Leaves showing CMV-like symptoms were obtained from 16 locations across five states (Oyo, Kwara, Edo, Enugu, and Akwa Ibom) in Nigeria (Figure 1). Sampling was performed on four economically important fruit vegetables in Nigeria: pepper, tomato, African garden eggplant, and watermelon. Plants mainly showing chlorosis and mosaic patterns were sampled (Figures 2A–C), while leaves were stored in small plastic vials containing silica gel and cotton wool at ambient temperatures of 25 ± 3°C until laboratory analyses.

Total nucleic acids were extracted from the leaf samples using a modified CTAB procedure as previously described by Abarshi et al. (2010). Partial RNA1 and complete CP gene fragments from the CMV genome were amplified from the aqueous extracts using primer pairs CMV-F (5′-CCGCTTGTGCGTTTAATGGCT-3′)/CMV-R (5′-GCCGTAA GCTGGATGGACAA-3′) (Wylie et al., 1993) and CMV-CP-F (5′-ATGGACAAATCTGAATCAACC-3′)/CMV-CP-R (5′-TCAGA CTGGGAGCACCCC-3′) (Hu et al., 2016), respectively, via one-step reverse transcription polymerase chain reaction (RT-PCR) procedures. Assays were performed in total volumes of 25 μL cocktails containing 5× GoTaq^® Flexi buffer (Promega, Madison, WI, United States), 0.2 mM dNTPs, 1.5 mM MgCl₂, 10 μM primers, 0.3 U of Taq polymerase (New England Biolabs, United States), 0.3 U of MMLV reverse transcriptase (New England Biolabs, United States), and sterile distilled water to make the final volume. Cycling conditions were an initial denaturation at 92°C for 3 min followed by 35 cycles at 95°C for 1 min, annealing at 60°C (RNA1) and 53°C (CP) for 1 min and extension at 2°C for 1.5 min with final elongation at 72°C for 7 min. Amplicons were analyzed on a 1.5% agarose gel stained with GelRedSafe (Promega, Madison, WI, United States) at 100 V for 1 h.

Products from both RT-PCR amplifications were ethanol-purified from selected CMV-positive samples and bidirectionally sequenced at Inqaba Biotec Company (South Africa). Nucleotides from both strands per amplicon were assembled and manually edited using BioEdit v.7.0.5 (Hall, 1999). Consensus sequences were obtained for each isolate and amplicon, verified by the nucleotide option of the basic local alignment search tool (BLASTn) (Altschul et al., 1990) and submitted to the GenBank database. Sequences of CMV RNA1 and CP genes from representative strains were downloaded from the National Centre for Biotechnology Information¹ and used for subsequent analyses (Supplementary Table S1). Our sequences were compared with existing CMV isolates available at GenBank, while percentage similarities within the Nigerian population and those from the database were calculated using SDT v1.2 (Muhire et al., 2014) with pairwise gap deletions. Deduced proteins from both partial RNA1 and complete CP genes were also compared, and identities were obtained.

Multiple alignments of nucleic acid and deduced amino acid sequences for both gene sets were separately performed on all CMV isolates used in this study (Supplementary Table S1) using ClustalW (Thompson et al., 1994) in BioEdit. The RNA1 and CP genes from peanut stunt virus strain P (accession numbers EU570236 and EU570238) were obtained as outgroups and included in the phylogenetic analyses. The base substitution models used for each gene set were selected using the jModelTest program (Darriba et al., 2012), supported by the corrected Akaike information criterion (AICc) values. Subsequently, the Tamura 3-parameter model, including heterogeneity of substitution rates among sites modeled using a discrete gamma distribution (T92 + G), was implemented for the RNA1 dataset, while the Kimura 2-parameter model, with the gamma-invariable mixture model (K2 + G + I), was adopted for the CP gene sequences. Bayesian phylogenetic trees were obtained using MrBayes v3.2.7 (Ronquist et al., 2012), with each analysis obtained from Monte Carlo Markov chain (MCMC) parameters set at ten million generations and sampled every 1,000 generations. Two independent runs were set per analysis with the minimum probability set at 0.05. Prior to convergence, the relative burn-ins were implemented at 25% for diagnostics, and trees generated were discarded. Chains were heated to 0.10, and trees were viewed using FigTree v1.4.4 (Rambaut, 2019).

Searches were conducted for recombination signatures within the partial RNA1 and complete CP gene populations using genetic algorithm recombination detection (GARD) and single breakpoint scanning (SBP) indices (Kosakovsky-Pond et al., 2006) as implemented by Datamonkey software (Weaver et al., 2018) at http://www.datamonkey.org. Evidence for the occurrence of recombination was further examined with algorithms within RDP v 4.13 (Martin et al., 2015) using the default settings and a Bonferroni corrected P-value cut-off of 0.05. Signals were confirmed true when detected by at least four detection methods with strong significance levels (P ≤ 0.05), in addition to the occurrence of putative parents across different phylogenetic clusters in test trees.

The isolates obtained from this study, together with other Nigerian CMV RNA1 and CP gene sequences from the database (as of July 1, 2021), were used to infer population genetic structures of CMV in Nigeria. The estimated parameters comprised statistics such as nucleotide polymorphism (π), number of variable sites (S), haplotype diversity (Hd), number of haplotypes (h), number of mutations (Eta), statistic θ from S (θ-W) and the mean number of nucleotide differences between sequences (k). The deviation of the Nigerian CMV populations from neutrality for both gene datasets was determined via Tajima’s D, Fu and Li’s D, and Fu and Li’s F statistics (Tajima, 1989). Negative values indicate the occurrence of redundant low-frequency polymorphisms within the population prompted by population expansions, genetic hitchhiking or background selection, while positive values suggest nominal levels of low- and high-frequency polymorphisms that show balancing selection and/or a decrease in population size (Alabi et al., 2011). All genetic variability calculations and tests of neutrality parameters were obtained using DnaSP v5.10.01 software (Librado and Rozas, 2009).

The existence of selection pressures within the sequence populations was obtained using the single-likelihood ancestor counting (SLAC) method (Kosakovsky-Pond and Frost, 2005) as implemented in the HyPhy package (Kosakovsky-Pond et al., 2020) domiciled in Datamonkey software (Weaver et al., 2018). The status of selection pressures was obtained by calculating the ratio of the mean number of nucleotide differences between sequences per non-synonymous site (d_N) and the mean number of nucleotide differences between sequences per synonymous site (d_S). Selection pressure within the sequence populations was considered purifying when d_N/d_S < 1, neutral when d_N/d_S = 1 and diversifying when d_N/d_S > 1 (Yang et al., 2000).

A total of 106 samples were collected from 16 locations (Figure 1), out of which 24 were CMV-positive (Table 1). The leaf samples obtained from tomato, pepper and watermelon tested positive for CMV via RT-PCR, although infection was only obtained from eight locations in Oyo and Kwara states. Samples from Edo, Enugu and Akwa Ibom all tested negative for CMV, and there was no amplification from all African garden eggplant samples obtained across the areas sampled. Twelve positive isolates were selected for genomic sequencing of the partial CMV RNA1 segment, out of which 7 were further sequenced for the complete CP gene.

Sequencing of the partial RNA1 amplicons obtained from 12 RT-PCR-positive samples produced genomic fragments with nucleotides ranging from 506 to 543 bp and proteins with 144 amino acids (Table 2). These were deposited in the GenBank database under accession numbers MH798804-MH798810 and MW655573-MW655577. For the seven complete CP gene fragments, products ranged from 667 to 669 nucleotides with protein lengths between 217 and 218 aa (Table 2). The nucleotide sequences were deposited under the accession numbers OK107526-OK107532.

BLASTn searches of the partial RNA1 genes revealed 92.90–99.20% nucleotide identities and 98–100% query coverage with CMV sequences present in GenBank (Table 2). These isolates shared 82.94–99.80% nucleotide and 85.42–100% amino acid identities with each other (Table 3) and 73.17–97.79% nucleotide sequence similarities with other CMV isolates from all subgroups worldwide (Figure 3A). Similar BLASTn searches using the complete CP gene populations produced 94.90–96.92% sequence identities with GenBank isolates, showing 98–100% query coverage (Table 2). The partial RNA3 nucleotides and the translated CP were 95.79–97.90% and 98.62–100% similar (Table 4) and shared 75.72–95.56% nucleotide identities with complete CP genes from global CMV isolates (Figure 3B). The translated proteins from both gene sets showed peculiar properties when aligned with other CMV subgroup I isolates, with nine unique amino acid changes observed within the RNA1 population (Supplementary Figure S1) and four amino acid alterations within the CP gene (Supplementary Figure S2).

Further analyses of the RNA1 nucleotides showed 82.34–97.79% similarity with CMV isolates in subgroup IB and 82.63–90.27% identity with strains in subgroup IA (Supplementary Table S2). Deduced amino acids also revealed similar patterns, with 86.81–99.31% similarities with those from subgroup IB and 85.42–98.81% identities with isolates from subgroup IA (Supplementary Table S2). The isolates shared only 73.17–77.90% nucleotide and 79.02–86.11% amino acid similarities with CMV subgroup II isolates. Similarly, for the CP gene sequences, our nucleotides shared 89.37–93.42%, 92.96–96.92%, and 75.72–76.92% similarities with CMV isolates from subgroups IA, IB and II, respectively (Supplementary Table S3). The complete CP from the Nigerian CMV population shared 93.54–98.25% identity with isolates from subgroup IA, 95.85–100% identity with subgroup IB isolates and 81.94–83.79% identity with subgroup II isolates (Supplementary Table S3).

Phylogenetic analyses obtained via Bayesian probabilistic inferences categorized the CMV sequences in this study into two groups. First, within the RNA1 population, eight isolates clustered with other subgroup IB strains irrespective of their locations of origin or host. An isolate, ‘KW-AJ-02’ (MW655554), from watermelon formed a close relationship with an Ugandan CMV sequence (MG021460), as shown in Figure 4A. However, four isolates from this study formed a unique divergent subgroup with well-supported probability values. Isolates ‘26P_Ibi’ from pepper (MH798809), ‘29T_Ibi’ from tomato (MH798810), ‘KW-LF-04’ (MW655576) and ‘KW-EY-05’ (MW655577), both from watermelon, formed a distinctly separate subgroup with previously characterized CMV isolates from Nigeria (MF447457, MF447457, MF447459) and Kenya (MH567352). Likewise, all the CP gene sequences from the isolates also clustered with the subgroup IB strains, and this separation into a single clade was well supported (Figure 4B).

The automated SBP and GARD tools in Datamonkey identified recombination signals within the CMV RNA1 sequences obtained in this study (data not shown). Further analyses using the algorithms implemented in the RDP4 software identified an isolate from tomato in Oyo State, ‘29T_Ibi’ (MH798810), as a putative recombinant (P ≤ 0.05), with subgroup IB isolates ‘N-Ta05’ (MF447459) and ‘1A’ (AB042292) as putative major and minor parents, respectively (Table 5). This was supported by recombination signals (Figure 5A) and phylogenetic reconstructions (Figure 5B) that categorized putative parents of the potential recombinant in separate clusters. No recombination signal was detected within the complete CP gene regions of the Nigerian CMV isolates.

All CMV populations originating from Nigeria (partial RNA1 and RNA3 genes), obtained via our present study and those existing within the GenBank database, were pooled and analyzed. The genetic diversity within the CMV RNA1 gene populations from Nigeria was highest compared to the worldwide isolates. First, the nucleotide diversity (π) value of 0.12109 was highest in isolates from Nigeria (n = 17), compared to 0.09345 from other subgroup IB isolates (n = 17) and 0.08126 in subgroup IA sequences (n = 17) (Table 6). Similarly, ‘Eta’ and ‘S’ values showed the highest values in the populations from Nigeria, at 234 and 171, respectively. However, haplotype number and haplotype diversity values within the Nigerian population were similar to both subgroups IB and II populations. The nucleotide polymorphisms within the isolates, as revealed by Tajima’s D, Fu and Li’s D, and Fu and Li’s F statistics (Tajima, 1989), identified negative values for all the sequence datasets that did not statistically deviate from zero (P > 0.10) (Table 6). These results indicate the superfluous occurrence of low-rate polymorphisms within the RNA1 gene population, which may be caused by population proliferation, background selection or genetic hitchhiking. However, within the CP gene datasets, genetic diversity within the Nigerian population was comparable with subgroup IA and IB isolates. For example, the ‘π’ value within the Nigerian population (0.04804) was similar to the values obtained for subgroups IA (0.04550) and IB (0.04227) (Table 6). Conversely, a lower ‘S’ statistic was obtained from the Nigerian population, while ‘Eta’ values in subgroups IA and IB were almost twice the values recorded for the Nigerian sequences. The haplotype diversity within the Nigerian CP gene population was equivalent to the value obtained within subgroup II (Table 6). Additionally, Tajima’s D estimates were negative for the Nigerian CP isolates, while Fu and Li’s D and Fu and Li’s F statistics were positive, indicating population subdivision or bottlenecks. Collectively, these results confirm genetic diversity within the Nigerian CMV populations, with contrasting dynamics within the RNA1 and CP genes.

The estimation of selection pressure occurring within CMV in Nigeria were determined via the ratios of non-synonymous substitution per non-synonymous sites (d_N) and synonymous substitutions per synonymous sites (d_S). Across all the CMV subgroups, d_N/d_S ratios were less than 1 across datasets from both genes, indicating the occurrence of purifying or negative selection. Although subgroups IA and IB from isolates worldwide recorded similar numbers of sites and d_N/d_S ratios within the partial RNA1 genomes, isolates from Nigeria and subgroup II recorded d_N/d_S ratios of 0.293 and 0.254, respectively, at a p-value threshold of 0.1 (Table 7). The Nigerian CMV population had a d_N/d_S ratio that was more than thrice that of other subgroup IB populations, with similar proportions also recorded for the number of sites for negative selection. However, the d_N/d_S ratio within the Nigerian CP gene population was approximately half the values for subgroups IA and IB, albeit with almost five times the number of negative selections when compared with other isolates from subgroup IB (Table 7).