Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
2 When and Why to Consider Bioconfinement INTRODUCTION Safety is an issue for all modern technology--from the design of auto- mobiles to the delivery of information over the Internet--and confinement of technology is frequently an important aspect of ensuring safety. Airbags and radial tires promote drivers' safety; highly developed software pro- motes computer users' privacy and security. The atomic energy industry works to confine radiation; the chemical industry, to confine toxic chemicals; the food industry, to prevent nonedible industrial rapeseed from contami- nating canola oil for human consumption; and the horticultural industry, to confine ornamental plants that might become invasive. The stewardship of those industries bolsters public confidence in technology and its products at the same time as it prevents the damage that can result from the movement of products and byproducts into arenas for which they are not intended. The success of any new technology depends on this attention to safety and confinement and on building public trust by protecting human health, the environment, and the security of intellectual and financial information. Biotechnology should enjoy the same benefits from attention to safety and confinement. As most traditionally improved organisms pose few safety problems that require confinement, it is likely that most of the vast array of proposed genetically engineered organisms (GEOs) will pose little threat to public health or the environment, and they will require minimal confine- ment, if any. However, as some traditionally improved organisms require confinement, some GEOs will need some or substantial confinement. If 29
30 BIOCONFINEMENT OF GENETICALLY ENGINEERED ORGANISMS field release of a GEO is proposed, it is important to consider whether confinement is necessary, and, if so, how to attain it. This chapter reviews the concept of risk and then reviews some of the effects that could be expected from the release of GEOs. Avoiding or mini- mizing damage generally requires that specific methods of confinement be considered and that the consequences of failure (inadequate confinement performance) are predicted. This chapter also includes a discussion of who bears responsibility for deciding whether and how to confine GEOs and their engineered genes. WHAT IS RISK? In many aspects of its deliberations, the Committee on the Biological Confinement of Genetically Engineered Organisms found it necessary to discuss "hazard," "harm," and "risk." The terminology used here is consis- tent with past reports of the National Research Council (NRC, 1983, 1996, 2002b). To quote a recent report (NRC, 2002b), "as set forth by the NRC (1983, 1996), a hazard is an act or phenomenon that has the potential to produce harm, and risk is the likelihood of harm resulting from exposure to the hazard ... risk is the product of two probabilities: the probability of exposure and the conditional probability of harm, given that exposure has occurred." Risk assessment typically involves asking questions: What can go wrong? How likely is failure? What are the consequences of failure? How likely are those consequences (assuming the triggering event occurs)? How significant are those consequences? How certain are we about this knowledge, whether it is qualitative or quantitative (NRC, 1996)? Risk assessment involves (1) identifying potential harms, (2) identifying potential hazards that might produce the harms, (3) defining exposure and the likelihood of exposure, (4) quantifying the likelihood of harm given that exposure has occurred, and estimating the severity of the harm (NRC, 2002b). The importance of social values to this determination is discussed below. Risk is sometimes described as a formula: exposure multiplied by hazard. Although this can be useful shorthand, risk is not easily estimated. Uncertainties can arise from random events (including human error) in the physical world, lack of knowledge about the physical world, or lack of knowledge about the applicability of risk-generating processes (NRC, 1996). Ecological harm (consequences) is difficult to quantify, including damage to an ecosystem or the extinction of a species, for example. Moreover, evaluating harm requires consideration of social values that will define the significance of predicted consequences. Clearly, exact quantification is impossible. An evaluation of risk must consider cumulative risk--the combined risks to human health or the environment posed by exposure to multiple
WHEN AND WHY TO CONSIDER BIOCONFINEMENT 31 agents or stressors (US EPA, 2003a). Without cumulative risk analysis, it is impossible to assess the hazard associated with bioconfinement failure. For example, suppose failure would result in the release of a contaminant into the environment. The hazard stemming from that failure could depend on whether the environment is totally free of the contaminant, the contami- nant already is present in the environment at a near-hazardous concentra- tion, the environment already is stressed by other factors such that its normal resilience is compromised, or there are some other substances in the environment that would neutralize the contaminant or exacerbate its effects. The risk of any new technology should be considered in the context of preexisting relevant technologies. Doing so would likely involve an assess- ment of relative risks and while this comparison is an important one for consideration, it is also a challenging task. In the context of this discussion, relative risk is, in theory, equal to the probability of harm utilizing the new technology divided by the probability of harm utilizing the preexisting technology. One challenge is that, as indicated above, risk assessment is not readily susceptible to exact mathematical calculation. Another challenge is identifying the appropriate preexisting technology to serve as a compara- tor. Given the diversity of GEOs and the farming systems where they might be used, there is a substantial amount of disagreement as to what pre- existing agricultural technologies are appropriate to serve as comparators. Furthermore, given the range and number of potential variables and risks associated with the application of the technologies, identifying the relevant risks for comparison would have to be done on a case-by-case basis. For the bioconfinement technologies described in this report that have yet to be fully developed or applied, discussion of relative risks is simply premature at this time. Nonetheless, as developers of GEOs and the confinement methods progress such comparisons will be attempted but will not be easy to make. Given the uncertainty involved in risk analysis and the fact that many variables cannot be quantified, the committee determined that an alternate, less formulaic, model for risk assessment would be valuable. Figure 2-1 is a risk assessment matrix that assumes that a hazardous event has occurred.. The hazard of greatest concern is that with a high risk (probability of occurrence) and high significance or severity of harm (black area). Social, ecological, and economic considerations influence the significance of conse- quences. Depending on the quality of information available, the axes could consist of continuous values or more discrete categories (e.g., a 3×3 matrix of highmediumlow rankings; Miller et al., in press). Earlier reports of the National Research Council contain lengthy dis- cussions of risk and approaches to its analysis. There is extensive consider- ation of risk in the report Understanding Risk (NRC, 1996), and Environ- mental Effects of Transgenic Plants (NRC, 2002a) contains a detailed
32 BIOCONFINEMENT OF GENETICALLY ENGINEERED ORGANISMS High Harm of Severity Low Low High Risk (probability of harm) FIGURE 2-1 A risk assessment matrix. The horizontal axis denotes the risk, that is, the probability of occurrence of a harm, and the vertical axis denotes the sever- ity of the harmful consequence. discussion of risk analysis, including analyses of two-part and whole- organism models, and fault-tree and event-tree approaches. Discussions of risk often distinguish among three related processes: risk assessment, risk management, and risk characterization. The 1996 report sets forth an elabo- rate description of risk characterization, which it defines as "a synthesis and summary of information about a potentially hazardous situation that addresses the needs and interests of decision makers and of interested and affected parties and which is a prelude to decision making and depends on an iterative, analyticdeliberative process" (NRC, 1996). The committee recognizes those earlier discussions and suggests it would be useful to describe a systematic approach to risk assessment and management, with- out delving into the issue of risk characterization other than to note its relevance and importance, particularly with respect to transparency and public participation. Table 2-1 presents a systematic approach to risk assessment and man- agement. It considers exposure, hazard, risk reduction and prevention,
WHEN AND WHY TO CONSIDER BIOCONFINEMENT 33 TABLE 2-1 Systematic Risk Assessment and Management Step Key questions Hazard What event posing harmful consequences could occur? identification Risk analysis How likely is the hazard? What would be the harms from realization of the hazard, and how severe are they, taking into account social values? What is the risk assessment as shown on a matrix of risk (likelihood of harm) plotted against severity of harm; see Figure 2-1, above)? Each cell of the matrix should be accompanied by a qualitative assessment of the response and a quantification of assurance needed to reduce harm if the cell's conditions were to occur. How well established is the knowledge used to identify the hazard, estimate its risk, and predict harms? Risk reduction What can be done (including bioconfinement and other planning and confinement) to reduce risk, either by reducing the likelihood or implementation mitigating the potential harms? Are there steps that can be taken to prepare for remediation? Risk tracking How effective are the implemented measures for risk reduction? (monitoring) Are they as good as, better than, or worse than planned? What follow-up, corrective action, or intervention will be pursued if findings are unacceptable? Did the intervention adequately resolve the concern? Remedial action What remedial action should be taken? Transparency How transparent should the entire process be? How much and and public what type of participation should there be in the steps above participation (and in risk characterization) by the public at large, by experts, and by interested and affected parties? SOURCE: Adapted from Kapuscinski, 2001. monitoring, remedial action, transparency, and public participation (adapted from Kapuscinski, 2001). Confinement should be undertaken in the context of an integrated confinement system (ICS)--one that considers whether confinement might be necessary from the beginning of GEO development and that has redun- dancy of confinement as an operating principle (Chapter 6). Elements of
34 BIOCONFINEMENT OF GENETICALLY ENGINEERED ORGANISMS this approach include a clear strategy that is adopted in advance of field testing or release of the GEO. Adequate training of those who are respon- sible for managing the confinement strategy is essential. Permanent employ- ees should maintain accountability and consistency of standard operating procedures. For pharmaceutical-producing GEOs, good management prac- tices also will be necessary. An ICS should be subjected to periodic audit, review, adjustment, and reporting. Each ICS should be supported by a rigorous, comprehensive, and credible regulatory regime that includes mechanisms for inspection and enforcement. The risk assessment techniques discussed here would have been effec- tive in predicting some notable events that have already occurred. Consider the StarLink incident (Box 2-1). In this case a variety of corn was deregu- BOX 2-1 Confinement Failure: StarLink Corn The StarLink incident illustrates the failure of a regulatory-agencyindustry attempt to prevent genetically engineered material from entering the human food supply. The story offers lessons about the confinement of engineered genes and their products. Genes from Bt that produce various Cry proteins have been individually engi- neered into plants to confer resistance to insect pests. These are commonly called Bt crops. The most widespread use of genes for Cry proteins has been to confer resistance to lepidopteran pests in maize. The trade name StarLink was applied to maize varieties produced by Aventis CropScience that were engineered to synthesize the Cry9 protein for resistance to the European corn borer. The Cry9 protein was found to be more heat resistant than other Cry proteins and potentially harder to digest based on in vitro studies, suggesting possible allergenicity (US EPA, 1999). Because of the inconclusive tests regarding possible allergic reactions in humans, the US EPA granted StarLink varieties a "split registration" in 1998, designating it for animal feed and industrial use, but not for human consumption (US EPA, 1998). In September 2000, the Washington Post reported that traces from transgenes of StarLink corn had been detected in the human food supply by an independent laboratory for Genetically Engineered Food Alert, a coalition of environmental and food safety organizations (Kaufman, 2000). Soon thereafter StarLink Bt gene was found in a great variety of products intended for human consumption, in the United States as well as in Japan and South Korea--major United States maize buyers. The effect on the United States food industry was substantial, ranging from the recall of food products to the temporary shutdown of grain mills (Taylor and Tick, 2003). Aventis voluntarily withdrew its registration, and to prevent further mixing of StarLink material into the human food supply, it agreed to a buyback program with USDA. The US EPA announced that it would no longer grant split registrations for GEOs (Taylor and Tick, 2003).
WHEN AND WHY TO CONSIDER BIOCONFINEMENT 35 lated with the proviso that it was not permitted to enter the human food supply, and yet it did so. If the following factors were initially considered-- the possibility of pollen dispersal, retention of prior genotypes in the soil seed bank, and the mixing of corn seeds in facilities that harvest, transport, store, and process corn seedit would have been obvious that without very stringent procedures confinement failure was inevitable. CONCERNS The committee adopted the definition of "concern" used in the report Animal Biotechnology: Science-based Concerns (NRC, 2002b). "Concern" is used here to mean an "uneasy state of interest, uncertainty, and appre- The potential sources of the widespread presence of StarLink material in the human food supply are many. Opportunities for mixing genes, seeds, and maize products start in the field and end at the store. Cleaning the facilities that harvest, transport, and store maize seed does not necessarily remove every seed. Pro- cessing also presents an opportunity to commingle products. If a farmer does not rotate crops, ungerminated seed from an earlier planting or seed that dropped from another year's crop could germinate and mix with the current year's variety. StarLink genes also were found in varieties into which it had not been engineered, strongly suggesting unintended cross-pollination from StarLink maize. It also is clear that some farmers did not comply with the terms of US EPA registration: Some admitted that they had sold their maize for human food, others stated that they did not know how their maize would be used when they sold it (Pollack, 2001). Although StarLink corn caused no proven adverse human health effects, other effects have been substantial. Despite the fact that StarLink represented less than 1% of the U.S. maize crop, millions of dollars were spent to buy back grain, con- sumer confidence in GEOs was shaken, and international trade relationships became strained. Once a gene escapes into populations for which it is unintended, the consequences can be enormous. On December 27, 2002, more than two years after Aventis stopped selling StarLink seed, the Washington Post reported that traces of StarLink material were detected in a shipment of U.S. maize on its way to Tokyo for use in food products (Fabi, 2002). More extensive information on the StarLink incident is presented in Post-Market Oversight of Biotech Foods (Taylor and Tick, 2003), which includes a substantial chronology of the incident. Each of the following publications concentrates on a different aspect of the incident: The StarLinkTM Situation (Harl et al., 2001), StarLink: Impacts on the United States Corn Market and World Trade (Lin et al., 2001), and Channeling, Identity Preservation and the Value Chain: Lessons from the Recent Problems with StarLink Corn (Ginder, 2001).
36 BIOCONFINEMENT OF GENETICALLY ENGINEERED ORGANISMS hension" (NRC, 2002b). Ethical concerns--a subset--are discussed in Chapter 1. The committee attempted to base its analyses of risk and concern on current natural and social scientific knowledge. As with its consideration of risk, the criteria include the likelihood that a given result will occur; its severity; and its significance, as would be determined at least in some measure by public opinion or societal values. Potential Effects Genetically engineered organisms are controversial. Dozens of scholarly reports have identified, reviewed, and evaluated the realized and potential effects of their use, particularly transgenic crops and fish (e.g., Carpenter et al., 2001; Colwell et al., 1985; Cook et al., 2000; Dale et al., 2002; Hails, 2000; Kapuscinski and Hallerman, 1991, 1994; Keeler and Turner, 1990; Marvier, 2001; McHughen, 2000; NRC, 1989b, 2000, 2002a, 2002b; Pew Initiative on Food and Biotechnology, 2003; Rissler and Mellon, 1996; Scientists' Working Group on Biosafety, 1998; Snow and Moran-Palma, 1997; Tiedje et al., 1989; Traynor and Westwood, 1999; Winrock Inter- national, 2000; Wolfenbarger and Phifer, 2000). As GEOs are developed and released, their specific effects can be scrutinized. Current and Future GEOs: A Brief Summary The first commercially grown transgenic plant--the Flavr SavrTM tomato (Kramer and Redenbaugh, 1994)--was released to the field a little more than a decade ago. Since that time, millions of acres have been planted in genetically engineered crops--mostly in the United States. Soybean, corn (maize), canola (oilseed rape), and cotton are the major species used world- wide, and most of those plants are restricted to three phenotypic classes: herbicide resistance, insect resistance, and viral resistance (James, 2002). Despite the narrow range of crops and phenotypes and despite the fact that most of the acreage is restricted to a few countries, their rapid acceptance is remarkable in the history of agriculture, outpacing even the acceptance of corn hybrids (James, 2002). Not all field-grown plants that are used to produce commercial products have been deregulated. In the United States, some products have been produced from transgenic crops and viruses in field tests that are regulated under the United States Department of Agricul- ture (USDA) Animal and Plant Health Inspection Service (APHIS) notifica- tion and permit processes (NRC, 2002a). Transgenic animals are largely under development, and some have been commercialized. In a few cases, transgenic mammals have been created to secrete commercial biochemicals (http://www.nexiabiotech.ca/en/01_tech/
WHEN AND WHY TO CONSIDER BIOCONFINEMENT 37 01.php). Those genetically engineered animals that have been commercial- ized have not been released to the field. Some microorganisms have achieved commercial success with applications in medicine and in food production. As yet, they have been used in industrial situations, rather than in situations calling for field release. For example, chymosin is the active enzyme in rennet, which is used in cheese production. The gene for this enzyme has been engineered into bacteria and yeast, and it now contributes substan- tially to the commercial production of hard cheese in Europe and North America (Mohanty et al., 1999). Prior to genetic engineering, chymosin was obtained exclusively from rennet extracted from the stomachs of calves. Benefits beyond the commercial success of transgenic crops also have been noted. In the United States, insect resistance and herbicide tolerance sometimes have permitted reduced pesticide use and have increased yields (Benbrook, 2001; Fernandez-Cornejo and McBride, 2002). The gains in yield and reductions in pesticide use have been modest in developed coun- tries but more substantial in developing nations (e.g., Qaim and Zilberman, 2003). Herbicide resistance probably has facilitated the increasing practice of "no-till" agriculture, which replaces mechanical weed control with chemi- cal weed control, resulting in reduced soil erosion--but the details of how and whether herbicide-resistant plants have improved soil quality require more study (Wolfenbarger and Phifer, 2000). The list of potential benefits attributable to transgenic plants, animals, and microorganisms is long and diverse, and it is difficult to name all of the beneficial phenotypes that have been proposed for transgenic organisms. However, some expected benefits have been discussed widely. One is the dramatic growth enhancement in several lines of transgenic fish that are approaching commercialization (NRC, 2002b). Also, some traits related to coping with abiotic stresses have been targeted, such as drought and salinity tolerance in crops (e.g., Garg et al., 2002) and cold tolerance in fish (e.g., Wang et al., 1995). Some phenotypes have been proposed to make food healthier by increasing its nutritional value, by eliminating allergens and antinutritional factors, and by extending the shelf life of fruits and veg- etables. Other phenotypes pertain to nonfood products processed from plants, such as the alteration of the chemical composition of wood fibers from trees to reduce the cost of paper production, and the production of expensive or hard-to-synthesize specialty chemicals, such as pharmaceuticals in transgenic crop plants (Pew Initiative on Food and Biotechnology, 2002), aquatic plants and moss (Spencer, 2003; Wagner, 2003), algae (Mayfield, 2003), and fish (Aquagene, 2003). Viruses and microorganisms that are natural pathogens of insects and other pests could be engineered into more effective agents of biological control. Transgenic plants, microorganisms, and algae have been proposed for use in environmental remediation to detoxify polluted soils and waters (Cai et al., 1999).
38 BIOCONFINEMENT OF GENETICALLY ENGINEERED ORGANISMS Some of the proposed uses are novel; others are natural extensions of current use. Nontransgenic bacteria already are used to degrade environ- mental pollutants. Table 2-2 lists some GEOs and phenotypes that have been proposed, are under development, or are in use. Only time will reveal which predicted benefits of those organisms are realized, and a thorough discussion of those benefits is beyond the scope of this report. Gene Dispersal and Persistence Before considering specific categories of GEOs that may be candidates for bioconfinement, it is useful to examine how genes and organisms move, and whether transgenes are likely to be favored by natural selection. The ecological spread of transgenes is accomplished by horizontal transfer, dis- persal of whole organisms, or dispersal of gametes that may move transgenes past the point of intentional release and into new environments and organ- isms. The persistence of transgenes depends largely on how they affect the organisms' evolutionary fitness, as discussed below. How Transgenes Disperse Horizontal transfer occurs when genetic material from one organism becomes incorporated into an unrelated organism (NRC, 2002a); it is natu- rally occurring genetic engineering. Gene transfer among unrelated bacteria is relatively common (Syvanen, 2002); transfer among unrelated eukaryotes happens less frequently. The rate of horizontal transfer that is detected as evolutionary change is extremely low--it is far less the mutation rate-- although it is surprisingly fast over evolutionary time. And some plants, for example, appear to have acquired genes from other types of organisms, even from other kingdoms (Adams et al., 1998). And some flowering plants apparently have acquired mitochondrial genes from fungi hundreds of times over the past 100 million years (Palmer et al., 2000). Horizontal transfer is largely viewed as a source of unanticipated con- sequences. Transgenic plant DNA can transform bacteria in sterile soil microcosms (Nielsen et al., 2000). There is no data yet, however, to suggest that the rate of horizontal transfer would be any faster for transgenic organisms (Syvanen, 2002). But relevant data remain few (Nielsen et al., 1998); further research to identify the mechanisms and ecological implica- tions of horizontal gene transfer (Traavik, 2002) would be helpful. Dispersal of whole organisms includes the movement of juvenile and adult animals, fertilized eggs and seeds, spores, and vegetative propagules such as offshoots and fragments of plants and algae. Dispersal can occur passively--in the same ways pollen and seeds are transported, for example-- or by anthropogenic means: during international trade, in the ballast water
WHEN AND WHY TO CONSIDER BIOCONFINEMENT 39 TABLE 2-2 Genetically Engineered Organisms Species Engineered Trait Application Development Finfish Mud loach Increased growth rate, Aquaculture Research improved feed conversion, and for human likely sterility after insertion of food mud loach growth hormone driven by mud loach -actin regulatory region (Nam et al., 2001a,b) Channel catfish Enhanced bacterial resistance Aquaculture Research after insertion of moth peptide for human antibiotic, cecropin B gene food (Dunham et al., 2002) Medaka Facilitation of better detection Industrial Research; of mutations (presumably caused uses; method has by environmental pollution) after environmental been patented insertion of a bacteriophage uses vector (serves as a mutational target). After exposure to mutagenic agent, vector DNA is removed and inserted into indicator bacteria where mutant genes can be easily measured (Winn, 2001a, b; Winn et al., 1995, 2000, 2001) Atlantic salmon Increased growth rate and food Aquaculture Method has conversion efficiency after for human been patented; insertion of Chinook salmon food seeking Food growth hormone gene that and Drug operates year-round, thereby Administration fostering steady growth through approval the year rather than summer growth (Cook et al., 2000; Hew and Fletcher, 1996) Red sea bream Increased growth rates after Aquaculture Research insertion of an "all fish" growth for human hormone consisting of ocean food pout antifreeze protein gene promoter and Chinook salmon growth hormone (Zhang et al., 1998) continued
40 BIOCONFINEMENT OF GENETICALLY ENGINEERED ORGANISMS TABLE 2-2 Continued Species Engineered Trait Application Development Rainbow trout Improved carbohydrate Aquaculture Research metabolism after insertion of for human human glucose transporter type I food; and rat hexokinase type II, Industrial cloned with viral (CMV) and uses piscine (sockeye salmon metallothionein-B and histone 3) promoters. Potentially allows giving fish feed that contains plant materials (Pitkänen et al., 1999) Steelhead trout Increased growth rate and food Aquaculture In use as a conversion efficiency via for human research model insertion of sockeye salmon food growth hormone gene (Devlin et al., 2001) Zebrafish Production of male-only Biological Research; offspring by injecting into fish control of In use as a eggs an altered gene that aquatic research model prevents aromatase enzyme from nuisance transforming reproductive species, hormone androgen into estrogen; such as lack of estrogen prevents carp development of female fish (Woody, 2002) Carp Improved disease resistance after Aquaculture Research insertion of a human interferon for human gene (Zhu, 2001) food Goldfish Increased cold tolerance after Aquaculture Research insertion of ocean pout antifreeze for human protein gene (Wang et al., 1995) food Tilapia Increased growth rate and food Aquaculture Seeking conversion efficiency after for human regulatory insertion of tilapia growth food approval hormone gene (Martinez et al., 2000) Tilapia Production of clotting factor Pharmaceutical Research after insertion of human gene for production clotting factor VII, for medicinal applications (Aquagene, 2003)
WHEN AND WHY TO CONSIDER BIOCONFINEMENT 41 TABLE 2-2 Continued Species Engineered Trait Application Development Tilapia Increased growth rate, food Aquaculture Research conversion efficiency, and for human utilization of protein after food insertion of chinook salmon growth hormone with ocean pout antifreeze promoter (Rahman et al., 2001) Mollusks Mollusks Potential improved disease Aquaculture Research; resistance and growth for human method has acceleration in mollusks by food been patented harnessing altered genetic material from a virus to introduce foreign DNA (Burns and Chen, 1999). Oysters Improved disease resistance by Aquaculture Research introduction of retroviral vectors. for human Researchers are determining most food effective method of insertion (Burns and Friedman, 2002; Lu et al., 1996) Marine Plants Seaweed Enhanced production of Industrial Research; carrageenan or agar (valuable uses method has to the food, pharmaceutical, been patented cosmetic industries) after introduction of foreign DNA (Cheney and Duke, 1995) Micro algae Potential improved nutritional Aquaculture Research (Spirulina) and medicinal value of for human commonly consumed Spirulina. food Method to achieve such trait changes recently confirmed via successful integration and expression of a genetically engineered marker gene (Zhang et al., 2001) continued
42 BIOCONFINEMENT OF GENETICALLY ENGINEERED ORGANISMS TABLE 2-2 Continued Species Engineered Trait Application Development Algae Enhanced ability to bind heavy Bioremedial Research metals after successful use expression of a foreign class-II metallothionein (chicken MT-II cDNA) (Cai et al., 1999) Marine Microorganisms Diatoms Reduced dependence on light for Industrial Research growth after insertion of human uses gene for biochemical involved in metabolism of sugar (Zaslavskaia et al., 2001) Crustaceans Crayfish Production of transgenic Aquaculture Research; offspring (in crayfish and live- for human in use as a bearing fish) after injection, in food research model parents' gonads, of replication- defective pantropic retroviral vector. Successful transgenic individuals expressed neomycin phosphotransferase gene (neoR) (Sarmasik et al., 2001) Kuruma prawns Potential improved growth rate Aquaculture Research through gene insertion. for human Researchers are currently food inserting marker genes to confirm most appropriate method (Preston et al., 2000) Terrestrial Plants Corn Production of a glycoprotein, Industrial Field grown avidin (Hood et al., 1997; uses, commercially NRC, 2002a) including under APHIS medical notification diagnostic procedure procedures Yellow Transformed with coat protein Agriculture Deregulated crookneck genes of three viruses to gain for human and squash resistance against them food commercialized (Schultheis and Walters, 1998)
WHEN AND WHY TO CONSIDER BIOCONFINEMENT 43 TABLE 2-2 Continued Species Engineered Trait Application Development Flax Sulfonylurea herbicide resistance Agriculture Deregulated based on a gene from for oil and and Arabidopsis thaliana oilseed commercialized (McHughen et al., 1997) production Tomato Flavr SavrTM; slowed ripening Agriculture Deregulated, by the introduction of an for human no longer antisense sequence of food commercially polygalacturonase gene from available tomato (Kramer and Redenbaugh, 1994) Rice Enhanced -carotene production Agriculture Research from phytoene synthase and for human lycopene -cyclase genes both food with introduced from Narcissus enhanced pseudonarcissus (Beyer et al., nutritional 2002) value Poplar Introduced bacterial gene confers Managed Field tests resistance to the herbicide forestry under APHIS glyphosate (Meilan et al., 2002) notification Loblolly pine Resistance to pine caterpillars Managed Research (Dendrolimus punctatus and forestry Crypyothelea formosicola) conferred by Bacillus thuringienesis. CRY1Ac insecticidal protein gene (Tang and Tian, 2003) Walnut Reduced flavonoid content and Nut Research enhanced adventitious root production formation conferred by a walnut antisense chalcone synthase gene (El Euch et al., 1998) Plum Resistance to plum pox virus Fruit Release permit using capsid gene (Ravelonandro production issued et al., 1998) Bartlett pear Resistance to fireblight bacterial Fruit Field tested disease conferred by a synthetic production under APHIS antimicrobial gene (Puterka et notification al., 2002) (20002001) continued
44 BIOCONFINEMENT OF GENETICALLY ENGINEERED ORGANISMS TABLE 2-2 Continued Species Engineered Trait Application Development Apple Resistance to apple scab disease Fruit Field tested conferred by antifungal chitinase production under APHIS genes from the biocontrol fungus notification Trichoderma atroviride (19982002) (Bolar et al., 2001) Papaya Resistance to papaya ring spot Fruit Deregulated Virus conferred by coat protein production for commercial gene (Cheng et al., 1996) use Banana Resistance to banana leaf spot Fruit Research disease conferred by the frog production Xenopus laevis gene for the antimicrobial peptide magainin (Chakrabarti et al., 2003) Microbes Pseudomonas Improved protection against Biocontrol Small-scale putida soil-borne pathogens after of soil-borne field test insertion of genes for production pathogens of antifungal compound phenazine-1-carboxylic acid or the antifungal and antibacterial compound 2,4-diacetylpholoroglucinol (Bakker et al., 2002; Glandorf et al., 200a) Pseudomonas Chromosomal insertion of two To identify Small-scale fluorescens reporter gene cassettes (lacZY effects on field test and Kanr-xylE) (De Leij et al., indigenous 1995) microbial populations in wheat Fluorescent Chromosomal insertion of lacZ To monitor Field testing pseudomonads and Kanr (De Leij et al., 1995) movement of from wheat genetically rhizosphere engineered bacteria in soil Agrobacterium A transfer-deficient mutant of Biocontrol Commercial tumefaciens the natural isolate (K84) was of crown use K1026 constructed that prevents gall disease transfer of the plasmid conferring resistance to toxin (Jones et al., 1988)
WHEN AND WHY TO CONSIDER BIOCONFINEMENT 45 TABLE 2-2 Continued Species Engineered Trait Application Development Metarhizobium Addition of gfp gene and/or Biocontrol Field testing anisopliae additional protease genes of insect (Hu and St. Leger, 2002) pathogens of plants Pseudomonas Ice nucleation-negative (ice Prevent frost Research syringae minus) mutants constructed injury to (Wilson and Lindow, 1993) plants Aspergillus niger Production of bovine chymosin Cheesemaking Commercial Colletotrichum Virulence to weeds Mycoherbicide Research and coccodes development Escherichia coli Production of human insulin by Treatment of Commercial cloned gene diabetes in production humans Insects Cochliomyia GFP in PiggyBac (transposon Improved Research hominovorax that inserts gene sequence of a pest control desired trait in the TTAA gene sequence of the insect) (Handler and Allen, unpublished) Culex pipiens GFP in Hermes transposon Reduced Research (Allen et al., submitted) vector competence Aedes aegypti GFP and rescue of eye pigment Reduced Research pathways in Hermes, Mariner, vector piggyBac (Coates et al., 1998; competence Jasinskiene et al., 1998) Tribolium sp. Fluorescent proteins, Genetic Research "informational molecules" in research PiggyBac (Berghammer et al., 1999) Bombyx mori GFP, human collagen in Improved or Application PiggyBac (Tamura et al., 2000) modified silk, in laboratory disease resistance continued
46 BIOCONFINEMENT OF GENETICALLY ENGINEERED ORGANISMS TABLE 2-2 Continued Species Engineered Trait Application Development Anopheles Fluorescent proteins in PiggyBac Reduced Research gambiae (Benedict, unpublished) vector competence Anopheles Eye color mutant rescue in Reduced Research stephensi Minos transposon (Catteruccia vector et al., 2000; Ito et al., 2002) competence Pectinophora Fluorescent proteins in PiggyBac Improved Research and gossypiella (Peloquin et al., 2000) pest control, field trials heterologous protein expression in mass-reared insects Anastrepha Fluorescent proteins in PiggyBac Improved Research suspensa (Handler and Harrell, 2001) pest control Musca Fluorescent proteins in PiggyBac Genetic Research domestica (Hediger et al., 2001) research Ceratitis White eye mutant rescues, Improved Research capitata Fluorescent proteins in PiggyBac, pest control Minos (Handler et al., 1998; Loukeris et al., 1995) of commercial ships, and in unintentional spilling of seed during transport from harvest to market (NRC, 2002a; Scientists' Working Group on Bio- safety, 1998). In the United Kingdom, some roadside feral oilseed rape populations (Brassica napus) apparently are replenished by seed that spills from vehicles on their way to an oilseed crushing plant (Crawley and Brown, 1995). It is difficult to quantify dispersal of whole organisms in detail. None- theless, dispersal can cover remarkable distances. For example, bird watchers in North America annually report sightings of dozens of individuals that are otherwise native to Europe. Hundreds of invasive species have success- fully colonized new regions after unintentional and deliberate anthropo- genic dispersal over hundreds or even thousands of miles (e.g., Mack and Eisenberg, 2002; Rosenfield and Mann, 1992; Williams and Meffe, 2000). Gamete dispersal provides an opportunity for the sexual transfer of transgenes to wild or domesticated relatives of the transgenic organism. For
WHEN AND WHY TO CONSIDER BIOCONFINEMENT 47 example, in the case of crop plants, partially or fully sexually compatible relatives would include other varieties of that crop, related crops, and wild relatives (NRC, 2002a). Virtually all farmed fish and shellfish lines can breed readily with other captive lines and with wild relatives. Hybridization of closely related fish and shellfish species is relatively common (Collares- Pereira, 1987; Turner, 1984); it occurs in at least 56 families (Lagler et al., 1977), and it frequently yields fertile hybrids. Motile gametes are typically associated with male function, such as in the sperm cells of pollen produced by seed plants and in animal sperm. The means by which gametes disperse varies by organism. For example, for insects and some vertebrates, sperm are delivered to the female by insemination. But, for seed plants and many aquatic animals (fish and shellfish), gametes are released independently of the paternal parent and either contact an egg under their own power, or require a vector. Wind and insects are agents that most often carry pollen from one seed plant to another; water often serves as the vector for marine and freshwater organisms that release their gametes. Some fraction of gametes released into the environment is expected to disperse, regardless of whether the species is largely outcrossing or mostly self-fertilizing. Bread wheat plants are highly self-fertilizing but can mate with plants some dis- tance away (Hucl and Matus-Cádiz, 2001). Although some crops typically are harvested before they flower, occasionally the plants flower prematurely (Longden, 1993) or are missed by harvesting equipment, and eventually flower. Only a very few domesticated plants--such as some potato varieties and some ornamental plants--are completely male-sterile and produce no pollen. The dispersal of gametes leading to successful fertilization rarely has been measured directly. The best data, which come from numerous experi- mental and descriptive studies of plants, show that it is not unusual for 1% or more of the seeds in a population to be sired by plants 100 m or even 1000 m distant (reviews by Ellstrand, 1992, 2003a list several of those studies). Evolutionary Persistence of Transgenes When left to their own dispersal devices, organisms, their genes, or both can move into locations for which they were not intended and then multiply. Population genetics theory can be applied to predict the fate of transgenes because the consequences of introducing a new transgene are essentially the same as are those for any new immigrant allele. An immi- grant allele's fate--whether its frequency increases, decreases, or stays the same--depends on several factors, the most important of which are the evolutionary fitness effects of that allele in its new population and the rate of recurring immigration (Slatkin, 1987). Other factors that could affect an
48 BIOCONFINEMENT OF GENETICALLY ENGINEERED ORGANISMS immigrant allele's expression in a population include counterbalancing gene flow from other sources, changes over time in the environment that alter the allele's influence on fitness, chance effects if the allele is introduced in a very low frequency, and if the population into which it is introduced is small (fewer than 100). Generally, if an allele confers a fitness advantage when introduced into a population, it is expected to increase in frequency, even if it is introduced only once (Wright, 1969). If an allele has no effect on fitness, and if it is introduced just once, its frequency will be static. That is, if a single immi- gration event results in a 10% frequency of the new allele, that frequency would be maintained indefinitely. If the neutral allele is introduced repeat- edly, however, its frequency eventually should evolve to match that of the source population: If the allele frequency in the source population is 100%, the frequency in the sink population eventually should be the same (Wright, 1969). Finally, if an allele is detrimental to an individual's fitness, the allele will go extinct in the new population if there is just a single instance of immigration. "Detrimental" is meant to apply to generation-to-generation demographic contributions of a genotype and not to any individual fitness component. If immigration is recurrent, the allele will remain as a poly- morphism in the population, maintained by a balance between selection and gene flow (Wright, 1969). The factors that might affect this result are the same as those that determine fitness effects. A basic understanding of gene dispersal and the fitness effects of specific transgenes is essential for evaluating whether bioconfinement is needed and for developing effective bioconfinement methods. The committee intentionally did not hazard to guess what the fitness effects of classes of transgenes might be in recipient populations because it is already clear that generalizations might be difficult to obtain. For example, experimental field studies have already shown that different pest- resistance transgenes introgressed into wild sunflower have drastically dif- ferent fitness impacts (Burke and Rieseberg, 2003; Snow et al., 2003). Furthermore, these fitness impacts appear to vary with biotic and abiotic factors. Some Concerns about Field-Released GEOs Although most GEOs are expected to carry little or no risks to human health or the environment, several categories of potential risk have been identified by United States regulatory agencies and others. An understand- ing of the potential problems is relevant to the efficacy of bioconfinement methods that are intended to alleviate them. The discussion of concerns about the release of transgenic organisms has focused primarily on three
WHEN AND WHY TO CONSIDER BIOCONFINEMENT 49 broad categories of environmental risk: consequences of movement of a transgene or transgenic organism into a population, into a community of wild species, or into a location for which it was not intended; effects of the transgene protein product on other organisms in the ecosystem, such as engineered plant pesticides that could harm nontarget organisms; and evo- lution of resistance in targeted pests and pathogens. Those categories can apply to conventionally-improved domesticated plants and animals (NRC, 1989b, 2000, 2002a, b). And in fact any new genotype--transgenic or not--can create concerns that are unique to that genotype (NRC, 2002a, b). The first category obviously invites discussion of the need for confine- ment, including bioconfinement. Although less obvious, there could be circumstances under which bioconfinement would help reduce the chances of occurrences in the other two categories. The movement of transgenes does not, in itself, constitute a risk (NRC, 1987, 2002a, b); it does however constitute the "exposure" component of a risk if a specific hazard is associ- ated with that spread (NRC, 2002a, b). Three concerns dominate the discussion about the unintended move- ment of transgenes. The first is whether transgenes will confer a benefit to the transformed organism itself or to weedy or invasive relatives, resulting in the evolution of weeds that are difficult to control or in the evolution of new invasive lineages that overrun and disrupt natural ecosystems. The second issue is the question of whether the wild relatives of transgenic organisms will suffer an increased risk of extinction because of hybridiza- tion with or competition from those organisms. A third issue is whether transgenes will spread to other domesticated varieties and whether this could lead to health, environmental, or regulatory concerns. Those con- cerns are directly relevant to decisions about bioconfinement. Weediness or Invasiveness The most publicized concern associated with transgene dispersal in plants is the evolution of weediness or invasiveness, particularly as a result of the sexual transfer of crop alleles to wild relatives (e.g., Ellstrand, 1988; Goodman and Newell, 1985; NRC, 2002a; Snow and Moran-Palma, 1997). It is not unusual for natural hybridization to occur when domesticated and wild relatives live in proximity (Ellstrand et al., 1999; Rhymer and Simberloff, 1996). In the United States, more than half of the top 20 crops are known to naturally hybridize with their wild relatives (Ellstrand, 2003a). One could imagine that genes engineered to confer pest resistance or other- wise increase fitness (such as herbicide resistance or tolerance to abiotic stresses) could contribute to the evolution of increased weediness, especially if the genes were to escape to an organism that already is a weed (the noxious weed johnsongrass is a close relative of the crop plant sorghum).
50 BIOCONFINEMENT OF GENETICALLY ENGINEERED ORGANISMS The problem is not unique to transgenics; hybridization between conven- tional crops and their wild relatives is known to have led to the evolution of increased weediness and invasiveness in several cases (Ellstrand, 2003a). The transgenic organism itself could become an environmental prob- lem if the transgenic traits it expresses alter its ecological performance such that it becomes an invasive or nuisance species. Many crop plants--espe- cially those that have had a long history of domestication--pose little hazard because traits that make them useful to humans also often reduce their ability to establish feral populations either in agroecosystems or in non- agricultural habitats (NRC, 1989b). However, feral and naturalized popu- lations are well known for some crops and domesticated animals, and in some cases those populations could become more problematic as a result of their acquiring new transgenic traits. In the United States, forage grasses, turf grasses, alfalfa, and many horticultural species have established free- living weedy populations. Escaped cats, pigs, dogs, and goats have become "feral and resulted in environmental disruptions" in many parts of the world (NRC, 2002b). In the same way, problems have occurred from fish or shellfish species that escape from aquaculture operations (Bartley et al., 1998; Carlton, 1992; Courtenay and Williams, 1992). Introduced tilapia have displaced native fish in African, Asian, and American aquatic eco- systems (Lever, 1996; Lowe-McConnell, 2000), and fish farm escapees are the putative cause of the upstream spread of two Asian species, black carp and silver carp, in the Mississippi River basin (Naylor et al., 2001). If transgenes confer the ability to overcome factors that limit wild popula- tions, the resultant genotype might be significantly more weedy or invasive than is its nontransgenic progenitor. The factors that limit the invasiveness of populations are not well understood (e.g., Parker et al., 1999). An allele that confers a fitness advan- tage will spread through a population, but it will not necessarily result in the evolution of invasiveness. Thus, the mere presence of a transgene should not be taken as certainty that the invasiveness of a population has been altered. Many crops are unlikely to become weedier by the addition of a single trait (Keeler, 1989). In a few cases, however, the consequences might be obvious. The evolution of herbicide resistance in a weed population that previously was controlled by that herbicide will force new consideration of options for its control. Extinction of Wild Taxa The spread of one taxon sometimes overwhelms related, locally rare taxa, either by competitive displacement or by hybridization, thus increas- ing the probability of extinction (e.g., Levin, 2003). The fraction of hybrids produced by the rare population can be so high that the population becomes
WHEN AND WHY TO CONSIDER BIOCONFINEMENT 51 genetically absorbed into the common species (genetic assimilation; Ellstrand and Elam, 1993). Also, hybrids can suffer from reduced fitness (because of outbreeding depression), and the rare species might be unable to maintain itself. For example, spontaneous hybridization between nontransgenic crops and their wild relatives has been implicated in the disappearance of wild coconuts (Harries, 1995) and in the genetic dilution of California's wild walnut populations (Skinner and Pavlik, 1994). Depressed fitness or local extinction of wild fish populations has resulted from introgressive hybrid- ization between an introduced population, often derived from fish farms or hatchery-stocking programs, and a local, genetically distinct wild popula- tion (e.g., Hallerman, 2003; Kapuscinski and Brister, 2001; Utter, 2003). If the intended phenotypic effect of genetic engineering permits a GEO to be grown more closely to a wild relative than previously--for example, because it now can better tolerate an environmental stress (saline soil)--the previ- ously isolated species then would be subject to increased interbreeding, which would increase the probability of extinction of the wild population by hybridization. In many other cases, however, GEOs are not expected to exacerbate problems with the conservation of endangered wild relatives. Gene Flow to Other Domesticated Organisms The scientific literature has given scant attention to the risk that attends the movement of transgenes from one managed population to another. Hybridization among different transgenic varieties of the same species can lead to the unintended natural "stacking" of transgenes. This already has happened: Hybridization among three canola varieties--each resistant to a different herbicide--has led to the evolution of triple-herbicide-resistant crop volunteers in Canada that are now more difficult to control than were volunteer plants in the past (Hall et al., 2000). Crops that are engineered to produce pharmaceutical or other industrial compounds can cross-pollinate with the same species grown for human consumption, with the unantici- pated result of new chemical components in the human food supply (Ellstrand, 2003b). The same issue could apply in the future to gene flow to edible algae from transgenic algae that are created to produce inedible compounds (Minocha, 2003; Zhang et al., 2001). Given current research and develop- ment on transgenic fish used to produce pharmaceuticals (Aquagene, 2003), the scenario could extend to fish species also grown for human consumption. A recent case illustrates the point: soybeans intended for market were contaminated by volunteer maize that had been transformed to create a pharmaceutical compound (USDA, 2002). Cross-pollination was not neces- sary for the risk to be realized: a persistent seed bank was sufficient. Another consequence of unintentional movement of transgenes among managed populations is the transfer of transgenes into crops or other organisms that
52 BIOCONFINEMENT OF GENETICALLY ENGINEERED ORGANISMS are intended to be "transgene-free," such as crops that are to be certified organic or sold to an international market that prohibits the sale of transgenic products. If zero tolerance for nontransgenic ingredients is required by the marketplace, the presence of transgenes in crops or crop products intended to be transgene-free could pose an economic hardship to the grower. Similar concerns apply to organic aquaculture, which is gaining interest and activity (Brister, 2001; Tacon and Brister, 2002), or to the movement of an animal transgene into a crop intended for consumption by vegetarians. Ethical aspects are discussed in Chapter 1. EFFECTS ON NONTARGET SPECIES Bioconfinement methods also could be used to prevent unintentional damage to nontarget species, as could occur when a transgene that is designed to interfere with the growth or viability of a pest species could alter other species nearby. Bacillus thuringiensis (Bt) corn has been devel- oped to control the European corn borer and the southwest corn borer, and corn plants disperse Bt pollen. Reports of the potentially toxic effects of Bt corn pollen eaten by monarch butterfly larvae (Losey et al., 1999) captured widespread attention, in part because the butterfly is so well known. The effects of Bt pollen on monarch mortality appear to be highly variable, depending on factors such as the density of the Bt corn pollen, the Bt genotype in the crops, and other environmental conditions (Sears et al., 2001; Wraight et al., 2000). It is now clear that current commercial varieties of Bt corn are not particularly toxic to monarch butterflies, but it also is clear that some transgene products could harm organisms not intended for control. Nontarget effects of crop resistance alleles that have naturally introgressed into wild populations have not been well researched (Ellstrand, 2003a). Nonetheless, if transgenes or transgenic organisms designed to be toxic to pests or pathogens move into locations or populations for which they were not intended, they could harm organisms other than the intended pest species. DELAYING THE EVOLUTION OF RESISTANCE Insects, weeds, and microbial pathogens frequently have evolved resis- tance to the controls used against them (Barrett, 1983; Georghiou, 1986; Green et al., 1990). As with conventionally bred domesticates, resistance evolution can occur in pests targeted for control by or associated with GEOs. Although the evolution of resistance is a continuous process, the evolution of resistant pests has been considered a potential environmental hazard of GEOs because more environmentally damaging alternative treat- ments could be needed for continued control. Insect resistance to transgenic
WHEN AND WHY TO CONSIDER BIOCONFINEMENT 53 Bt crops is considered inevitable (NRC, 2000), and the United States Envi- ronmental Protection Agency (US EPA) has issued guidelines on the cultiva- tion of transgenic crops, mandating that farmers plant refuges of non-Bt crops along with Bt crops to prevent or decrease the rate of resistant evolu- tion (US EPA, 1999). In some cases, the bioconfinement methods described in this report could help address needs for resistance management. FOOD SAFETY AND OTHER ISSUES Beyond the environmental concerns described above, two more topics have been widely discussed. The first is food safety. Although no adverse health effects have been identified after a decade of commercial production of genetically engineered food crops in the United States, initial general concerns about their consequences for human health have been replaced by specific questions: Will some transgenic products prove allergenic? Will transgenic products that are not intended for human consumption end up in foods? Genetically engineered food crops that produce pharmaceutical and industrial compounds pose a special challenge to ensure that those crops do not commingle with crops of the same species intended for food (Pew Initiative on Food and Biotechnology, 2002). There also has been debate about the social and economic consequences of GEOs. The possibil- ity that GEOs could alleviate hunger in less developed nations by increasing productivity is balanced by the concern that genetically engineered crops-- like past advances in agricultural technology (Evenson and Gollin, 2003)-- will have complex socioeconomic impacts that benefit some farmers and adversely affect others. WHEN AND WHY TO CONSIDER BIOCONFINEMENT: THE NEED FOR PREVENTIVE ACTION It is essential to consider preventive action before a failure occurs and even before confinement techniques are chosen. Prevention typically is less expensive and more effective than is remedial action, and some conse- quences--death of a human, extinction of a species, destruction of a large ecosystem--cannot be reversed. The choice of confinement technique, and even the decision to proceed with a proposed GEO, should be informed by an analysis of possible preventive actions because the use of confinement is itself a precautionary measure. The committee concludes that it is essential to consider--from the very beginning of the process of developing a GEO and its possible confine- ment--the risks and consequences of failure, the means of failure preven- tion (particularly by bioconfinement), and the potential for postfailure remediation, to determine what, if any, bioconfinement measures to take.
54 BIOCONFINEMENT OF GENETICALLY ENGINEERED ORGANISMS Precisely how to identify what combination of confinement measures to undertake--if any--is addressed above. HOW MUCH CONFINEMENT IS ENOUGH? If some type of confinement is deemed necessary, careful consideration should be given to how much will be sufficient. In some cases, appropriate confinement might be obtained by conventional, non-biological, methods. For example, sufficient isolation might be achieved by growing a minor crop far from stands of the same variety and away from populations of wild relatives. That might reduce viable pollen flow to acceptable levels. Other steps would ensure that seed and vegetative propagules did not find an environment where they could become established. The most stringent confinement is necessary when field releases of GEOs or their transgenes have sufficient potential to create substantial problems. If stringent confinement is to be applied to released organisms, the standard methods of spatial or temporal isolation will not suffice. For example, the standard contamination tolerated by breeders of high-quality "foundation" seed generally is 10 (NRC, 2000). That purity is attained by 3 spatial and temporal isolation, often in conjunction with the use of border or barrier crops that will interfere with pollen flow (Kelly and George, 1998). Generally a 660-ft buffer is considered sufficient to reduce back- ground contamination of maize fields to 0.10%--often considered accept- able. However, almost one-third of some 300 maize fields in the Corn Belt exhibited background contamination that ranged from 1.5% to 15.6% at that distance (Burris, 2001). Redundancy of methods is usually necessary to achieve stringent con- finement levels. The 2002 APHIS requirements for growing transgenic maize for pharmaceutical compounds in the field call for substantial spatial and temporal isolation (USDA, 2002). Those requirements, however, are only for preventing dispersal of genes by pollen. APHIS requires additional methods to prevent gene movement by seed or gene persistence in a soil seed bank. Maximum confinement will not be necessary for most organisms. GEOs that pose no hazard or whose risk level is so low as to be tolerable would not require containment. The need for confinement might vary with the fitness impact of the allele in question as well as with whether immigration is anticipated to be recurrent or a single event. What if the allele will have no impact on fitness? If the allele is expected to have a neutral impact on fitness but immigration is expected to be recurrent, then the allele is expected to increase, generation by generation. If such an allele is predicted to create a significant hazard in locations for which it was not intended and that recurrent immigration is going to occur, then the most stringent confine-
WHEN AND WHY TO CONSIDER BIOCONFINEMENT 55 ment conditions are advised. If immigration would occur once or a few times, then less stringent conditions might be permissible, depending on the acceptable frequency for that allele. What if the allele would prove detrimental in populations or environ- ments for which it was not intended? If only one or a few immigration events are anticipated, the allele will not persist in a population, and if immigration is recurrent, the allele will be maintained at a frequency between zero and one. In this case, the acceptable allele frequency is important. The fact that the allele confers a fitness disadvantage will result in a decline in the average fitness of the recipient population. The fitness change also could be so severe as to drive the affected population to the point of extinction (e.g., Huxel, 1999; Wolf et al., 2001). Extinction by hybridization can result from more complicated situa- tions. If an immigrant allele reduces viability but dramatically increases mating success, the antagonistic effects on different fitness components can lead to eventual extinction after a single immigration event (Muir and Howard 1999, 2001, 2002). Whether such local extinction would be a problem or a benefit would depend on the population. For example, an increased extinction risk would be a problem for an endangered species, but a benefit for one considered a noxious invasive. Clearly, to reduce chances of escape, there is a need for as many cost- effective tools as possible. The appropriate amount of confinement might be possible now, but the method could be so expensive as to preclude its use. Bioconfinement methods offer an opportunity to expand the number and diversity of tools available. Finally, there could be organisms for which the possible hazard is so great that it would be best never to release the product. A recent document (USDA, 2002) states that some species are "inappropriate for the produc- tion of pharmaceuticals" when grown under field conditions. The cited example is the oilseed rape species Brassica rapa. Its traits include multiple- year seed dormancy, bee pollination, and sexual compatibility with weed species that could be found in adjacent fields. NEED FOR BIOCONFINEMENT What sorts of GEOs might require confinement? Some GEOs may raise environmental concerns, as described throughout this report. When great harm to the environment is probable, confinement is warranted. Food safety and food purity issues also could motivate confinement. Other reasons could be social, ethical, political, or economic. For example, the security of intellectual property has long been recognized as a motivation for prevent- ing the unintended escape or theft of living biological material of value (e.g., Ellstrand, 1989). Transgenic crops that are commercialized in the
56 BIOCONFINEMENT OF GENETICALLY ENGINEERED ORGANISMS United States might be prohibited elsewhere, leading to the need to segregate genetically engineered products for export. In addition, APHIS currently requires strict confinement for GEOs that are field-released under "Notifi- cation" and "Permit" as part of its performance standards (USDA, 1997). New, more stringent requirements for organisms that produce products intended for use in pharmaceutical and other industrial compounds were released by APHIS in early 2003 (USDA, 2003). The StarLink incident of 2000 illustrates the need for effective confine- ment in maize (see Box 2-1). Although StarLink corn was released for animal consumption only, it rapidly entered the general maize supply and within a year its presence was detected "in nearly one-tenth of 110,000 grain tests performed by United States federal inspectors" (Haslberger, 2001). Generally, it is thought that mixing of seed was responsible for most of the unintended movement. However, the Cry9C protein also was detected in other, supposedly genetically pure, non-StarLink varieties, suggesting that cross-pollination was partly responsible for its spread (Taylor and Tick, 2003). Presence may or may not indicate a high level of contamina- tion. Current detection techniques are quite sensitive, capable of detecting a single contaminating grain out of thousands in a bulk sample. It is highly unlikely that the Cry9C protein produced by this variety poses any kind of risk, but the fact that the gene moved so rapidly demonstrates how quickly unintentional movement can occur. In any case, regulatory agencies have seen that granting approval to genetically engineered plants that are intended solely for animal consumption is inadvisable if other varieties are used in human food production. PREDICTING THE CONSEQUENCES OF FAILURE Failure of bioconfinement presents several challenges. The most obvious apply to the unintentional escape of transgenes. The specific problem will be the risk that is intended to be averted by the bioconfinement method. Thus, the problem could be environmental; for example, the creation of a new or more difficult-to-control pest that is developed because of the intro- duction of a gene that makes an organism more invasive. There also could be effects on human or animal health. A gene introduced into a crop for the production of industrial chemicals might inadvertently move into crops intended for human consumption. The problem could be economic. A pat- ented transgene, considered intellectual property, can be stolen. The prob- lem could be the serious decline, displacement, or extinction of a species with social or cultural significance. One could quantify specific risk as the likelihood of failure and the magnitude of escape. Other problems are subtler. The bioconfinement phenotype itself could cause havoc in organisms for which it is not intended, as in the following example. Because of the
WHEN AND WHY TO CONSIDER BIOCONFINEMENT 57 automatic and substantial fitness advantage of asexual genotypes relative to sexual genotypes due to the evolutionary "cost of sex" (e.g., Charlesworth, 1989; Williams, 1975), the application of seed apomixis--an often-touted method of bioconfinement--could result in the rapid spread of an asexual genotype through wild, sexually reproducing populations (see Chapter 3 for a detailed discussion). The consequence could be a drastic reduction in genetic diversity and in the potential for evolutionary response (van Dijk and van Damme, 2000). The failure itself could result in altered public perception of biosafety, decreasing the credibility of the biotechnology industry or government regu- lators, or both. The StarLink incident led to the realization that current systems do not provide for the segregation of genetic material. A flurry of attention from the popular media about a bioconfinement method that fails could result in similar public mistrust for methods that are designed to keep transgenes in their place, leading to a loss of public confidence in the food supply and damage to the viability of the biotechnology industry as a whole. Clearly, the failure of GEO bioconfinement can, in some circumstances, result in substantial consequences. What factors must be considered to predict some of the risks that could result from the failure of bioconfinement? The consequences of bioconfinement failure must be assessed on a case- by-case basis. A science-based risk analysis of the consequences of bio- confinement failure should consider at least the following factors: the organ- ism involved; the trait or traits that have been introduced into the organism; the genomic, physical, and biotic environments in which the failure could occur and that could experience the effects of a failure; the possible effects on human health; the bioconfinement techniques involved; and the social and behavioral factors that could affect consequences. The most important factors in determining environmental consequences could be the ecological phenotype and ecological novelty of the GEO. For example, mud loach (Misgurnus mizolepis) (Chinese weatherfish) is an aquacultural species native to China and Korea. It has been genetically engineered such that no extraneous DNA was used (by inserting a gene and a promoter that originated from the same species--the mud loach itself; Nam et al., 2001a, 2001b). Hatchlings showed dramatically accelerated growth--at a maximum, 35-fold faster than non-genetically-engineered sib- lings. The largest hatchling weighed 413 g, and, with a length of 41.5 cm, exceeded the size of 12-year-old normal broodstock (89 g, 28 cm). The time required to attain marketable size (10 g) was 3050 days after fertilization; nonengineered fish require at least 6 months. There also was significantly improved feed-conversion efficiency, up to 1.9-fold. There appeared to be no gross abnormalities other than the size increase, but most individuals died after their body weight exceeded 400 g. Thus, despite the fact that the
58 BIOCONFINEMENT OF GENETICALLY ENGINEERED ORGANISMS transgenic organism has no exogenous genes, it has a novel phenotype that is obviously (ecologically) quite different from that of the original (Nam et al., 2001a, 2001b). The degree of transgenic novelty (the number of geneti- cally engineered changes) and the taxonomic or phylogenetic distance between the host organism and the novel genes are not likely to determine conse- quences (NRC, 2002b). This is also consistent with a previous NRC report where it was noted that both small and large genetic changes can have significant environmental consequences and that the consequences of biotic novelty are strongly influenced by the "genomic environment, physical environment, and biotic environment" (NRC, 2002a). The significance of undefined consequences depends in part on social values and the context in which the failure occurs. Potential loss of cultur- ally symbolic varieties or species is often a focus for social action because they represent social or spiritual values. For example, a failure leading to the decline of a species will have greater significance if that species has high symbolic importance in that location (sugar maples in New Hampshire, blue crabs in the Chesapeake Bay, corn to the Hopi Tribe, wild rice (Zizania) to the Ojibwa and Menomonii, or salmon to sport and commercial fishing and to Native Americans in the Pacific Northwest) than if it has economic or ecological importance alone. In addition, the decline of a symbolic species is a likely indicator of environmental harm. WHO DECIDES? Decisions about bioconfinement involve asking whether bioconfinement measures should be applied in a given case and, if so, which measures should be adopted and whether they should be applied alone or in conjunc- tion with other types of confinement. The decision makers come from the genetic engineering industry, including the GEO developers (including aca- demic and government scientists), related industries (such as the wholesale and retail food industries) that could be affected by an escape or by any resulting loss of public confidence; insurance companies; government regu- lators; and private citizens who might sue to enforce environmental laws. Decisions about private legal action arising from damage caused by GEOs might indirectly affect whether bioconfinement or other confinement mea- sures are undertaken. Industry Bioconfinement should be considered early in the development of a GEO. Whether public or private, for-profit or nonprofit, the enterprise that develops a GEO is in the best position to determine the advisability and viability of bioconfinement because that person or group uniquely possesses
WHEN AND WHY TO CONSIDER BIOCONFINEMENT 59 the information necessary to conduct such analyses. The enterprise has an interest in achieving regulatory approval for field testing and marketing so that it can avoid negative publicity and to protect itself from liability. The industry has an interest in bioconfinement because the escape of one or a few GEOs could jeopardize the viability of the entire industry. The com- mittee is not aware of any industrywide standards (binding or nonbinding) for bioconfinement. Related industries that use or depend on GEOs, such as the wholesale and retail food industries, also have an interest in maintaining the safety of their products and the public confidence in that safety. Thus, they have an interest in confinement and, by extension, in bioconfinement. The U.S. food industry, for example, supports strong regulations that would ensure the segregation of pharmaceutical crops from those that enter the U.S. food supply. After the Prodigene incident, the Grocery Manufacturers of America met with senior USDA officials and congressional staff members to call for stricter regulation of pharmaceutical crops (Fox, 2003). Insurance Companies To the extent that GEO developers can obtain insurance against the possibility of escape or failure of bioconfinement, the risk of loss shifts to the insurer, who then would have an interest in adequate confinement, including bioconfinement. Apparently, insurance is available for genetic engineering under liability insurance policies, and only a few markets con- tain specific coverage or exclusions of genetic engineering applications (Epprecht, 1998). There is scant experience with losses involving GEOs. At the same time, societal views about the acceptability and value of GEOs-- and even of particular bioconfinement methods (such as the use of termina- tor genes)--are in flux and can influence the insurance risk (e.g., with respect to product liability). Furthermore, the risk associated with some escapes could be extraordinarily high. It thus would seem that the insurance industry operates in a sea of uncertainty. The committee is unaware of any confinement or bioconfinement requirements imposed by insurance compa- nies individually or collectively. Government In the United States, GEOs and associated bioconfinement measures are regulated by a mosaic of laws and agencies. The Coordinated Frame- work for the Regulation of Biotechnology Products was adopted by federal agencies in 1986 (51 Fed. Reg. 23302 [June 26, 1986]) in response to concerns about how best to provide federal oversight for products of bio- technology. Where those laws apply, they take precedence over the private
60 BIOCONFINEMENT OF GENETICALLY ENGINEERED ORGANISMS decisions described above, with the exception of decisions concerning pri- vate lawsuits for damages. The Coordinated Framework is based on two premises. The first, which is consistent with the judgment of several reports (NRC, 1989a,b, 2000, 2002a), is that the potential risks associated with GEOs fall into the same general categories as those established for traditionally bred organisms (but see NRC, 2002b, for another judgment). The second is that the statutes written for non-GEOs should provide an adequate basis for regulating GEOs. The coordinated framework is intended to provide a harmonized regu- latory approach and to ensure the safety of biotechnology research and products by using existing statutory authority and building on agency expe- rience with agricultural, pharmaceutical, and other products developed through traditional techniques of genetic modification. The development of the framework anticipated that agencies might need to develop specific regulations or guidelines under existing statutory authority. It also antici- pated institutional evolution in accord with experience, including modifica- tions made through administrative or legislative action. Finally, the coordi- nated framework specifies that interagency coordination mechanisms are necessary to address all manner of policy and scientific questions (Council on Environmental Quality and the Office of Science and Technology Policy, 2001). The regulatory approach articulated by the coordinated framework invokes many statutes, as well as their implementing regulations and guide- lines that could apply to GEOs. Some apply to specific products or activities and are administered by a single agency; others apply generally and are thus of interest to virtually all agencies. The committee finds that the complexity of federal oversight could hamper the effectiveness of bioconfinement imple- mentation, as discussed below. Determining which law applies to a GEO under the coordinated frame- work can entail a complicated analysis, involving such factors as the stage of development (Is the GEO contained in the laboratory? Is it being field tested? Is it ready for commercial use?), its uses (Is it intended for bio- remediation of pollution or for biocontrol of another organism? Is it intended to be a human food or drug or an animal biologic? Might it eventually be used as food even though that is not its current use?), the type of possible hazards (Could it harm plants? Could its genetic material cause a plant to become a noxious weed? Could it release pollutants into the atmosphere or water?), the type of organism (Is it an animal, plant, or microorganism?), and whether regulatory agencies have reached consensus on how the GEO should be regulated. Depending on that analysis, many laws and several agencies could be involved in regulating a particular GEO or bioconfinement method, and
WHEN AND WHY TO CONSIDER BIOCONFINEMENT 61 several other statutes that are not currently used to regulate GEOs could be brought to bear. Some of them are now applied to invasive species, and the experience with and laws that regulate nonindigenous species might be helpful in regulating GEOs (Council on Environmental Quality and the Office of Science and Technology Policy, 2001). Federal statutes and guidelines embody different approaches and con- tain different authorities, standards, and enforcement provisions. Other federal agencies, including the Office of Management and Budget, for example, affect the way regulatory agencies interpret and apply the statutes for which they are responsible. Congress also weighs in by controlling funds. Congress prohibited US EPA from expending any money to enforce a regulation issued under the Clean Air Act. Thus, each agency exercises regulatory authority differently. To the extent that laws regulate confine- ment and bioconfinement, they trump the decisions of the private parties discussed above. Several agencies impose specific confinement requirements. APHIS, for example, requires physical confinement where crops genetically engineered to produce industrial chemicals or pharmaceuticals are field tested. The confinement requirements include a 50-ft perimeter fallow zone around the field test site; restriction on the production of food and feed crops in the field test site and in the fallow zone the next season if volunteer plants could be inadvertently harvested with that season's crop; the use of dedication of mechanical planters and harvesters for the duration of the test, and cleaning of that equipment in accordance with protocols for tractors and tillage attachments; the use of dedicated facilities for storing equipment and the regulated GEO; and, for field tests of open-pollinated pharmaceutical corn, a prohibition against growing any other corn one mile of the field test site for the duration of the field test (Federal Register, 2003). Another example is that the US EPA sometimes requires refuges or buffer zones for certain genetically engineered plants (Bt corn and cotton) (US EPA, 2002; Elias, 2002). In 2002, the US EPA filed complaints against two companies for noncompliance with agency requirements not to grow experimental GE corn too close to other crops and to use trees and other corn to form a windblock next to the GE crops. The complaints were settled through legal agreements between EPA and both companies, with each company agreeing to pay a fine (Elias, 2002). Additionally, EPA fined one of those companies a second time, for failing to notify the agency of test results indicating the presence of the experimental gene in seeds grown near the experimental plants, and for failing to submit maps identifying the location of such seeds (EPA, 2003b). The Food and Drug Administration (FDA) may require confinement (including bioconfinement) of transgenic animals under the terms of the Federal Food, Drug, and Cosmetics Act (FFDCA), 21 U.S.C. 321-397. The
62 BIOCONFINEMENT OF GENETICALLY ENGINEERED ORGANISMS U.S. Fish and Wildlife Service and the National Marine Fisheries Service are permitted by the Endangered Species Act (16 U.S.C. 1531-1544) to require bioconfinement for protecting a listed species. Laws concerning confidentiality of information also affect how GEOs and their confinement are regulated by the federal government. Trade secrets and confidential commercial informationoften called confidential busi- ness information (CBI)are protected under the statutes that are used to regulate GEOs. The CBI provisions in those statutes differ considerably, as do the regulations issued thereunder and the procedures used by the agencies involved. For example, FFDCA prohibits FDA from sharing CBI with any other federal agency, even if that agency is engaged in evaluating or regulat- ing the same GEO. In contrast, other statutes used to regulate GEOs do not contain any such prohibition. Each agency administers its own program for protecting CBI on a statute-by-statute basis. Thus, US EPA has two separate CBI programs, and authorization for access to CBI under one does not allow access under the other. The result is that a scientist involved in regulating a GEO under one statute cannot share CBI, including confine- ment-related CBI, with a scientist involved in regulating the same GEO under another statute unless each is qualified to know CBI under both statutes. Similarly, each agency applies different amounts of scrutiny to assertions by business entities about what constitutes CBI. CBI is not available to the public (for examples of the use of CBI, see Chapter 3, Tables 3-2 and 3-3). The regulatory system described above is obviously complex, leading to differing legal authorities and responsibilities, potentially overlapping juris- dictions, and differing statutory standards for regulating GEOs and deter- mining confinement (including bioconfinement). This could lead to a host of problems in regulating confinement. For example, US EPA has no inde- pendent authority to require information about where a GEO granted nonregulated status by USDA is grown, which impedes US EPA's ability to monitor confinement requirements for that GEO. US EPA does not have the regulatory authority to enforce the confinement requirements built into the Federal Insecticide, Fungicide, and Rodenticide Act (FIFRA) in any event (Bratspies, 2002). US EPA may set regulatory thresholds for GEOs, but FDA has no such authority (Taylor and Tick, 2003). Coordination is needed, but there are no central mechanisms to ensure consistency in confinement, including bioconfinement, or to ensure coherence in setting priorities (Bratspies, 2002). The splintered approach to protecting CBI further interferes with agencies' ability to take a coordinated approach to GEO confinement. Indeed, a mismatch exists between the application of existing laws and the actual practice of regulating GEOs. Because the statutes initially were
WHEN AND WHY TO CONSIDER BIOCONFINEMENT 63 enacted for radically different purposes, there is no unifying vision of or approach to the questions and challenges posed by GEOs and their confine- ment and there is no concordance with the coordinated framework's evident goal of smoothing the path for genetic engineering technology (Bratspies, 2002). The approaches normally used in carrying out those statutes do not always fit GEO confinement. For example, US EPA typically applies split registration for traditional chemical pesticides by requiring labeled con- tainers with a clear indication of registered use. This practice cannot be extended to Bt crops, for example, because they do not stay in neatly labeled bottles, and the postharvest distribution, storage, and processing of the corn do not parallel the use of a typical pesticide. Citizen Suits to Enforce Environmental Laws Private citizens have a role in enforcing relevant legal authority: Some statutes allow them to bring action to enforce environmental laws, either to direct the agency involved to enforce a law or to seek a civil penalty (which, if recovered, would go to the government). Citizens can sue to enforce the National Environmental Policy Act and the Endangered Species Act as they apply to the bioconfinement of GEOs. Citizen suits seeking to compel government agencies to enforce the law are permitted by other environ- mental statutes including the Clean Water Act (sections 304 & 305, 42 U.S.C. sec. 7604, 33 U.S.C. 1365), and cases frequently are brought under such provisions. Of the statutes used to regulate GEOs, only the Toxic Substances Control Act (TSCA) contains such a provision (U.S.C. sec. 2619). In a related type of action, however, it is possible to petition USDA, FDA, and US EPA, respectively, under the terms of the Plant Protection Act, FFDCA, and the FIFRA regarding some GEO-related activities and to chal- lenge the agencies' decisions regarding those petitions in court under the judicial review provisions of those statutes. Private Action for Damage An indirect decision maker about bioconfinement is found in private causes of action for compensation for damage that results from escape, which typically would be subject to state law. One example with respect to nonbiological confinement is a class action lawsuit brought by farmers who claimed they were harmed when StarLink corn was discovered in the human food supply in 2000. The biotechnology companies that created and dis- tributed StarLink agreed to pay $110 million to settle the case (Pesticide and Toxic Chemical News, 2003). Private legal action also can be used to enforce or invalidate patent rights, including in the context of a failure of
64 BIOCONFINEMENT OF GENETICALLY ENGINEERED ORGANISMS confinement (Monsanto Canada Inc. & Monsanto Co. v. Percy Schmeiser & Schmeiser Enterprises Ltd., 2001 FCT 256, confirmed in Percy Schmeiser & Schmeiser Enterprises Ltd. v. Monsanto Canada Inc. & Monsanto Co., 2002 FCA 309, on appeal to the Canadian Supreme Court as of May 2003). Such law is federal, thus implicating lawmakers at the federal level.