POTENTIAL ENVIRONMENTAL RISKS AND HAZARDS OF BIOTECHNOLOGY
Part I: Risks and Hazards
Part II will look at methods to address those hazards
Harm, Risk, and Hazard
A concern related to genetically modified (GM, transgenic) organisms is the potential environmental harm if these organisms escape or are released into the environment. Harm can take many different forms from transient to permanent in time frame and from local to global in scope. Thus, to define harm it is first necessary to distinguish between the terms risk and hazard, which are often confused. In this context, William Muir and Richard Howard (Purdue University, Lafayette, Indiana) define transgene risk as the probability that a transgene will spread into natural populations once released and hazards as the probability of species extinction, displacement, or ecosystem disruption given that the transgene will spread into the population.1 To show lack of harm from transgenic organisms, either the risk [Risk = P(E) where P(E) represents the Probability that Exposure will occur] or hazard [Hazard = P(H/E) where P(H/E) is the conditional Probability of a resulting Harm (H) given that exposure has occurred] must be close to zero; that is, P(E) 0 or P(H/E) 0. Long-term hazards to the ecosystem are difficult to predict because not all non-target organisms may be identified, species can evolve in response to the hazard, and a nearly infinite number of direct and indirect biotic interactions can occur in nature. Muir and Howard conclude the only way to ensure that there is no harm to the environment is to release only those transgenic organisms whose fitness is such that the transgene will not spread, i.e., P(E) 0, in which case the hazard, P(H/E), is irrelevant because the transgene is lost from the population.1
Factors Affecting Risk
In this context, long-term ecological risk can be determined from the probability that an initially rare transgene can spread into the ecosystem. Spread of the transgene into natural populations may result in a number of ways, including 1) vertical gene transfer as a result of matings with feral animals, 2) invasion of new territories as with introduction of an exotic species, and 3) horizontal gene transfer mediated by microbial agents, or a combination of these factors. The relative importance of each factor is dependent on species, transgene inserted, and method used to insert the transgene, respectively.
Vertical Gene Transfer: The first mechanism of spread, vertical gene transfer, is dependent on species modified. Highly domesticated stock developed for poultry, swine, and cattle are not well adapted to the natural setting and may not be able to survive and reproduce there. However, if feral populations are locally available, then local adaptation is not a major barrier to gene spread, as the domesticated GM stock may be able to mate with the highly adapted native populations. Aquatic species present the greatest concern in this regard because aquatic environments are highly connected throughout the world and readily available feral populations exist for all domesticated species. Although feral populations do not exist locally for every domesticated species, if the GM organism has an economic advantage, we must assume that human intervention will transport such organisms to area(s) of the world where native populations exist.
Invasion of New Territories: The second mechanism of spread, invasion of new territories, depends on the functionality of the transgene. The anthropogenic introduction of any exotic organisms into natural communities is a serious ecological concern because exotics could adversely affect communities in many ways, including eliminating populations of other species.2 The release of transgenic organisms into natural environments, however, poses additional ecological risks—although transgenic individuals retain most of the characteristics of their wild-type counterparts, they may also possess some novel advantage. A transgene for enhanced environmental adaptation, such as heat tolerance, would allow cold water fish with this gene to invade cool and warm water environments while maintaining populations in current habitats. As such, GM fish could reproduce at a faster rate; their population may increase unchecked and adversely affect other species. As a consequence, transgenic organisms might threaten the survival of wild-type conspecifics as well as other species in a community.3
Horizontal Gene Transfer: The third mechanism of spread, horizontal gene transfer, occurs naturally through viruses and transposons, but at such low rates that it would not normally be an additional concern. However, if a virus or transposon is used to insert the transgene construct, even if the virus is disabled, it may be possible for the element to recombine with other naturally occurring viruses and spread into new hosts.
Regardless of the mechanism of gene spread, the ultimate fate of the transgene will be determined by the same forces that direct evolution, i.e., natural selection acting on fitness. Thus, risk assessment can be accomplished by determining the outcome of natural selection for increased fitness. This conclusion assumes that the natural populations are large enough to recover from such introductions, i.e., natural selection will have time to readjust the population to its previous state. Fitness in this context is not simply survival to market age but all aspects of the organism that result in spread of the transgene. Muir and Howard reduced these aspects to six net fitness components: juvenile and adult viability, age at sexual maturity, female fecundity, male fertility, and mating success.1,4,5 Mating success is often overlooked because it is not a factor in artificial breeding programs but is often the strongest factor driving natural selection.6
Extinction Hazard: Muir and Howard found that pleiotropic effects of transgenes that have antagonistic effects on net fitness components can result in unexpected hazards, such as local extinction of the species containing the transgene.7,1 Such transgenes were referred to as Trojan Genes. A Trojan Gene is a gene that drives a population to extinction during the process of spread as a result of destructive self-reinforcing cycles of natural selection. For example, if a transgene enhances mating success while reducing juvenile viability, the least fit individuals obtain the majority of the matings while the resulting transgenic offspring do not survive as well. The result is a gradual spiraling down of population size until eventually both wild-type and transgenic genotypes become locally extinct.7 These results were later theoretically verified by Hedrick.8 Local extinction of a wild-type population from a transgenic release could have cascading, negative effects on the rest of the community.
The interaction of mating success and juvenile viability is not the only mechanism that can produce a Trojan Gene effect. Muir and Howard have shown that there are other ways in which a Trojan Gene can result, such as if the transgene increases male mating success but reduces daily adult viability, or the transgene increases adult viability but reduces male fertility.1 The latter case is of particular interest because transgenes for disease resistance or stress tolerance can increase offspring viability and transgenes can also reduce male fertility, as has been reported for transgenic tilapia containing the growth hormone (GH) gene.9 Extinction hazards predicted in this case parallel the use of sterile males to eradicate pest insects. However, in the latter program, males are completely sterile and must be reintroduced repeatedly to cause extinction. In effect, the viability of sterile males is near 1.0 (due to repeated introduction) while male fertility is 0%. Such population extinction, as a result of the antagonistic pleiotropic effects of transgenes on viability and fertility, represents a new class of Trojan Genes, which suggests that attempts to reduce transgenic male fertility that do not result in complete male sterility may increase hazard rather than reduce it.9
Invasion Hazard: Muir and Howard also confirmed that, as expected, if any of the net fitness components are improved by the transgene, while having no adverse side effects, the transgene will invade a population.1,4 However they showed that advantages in one fitness component can offset disadvantages in another and still result in an invasion risk. Experimental evidence that transgenes have multiple effects on fitness components was presented by Muir and Howard with the Japanese rice fish, medaka (Oryzias latipes).4 They found that insertion of a growth hormone gene resulted in a 30% reduction in juvenile viability, a 12.5% reduction in age at sexual maturity, and a 29% increase in female fecundity, relative to wild type. Our model predicted that advantages in both age at sexual maturity and fecundity are sufficient to overcome the viability disadvantage produced by the transgene and would present an invasion risk if released. The model also predicted that for a wide range of parameter values, transgenes could spread in populations despite high juvenile viability costs if transgenes also have sufficiently high positive effects on other fitness components.
This research clearly shows that all six net fitness components must be estimated to determine risk. Simple models, such as those presented by Mclean and Laight that are based on viability or other single fitness components, are very misleading.10 Also, those components need to be integrated into a model that combines them into one prediction of risk. In the next part, I (W. M.) will examine experiments to estimate net fitness components and review development of the model.
1. Muir WM and Howard RD. 2001. Environmental risk assessment of transgenic fish with implications for other diploid organisms. Transgene Research. In press.
2. Bright C. 1996. Understanding the threat of biological invasions. In State of the World 1996: A World Watch Institute report on progress toward a sustainable society, ed. L Starke, 95–113. New York: WW Norton.
3. Tiedje JM et al. 1989. The planned introduction of genetically engineered organisms: Ecological considerations and recommendations. Ecology 70: 298–315.
4. Muir WM and Howard RD. 2001. Fitness components and ecological risk of transgenic release: A model using Japanese medaka (Oryzias latipes). American Naturalist 158: 1–16.
5. Muir WM and Howard RD. 2001. Methods to assess ecological risks of transgenic fish releases. In Genetically engineered organisms: Assessing environmental and human health effects, eds. DK Letourneau and BE Burrows, 355–383. CRC Press.
6. Hoekstra HE et al. Strength and tempo of directional selection in the wild. PNAS USA 98: 9157–9160.
7. Muir WM and Howard RD. 1999. Possible ecological risks of transgenic organism release when transgenes affect mating success: Sexual selection and the Trojan Gene hypothesis. PNAS USA 24: 13853–13856.
8. Hedrick PW. 2001. Invasion of transgenes from salmon or other genetically modified organisms into natural populations. Canadian Journal of Fisheries and Aquatic Sciences 58: 841–844.
9. Rahman MA and Maclean N. 1999. Growth performance of transgenic tilapia containing an exogenous piscine growth hormone gene. Aquaculture 173: 333–346.
10. Maclean N and Laight RJ. 2000. Transgenic fish: An evaluation of benefits and risks. Fish and Fisheries 1: 146–172.
POTENTIAL ENVIRONMENTAL RISKS AND HAZARDS OF BIOTECHNOLOGY
Part II: Methods to Estimate Risks and Hazards
In Part I, William Muir and his associate Richard Howard conclude that the risk of releasing transgenic organisms to the environment can be assessed by addressing the probability of exposure to the hazard, P(E), rather than the probability of harm given exposure, P(H/E) (see ISB News Report, November 2001, http://www.isb.vt.edu/news/2001/news01.nov.html#nov0105 ). The probability of exposure is equal to the expected long-term outcome of natural selection for the transgene, given that the transgene has escaped into a natural environment. Although escape results in initial exposure, harm ultimately results from long-term exposure—the transgene may increase in frequency until it becomes the norm rather than the exception (P(E)=1), or, with time, the transgene will be culled from the population (P(E)=0). The exception occurs if there is a massive escape and/or the wild population is very small.
Natural selection occurs as a result of the differential ability of genotypes to produce offspring for the next generation. This differential ability is termed overall or net fitness. Muir and Howard have reduced overall fitness to six components: juvenile and adult viability, age at sexual maturity, female fecundity, male fertility, and mating success.1,2,3 They then incorporated these components into a mathematical model that integrates them into a single prediction of risk. The discussion that follows will explain how the fitness components are estimated and demonstrate use of the model using Japanese medaka fish.
Estimation of Fitness Components
The model is based on the assumption that transgene expression is completely dominant and that individuals hemizygous for the transgene (Tw) have the same phenotype as homozygous (TT) individuals. Heterozygous individuals are termed hemizygous because there is no complementary allele for the transgene. Nevertheless, the absent allele at that locus may be represented as w and the transgene allele may be represented as T. In the following notation, the transgenics' genotypes TT and Tw are designated as subscripts 2 and 3, respectively.
Juvenile Viability (vj )
Juvenile viability is simply defined as survival from the embryo to the age of sexual maturity (or approaching sexual maturity). There are several ways to estimate this component. The simplest experiment would be to establish two pure breeding lines (TT and ww) and, starting with a known number of fertile eggs, count the number that survive to sexual maturity. The experiment should be conducted under environmental conditions that closely approximate the natural environment into which they might escape. This experiment should be replicated several times (2 to 10). The average percent survival of each genotype (v'j) is converted to per day viabilities (vj) occurring between consecutive census time periods (at+1 and at) by assuming a log-linear reduction in daily viability between time periods. Viability can then be described by the following equation.
For example, if 10 replicates of 1000 fry results in an average of 175 transgenic and 250 wild-type individuals surviving to 56 days of age, the per day survival rate is calculated thus:
An alternative method that addresses the issue of background genotype, but does not require the offspring to be genotyped, is given by Muir and Howard.1,2 This method is based on the theory that the only difference between survival of an intercross and a backcross is the segregation ratio of 3:1 vs. 1:1. If the viabilities of each genotype are the same, then the expected survival for each cross would be the same. If viability of the transgenic genotype is less than that of wild type, then total survival of the intercross will be less than that of a backcross. Preferably the wild-type line is representative of the native fish in the area into which the fish might escape. In this way, the background genotype of both transgenic and wild-type fish are taken into account.
The procedure is to cross the homozygous transgenic line with a wild-type strain to produce the F1 generation. The cross F1 is then intercrossed to produce the F2 generation and the F1 is also backcrossed to the wild type stock to produce the BC1 generation. The number of fish that survive from hatching to sexual maturity (or approaching sexual maturity) is recorded. The relative viabilities are then found by the method of maximum likelihood as shown by Muir and Howard.1
Adult Viability (uj )
Ideally, adult survival of each genotype would be measured in as natural an environment as possible until most fish die. Because this may take a very long time, an alternative method is to assume a log-linear reduction in daily viability and observe only enough time periods to establish a trend.
For example, assume 1000 fish of each genotype are observed for 100 days past sexual maturity. The proportion surviving at the termination of the experiment of each genotype (u'j) is 90% for wild-type and transgenic individuals. The daily reduction in survival (uj) is determined using the following equation:
In our example, at+1- at = 100 days. Thus
Age at Sexual Maturity (sj )
Age at sexual maturity can be straightforward or difficult, depending on the species. For medaka, age at sexual maturity was recorded as the age at which females first produced eggs. We assumed the males to be mature at the same age. For other species, it may be necessary to sacrifice the animals at various ages and observe gonadal development. Assume for this example that ages at sexual maturity are 56 and 49 days after hatching for wild type and transgenic fish, respectively.
Female Fecundity (cj )
Fecundity is also straightforward. For medaka, estimating fecundity is a simple matter of counting the number of eggs produced from each genotype. The genotypes should be the same age and several fish should be measured. For this example, assume wild-type fish produce an average of 8.8 eggs per spawn and transgenic produce 11.4.
Male Fertility (rj )
Fertility is more difficult to determine than fecundity and is inferred from the number of eggs fertilized by alternative genotypes. We examined the ability of the male genotype to fertilize eggs with a simple, completely randomized design experiment in which 10 transgenic males and 10 wild-type females were randomly single-pair mated with wild-type females in separate 40-liter tanks for eight days. The first three egg masses produced by each female in that eight-day period were collected and incubated in a hatching tank. Twenty-four hours later, the eggs were examined under a dissecting microscope and classified as fertile or infertile based on presence or absence of embryo development. Assume for this example that both wild-type and transgenic males have fertility rates of 95%.
Relative Mating Success (mj ) and (fi )
As with juvenile viability, mating success of each genotype can be determined either directly or indirectly. Direct observation is the simplest, but may not be possible in all species. With direct observation, staged mating trials are conducted that allow for both mate competition and mate choice. The mating frequency of each genotype is recorded.4,5 Because the effect of the transgene on mating success can vary with its relative frequency, such trials should also be performed using different ratios of genotypes.
With indirect observation, transgenic and wild type males and females in various ratios are placed together in a setting and allowed to mate. Muir and Howard2 call these trials `mating sets.' Several different mating sets, and replications of mating sets, are conducted. After mating, the identity of the male parent is inferred from the genotypes of the progeny. In the simplest case, a single wild-type female would be introduced into a tank containing a transgenic and wild-type male. Any transgenic offspring would immediately identify the genotype of the male parent. Assume for this example an equal mating success of transgenic and wild-type males and females.
The Model and Prediction of Exposure
Table 1 presents a summary of the estimates for net fitness components of Japanese medaka.
|Table 1. Fitness component||Wild type (ww)||Transgenic (TT,Tw)|
|Juvenile survival to sexual maturity (v'j)||25.0%||17.5%|
|Daily juvenile viability (v'j)||0.9755||0.9649|
|Adult survival 100 days past sexual maturity (u'j)||90%||90%|
|Adult viability (uj)||0.9989||0.9989|
|Male fertility (rj)||95%||95%|
|Relative male mating success (mj)||100%||100%|
|Relative female mating success (fi)||100%||100%|
|Female fecundity (eggs/clutch) (cj)||8.8||11.4|
|Age at sexual maturity (si)||56 days||49 days|
The model, which incorporates all these effects, is given by Muir and Howard1,2 and is programmed to allow for any combination of fitness components. An interactive trial model is available at http://www.isb.vt.edu/nfca/nfca1.cfm.
Before starting to calculate fitness, one usually assumes a stable wild-type population age distribution. The initial age distribution is set using an exponential decay parameter `b'. The value of this parameter is found by trial and error—the fitness values are set for transgenic fish the same as for wild-type, i.e., the initial population consists only of wild-type fish, or at least fish that do not differ in their fitness. The value of b is found such that the initial and stable age distribution, determined after many generations, is similar. This parameter is not critical to the program, but establishes at least a reasonable starting distribution. Using these above components, the constant value of b for a stable age distribution was found to be 0.93.
Assuming an initial population size of 60,000 and that 60 transgenic individuals were introduced at 56 days of age, Figure 1 gives the predicted gene frequency over the first 40 generations. The increase in transgene frequency suggests a high risk for spread of the transgenic organism, i.e., the transgene will eventually become fixed.
1. Muir WM and Howard RD. 2001. Fitness components and ecological risk of transgenic release: A model using Japanese medaka (Oryzias latipes). American Naturalist 158: 1-16.
2. Muir WM and Howard RD. 2002a. Methods to assess ecological risks of transgenic fish releases. In Genetically Engineered Organisms: Assessing Environmental and Human Health Effects, eds. DK Letourneau and BE Burrows, 355-383. CRC Press.
3. Muir WM and Howard RD. 2002b. Environmental risk assessment of transgenic fish with implications for other diploid organisms. In press.
4. Hallerman E and Kapuscinski A. 1993. Potential impacts of transgenic and genetically manipulated fish on natural populations: Addressing the uncertainties through field testing. Genetic Conservation of Salmonid Fishes, eds. JG Cloud and GH Thorgaard. New York: Plenum Press.
5. Howard RD et al. 1998. Mate choice and mate competition influence male body size in Japanese medaka. Animal Behaviour 55: 1151-1163.
William M. Muir
Departments of Animal and Biological Sciences