Viral metagenomics of aphids present in bean and maize plots on mixed-use farms in Kenya reveals the presence of three dicistroviruses including a novel Big Sioux River virus-like dicistrovirus

Abstract

Background: Aphids are major vectors of plant viruses. Common bean (Phaseolus vulgaris L.) and maize (Zea mays L.) are important crops that are vulnerable to aphid herbivory and aphid-transmitted viruses. In East and Central Africa, common bean is frequently intercropped by smallholder farmers to provide fixed nitrogen for cultivation of starch crops such as maize. We used a PCR-based technique to identify aphids prevalent in smallholder bean farms and next generation sequencing shotgun metagenomics to examine the diversity of viruses present in aphids and in maize leaf samples. Samples were collected from farms in Kenya in a range of agro-ecological zones.

Results: Cytochrome oxidase 1 (CO1) gene sequencing showed that Aphis fabae was the sole aphid species present in bean plots in the farms visited. Sequencing of total RNA from aphids using the Illumina platform detected three dicistroviruses. Maize leaf RNA was also analysed. Identification of Aphid lethal paralysis virus (ALPV), Rhopalosiphum padi virus (RhPV), and a novel Big Sioux River virus (BSRV)-like dicistrovirus in aphid and maize samples was confirmed using reverse transcription-polymerase chain reactions and sequencing of amplified DNA products. Phylogenetic, nucleotide and protein sequence analyses of eight ALPV genomes revealed evidence of intra-species recombination, with the data suggesting there may be two ALPV lineages. Analysis of BSRV-like virus genomic RNA sequences revealed features that are consistent with other dicistroviruses and that it is phylogenetically closely related to dicistroviruses of the genus Cripavirus.

Conclusions: The discovery of ALPV and RhPV in aphids and maize further demonstrates the broad occurrence of these dicistroviruses. Dicistroviruses are remarkable in that they use plants as reservoirs that facilitate infection of their insect replicative hosts, such as aphids. This is the first report of these viruses being isolated from either organism. The BSRV-like sequences represent a potentially novel dicistrovirus infecting A. fabae.

Keywords: Aphid; Dicistrovirus; Epidemiology; Metagenomics; Phylogenetics; Potyvirus; Recombination; Vector.

Fig. 1

Sampling sites in Kenya. Crosses indicate sampling sites in the Central, Eastern and Rift Valley regions of Kenya where farmers’ fields were sampled for aphids (blue) and maize leaf (red). Black lines demarcate county boundaries. Three farms per site were sampled for aphids in Ndeiya, Kiambu county (humid, highland), Oloirien, Kajiado county (semi-arid, highland), Katumani, Machakos county (semi-arid, lowland) and Kaiti, Makieni county (Humid, lowland). For the maize samples, 14 farm sites in four counties were sampled for maize leaf samples. These were in Baringo (semi-arid, highland), Machakos (semi-arid, lowland), Kitui (semi-arid, lowland) and Nakuru (humid, highland). The details of the sampling sites and GPS coordinates are provided in Additional File 1: Table S1

Fig. 2

Genome organization and mapping coverage of Aphid lethal paralysis virus (ALPV) RNA sequences from samples P7 and P9. The positive-sense single-stranded RNA genome of ALPV has a length of approximately 9800 nucleotides (nt). A protein molecule (VPg: virus-protein-genome linked) is covalently attached to the 5′-end of the genomic RNA, which has a 3′-poly(A) tail indicated by A_n. The 5′-untranslated region (5′-UTR) and the intergenic region (IGR) are 338 and 197 nt, respectively, and each function as an internal ribosome entry site (IRES) for translation of the two open reading frames (ORFs: gray boxes). ORF1 is 6111 nt long and is predicted to encode a 2037 amino acid precursor for the viral non-structural proteins. The ORF1 product contains the conserved motif for the 3C–like protease domain between residues 1395–1443 of the ORF1 product. Only putative functional domains in the virus-encoded proteins that can be predicted with a high degree of certainty are shown in the genome map (upper panel). Predicted helicase and RNA dependent RNA polymerase (RdRp) domains (indicated by blue boxes) were determined by searching for sequence homology in the NCBI conserved domain database. ORF2 is 2403 nt long and is translated into an 801 amino acid polyprotein that self-processes into four mature structural proteins VP2, VP4, VP3 and VP1 (indicated by blue boxes). The cleavage sites for the ALPV structural proteins have not yet been experimentally determined. However, their deduced positions based on amino acids alignments have been proposed [36]. By searching for the conserved motifs the cleavage position was determined for VP2/VP4 (I₂₂₈ AATAQ/VGTEAI₂₃₈) to be between residues 228–238 and VP3/VP1(I₅₅₃ to RGVAQ/VNVAES₅₆₃) to be between residues 553–563 of the amino acid sequence of ORF2 of ALPV. Sequence reads from samples P7 and P9 were mapped against an ALPV sequence from GenBank (reference NC004365) in CLC workbench version 5.1. The pink traces represent the depths of coverage at each nucleotide position which was × 1997 and × 1983 for samples P7 and P9, respectively

Fig. 3

Phylogenetic analyses and pairwise sequence similarity comparisons of Aphid lethal paralysis virus (ALPV). a ALPV isolates identified from aphids in Kenya (KE) this study (KE Aphid P7 and KE Aphid P9) and an isolate from this study found in maize plants in Kenya (KE Maize) were analysed alongside six other ALPV genomic sequences available at GenBank (from China, South Africa, Israel, East Timor USA and Spain). Analysis revealed two main clades. The isolates KE Aphid P7 and KE Aphid P9 clustered in a subclade together with the ALPV isolate from Israel in the larger clade and KE Maize clustered with isolates from China, East Timor and South Africa (orange dashed box). Drosophila C virus (DCV) was used as the out-group. b The arrangement of the sequences in the pairwise comparison is based on a neighbour-joining tree from the aligned sequences. Pairwise alignments revealed similarities as low as 82% when isolates in different clades are compared. In both analyses, the sequences were aligned using MUSCLE. The phylogenetic tree was constructed using the Neighbor-Joining method with 1000 bootstraps in MEGA6. Asterisks indicate ALPV isolates identified in this study

Fig. 4

Pairwise comparison of Big Sioux River virus (BSRV)-like sequences to ALPV and RhPV sequences. The BSRV-like sequences shared up to 99% sequence similarity among each other. Similarly, the other species in the analysis showed close intra-species similarity. Interspecies comparisons revealed a clear demarcation between the different species. The sequences were aligned using MUSCLE and the arrangement of the samples in the matrix is based on phylogenetic clustering of neighbour-joining tree constructed from the multiple sequence alignment. RhPV sequences (marked with two asterisks) used in this analysis were from GenBank while the ALPV sequences used (marked with a single asterisk) were from this study

Fig. 5

Phylogenetic analysis of Big Sioux River virus (BSRV)-like sequences identified in this study. The BSRV-like sequences from our study are (in the dashed box) clustered most closely with Rhopalosiphum padi virus (RhPV) in the genus Cripavirus [57], which include the type species Cricket paralysis virus (CrPV), RhPV, as well as Aphid lethal paralysis virus (ALPV) and Drosophila C virus (DCV). The second clade shown has sequences from the genus Aparavirus [57] including the type species Acute bee paralysis virus (ABPV), Kashmir bee virus (KBV), and Israeli acute paralysis virus (IAPV). The phylogenetic tree was constructed using the Neighbour-Joining method with 1000 bootstraps. ALPV isolates identified in this study are indicated with red asterisks

Fig. 6

Deduced genome organization of the (BSRV)-like virus. The positive-sense single stranded RNA genome of the BSRV-like virus is approximately 10.2Kb and 3′-polyadenylated. The 5′-untranslated region (5′-UTR) and intergenic region (IGR) are 680 and 594 nucleotides (nt) long, respectively, and are internal ribosome entry sites (IRESs) for translation of open reading frames (ORFs) 1 and 2 (gray boxes). A VPg molecule is likely attached to the 5’end of the genomic RNA. The IGR-IRES for this virus is longer than that of other dicistroviruses. By comparison, RhPV is the only other dicistrovirus with a long IGR-IRES (533 nt), while the ALPV IGR-IRES is 197 nt long (see Fig. 2). ORF1 (6024 nt) is predicted to encode a non-structural protein precursor of 2008 amino acid residues. ORF2 is 2262 nucleotides long and is translated into a 754-amino acid long structural protein precursor and if consistent with other dicistroviruses, self-cleaves into mature capsid proteins. Locations of putative helicase and RdRp domains encoded by ORF1 and three putative structural proteins encoded by ORF2 (shaded blue) were obtained by searching for homology in the NCBI conserved domain database. As in Fig. 2, only those putative functional domains in the encoded proteins that can be deduced with a high degree of certainty are shown