Recurrent speciation of a tomato yellow leaf curl geminivirus in Portugal by recombination

Recurrent evolution can involve interspecific interactions, recognized to play a primary role in the diversification and organization of life. Both in the plant and animal kingdoms, the recurrent formation of allopolyploid species has been described. In the virosphere, recombination between isolates of different species has been shown to be a source of speciation. In this work, complete genome analysis showed that speciation through recombination of an emergent DNA plant virus, tomato yellow leaf curl Malaga virus (genus Begomovirus, family Geminiviridae), has occurred independently in Portugal and Spain, confirming previous observations with tomato yellow leaf curl Axarquia virus, also originated independently in Spain and Italy. These results will guide future research to discover new cases of recurrent emergence of recombinant virus species in geographical areas where the putative parents co-exist or can be introduced. This will reveal the role that recurrent speciation through recombination plays in the evolution of the virosphere and will help to understand the consequences of this phenomenon on the diversification of life.

Scientific RepoRts | (2019) 9:1332 | https://doi.org/10.1038/s41598-018-37971-z first reports of infections by TYLCVs were from the early 1990s, associated with the presence of the Spain (ES) strain of Tomato yellow leaf curl Sardinia virus (TYLCSV) 22 . Later, isolates of the Mild (Mld) and type [aka, and henceforth, Israel (IL)] strains of TYLCV were introduced in the country [23][24][25] . A few years after the detection of TYLCV, two novel recombinant viruses emerged: tomato yellow leaf curl Malaga virus (TYLCMaV) (as a result of a genetic exchange between isolates of the ES strain of TYLCSV and of the Mld strain of TYLCV) 9, 26 and tomato yellow leaf curl Axarquia virus (TYLCAxV) (as a result of a genetic exchange between isolates of the ES strain of TYLCSV and the IL strain of TYLCV) 7 . Both recombinant begomoviruses exhibited biological properties that suggested an ecological fitness higher than either parental begomovirus, including a wider host range, thus supporting the consideration of these recombinant viruses as bona fide species 14 . Recombinant TYLCVs have also been detected associated with mixed infections of the parental viruses in Italy, Jordan and Morocco 16,[27][28][29] . TYLCMaV from Morocco and TYLCAxV from Jordan have been probably introduced from Spain and Italy, respectively. However, the TYLCAxV isolates from Italy have been originated by recombination between Italian TYLCV and TYLCSV isolates 16 . Also, sequencing of the intergenic region showed the presence of recombinant begomoviruses, involving TYLCV and TYLCSV as parents, in tomato and S. nigrum plants in Tunisia 30 , although a correct taxonomic adscription of them awaits the availability of complete viral sequences. The TYLCMaV and TYLCAxV isolates characterized in the Mediterranean basin and close regions share a number of features: i) contain two recombinant fragments of similar length originated from each parental begomovirus (TYLCV and TYLCSV), ii) one of the recombination break points is close to the conserved stem-loop contained in the intergenic region and iii) the presence of a moiety containing the virion-sense genes originated from TYLCSV and a moiety containing the complementary-sense genes originated from TYLCV. Geographical distribution of TYLCV-TYLCSV recombinant begomoviruses is shown in Fig. 2.
In Portugal, although both strains of TYLCV, IL and Mld, seem to be present 31,32 , only one complete genome sequence of TYLCV-Mld was available when this work was initiated. The presence of TYLCSV in Portugal,  however, has not been described to date. Unpublished data available in our laboratory suggested the presence of mixed infections of TYLCV and TYLCSV and/or recombinants originating from them in southern Portugal (Faro district, formerly Algarve province). This was based on molecular hybridization with specific probes of DNA extracts of some symptomatic tomato plants. In this work, we revisited our preliminary results and extended the analysis to a high number of tomato samples as well as to symptomatic samples of S. nigrum and other solanaceous weed species frequently associated with tomato crops, confirming the presence of the recombinant begomoviruses TYLCMaV and TYLCAxV in Portugal. Genealogical analysis suggested that TYLCAxV isolates from Portugal have been introduced from Spain. In contrast, the Portuguese TYLCMaV isolates have most probably originated independently from the Spanish isolates, thus providing strong evidence for the recurrent generation and establishment of this begomovirus species in different areas and periods of time.

Results and Discussion
TYLCV and TYLCSV are widely spread in southern Portugal. Two hundred sixty-three out of 350 tomato (75.1%) and 10 out of 28 weed (35.7%) samples were shown to be infected by begomoviruses belonging to the tomato yellow leaf curl virus complex after dot-blot molecular hybridization with probes specific for TYLCSV and TYLCV, respectively (Table 1, Supplementary Tables S1 and S2). Two hundred thirty-three tomato samples were positive for only one probe (16 with TYLCSV and 217 with TYLCV). Thirty tomato samples were positive with both probes. Nine out of 19 S. nigrum samples were positive for the TYLCV probe, two of them were also positive for the TYLCSV probe. The only D. stramonium sample analyzed was positive for the TYLCV probe. The existence of samples that hybridized both probes is consistent with mixed infections and/or with the presence of recombinant begomoviruses containing a recombination breakpoint within the intergenic region contained in the probes 7,9,33 . PCR using primers specific for TYLCMaV and TYLCAxV was carried out with the 32 samples positive for both probes in the hybridization analysis. PCR using TYLCMaV specific primers (i.e. forward primer specific for TYLCV-Mld and reverse primer for TYLCSV-ES) showed amplicons from two tomato samples [150301/19-2 (#2) and 150301/19-6 (#6)]. The TYLCAxV specific primers (i.e. forward primer specific for TYLCV-IL and reverse primer for TYLCSV-ES) showed amplification in one S. nigrum sample [300902/38 (#38)]. Digestion of the rolling-circle amplification (RCA) product with BamHI yielded two fragments of approximately 2.8 kbp with barely distinguishable electrophoretic mobilities in each tomato sample and one fragment in the S. nigrum sample, also of approximately 2.8 kbp. These fragments, each putatively corresponding to a full-length monopartite begomovirus genome, were cloned and sequenced. Nucleotide sequence analysis of two clones from tomato samples #2 (2776 nt, Genbank accession number JN859134) and #6 (2776 nt, JN859136) confirmed the presence of a monopartite begomovirus in both samples, showing the highest nucleotide identities (98.8% and 99.2%, respectively) with a TYLCSV-ES isolate from Spain (AJ519675). The two clones showed a nucleotide identity of 99.5% between them. Other clones from samples #2 (2782 nt, JN859135) and #6 (2782 nt, JN859137) showed the highest nucleotide identities (97.1% and 98.5%, respectively) with a TYLCMaV isolate from Spain (AF271234) 9 . Both clones showed a nucleotide identity of 97.3% between them. S. nigrum sample #38 was also infected by a begomovirus (2772 nt, JN859138) that showed the highest nucleotide identities with TYLCAxV isolates infecting S. nigrum in Spain (99.4%, AY227892) 7 and tomato in Italy (92.4%, EU734831) 16 . In accordance with begomovirus species demarcation and nomenclature 14  Recombination analysis of the TYLCMaV and TYLCAxV isolates from Portugal obtained in this work confirmed their recombinant nature, as suggested by the high nucleotide identity showed with isolates of those begomoviruses from Spain that are known to be recombinants 7,9 . Isolate TYLCMaV-[PT-Alg2-01] showed a recombination event involving TYLCV-Mld (AF105975) as major parent and TYLCSV-ES[PT-Alg2-01] as minor parent (Fig. 1, Table 2). Isolate TYLCMaV-[PT-Alg6-01] showed a recombination event involving TYLCV-Mld (AF105975) as major parent and TYLCSV-ES[PT-Alg6-01] as minor parent. However, TYLCAxV from Portugal showed a recombination event involving two isolates from Spain: TYLCSV-ES (Z25751) as major parent and TYLCV-IL (AJ489258) as minor parent (Fig. 1, Table 2). One of the recombination breakpoints of all three isolates was located in the stem-loop sequence conserved in the IR. It is known that this stem-loop is a hot-spot for recombination in begomoviruses and other geminiviruses 8,34 .
To our knowledge, this is the first report confirming the presence of TYLCSV, TYLCMaV and TYLCAxV in Portugal, where TYLCV was the only begomovirus associated with TYLCD to date 31,32 .
TYLCMaV, but not TYLCAxV, originated in Portugal. TYLCMaV is a species arisen from recombination between TYLCSV-ES and TYLCV-Mld ( Table 2, Supplementary Fig. S1). Despite TYLCMaV sequences from Portugal having high nucleotide identity with the Spanish isolate, the moiety of the TYLCMaV genomes from Portugal originated from TYLCSV were more similar to TYLCSV-ES isolates from Portugal (0-4 nt substitutions) than to TYLCSV-ES/TYLCMaV isolates from Spain (4-13 nt substitutions), as was shown by the genealogical network analysis (Fig. 3a). Also, the moiety of the TYLCMaV genomes from Portugal originated from TYLCV-Mld were more similar to TYLCV-Mld isolates from Portugal (6-10 nt substitutions) than to TYLCV-Mld/TYLCMaV isolates from Spain (11-16 nt substitutions) (Fig. 3b). This strongly suggests that the TYLCMaV isolates characterized in this work were not introduced from Spain but arose in Portugal as a result of an independent recombination event. In contrast, comparison of the moiety of the Portuguese TYLCAxV isolate (arisen from recombination between TYLCSV-ES and TYLCV-IL; Table 2, Supplementary Fig. S1) that originated from TYLCSV-ES showed that was more similar to TYLCSV-ES/TYLCAxV isolates from Spain (7-10 nt substitutions) than to TYLCSV-ES isolates from Portugal (13-15 nt substitutions) (Fig. 3c), thus suggesting that TYLCAxV most probably emerged in Spain and moved to Portugal, where it was also found infecting S. nigrum, a wild plant widely distributed in the Mediterranean basin that has been suggested to be an optimal niche for the development of recombinant begomoviruses 7 . This hypothesis was reinforced by the fact that TYLCV-IL, one of the parents of TYLCAxV, has never been reported in the Faro district in southern Portugal. The only report of TYLCV-IL in this country was based on a partial sequence from Portalegre, a district in central Portugal 31 . The results of the genealogical network analysis were confirmed by phylogenetical analysis (Fig. 4). Thus, both moieties of the Portuguese TYLCMaV isolates were clustered with the putative parental begomovirus isolates from Portugal (TYLCSV-ES, TYLCV-Mld); in contrast, the Portuguese TYLCAxV isolate clustered with the TYLCAxV isolate from Spain.
Sequence comparisons and phylogenetic analysis have shown that Italian TYLCAxV isolates [classified within Sicily1 (Sic1) or Sicily2 (Sic2) strains; Supplementary Fig. S1, Table 3] originated from parental viruses present in Italy, including the Sicily (Sic) strain of TYLCSV 16,27 , thus providing additional evidence for recurrent speciation in the TYLCV complex. In contrast, the TYLCAxV isolates from Jordania 28 and the TYLCMaV isolates from Morocco 29 were very probably introduced from Italy and Spain, respectively ( Supplementary Fig. S1).
The emergence of TYLCMaV isolates in Portugal as the result of recombination between the parental begomoviruses TYLCV-Mld and TYLCSV, a phenomenon previously described in Spain 9 , suggests that wherever the parental viruses have contact through mixed infections, the same recombinant virus will probably be generated. Both TYLCMaV and TYLCAxV isolates generated in Spain exhibited higher ecological fitness than either parental virus, including a wider host range 7,9 . In this regard, Spanish TYLCMaV isolates, that were initially detected in tomato plants, were found later in common bean crops where they displaced the parental begomoviruses 9 . This behavior highlights the relevance of the recurrent generation of recombinant begomoviruses causing the TYLCD in new geographical areas and agrosystems.
The results obtained in this work will guide future research to discover new cases of recurrent appearance and emergence of recombinant virus species in geographical areas where the putative parents co-exist or can be introduced, which is highly probable not only in the case of geminiviruses but in other plant, animal and human viruses having highly recombinogenic genomes. This will reveal the actual role that recurrent speciation through recombination plays in the evolution of the virosphere and will help to understand the consequences of this phenomenon on the diversification of life as a whole.

Materials and Methods
Plant material, DNA extraction and molecular hybridization. Tomato (n = 350, Supplementary  35 and kept at −20 °C until they were used for dot-blot hybridization. For that, 1 μL of DNA extract from each sample was spotted on a positively charged nylon membrane (Roche Diagnostics, Mannheim, Germany). Squash-blotted cross-sectioned leaf petioles of healthy and TYLCMaV-or TYLCAxV-infected tomato  PCR, rolling-circle amplification (RCA) and cloning. Plant samples that were positive for both probes were analyzed by PCR using the primers MA117 (5′-TAAGGAGCACTTAGGATATG-3′) and MA116 (5′-GTAGGGCCCACTACTTTATC-3′) specific for TYLCMaV 9 and MA250 (5′-GGTGTCCCTCAAAGC TCTATGGCAATCG-3′) and MA116 specific for TYLCAxV 7 . PCR reactions were carried out using BIOTAQ DNA polymerase (Bioline, London, UK). The amplification program started with an initial denaturing step at 94 °C for 2 min, followed by 30 cycles of 94 °C for 1 min, 50 °C for 1 min and 72 °C for 1 min, plus a last elongation     37 . Phylogenetic evidence of recombination was revealed by using the neighbor-net method in the program SPLITSTREE4 38 and statistically verified by using the pairwise homoplasy index (PHI) test 39 . The identification of potential recombinant fragments was carried out using the nine methods included in the RDP4 package with default settings (primary scan with RDP, GeneConv and MaxChi and secondary scan with BootScan and SiScan) 40 from the alignment generated by CLUSTAL X (v.2.07) 41 . The common sequences of the TYLCSV and TYLCV moieties identified in the recombinant begomoviruses were analyzed using statistical parsimony 42 with the program TCS (v.1.21) 43 . The resulting genealogical networks identify both the relationship between the different sequences as well as the number of nucleotide substitutions connecting them. Both begomovirus moieties were also submitted to phylogenetic analysis using maximum likelihood (ML) after selecting the best-fit model of nucleotide substitution based on the corrected Akaike information criterion (AICc) and Bayesian information criterion (BIC) as implemented in MEGA X 44 . Details of the begomoviruses belonging to the TYLCV complex analyzed, included those characterized in this work, are given in Table 3. Table 4. Primers used to generate DNA probes specific to the IR of TYLCSV and TYLCV. *Nucleotide coordinates considering as first nucleotide that after the nicking site within the conserved nonanucleotide at the origin of replication.