a crucial issue in biosafety assessment. Fitness is a measure of the competitive success of a given plant, expressed in either vigour or number of progeny generated. Spraying with phosphinothricin creates a clear selective advantage for the transgenic crop in the field, but it is unlikely that such selective conditions will be found outside agricultural production fields. In the absence of phosphinothricin, the presence of the bar gene and the PAT protein will not contribute to any enhanced competitiveness or weediness of the crop itself.

(2) Will the bar gene spread to wild relatives or other organisms that could result in problematic species/organisms? To answer this question, the likelihood of outcrossing of genes transferred and gene flow has to be assessed. Outcrossing means cross-pollination with a plant that is genetically different from the initial crop and results in a hybrid. If such a hybrid continues to outcross to relatives, the accompanying spread of genes is referred to as gene flow. At present, the overall conclusion is that in some crops, under some conditions and at some locations, gene flow, including the bar gene, may occur from a transgenic phosphinothricin-tolerant crop to a wild relative. For example, oilseed rape (Brassica napus) is reported to outcross to several wild relatives such as B. rapa, (a parental species of oilseed rape), wild radish (Raphanus raphanistrum) or hoary mustard (Hischfeldia incana). The likelihood and distances of outcrossing may differ, and can be estimated. Nevertheless, the relevant question regarding any wild relative remains whether the presence of the bar gene affects the recipient plant’s fitness. As long as the herbicide is not used outside agricultural production fields, there is no selection pressure favouring the bar gene-containing hybrid. For the same reason, centres of origin of the crop would not be affected either. Outcrossing could, however, yield a phosphinothricin-tolerant weed hybrid that, if it moves back into the field, cannot be controlled with phosphinothricin anymore. This would decrease the selectivity of phosphinothricin. It is difficult to predict how quickly such a putative loss of selectivity might occur. Yet this would seem to be an economic rather than a biosafety issue, and one that can be prevented by responsible use of the trait in agronomy.

Commercialized transgenic herbicide tolerant crops
Common name	Latin name	Herbicide	Company
Canola/oilseed rape	Brassica napus	Phosphinothricin	AgrEvo (Germany)
		Glyphosate	Monsanto (USA)
Chicory	Cichorium intybus	Phosphinothricin	Bejo Zaden (the Netherlands)
Cotton	Gossypium hirsutum	Glyphosate	Monsanto
Maize	Zea mays	Phosphinothricin	AgrEvo,  Novartis (Switzerland),  Monsanto
Soybean	Glycine max	Phosphinothricin	AgrEvo
		Glyphosate	Monsanto
Sources: OECD Biotech database; The Gene Exchange Fall/Winter 1998.

(3) Does the presence of the bar gene or the PAT protein compromise human consumption? DNA itself is not toxic. The likelihood and consequences of a human intestinal cell acquiring the functional bar gene seem sufficiently minute. Undesirable effects could result from the presence of the PAT protein itself, its enzymatic activity, or any product derived from it. Using a variety of criteria, no allergenicity or toxicity of the PAT protein or its degradation products has been reported.

(4) Are there any unpredictable and undesirable effects associated with transgenic phosphinothricin tolerance? The presence of the bar gene, its PAT protein product or any of its metabolites or the method by which the bar gene was introduced, may in some cases lead to unexpected, so-called ‘pleiotropic’, effects. These effects may alter any of the ecological relationships or toxicological characteristics of the crop or any wild relative derived from outcrossing. In general, it is currently unclear whether such pleiotropic effects do occur to the extent that any effect can be measured. If any effect can be measured, it is unclear whether the effect has any biological relevance compared to current agronomic practice. And if the effect has any relevance, it is not certain whether the outcome is an adverse effect. It can be argued that the dynamics and self-regulatory properties of ecosystems and consumers would create sufficient ‘noise’ so that pleiotropic effects due to the presence of the bar bene will be of minor or no importance.

(5) To what extent are consumers going to be exposed to the herbicide or its metabolites? The novel use of phosphinothricin as a selective herbicide implies that phosphinothricintolerant plants or products derived from them can now contain traces of the herbicide. Obviously, this depends on whether, when, and how much the plant was sprayed prior to consumption and what part of the plant is actually used for consumption. Without spraying, for example because the phosphinothricin tolerance is only used as a selection system in the laboratory phase, the situation reverts to question number 3 above. With spraying, the additional metabolites and degradation products of the herbicide need to be monitored on a product-by-product basis for products to be marketed. To date, the likelihood of exposure and the toxicological impact of such exposure are not sufficiently clear and should be covered by regulations for herbicide use.

(6) What is the environmental impact of phosphinothricin? The application of transgenic phosphinothricin-tolerant plants will increase the use of this particular herbicide, but will replace currently used substances.For example, regular sugar beet growing in the Netherlands requires about 4 kg active ingredient (AI) per ha per season, composed of combinations of various compounds such as Pyramin and Tramat. On the other hand, for transgenic phosphinothricin-tolerant sugar beet, about 1.5 kg AI phosphinothricin per ha would be sufficient. The environmental impact of that 1.5 kg phosphinothricin is estimated to be considerably lower than that of the current cocktail of 4 kg. For other crops or environments, the required amount of phosphinothricin can be the same or even higher than the amount of current mixtures. Still, the environmental impact of this herbicide is considered less than that of currently used cocktails of herbicidal compounds. Inasmuch as the use of such crops replaces currently used herbicides with phosphinothricin, their use has a less adverse impact on the environment.

Essentially the same questions are valid for any transgenic glyphosate-tolerant crop. The toxicological consequences of glyphosate’s hypothetical presence in plant material also need attention, depending on the precise method used to obtain the transgenic trait (see box). To manage the risks of biotechnology the Organisation for Economic Co-operation and Development (OECD) has suggested the concepts of ‘familiarity’ for the ecological assessments, and ‘substantial equivalence’ for toxicological assessments. Yet both concepts seem to be too poorly defined to be applied in a case-by-case approach (see also the article by van Dommelen).
Transgenic phosphinothricin or glyphosate tolerance are new traits and unfamiliar to ecosystems. Therefore, the concept of familiarity does not help, unless one considers ‘herbicide tolerance’ as the essence of the trait, which, as a general concept, is a known phenomenon. In the case of the introduction of a mutated plant-derived EPSPS (see box) to obtain glyphosate tolerance, it might be justified to talk about substantial equivalence. However, the EPSPS enzymes used in transgenic plants vary in structure and activity. Substantial equivalence would require that this variation is similar to the variation of EPSPS activity in untransformed plants. For this, however, more insight in the naturally occurring variation in EPSPS activity would be required.

Transgenic herbicide-tolerant crops: technical background

Transgenic phosphinothricin tolerance. The herbicide phosphinotricin originates from the microbe (actinomycete) Streptomyces viridochromogenes and several other Streptomyces species. Phosphinothricin is an amino acid-like compound that is currently synthesized industrially. Its herbicidal activity is based on inhibiting the key enzyme in nitrogen metabolism glutamine synthetase (GS). Obtaining phosphinotricin-tolerant crops is based on the strategy of the microbial producers to protect themselves against their own compound. The Streptomyces species protect themselves by producing an enzyme that deactivates the phosphinothricin. This enzyme is called phosphinothricin-N-acetyltransferase (PAT). The gene coding for PAT is called the bar gene. From different Streptomyces sources, bar genes were obtained. The bar gene was successfully introduced in a large number of crops, making them transgenic herbicide tolerant crops.

Transgenic glyphosate tolerance. The herbicide glyphosate is chemically a simple tertiary amine. Its herbicidal activity is based on inhibition of the enzyme 5-enolpyruvylshikimate-3-phosphate synthase (EPSPS). This enzyme operates in the shikimate pathway that yields aromatic amino acids and secondary plant products. Two different strategies to obtain transgenic glyphosate tolerance are of agronomic relevance:  • introduction of a glyphosate-tolerant EPSPS. Genes encoding EPSPS with a reduced affinity to glyphosate were isolated from a number of microbial and plant sources. Furthermore, laboratory-originated mutations of EPSPS are used.  • introduction of a glyphosate-degrading enzyme known as glyphosate oxidoreductase (GOX). The gox gene was isolated from an Achromobacter bacterial strain.  Various glyphosate-tolerant EPSPS genes and the gox gene, alone or in combination, have been successfully introduced into crops.

A role for science-based assessments?
At the moment it is unclear whether the transgene-centred approach to biosafety as outlined above will make an impact on assessment and will contribute to the public’s appreciation of transgenic herbicide-tolerant products. If it is the technology itself, rather than the resulting herbicide-tolerant product, that is mistrusted, any assessment of that product will be in vain. In such a situation, science-based biosafety assessment becomes a paradoxical exercise that will never be able to satisfy its sceptics.
What tends to be overlooked in most controversies is, that the decision to disapprove the use of such transgenic plants also has its consequences and basic assumptions, for instance that the currently accepted situation is ‘natural’ and with less or better acceptable risks. Obviously, transgenic herbicide-tolerant products should meet the highest possible quality standards. In that way, they are no different from any other product, transgenic or not. Consumers and societies have the right to know what they are eating and the right to demand safety.
The development of transgenic herbicide-tolerant plants was intended as a solution to some of the current problems in plant breeding and agricultural production, but not necessarily as the only solution. If one does not accept the use of any herbicide in agriculture, the replacement of current-day herbicides with the herbicides phosphinothricin or glyphosate is obviously pointless. But if one accepts the current rating of their environmental impact, the substitution of more harmful herbicides with either one of these two herbicides must be judged positively for the environment. Although herbicide resistant crops that encourage this replacement may not be the best solution, and they are unlikely to be the final solution, they are an improvement on the current situation that gives societies time to come up with even better solutions.
The priority issue that should be resolved is the level of exposure of consumers to the herbicide and/or its degradation products, and the toxicological impact of such exposure, if any. Such assessments could be done in the industrialized world and should generate data in the public domain that are useful for all countries where such transgenic herbicide-tolerant crops will be grown and sprayed. If this shows that the likelihood and/or toxicological impact of exposure is sufficiently low, developing countries could concentrate their capacity on biosafety testing for individual crops for their specific environments. This pragmatic approach could ensure that such crops can also be adapted in agricultural production systems in developing countries. It may enable them to balance risks with benefits better, while keeping an eye on the added value of the new technology for the next Green Revolution.
Jan-Peter Nap

Department of Molecular Biology, DLO-Centre for Plant Breeding and Reproduction Research, CPRO-DLO, P.O. Box 16, 6700 AA Wageningen, the Netherlands.
Phone (+31) 317 477 169; Fax (+31) 317 418 094; E-mail J.P.H.Nap@CPRO.DLO.NL

Sources
European Federation of Biotechnology Task Group (1999), Ethical Aspects of Agricultural Biotechnology. The Hague, the Netherlands: CBC.

De Kathen, A. (1998), "The debate on risks from plant biotechnology: the end of reductionism?" Plant Tissue Culture and Biotechnology 4, pp. 136-147.

Lutman, P.J.W. (ed.), (1999), Gene Flow and Agriculture. Relevance for transgenic crops. Nottingham, UK: British Crop Protection Council.

Metz, P.L.J., Stiekema, W.J. and Nap, J.P. (1998), "A transgene-centered approach to the biosafety of transgenic phosphinothricin-tolerant plants." Molecular Breeding 4, pp. 335-341.

Nap, J.P., Metz, P.L.J. and Stiekema, W.J. (1996), A transgene-centered evaluation of genetically modified plants, Part 3, Biosafety of genetically modified glyphosate-tolerant plants. Wageningen, the Netherlands: CPRO-DLO.

Personal communication with L. Gilissen (CPRO-DLO).