Risk assessment and risk management of LMOs with stacked genes or traits (part I)

It is useful to consider the question of stacked genes in context.  Traditional breeding methods often introduce genes from species that do not normally cross in nature. These methods, in effect, introduce hundreds, or even thousands, of stacked genes.  With these methods, we cannot assess the safety of the stacked genes, and the interaction of these genes with each other, on a gene-by-gene basis.  There are too many genes involved, and for most of the genes, their functions, and their interactions with other genes are unknown.  However, plant breeders assess these stacked gene introductions and their interactions on a more global basis by testing plant varieties for productivity, resistance to abiotic and biotic stresses and quality in many environments on a trait or product basis. i.e. the safety and productivity of a plant variety, whether transgenic or not, is typically tested in multiple environments over multiple generations, genetic backgrounds and years.  In cases where biochemical pathways or specific analyzable characteristics may be known to be adversely affected, these characteristics are analyzed.  Any detrimental characteristics are eliminated during the breeding process when commercial varieties are developed.  Due to the rigorous testing in plant breeding there have only been a few documented cases of safety hazards in crop plants. As a result crops traditionally bred have provided a safe food supply for a century and are not subjected to risk assessment by most countries, These methods of assessment have been or are  applied to transgenic crops, and achieve the same level of safety.

However, LMOs are subjected to additional assessments and therefore, those that have been approved for commercial release using the risk assessment framework set forth in documents such as Annex III of the Cartagena Protocol, achieve an even higher level of safety that traditionally bred varieties.  This conclusion is supported in a publication by the European Commission in 2001, reporting the results of 81 Research Projects performed over 15 years:

“Research on the GM plants and derived products so far developed, following usual risk assessment procedures, has not shown any new risks to human health or the environment, beyond the usual uncertainties of conventional plant breeding.  Indeed, the use of more precise technology and the greater regulatory scrutiny probably make them safer than conventional plants and foods…”

The existing framework for risk assessment can be readily applied to the assessment of stacked genes introduced into crop varieties.  The assessment of these genes provide the advantage that they are well characterized in comparison to those introduced by traditional breeding.  For clarity, we are using the term “stacked” in this posting to mean combinations of transgenes achieved either by the cross-breeding of individual LMO lines or the direct introduction of multiple genes via transformation.  In both situations, the sequences and functions of the introduced genes are known, and the interactions between the introduced genes can be better predicted.  Where knowledge of these genes and their interactions is not sufficient, gaps can be filled prior to commercial release by conducting the appropriate tests in field trials.  This increased information enables us to do a risk assessment with these stacked genes, which we could not do with traditionally bred varieties.  The additional risk assessment goes beyond the already safe and proven system now used for plant breeding.

Annex III of the Cartagena Protocol provides a robust framework for risk assessment.  It provides a means by which LMO’s can be assessed for safety, whether they contain a single or multiple transgenes, and whether multiple transgenes have been incorporated by traditional breeding methods or by transformation via a single vector.  It should be pointed out that this framework applies to environmental effects and impacts, and does not include questions of food safety, which are not covered under the Protocol.

Adequate guidance in the risk assessment of stacked genes has been published by various regulatory agencies.  Examples are given in the supplementary readings posted by the BCH in connection with this online forum.  PRRI considers these guidance documents as good sources of further detail regarding risk assessment within the framework of Annex III.  Regulatory decision documents covering the risk assessment of various stacked events or varieties can also serve as examples for conducting risk assessment with stacked genes.  These documents are available on the web sites of many regulatory agencies.
PRRI has published a guide to conducting risk assessments that is also consistent with Annex III, and which can be applied to the assessment of stacked genes.  This guide takes a step-by-step approach to risk assessment, and can be applied to any number of genes that may be present in a specific LMO.  This guide can be downloaded from the internet at the following URL:
http://pubresreg.org/index.php?option=com_docman&task=cat_view&gid=48&Itemid=58

PRRI recommends that in the case of LMO’s containing genes stacked by direct transformation or crossing, in which the genes have already assessed previously in individual events, complete safety assessments on the stacked-gene LMO need not start from a blank slate.  The information generated in connection with the individual genes can be used to conduct a risk assessment of the combined genes, thus focusing the risk assessment on the potential interactions between genes or gene products.  Because the genes and their mode of action are well known, the interactions between introduced genes can be assessed in a precise way.  The first step in such an assessment is to determine whether, based on the known characteristics of the gene or gene products, there is a potential for interaction.  One example of an identified potential interaction would be the case where the two genes being combined are part of the same metabolic pathway.  If the answer is no, then the risk assessment can rely on the previous assessments done on the individual genes.  If the answer is yes, then the potential for a harmful or beneficial interaction can be tested.  The interactions of these genes with the endogenous genes in the plant can then be assessed in the light of what is known about the interactions of the individual genes and plant genotypes, available from the previous assessments of individual lines.  In addition, the assessment can consider information collected from agronomic or other observations that are normally done by breeders.  This is a process that has been considered, over many decades of experience, to provide an acceptable level of safety.

With all due respect I like to contradict the comparison of traditional breeding and genetic engineering. I think this comparison is a categorical fault. Traditional breeding techniques are based on crosses which can occur naturally and combine genes in their existing regulation networks and epigenetic context while genetic engineering has the potential and perceives its attractiveness out of the possibility to introduce new gene cassettes stemming from sources never able to exchange genetic material.  (see also definitions in the Cartagena Protocol). There are other laboratory based techniques which have also the potential to combine genomes or part of genomes which can not crossbreed naturally (like cell fusions “beyond taxonomic families, that overcome natural physiological reproductive or recombination barriers and that are not techniques used in traditional breeding and selection” Art.3 Use of terms, Cart-agena Protocol) . According to the Protocol these should be assessed with the same scrutiny as a GMO.
According to European law plants with stacked genes have to be assessed on a case by case approach. As Helmut Gaugitsch pointed out in his contribution there are a number of open questions and challenges when assessing plants with stacked genes. Especially the potential interactions – synergistic or antagonistic -  between stacked genes and of the stacked genes with their genomic background is methodologically spoken not an easy task to explore. Non GM parent lines, single GM lines and stacked lines have to be compared. In order to assess biological interactions (e.g. effects on non target organism) test designs have to be developed which are based on the use of the whole transgenic plant taking into account the potential receiving environment.
Risk assessors need reliable and comparable data. This needs some form of standardisation of test procedures. There is development work and guidance needed how to choose the right test organisms or develop the right test design, because the current ecotoxicological testing is based on single chemical compounds (pesticide testing). That is a useful starting point but does not suffice to assess complex organism with interacting possibly fortifying or modulating characteristics.

In response to comment 880, it is important to remember that traditionally bred varieties and transgenic varieties do not differ fundamentally in the nature of the risks they may present.  This widely recognized and acknowledged fact provides the foundation of risk assessment.  First laid out by the OECD1 after many years of expert deliberations, numerous scientific associations worldwide have re-affirmed this principle.  It is appropriately embraced in Annex III of the Cartagena Protocol, which states that the risks of LMO's should be "considered in the context of the risks posed by the non-modified recipients or parental organisms".  The assumption that traditional breeding combines genes in their pre-existing genetic backgrounds, thus assuring unperturbed gene interactions, is not supported by the facts.  There is an enormous body of scientific literature that interspecific and even intraspecific hybrids, an important tool in traditional breeding, often have reduced fitness.  This phenomenon is an indication that these genomes, with their co-adapted gene complexes, do not interact perfectly.  In fact, the combination and recombination of thousands of genes at a time disrupts regulatory patterns far more than does the introduction of a single gene.  This has been confirmed numerous times, with different lines of evidence, most recently by the comparison of metabolic and proteomic profiles between traditionally bred varieties and transgenic varieties (see references 2,3 and 4 for examples).  Despite the enormous variety of significant and sometimes drastic genetic manipulation in the course of traditional plant breeding, such breeding has (rightly) been regarded as so safe as to require no regulation in most countries. In fact plant breeders are manipulating thousands of genes routinely without knowing their individual effects or interactions a priori.  However, through several cycles of crossing, selection and testing in the field and lab, based on measured traits, productive high quality varieties have been delivered with an acceptable level of safety for centuries.  Such an approach can also be applied to transgenic plants.

This discussion of traditional breeding and its impacts serves to provide the baseline for comparison that is the starting point for risk assessment as provided in Annex III of the Cartagena Protocol.  The risk assessment that is carried out can then proceed under the framework of Annex III, which can be applied to both single as well as stacked genes. It is hard to imagine what additional information beyond what is already in Annex III would provide improvements in the risk assessment.
  1 See OECD 1986: Recombinant DNA Safety Considerations – Safety considerations for industrial, agricultural and environmental applications of organisms derived by recombinant DNA techniques. ISBN 92-64-12857-3; and OECD, 1993. Safety Evaluation of Foods Derived by Modern Biotechnology: Concepts and Principles. ISBN: 9789264138599. Paris.
2 Lehesranta, S.J. et al., 2005. Comparison of Tuber Proteomes of Potato Varieties, Landraces, and Genetically Modified Lines.  Plant Physiology 138: 1690–1699.

3 Catchpole, G.S. et al., 2005. Hierarchical metabolomics demonstrates substantial compositional similarity between genetically modified and conventional potato crops.  PNAS 102: 14458 –14462.

4 Batista, R. et al., 2008. Microarray analyses reveal that plant mutagenesis
may induce more transcriptomic changes than transgene insertion.  PNAS 105:  3640 –3645.