Quantcast
Last updated on February 12, 2012 at 7:34 EST

Defective Interfering DNAs of Plant Viruses

March 15, 2006

By Patil, Basavaprabhu L; Dasgupta, Indranil

Several plant DNA viruses produce significant quantities of deleted versions of their DNA in the infected plants, which is generally correlated with a slowing down of replication process of the viral DNA. These deleted versions of the viral DNA are called Defective-Interfering (DI) DNA because of their inhibitory effect on the helper virus. The sizes of the DI-DNAs for different plant viruses can vary from one tenth of the size of the viral genome to one half and are mostly encapsidated. Sequence analysis suggests that DI-DNAs are formed by deletion, duplication, inversion, rearrangement and sometimes by insertion of non-viral DNA sequences involving the viral genome and its satellites. The role of the host plant in the formation of DI-DNAs is also important as DIDNAs are readily formed in experimental hosts rather than their natural hosts. Symptom modulation by DI-DNAs is believed to be by competition for essential viral and host factors, thereby reducing the levels of the helper virus. Evidence is also emerging of its role in activating PTGS in the plant against viral transcripts, which is also likely to contribute towards symptom amelioration. The possibility of DI-DNAs playing a role towards the integration of pieces of viral DNA into plant genomes also exists. Most importantly, DI-DNAs have the potential to act as a tool in developing novel control strategies against viruses of crop plants and acting as gene expression/ silencing vectors.

Keywords DI-DNA, interference, plant virus, PTGS, recombination, replication, symptoms

Table of Contents

I. INTRODUCTION

Other than the full length viral genomes, plants infected with DNA viruses often contain one or more types of smaller DNAs (subviral molecules) derived from them. Often, these smaller versions of the viral DNAs accumulate to significant levels in the infected tissues and in most cases interfere with viral multiplication resulting in symptom amelioration and hence, interference. In many cases, their interfering natures are not well- established and are consequently called defective DNA or subgenomic DNA. All of the above lack the full complement of genes necessary for a complete infectious cycle and thus need the original/helper virus to provide the missing functions, like replication and encapsidation.

II. PLANT VIRUSES HAVING A DNA GENOME

Plant viruses with a DNA genome fall into two general types, those with circular double-stranded DNA (dsDNA), which replicate by reverse transcription through an RNA intermediate (caulimoviruses and badnaviruses), and those with circular single-stranded DNA (ssDNA), which replicate through a dsDNA intermediate (the geminiviruses, nanoviruses and associated DNA satellites) by rolling circle replication in the nuclei of infected cells (Hanley-Bowdoin et al., 1999) and also by recombination-dependent replication (Jeske et al., 2001). Defective DNA molecules have been reported for both the above types, which fall into three families, namely Geminiviridae, Nanoviridae and Caulimoviridae. In addition, geminivirus-associated satellite DNAs also give rise to defective DNAs (Briddon et al., 2003; Usharani et al., 2004).

Geminiviruses are a large and diverse group of plant viruses characterized by bisegmented (geminate 30 20 nm) particles, by their circular ssDNA genome that replicates in the host cell nucleus and by being transmitted in a persistent manner by insect vectors. The family Geminiviridae is divided into four groups based upon genome organization, host range and in sect transmission. Begomoviruses are transmitted by the whitefly Bemisia tabaci, have bipartite genomes (A and B components, approximately 2.7 to 2.8 kb each), except in a few cases, and infect dicotyledonous plants. The A component encodes for the coat protein (CP or AVl) and a cell-to- cell movement protein (AV2), not present in “New World” begomoviruses (Padidam et al., 1996), on the viral sense strand, and a set of four proteins, replication associated protein (Rep or ACl), a transcriptional activator protein (TrAP or AC2), replication enhancer protein (REn or AC3) and C4 (AC4), the least conserved protein, on the complimentary-sense strand. AC4 seems to be functional only in begomoviruses that lack DNA-B component as a movement-associated protein and has also been recently implicated in countering the host response to infection (van Wezel et al., 2002). The B component encodes two proteins, one in virion sense and other in complementary sense. Both gene products are involved in virus movement within the host plant and act cooperatively. The BVl protein (Nuclear Shuttle Protein; NSP) controls the transport of viral DNA between the nucleus and cytoplasm, and the BCl protein (MP) mediates its cell-to-cell movement (Lazarowitz and Beachy, 1999). Mastreviruses (Maize streak virus, MSV, as type member) are transmitted by a variety of leafhopper (Cicadellidae) species, have a monopartite genome and, generally, infect monocotyledonous species. Four proteins are encoded by the Mastrevirus genome: the movement protein (MP) and the capsid protein (CP), on the viral- sense (v-sense) strand, and the RepA protein, exclusive of this genus, and the Rep protein, on the c-sense strand. Curtoviruses (Beet curly top virus, BCTV, as type member) are also transmitted by leafhoppers, have a monopartite genome, although with a genetic organization different from that of mastreviruses, and infect dicotyledonous plants. They apparently occupy an intermediate phylogenetic position. In addition to MP and CP, their genome also encodes a V2 protein on the v-sense strand, whereas on the c-sense strand, four genes exist, namely Rep, C2, REn, and C4. Topocuvirus (Tomato pseudo-curly top virus, TPCTV, as type member) is a recently- designated genus that has been split from Curtovirus and has similar genome organization, but is transmitted by treehopper.

All geminiviruses have a ~200 nucleotide non-coding intergenic region, which has a conserved nonanucleotide (TAATATT[arrow down]AC) present in the loop of a stem-loop structure. Members of the family Mastrevirus contain two intergenic regions, one large (LIR) and another small (SIR), located at opposite sides of the viral genome, the invariant nonanucleotide sequence being located in the LIR. Two features are unique to Mastrevirus: (i) the presence of a ~ 80 nt- long DNA fragment annealed to a region within the SIR and (ii) the occurrence of a splicing event of the complementary sense (c-sense) transcript. The common region of geminiviruses also contains sequence elements 6 to 12 nt in length that are repeated 2 to 5 times, sometimes imperfectly and sometimes in inverted orientation. These “itrons” are binding sites for Rep and have a role in the initiation of DNA replication (Arguello-Astorga et al., 1994; Fontes et al., 1994). The intergenic region contains several regulatory elements, including two TATA motifs (Harrison and Robinson, 1999).

Two other DNA molecules have been reported to be associated with several begomoviruses, namely DNA β and DNA 1. DNA β molecules are a diverse set of symptom modulating, single-stranded DNA satellites, associated with many begomoviruses. They are present in a diverse range of crops, mainly in the Solanaceae and Malvaceae, as well as many weed and ornamental species (Briddon et al., 2003; Zhou et al., 2003; Bull et al., 2004). These components are typically 1350 nucleotides in length, half that of their helper begomoviruses. They encode only a single open reading frame designated Cl gene, which is conserved among these molecules in position and sequence and is believed to be a determinant of pathogenicity (Saeed et al., 2005). In addition, DNA β molecules contain an A-rich region that shows high level of sequence conservation called the satellite conserved region, SCR (Briddon et al., 2003). DNA molecules are believed to be transreplicated by the cognate helper virus, but the mechanism may be different from the strict requirements of compatibility of the Rep protein with the iteron sequences seen in the case of DNA B components of bipartite begomoviruses. DNA β molecules do not possess the iteron sequences of the helper begomoviruses, suggesting a less specific interaction between begomoviruses and DNA β (Lin et al., 2003).

DNA 1, reported from cotton plants infected with cotton leaf- curl disease, is a satellite like, single-stranded DNA molecule approximately 1350 nucleotides in length. It encodes a rolling- circle replication initiator protein (Rep) which shows high levels of sequence similarity to the Reps encoded by components of nanoviruses (Mansoor et al., 1999; Saunders and Stanley, 1999). It is believed that the DNA 1 may have been derived from a nanovirus and acquired by begomoviruses by a process of “component capture.” Apart from the Rep coding sequence, DNA 1 molecules contain two further conserved features; a predicted hairpin structure with the nonanucleotide sequence loop and an A-rich region. DNA 1 and DNA β components, being half the size of the viral component, are encapsidated in monomeric, rather than geminate particles.

The nanoviruses are a family of multi-component, single-stranded DNA viruses that are transmitted by aphids (Gronenborn, 2004). Nanoviruses, such as the economically important virus Banana bunchy top virus (BBTV) consist of multiple circular ssDNA molecules, t\heir number varying from 6 to 11, of approximately 1 kb in size, encapsidated in icosahedral virions. They exhibit rolling circle replication, similar to geminiviruses. All DNA components appear to be structurally similar in being (+) sense, transcribed in one direction and containing a conserved stem-loop structure.

Caulimoviruses contain double-stranded circular DNA as their genome of approximately 8 kbp in size and have isometric virus particles, which are transmitted by aphids. It has a single gap in one strand (α- or minus) and two in the complementary strand (plus) and replicate their DNA by reverse transcription of a full- length transcript. The DNA encodes six to eight genes, specifying the coat-protein, movement protein, reverse transcriptase- ribonuclease H (RT-RNase H), aphid-acquisition factor, transcriptional transactivator, inclusion-body proteins, etc. Two 3′ co-terminal transcripts, the 35S and the 19S are formed due to the activity of the two corresponding promoters. The longer transcript (35S) acts as a template for the synthesis of cDNA using a cytoplasmic met-initiator tRNA as the primer, which binds to a homologous sequence, approximately 200 bp downstream to its 5′-end. The enzyme synthesizing the cDNA, RT-RNase H, simultaneously degrades the RNA template and jumps over to the complementary 3′- end of the transcript, which is terminally redundant. Completion of the cDNA synthesis results in the formation of the negative-sense strand of the viral DNA, which in turn, synthesizes the positive- sense strand (Guilley et al., 1983; Hull and Covey, 1983; Pfeiffer and Hohn, 1983).

Defective DNA/RNAs associated with animal viruses (von Magnus, 1954; Huang and Baltimore, 1977; Mankertz et al., 1997; Mankertz et al., 1998; Russo et al., 1994; Kitajima and Muller, et al., 1972) and defective RNAs of plant viruses (Roux et al., 1991; Graves et al., 1996; White, 1996; White and Morris, 1999; Simon et al., 2004) have been widely reported. Defective DNAs have been reported for the geminiviruses Tomato golden mosaic virus TGMV (Hamilton etal., 1982; MacDowell et al., 1986), African cassava mosaic virus, ACMV (Stanley and Townsend, 1985), Wheat dwarf virus, WDV (MacDonald et al., 1988), Potato yellow mosaic virus, PYMV; Roberts et al., 1988), Tomato yellow leaf curl virus, TYLCV (Czosnek et al., 1989), Beet curly top virus, BCTV (Frischmuth and Stanley, 1992; Stenger et al., 1992), Ageratum yellow vein virus, AYVV (Stanley et al., 1997), various begomoviruses associated with cotton leaf curl disease, CLCuD (Liu et al., 1998; Nadeem et al., 1997), Sweet potato leaf curl virus, SPLCV (Lotrakul and Valverde, 1999), Tobacco leaf curl Zimbabwe virus, TbLCZWV (Paximadis and Rey, 2001), Tobacco leaf curl virus, TLCV; Zhou et al., 2001) and Maize streak virus, MSV (Casado et al., 2004). Caulimoviruses and nanoviruses (CaMV and BBTV respectively) also have been reported to be associated with defective DNAs (Table 1).

In this review, certain aspects of DI-DNAs have been discussed, with special emphasis on their structure, origin, encapsidation, transmission, interaction with the host plant, symptom modulation and evolutionary significance.

TABLE 1

Defective DNAs found in plant DNA viruses

III. STRUCTURE AND SEQUENCE COMPOSITION

A. Geminiviral Defective DNAs

Stanley and Townsend (1985) first characterized the subgenomic DNA forms associated with ACMV, which was known as “cassava latent virus” during that time. Sequence analysis of these forms revealed them to be derived from DNA B component by deletions within one half of the molecule. These deletions were found to disrupt both major open reading frames (ORFs) of DNA B, completely excising the NSP and removing the C terminal half of the MP (Figure 1). Interestingly, small repeat sequences were found at the deletion points of all the subgenomic forms studied, suggesting their production as a result of recombination events. MacDowell et al. (1986) analyzed similar DNA molecules associated with TGMV and reported them to be 1.2 kb in size, derived from DNA B, again by deletion of NSP and the C- terminal portion of MP, similar to those found for ACMV. Further analysis of BCTV subgenomic DNAs revealed that a majority of them were derived from the DNA B component, obtained by complete deletion of NSP and the C-terminus of MP (Frischmuth and Stanley, 1994). Subgenomic DNA B of Potato yellow mosaic virus, PYMV (Roberts et al., 1988) and Cowpea golden mosaic virus, CPGMV (Winter et al., 1998) were also shown to be approximately half the size of the viral genomic components. Since all the above molecules were apparently being transreplicated by the Rep of the helper virus, it can be assumed that all of them contained the iteron sequence recognized by the replication complex with variable degrees of deletions, mostly involving NSP in the DNA B component.

FIG. 1. Genome maps of ACMV and BCTV showing the position and orientation of virus-sense (V) and complementary-sense (C) genes. The extent of deletions within naturally occurring defective DNAs (Stanley and Townsend, 1985; Frischmuth and Stanley, 1992) are indicated by arcs. CR, common region; IR, intergenic region; LIR, large intergenic region; SIR, small intergenic region. (Reprinted from Seminars in Virology, Volume 4. Frischmuth, T., and Stanley, J., Strategies for control of geminiviral diseases, pp. 329-337, 1993, with permission from Elsevier).

Formation of subgenomic DNA from only DNA B may reflect differences in the primary, secondary or tertiary structures of DNA A and DNA B. Significant differences in the structure of the supercoiled forms of dsDNA A and B were revealed by the much greater sensitivity of DNA A to S1 nuclease digestion (Sunter et al., 1985). This might partly explain enhanced generation of defective DNA specific to B than to A.

Stanley et al. (1997) reported defective DNA from the monopartite begomovirus, AYVV isolated from the naturallyinfected weed Ageratum conyzoides. As reported in bipartite begomoviruses, all the AYVV subgenomic DNAs were also approximately half the size of the viral component, and contained the intergenic region and variable amounts of 5′-ends of Rep, C4 and V 2 ORFs. In addition, some of the defective/subgenomic DNA also contained sequences that were not from AYVV. The chimeric nature of this defective DNA distinguished it from all previously characterized components. In this respect, the AYVV defective DNA resembled Turnip crinkle virus (TCV) RNA C, which contains sequences of the helper virus in addition to unrelated sequences (Simon and Howell, 1986). It was initially suggested that the non-viral sequences were of host plant origin, although it was not possible to confirm this by genomic blot hybridisation and database searches. Defective DNAs were also produced in N. benthamiana following inoculation of the cloned AYVV DNA component, but were found to be significantly smaller than those produced in A. conyzoides (Stanley et al., 1997). A database search revealed that the non-viral sequences of one such defective component (def19) showed significant homology with sequences of nanovirus components that encode replication-associated proteins (Saunders and Stanley, 1999). Bigarre et al. (2001) reported defective molecules associated with Holleyhock leaf crumple virus (HLCrV), a monopartite begomovirus, which were approximately 1400 nucleotides in size. Sequence analysis revealed that they contained the intergenic region of HLCrV and parts of the Rep and AV1 ORFs that had resulted from deletions and recombination events. Non-viral DNA sequences of 300 to 500 bp were found inserted in these defective forms, which were of unknown origin. Such non-viral sequence (172nt) has also been found in defective DNA of TbLCZWV (Paximadis and Rey, 2001). Similarly association of defective DNA A molecules is reported in TLCV (Zhou et al., 2001), SPLCV (Lotrakul and Valverde, 1999), CLCuD- associated begomoviruses(Liu et al., 1998; Nadeem et al., 1997), TbLCZWV (Paximadis and Rey, 2001) and TYLCV (Rezaian, 2004).

For CLCuD-associated begomoviruses, eleven sites in the five small DNA molecules sequenced represented deletions, presumably produced by recombination. Analysis of the coding potential of the above molecules showed that notably, in every case, the 5′ part of the Rep coding sequence was present and one of them contained a complete AC4 gene. Various other ORFs were also present, some of which were entirely novel and others included portions of in-frame coding sequences of genes of the helper virus (Liu et al., 1998).

Stenger et al. (1992) reported the presence of seven classes of subgenomic DNAs in N. benthamiana derived from BCTV, a curtovirus, ranging in size from 800 to 1800 nucleotides. Sequence analysis of one of them showed two deletions, the deletion boundary showing a direct repeat of 2 to 6 nucleotides. Frischmuth and Stanley (1992) cloned defective DNAs from N. benthamiana plants infected with BCTV and two major population of subgenomic DNAs were isolated, whose size ranged from 1180 to 1700 nucleotides in group I and 680 to 880 nucleotides in group II (Figure 1). The deletions were distributed largely throughout the genome apart from the region between nucleotides 2946 to 410, which was present in all clones, containing cis-acting elements required for their replication.

Subgenomic DNA found associated with Maize streak virus (MSV) varied in size from 200 to 1600 nucleotides and was classified into three major classes based upon their size and sequence differences (Casado et al., 2004). All the cloned DNAs invariably contained the replication signals. The plus and minus strand replication origins, located in the large and small intergenie regions (LIR and SIR, respectively), were both present in all class 1 clones (>1.34 kb), but only one of the origins were present in class 2 (1.00 to 1.34 kb) and class 3 clones (<1.00 kb). Mostly, suchcloned DNAs retained the conserved stem-loop structure, required for replication of MSV DNA and the majority contained the loop structure (and therefore the nick site required for initiation of rolling circle replication) as well.

In summary, defective DNAs derived from the DNA A component contain the intergenic region with all the cis-acting sequences required for replication, and 5′ portions of the Rep and sometimes a small part of gene encoding CR Those derived from the DNA B component usually have part of MP but NSP is completely absent.

B. Caulimoviral Defective DNAs

The defective DNAs for caulimoviruses have been only reported in association with Cauliflower mosaic virus, CaMV (Olszewski and Guilfoyle, 1983). The size range of CaMV defective DNA was much more as compared to those reported from geminiviruses. These DNA species were present as covalently-closed circles, relaxed circles and linear molecules. It was presumed that they are formed by deletions during each infection cycle. No specific region of the CaMV genome appeared to be preferentially deleted and the different subgenomic species were presumed to result from random deletions. The involvement of intermolecular recombination or aberrant viral replication also could not be ruled out. Three distinct subgenomic species appeared to accumulate preferentially in nuclei isolated from infected tissue. Two of these species were circular and the other linear. Some also appeared to be in the form of minichromosomes. A 630-base pair linear subgenomic dsDNA species was found in the nucleus, which was the only species encapsidated in virus particles (Guilfoyle et al., 1983).

C. Defective DNA from Nanoviruses

The only report of a defective DNA in nanoviruses was by Su et al. (2003) for Banana bunchy top virus (BBTV). This DNA had a size of 537 nucleotides, almost half the size of the original genome and was produced by two deletions.

D. Defective Satellite DNA

Full-length satellite DNA β molecules have been found associated only with Old World monopartite begomoviruses. Usharani et al. (2004) reported association of defective DNA β molecules, with a bipartite genome of Old World begomovirus Potato Leaf Curl Virus. Sequence analysis revealed that it lacked the A- rich region and the Cl open reading frame, while maintaining the satellite conserved region (SCR) of ca. 115 bp. Dry et al. (1997) showed the presence of a 682 bp satellite DNA (which was rather a defective DNA β molecule lacking the Cl ORF) associated with TLCV. Defective molecules are also reported for DNA β associated with Tomato leaf curl virus (692 nt), AYVV (892 nt), CLCuD-associated viruses (902 to 1248 nt), Potato leaf curl virus (705 nt), Zinnia leaf curl virus (1349 nt), Honeysuckle yellow vein mosaic virus (426 nt), Bhendi yellow vein virus (759 nt) and Hollyhock leaf crumple virus (660 nt) (Table 2). In addition, recombinant molecules between DNA A and DNA β and between DNA A and DNA 1 and/or 2 have also been reported (Saunders et al., 2000; Saunders et al., 2002; and Briddon et al., 2001), which have been generally referred to as chimeric DI-DNA (Table 2).

IV. ENCAPSIDATION AND INSECT TRANSMISSION

A. Geminiviruses

Frischmuth et al. (2001) reported ACMV particles of different sizes as analyzed on a sucrose gradient. The top fraction was associated with DI-DNA, the middle fraction with both genomic and DI- DNA, and the bottom fraction with multimeric, genomic and DI-DNA. Electron microscopy studies of these fractions indicated that DI- DNA was encapsidated in isometric particles of 18 to 20 nm whereas the genomic DNA was encapsidated in geminate particles (Frischmuth et al., 2001). The association of defective DNA with virus particles has also been reported for TYLCV (Rezaian, 2004; Czosnek et al., 1989), ACMV (Sequeira, 1982), PYMV (Roberts et al., 1988) and MSV (Casado et al., 2004). Immunotrapping experiments also showed the association of TLCV defective DNA molecules with virus particles (Zhou et al., 2001). The above results suggest that the geminivirus coat proteins can encapsidate circular ss-DNA molecules of 1.4 to 1.5 kb to form isometric particles. Defective DNAs, associated with geminiviruses, maintaining a size approx. half that of a complete genomic component (or genome) may suggest a stringent size selection for encapsidation.

It has also been suggested that the size of DNA rather than its sequence is important for encapsidation (Mansoor et al., 1999). As defective molecules are encapsidated, this implies that they must retain signals, if any, necessary for this process. It is also important to note that preparations of isometric particles associated with geminiviruses are enriched in defective DNA (Bottcher et al., 2004). However, the possibility that two subgenomic-sized DNAs can be packaged in one twinned particle cannot be excluded (Liu et al., 1998).

The encapsidated defective DNAs (Stanley and Townsend, 1985; Bottcher et al., 2004) are also insect-transmitted (Czosnek etal, 1989, Zhou et al., 2001, Frischmuth et al., 2001). The DI-DNA population associated with AYVV infection was maintained during whitefly transmission between A. conyzoides plants, implying that they spread systematically throughout plants and suggesting that they are encapsidated in order to be transmitted in this manner. Defective DNAs associated with ACMV and TGMV infection (Stanley and Townsend, 1985; MacDowell et al., 1986) were also of similar size as those of AYVV and were also maintained during plant-to-plant transmission (Stanley et al., 1990). The defective DNAs of CLCuV were also transmitted between tobacco and tomato plants both by grafting and by the whitefly vector (Liu et al., 1998). The identity of encapsidation signals, if any, has not yet been established, but Qin et al. (1998) suggested that the stem-loop region, conserved in all geminivirus genomes, might have a similar function.

TABLE 2

List of defective β DNAs and chimeric defective DNAs associated with different geminiviruses and their details.

B. Caulimoviruses

In contrast to geminiviruses, defective DNAs derived from caulimoviruses are generally not encapsidated, suggesting that these defective DNAs might lack the signals essential for this process (Covey and Turner, 1993). The 630bp subgenomic, reportedly encapsidated linear DNA species (Olszewski and Guilfoyle, 1983; Guilfoyle et al., 1983) needs further analysis to throw more light on the proposed encapsidation signals.

V. MECHANISM OF GENERATION OF DEFECTIVE AND/OR DI DNA

A. Lessons from Defective RNA

The mechanisms responsible for the formation of defective RNA, associated with RNA viruses infecting plants and animals have been studied in detail. Two models have been proposed for animal viruses. The first involves a molecular exchange event in which enzymatic or autocatalytic splicing reactions cause RNA rearrangements by a breaking and joining mechanism. The second involves a “copy choice” mechanism in which viral RNA polymerase makes an occasional replication error by “jumping” to either a different portion of the same template strand or a new template strand (Lazzarini et al., 1981).

The mechanisms involved in the generation and evolution of defective RNAs of Tomato bushy stunt tombusvirus (TBSV) were studied using a protoplast system (Jones et al.. 1990). Synthetic transcripts corresponding to different naturally-occurring TBSV defective RNAs, or to various artificially-constructed TBSV defective RNAs, were analyzed for their competitiveness by co- inoculating them with helper genomic RNA transcripts and monitoring their accumulation. Studies were also performed to assess the contribution of stability of the naked defective RNA and their encapsidation efficiency to the observed levels of competitiveness. In addition, the ability of various defective RNAs to produce alternative forms was tested by serially passaging protoplast infections initiated with transcripts corresponding to helper genomic RNA and a single type of defective RNA. These studies, and the analysis of the sequences of observed recombinants, indicated that (i) replication-competence is a major factor dictating DI-RNA competitiveness and is likely to be a primary determinant in DI-RNA evolution; (ii) DI-RNAs are capable of evolving into both smaller and larger forms, and the rates at which various transitions occur, differ; (iii) DI-RNA intermolecular recombination and/or rearrangement is responsible for the formation of the new RNA molecules; and (iv) sequence complementarities between positive- and negative-sense strands in the regions of the junctions suggest that, in some cases, base pairing between an incomplete replicase- associated nascent strand and acceptor template may mediate selection of recombination sites. On the basis of these results, a stepwise deletion model to describe the temporal order of events leading to the formation of tombusviral DI-RNAs was proposed (Jones et al., 1990). The importance of effects of various secondary structure elements as well as GC content has also been noted (Simon et al., 1994). Interestingly, sequences within proteins involved in virus replication were also found to affect the pattern of recombination (Figlerowicz et al., 1997; Nagy et al., 1995).

Recombination in the DNA viruses can be by homologous or nonhomologous (illegitimate) recombination. In geminiviruses, only DNA-DNA recombination can occur but for caulimoviruses, both DNA- DNA recombination, in the nucleus as well as RNA-RNA recombination during replication (Bisaro, 1994) are feasible.

B. Geminiviruses

The mechanisms by which the defective DNA molecules are generated are not well established but several hypotheses have been proposed based on indirect evidence. In geminiviruses, they could be formed by jumping of DNA polymerase during conversion of ssDNA to dsDNA or during rolling circle ssDNA synthesis by recognition of “pseudo- or\igin” of replication or sometimes by intramolecular recombination in dsDNA catalysed by the replication-associated enzymes Topoisomerase I (Topo I) and Topoisomerase II.

A high replication rate has also been proposed to be a major factor for the high recombination observed for geminiviruses, similar to that reported for baculoviruses (MacDowell et al., 1986; Holland, 1985). Such a correlation has been experimentally demonstrated for baculoviruses by Kamita et al. (2003) in Autographa californica nuclear polyhedrosis virus (AcM-NPV) and Bombyx mon nuclear polyhedrosis virus (BmNPV) by studying the timing and frequency of homologous recombination between two BmNPV strains and between BmNPV and AcMNPV after co-infection. In addition, on the basis of in vitro transposition studies using AcMNPV sequences, Martin and Weber (1997) also found that high-frequency recombination is strictly dependent upon viral DNA replication. More recently, Crouch and Passarelli (2002) have shown by co-transfection studies that a subset of genes that are involved in DNA replication is required for high-frequency homologous recombination of AcMNPV. Liu et al (1998) reported that tobacco and tomato plants inoculated with CLCuD-associated viruses contained substantial amounts of defective DNA A after a period of nine months. Thus, it was thought that generation of these small DNAs was a feature of both high replication rates as well as long-term, chronic infection, but once formed they could be transmitted by whiteflies.

Deletions observed in DI-DNA presumably arise during DNA replication although the precise mechanism is not understood. Presence of repeat sequences at the deletion points indicates the involvement of recombination events (Stanley and Townsend, 1985). It is essential that a selective advantage is conferred on such deleted DNA over helper viral DNA, often at the level of replication or possibly encapsidation, to maintain their numbers within the virus population (Stanley et al., 1985).

A number of explanations have been put forward to explain the predominance of defective DNAs derived from the B component over A. In TGMV, the DNA B accumulates to almost three times the levels of DNA A in infected tissues (Orozco et al., 1998; Hung and Petty, 2001), which might partly explain the above. Alternately, subgenomic DNAs derived from DNA A may possibly interfere with virus replication to such an extent that they are selected strongly against. Subgenomic DNA from B component delays the time of symptom formation and is probably analogous to defective interfering nucleic acids of animal viruses (Holland, 1985). Subgenomic DNAs which attenuate symptoms may be selected because they may prevent the virus from killing the host plant. On the other hand subgenomic DNAs which strongly inhibit viral DNA replication could be lethal to the virus.

The eukaryotic Topo I makes single-stranded breaks in dsDNA and cleavage can occur on average at about 1 in 7 residues with no absolute sequence specificity but with some preference for CTT or GTT immediately 5′ to the cleavage site (Been et al., 1984). It was seen that CTT or GTT sequences were present at, or close to, the subgenomic DNA deletion boundaries for ACMV DNA B (Stanley and Townsend, 1985), WDV (MacDowell et al., 1986) and TLCV (Sharma et al., 1998; Zhou et al., 2001). Similar observation have also been reported for the animal virus SV40, where the sequences CTT and GTT have been observed near the cleavage sites of Topo I-mediated nonhomologous recombination (Bullock et al., 1985). The above evidence indicates the close involvement of at least Topo I in generating DI-DNAs.

Jeske et al. (2001) proposed recombination-dependent replication (RDR) to be an important mode of replication in the case of Abutilon mosaic virus (AbMV), a bipartite begomovirus, along with rolling circle replication (RCR) mechanism. This study used electron microscopy and two-dimensional gel electrophoresis analysis of viral DNA forms. During development of naturally infected leaves, viral DNA intermediates compatible with both the above models appeared simultaneously, whereas agroinfection, a process of introduction of cloned viral DNA to plants via Agrobacterium, of leaf discs led to an early appearance of RDR forms but no RCR intermediates. Some of these RDR intermediates were proposed to act as precursors for the generation of defective DNAs.

A template-switching model in which specific sequences are targeted by the replicase during re-initiation of RNA synthesis has been proposed for the generation of defective RNAs and the satellite/ defective hybrid RNAs of Turnip crinkle virus, TCV (Cascone et al., 1990). Certain conserved sequence motifs were found at the 5′ ends of TCV genomic RNA and defective RNAs, as well as near the initiation site of 1.45 kb subgenomic RNA. Template-switching by the replicase has been suggested to occur during replication of minus- strand RNA. Such a step has a high probability of resulting in recombinant products. A similar mechanism may also be responsible for generation of Cymbidium ringspot virus (CymRSV) defective RNAs, since there is sequence similarity at the right side of all junctions with the 5′ end of the helper viral genomic RNA (Burgyan et al., 1989).

Porcine circovirus, whose replication strategy is similar to geminiviruses, displays a strong requirement for a stable stemloop cruciform structure for efficient replication. Introduction of artificially deleted version, lacking a complete stem loop has been reported to result in recombination-mediated recovery of the deleted regions by template switching (Cheung, 2004). Such phenomena can also be expected for geminiviruses. In fact, recombination between DNA A and DNA B CR sequences has been reported for ACMV (Roberts and Stanley, 1994) and the stem-loop sequence was implicated in the production of some recombinants.

C. Caulimovirus

Subgenomic populations of CaMV DNA are detected in infected tissue and probably arise anew in each infection event. They do not accumulate as the result of serial propagation of the virus over many plant generations. Purified CaMV virion DNA does not contain this population of subgenomic DNA (Guilfoyle et al., 1983). The defective molecules in caulimoviruses are believed to be formed either by DNA-DNA recombination in nucleus and/or RNA-RNA recombination in cytoplasm (Walden and Howell, 1982). They are mainly formed when the reverse transcriptase reaches a break in the transcript being replicated and the nascent strand is transferred to a homologous region within the genome (Olszewski et al., 1983; Coffin, 1979). The presence of secondary structures can also influence their formation. The rapid accumulation of subgenomic CaMV DNA species by intramolecular recombination events could also be because of its replication by reverse transcription. It is also possible that the subgenomic CaMV DNA species may represent dead- end products of aberrant CaMV replication.

CaMV nucleic acids accumulate in the cell in different structural conformations related to their roles in gene expression, replication and virion assembly. Covey and Turner (1993) have characterized changes in the population of CaMV DNA and RNA replication products which occur following culture of infected turnip leaves under conditions where callus proliferates. The change in pattern of OC (open circular) and SC (supercoiled) DNA forms during callus proliferation suggests a possible precursor/product relationship involving generation of deleted molecules from gap-containing virion DNA-like molecules followed by sequential repair of the gaps to produce SC DNA. The presence of CaMV SC DNAs in replicating cells might also enhance illegitimate integration into host chromosomes, as hybridization of CaMV DNA to high molecular weight DNA was observed (Covey and Turner, 1993).

Walden and Howell (1982) demonstrated intergenomic recombination events among pairs of defective CaMV genomes in planta. Pairs of defective CaMV genomes co-inoculated in plants were rescued at high frequency by homologous, intergenomic recombination events under strong selection pressures. Such events may take place during the normal viral replication processes however, their relevance in the absence of strong selection favouring the recombinant products has to be assessed.

VI. ROLE OF HOST AND HELPER VIRUS IN DI-DNA GENERATION AND ACCUMULATION

Defective DNAs are usually formed at high levels in the experimental hosts and rarely in the natural hosts for the helper viruses. Agroinoculation of N. benthamiana plants with BCTV (Logan strain) consistently resulted in the de novo production of subgenomic DNAs of several different sizes and forms on initial passage. However, extracts from sugar beet and tomato infected with the same strain contained variable but usually smaller amounts of subgenomic DNAs, suggesting that their production may be influenced by the host species (Stenger et al., 1992; Frischmuth and Stanley, 1992). Infectivity and the formation of defective molecules were studied in Spinach curly top virus (SCTV) in different experimental hosts. The amount of DI-DNA varied among hosts, ranging from strong in N. benthamiana and Arabidopsis thaliana to very weak in carrot. Low amounts of defective DNA appeared in other hosts experimentally inoculated with the virus and in its natural host (Baliji et al., 2004) and also in cassava infected with ACMV (Coutts and Buck, 1987). Defective DNA forms are readily produced de novo by other curtoviruses using cloned DNA in experimental hosts but at lower levels in field samples (Stenger, 1995). The only exception of high levels of DI-DNA in natural hosts is seen in A. conyzoides infected with AYVV (Stanley et al., 1997). Stanley et al. (1997) observed that the DI-DNAs obtained from naturally-infected A. conyzoides contained non-viral sequences, but in contrast, using the same cloned DNAs, they we\re not detected in the agroinoculated N. benthamiana plants. Low levels of subgenomic DNAs of CPGMV DNA B genomic component were observed in naturally-infected cowpea (Winter et al., 1998).

Romero et al. (1993) studied the effect of host on accumulation and encapsidation of DI-RNAs in the RNA virus Broad bean mottle virus BBMV, which encapsidates DI-RNA components that are formed by in-frame deletions of the helper virus. Biological characteristics of these DI-RNAs include exacerbation of the severity of symptoms on certain hosts and the lack of DI-RNA in other hosts. In order to study the molecular aspects of the relationship between DI-RNA transmission and symptoms, their accumulation and encapsidation in various local lesion or systemic host plants under various environmental factors was investigated by Llamas et al. (2004). Serial passages of BBMV through selected plant species revealed that, with low multiplicity inocula, some systemic hosts (Vicia faba, N. clevelandii and N. tabacum cv. Samsun) supported DI-RNA accumulation after first passage cycle but other hosts (Phaseolus vulgaris, Pisum sativum and Glycine max) did not. However, several passages with high multiplicity inocula could generate DI-RNAs in P. sativum as well. Local lesion hosts (Chenopodium quinoa, C. amaranticolor and C. murale) remained free of DI-RNA components. The size of de novo-formed DI-RNAs depended on host and environmental conditions. For instance, broad bean plants cultivated in a greenhouse or a growth chamber at 20C accumulated DI-RNAs of 1.9 kb or 2.4 kb, respectively, and a reverse trend was observed in pea plants, i.e., 2.4 kb was detected when plants were grown in a growth chamber under the same conditions and 1.9 kb in glasshouse. Lower temperature greatly facilitated formation of DI-RNAs in broad bean and pea hosts after first passage (Romero et al., 1993). In addition to the kind of host, the studies with DI-RNA of Cucumber mosaic virus (CMV) have shown tissue-specificity in replication, systemic movement and encapsidation (Romero et al., 1993).

To explain these results, it was speculated that the host affects the RNA conformation under given environmental conditions, so that certain sites of the parental viral RNAs are exposed for crossover events. Alternatively, certain steps in the virus life cycle or changes in physiological conditions of the host can alter the formation of DI-RNAs. Interestingly, when the virus preparation was derived from one host and then inoculated on the second host (e.g., pea to broad bean), initially both DI-RNA types coexisted (the parental one and the one characteristic for the given host). However after the second passage, the DI-RNAs of cognate size predominated and, after more passages, they became exclusive and the parental type became undetectable. The local lesion hosts did not support the accumulation of DI-RNAs, which could be because of the limited number of replication cycles in it or because of the plant defense mechanisms.

To explain the above results, it was hypothesized that the RNA polymerase functions at suboptimal conditions at 12C, causing the enzyme to make more errors (i.e., RNA-RNA crossovers) during replication. Also, lower temperatures could preserve secondary structures of virus RNA templates that function as the polymerase pausing sites and thus favor the formation of DIRNAs, change the host defense, or alter selection pressure. Similar kind of studies with plant DI-DNAs in different experimental host plants may unravel similar phenomena and help to understand the factors responsible for the generation of DI-DNA.

VII. SYMPTOM MODULATION BY DI-DNA

Symptom modulation by DI-DNA is manifested by a general decrease in symptom severity accompanied by fall in the levels of helper viral DNA (Figure 2). The DI-RNAs associated with the RNA virus Tomato bushy stunt virus, TBSV (Hillman et al., 1987) and CymRSV (Burgyan et al., 1989) significantly attenuate the symptoms of their helper counterparts. Plants inoculated with purified virus and DI- RNA exhibit local necrotic lesions only on inoculated leaves followed by a persistent infection. Symptom amelioration is correlated with both increase in the accumulation of DI-RNA and decrease in amount of helper virus. Studies using the 0.7 kb DI-RNA of TBSV in the tobacco protoplasts indicated 65% reduction in the rate of helper virus genomic RNA synthesis (Jones et al., 1990). By analyzing the rate of genomic RNA synthesis they were able to discount the possibility of accumulation of DI-RNA due to the enhanced degradation of genomic RNA or selective suppression of subgenomic RNA synthesis.

FIG. 2. Mobilzation of geminivirus DNA from integrated templates. A. DNA B (large shaded disc) is mobilized from an integrated dimeric template and amplified when the transgenic plant is challenged with cognate DNA A (non shaded disc) to produce a systemic infection. B. DI-DNA (small shaded disc) is similarly mobilized from an integrated dimeric template when the transgenic plant is challenged with cognate virus. Amplification of the DI-DNA protects the plant against infection by reducing the level of full-length genomic components. (Reprinted from Seminars in Virology, Volume 4, Frischmuth, T., and Stanley, J., Strategies for control of geminiviral diseases, pp. 329-337, 1993, with permission from Elsevier).

Unlike DI-RNAs of other plant and animal viruses, DI-RNAs of TCV markedly intensify the symptoms of the helper virus (Altenbach and Howell, 1981; Li et al., 1989). The intensification of symptoms by certain satellites or DI-RNA of TCV is completely dependent on the helper virus genomic RNA (Li and Simon, 1990). Similar results have been obtained for several CMV satellites (Sleat and Paulkaitis, 1992). These observations suggest that an interaction between satellite RNAs and viral genomic RNA or encoded products is necessary for symptom amelioration.

It was noted that the TbLCZWV DI-DNAs from two different plants, “Mild” and “HG” (Paximadis and Rey, 1997) were different in structure, and since defective DNAs are known to cause symptom amelioration, it could be hypothesized that the defective DNAs may be playing a role in symptom modifications (Paximadis and Rey, 2001).

The DI-RNAs of TBSV (Carpenter and Simon, 1996a), CyRSV (Ahlquist, 2002), and Cucumber necrosis virus (CNV) (Havelda et al., 1998) decrease severity of symptoms caused by the respective helper viruses. Reduction of the levels of the helper viral RNA is mainly due to competition of DI-RNAs with the helper virus for available replicase protein (Carpenter and Simon, 1996b). Later studies with TBSV DI-RNAs showed that their effect could also be due to an inhibition in the expression of the viral proteins P22 (responsible for cell-to-cell movement) and P19 (associated with tissue necrosis), in addition to the reduction in accumulation of viral genomic RNA (Havelda et al., 1998).

Symptom amelioration by DI-DNA was used as a strategy against ACMV by producing transgenic N. benthamiana plants containing integrated copies of DI-DNA derived from DNA B of the virus. The plants showed amelioration of the symptoms, upon ACMV infection (Frischmuth and Stanley, 1993). In the above case the DI-DNA generated due to mobilization of the integrated copy brought about reduction in the titres of DNA B of the infecting virus. The disproportionate reduction in level of full-length components (DNA A, 20% reduction; DNA B, 70% reduction) indicated that the episomally-replicating subgenomic DNA had been amplified at the expense of full length DNA B to three times the level of the latter. Serial infection of transformants resulted in a further decrease in symptom severity and viral DNA levels. No differences were observed in the severity of symptoms or levels of viral DNA when transformants and controls were challenged with the related geminiviruses BCTV and TGMV, demonstrating the specific nature of the interaction. Analysis of the infected tissue showed that TGMV was unable to amplify subgenomic DNA. It was speculated that it was possible to enhance the effect of the subgenomic DNA by further increasing the copy number. This was proposed to be achieved by reducing the size of DNA while ensuring that cis-acting sequences responsible for replication remained intact (Stanley et al., 1990).

When the concentration of DI-DNA was increased in the inoculum containing DNA A and B of ACMV, there was a decrease in the number of plants showing symptoms as well as the time required for symptom development (Stanley and Townsend, 1987). Cloned TbLCZWV DNA A, a monopartite begomovirus, upon agroinoculation, gave rise to a variety of symptoms in N. tabacum cv. Samsun, which was mostly because of the varied amount and type of DI-DNAs present (Paximadis and Rey, 2001).

In addition to symptom amelioration, AYVV DI-DNA influences the pattern of chlorosis and leaf curling (Stanley et al., 1997). The DI- DNA altered the symptom phenotype and caused a delay in the accumulation of DNA of the helper virus when reintroduced into N. benthamiana in the presence of AYVV DNA (Stanley et al., 1997). The severe upward leaf roll symptom that is characteristic of infection of N. benthamiana by single component geminiviruses such as AYVV and BCTV (Stanley and Latham, 1992; Tan et al., 1995) developed only rarely and in a very mild form in presence of DI-DNA. Leaf curl phenotype is determined by the ability of virus to affect certain tissues at an early stage of their development rather than by overall level of virus that accumulates in these tissues (Stanley et al., 1997). Thus, AYVV DI-DNA may play a biological role by slowing down the infection process that may otherwise be deleterious to the plant and, hence, disadvantageous for maintenance of virus population.

To further understand the molecular basis of the interference phenomenon, Frischmuth et al. (1991) have compared the ability of ACMV a\nd TGMV genomic DNA to replicate in the leaf discs derived from the DI-DNA-containing transgenic plants. Results indicated that the DI-DNA of ACMV interfered with the replication of both the genomic DNAs of ACMV to similar extent. TGMV DNA A replicated to normal levels in transformed leaf discs and plants because it was unable to mobilize and amplify ACMV DI-DNA. The results indicated that the DI-DNA replicated at the expense of both genomic components and suggested that the basis of the interference phenomenon was the competition for rate limiting amounts of trans-acting proteins required for replication, possibly virus encoded AC1, which may be true for several other viruses (Etessami et al., 1991; Frischmuth and Stanley, 1991, 1992; Paximadis and Rey, 2001).

Various evidences point towards the DNA B as the component primarily responsible for symptom development in bipartite begomoviruses. The common strain and the yellow vein strain of TGMV show marked phenotypic differences in N. benthamiana, following infection with cloned viral genomes. Pseudorecombinants between the two strains, produced by exchange of genome components (DNA A and B), established that the difference in symptoms is determined by DNA B (von Arnim and Stanley, 1992a). Recombinants produced in vitro between the DNA B components showed the determinants of symptom development map to the common region and gene BC1 (von Arnim and Stanley, 1992b). DNA B is known to carry functions necessary for spread of viral DNA through the host plant. Intermediate or completely novel symptoms have been a feature in recombinant experiments with other plant DNA viruses like CaMV and MSV (Stratford and Covey, 1989; Boulton et al., 1989). Similarly, pseudorecombinant experiments between ACMV strains suggested that DNA B influenced the severity of symptoms induced in N. bethamiana and N. clevelandii (Morris et al., 1990). This was irrespective of whether the viral DNA was introduced into plants by mechanical inoculation of cloned inserts or by agroinoculation of partial repeat sequences. However, the response of N. benthamiana to ACMV infection has also been correlated with large fluctuations in the levels of DNA B (Stanley et al., 1990). In addition to modulation of the symptoms, the results demonstrated that ACMV DNA B also influenced the pattern of chlorosis (von Arnim and Stanley, 1992a) in N. benthamiana.

Symptom amelioration was associated with the mobilization of integrated DNA by homologous recombination and an amplification of extra-chromosomal copies of the DI-DNA with a concomitant decrease in the level of DNA. The results also indicated that the resistance is not conferred simply by the presence of an integrated copy of partial repeat of the DI-DNA, but by its ability to be mobilized and transreplicated by the infecting virus (Frischmuth and Stanley, 1994).

The importance of mobilization of integrated DI-DNA in symptom amelioration was illustrated by Frischmuth and Stanley (1994), when N. benthamiana plants transformed with a repeat of BCTV DI-DNA remained susceptible to infection but produced ameliorated symptoms when agroinoculated with BCTV. To further understand the molecular basis of the interference phenomenon, Frischmuth et al. (1997) compared the ability of BCTV to replicate and accumulate in leaf discs of resistant and non-resistant transgenic plants obtained by using two different sized DI-DNAs of the same virus for transformation. If the DI-DNA is preferentially amplified as a result of its smaller size relative to the full-length genomic component, it might be possible to increase its potential to interfere with virus proliferation by using smaller subgenomic DNAs. Both the subgenomic DNAs were able to interfere with virus replication but resistance was found only in some plants, in which the there was accumulation of the DI-DNA. Because in both transformed plant lines subgenomic DNA was amplified, a mere amplification of any subgenomic DNA cannot be the primary reason for BCTV DIDNA-mediated resistance. As a consequence, in order to use subgenomic DNAs of BCTV as a tool to mediate resistance in transgenic crops, it appears that the size of such DNA molecules is crucial and should be limited to approximately half genomic size of the helper viral DNA.

The DI-DNAs derived from the DNA β are also believed to be responsible for the attenuated symptoms (Saunders et al., 2001). Defective DNA derived from the monoparti te begomovirus AYVV altered the symptom phenotype and caused a delay in the accumulation of viral DNA, when reintroduced into N. benthamiana in the presence of the genomic DNA of helper virus. The severe upward leaf roll symptom that is characteristic of infection of N. benthamiana by AYVV and BCTV (Stanley and Latham, 1992; Tan et al., 1995) developed only rarely and in a very mild form in the presence of the DI-DNA (Stanley et al., 1997). This suggests that leaf curl phenotype is determined by the ability of the virus to affect certain tissues at an early stage of the development rather than by the overall level of virus that accumulates in these tissues. Similarly, infection ofAlthea rosea with Hollyhock leaf crumple virus (HLCrV) gives rise to a range of symptoms. This may be attributed to differences in the level of DI-DNAs found naturally (Bigarr’e et al., 2001).

VIII. INTERACTION OF DI-DNA PRODUCTS WITH VIRAL AND HOST COMPONENTS

A. Inhibition of Rep

It is clear that the DI-DNAs exert their symptom-modulating influence through interactions directly with a number of other viral or cellular components or indirectly through their products. Since one of the major effects is the slowing-down of the helper virus replication, the Rep protein, which is absolutely required for initiation of viral DNA replication was studied. Several DI-DNAs produce C-terminal truncated Rep proteins. The N-terminal part of Rep protein (amino acid residues 1-160) of ToLCNDV has been shown to be involved in the specificity of origin recognition and binding (Chatterji et al., 2001). Considering the various activities that are associated with Rep (1-160), it is possible to suggest the mode of action of Rep (1-160) in limiting virus DNA accumulation. One possibility is that the Rep (1-160) protein reduces replication by competing with the viral Rep protein for binding the iteron sequences in the origin (Chatterji et al., 2001). The truncated Rep protein may therefore behave as a dominant negative mutant (Herskowitz, 1987) and block virus replication. Another possibility is that the truncated Rep protein does not contain the NTP binding domain present on the C-terminus of the Rep protein. The NTP binding domain is required for replication (Desbeiz et al., 1995), and the lack of this region may interfere with the normal replication process of the virus.

B. Activation of PTGS

Post-transcriptional gene silencing (PTGS) is a specific RNA degradation mechanism, believed to be universally present in living organisms. In plants, it has been implicated in various phenomena including virus resistance (Ratcliff et al., 1999). PTGS is most strongly triggered by dsRNA (including hairpin RNA) and targets homologous RNA (either endogenous cellular RNA, transgene RNA or viral RNA), resulting in its degradation to 21-23 base RNA known as siRNA. This specific degradation is undertaken by a complex of proteins and RNA, known as the RISC complex. The RNA components of the RISC complex are derived from siRNA and function in the specificity of target recognition, known as “guide RNA.” When a plant virus infects a host cell it activates the above RNA-based defence that is targeted against the viral genome (Ratcliff et al., 1999). Many plant viruses encode proteins that are suppressors of this RNA silencing process (Voinnet, 2001) and influence the final steady-state level of virus accumulation.

The role of PTGS in DI-RNA induced symptom modulation was investigated by analyzing the composition of virus-derived siRNAs in CymRSV infected plants. The presence of DI-RNAs caused dramatic increase in the siRNA corresponding to sequences encompassed by the DI-RNAs (Szittya et al., 2002). A comparison of target activity of different DI-RNAs and viral genomic regions that are not included in DI-RNA molecules, suggested that the target activity of different regions of the viral genome can vary significantly. The finding that short DI-RNAs were targeted poorly by the helper virus-induced PTGS suggested that they may accumulate in virus-infected plants. It was also demonstrated that the poor target activity of short DI-RNAs was not caused by either the low level of the DI-RNA-specific guide RNAs or the size of the target molecule. It is also possible that short DI-RNAs, which contain almost exclusively cis-acting sequences required for replication (Havelda et al., 1998), may bind efficiently to the replication complex and thus are not accessible to the RISC degradation complex of PTGS.

The target activity of endogenous genes towards triggering PTGS also might vary. The capacity of distinct regions from the phytoene desaturase (PDS) gene expressed by PVX to confer virus-induced gene silencing differed significantly (Thomas et al., 2003). Furthermore, certain genes were resistant to RNAi inactivation in Caenorhabditis elegans (Fraser et al., 2000) and in HeLa cells (Elbashir et al., 2001), even though guide RNAs were abundant in both systems. It is likely that the low target activity of these endogenous genes contributed to their resistance against RNAi-mediated inactivation.

CymRSV-infected plants displayed a number of DI-RNAs, some large (DI-13) and some small. In contrast to the short CymRSV DI-RNAs, the large DI-13 RNA was a good target for PTGS induced by the helper virus. It was also demonstrated that a stretch of 235 nucleotides (Ml), the only region present in DI13 and absent in short DI RNAs, was responsible for the good target activity of \DI-13 RNA (Szittya et al., 2002). Previously, it was reported that Ml was deleted preferentially during the evolution of CymRSV DI-RNAs (Havelda et al., 1998). These coinciding results suggest that PTGS provides selective pressure on DI-RNA accumulation by eliminating those molecules containing PTGS target sequences. These observations suggest that short DI-RNAs are weak targets of PTGS. Therefore, it is suggested that, in addition to a size-related replication advantage (Roux et al., 1991; White, 1996), PTGS-mediated selection also plays a decisive role in the selective accumulation of short DI- RNAs. Such phenomena may be operational even in the case of DI- DNAs.

IX. EVOLUTIONARY SIGNIFICANCE

In infected A. conyzoides, in addition to a single AYVV genomic component (DNA A), infected plants contained chimeric defective viral components, comprising DNA A and non-viral sequences, acting as DI-DNAs. One such DI component (def 19) showed significant homology with sequences of nanovirus components (Saunders and Stanley, 1999). This can be considered to be a direct evidence for recombination between distinct families of plant single-stranded DNA viruses and suggests that co-infection by geminivirus and nanovirus- like pathogens may be a widespread phenomenon. The ability of plant DNA viruses to recombine in this way may greatly increase their scope for diversification (Saunders and Stanley, 1999). Similarly, some of the molecules reported to be associated with the CLCuDcausing viruses were found to be chimeras of sequences derived from CLCuMV DNA A and other satellites and satellite-like molecules (Briddon et al., 2001; Varma and Malathi, 2003). Another important example was the report of a yellow vein phenotype produced by co- inoculating A. conyzoides with AYVV DNA A and recDNA-Ab!7, a naturally occurring recombinant of approximately the same size as DNA β that contains sequences from both DNA A and DNA β. Symptoms induced by DNA A and recDNA-Aβ17 in A. conyzoides and N. glutinosa were reported to be qualitatively similar to those associated with DNA A and DNA β, although milder. This recombination was observed after whitefly transmission of the disease in A. conyzoides (Saunders etal., 2001 ). Hence, such recombination events are likely to occur frequently, implying that recombinants will normally be associated with this type of disease complex in the field.

X. ECONOMIC IMPORTANCE

The use of transgenically mobilized DI-DNAs for resistance against ACMV has already been discussed (Frischmuth and Stanley, 1993). Geminiviral vectors can also be used to express heterologous proteins in plants (Timmermans et ai, 1994). In addition, since generation of DI-DNAs inhibits viral multiplication, it may be of interest to suppress the generation of such molecules to ensure high accumulation and expression of episomally replicating geminiviral vectors. Such an approach may be feasible, based on the strategies adopted for reducing the generation of DI-DNAs against the expression vectors based on baculoviruses to express proteins in insect cell lines. These approaches are based on eliminating specific sequences which are necessary for the accumulation of DI- DNAs against baculoviruses (Pijlman et al., 2001, 2002). A similar situation can be anticipated and resolved in construction of high- expression gene vectors on the basis of geminiviruses. CaMV can also be used as a vector to carry foreign DNA into plants; however, the foreign DNA which can be maintained has a maximum size of 450 bp (Gronenborn et al., 1981). If the smaller sized subgenomic DNA can replicate, they may have the potential to act as vectors which can accommodate larger sized foreign DNA than the full genome length CaMV DNA.

Virus-induced gene silencing (VIGS) is a recently-developed technique to specifically silence endogenous genes by the help of replicating viral nucleic acids and thus study gene function (Baulcombe, 1999). Infectious clones of such nucleic acids are used as independently replicating vectors, containing part of the gene to be silenced (Kumagai et al., 1995; Ruiz et al., 1998; Jones et al, 1999; Ratcliff et al, 1999; Turnage et al, 2002; Liu et al, 2002a, 2002b). Since viral nucleic acids commonly involve double-stranded RNA intermediates during replication, it acts as a trigger for an amplifying and spreading silencing signal, using the mechanism of post-transcriptional gene silencing (PTGS) to specifically bring about degradation of the target gene transcripts (Kjemtrup et al, 1998). Geminiviral DNA has also been used as VIGS vectors to study gene function (Kjemtrup et al, 1998; Peele et al, 2001; Turnage et al, 2002; Fofana et al, 2004). DI-DNAs of geminiviruses open up the possibility of enhancing the effectiveness of such vectors because they would be able to replicate better and accommodate larger inserts because of their smaller sizes. Even though such an approach is yet to be tried, it opens up the possibility of silencing entire metabolic pathways by simultaneously silencing several genes, cloned in such DI-DNA vectors.

Xl. CONCLUSION

Defective DNAs are associated with all families of DNA containing plant viruses. They are not restricted to a particular family and thus appear to be a natural by-product of virus infection. The majority of the defective DNAs are derived from the B genomic component of bipartite begomoviruses. It may be possible that – derived and B-derived defective DNAs are formed, but only the latter accumulate, the former being not selected. The notion that there are structural constraints against accumulation of defective DNAs derived from A is also plausible, but seems to be not adequately supported by experimental evidence. One common assumption is that the B genomic component is partially dispensable and the A component is capable of deploying some of its own genes to carry out the functions of the