SCOPE 44 - Introduction of Genetically Modified Organisms into the Environment

9.1 INTRODUCTION

In recent years, a number of classes of genetically engineered microorganisms (GEMs) have been developed for biotechnological processes that require introduction of the modified organisms into the environment in substantial numbers. This differs from earlier and the majority of current applications of GEMs, which involve GEM containment and ultimate destruction, and has elicited concern about potential risks for the environment. The majority of GEMs do not constitute particularly novel genotypes, nor are they intended for introduction into ecological situations that are unusual for the parental organism. Such GEMs do not therefore differ significantly from traditionally selected microbes with modified properties (e.g. vaccine strains of polio virus and Salmonella typhi Ty21a; nitrogen-fixing, plant root nodulated bacteria such as Rhizobium) that have been introduced in substantial quantities into the environment for extended periods of time without any untoward effects being detected. Nevertheless, the large number of different GEMs currently under development for introduction into the environment, a few of which may constitute novel genotypes, render dependence upon generalizations unjustified, particularly in view of the limited information available on the fate and ecological consequences of such organisms once introduced into target ecosystems.

There is therefore a need, on the one hand, to avoid irrational concerns that ignore the significant body of case history which indicates a lack of adverse effects on the environment by the vast majority of introduced modified microorganisms and, on the other, to augment substantially our base of relevant data so that we can assess more accurately the likely fate of a wider range of modified microbes. There is also a need to develop strategies to build into particularly novel GEMs, whose fate may be less easily predicted, or into GEMs perceived to represent some degree of risk, containment features that preclude the development of adverse consequences following introduction.

1. Its ability to survive and multiply in the ecosystem into which it is introduced
2. The stability of its new genetic material and the potential of this material to transfer horizontally to indigenous organisms
3. The ability of the GEM to carry out effectively the function for which it was designed under the conditions prevailing in the ecosystem
4. The effects, if any, of the GEM on the structure and function of the ecosystem into which it is introduced

9.2 SURVIVAL AND PROPAGATION OF A GEM IN ITS TARGET ECOSYSTEM

The most crucial property of a GEM which governs both its efficacy of function and its risk potential is its ability to survive and multiply in the ecosystem into which it is introduced. As a general rule, microbes well adapted to cultivation in the laboratory, through repeated passage in laboratory media, become less robust and persist less well upon return to the natural environment. Similarly, organisms originating from an ecosystem (e.g. topsoil) different from the target ecosystem (e.g. groundwater) are less likely to persist in comparison to those originally derived from the target ecosystem. The behaviour of a GEM constructed by modification of an organism derived from the target ecosystem will be more predictable and will most likely represent a lower risk than that of a GEM constructed from an organism isolated from a different ecosystem. These considerations lead to the conclusion that, other things being equal, selection of an organism from the ecosystem which is to be the target of the modified derivative of the organism, with as little propagation in the laboratory as possible before, during and after its modification, will provide a GEM of greatest efficacy and predictability and of lowest risk.

9.2.2 TECHNIQUES FOR TRACKING GEMS

In order to follow the fate of a GEM in the environment, it is necessary to detect and quantify it in time and space. Both the organism and the genetic information that constitutes the modification must be tracked simultaneously and independently so that both loss of the new information from the GEM and its possible lateral transfer to indigenous microorganisms can be assessed. The principal detection and quantification procedures for monitoring the survival and propagation of a GEM following its introduction into a particular ecosystem are listed in Table 9.1. These differ substantially from one another in specificity and sensitivity. A method exhibiting a high degree of specificity is clearly required for detecting a GEM in an, ecosystem containing closely related organisms. On the other hand, a highly sensitive procedure is needed in those cases where persistence of a GEM at very low numbers is important (e.g. in the case of a live vaccine strain whose principal aspect of risk is the possibility of it acquiring greater pathogenic potential). The method(s) used to track a GEM must therefore be selected on the basis of the quality of the information needed. Note that, in general, two procedures must be used in order to track both the organism itself and the newly acquired genetic information (Table 9.1).

Traditional methods for the detection and enumeration of specific microbes generally involve sample dilution and plating for single colonies on solidified selective media. The medium may be selective for a natural property of the organism or for a newly acquired property (e.g. lactose utilization, resistance to nalidixic acid) that has been introduced specifically for the purpose of tracking the GEM (Table 9.1) and which differs from the introduced property that constitutes the crucial new functional aspect of the GEM. Exceptionally, this latter property; may also serve as a basis for specific selection or detection of the GEM on solid media (Rojo et al., 1987), thereby enabling both the organism and the newly acquired genetic information to be independently tracked by plating techniques (D.F. Dwyer, S.W. Hooper, F. Rojo, and K.N. Timmis, submitted for publication). Where the newly acquired information does not itself confer a phenotypic property detectable by plating procedures, it can be directly linked to a marker that does (e.g. genes encoding lactose utilization (Barry, 1986; Drahos et al., 1986), catechol 2,3-dioxygenase, etc. (Mermod et al., 1986a)).

Table 9.1 Methods for tracking GEMs

Despite the simplicity of plating procedures, they do have a number of limitations in that they not only require the organism to be efficiently recoverable from environmental samples but also to be able to grow and develop colonies on the selective medium. Many organisms present in natural environments do not form colonies when directly plated on standard laboratory media, let alone strongly selective media, and require special resuscitation media and growth conditions in order to propagate themselves in the laboratory (Palmer et al., 1984; Lechevallier et al., 1987). It is also possible that a GEM may not express the genes whose functions are necessary for survival on the selective medium employed or that these genes may be lost while the GEM is in the environment.

The problems of recoverability of organisms from natural environments and their growth on laboratory media can be avoided by direct examination of samples or extracts of samples using an antibody specific for an antigen produced by the GEM. Direct enumeration of microbes may be effected by microscopic examination using fluorescent antibodies directed against a surface antigen (Bohlool and Schmidt, 1980) or indirect quantitation may be carried out on extracts of surface or non-surface antigens using standard immunological procedures. If monoclonal antibodies are employed, this method can have both high sensitivity and high specificity. Since GEMs constructed for introduction into the environment will generally express the newly acquired genetic information, immunological methods can be used to track the organism and the new information simultaneously. Note, however, that precise and reproducible results require a uniform level of synthesis of the marker antigens (and exposure on the cell surface in the case of surface antigens) under all environmental conditions encountered, and this may not always be the case.

The problem of variable expression of marker antigens is obviated by the use of nucleic acid hybridization procedures in conjunction with oligonucleotide probes for homologous sequences in the GEM. Oligonucleotide probes can have homology to a specific DNA sequence of the organism (e.g. to a specific sequence of the 23S RNA gene) or to the added genetic information. Alternatively, it can have homology to a short, rare oligonucleotide (fingerprint) introduced into the chromosome or into (or closely linked to) the newly acquired information, specifically for the purpose of tracking. Nucleic acid probes can be used in various ways. Colony hybridization methods (Grunstein and Hogness, 1975) involve the growth of bacteria isolated from environmental samples on laboratory media and the scoring of these colonies with the probe. This procedure has been shown to be fairly sensitive for the detection of specific bacteria added to environmental samples (Sayler et al., 1985) and for detecting the presence of specific genes in bacterial communities (Barkay et al., 1985). However, it requires the successful isolation of bacteria and their growth in the laboratory. Alternatively, probes can be used to analyse bacterial DNA directly isolated from environmental samples (Kuritza and Salyers, 1985). This procedure has no requirement for the prior laboratory growth of bacteria carrying the gene of interest and allows for quantification of particular genes. In this way, the size of the bacterial population containing the gene can be estimated. When used together the two methods can give a good estimate of the size of the population harbouring the homologous genes.

9.3 STABILITY OF ACQUIRED HETEROLOGOUS DNA AND ITS TRANSMISSION TO INDIGENOUS ORGANISMS

9.3.1 INSTABILITY OF STRUCTURE AND INHERITANCE OF CLONED DNA

Many hybrid plasmids tend not to be inherited stably in populations of actively dividing bacteria in the absence of continuous selection conditions that kill plasmid-free bacteria. There are several reasons for this (Table 9.2), including disequilibrations in cellular metabolism and physiology or the perturbation of membrane structure and function (in the case of cloned genes of membrane proteins), resulting from the synthesis of a heterologous protein or proteins (or their products), or the higher-than-normal rate of synthesis of a heterologous or homologous protein (Mermod et al., 1986a). The perturbation in optimal cellular functioning caused by a hybrid plasmid constitutes a selection pressure for elimination of the cloned gene. This may occur by insertional inactivation of the gene, deletion of the gene through recombination between repeated sequences bracketing the cloned sequence (or part of it) or through the activity of deletogenic sequences in the neighbourhood of the cloned gene, or by loss of the hybrid plasmid. Which of these processes is dominant in any situation depends upon the nature of sequences in the hybrid plasmid, the properties of the host strain, and particularly the inherent instability of the hybrid plasmid in growing cells. The molecular basis of instability of an acquired property can often be determined precisely by structural analysis of the cloned genes in individual clones or in mixed populations through the Southern hybridization procedure (Southern, 1975; Meinkoth and Wahl, 1984), using the cloned gene as a probe.

Table 9.2 Some causes of instability of DNA sequences cloned in plasmid vectors 

9.3.2 STRATEGIES TO INCREASE STABILITY: INSERTION INTO THE CHROMOSOME

The main causes of instability of a hybrid plasmid, namely the high-level expression of cloned genes and inherent instability of many current cloning vectors, may be addressed in several ways. Whereas high-level expression may result from the existence of an unnatural constitutive or partially constitutive promoter upstream of and oriented towards the cloned gene, it often reflects a gene dosage effect of DNA sequences cloned into high copy number vectors, such as pBR322 and pUC derivatives. The cloning of genes together with their native promoters and natural regulatory determinants into low copy number vectors such as pRK290 (Ditta et al., 1980) or pLG339 (Stoker et al., 1982) generally produces hybrid plasmids with improved stability. The best stability is, however, obtained in most cases through eliminating altogether the gene dosage phenomenon, i.e. through reducing the copy number of the gene to one per chromosome, the natural level for many genes, by inserting the gene into the chromosome itself (Rojo et al., 1987). This generally has three important positive consequences: (1) reduction in expression of the cloned gene to its normal level; (2) insertion of the gene into an essential replicon that cannot by definition be lost from a viable cell; and (3) reduction in the tendency of the gene to transfer laterally to other organisms (see below).

There are basically two strategies for inserting cloned genes into host chromosomes. The most versatile strategy is to insert the genes into a transposon present on a suitable delivery system and then to cause transposition of the hybrid transposon into the chromosome of the selected host strain (Simon et al., 1983; Tait et al., 1983). Typically, the transposon into which the gene of interest is cloned is located on a plasmid that can be propagated stably in one host (e.g. Escherichia coli) but not in target bacteria (e.g. Pseudomonas putida) but that can conjugally transfer at high frequency to the latter (Mermod et al., 1986a). The hybrid plasmid is then transferred by conjugation to the target organism, and transconjugants expressing a marker of the transposon are selected. Since the plasmid itself cannot survive in the target organism, the only means by which the transposon marker can persist is through transposition of the hybrid transposon from the donor plasmid to a stable replicon (e.g. the chromosome) of the recipient. Although some derivatives obtained in this way will contain the hybrid transposon in an important gene, and will thus be compromized in their metabolism and growth, others will not and will be healthy hybrids. The versatility of this approach is that a single transposon can be used for many different target bacteria. However, much development of transposon cloning vectors remains to be done.

The other strategy is to clone the gene into a previously cloned segment of the chromosome of the host to be used and then to affect targeted insertion of the gene by recombination between the regions of homology on the chromosome and the introduced hybrid plasmid (Harayama and Don, 1985). Since the site for insertion on the chromosome is selected for an absence of deleterious effects of the manipulations, all targeted insertions obtained should be healthy. However, with this strategy a new cloning vector must be constructed essentially each time a different host is used.

9.3.3 HORIZONTAL TRANSFER OF CLONED DNA

A critical aspect of a GEM is whether it horizontally transfers its new information to indigenous microorganisms and, if so, to which members and at what frequency. Non-GEMs transfer DNA in soil culture (Graham and Istock, 1978; Weinberg and Stotzky, 1972) and thus the potential for transfer by GEMs in nature exists. It should be emphasized that horizontal transfer of such information is not necessarily undesirable because it may lead to the evolution of organisms that carry out the desired activity better than the original GEM. However, since horizontal transfer in the environment cannot be controlled and its consequences may not be predictable a priori, effort should be made to reduce it to a minimum until sufficient data are obtained to permit valid predictions.

A number of current cloning vectors, particularly broad host range vectors that are used extensively for genetic engineering of soil bacteria, are readily transmissible when present in a cell concommitantly carrying a conjugative plasmid such as F, ColI, RP1/RP4/RK2, etc., because they retain the mobilization functions (mob and oriT) of the original replicon from which they were constructed (Bagdasarian et al., 1979; Bagdasarian and Timmis, 1982). In the case of the pKT series of vectors, transfer-deficient (Mob) derivatives are available that, in combination with Pseudomonas putida host strain KT2440, are certified by the Office of Recombinant DNA Activities, National Institutes of Health, USA, as the cloning vectors of an HV1 (hostvector 1) system (Bagdasarian et al., 1981). However, even with these and with Mob HV1 vectors of E. coli such as pBR322, low-frequency conjugal transfer can be demonstrated under laboratory conditions (Guyer, 1978). There is, however, only limited information about the ability of either hybrid plasmids engineered for a particular function in the environment or naturally occuring plasmids to transfer to indigenous organisms in their target environment or in model ecosystems that closely mirror target environments (Gowland and Slater, 1984; Gealt et al., 1985; Mancini et al., 1987).

Horizontal transfer can also occur by transduction, following infection of the donor by a virus, or by transformation, following death and lysis of the donor organisms. Again there is little information available on frequencies of transfer of specifically engineered (i.e. non-model) plasmids or natural plasmids by these processes in natural environments or model ecosystems (Saye et al., 1987). It seems, however, likely that transfer frequencies under such conditions will reflect both the copy number of the new gene and the ability of it either to become a relatively small replicon capable of independent propagation (i.e. a plasmid) or to become integrated into an existing replicon (e.g. the chromosome) in the new host. It is therefore to be expected that, all things being equal, a gene on a high copy number plasmid will be transferred at a higher frequency than one on a low copy number plasmid, and that a gene integrated in the chromosome will be transmitted at the lowest frequency, both because it is present as a single copy and because it constitutes only a tiny part of a very large replicon that in all probability will be extremely inefficient in transferring to a new organism and integrating into its chromosome.

9.3.4 STRATEGIES TO DIMINISH HORIZONTAL TRANSFER OF DNA FROM GEMS

Although the institution of additional barriers to horizontal transfer has not been extensively explored, it is quite conceivable that this would be readily achievable (Table 9.3). In the case of the conjugal transfer of bacterial DNA, although it is known that most of the functions needed for conjugation and mobilization are plasmid-encoded, host mutants have been described that are blocked in conjugal transfer (Willetts and Wilkins, 1984). The introduction of such defects into GEMs carrying conjugation- and mobilization-deficient vectors, and the analysis of transfer frequencies from such bacteria with combined transfer defects, would seem well worth exploring. Moreover, it is not beyond the realms of current capabilities to take a GEM system exhibiting low transfer rates (by any means) and to engineer it in such a way that any recipient of the modified DNA segment would rapidly die. There are a number of genes (e.g. genes of restriction endonucleases, colicins, etc.) that, when expressed in the absence of another gene (e.g. genes of modification enzymes, colicin immunity proteins, etc.), cause killing of the cell (e.g. see Kuhn et al., 1986). If the 'kill' gene is inserted close to the DNA fragment in question and the 'protection' gene inserted at a site far from this fragment or, preferably, into another replicon in the cell, then the GEM itself would be viable whereas a recipient of the 'kill' gene would be killed because co-acquisition of the `protection' gene in a low transfer system would be exceptionally low. The potential utility of a variation on this theme, involving the plasmid Rl hok gene, was recently demonstrated (Molin et al., 1987). Note that engineering a GEM to produce an active and stable nuclease may greatly diminish the transforming ability of its DNA released after cell death (e.g. see Timmis and Winkler, 1973).

Table 9.3 Horizontal transfer of cloned genes and possible means of prevention

 The assessment of horizontal transfer of cloned genes can be accomplished with a variety of approaches. The potential for transfer under different conditions can be studied conveniently in the laboratory. Conjugal transfer of the new information of the GEM to selected recipients using filter matings and strongly selective conditions for the isolation of transconjugants enables detection of low transfer frequencies under optimal conditions. Similarly, the influence on low transfer frequencies of a third partner carrying a promiscuous conjugative plasmid can be examined using triparental matings. Transfer frequencies between pairs and triples of bacteria on more natural surfaces (sand, clay, etc.) can be analysed following adsorption to or biofilm formation on such surfaces by the partners. The potential influence on transfer rates of nutrients, temperature, pH, ionic strength, the presence of noxious compounds, etc., can all be assessed in such types of experiments.

The isolation of phages able to grow on the donor strain enables assessment of the potential for transductional transfer of the new information to selected recipients. The isolation of total DNA and its application to selected recipients present in different media and metabolic states provides information on the potential for the transfer of new information by transformation.

Actual transfer rates in natural environments are less readily assessed because conditions are not easily controlled and because all non-GEM bacteria present in a target ecosystem, most of which cannot be individually selected, are potential recipients. Simultaneous addition of the GEM and a target recipient to an ecosystem, with assessment only of target recipients having acquired the new information, is a means of obtaining precise data, but this information may have limited significance.

Model ecosystems offer a useful intermediate stage between a natural ecosystem and laboratory experiments, in that conditions can be more carefully controlled and the system can be fully contained, thus avoiding the possibility of introducing GEMs into the environment before they or their potential effects are sufficiently characterized (Sagik et al., 1981).

Procedures available to detect and measure gene transfer range from the direct selection of transconjugants using selection markers of the recipients and of the transferred DNA to indirect detection of transconjugants using nucleic acid and/or antibody probes and/or plating techniques to separately detect the donor and the new genetic information, and thereby to identify non-donor organisms carrying the new information. Whereas the direct selection of transconjugants detects low transfer frequencies, indirect detection systems are only able to reveal high transfer frequencies because of the presence of donor bacteria. Since donor bacteria need not be present in experiments designed to evaluate transduction (application of phages that were propagated on the donor) and transformation (application of donor DNA), detection systems here can be more sensitive because of the lower background. However, the absence of donor bacteria is unnatural and may negatively influence transduction and transformation frequencies, particularly if proximity of donor and recipient (as is presumably the case with bacteria absorbed to soil particles or growing in biofilms) is important for such mechanisms of DNA transfer.

9.4 GEM FUNCTIONING

In order to be useful, a GEM must survive for a minimal period of time and express its new genetic information (and most probably a number of other non-modified key functions) for at least this period of time and at some minimal level, in the target environment in order to successfully accomplish the job for which it was designed. In this regard, it is encouraging that aromatic-degrading bacteria which have either been isolated from activated sewage sludge (Focht and Shelton, 1987) or which have modified genotypes obtained by conjugative plasmid transfer (Jain et al., 1987) both survive and function as aromatic degraders upon introduction into soil and groundwater microcosms.

Clearly there are a large number of variables that will influence these properties of the GEM and that must be evaluated. The most obvious of these is the ability of the GEM to survive in its target environment, and parameters important for this have been alluded to above. Similarly, the stability of the modified DNA in the GEM is of crucial importance and some strategies to optimize stability, including retention of native gene regulation systems, have been considered (Table 9.4).

For many applications, there will be no specific environmental signal that can be used to switch on expression of cloned genes when the engineered function is needed and to switch them off when not needed. Here, the cloned genes will need to be expressed constitutively and this may represent an energetic (and perhaps a physiological) burden on the GEM that will reduce its survival or the survival of the modified genetic information (see above). In other applications, however, it will be possible and preferable to retain native regulatory circuits that do respond to appropriate environmental signals: e.g. the expression of adhesion antigens and other virulence associated factors at 37°C (body temperature; e.g. see Timmis and Manning, 1986) by live vaccines and the expression of catabolic genes in response to the presence of specific pollutants in microbes designed to degrade such pollutants (Ramos et al., 1987). In such cases, the lack of or low-level expression of the cloned genes when their products are not required will diminish their energetic and physiological burden on the GEM and will have less effect on its survival.

Table 9.4 Engineering GEMs for function

The use of microbes to degrade chemicals toxic at the p.p.b. level poses a particular problem because organic chemicals typically induce synthesis of enzymes for their catabolism when present only in p.p.m. concentrations (Mermod et al., 1986b). Thus, catabolic genes whose expression is controlled by native regulatory circuits will stop being expressed and the desired activity of the GEM will cease to function when the level of the pollutant drops to a level that still constitutes a significant environmental hazard. Maintenance of the synthesis of catabolic enzymes at p.p.b. levels of the pollutant can be accomplished through having constitutive synthesis of the catabolic enzymes, but since catabolic pathways typically involve many enzymes, this would constitute a considerable drain upon the energy resources of the cell. It is therefore preferable to maintain the native regulatory circuits but to engineer into the system a back-up expression circuit that would switch on low-level synthesis of the catabolic enzymes when the cell finds itself in substrate-poor environments, i.e. in response to a starvation or stress signal (Table 9.4). Further means of promoting elimination of pollutants present at low levels include the incorporation or engineering of high-affinity transport proteins for certain compounds and the manipulation of chemotactic responses to accomplish directional movement of bacteria towards substrates adsorbed to particulate matter.

GEM functioning (Table 9.4) might also be improved by the incorporation of properties that permit specific adhesion to surfaces upon which the organism would function optimally or that provide protection from certain predators active in the target environment. Where the function of the GEM involves metabolism of a particular substrate that causes imbalances in parameters like NADH production, engineering of these secondary parameters to compensate the imbalances may be useful if feasible (Ingram et al., 1987).

9.5 EFFECT OF THE GEM ON STRUCTURE AND FUNCTION OF TARGET ECOSYSTEM

At present we have rather limited information on microbial ecology and about microbial ecosystem structure and function. In some, particularly anaerobic ones, a significant proportion of the bacteria present have neither been satisfactorily classified nor named, nor are they readily isolated by standard procedures. It is therefore rather difficult at present to examine effects on the structure and function of some ecosystems provoked by introduction of a GEM. However, the techniques necessary for studying microbial ecosystems are now available and we can anticipate great progress in this area over the next few years. Moreover, a bank of data on the interactions of microbes with their environment and with each other is growing (Liang et al., 1982; Stotzky and Krasovsky, 1981; Stotzky and Babich, 1984) and may soon enable concepts and mathematical models of microbial interactions to be formulated. Until this time, studies on the influence of introduced GEMs on ecosystem structure and function will rely primarily on measurement of transfer of modified DNA from the GEM to indigenous organisms, using procedures outlined above, indirect assessment of population complexity, and analysis of general aspects of community functioning.

9.5.1 METHODS TO FOLLOW THE FATE OF INDIVIDUAL MICROORGANISMS IN AN ECOSYSTEM

In simple communities consisting of only a few distinct species, all of which can be isolated and propagated in pure culture, it will be relatively straightforward to generate specific antibody or nucleic acid probes that will permit each member to be quantified in time and space. Even simpler perhaps will be the use of pulsed field electrophoresis (Schwartz and Cantor, 1984) and contour clamped homogeneous field electrophoresis (Chu et al., 1986) to analyse total DNA extracted from samples and cut with an infrequent cutting restriction endonuclease. A knowledge of the fragmentation pattern of individual members may permit quantification of all members from the patterns obtained with total DNA extracts. This method should even be applicable to larger communities if combined with nucleic acid hybridization using a mixture of probes, each one specific for one of the members. Another simple analytical procedure that may be useful is gas chromatography (GC) of fatty acids (Smith et al., 1986). Different species contain different fatty acids that appear as a distinct GC profile. Provided that the community is not complex, GC analysis of total fatty acids will be a rapid means of quantifying all members. Another approach is to identify certain bacteria by determining the sequences of unique segments of 5S rRNA (MacDonell and Colwell, 1984).

In more complex communities, higher resolution is required and the preparation of individual reagents specific to each member may be impractical. In such cases, which are likely to be the majority, indirect assessment of community complexity and the individual tracking of uncharacterized organisms by means of probes may be necessary. One promising approach is the raising of polyclonal antibodies against cell surface antigens of entire microbial populations, and their use to analyse by Western blotting (Towbin et al., 1979) total proteins extracted from community samples and fractionated on two-dimensional polyacrylamide gels (O'Farrell, 1975; Neidhardt et al., 1983). Subsequent probing of such blots with antibodies to one or several specific antigens would provide additional information about subpopulations in the community.

9.5.2 ASSESSMENT OF ECOSYSTEM FUNCTIONING

To predict the impact of a GEM upon a given ecosystem requires knowledge about the reciprocal effects of the GEM and its environment and of the physical and biological parameters that control certain activities (Pritchard et al., 1987). Ecological factors such as temperature, soil type, pH, nutrient and moisture availability, competition, and predation by indigenous microorganisms, etc. (Stotzky and Babich, 1984; Stotzky and Krasovsky, 1981), which affect survival and activity as well as the stability and expression of introduced genetic information of GEMs and indigenous microorganisms, need to be delineated. In certain cases it will be possible to assess direct competitive effects, e.g. the exclusion of epiphytic bacteria by introduced Ice mutants of Pseudomonas syringae (Lindow, 1987). In others, however, possible effects may only be detected, indirectly by measurement of alterations in global ecosystem processes such as energy flow and nutrient cycling (Levin and Harwell, 1986).

Determination of a GEM's possible effect on an ecosystem, as well as its ability to function effectively therein, may be conveniently ascertained using model systems. A data base covering the interplay of model GEMs and various types of model ecosystems would provide: (1) a predictive means for evaluating whether introduction of a given GEM into the environment would be accompanied by significant risk, (2) guidelines for constructing GEMs to meet certain environmental standards for activity and survival, and (3) methods for determining the fate of a GEM and of its modified genetic component(s) following environmental introduction (Levin et al., 1987).

9.6 ENGINEERING GEMS FOR ENVIRONMENTAL SAFETY

Two aspects relating to the safety of GEMs to be introduced into the environment are the risk that exists because sufficient information on the behaviour of the GEM in the target ecosystem is lacking and the more serious risk that is associated with the use of GEMs derived from microbes that exhibit a known hazard, such as live vaccine strains derived from pathogenic microorganisms. In the first instance, prudence indicates that the new genetic information should not be able to transfer laterally at significant frequencies and the GEM should not be able to propagate itself at a rate that allows it to significantly disturb the normal functioning of the ecosystem or to displace any members of the community. In the second, the GEM must additionally be modified so that it is unable to regain its prior hazard potential and so that it has only limited capacity to survive in its target ecosystem.

Currently available information suggests that most GEMs will not be as competitive as indigenous members of an ecosystem, will tend to be maintained at low levels, and thus will not have a significant impact on the ecosystem structure or function. This appears to be the case for Pseudomonas sp. B13 derivatives able to degrade chlorinated benzoates and phenols that have been studied in a model activated sludge ecosystem (D.F. Dwyer, S.W. Hooper, F. Rojo, and K.N. Timmis, submitted for publication). When such bacteria are introduced into this ecosystem at a level equal to that of all other bacteria in the ecosystem able to form colonies on laboratory media, their numbers quickly fall to 0.10.01% of this level and then stabilize. These numbers only increase significantly if the ecosystem is placed under strong selection pressure for the GEM function and even then the increase is modest and not maintained. The GEM nevertheless functions in the way for which it was designed. Thus, as indicated above, an organism taken from the target environment and exhibiting no pathogenic potential for animals or plants, and modified in the laboratory to perform some specific task, is unlikely to provoke changes in the community structure and function when reintroduced into its environment. This is of course provided that steps are taken to minimize lateral transfer of its modified DNA to indigenous organisms.

If the host strain is judged to be safe but there seems to be a risk associated with the modified or new DNA, then the disadvantages of having the DNA on a plasmid vector, in terms of instability of the GEM, can constitute advantages in terms of safety (Table 9.5). In this case, vector plasmid instability can be increased further by deleting DNA sequences (par) necessary for precise partitioning of the vector between daughter cells at cell division (e.g. see Cohen et al., 1986).

If the GEM itself is judged to represent a significant risk there are two general strategies to diminish its risk. One is to reduce its survival capacity such that it can survive for only a limited period of time outside of the laboratory. For example, strains of Salmonella currently under development as live vaccines have been attenuated by the introduction of non-reverting mutations that inactivate either the ability to synthesize aromatic amino acids (Levine et al., 1987) or the cAMP-CRP global regulatory circuit of the cell (Curtiss and Kelly, 1987). Other global circuits that might be good targets for reducing the survival of GEMs include the RecA recombination system and the Rel stringent response.

Table 9.5 Engineering for safety/containment

1.	Unstable inheritance of introduced genes (unstable plasmid, par)
2.	Poor host survival (auxotroph, recA, defective in global regulation of metabolism)
3.	Autodestruct mechanism (e.g. restriction+ , bacterocin+)

A second approach is to engineer the GEM in such a way that it dies rapidly once its specific function is no longer required. The construction of such an autodestruct GEM is possible where expression of the function is regulated by an environmental signal, such as the presence of a pollutant, a temperature of 37^°C, etc. One strategy here would be to take a bicomponent 'kill-protection' system (e.g. restriction-modification; bacteriocinbacteriocin immunity) and to arrange for the `kill' gene to be expressed constitutively and the 'protection' gene to be expressed from the same regulated promoter that is responsible for expression of the GEM function and whose activity is modulated by an environmental signal. For example, if the protection gene in a vaccine strain is expressed from a promoter that is active at 37°C but not at temperatures below 30^°C, then the GEM will be viable while in the individual being vaccinated but not when it passes out of the body, where it is exposed to lower temperatures and expresses only the kill gene (see also the use of the hok system; Molin et al., 1987).

9.7 CONCLUDING REMARKS

The introduction of GEMs into the environment carries with it not only the promise of health, ecological and economic benefits, but also the responsibility to protect the environment from any potential adverse consequences. It is therefore essential to devise strategies and develop methods for obtaining qualitative and quantitative data concerning survival, competitiveness, etc., of GEMs and of their ability to function in the ecosystems into which they are introduced.

In this overview of experimental approaches to construct safe and effective GEMs for environmental applications, we have attempted to point out those criteria that in our view are important for making strategic choices. On the one hand, we have been dogmatic where we feel a particular choice is dictated. On the other, we have in some instances been highly speculative because new approaches and new applications of developing technologies are clearly needed and we do not yet know with any certainty which of these will ultimately turn out to be most useful.

What is apparent is that, although much basic information is urgently needed about microbial ecology and about the possible influence on the structure and function of an ecosystem into which a GEM is introduced, the experimental techniques to obtain such information are now available or can readily be developed. The application of modern molecular biological and immunological techniques, as well as analytical instrumentation for GC, high-pressure liquid chromatography, mass spectrometry, nuclear magnetic resonance, etc., will undoubtedly revolutionize microbial ecology during the next few years and lead to significant advances in our understanding of fundamental processes of ecosystem functioning and of the important attributes desirable in a GEM destined to be introduced into the environment. 

REFERENCES

Bagdasarian, M., Bagdasarian, M.M., Coleman, S. and Timmis, K.N. (1979) New vector plasmids for gene cloning in Pseudomonas. In Timmis, K.N. and Pühler, A. (Eds.) Plasmids of Medical, Environmental and Commercial Importance, Elsevier/North Holland, Amsterdam, pp. 411-22.

Barkay, T., Fouts, D.L. and Olson, B.H. (1985) Preparation of a DNA gene probe for detection of mercury resistance genes in gram-negative bacterial communities. Appl. Environ. Microbiol. 49, 686-92.

Bohlool, B.B. and Schmidt, E.L. (1980) The immunofluorescence approach in microbial ecology. Adv. Microbiol. Ecol. 4, 203.

Cohen, S.N., Biek, D.P. and Miller, C.A. (1986). Chromosomal and extrachromosomal functions that affect plasmid stability in E. coli. In:Levy, S.B. and Novick, R.P. (Eds.) Banbury Report 24: Antibiotic Resistance Genes: Ecology, Transfer, and Expression, Cold Spring Harbor Laboratory, Cold Spring Harbor, New York, pp. 247-61.

Ditta, G., Stanfields, S., Corbin, D. and Helinski, D.R. (1980) Broad host range DNA cloning system for gram-negative bacteria: construction of a gene bank of Rhizobium meliloti. Proc. Natl Acad. Sci. USA. 77, 7347-51.

Focht, D.D. and Shelton, D. (1987). Growth kinetics of Pseudomonas alcaligenes C-O relative to inoculation and 3-chlorobenzoate metabolism in soil. Appl. Environ. Microbiol. 53, 1846-9.

Gowland, P.C. and Slater, J.H. (1984) Transfer and stability of drug resistance plasmids in Escherichia coli K12. Microbiol. Ecol. 10, 1-13.

Grunstein, M. and Hogness, D.S. (1975) Colony hybridization: a method for the isolation of cloned DNAs that contain a specific gene. Proc. Natl Acad. Sci. USA 72, 3961-5.

Harayama, S. and Don, R.H. (1985) Catabolic plasmids: their analysis and utilization in the manipulation of bacterial metabolic activities. In Setlow, J.K. and Hollaender, A. (Eds.) Genetic Engineering, Vol. 7, Plenum Press, New York, pp. 283-307.

Jain, R.K., Sayler, G.S., Wilson, J.T., Houston, L. and Pacia, D. (1987) Maintenance and stability of introduced genotypes in groundwater aquifer material. Appl. Environ. Microbiol. 53, 996-1002.

Kuritza, A.P. and A.A. Salyers (1985) Use of a species-specific DNA hybridization probe for enumerating Bacteroides vulgatus in human feces. Appl. Environ. Microbiol. 50, 958-64.

Lechevallier, M.W., Camper, A.K., Broadaway, S.C., Henson, J.M. and McFeters, G.A. (1987) Sensitivity of genetically engineered organisms to selective media. Appl. Environ. Microbiol., 53, 606-9.

Levin, M.A., Seidler, R., Borquin, A.W., Fowle, J.R. and Barkay, T. (1987) EPA developing methods to assess environmental release. Bio/Technology 5, 38-45.

Levine, M.M., Herrington, D., Morris, J.G., Lasonsky, G., Murphy, J., Tall, B. and Stocker, B. (1987) Safety, inactivity, immunogenicity, and in vivo stability of two attenuated auxotrophic mutant strains of Salmonella typhi 541Ty and 543Ty, used as oral vaccines in man. J. Clin. Invest. 79, 888-902.

Lindow, S.E. (1987) Competitive exclusion of ephiphytic bacteria by Ice-Pseudomonas syringae mutants. Appl. Environ. Microbiol. 53, 2520-7.

Mancini, P., Fertels, S., Nave, D. and Gealt, M.A. (1987) Mobilization of plasmid pHSV106 from Escherichia coli HB101 in a laboratory-scale waste treatment facility. Appl. Environ. Microbiol. 53, 665-71.

Meinkoth, J. and Wahl, G. (1984) Hybridization of nucleic acids immobilized on solid supports. Analyt. Biochem. 138, 267-84.

Mermod, N., Ramos, J.L., Lehrbach, P.R. and Timmis, K.N. (1986b) Vector for regulated expression of cloned genes in a wide range of gram-negative bacteria. J. Bacteriol. 167, 447-54.

Neidhardt, F.C., Vaughn, V., Phillips, T.A. and Bloch, P.L. (1983) Gene-protein index of Escherichia coli K-12. Microbiol. Rev. 47, 231-84.

Palmer, L.M., Baya, A.M., Grimes, D.J. and Colwell, R.R. (1984) Molecular, genetic and phenotypic alteration of Escherichia coli in natural water microcosms containing toxic chemicals. FEMS Microbiol. Lett. 21, 169-73.

Ramos, J.L., Mermod, N. and Timmis, K.N. (1987). Regulatory circuits controlling transcription of TOL plasmid operon encoding meta-cleavage pathway for degradation of alkylbenzoates by Pseudomonas. Molec. Microbiol. 1, 293-300. 

Rojo, F., Pieper, D.H., Engesser, K.-H., Knackmuss, H.-J. and Timmis, K.N. (1987) Assemblage of ortho cleavage route for simultaneous degradation of chloro-and methylaromatics. Science 238, 1395-8.

Saye, D.J., Orgunseitan, O., Sayler, G.S. and Miller, R.V. (1987) Potential for transduction of plasmids in a natural freshwater environment: effect of plasmid donor concentration and a natural microbial community on transduction in Pseudomonas aeruginosa. Appl. Environ. Microbiol. 53, 987-95.

Schwartz, D.C. and Cantor, C.R. (1984) Separation of yeast chromosome-sized DNAs by pulsed field gradient gel electrophoresis. Cell 37, 67-75.

Smith, G.A., Nickels, J.S., Kerger, B.D., Davis, J.D., Collins, S.P., Wilson, J.T., McNabb, J.F. and White, D.C. (1986) Quantitative characterization of microbial biomass and community structure in subsurface material: a prokaryotic consortium responsive to organic contamination. Can. J. Microbiol. 32, 104-11.

Stoker, N.G., Fairweather, N.F. and Spratt, B.G. (1982). Versatile low-copy-number plasmid vectors for cloning in Escherichia coli. Gene 18, 335-41.

Stotzky, G. and Krasovsky, V.N. (1981) Ecological factors that affect survival, establishment, growth and genetic recombination of microbes in natural habitats. In: Levy, S.B., Clowes, R.C. and Koenig, E.L. (Eds.) Molecular Biology, Pathogenicity, and Ecology of Bacterial Plasmids, Plenum Press, New York, pp. 31-42.

Timmis, K.N. and Manning, P.A. (1986) Virulence factors in enteropathogenic bacteria. In: Falmagne, P., Alouf, J.E., Fehrenbach, F.J., Jeljaszewicz, J. and Thelestam, M. (Eds.) Bacterial Protein Toxins, Gustav Fischer Verlag, Stuttgart, pp. 301-13.

Towbin, H., Staehelin, T. and Gordon, J. (1979) Electrophoretic transfer of proteins from polyacrylamide gels to nitrocellulose sheets: procedure and some applications. Proc. Natl Acad. Sci. U.S.A. 76, 4350-4.

Willetts, N. and Wilkins, B. (1984) Processing of plasmid DNA during bacterial conjugation. Microbiol. Rev. 48, 24-41.