SCOPE 44 - Introduction of Genetically Modified Organisms into the Environment

Most biologists agree that there there is no special hazard associated with engineered organisms as a group. Most of us also agree that one could (if one wished) design organisms expected to cause problems if released, at least at the local level. The major remaining doubts therefore focus on the possibility of inadvertently constructing something that is undesirable for reasons that are unanticipated. Epistasis and pleiotropy cover some interactions between genes or their products that are more complex (and therefore sometimes harder to predict) than cases where independent unit functions are simply added to or subtracted from genomes. Both terms were invented to describe mutations that are found in natural (nonengineered) populations.

The term epistasis refers to situations where the phenotypic consequences of mutation at one locus are obscured by the state of a second genetic locus in the same cell. One relevant bacterial example is the determination of antibiotic resistance. Streptomycin-resistant mutations in Escherichia coli affect the target of streptomycin action, a ribosomal protein; in bacteria carrying a duplication of the chromosomal gene that encodes this protein, sensitivity is dominant over resistance, because a cell in which 50% of the target protein is inactivated does not survive. E. coli cells can also become streptomycin resistant through acquisition of plasmids encoding enzymes that destroy the antibiotic. In a cell carrying such a plasmid, the state of the chromosomal locus has no effect on response to streptomycin. The plasmid gene is epistatic to the chromosomal locus.

A pleiotropic mutation is a single genetic change that affects more than one phenotypic trait. There are many bases for pleiotropy. A single protein or a single product may function in multiple pathways; thus mutations of E. coli that inactivate enzymes in the pathway for molybdopterin synthesis cause simultaneously inability to reduce nitrate and some organic sulfoxides, inability to oxidize trimethylamine, and resistance to added chlorate because nitrate reduction, trimethylamine oxidation, and chlorate sensitivity all depend on enzymes that use a common molybdopterin cofactor. Missense mutations in proteins that function in central biosynthetic processes can affect the fidelity of macromolecular synthesis and generate a host of incidental changes, as seen in mutations to streptomycin resistance (protein synthesis) and rifampicin resistance (RNA synthesis). The multiple effects of mutations to streptomycin dependence served as one of the first clues to the nature of the affected protein, before it was known to be ribosomal (Spotts and Stanier, 1961). Mutations affecting components of global regulatory systems can simultaneously affect expression of all the genes controlled by that system. Thus, mutants unable to make cyclic AMP fail to turn on many separate genes for fermentation of different sugars, while recA mutations reduce genetic recombination and expression of the SOS repair system as well as spontaneous production of phage by lysogenic bacteria. Impairment of critical steps in the developmental program of multicellular organisms can create bizarre monsters such as homeotic mutants in which whole organs are misplaced.

Biologists have little difficulty incorporating pleiotropic mutations into the conceptual framework of cellular and organismal determination. Both metabolism and development are affected by complex arrays of interacting components. The effects of individual genetic changes are frequently an incisive tool for detecting those interactions (Botstein and Maurer, 1982). On occasions (as in the early days of biochemical genetics), geneticists have deliberately avoided pleiotropic mutations because they complicated the straightforward analysis of gene action. This has been the shortest course to determining basic mechanisms, but has sometimes created a textbook genetics highly biased toward mutations and genes with simple effects.

Some developmental mutants have such dramatic effects on organismal form that they force reflection on the determinative role of the genes that have mutated. If changing a single gene causes a leg to project from a fly's head, obviously that gene cannot, by itself, specify all the information required for leg development. The same distinction between induction and specification arises in experimental embryology. Tissues can be induced to differentiate along certain lines by natural biological signals, such as cellular contact or gradients of morphogens, but also sometimes by very simple chemical or physical stimuli, such as ion concentration, pH, or even contact with glass beads. Clearly the information for the developmental pathway is specified not by the inducing signal but by the induced tissue (ultimately, by its genome). The inducing stimulus selects one pathway from a limited number of alternatives prespecified by that genome. Likewise, a gene whose alteration redirects development into a different pathway cannot be the major source of information for that pathway, but must control some switch that activates many other genes.

In engineered organisms, we may ask whether foreign genes exert epistatic or pleiotropic effects and, if so, what those effects might be. Usually, the effects will be deleterious rather than advantageous to the organism in question and therefore limit its spread, but there might be exceptions. In principle, answers might either come from experiment (based on surveying the engineered organisms isolated to date) or from theory. As to experiment, I know of no facts to indicate that genetic manipulation has produced any radically surprising phenotypes. On the theoretical side, we may at least try to define the issue. In the early days of recombinant DNA research, some champions of the technology emphasized that an E. coli cell harboring a small segment of exogenous DNA is, after all, still E. coli; that little piece of exogenous DNA could hardly change it into a totally new organism. Our discussion of pleiotropy indicates the difficulty with literal acceptance of that position. If a very small change in the E. coli genome can effect multiple alterations in phenotype, one can hardly doubt that introduction of some foreign genes would have similar drastic effects.

Nevertheless, there is some justification for emphasizing the small size of the added piece. Even if it does generate a radical change in phenotype, it cannot do so by adding large amounts of new information; rather, like known pleiotropic mutations, it must mainly select prespecified but unexpressed potentialities of the resident genome. In that respect, I would expect that most such effects would not be far removed from changes that can occur by mutation in the nonengineered host.

One highly visible effect of genetic manipulation has been the development of giant mice by deliberate amplification of growth hormone genes (Palmiter et al., 1983). Any fantasies of the earth being overrun by giant rodents escaped from the laboratory are dispelled by knowledge that gigantism and dwarfism are traits that arise in all mammals by spontaneous mutation (perhaps sometimes by gene amplification) and that have not won out in natural selection. We should expect that introduction of heterologous genes for developmental hormones would sometimes generate bizarre monstrosities of unpredicted phenotype, but we should also expect similar abnormalities to arise by spontaneous mutation.

4.1 MICROEVOLUTION VERSUS MACROEVOLUTION

The importance of pleiotropy spills over into a question about natural evolution that has long been debated. For most of the present century, the dominant figures of population genetics have belonged to the microevolutionary school, which attributes most evolution to the gradual accumulation of small genetic changes that collectively produce major alterations in phenotype. Throughout the same period, a minority group, but one that has included some geneticists of great distinction, has proposed that sudden large discrete changes in genome organization are the pivotal evolutionary events. The most explicit advocate of macroevolution was Richard Goldschmidt, but some of the ideas propounded more recently by Barbara McClintock, as well as the recent vogue of the `punctuated equilibrium' concept of paleontologists, bear some of the same stamp (Maynard Smith, 1983).

Personally, I have found many discussions of this subject hard to follow because the issues are seldom explicitly defined. When microevolutionists speak of small changes, do they mean small changes in DNA sequence (such as simple base substitutions compared to complex rearrangements) or do they mean changes with small phenotypic effects (such as typical missense mutations or amplifications of genes already present in multiple copies), as contrasted to highly pleiotropic mutations of the types discussed here? The distinction was probably harder to make in the premolecular era, but it seems absolutely central. To my mind, the most important distinction is between simple and complex changes in DNA sequence or, more precisely, between changes that occur readily (which includes not only replication errors but also rearrangements facilitated by specific recombinases and transposases) contrasted to very rare or multiple events. Changes in the former category, however substantial their phenotypic effects, are happening all the time and therefore are already represented in the gene pools of existing species at frequencies set by rates of mutations and selection and parameters of population structure. For that reason, they seem readily assimilable into the general framework of microevolutionary theory. On the other hand, if DNA changes sufficiently rare to be effectively unique are major evolutionary determinants, a truly different, far more stochastic, description of evolution is required.

Macroevolution (especially in the sense of rare DNA changes indicated here) remains a minority viewpoint at best, for reasons I will not detail here. As the introduction of a foreign DNA segment constitutes a genuine macromutation, the possibility that such events are important in natural evolution is at least worth considering.

4.2 CONCLUSIONS

Small changes in genomes, either by mutation of critical components or by artificial manipulation, can effect major alterations in phenotype. However, even when the changes create new forms with a high degree of organization, the principal determinants of form lie in the residual genome rather than at the site of genetic alteration. The altered elements act mainly to elicit unexpressed potentialities of the genome. Two consequences are relevant to engineered organisms. First, engineering small alterations in genotype may have large (sometimes unpredicted) phenotypic consequences; second, equivalent changes are likely to have occurred also in natural populations without genetic manipulation. Assessment of the possible impact of engineered organisms depends on the extent to which evolutionary rates are limited by mutation rates and on the relative importance of microevolution versus macroevolution.

REFERENCES

Botstein, D. and Maurer, R. (1982) Genetic approaches to the analysis of microbial development. Ann. Rev. Genet. 16, 61-84.

Maynard Smith, J. (1983) The genetics of stasis and punctuation. Ann. Rev. Genet. 17, 11-25.

Palmiter, R.D., Norstedt, G., Gelinas, R.E., Hammer, R.E. and Brinster, R.L. (1983) Metallothionein-human GH fusion genes stimulate growth of mice. Science 222, 809-14.

Spotts, C.R. and Stanier, R.Y. (1961) Mechanism of streptomycin action on bacteria: a unitary hypothesis. Nature 192, 633-7.