Here, we report the formation of pseudorecombinants between two distinct, albeit closely related, begomoviruses. The reciprocal pseudorecombinants ToRMV-A + ToSRV-B and ToRMV-B + ToSRV-A were both viable. Their viability is most likely due to the high sequence identity of their CRs and Rep proteins, as well as the identical sequences of their iterons (GGTAG). The latent period and symptoms were the same in all combinations compared to the parental viruses, which is not surprising since the severity of symptoms is associated with the DNA-B, and these are almost identical between the two viruses.
When DNA-A and DNA-B components of both ToSRV and ToRMV were inoculated, mosaic symptoms identical to those induced by each virus alone appeared at 14 dpi (Figure
1 and Table
1). This suggests that synergism does not occur between these two viruses. On the contrary, ToRMV had a negative interference over ToSRV. In most cases in which three or four DNA components were inoculated, the ToRMV components were always detected, whereas the ToSRV components were rarely detected. Negative interference exerted by ToRMV in mixed infection with ToYSV has already been reported
. Early in infection ToRMV negatively affected the accumulation of ToYSV, but when the infection was established the negative interference ceased
. Interestingly, although ToRMV negatively affected ToSRV in our experiments, ToSRV behaved similarly to ToRMV in a single infection, both in terms of latent period and symptoms.
Considering that their DNA-B components are almost identical, and that their iterons are identical, we hypothesized that the preferential replication of ToRMV over ToSRV would be due to differences in their Rep proteins, most likely in their virus-specific recognition domains. Begomovirus iterons are composed of an invariable GG sequence followed by three nt (identified as N1, N2 and N3) that vary for each virus. The virus-specific recognition domain of the Rep protein maps to its N-terminal region
[40, 41] and includes the conserved motif 1 of rolling-circle replication-initiator proteins
. The IRD sequence is conserved among begomoviruses with identical iterons, but varies among species with different iteron sequences. Predicted nt-aa pairing would occur between N1 and the eighth aa of the IRD, N2 with the sixth IRD aa, and N3 with the first or third IRD aa, depending on the iteron sequence
Although ToRMV and ToSRV have identical iterons, their IRDs are different at positions 2 and 3. The change of lysine to arginine at IRD position 3, which pairs with N3, should not promote major changes in the protein, since the two amino acids have similar properties. Therefore, it is reasonable to assume that this difference is not responsible for the preferential replication of ToRMV over ToSRV. In fact, this change does not affect iteron recognition, since the two Rep proteins have the ability to replicate viral DNA. This is observed when any combination of a DNA-A and a DNA-B component (cognate or pseudorecombinant) is inoculated: both components are always detected (Table
Further analysis revealed small differences between ToRMV and ToSRV in the CR and in the Rep protein. In the CR, nine nucleotide changes occur among the four components, with ToRMV DNA components having the largest number of modifications compared to ToSRV. It is possible that these changes in the CR sequence, as well as changes in IRD aa 2 and 3, have enabled the ToRMV Rep protein to recognize the iteron sequence GGTAG with higher affinity compared to ToSRV. These same differences could cause the ToSRV Rep protein to replicate the DNA-A and -B of ToRMV preferentially. More likely is that the combination of minor changes in both Rep and CR may be involved in the preferential replication of ToRMV components. A detailed mutagenesis analysis of the IRD region and of the iterons of ToRMV and ToSRV will be required to confirm this hypothesis.
Interestingly, ToSRV is more prevalent in the field compared to ToRMV
[27, 35]. This could be due to differences in transmission by the insect vector, since the coat protein, which is considerably distinct between the two viruses, is the major viral determinant of insect transmission.
The detection of a recombination event involving the DNA-A of the two viruses was not surprising, since it is well established that begomoviruses are recombination-prone. The region transferred from ToSRV to ToRMV comprises practically the entire Rep ORF as well as part of the CR including the TATA box, G-box, the conserved nonanucleotide (TAATATTAC) and the iterons. This recombination event between ToSRV and ToRMV was detected also by Ribeiro et al.
, who proposed that ToRMV would be a recombinant between ToCMoV and ToSRV.
The specific genomic region involved in a recombination event is an important aspect in the maintenance and adaptation of the recombinant virus in the population. It is known that the begomovirus genome has a number of recombination hot spots, including the CR, the CP/Ren interface and the 3′ region of the Rep gene, while the remaining coding regions are recombination cold spots
[28, 43–45]. The breakpoints of the recombination event involving ToSRV and ToRMV map to two hotspots, the CR and to the 3′ region of the Rep gene, near the interface with the Trap gene. Recombination in coding regions probably occurs due to the conflict between the machineries of transcription and replication
[45, 46]. Lefeuvre et al.
 suggest that selection apparently disfavors recombinants with breakpoints in coding regions, and that the probability of such a recombinant to remain in the population is dependent on the specific position of the breakpoint: the peripheries of coding regions tend to be more favorable to recombinantion than central regions. However, recent experimental evidence demonstrated that the coding regions of the begomoviruses Tomato yellow leaf curl virus-Mld (TYLCV) and Tomato leaf curl Mayotte virus (ToLCYTV) are recombination prone, and that all recombinants between TYLCV and ToLCYTV generated using L-DNA-shuffling technology displayed infectivity and viral accumulation similar to the parental begomoviruses
Recombination events involving DNA-A components are considered a major source of molecular variation for begomoviruses and may result in a gain of virulence
[49, 50]. However, predicting the epidemiological outcome of a recombination event is not a straightforward task. For example, van der Walt et al.
 demonstrated that infection by a recombinant between different MSV isolates led to increased symptom severity in maize, indicating greater adaptation of the recombinant. On the other hand, Davino et al.
 demonstrated that a recombinant between TYLCV and TYLCSV coexists with the parental viruses and has a lower replicative capacity, a fact that might help to explain why plants infected only by the recombinant have not been found. In the case of the recombinant ToRMV, there was an adaptive advantage over the parental ToSRV reflected in the preferential replication of ToRMV over ToSRV, nevertheless ToSRV is more prevalent in the field compared to ToRMV. We suggest that ToSRV is maintained in nature by infecting plants in the absence of ToRMV. If the occurrence of mixed infection between ToRMV and ToSRV was frequent, ToRMV would be the prevalent virus.
Another interesting aspect is that the two DNA-B components are almost identical (98.2% nt identity). This supports the hypothesis that, following the transfer of practically the entire Rep ORF as well as part of the CR from ToSRV to ToRMV, ToRMV captured the DNA-B of ToSRV. According to the current ICTV-supported criteria for species demarcation, a begomovirus is considered to represent a new species when its DNA-A sequence is less than 89% identical to that of other begomoviruses
. ToSRV and ToRMV DNA-A components have nucleotide sequence identity of 86.3%, which clearly separates them as members of different species based on this criterion. However, these two viruses are very similar in terms of biological and molecular properties, including: (i) high sequence identity of their CRs and their Rep coding regions; (ii) nearly identical Rep binding sites in their CRs, with identical iterons; (iii) the same DNA-B; and (iv) induction of nearly identical symptoms in tomato, with the same latent period. We thus propose that ToRMV and ToSRV should be considered to be strains of the same viral species, named ToRMV since this virus was described first.
The fact that begomovirus taxonomy ignores the DNA-B was addressed by Briddon et al.
, who performed a hypothetical taxonomy based on the DNA-B sequence alone. The authors show that 85% of the sequences analyzed remain unaltered in their classification based on the DNA-A, providing good support for a taxonomy based only on the DNA-A. However, in specific cases such as the one presented in this work, the criterion based on the DNA-A could be less restrictive and/or the DNA-B could be considered in the analysis for species assignment.