Many crops bearing transgenes coding for highly specific enzymes that metabolically catabolize herbicides have been generated (Cole and Rodgers, 2000). These include, for example, bromoxynil resistance crops bearing a nitrilase (Stalker et al., 1996), glufosinate resistance crops bearing an ace-tyltransferase (Vasil, 1996), 2,4-D-resisting crops bearing a highly specific soluble cytochrome P-450 monooxygenase (Streber and Willmitzer, 1989) phenmidipham-resisting crops bearing a bacterial gene (Streber et al., 1994), and dalapon-resisting crops bearing a dehalogenase (Buchanan-Wollaston et al., 1992). Of these, only the bromoxynil- and glufosinate-resistant crops have reached commercialization. All of the genes used commercially are of bacterial or actinomycete origin, despite the fact that plants contain genes for herbicide resistance, which is the basis for most natural metabolic selec-tivities used for 45 years. Nevertheless, plant genes conferring metabolic resistance have not yet been used commercially. There are reports on using nonprokaryotic genes to confer resistance, but none of this work is yet commercial, and whether they confer sufficient resistance is not clear. A rabbit esterase gene conferred resistance to thiazopyr via degradation (Feng et al., 1998). The expression of plant and animal P-450 transgenes conferred phe-nylurea resistance (Inui et al., 1999; Siminsky et al., 1999). Transgenes encoding maize glutathione transferases increased the level of herbicide resistance (Jepson et al., 1997).
Unlike the target site resistances that are fraught with problems (see later), the crops generated with metabolic resistances seem to be problem free, with little metabolic load conferred by generating the small amount of enzyme needed. The toxicology is simplified because the transgene product typically initiates a cascade of events whereby the herbicide eventually disappears.
There has been an assumption that one cannot use catabolic enzymes to confer resistance to fast-acting herbicides. This view is of doubtful rele vance as biotechnologists should be able to perform as well as plants. Inhibitors of protoporphyrinogen IX oxidase (protox), which actually cause the accumulation of the photodynamically toxic product (Boger and Waka-bayashi, 1999), induce photodynamic death of plants within 4-6 hours in bright sunlight. Beans are immune to some members of this group, e.g., acifluorfen, because they possess a specific homoglutathione transferase and contain enough homoglutathione to degrade these herbicides stoichiometri-cally before they can damage the crop (Skipsey et al., 1997). Similarly, strains of Conyza bonariensis contain a complex of enzymes capable of detoxifying the reactive oxygen species generated by the photosystem I blocker paraquat and keeping the plants alive until the paraquat is dissipated (Ye and Gressel, 2000).
As almost all herbicides are degraded either in the soil or in some plant species, one should be able to find more genes for catabolic resistance to those herbicides and then rapidly generate herbicide-resistant crops with metabolic resistance rather than with target site resistance.
Biotechnology has been used to generate target site HRC crops by two means—transfer of field- or laboratory-generated mutants into crop varieties by genetic means or transgenically.
One must assume a fitness penalty with any target site HRC. This is simply due to the fact that if the resistant mutation was neutral or very nearly neutral there would be naturally resistant populations preexisting in the wild due to Sewall Wright drift. As all the mutations were found at a low frequency, one in a million or less, the resistant traits are not near neutral fitness. This was most obvious with triazine resistance, which was crossed from weedy populations of Brassica campestris (-B. rapa) that evolved resistance into oilseed rape (B. napa) (Souza-Machado, 1982). This psbA gly264ser mutation is plastid inherited, allowing researchers to test fitness in the near-isonuclear eighth backcross (Gressel and Ben-Sinai, 1985), and in isonuclear reciprocal hybrids (Beversdorf et al., 1988). There was a consistent 20% productivity loss when the crop was grown alone and a much greater fitness loss when the resistant biotypes were grown in competition with the sensitive biotype. A 20% yield loss does not mean that such crops are without value; this triazine-resistant oilseed rape is widely grown in Australia because some pernicious weeds can be inexpensively controlled. Still, the triazine-resistant varieties will probably eventually be supplanted by transgenics, which do not have the fitness penalty.
Despite a plethora of papers claiming no fitness penalty in acetolactate synthase mutant and transgenic plants (see Saari et al., 1994), elegant ex periments by Bergelson et al. (1996) with transgenic Arabidopsis did indeed show that such a penalty exists. Previous competitive fitness experiments started with transplants and/or did not measure seed output. The two most competitive phases in a plant's life cycle are establishment, when large populations of seedlings self-thin to a single plant per parent, and fertilization, when thousands of pollen grains compete on stigma and through style. If competition is not measured seed to seed, then the fitness penalty is unknown.
The natural resistance mutations to ALS and ACCase derived from maize tissue cultures or via pollen mutagenesis and then backcrossed into inbreds seem to have produced normally yielding hybrids. There have been more problems with the transgenics than the mutants, as successful as they have been in the marketplace. The magnitude of resistance with transgenics is not as high as it is with the natural mutations, with both ALS and EPSP synthase target site resistances. The first single-site mutant prolOlser EPSP synthase transgenic plants tested were insufficiently resistant to glyphosate for field use. Thus, considerable effort was made to increase the level of resistance and repair the huge shift in Km toward natural substrates by both adding second site-directed mutations as well as enhanced promoters (Pad-gette et al., 1996; Lebrun et al., 1997). The necessity to do this was the basis for the claimed invincibility of glyphosate-resistant crops from evolving resistance in weeds; it would take too many simultaneous mutations for resistance to evolve (Bradshaw et al., 1997). This thesis was published after there was already known natural metabolic resistance in legumes (Komossa et al., 1992) as well as natural resistance due to enhanced levels of the target enzyme in Lotus corniculatus (Boerboom et al., 1990).
These flaws in the invincibility theory were pointed out by Dyer (1994) and reiterated and updated by Gressel (1996). The only difference that could be found between a recently evolved glyphosate-resistant Lolium rigidum population having a sevenfold increase in the I50 value and the wild type was the presence of a doubled level of EPSP synthase (Grays et al., 1999). Lotus corniculatus with a doubled level of EPSP synthase also had a sevenfold increase in glyphosate resistance (Boerboom et al., 1990). This is not the only case in which it has been possible to confer herbicide resistance by overexpression of the gene encoding its target site. Glufosinate resistance was achieved in tissue culture by this means with overproduction of gluta-mine synthase (Deak et al., 1988). Similarly, poplars were transformed for elevated glutamine synthase to enhance nitrogen utilization (Gallardo et al., 1999). The transgenic poplars are more resistant to glufosinate than the wild type (F. M. Cánovas, personal communication). In addition, transgenic tobacco plants with fivefold overexpression of protoporphyrinogen IX oxidase in chloroplasts are resistant to a discriminatory dose of 20 /xM
acifluorfen, which severely inhibited the wild type (Lermontova and Grimm, 2000). Unfortunately, in the latter case there are no quantitative data on the magnitude of resistance.
The nail in the coffin of invincibility came with the identification of glyphosate-resistant Eleucine indica in Malaysia (Lee and Ngim, 2000). This material was resistant to far greater than typical field application rates of glyphosate. It was disclosed that this material bore a target site resistance to glyphosate (Tran et al., 1999). The single mutation in the binding domain of this material was at a position equivalent to the prolOlser in the Salmonella typhimurium aroA resistance gene (Tran et al., 1999), the same position that was regarded as insufficiently resistant in trangenic resistant genes (Padgette et al., 1996).
In retrospect, there is good reason to expect that transgenics will be less resistant than target site mutants bearing the same transversion. Recurrent selection of a natural mutant will select for homozygous resistant individuals. This cannot happen with the present generation of transgenics. They bear and express both the native and transgenic enzymes (although the ratios of the two have never been published to the best of my knowledge). Thus, transgenic plants with target site resistance are functionally heterozygous and will remain so despite recurrent selection. Homozygosity can be achieved at present only when totally exogenous genes such as those for metabolic resistance are used, as they do not compete with native genes. When the herbicide is applied, target site HRC must depend on the transgenic derived enzyme while the native is inhibited, perhaps even causing phytotoxic precursors to accumulate.
There have been continuing problems with some glyphosate-resistant crops, suggesting that the critical balance of transgenic and native enzymes is not optimal. Initial varieties of maize, about to be released 4 years ago, were retracted because late-season herbicide application caused pollen sterility. Similar problems of obtaining vegetative (only) glyphosate-resistant cotton plants continue to plague cotton growers. The period of pollen production is well known to be subject to inhibition by minor chemical, metabolic, or environmental perturbations. There have been reports of stem brittleness and cracking of untreated glyphosate-resistant soybeans (Anonymous, 2000), suggesting an overproduction of product leading to increased lignin formation when both the native and transgene-derived enzymes are operative.
One does not know whether such glyphosate-resistant transgenic plants are transiently weakened until the herbicide is dissipated. This period after glyphosate application is analogous to sublethal treatments of nontransgenic plants with glyphosate. When legumes are given sublethal applications of glyphosate, they are unable to produce phytoalexins after elicitation and are far more susceptible to disease (Sharon et al., 1992). A study purporting to ascertain whether phytoalexin biosynthesis was compromised after glyphos-ate treatment of glyphosate-resistant soybeans (Lee et al., 2000) was arti-factual. The authors measured only constitutive levels of glyceollin, without elicitation, and their conclusions, concerning phytoalexins, which are by definition elicited, are without basis.
There are at least two apparent solutions to the problems arising from the functional heterozygote status of transgenic target site resistant plants. One solution is to try to enhance the metabolism of the herbicide, obviating the problem. This may have been done using the gox gene coding for an enzyme-degrading glyphosate together with target site resistance (Mannerlof, 1997). There are no data indicating how quickly the transformants degraded the glyphosate in this case. A better solution might be to begin with a "clean slate." Target site resistant plants should be generated via replacement, i.e., in plants that have had the native gene deleted or functionally irreversibly "knocked out." This can be done by classical deletion mutagenesis with ultraviolet (UV) or gamma irradiation, but this causes multiple lesions and the screening for particular mutants is not easy. Molecular techniques show far more promise as they are more "surgical," i.e., there is less likelihood of multiple deletions along a chromosome, leading to complications from linked deletions. These techniques include T-DNA tagging (Zupan and Zam-bryski, 1995; Choe and Feldmann, 1998) and transposon mutagenesis (Tis-sier et al., 1999). There are an estimated 3000 genes that might be target sites of herbicides (Berg et al., 1999), giving mutations that will have to be rescued by the specific gene products of interest as the selectable markers. The most elegant new technique is with specific viral vectors spliced to the gene to be silenced. After cloning into the crop, some "black magic" mechanism virus-induced gene silencing turns off both the viral and the endogenous gene (Baulcombe, 1999). At least one of the virus vectors used is transmitted through seeds and by pollen, so following generations remain with the gene suppressed (Lister and Murant, 1967), but whether the gene is never reactivated is not certain.
After the native gene is suppressed, it should be possible to obtain transgenics with various levels of transgene expression in functional homozygotes and choose those with the optimal expression level for agricultural release. The gene used may have to be very different in sequence from the native gene as viral-induced gene silencing can suppress the native gene.
V. RESISTANCE MANAGEMENT—THE NEED FOR STACKED GENES
Probably >10 million ha of agricultural lands are infested with weeds that are resistant to one or more classes of diverse herbicide chemistry (Heap,
2000). Proactive resistance management is finally being considered; previously, there was a smug assumption that industry would continue to develop many new chemicals. Now there is a realization that T-HRC is almost all there will be that is new for a long while, and measures should be instituted to delay weeds from evolving resistance. One way to delay the evolution of herbicide resistance in weeds is to stack two genes for herbicide resistance and require that the farmers use only a mixture of herbicides. This is useful because it considerably lowers the mutation frequency for resistance in the weed. For example, if the frequency of resistance to one herbicide is 10 h and the other 10 the resistance to them stacked together is 10 14 on condition that both herbicides are used. The use of stacking is very important where one wishes to preserve the ability of an excellent herbicide such as glyphosate and one has a pernicious weed with a propensity to evolve resistance, e.g., Loliitm rigidum in Australia.
There are two hazards that must be assessed with glyphosate and other BD-HR wheats: the risk of grass weeds such as Lolium rigidum evolving resistance to the herbicide and the risk of introgression of the gene into related weeds. If glyphosate resistance is to be engineered into wheat, it should be stacked with a gene coding for resistance to a second unrelated graminicide. The herbicide mixtures should always be used when such wheat is cultivated. Glyphosate used alone will clearly further engender evolution of glyphosate resistance (Gressel, 1996; Powles et al., 1998; Pratley et al., 1999), and/or bring about a shift in weed spectra toward weeds that have never been controlled by glyphosate (Owen, 1997). The use of stacked BD-HR wheat and herbicide mixtures should delay the resistance of Lolium to glyphosate. The risk of introgression of stacked genes into wheat-related weeds will be discussed later.
When stacked resistance genes and herbicide mixtures are used, it is possible that there will be cases of enhanced weed control by the mixture. Still, some mixtures may be contraindicated vis-à-vis resistance management. Simply combining herbicides with different modes of action will not result in delaying resistance if the efficacy and temporal activity characteristics of the mixed herbicides do not match (Roush et al., 1990). Both mixing partners must effectively inhibit the weeds most sensitive to the vulnerable herbicide because the greater selection pressure weed species are the most likely to evolve target site resistance (Gressel and Segel, 1982; Maxwell et al., 1990). Resistance could quickly evolve in a weed species that is naturally resistant to one of the herbicides in a mixture.
The components of the mixture need to have similar persistence or the mixing partner with a low mutation frequency must have longer persistence than the vulnerable one with a high mutation frequency of resistance. Otherwise, there will be a period when only the vulnerable partner is present, and it will select for resistance in the target weed as if there were no mixture at all. This would be the case when a persistent resistance-vulnerable ALS inhibitor is mixed with a short-lived phenoxy herbicide (Wrubel and Gressel, 1994). Unlike crops, which have been selected to germinate uniformly shortly after planting, seeds of many weed species display many flushes of germination during a cropping season (Bewley and Black, 1982). If a resistance-prone weed species has multiple flushes during the season and the vulnerable herbicide has a longer period of activity than the mixing partner does, then the vulnerable herbicide selects for individuals resistant only to it after the mixing partner has dissipated. A mixture that is not well matched for persistence can still be effective if all the weeds germinate over a short period of time and both herbicides outlast the germination.
The ideal mixing partner should have three other properties (Wrubel and Gressel, 1994) in addition to equal persistence:
1. It should have a different target site of action from the vulnerable herbicide.
2. The mixing partner should not be degraded in the same manner as the vulnerable herbicide. For example, if the vulnerable herbicide is degraded in the crop by a glutathione transferase, the mixing partner should have no chemical site that can be attacked by the same enzyme.
3. Another useful attribute in a mixing partner would be to possess negative cross-resistance, i.e., where individuals resistant to the vulnerable herbicide are more susceptible to the mixing partner than the wild type. This would actually reduce the frequency of resistant alleles in the weed population. This strategy was first proposed for herbicides on the basis of laboratory data (Gressel and Segel, 1990), and information on the existence of negative cross-resistance has been published at the whole plant level (Gad-amski et al., 2000).
Was this article helpful?