Distribution of the phenotypic effects of random homologous recombination between two virus species

Abstract

Recombination has an evident impact on virus evolution and emergence of new pathotypes, and has generated an immense literature. However, the distribution of phenotypic effects caused by genome-wide random homologous recombination has never been formally investigated. Previous data on the subject have promoted the implicit view that most viral recombinant genomes are likely to be deleterious or lethal if the nucleotide identity of parental sequences is below 90%. We decided to challenge this view by creating a bank of near-random recombinants between two viral species of the genus Begomovirus (Family Geminiviridae) exhibiting 82% nucleotide identity, and by testing infectivity and in planta accumulation of recombinant clones randomly extracted from this bank. The bank was created by DNA-shuffling-a technology initially applied to the random shuffling of individual genes, and here implemented for the first time to shuffle full-length viral genomes. Together with our previously described system allowing the direct cloning of full-length infectious geminivirus genomes, it provided a unique opportunity to generate hundreds of "mosaic" virus genomes, directly testable for infectivity. A subset of 47 randomly chosen recombinants was sequenced, individually inoculated into tomato plants, and compared with the parental viruses. Surprisingly, our results showed that all recombinants were infectious and accumulated at levels comparable or intermediate to that of the parental clones. This indicates that, in our experimental system, despite the fact that the parental genomes differ by nearly 20%, lethal and/or large deleterious effects of recombination are very rare, in striking contrast to the common view that has emerged from previous studies published on other viruses.

Figure 1. Schematic representation of 47 recombinant viral genomes generated by L-DNA-shuffling with the parental genomes of Tomato yellow leaf curl virus (Tyx) and Tomato leaf curl Mayotte virus (Tox).

For convenience, each circular genome is presented in linear form beginning with the cloning site XhoI. (A) The 6 ORFs encoded by both parental genomes are presented above the aligned genomes: V2 [movement protein (MP)], V1 [coat protein (CP)], C3 [replication enhancer protein (Ren)], C2 [transcription associated protein (TrAP)], C1 [replication associated protein (Rep)], and C4 (symptom determinant implicated in cell cycle control). The percentage of amino acid identities between Tyx and Tox is indicated within parentheses for each ORF. (B) Recombinant genomes are represented in red for Tyx-derived fragments and in blue for Tox-derived fragments. Nucleotide positions are indicated above the aligned genomes.

Figure 2. Infectivity of the 47 recombinants and their 2 parental genomes.

Infectivity was determined on plant samples collected at 15 (A) and 22 days (B) post-inoculation. Infectivity was defined as the proportion of infected plants (virus detected in systemic leaves) of the total number of virus-inoculated plants. Within the boxes, the horizontal line indicates the median value (50% quantile), the box itself delimits the 25% and 75% quantiles, and the dashed lines represent the normal range of the values. The numbers on the top line indicate the number of infected plants for each viral clone. The blue box at the left end corresponds to the parental genome Tox and the red box at the right end to the parental genome Tyx. The recombinant genomes are ordered from left to right by increasing nucleotide identity with Tyx genome. White boxes correspond to recombinants that are not significantly different from either parent, light red boxes to recombinants that are significantly less infectious than Tyx, and light blue boxes to recombinants that are significantly more infectious than Tox.

Figure 3. Virus accumulation within infected plants for the 47 recombinants and their two parental genomes.

Virus accumulation was measured on plant samples collected at 15 (A) and 22 days (B) post-inoculation. Viral DNA was quantified with real-time PCR. The logarithm of the Calibrated Normalized Relative Quantity (logNRQ) reflects virus accumulation. Within the boxes, the horizontal line indicates the median value (50% quantile), the box itself delimits the 25% and 75% quantiles, and the dashed lines represent the normal range of the values; the points above and/or below correspond to outlying values. The blue box at the left end corresponds to the parental genome Tox and the red box at the right end to the parental genome Tyx. The recombinant genomes are ordered from left to right by increasing nucleotide identity with Tyx genome. White boxes correspond to recombinants that are not significantly different from either parent, light red boxes to recombinants that are significantly less infectious than Tyx, and light blue boxes to recombinants that are significantly more infectious than Tox.

Figure 4. Distribution of the phenotypic effects of random recombination.

The effect on the phenotype was assessed through infectivity (A) and virus accumulation (B), at 22 days post inoculation. The estimated effects correspond to the coefficients of the recombinant clones in the corresponding (generalized) linear model. Parental genome coefficients are just superimposed as vertical lines: Tox (blue) and Tyx (red). In each bin, the number of recombinants that differ significantly from Tyx and Tox are represented in light red and light blue, respectively.