The family Geminiviridae comprises viruses with a circular single-stranded DNA genome (2.6–5.2 kb) encapsulated in geminated quasi-icosahedral particles [1,2]. This family is divided into 14 genera based on their genome organization, type of vector, phylogenetic relationship, host range, and nucleotide sequence identity . The genus Begomovirus includes more than 400 species currently recognized by the International Committee on the Taxonomy of Viruses (ICTV, http://www.ictvonline.org/ virusTaxonomy.asp), emphasizing its importance within the family . Begomoviruses have one or two genomic components known as DNA-A and DNA-B, are transmitted by whiteflies of the Bemisia tabaci complex [4,5], and infect a wide range of cultivated and non-cultivated hosts, mainly in tropical and subtropical regions .
Several species of begomoviruses have been reported in non-cultivated plants in the family Euphorbiaceae [7,8,9,10,11,12,13]. In Cnidoscolus urens, a wild host largely found in northeastern Brazil, only two viruses were reported: Cnidoscolus mosaic leaf deformation virus (CnMLDV)  and Cnidoscolus blistering yellow mosaic virus (CnBYMV) [not published]. These reports reinforce the role of non-cultivated plants as viral reservoirs. In this study, the molecular characterization of a new bipartite begomovirus infecting Cnidoscolus urens in Pernambuco, Brazil was described.
A sample of C. urens, showing mild mosaic and leaf deformation symptoms, was collected in 2019 in Cabo de Santo Agostinho, Pernambuco State, Brazil (Figure 1). Total DNA was extracted using the CTAB method , and begomovirus infection was confirmed via PCR using the degenerate primers PAR1c496/PAL1v1978 . Complete viral genomes were amplified by rolling-circle amplification (RCA) , cleaved with restriction enzyme ClaI, ligated into the pBluescript KS+ (Stratagene) plasmid vector, and transformed into Escherichia coli DH10B by heat shock . Viral inserts were sequenced commercially by primer walking (Macrogen Inc., Seul, South Korea).
The complete DNA-A and DNA-B genomic components were assembled using CodonCode Aligner v. 6.0.2 (www.codoncode.com) and initially analyzed using the BLASTn algorithm  to determine the viral species that shared the highest nucleotide identity. Pairwise sequence identity comparisons were performed using Sequence Demarcation Tool software (SDT) v.1.2 .
Multiple sequence alignments were prepared using the MUSCLE algorithm implemented in MEGA6 . Phylogenetic analyses were performed on the CIPRES Science Gateway  using MrBayes v.3.3.3 . The best nucleotide substitution model (GTR+I+G) was used for the DNA-A and DNA-B component datasets. The analysis was carried out using two replicates with four chains each for 10 million generations and sampling every 1000 steps (a total of 10,000 trees). The first 2500 trees were discarded as a burn-in phase. Trees were viewed and edited using Figtree v.1.4 (ztreebio.ed.acuk/software/figtree).
A total of five sequences were obtained: three for the DNA-A component and two for the DNA-B component. The sequences ranged from 2657 to 2692 nt in length for DNA-A (GenBank accession MZ465527, MZ465586, and MZ465587) and 2622 nt in length for DNA-B (GenBank accession MZ465585 and MZ46526). They showed a typical New World begomovirus gene organization, with DNA-A encoding one protein (CP) in the virion-sense strand and four in the complementary-sense strand (Rep, TrAP, Ren, and AC4), while DNA-B encoded two proteins, NSP in the virion-sense strand and MP in the complementary-sense strand (Supplementary Table S1). Analysis of the common region (CR) showed a conserved nona-nucleotide (5'-TAATATT/AC-3') and identified three putative iterons (GGGT), one of which was inverted (Supplementary Figure S1). The CR (200 nt) showed 96% nucleotide sequence identity between the two components, suggesting that they are a cognate pair.
Using pairwise comparisons and established criteria for species demarcation within the genus Begomovirus , clones BR_Ca96_19A (GenBank accession MZ465527), BR_Ca112_19A (MZ465586), and BR_Ca113_19A (MZ465587) showed a nucleotide identity between them ranging from 98.4 to 100%. They shared 79.8% similarity with CnBYMV (MT553995) and 88.6–88.9% similarity with CnMLDV (NC_038982; Supplementary Figure S2), which are begomoviruses isolated from C. urens in the Piauí and Alagoas States, respectively . Therefore, these isolates represent a new begomovirus species, and the name Begomovirus caboniensis is proposed for this virus, adopting the binomial nomenclature system established by ICTV .
Bayesian inference phylogenetic trees were constructed from nucleotide sequences of the DNA-A and DNA-B components obtained in the present study, with the most related begomoviruses from cultivated and non-cultivated plants from South America (Figure 2). The analysis for the DNA-A component indicated that the B. caboniensis isolates cluster with CnMLDV (NC_038982) and CnBYMV (MT553995) isolates, both obtained from C. urens from the northeast region of Brazil. For the DNA-B component, the isolates cluster with the CnMLDV isolate (KT966772), thus reinforcing the phylogenetic relationship between the isolates obtained from C. urens.
Recombination analysis of DNA-A and DNA-B genomic components of the B. caboniensis isolates and the complete sequences from the database were conducted using RDP, GENECONV, BootScan, MaxChi, Chimera, SiScan, and 3Seq methods implemented in the Recombination Detection Program (RDP) v.4.0 . However, no evidence of recombination events was observed.
The occurrence of begomoviruses in C. urens has been reported for over ten years in Alagoas State . However, it was only in 2016 that the first complete genome of a begomovirus occurring naturally in this host was characterized, which was identified as CnMLDV . Recently, a new species, CnBYMV, was reported in C. urens in Piauí State. In the present study, a new bipartite begomovirus was characterized, demonstrating the potential of C. urens as a natural reservoir of begomoviruses.