MSV Variation

The following is an expansion of the talk given at the (First!) Maize Streak Disease Symposium in September, 1997, at the Sanbonani Lodge in Hazyview, South Africa.

STUDYING THE DIVERSITY OF MAIZE STREAK VIRUS

EP Rybicki*, J Willment, FL Hughes, G Napier, H Jones, S Dennis and MB von Wechmar.

Abstract

For many years after the discovery of maize streak virus (MSV), it was believed that there was only a very limited genotypic range of the viruses, and that viruses infecting wild grasses and sugarcane were "host-adapted" strains of MSV. This finding was largely an artifact of the limited and mainly biological methods used at the time; however, serological work on the viruses tended to reinforce the assumption. The first complete nucleotide sequences obtained in the mid-1980s for severe maize-infecting viruses also showed very little (<3%) variation. Use of other nucleic acid-based techniques in the 1980s, however, revealed unexpected diversity: differential hybridisation and whole-genome restriction endonuclease mapping showed that there were very distinct groups of "African cereal streak viruses"; polymerase chain reaction-based DNA amplification and partial sequencing of a wide variety of viruses indicated both host-related as well as geographic variation among viruses, reinforced by total genome sequencing. Phylogenetic analysis of total and partial sequences shows that there is a cluster of closely-related viruses causing severe symptoms in maize, a distinct group of Panicum spp.-infecting viruses, and another distinct sugarcane streak virus (SSV) group. We have also defined a distinct MSV-related cluster of viruses infecting grasses and wheat in South Africa, all of which could be defined as "mild MSVs" in maize. Use of a differential panel of maize and other cereal hosts in leafhopper transmission or Agrobacterium tumefaciens-mediated "agroinoculation" experiments also shows that genetically diverse viruses also vary in their symptom expression and host ranges . The relative merits of different approaches for examining MSV and related virus diversity will be discussed, with examples.

Introduction

Maize streak virus (MSV) is the type member of the newly-renamed Mastrevirus genus of the family Geminiviridae: these are a group of leafhopper-transmitted viruses with geminate particles, and single-component single-stranded circular DNA genomes. The genus - formerly known as Subgroup I - also includes the dicot-infecting tobacco yellow dwarf (TYDV) and bean yellow dwarf viruses (BeYDV; Liu et al., 1997) from Australasia and South Africa respectively; chloris striate mosaic virus (CSMV) from Australia; digitaria streak virus (DSV) from Vanuatu; miscanthus streak virus (MiSV) from Japan; panicum streak virus (PanSV) from Africa; sugarcane streak virus (SSV) from Africa and Mauritius; and wheat dwarf virus (WDV) from Europe (Briddon and Markham, 1995); ICTV Executive Committee Proceedings, May, 1997). MSV, PanSV and SSV (and surprisingly, DSV) constitute a cluster of viruses termed the "African streak virus group" (Hughes et al., 1992): these are more closely related to one another than to any of the other Mastreviruses (Rybicki, 1994). Similarly, there is an "Australasian striate mosaic virus group", of viruses related to CSMV (Pinner et al., 1992).

Maize streak disease was first described - as "mealie variegation" - in the Natal Province of South Africa at the turn of the century (Fuller, 1901). The disease occurs only in Africa and adjacent Indian Ocean islands, where it is one of the worst occurring in maize. Breeding projects in Nigeria, South Africa, Zimbabwe, La Reunion and elsewhere have resulted in the development of germ plasm with a high degree of resistance to the virus; however, it continues to be a problem wherever small farmers cannot afford to use insecticides or tolerant varieties of maize.

The rationale for studying the diversity of MSV is to biologically define the range of the pathogen in terms of its potential for causing epiphytotics in, and yield reduction in, maize crops. Various studies have been performed on the genetic diversity of the streak virus of maize, starting with Storey’s pioneering work early this century, and continuing with work by McClean and others (see Rybicki, 1988; also McClean, 1947; and Storey and McClean, 1930). These almost exclusively biological investigations revealed that there was a relatively wide range of viruses to be found in maize and grasses and sugarcane; however, much of what was found caused only mild symptoms in maize. So-called "A" and "B" types of MSV were described; additionally, sugarcane and grass streak viruses were supposed to be "host-adapted strains" of MSV (Storey and McClean, 1930; Bock and Bailey, 1989). It is obvious, however, that as maize and sugarcane are both introduced hosts, the virus(es) must have or have had an original indigenous grass host or hosts.

Much of the work which predated the discovery of the circular ssDNA genome of the virus is flawed, in that these workers had no proof other than biological properties or single-insect transfers that they had "pure" isolates of virus. Additionally, they had no measures of difference of the virus(es), due to the lack of defined properties. Indeed, it was only in 1974 that any even preliminary molecular information was published on any geminivirus (Bock et al. 1974), and only in 1977 that it was determined that they had ssDNA - and not RNA - genomes, and a single coat protein (Harrison et al. 1977). The development of better means of virus characterisation and better purification schedules allowed more sophisticated means of analysis of diversity, which are the subject of this paper. It is our intention to compare and evaluate, for their relevance and applicability to MSV and other Mastrevirus isolate differentiation, techniques such as serology, nucleic acid hybridisation, restriction fragment length polymorphisms (RFLPs), restriction endonuclease mapping, polymerase chain reaction (PCR) DNA amplification, and partial and complete genome sequencing. In addition, we will discuss the continuing value of host range and virulence determinations on individual viruses.

Host range and symptom determination

Studies done in recent times - with the benefit of molecular techniques as backup - have to a large extent simply confirmed what good work the pioneers did, in the almost complete absence of any facilities worth the name other than insect isolation chambers. Thus, Pinner et al. (1988) described a wide range of "MSV" genotypes in terms of their biological effects in a susceptible maize cross (Golden x Bantam), as well as by leafhopper (Cicadulina mbila Naude) transmission to a number of other hosts. They essentially concluded that there were distinct "strains" of MSV infecting hosts such as Panicum maximum and sugarcane, and a group of similar isolates causing severe disease in maize, and that there were differences in host range between the different groups of viruses. In later work, both Mesfin et al. (1992) and von Wechmar and Hughes (1990) showed that local isolates of grass and maize viruses could be reliably differentiated using C. mbila transmission to "differential panels" of locally-available maizes and grasses, and that only viruses isolated from maize reliably caused severe symptoms in tolerant maize varieties. In the latter case, it could be shown that mild isolates of MSV and grass viruses could be differentiated from severe maize isolates of MSV by differential genomic hybridisation as well, indicating substantial differences in sequence. There were also almost always subtle but reproducible differences in the symptoms produced by severe or moderately severe MSVs on different maize lines, allowing their differentiation from one another. Thus, the picture that emerged was of a wide diversity of viruses in the environment, only some of which are capable of causing severe disease in maize.

Most recent work from our laboratory (Schnippenkoetter et al., 1997, this volume; Martin and Rybicki, 1997; this volume, and unpublished results) has involved the use of virus isolates made "agroinfectious" - that is, clones of partial or complete dimers of viral genomes in Agrobacterium tumefaciens, used to recreate the native virus by injection into a young maize plant (Grimsley et al. 1987) - to obtain truly clonal virus preparations. These were subsequently used for leafhopper transmissions to determine host ranges as detailed by von Wechmar and Hughes (1990) (and W Schnippenkoetter et al., unpublished), or used as agroinfectious preparations to determine host ranges by direct injection of seedlings (Martin et al., 1997, this volume). All of our work to date has shown that the cloned genomes we have tested as being characteristic of a virus isolate have the same biological characteristics as the "natural isolate" of the virus. This has possibly been the first instance of testing whether a cloned virus has the same leafhopper-transmitted host range as the "natural" or parent isolate, and almost certainly one of the first detailed tests of how agroinfection compares to leafhopper transmission in terms of determination of host ranges and symptom severity in individual maize lines and other hosts.

Serological techniques

Results obtained using enzyme linked immunosorbent assay (ELISA)-based serological techniques initially reinforced earlier notions, in that all MSV-like maize and grass viruses appeared very similar, and in the case of maize viruses, effectively serologically identical (Dekker et al. 1988). Thus, these authors reinforced the impression of Pinner et al. (1988), that all streak viruses were strains of MSV. Later work from the same group (Pinner and Markham, 1990; Pinner et al. 1992) took into account proven sequence differences between viruses, and produced results which strongly reinforced conclusions drawn from PCR amplification and sequencing of different viruses (see below). These workers and Clarke et al. (1989) and Peterschmitt et al. (1991) could all differentiate to some small extent between severe maize types of MSV, but these differences - even using monoclonal antibodies (MoAbs; Dekker et al. 1988; Peterschmitt et al. 1991) - were slight enough to indicate that serology was better suited to showing similarities between distinct viruses, than to differentiating closely related isolates. A composite relationship dendrogram - derived from the work of Pinner et al. (1992) and Rybicki (1991) - indicating the relative scale of the differences between isolates and distinct viruses, is shown in Figure 1.

Differential hybridisation

Simple hybridisation of radioisotope-labelled cloned MSV DNA to genomic DNA immobilised onto membranes was probably first used for detection purposes by Boulton and Markham (Boulton and Markham, 1986): part of this work was the ingenious use of leafhopper "squash blots" to demonstrate that MSV probably does not replicate in the insect vector. Rybicki et al. (1989) also demonstrated the use of simple hybridisation techniques for MSV detection in "dot blots" of sap extracts.

Under a defined set of reaction conditions, the extent of DNA cross-hybridization between two viral DNAs is determined by the number of regions of shared sequence similarity and the extent of sequence similarity within those regions. Thus, stable hybridization between viral DNAs having a particular level of overall similarity to be to one another can be obtained by manipulating the stringency of the hybridization and post-hybridization washing conditions used. While this technique is not strictly quantitative, the results correlate well with known levels of sequence similarity between geminiviruses. For example, under the stringency conditions used by Hughes (1991), digitaria streak virus (DSV; from Vanuatu), MSV, and sugarcane streak virus from Natal (SSV-N) are mutually non-cross-hybridizing, in hybridisations on whole replicative form (RF) dsDNA isolated from infected plants using different distinct digoxygenin-labelled cloned DNAs as probes. This correlates with the extent of sequence homology, as each shares approximately 60% total DNA sequence with any of the other viruses (Hughes, 1991). It was also shown in this study (Hughes et al. 1992) that one could differentiate between maize-type MSVs, sugarcane streak virus (SSV)-type viruses, and Panicum-infecting viruses (PanSVs). Thus, differential hybridisation could be used to differentiate between distinct African grass-infecting Mastreviruses, but not between maize-type isolates.

Studies such as the ones mentioned above have shown that increasing severity of disease symptoms in maize seems to be correlated with increasing concentrations of virus particles (Mesfin et al., 1992), and especially of replicative form dsDNA (Hughes, 1991): it is possible that "milder" strains / variants of MSV either do not replicate as well as more virulent types, or do not spread as well in the infected plant, or both.

Restriction fragment length polymorphisms (RFLPs)

A relatively simple technique for demonstrating even quite limited sequence differences between two stretches of double-stranded (ds) DNA is to cut them into fragments with a panel of restriction endonucleases, electrophorese the fragments, and either view them on ethidium bromide-stained agarose gels directly, or reveal their presence by "Southern blotting", or hybridisation with a labelled nucleic acid probe. It was apparent from early studies on MSV DNA in infected plants that large amounts ds replicative form (RF) DNA was made in especially severely affected maize: thus it is an obvious extension of hybridisation studies on whole genomic DNA to cut the abundantly-available dsRF DNA with enzymes to show up subtle genotypic differences (Clarke et al. 1989; Hughes, 1991; EP Rybicki et al., unpublished). Although it is useful to use hybridisation for detection of MSV-specific DNA if one cuts total infected maize plant DNA, it is possible to quite quickly and simply purify sufficient dsRF DNA to allow excellent results by simple ethidium bromide staining (Hughes, 1991; Palmer et al. 1997). The latter reference in particular gives simple protocols for such isolation. Hughes (1991) and Hughes et al. (1992) detailed the uses and limitations of this approach; it suffices to say that nearly all MSV isolates, even those causing severe symptoms in maize, could be found to differ in RFLP profiles generated by at least one restriction enzyme. While it is possible to roughly quantitate differences between virus genomes in terms of numbers and positions of DNA fragments produced, this is not a very reliable method for assessing genomic similarities/differences; it is far more useful to determine genomic maps as described below.

We have found the RFLP approach very useful in investigating virus diversity in field infections: we have used the technique to (i) investigate the diversity of MSVs in individual plants early and late in an infected field (E Edge and EP Rybicki, unpublished), and (ii) to investigate the clonal diversity of viral genomes within single plants infected with a severe maize isolate (MSV-RSE) and a grass isolate (MSV-Setaria) (Hughes, 1991). In the first application, it was possible to show that the viral diversity early in a disease incidence (with young plants) was greater than that found later in mature plants in the same field: it is well known that older plants are more tolerant of MSV infection; these results were taken as indicating that younger plants can be symptomatically infected with a wider range of viruses - with a wide range of virulence - than older plants. In the second, using as many full genomic clones of viral DNA in E coli as possible, it was shown that individual infected plants contained a predominant RFLP type (to >80% of samples), but that several other minor genotypes were also present. Taken together, these results indicate that naturally-infected plants appear to contain a predominant genotype of virus together with other minor types, and that severe maize disease in older plants is associated with a narrow range of predominant genotypes of virus. However, the main finding is that MSV "isolates" are almost certainly a "quasispecies", or population of virus genotypes, even if isolated from a single plant: this conclusion is reinforced by later findings by Isnard et al. (1997, this volume, and see below).

Restriction endonuclease mapping of RF or cloned DNA

It is possible, with very little extra work compared to what is required to generate RFLPs of virus isolates, to "map" restriction endoclease cleavage sites in a virus genome, either isolated as an RF dsDNA, or as a clone in an E. coli vector plasmid. Such maps may be compared to one another to give quantitative estimates of sequence divergence, and even to construct relationship dendrograms, as has been done for MSV isolates in our laboratory (Clarke et al. 1989; Hughes et al. 1992). An important advantage of the use of restriction maps for comparison of viruses is that the data may be compared between laboratories without any exchange of materials, such as virus preparations or antisera: restriction sites are positions on a number line, and as long as different labs use the same panel of enzymes, maps will be immediately comparable. Our work has shown that mapping can be quite an accurate exercise: maps we have determined from the use of enzymes compare very well to theoretical maps derived from sequence data for the same virus isolate; to within 2% of position and with the order unchanged for sugarcane streak virus (SSV) strain Natal (Hughes et al. 1993; Hughes et al. 1992; Hughes et al. 1991). We have used mapping to demonstrate that maize isolates of MSV are all more closely related to one another than they are grass or barley or wheat isolates, and that nearly all isolates can be differentiated from one another on the basis of at least one restriction site difference (Hughes et al. 1992). The technique is not too useful for isolates differing in sequence by more than 20% (Hughes et al. 1992); however, as all maize isolates of MSV found to date differ by less than this, it remains useful for differentiation of closely-related viruses.

We have constructed a database of virus maps - derived both from sequence and from mapping exercises - as a demonstration of the utility of the use of map data for virus isolate comparisons. The database uses the computer program Resolve (EH Harley, Dept Chemical Pathology, Univ Cape Town), which is a multifunctional mapping, map storage and map manipulation package. Maps are stored as a series of positions on a number line; they can be represented graphically, and compared with one another in a number of ways for determination of relationships, including both "phenetic" or distance measures, and "cladistic" or homology measures. Figures 2a and 2b illustrate different aspects of the package.

As an object example of the ease of exchange of data, we obtained a map of a cloned virus isolate (MSV-Zim; from Rob Briddon, John Innes Centre, Norwich, UK), and compared it with local and sequenced viruses: results are shown as a neighbour-joining relationship dendrogram in Figure 3. It was easy to determine that the MSV-Zim isolate was closely related to other maize-type viruses, distinct from other grass- and cereal-infecting viruses, and that the maize-infecting MSV types represent a tightly-clustered group.

Polymerase chain reaction DNA amplification for detection and characterisation of virus isolates

The use of a degenerate oligonucleotide primer pair for the amplification of C2 ORF DNA - the most conserved part of the Mastrevirus genome - from potentially the entire range of grass- and cereal-infecting Mastreviruses, was described by Rybicki and Hughes (1990). A pair of 17-base primers - one with four and one with five 4-base degeneracies - could be used to amplify an approximately 250 base pair piece of DNA from almost the entire range of southern African streak viruses; later work proved that even DNA from the very distantly related chloris striate mosaic virus (CSMV) from Australasia could be amplified. The technique was exquisitely sensitive for MSV detection: it could be used successfully to detect viral DNA in sap extract that had been diluted by a factor of 10⁹; a similar technique has recently been shown to be usable for virus detection for samples with a single infected leaf in a pool of 1000 leaves (Jonker et al., 1997, this volume).

Our earlier published work described the use of the technique in conjunction with differential hybridisation and direct PCR product sequencing to show that maize MSVs were closely related, and to define SSV-Mauritius as a distinct virus (Rybicki and Hughes, 1990). Our latest work on sequencing products obtained with the same primers has revealed a closely-related group of viruses infecting grasses and barley and wheat, which is distinct from other grass viruses and from a group of viruses which infect maize (see Figure 4). Especially noteworthy in this Figure is the demonstration that viruses collected from the same location in different seasons in maize and in wheat (MSV-VM and MSV-VW respectively) are obviously very dissimilar. Although this is apparently an upset to currently-held views on MSV epidemiology - as conventional wisdom would have it that the maize MSV overwinters in the wheat - it may actually be evidence of different crops selecting different "most fit" viruses out of a common background population of viruses. Thus, the predominant virus infecting different crops in the same location is selected by the crop rather than the location.

Other workers have also used PCR amplification and sequencing for characterisation of MSVs: Briddon et al. (1994) amplified and sequenced 12 different viral coat protein (CP) genes and short intergenic regions (SIRs) in an effort to determine genetic variability of MSV isolates and strains. They found that the greatest diversity between any isolates was linked to geographic isolation, in that Mauritius and La Reunion Island maize isolates were the most different of any maize types sampled; however, all of the maize-type viruses were very closely related.

Our laboratory presently has a new set of oligonucleotide primers which potentially amplify a DNA fragment from the entire range of African streak viruses: the fragment includes most of the movement protein (MP) gene, the whole long intergenic region (LIR) and most of the RepA ORF (J Willment et al., unpublished). The LIR and MP gene represent the most variable region of sequence in these viruses; thus, partial sequencing across this region or RFLP analysis would be a good measure of virus diversity. An illustration of the latter approach is shown in Figure 5: this demonstrates that it is possible, using a single restriction enzyme, to differentiate distinctly different groups of viruses.

Complete genome sequencing and analysis

The "gold standard" of virus comparisons has to be comparisons of their entire genomic sequences: this is the information that encodes the entire range of interactions of the virus with the host; that specifies host range, symptom severity and vector interaction. With Mastreviruses this is attainable to a degree possible for few other virus genera: the entire genome of all members of this genus characterised so far does not exceed 2800 bases; the only smaller eukaryote-infecting viruses known are the circoviruses and plant circo-like viruses. Consequently, a relatively large number of Mastrevirus genome sequences are currently known, including CSMV, wheat dwarf virus (WDV), miscanthus streak virus (MiSV), digitaria streak virus (DSV), sugarcane streak virus Natal (SSV-N; Hughes et al. 1993), panicum streak virus Kenya (see (Rybicki, 1994), and the unpublished panicum streak virus Karino from South Africa (W Schnippenkoetter et al., in preparation). Four complete severe maize-type MSV sequences have been published: these are MSV-Kenya (Howell, 1984); MSV-Nigeria severe (Mullineaux et al. 1984); MSV-South Africa (Potchefstroom, (Lazarowitz, 1988); and MSV-Reunion severe (Peterschmitt et al. 1996); the last three are all of infectious virus clones. My group has put the complete infectious sequences of a moderately severe maize type (MSV-Komatipoort) and a Setaria sp. infecting strain (MSV-Set) onto the GenBank sequence database (W Schnippenkoetter et al., in preparation); in addition, we have complete infectious genomic sequences available for another extremely severe maize type (MSV-Matabeleland; D Martin et al., in preparation), and a wheat-infecting isolate (MSV-Tas; J Willment et al., in preparation). Ngwira et al. (1997, this volume) have also completely sequenced a mild (MSV-KL) and a severe (MSV-Ks1) MSV, and demonstrated - as expected - that differences were concentrated in the long intergenic regions (LIRs). Isnard et al. (1997, this volume) have completely sequenced seven MSV genomes derived from a mixed population ("quasi-species") contained in their very severe "screening isolate". Additionally, our group (J Willment et al., unpublished) and Ngwira et al. (1997, this volume) have a number of partial sequences of MSV strains and isolates, encompassing the LIR (and CP genes for Ngwira et al.). Thus, there exists a sequence library of 17 complete MSV sequences, mostly of severe maize types, as well as at least 30 partially sequenced genomes, counting the PCR-generated sequences mentioned earlier.

A relationship dendrogram drawn from alignments of (nearly) all sequences available to us is presented in Figure 6. It is immediately apparent that, except for the case of MSV-Mauritius and MSV-Reunion, geographic location of a viral isolate has little bearing on its relationship to other MSVs: thus MSV-E from Egypt appears most closely related to MSV-U from Uganda, -Nb from Nigeria, and -Gb from Gambia; MSV-Mat from Zimbabwe is equidistantly related to these viruses, and cluster from Zambia, Zimbabwe and Mozambique, and so on. The only clear division between isolates appears to be based on host plant: thus, the severe maize isolates all cluster together, while viruses infecting grasses cluster separately from this group, and from one another in most instances. The most distantly related viruses causing severe disease in maize are separated by only 4% or so of total sequence (eg: MSV-Reunion and MSV-Kom); it is noteworthy that viruses infecting grasses and other cereals are more diverse. It is heartening to note how sequence comparison backs up biological observations, such as those of Pinner et al. (1988) and Mesfin et al. (1992), and even those of earlier pioneers (eg: McClean, 1947): there are clearly distinct groups of viruses infecting different hosts, such as Panicum maximum (panicum streak viruses, PanSV), sugarcane (sugarcane streak viruses, SSV), and the maize and grass MSVs.

Discussion

The relative merits and different areas of application of the different techniques available for the differentiation / characterisation of maize streak virus isolates and strains are quite easy to determine. Although host range determination and biological characterisation may be though to be archaic in this day and age, we and others have found the use of a "differential panel" of maize and other grasses to be most useful for the fine differentiation of virus isolates, and especially of viral infections resulting from agroinoculation using cloned virus genomes. Isnard et al. (1997, this volume) used agroinfection to determine that none of their cloned isolates gave the same symptom expression / severity as their biological "isolate" N2A, but that it was possible to complement severity by mixing agroinfectious clones; we have used both agroinoculation and subsequent leafhopper transmission of cloned isolates to prove that our cloned genomes are representative of the biological isolates from which they came. Njuguna et al. (1997, this volume) used transmission to different hosts as a means of determining that a virus isolate maintained in Coix lachryma-jobi had in fact become less virulent. Thus, determination of biological parameters such as symptom severity, leafhopper transmissibility and host range still have much relevance in the study of MSD - especially when coupled with modern techniques such as the image analysis system described by Martin et al. (1997, this volume), and with the use of agroinfectious clones of well-characterised viruses.

Serology is probably the least useful technique for differentiating MSV isolates: apart from the fact that most maize isolates are effectively serologically identical, there is the problem that no two antisera or MoAbs will react similarly, and that reagents have to be shared (as well as antigens) if any two laboratories wish to obtain comparable results. Thus, its use should be discounted for any except the more esoteric investigations, such as probing structural changes in viral capsids, or whether defined mutations in coat proteins affect the viral surface.

Of the different techniques related to viral nucleic acid manipulation/analysis, probably the most widely useful would be RFLP analysis: it is relatively easy, relatively cheap, can be performed on a large number of samples without too many problems, and is capable of distinguishing at a fine level between virus isolates. We have shown in our laboratory that it is possible to differentiate between viruses using restriction digests of unfractionated whole infected plant DNA isolates, by hybridisation and even by simple ethidium bromide staining (W Schnippenkoetter et al., unpublished). Coupled with PCR (see Figure 5), it becomes a slightly more expensive but still simple technique for the quick characterisation of virus isolates, or of the predominant genotype in an isolate.

While we have had good results with differential hybridisation and restriction mapping as tools for revealing diversity (eg: see Hughes et al. 1992), we feel that these techniques will never be as generally applicable as RFLP analysis; partly for reasons of cost, but also because of the extra work involved in both, and the fact that MSV isolates from maize especially are all so closely related as to be indistinguishable by the former technique. We will continue to maintain and to use our restriction map library, however, and would encourage others to map their viruses similarly so as to be able to compare them with ours, without the necessity for exchanging viruses or their DNA. A vindication of the mapping approach in terms of a survey of virus diversity was published recently for the very distantly related beet curly top virus (BCTV), a member of the Curtovirus genus of geminiviruses: Stenger and McMahon (1997) have described a restriction mapping approach to studying BCTV diversity in the western US, in which they examined 66 biological isolates by cloning and mapping. They determined that 43% of their isolates contained more than one virus genotype, that 11% contained mixtures of strains, and they used similar approaches to ours to characterise genotypes into strain groupings.

The use of techniques such as PCR amplification and total and partial genome sequencing has lately become prevalent throughout the world of plant virology - and the nature and size of the MSV genome has meant it is particularly amenable to characterisation in these ways. The large and growing sequence library of MSV strains and isolates - not to mention other Mastreviruses - represents an invaluable resource, both for illumination of the extent of virus diversity, but also as a "germplasm bank" for the study of virus-host interactions. Sequence comparison of all the sequenced viruses - similar to what was attempted by Ngwira et al. (1997, this volume) - could allow easy predictions of regions which must be involved in different activities, such as binding to host transcription factors, interaction with coat/movement proteins, etc. As far as biological relevance goes, for example, we have a panel of four different cloned and sequenced MSVs with known biological properties, which display different ranges of "penetration" (in terms of symptom causation) of virus-tolerant/resistant maize lines. By swapping segments of DNA between these viruses, we can determine which viral sequences are responsible for the differing phenotypes, and possibly eventually even which host genes are responsible for limiting viral spread and/or replication (van der Walt et al., 1997, this volume, and D Martin et al., unpublished). We are also engaged in attempting to genetically engineer pathogen-derived resistance to MSV infection into maize (Mangwende et al., 1997, this volume): it is of great interest to us to determine which genotypes of MSV are protected against / can break this resistance, and possession of as wide a diversity of sequenced and infectious viruses will help in this regard. As suggested in the MSD Symposium address on this subject, the use of a panel of sequenced agroinfectious viruses with different virulence or aggressivity could potentially also aid in classical breeding for MSV resistance (eg: Dintinger et al., 1997; Pixley et al., 1997, this volume).

It would do this by providing a known range of viruses from all over Africa, as a challenge to any particular breeding line: this might obviate the some of the need to test lines developed in different centres in different locations all over Africa, in an effort to determine whether viruses in the different centres are capable of breaking resistance developed in other centres.

However, there is one obstacle to this goal: the fact that many of the sequences have not yet been published, and are consequently not in one of the various interlinked genome sequence databases. This is a handicap, and one which could be overcome by establishing a dedicated Mastrevirus/MSV sequence database, similar to the Potyviridae database (Berger and Rybicki, 1996). The idea of this would be to have an easily accessible, centrally-located resource, continuously updated with sequences and analyses of genome diversity, as a service to the community of researchers working on these viruses. A start to such a database will be made available on the World Wide Web at http://www.mcb.uct.ac.za/MSV/mastrevirus.htm.

Part of the preamble to the talk on this subject given at the MSD Symposium involved the small farmers’ vulnerability to MSD, compared to the larger commercial farmer. If studying the diversity of MSV can help in any way to alleviate this vulnerability - for example, by allowing better selection of tolerant maize genotypes, or allowing development of a useful genetically-engineered product - then we will have elevated this work from the level of legitimate academic curiosity, to being a genuinely useful application. We firmly believe that we are well along the way to this goal.

Acknowledgements

We wish to acknowledge all those who have provided samples over the years to our group; EPR would especially like to thank all those generations of students who have done the work, and to thank Dr GDJ van Rensburg for the leafhoppers all those years ago, and Koos "JB" van Rensburg, Brad Flett and Mossie Mostert for an excellent meeting.

Copyright Ed Rybicki, November 1997
unless otherwise specified.