W. Allen Miller, Gennadiy Koev, and B. R. Mohan Iowa State University, Ames
!RE 4HERE 2ISKS !SSOCIATED WITH 4RANSGENIC 2ESISTANCE TO ,UTEOVIRUSES Engineering crops for disease resistance is one of the first and most successful examples of the application of plant genetic engineering for crop improvement. Numerous, diverse approaches have been used to genetically engineer virus resistance in plants. These will not be reviewed in this article; instead the reader is advised to read other reviews (4,25,49,78). By far the most common and successful general strategy has been pathogen-derived resistance, in which plants are transformed with a gene or sequence from the virus. These plants show varying levels of resistance, by a variety of mechanisms, to the virus from which the gene was derived. The two most common viral genes used are those coding for the coat protein and the RNA-dependent RNA polymerase. The polymerase, also known as replicase, is the enzyme that copies the viral genome. (Although the two terms differ somewhat in meaning, for the purposes of this review they will be used interchangeably.) The coat protein gene
Dr. Miller’s address is: Plant Pathology Department, 351 Bessey Hall, Iowa State University, Ames 50011; 515/294-2436, Fax: 515/294-9420, E-mail:
[email protected], W3: http://www. public.iastate.edu/~wamiller/
Publication no. D-1997-0523-04F © 1997 The American Phytopathological Society
700
Plant Disease / Vol. 81 No. 7
tends to confer resistance to a broader range of related viruses, but resistance is often incomplete, with the plant showing delayed and milder symptoms than the untransformed plant. Resistance can be overcome by extremely high levels of inoculum or by inoculation with naked viral RNA (31). In contrast, polymerase (replicase) gene-mediated resistance can confer complete immunity, but only to virus strains with very high sequence homology to the one from which the transgene was derived (28). Because transgenic resistance using these two viral genes is the most widely used, including against luteoviruses, most of the discussion of risk will deal with these two approaches. Like many other major technological advances, transgenic resistance to pathogens offers not only advantages but also potential risks (71). These potential risks have aroused controversy (18,35). We use a simple definition of risk paraphrased from Goy and Duesing (29): (probability of an event occurring) × (potential damage or loss resulting from the event) = risk. Throughout the paper, we will distinguish between probability of an event happening and the potential damage, because they are separate aspects of any interaction. This review is predicated on the notion that (i) comparison of luteoviral genome sequences, (ii) understanding the replication mechanisms, (iii) observation of interac-
tions of luteoviruses with each other and with their hosts, and (iv) observation of cross-hybridization between luteovirus host plants and weedy relatives allow prediction of potential risks in virus-derived transgenic resistance strategies. We discuss three categories involving risk. The first comprises interactions that do not involve any genomic rearrangements between the transgene product and an invading virus. The second category is recombination between the transgene and the invading virus that can create new strains or viruses. The third category is the potential escape of the transgene via pollen to weedy relatives of the transgenic host. Discussion of the first two points is particularly relevant because we will provide evidence that the probability of these events occurring may be greater for luteoviruses than for most other plant viruses. Luteoviruses. Luteoviruses represent one of the most economically significant groups of plant viruses. The most-studied members include barley yellow dwarf viruses (BYDVs), beet western yellows virus (BWYV), and potato leafroll virus (PLRV), all of which cause serious losses on their hosts and are worldwide in distribution (44,47). Luteoviruses (62) are phloem-limited, spherical viruses that often cause yellowing, reddening, and/or stunting of their hosts (47) (Fig. 1). They are not mechanically transmissible. Instead, they are transmitted by aphids (Fig. 2) in a circulative, nonpropagative manner only by certain aphid species. Intimate interactions between luteoviruses and their vectors have co-evolved (59). The viral genome consists of a single RNA molecule that is about 5.5 kb long and codes for 5 to 6 proteins (46,48). Efforts are under way to genetically engineer resistance to these three viruses and to other luteoviruses such as beet mild yellowing virus, cucurbit aphid-borne yellows virus, and groundnut rosette assistor virus. Based on cytopathology, serology, and genome sequences, luteoviruses have been divided into two distinct subgroups. This distinction is revealed when the genome organizations of the subgroups are compared (Fig. 3). Two essential genes, those coding for the RNA-dependent RNA polymerase and the coat protein, have very different origins, depending on the subgroup (26,48). The polymerase genes of subgroup I luteoviruses are more closely related to those of the dianthovirus (e.g., red clover necrotic mosaic virus), umbravirus (e.g., carrot mottle virus), and carmovirus (e.g., carnation mottle virus) groups
than they are to genes of the subgroup II luteovirus polymerases. The polymerases of subgroup II luteoviruses are most closely related to those of the sobemoviruses (e.g., southern bean mosaic virus). Phylogenetic analyses reveal that the RNA-dependent RNA polymerase genes of subgroup I and subgroup II luteoviruses are as different as such genes can get (40,80). No other virus group in any kingdom has such an extreme dichotomy in polymerase gene origins (40,80). In contrast, the coat protein genes of all luteoviruses are much more closely related to each other than to the coat protein genes of viruses in any other group. The blue shading in Figure 3
demarcates the region of similarity between the subgroups. The other luteovirus genes that share intragroup homology include an extension to the coat protein that is probably required for aphid transmission (AT, Fig. 3) (6,9) and possibly cell-to-cell movement (MP?) (6), and an overlapping gene (MP?, Fig. 3) that is required for systemic infection by BYDV (9) but not by BWYV (82). It is likely that these genes in this common region confer on luteoviruses their distinctive properties, including icosahedral particle shape, circulative, nonpropagative transmission by aphids, serological cross-reactivity, and confinement to the phloem.
Interactions Between Transgene Products and Virus Heterologous encapsidation. Coat protein–mediated resistance is probably the most widely used form of virus-derived transgenic resistance (4). This strategy has been applied successfully against viruses of many groups, including the luteoviruses (38). The first virus-resistant transgenic plant to be marketed was squash expressing coat protein of zucchini yellow mosaic virus (2), and it is likely that coat protein– mediated resistance will be used widely in the near future. In addition to its obvious roles in encapsidating and protecting the viral genome, plant virus coat proteins can
Fig. 1. Luteovirus disease symptoms and synergistic interactions. H indicates uninoculated plants. (A) Natural field infection of oats (Avena sativa) by unidentified strain(s) of barley yellow dwarf virus (BYDV). Severely infected plant on left; moderate symptoms on right. Note the sterile florets (white, wispy “dead heads”) that produce no grain and the reddening or yellowing of leaves. (B and C) Oats infected with different strains of RPV (R), PAV (P), or both (R+P) BYDVs. In panel B, the host is GAF/Park oats on which the RPV-NY isolate is mild, and the PAV-IL isolate is so severe that the mixed infection is only slightly more severe than PAV alone. In panel C, on an Australian cultivar, stunting is the most obvious symptom caused by Australian isolates, and the mixed infection causes more stunting than either isolate alone (photo by P. M. Waterhouse). (D and E) Carrot motley dwarf disease caused by mixed infection of carrot red leaf luteovirus and carrot mottle umbravirus (photos by B. W. Falk). (F) Shepherd’s purse (Capsella bursa-pastoris) plants infected with a mild strain of beet western yellows virus (BWYV) (B) or the ST9 strain which contains mild BWYV RNA plus ST9-associated RNA (B+ST9a) (photo by B. W. Falk). Plant Disease / July 1997
701
affect important biological properties. The coat protein plays a role in virus movement within the plant, in manifestation of disease symptoms, and in determining aphid transmission properties. Does expression of a viral coat protein in transgenic plants pose a risk? One possibility is heterologous encapsidation (Fig. 4). If a transgenic plant expressing a coat protein of virus A becomes infected with virus B, there is a chance that genomic RNA of virus B can get encapsidated in the transgenically expressed virus A coat protein and thereby acquire characteristics determined by virus A coat protein, including the ability to
move in the plant or altered vector specificity. Examples that support this scenario are known. When coat protein–defective mutants of tobacco mosaic virus (TMV) were complemented by the coat protein produced in transgenic tobacco plants, the mutant acquired the ability to spread systemically in the plant, which the nonencapsidated mutant virus was unable to do alone (55). Also, an aphid-nontransmissible mutant of zucchini yellow mosaic virus acquired aphid transmissibility due to heterologous encapsidation upon inoculation to a plum pox virus coat protein–expressing plant (42). Heterologous encapsidation
Fig. 2. Rhopalosiphum padi (oat bird-cherry aphid), the vector for PAV and RPV barley yellow dwarf viruses.
Fig. 3. Genome organizations of luteovirus subgroups. Representative members of each subgroup are listed below each map. Solid black lines represent the viral genomic RNA. Boxes indicate genes. Blue shading, genes with sequence similarities between subgroups; yellow, sequence similarity to umbra-, diantho-, and carmoviruses; green, sequence similarity to sobemoviruses. POL, RNA-dependent RNA polymerase; PRO, putative protease; CP, coat protein; MP?, putative movement protein; AT, read-through domain of the coat protein gene required for aphid transmission. BYDV, barley yellow dwarf virus; SDV, soybean dwarf virus; SCRLV, subterranean clover red leaf virus; BWYV, beet western yellows virus; BMYV, beet mild yellowing virus; CABYV, cucurbit aphid-borne virus; PLRV, potato leafroll virus; GRAV, groundnut rosette assistor virus; CRLV, carrot red leaf virus. 702
Plant Disease / Vol. 81 No. 7
was also observed in transgenic potato plants expressing the coat protein gene of the O strain of potato virus Y (PVY O) upon infection with the N strain of the same virus (PVYN) (20). In natural and laboratory settings, heterologous encapsidation has been observed in mixed luteovirus infections (barley yellow dwarf luteoviruses) which resulted in altered vector specificities (64,77). The numerous and widespread examples of heterologous encapsidation interactions among luteoviruses suggest that this may be a natural determinant of luteovirus epidemiology (65). Thus, it has been speculated that transgenic heterologous encapsidation interactions might occur between a genome of an infecting luteovirus (or other virus) and the transgenically expressed coat protein of another luteovirus or serotype (Fig. 4). The risk of transgenic heterologous encapsidation is extremely low, in both the probability and potential damage variables of the risk equation. Regarding probability, in all reported cases of transgenic expression of luteovirus coat protein, the level of coat protein is so low as to be undetectable or nearly so (3,38,39). The coat protein produced by the invading virus would be orders of magnitude greater in concentration and would thus overwhelmingly favor encapsidation in the invading virus’s own coat protein. Secondly, it is possible that the coat protein does not confer all of the aphid specificity determinants. A lowabundance form of the coat protein that contains a long extension at one end, produced by read-through of the stop codon during translation, is probably required for aphid transmission (6,10,21,36,75) (Fig. 1). The extended region may also confer vector specificity. This extended form has not been used in constructs employing coat protein–mediated resistance. Thus, any heterologously encapsidated RNA may not acquire the vector specificity of the virus from which the transgene is derived. Even if the invading RNA did acquire new vector specificity from the transgenic coat protein, the potential damage is low; the heterologously encapsidated virus could be transmitted only once by the new aphid vector because the genetic material (RNA) in the heterologously encapsidated virus still encodes its own vector specificity. The spread would be limited to a rare transmission event (which could happen anyway because aphid vector specificity is not absolute) within the field of transgenic plants and adjacent plants. Additional research is forthcoming to determine whether vector specificity is determined by the coat protein or by the fused read-through protein. Future research strategies could then further minimize risk by deleting or altering the vector transmission determinant(s) in the transgenic coat protein gene. An intriguing new perceived risk has come to light with the observation that
mixed infections of potato spindle tuber viroid (PSTVd, an infectious RNA with no coat protein) and PLRV can result in encapsidation of PSTVd RNA in PLRV virions and subsequent aphid transmission of PSTVd (61). This could potentially increase the spread of PSTVd, which has no natural vector and is transmitted only by mechanical means. However, the encapsidation is highly inefficient, with one PSTVd molecule encapsidated for every 3,000 to 5,000 PLRV genomes (61). Furthermore, due to the nearly undetectable levels of transgenically expressed coat protein and to the absence of the coat protein read-through domain, such an event need not be considered a significant risk. Synergistic interactions. Of greater concern should be the possibility for synergistic interactions between the virusderived transgene product and a challenging virus. In certain combinations, mixed infections of two viruses produce symptoms much more severe than those caused by either of the viruses alone. Generally, these viruses are unrelated or distantly related, as closely related viruses tend to cross-protect against one another (24). Such synergisms are quite common among luteoviruses and their relatives. As previously described, luteoviruses can be divided into two subgroups based on the homologies of the polymerase genes (Fig. 1). This dichotomy can be extended to related luteo-like viruses and RNAs. These include the umbraviruses, the enamoviruses, and the ST9-associated (ST9a) RNA that is associated with the ST9 isolate of BWYV. ST9a RNA enhances BWYV replication and greatly exacerbates disease symptoms (Fig. 1F) (19). This RNA codes for a subgroup I–like polymerase (11) and can replicate autonomously in laboratory experiments (56), but it lacks genes for many luteoviral functions, including a coat protein. The coat-proteinless umbraviruses also have subgroup I–like polymerase genes, are capable of autonomous replication, and are invariably found associated with subgroup II luteoviruses, upon which they depend for aphid transmission (54). Carrot motley dwarf disease results from a mixed infection of carrot mottle umbravirus and carrot red leaf luteovirus (Fig. 1D and E). The only known enamovirus is pea enation mosaic virus (PEMV). PEMV contains two RNAs, each of which can replicate autonomously in plant cells, but which depend on each other for cell-to-cell movement and encapsidation functions (13). All of these viruses and RNAs have been found in various pairs in which an RNA coding for a subgroup I–like RNA polymerase enhances replication of an RNA coding for a subgroup II–like polymerase which, in turn, benefits the subgroup I–like RNA (Table 1). These viruses and RNAs seem to represent an evolutionary continuum ranging from (i) luteoviruses that replicate on their own but repli-
cate even better when combined with a synergistic partner of the other subgroup, to (ii) symbiotic RNAs that rely on a luteovirus for a function such as encapsidation and in exchange somehow enhance the luteovirus’s accumulation (ST9a RNA and umbraviruses), to (iii) two luteoviruslike RNAs that have become so interdependent that they have become a single bipartite virus (PEMV). Can individual viral transgenes that confer resistance to one virus act synergistically with unrelated infecting viruses to exacerbate symptoms? This is not yet known for luteoviruses, but the work of Vance et al. (74) demonstrates clearly that this can occur for a different set of synergistically interacting viruses. The mixed infection of two unrelated viruses, potato virus X (PVX, a potexvirus) and a potyvirus such as potato virus Y (PVY), tobacco vein mottling virus (TVMV), or tobacco etch virus (TEV), is more severe in to-
bacco plants than infection by either virus alone (74). Transgenic plants expressing a specific portion of either the TEV or the TVMV genome showed the severe symptoms characteristic of the synergistic mixed infection upon inoculation with PVX alone. Thus, plants expressing a transgene from one virus actually were more severely affected when infected by an unrelated virus. This type of risk could be controlled by simply removing the transgenic crop variety from production. However, the consequences are by no means trivial. The widespread use of crop plants that contained a susceptibility gene analogous to the described synergy events led to one of the worst epiphytotics in U.S. history. The 1970 southern corn leaf blight epidemic resulted from millions of acres being planted with corn containing the T-cytoplasm, which confers cytoplasmic male sterility (cms). Unfortunately, the same gene that confers cms (t-urf13) also con-
Fig. 4. Potential heterologous encapsidation of an invading virus by a transgenically expressed coat protein. Colored lines represent RNA of invading virus (green) or transgenically expressed coat protein message (red). Ribosomes (gray spheres) translate coat protein mRNA (colored lines) to produce coat protein subunits (colored spheres), which assemble on viral RNA to form virions shown at the bottom. As an example, subgroup II barley yellow dwarf virus (BYDV)-RPV (green) could probably infect a plant transformed with subgroup I BYDV-PAV coat protein gene (red), because coat protein amino acid sequences of these viruses are only about 50% identical. If an RPV RNA was encapsidated in enough PAV coat protein, it could acquire the ability to be transmitted by the English grain aphid, Sitobion avenae, which is a vector for PAV but not for RPV. Drawing indicates how coat protein expressed from abundant replicating viral mRNA would be predicted to accumulate at much higher levels than transgenically expressed coat protein. Thus, phenotypic mixing and transcapsidation with significant levels of transgenic coat protein would be rare (indicated by
