Characterization of DNA{beta} associated with begomoviruses in China and evidence for co-evolution with their cognate viral DNA-A

Geminiviruses are a group of plant viruses characterized by their geminate shape and the size of their particles, which encapsidate a circular single-stranded DNA genome. The family Geminiviridae is divided into four genera (Mastrevirus, Topocuvirus, Curtovirus and Begomovirus) with the majority of described geminiviruses belonging to the genus Begomovirus. Most begomoviruses have bipartite genomes, referred as DNA-A and DNA-B components. A few species have only a single genomic component that resembles DNA-A.

Full-length clones of monopartite begomoviruses, Ageratum yellow vein virus (AYVV) from Singapore and Cotton leaf curl Multan virus (CLCuMV) from Pakistan, although infectious, were unable to induce typical symptoms of yellow vein in Ageratum conyzoides and leaf curl in cotton, respectively, and novel molecules, named DNAβ, were shown to be associated with both viruses, and to be essential for induction of characteristic symptoms in Ageratum and cotton (Saunders et al., 2000; Briddon et al., 2001). Analysis of DNAβ molecules revealed that they are approximately half the size of the genomic DNA-A and except for a conserved hairpin structure and a TAATATTAC loop sequence, have little sequence similarity to either DNA-A or DNA-B molecules of begomoviruses. DNAβ requires begomovirus DNA-A for replication, encapsidation, insect transmission and movements in plants (Saunders et al., 2000; Briddon et al., 2001).

In China, several begomoviruses have been reported infecting squash, tobacco and tomato (Zhou et al., 2001a; Yin et al., 2001; Xie et al., 2002). DNA-A molecules of 15 begomovirus isolates from tobacco, tomato, squash and weed species have been sequenced in our laboratory (X. Zhou and others, unpublished results). However, attempts to find DNA-B components by PCR using DNA-B primers and Southern blotting have been unsuccessful in all samples tested. In order to determine if these begomoviruses were associated with DNAβ-like molecules, DNAβ-specific primers were designed and used for PCR amplification. As a result, DNAβ molecules were found to be associated with many begomovirus isolates in China. In this paper, we describe the genomic structure and the molecular variation of 18 DNAβ molecules, and provide evidence of co-evolution of these DNAβ molecules with the DNA-A molecules of their helper viruses. In addition, we also demonstrate that the C1 ORF has evolved similarly as DNAβ molecules. The role of the C1 ORF of DNAβ in symptom induction is examined and the existence of species of DNAβ molecules is proposed. Furthermore, recombination between DNAβ molecules is documented.

Virus sources and DNA extraction.
Between 1999 and 2002, young seedlings were collected from naturally infected plants of squash, tobacco, tomato, Malvastrum coromandelianum and Siegesbeckia orientalis, which were showing begomovirus-like infection symptoms, in Yunnan province, China, separated by 700 km. Viral DNA from the following samples (years of collection in parentheses) was extracted as described by Xie et al. (2002).

(a) From tobacco with stunting, leaf curl, vein swelling, vein darkening or curly shoot symptoms: Y2 Baoshan, 580 km west of Kunming (1999); Y3 Baoshan, 580 km west of Kunming (1999); Y35 Baoshan, 580 km west of Kunming (2001); Y92 Baoshan, 580 km west of Kunming (2002); Y98 Baoshan, 580 km west of Kunming (2002); Y115 Baoshan, 580 km west of Kunming (2002).

(b) From tobacco with leaf curl and enation symptoms: Y5 Dali, 400 km north-west of Kunming (1999); Y8 Honghe, 160 km south of Kunming (1999); Y10 Honghe, 160 km south of Kunming (1999); Y11 Baoshan, 580 km south of Kunming (2000); Y36 Honghe, 160 km south of Kunming (2001); Y38 Honghe, 160 km south of Kunming (2001); Y43 Dali, 400 km north-west of Kunming (2001); Y45 Honghe, 160 km south of Kunming (2001); Y87 Baoshan, 580 km west of Kunming (2002); Y88 Baoshan, 580 km west of Kunming (2002).

(c) From tomato with leaf curl and enation symptoms: Y25 Chuxiong, 180 km north of Kunming (2000).

(d) From squash with leaf curl symptom: Y23 Jinghong, 700 km south of Kunming (2000).

(e) From Malvastrum coromandelianum with yellow vein symptom: Y47 Honghe, 160 km south of Kunming (2001).

(f) From Siegesbeckia orientalis with leaf curl symptom: Y64 Honghe, 160 km south of Kunming (2001).

PCR and sequence determination.
Comparison of the reported DNAβ sequences of AYVV and CLCuMV was performed and a conserved region (nt 12481347 of AYVV DNAβ) was found. Based on the conserved nucleotide sequences, abutting primers beta01 (5'-GGTACCACTACGCTACGCAGCAGCC-3') and beta02 (5'-GGTACCTACCCTCCCAGGGGTACAC-3') were designed and used for amplification of the possible full-length DNAβ; a unique KpnI restriction endonuclease site (underlined) was introduced into these primers (Briddon et al., 2001). Additional primers designed on the basis of the subsequently determined sequences were used to amplify fragments which cover the region of abutting primers beta01/beta02. PCR was carried out as described by Zhou et al. (1998). PCR products were recovered, purified and cloned into pGEM-T Easy vector (Promega) as described by Zhou et al. (1998). Sequences were determined using the automated model 377 DNA sequencing system (Perkin Elmer).

Clone construction and plant inoculation.
DNAβ from begomovirus isolate Y10 (Y10β) was used for construction of an infectious DNAβ clone. A complete genome unit of Y10β was amplified using beta05 (5'-GAAACCACTACGCTACGCAGCAGCC-3')/beta02 and the fragment inserted into pGEM-T Easy vector to produce clone pGEMβ. Subsequently, another full-length genome copy of Y10β from clone pGEMβ with beta01 and beta 02, was digested with KpnI and inserted into the unique KpnI site of pGEMβ to produce pGEM2β. Clone pGEM2β was digested with EcoRI and inserted into the binary vector pBinPLUS to produce pBinPLUS-2β, yielding a tandem repeat of Y10β.

Two independent PCRs were performed with primer pair beta05/C1R and beta02/C1F using pGEMβ as the template for construction of a C1 in-frame ATG mutation (nt 564566). The primer sequences covering and flanking the mutation sites (underlined) were 5'-AGTTCAGTTTATTTGTTGTGG-3' (C1F, sense) and 5'-CCACAACAAATAAACTGAACT-3' (C1R, complementary). PCR products were fused and amplified using an overlap extension-PCR as described by Tao et al. (2002), and the overlapping PCR product inserted into pGEM-T Easy vector to produce clone pGEMC1mβ. The same strategy was then used for construction of a tandem repeat of the C1 in-frame ATG mutation of Y10β (pBinPLUS-C1m2β).

Agrobacterium tumefaciens strain EHA105 was transformed with pBinPLUS-2β or pBinPLUS-C1m2β by triparental mating. The infectious clone pBinPLUS-1.7A, containing partial repeats of Y10 DNA-A, was constructed previously (X. Zhou and others, unpublished). Agrobacterium tumefaciens cultures were grown at 28 °C for 48 h (OD₅₅₀=1), after which a fine needle was used to inject 0·2 ml of culture into stems or petioles of plants at the six-leaf stage. Nicotiana benthamiana, N. glutinosa and Lycopersicon esculentum plants were agro-inoculated, either with pBinPLUS-1.7A or with pBinPLUS-2β and pBinPLUS-1.7A, while N. benthamiana plants were also agroinoculated with pBinPLUS-C1m2β and pBinPLUS-1.7A. Inoculated plants were grown in an insect-free cabinet at a constant temperature of 25 °C with supplementary lighting corresponding to a 16 h day length.

Sequence analysis.
Sequence data were assembled and analysed with the aid of DNAStar software. Other reported DNAβ molecules used for comparison were DNAβ of AYVV (AYVVβ, AJ252072), Bhendi yellow vein mosaic virus (BYVMV) from India (BYVMVβ, AJ308425), CLCuMV from Pakistan (CLCuMVβ-01, AJ292769; CLCuMVβ-02, AJ298903) and Cotton leaf curl Rajasthan virus (CLCuRV) from India (CLCuRVβ, AY083590). The database accession numbers of the begomovirus DNA-A sequences used for comparison are listed as follows: AYVV (X74516); BYVMV (AF241479); CLCuMV (AJ132430); CLCuRV (AF363011); Malvastrum coromandelianum begomovirus isolate Y47 (AJ457824); squash begomovirus isolate Y23 (AJ420319); Squash leaf curl China virus (SLCCNV, AB027465); Siegesbeckia orientalis begomovirus isolate Y64 (AJ457823); tobacco begomovirus isolates Y5 (AJ319674), Y8 (AJ319677), Y10 (AJ319675), Y11 (AJ319676), Y36 (AJ420316) and Y38 (AJ420317); Tobacco curly shoot virus (TbCSV) isolates Y1 (AF240675), Y2 (AF240676) and Y35 (AJ420318); Tobacco leaf curl Yunnan Virus (TbLCYNV) isolate Y3 (AF240674); tomato begomovirus isolate Y25 (AJ457985) and Tomato yellow leaf curl China virus (TYLCCNV, AF311734).

DNAβ associated with begomoviruses in China
Sixteen DNA samples from tobacco, one from tomato, one from squash, one from Malvastrum coromandelianum and one from Siegesbeckia orientalis showing begomovirus-like infection symptoms, and which had been collected from widely separated locations in Yunnan province between 1999 and 2002, were tested by PCR with abutting primers beta01 and beta02 for the presence of DNAβ fragments. Approximately 1·4 kb was consistently amplified from all samples, except isolates Y3 and Y23. Amplification with nucleic acid extracts produced from healthy control plants produced no product, indicating that DNAβ is widely associated with begomoviruses in Yunnan, China. The Y3 isolate from tobacco has previously been named as Tobacco leaf curl Yunnan virus (TbLCYNV) (Zhou et al., 2001a), while the sequence of the Y23 DNA-A from squash indicates that it is also a new begomovirus species, for which the name Squash leaf curl Yunnan virus (SLCYNV) is proposed (Y. Xie and others, unpublished results).

Sequence analysis of DNAβ
The complete nucleotide sequences of DNAβ molecules associated with 18 virus isolates were determined to be 1333 to 1355 nt in length. These sequences have been submitted to GenBank, and are available under accession nos AJ42031315, AJ42148285, AJ42161923, AJ45781822 and AJ506791. Sequences of the DNAβs obtained from the above samples are named corresponding to their associated begomovirus: thus Y2β refers to DNAβ from isolate Y2, etc. Nucleotide numbering for DNAβ, as for geminivirus genomic components, proceeds from the 3' A residue in the nonanucleotide sequence TAATATT/AC.

Comparison of these sequences shows that they are of three main types (Table 1). Type 1 includes 13 virus isolates from tobacco, tomato and Siegesbeckia orientalis. The sequences in this type show 7299 % overall nucleotide sequence identity, while the isolates Y8β, Y36β, Y38β, Y45β and Y64β show 9799 % sequence identity with each other. Type 2 includes four virus isolates from tobacco, for which the overall nucleotide sequence identity is 8398 %. Type 3 includes the virus isolated from Malvastrum coromandelianum. Only 5257 % overall nucleotide sequence identity was found between Type 1 and Type 2 DNAβ molecules. Type 3 is relatively different from the other DNAβ molecules with less than 42 % overall nucleotide sequence identity with DNAβ molecules within Type 1 and Type 2. Comparison between Type 1, Type 2 and Type 3 DNAβ and reported DNAβ molecules (AYVVβ, CLCuMVβ, CLCuRVβ and BYVMVβ) shows low sequence identity (3543 %), except that Y47β has relatively high overall sequence identity (6267 %) with CLCuMVβ and CLCuRVβ (Table 1).

Table 1. Percentage complete nucleotide (top right) or C1-encoded amino acid (bottom left) sequence identities among 18 DNAβ molecules associated with begomoviruses in China and five other reported ones

Structural features of DNAβ
The 18 DNAβ molecules isolated in this study have no obvious sequence homology with begomovirus genomic components, except for the nonanucleotide sequence TAATATT/AC. DNAβ and DNA-A do not share a common region, although DNAβ clearly must possess sequences that are recognized by the DNA-A-encoded replication-associated protein (Rep) in order to allow DNA-A-mediated replication of the DNAβ (Saunders et al., 2000). Neither the reiterated sequences in DNA-A (iterons), which participate in Rep binding (Fontes et al., 1994; Chatterji et al., 1999, 2000), nor conserved TATA motifs are present in DNAβ. Although the 18 DNAβ molecules isolated in this study have relatively low sequence identity with the other reported DNAβ molecules, comparison shows that all 18 DNAβ molecules do have a very conserved region of 115 nt which has 93100 % identity with the DNAβ consensus sequence reported elsewhere (Saunders et al., 2000; Briddon et al., 2001). An alignment of this conserved region is shown in Fig. 1. It contains the conserved nonanucleotide sequence TAATATT/AC in the loop of a putative stemloop structure. In geminiviruses, this sequence contains the nick site for initiation of virion-sense DNA replication (Laufs et al., 1995). The conserved region contains a very high GC percentage (70 %, not including the nonanucleotide) and is potentially highly structured (data not shown). Sequence variations in this region are located mainly in the stem portion of a hairpin structure surrounding the TAATATTAC sequence. All the DNAβ molecules contain an A-rich region (>56 %) between nucleotides ±760 and ±1000.

(71K): Fig. 1. Multiple alignment of nucleotide sequences of the conserved region of 18 DNAβ molecules associated with begomoviruses in China and other reported DNAβ molecules.

Sequence analysis identified several open reading frames (ORFs) with a predicted coding capacity above 4 kDa on the virion or complementary strands. However, the number, positions and sizes of all but one these ORFs varied substantially. The one consistent ORF was on the complementary strand and is here designated C1. The C1 ORF has similar start positions (between 195 and 209 nt) and end positions (between 544 and 570 nt) on all DNAβ molecules and potentially encodes a protein with 118 amino acids. Some isolates have an extra in-frame upstream ATG, giving an extra 8 (Y38β, Y64β, Y45β, Y36β, Y8β, Y10β, Y11β, Y88β and Y92β), 11 (Y47β) or 22 (Y87β and BYVMVβ) possible N-terminal amino acids. Considering that the C1 ORF is extremely conserved in position and length for all described DNAβ molecules, this would suggest that C1 encodes a functional protein. There is considerable sequence variation, but 18 amino acids (not including the start methionine) are absolutely conserved (data not shown). The C1-encoded proteins in Type 1 DNAβ molecules have 79100 % amino acid sequence identity with each other, while 92100 % amino acid sequence identity is found in Type 2 DNAβ molecules (Table 1). All 23 sequences fall into two main branches. The first contains Type 1 and Type 2 DNAβ molecules and AYVVβ, with more than 56 % amino acid sequence identity between them, and the second contains the Type 3 DNAβ molecule, CLCuMVβ, CLCuRVβ and BYVMVβ, with more than 46 % amino acid sequence identity between them. Only 2636 % amino acid sequence identity was found between the members of the two branches (Table 1).

Infectivity and symptoms induced by DNAβ
To determine if the DNAβ molecules associated with begomovirus isolates in China are also involved in symptom induction, a tandem repeat of Y10β was inserted into the binary vector pBinPLUS, and an infectious clone, pBinPLUS-2β, obtained. This plasmid was shown to be infectious in N. benthamiana, N. glutinosa and L. esculentum when co-inoculated by agro-inoculation with the previously constructed infectious clone pBinPLUS-1.7A, containing partial repeats of Y10 DNA-A. Symptoms in these plants were compared with those induced by inoculation of pBinPLUS-1.7A alone. All plants co-inoculated with DNA-A and DNAβ produced systemic symptoms of severe downward leaf curl, vein darkening, stunting and enations, which are identical to those observed in the field. No symptoms were obtained in N. glutinosa or L. esculentum while only very mild symptoms were observed in N. benthamiana when agro-inoculation took place with DNA-A alone (Fig. 2A). These results clearly show that the leaf curl disease induced by isolate Y10 in tobacco is a result of co-infection by DNA-A and DNAβ. It is possible that other DNAβ molecules associated with Chinese begomovirus isolates may have the same function, but this remains to be proven.

(97K): Fig. 2. (A) Symptoms associated with Y10 DNA-A alone and in the presence of wild Y10 DNAβ in N. benthamiana (a), L.esculentum (b) and N. glutinosa (c). (B) Symptoms associated with Y10 DNA-A alone and in the presence of Y10 DNAβ with C1 in-frame ATG mutation (DNAmβ) or wild DNAβ in N. benthamiana.

To determine if the C1 ORF is functional, an in-frame-ATG mutation (nt 564566) in the C1 ORF of Y10β was made and the infectious clone pBinPLUS-C1m2β was obtained. This was shown to be infectious in N. benthamiana when co-inoculated with pBinPLUS-1.7A, but the symptoms were much milder than those induced by DNA-A and DNAβ (Fig. 2B), indicating that the C1 ORF contributes to symptom production. The nucleotide sequence of DNAβ from infected N. benthamiana confirmed the presence of the corresponding C1 in-frame ATG mutation and therefore that the severe phenotype on N. benthamiana was induced by the C1 ORF of DNAβ.

Co-evolution of DNAβ molecules with their helper virus genomes
Complete and partial DNA-A sequences of several Chinese begomovirus isolates are available. Comparison of these sequences shows that all the DNA-A sequences associated with Type 1 DNAβs are related to the recently reported TYLCCNV, with 8895 % nucleotide sequence identity (Yin et al., 2001). Consequently, these virus isolates are named as TYLCCNV isolates (X. Zhou and others, unpublished results). All Type 2 DNAβs are associated with TbCSV DNA-A molecules (Xie et al., 2002), whereas Type 3 DNAβ is associated with a new begomovirus species, for which the name Malvastrum yellow vein virus (MYVV) is proposed (X. Zhou and others, unpublished results).

A phylogenetic tree of DNAβ sequences associated with 18 Chinese isolates of begomoviruses AYVV, CLCuMV, CLCuRV and BYVMV was constructed and compared with that of the complete nucleotide sequences of their cognate DNA-A molecules (Fig. 3A). From Fig. 3(A), we can recognize the three main DNAβ Types associated with the three Chinese begomoviruses described above. Type 1 molecules can form a further four separate branches, one branch containing six isolates (Y8, Y10, Y36, Y38, Y45, Y64) originally from Honghe district, the second branch consisting of two isolates (Y5, Y43) from Dali district, the third branch containing one isolate (Y25) from Chuxiong district and the fourth branch containing four isolates (Y11, Y87, Y88, Y92) from Baoshan district. Type 2 molecules include four isolates from Baoshan district, while Type 3 contains one isolate from Honghe district (Y47), which also clusters with CLCuMVβ and CLCuRVβ. In addition, comparison shows that the variability of DNAβ sequences is highly related to the variability of DNA-A sequences, and clustering of DNA-A sequences corresponds to the clustering of DNAβ sequences (Fig. 3A). The linear correlation (R²=0·8772) between pairwise nucleotide sequence identities of DNA-A and DNAβ among these Chinese isolates clearly demonstrates that DNAβ molecules have co-evolved with their cognate DNA-A molecules (Fig. 4A).

(38K): Fig. 3. (A) Phylogenetic trees of complete nucleotide sequences of 18 DNAβ molecules associated with begomoviruses in China and other reported DNAβ molecules and of complete DNA-A sequences of their cognate virus isolates. (B) Phylogenetic trees of putative amino acid sequences of the C1 ORF of 18 DNAβ molecules associated with begomoviruses in China and other reported DNAβ molecules and of complete nucleotide sequences of their DNAβ molecules. The trees were generated using the MegAlign program available with the DNAStar package. Vertical distances are arbitrary and the scale below the trees measures the distance between sequences. The trees are unrooted, but a random sequence has been used and served as the root for bootstrap analysis. A bootstrap analysis with 1000 replicates was performed using PAUP v3.1.1 and the bootstrap percent values more than 50 are numbered along branches.

(26K): Fig. 4. (A) Correlation curve of DNA-A and DNAβ molecular variability expressed as pairwise sequence identities of each type of molecule. The viruses used for this curve are TYLCCNV-Y36, Y38, Y64, Y10, Y5, Y25 and Y11, TbCSV-Y2 and Y35, MYVV-Y47, AYVV, CLCuMV, CLCuRV and BYVMV. (B) Correlation curve of C1-encoded 118 amino acid sequences and DNAβ molecular variability expressed as pairwise sequence identities of each type of molecule. All the 18 DNAβ molecules associated with begomoviruses in China and five other reported DNAβ molecules were used for comparison. The curves are made based on pairwise sequence comparison percentages obtained from multiple alignments done with the Clustal analysis of the MegAlign 3.11 software (DNAStar Package). The best regression curve obtained was a linear regression for the A correlation (coefficient of R2=0·8772) and a polynomial regression for the B correlation (coefficient of R2=0·9668).

Generally, DNAβ molecules of the same virus species show 7299 % sequence identity while less than 57 % sequence identity was found between DNAβ molecules corresponding to different virus species (Table 1 and Fig. 3A). However, CLCuMVβ and CLCuRVβ have more than 91 % sequence identity, despite their DNA-A molecules sharing only 84 % identity and being considered as different species. Close inspection of the CLCuRVβ sequence shows that it can be divided into two parts. Part A includes 184 nt between 10651269, which is considerably diverged from CLCuMVβ with 62 % identity, whereas the remainder (part B, including the conserved region) is extremely similar to that of CLCuMVβ with 98 % identity. These data are strong evidence that either CLCuRVβ or CLCuMVβ has arisen by recombination. MYVVβ-Y47 also has a relatively high sequence identity with CLCuMVβ (67 %), indicating that these DNAβ molecules may have a common ancestor and evolved afterwards.

A phylogenetic tree based on alignment of C1-encoded 118 amino acid sequences was also constructed and compared with that of the sequences of their DNAβ molecules. Comparisons show that the variability of C1-encoded 118 amino acid sequences is extremely high related to the variability of DNAβ sequences, and clustering of C1-encoded 118 amino acid sequences corresponds to the clustering of DNAβ sequences (Fig. 3B). The polynomial correlation (R²=0·9668) between pairwise amino acid identities of C1-encoded 118 amino acid and nucleotide sequence identities of DNAβ among these Chinese isolates clearly demonstrates that the C1 ORF has evolved similarly as DNAβ molecules (Fig. 4B).

To date, DNAβ components have only been found associated with monopartite begomoviruses such as AYVV (Saunders et al., 2000) and CLCuMV (Briddon et al., 2001). By using abutting primers beta01 and beta02, we have demonstrated that DNAβ is widely associated with begomoviruses from tobacco, tomato, Malvastrum coromandelianum and Siegesbeckia orientalis in China. A DNAβ molecule was not detected in only one isolate (Y3) of TbLCYNV and another isolate (Y23) of SLCYNV. The existence of DNAβ molecules and failure to find DNA-B components indicates that many begomoviruses in Yunnan, China are monopartite begomoviruses. Agro-infection has also shown that DNA-A alone of TYLCCNV from Guangxi can systemically infect tobacco and induce severe yellow leaf curl disease symptoms (Yin et al., 2001).

The complete nucleotide sequences of 18 DNAβ molecules were determined. Comparison showed that three Types of DNAβ molecule were present among the isolates and that they are associated with the three begomovirus species TYLCCNV, TbCSV and MYVV. Within one Type, sequence variation was geographically related, and DNAβ molecules associated with virus isolates from the same region are clustered closely, while virus isolates from different regions were more distantly related (Fig. 4A). Geographically related antigenic and molecular variation in begomoviruses has been well-documented (Harrison & Robinson, 1999; Padidam et al., 1995), but this is the first such report for DNAβ molecules. Nucleotide sequence variation among DNA-B components of begomoviruses is greater than that of their associated DNA-A components (Harrison & Robinson, 1999). Sequence comparisons carried out in this study also show that sequence variation of DNAβ is greater than that of DNA-A molecules (Table 1). The data also indicated that the DNAβ molecules reported here are unrelated to other known geminivirus or nanovirus components, and that their evolutionary origin remains to be determined. DNAβ of Type 1 and Type 2 was found in different years and Type 1 molecules were found in different crops. We conclude, therefore, that several distinct forms of DNAβ occur in Yunnan, China, and have reached some genetic stability. It will be interesting to see if TYLCCNV originating from Guangxi province (Yin et al., 2001) also contains a DNAβ.

All the DNAβ molecules isolated in this study were found to be approximately half the length of their associated DNA-A molecules. Comparison among these DNAβ molecules, together with those reported and deposited in GenBank, allowed their structural features and their relationships to be explored in more depth. Overall nucleotide sequence of different kinds of DNAβ molecules share less identity than their cognate DNA-A molecules. However, all the DNAβ molecules possess a very conserved 115 nt region (Fig. 1). This region should have important function for DNAβ trans-replication and gene expression or for interaction with their cognate DNA-As. It is most important, therefore, to determine what functions this region may have, and how it interacts with virus- or host-encoded proteins. The two genomic molecules (DNA-A and DNA-B) of bipartite begomoviruses share a common region of approximately 200 nt including the nonanucleotide motif (TAATATT/AC), which contains cis-acting elements for replication and gene expression. Unlike DNAβ, this region is particularly prone to variation among Begomovirus species, and no conserved region similar to that in DNAβ is found among them. Different begomovirus DNA-A sequences possess different iteron sequences in their intergenic region, which facilitates sequence-specific Rep binding to initiate rolling circle replication (Fontes et al., 1994; Chatterji et al., 1999, 2000). Available evidence indicates that DNAβs do not contain iteron sequences, and yet they still depend on DNA-A for their replication (Saunders et al., 2000; Briddon et al., 2001). The DNA-A components of the bipartite begomoviruses African cassava mosaic virus and Indian cassava mosaic virus are able to trans-replicate AYVVβ in N. benthamiana, but were unable to functionally interact with AYVVβ to produce a symptomatic systemic infection (Saunders et al., 2002), indicating that there is less replication specificity for DNA-A-mediated replication and that many species of Begomovirus could potentially replicate many kinds of DNAβ. It is important to know how DNAβs are recognized by the DNA-A-encoded Rep proteins and how they control DNA-A-mediated replication.

A-rich regions were found in all DNAβ molecules, but their positions and arrangements varied. It is proposed that such variation may have originated from sequence duplications in order to satisfy size requirements for encapsidation (Saunders et al., 2000). The fact that DNAβ and defective DNAs associated with geminiviruses have a size maintained at approximately half that of the genomic components suggests a stringent size selection for encapsidation, and that smaller DNA molecules have adapted their size to allow encapsidation within geminate particles (Zhou et al., 2001b).

Like AYVVβ and CLCuMVβ, Y10β has been shown to be indispensable for the induction of typical disease. It would be informative to determine if the three kinds of DNAβ, corresponding to different Types of DNAβ, could be trans-replicated by different species of DNA-A, and/or if their interaction could extend host range and/or change symptoms. It has already been demonstrated that the interaction between DNA-A of Sri Lankan cassava mosaic virus (SLCMV), a bipartite begomovirus, and AYVVβ can induce severe stunting, leaf curl and chlorotic symptoms in N. glutinosa and extend the host range of SLCMV DNA-A to include A. conyzoides with the symptoms resembling those associated with AYVV (Saunders et al., 2002).

The position and size of the C1 ORF are conserved in all 23 DNAβ molecules. C1 of CLCuMVβ and AYVVβ were predicted by TESTCODE to be functional for DNAβ molecules (Briddon et al., 2001). A natural recombinant associated with AYVV, encompassing nt 1977 of AYVVβ and including C1 ORF, was sufficient to induce yellow vein symptoms in A. conyzoides (Saunders et al., 2001). The correlation between pairwise amino acid identities of C1-encoded 118 amino acid sequences and nucleotide sequence identities of DNAβ implies that C1 is important for the function of DNAβ molecules. Our preliminary results show that mutation of the start codon of C1 of Y10β results in very mild symptoms on N. benthamiana. The very mild symptoms associated with this mutant could be attributed to leaky expression of the C1 gene translated from the upstream or/and downstream start codons. Further expression of C1 in tobacco plants and protoplasts is being carried out in order to elucidate the function of this ORF in pathogenicity.

We demonstrated here that there is a high correlation between the variability of the DNAβ molecules and their cognate DNA-A components. This has two implications. Firstly, there has been a co-evolution of these two types of molecule to become functionally dependent on each other in the same host, and secondly, we could apply to DNAβ molecules the species concept that has been applied to their cognate geminiviruses. Comparison of DNA-A sequences shows that begomoviruses are grouped into three clusters, the stain/isolate cluster with 89100 % sequence identity, the species cluster with 6089 % sequence identity when viruses belong to the same geographical region and a species cluster with 4259 % sequence identity when viruses originate from different parts of the world (Fauquet, 2002). Thirteen DNAβ molecules are now known to be associated with TYLCCNV, four with TbCSV and two with CLCuMV, respectively. DNAβ molecules of the same virus species share 7299 % sequence identity, while usually 3657 % sequence identity was found between DNAβ molecules belonging to different virus species, with the exceptions of MYVV-Y47β (6267 %) and the recombinant CLCuRVβ. We here propose that DNAβ sequence identity could be used as a taxonomic criterion for species demarcation of DNAβ molecules and that the nomenclature to be used will be the name of the virus species followed by the Greek letter β: for example, TYLCCNVβ, TCSVβ, MYVVβ, etc. For different isolates of the same virus species, it will be the name of the DNAβ species followed by a dash and the isolate name: for example, TbCSVβ-Y2, TbCSVβ-Y35, etc.

Begomoviruses are highly recombinogenic with interspecific recombination events and recombination within members of other genera and families having been reported (Briddon et al., 1996; Zhou et al., 1997; Padidam et al., 1999; Saunders & Stanley, 1999). A natural recombination event between AYVV DNA-A and AYVVβ has been reported to produce a viable DNAβ recombinant (Saunders et al., 2001). In addition, either CLCuRVβ or CLCuMVβ has apparently arisen by recombination between another unidentified DNAβ molecule and CLCuMVβ or CLCuRVβ. This is the first example of such an event having occurred between DNAβ molecules, indicating that recombination is a powerful factor in evolution of DNAβs.

DNAβ molecules depend on the helper geminivirus for replication and for transmission via trans-encapsidation in geminivirus particles. Begomoviruses are known to cause disease in plants either as a monopartite or a bipartite virus. However, there is also a new category of monopartite begomoviruses associated with DNAβ. We hypothesize that those kinds of monopartite begomoviruses are capable of a low level of replication but do not induce symptoms unless a satellite DNAβ molecule is also present. The DNAβ permits the high levels of DNA-A accumulation required to induce disease and subsequently, whitefly transmission. The vectors then transmit both molecules maintaining a mechanism for co-evolution. The fact that the C1 ORF has the same degree of variability as the entire DNAβ molecule and that DNAβ molecules and DNA-A components are known to have a similar evolution implies that they were subjected over time to the same types of pressure, including mutation and recombination.

This research work was supported by National Outstanding Youth Foundation (Grant No.30125032) and the Teaching and Research Award Program for Outstanding Young Teachers in Higher Education Institutions of MOE. Yan Xie and Xiaorong Tao contributed equally to this paper.

Footnotes

The GenBank accession numbers of the sequences reported in this paper are AJ42031315, AJ42148285, AJ42161923, AJ45781822 and AJ506791.