Genome-wide identification of endogenous viral sequences in alfalfa (Medicago sativa L.)

Endogenous viral elements (EVEs) are partial or entire viral genes or genomes integrated into host chromosomes and inherited as alleles [1, 2]. EVEs may be eventually removed from the host genomes or reach fixation and retained for millions of years [1, 3, 4]. They are instrumental in a gene flow between viruses and eukaryotes and may have a substantial role in the evolution of their hosts [1, 5, 6]. While the majority of retroviral EVEs originate from a mandatory genome integration stage in the life cycle of reverse-transcribing viruses [1], insertion mechanisms of plant viruses is debated because integration of a DNA copy into the host genome is not required for their replication [7].

In plant genomes, two categories of EVEs are recognized: endogenous pararetroviruses (EPRVs), derived from the reverse-transcribing double stranded (ds)DNA viruses of the family Caulimoviridae and endogenous non-retroviral elements (ENREs), originating from dsRNA, ssRNA, and ssDNA viruses [4, 5, 7‐9]. Several mechanisms for the integration of EPRVs into the plant genome were proposed, among them homologous and illegitimate recombination, integration through reverse-transcribed replication intermediates and via short AT-rich motifs flanking viral segments [8, 10, 11].

Insertion mechanisms of the ENREs remain largely uncharacterized and may involve reverse transcription by reverse transcriptase encoded by retrotransposons, recombination events, and transposon-mediated integration [5]. Localization and distribution of EVEs in plant genomes varies; it was reported that they might be located closer to transposable elements, overlap with genes [12], reside between the genes or within introns [5]. Numerous EVEs of both classes have been identified in the genomes of different plant species, including a close relative of alfalfa, Medicago truncatula [5, 7, 8, 12, 13].

Two M. sativa genomes, tetraploid [14] and diploid (http://​www.​medicagohapmap.​org/​downloads/​cadl), were used to create a standalone BLAST [15] database. At the time of this project’s initiation, the tetraploid genome was assembled to the chromosome level, while the diploid genomic sequences were available as scaffolds. To identify possible endogenous viral elements in the M. sativa genome we first used plant viral protein sequences from the Uniprot database (https://​www.​uniprot.​org). These sequences were downloaded and piped through the M. sativa tetraploid and diploid genomes using the tBLASTn program. The standalone BLAST search was also used against reference viral database (https://​www.​ncbi.​nlm.​nih.​gov/​genome/​viruses/​).

The BLAST analyses revealed sequences resembling two plant virus species of the family Caulimoviridae and integrated into both tetraploid and diploid genomes of M. sativa: Soybean chlorotic mottle virus (SbCMV) of the genus Soymovirus and Figwort mosaic virus (FMV) of the genus Caulimovirus. The resultant parameters of the BLASTn output and the coordinates of the detected homology regions in the M. sativa genomes are shown in Table 1 and Additional file 2: Table S1. The two viral genomes (for Genbank accessions see Table 1) were further used in BLASTn search to confirm the location of the integrated viral sequences. All EVEs found in M. sativa tetraploid and diploid genomes are shown in the Additional file 1: File S1.

The SbCMV-like EPRVs were homologous to the genes encoding reverse transcriptase and movement protein (MP) of SbCMV. Since reverse transcriptase sequences were most probably derived from the multiple integrated transposons and could intervene with the analysis and identification of EPRVs, they were excluded from consideration. The SbCMV-like EPRVs found in both tetraploid and diploid genomes and corresponding to the MP overlapped (percent identity 90%). The top BLAST hit for the longest insert from the diploid genome (563 bp) was SbCMV isolate from Japan (X15828.2; 70%; query cover 100%; E-value = 2e−63). Translation of the SbCMV-like sequence resulted in 163 amino acid-long fragment that had 59% identity with SbCMV movement protein (NP_044299.1; query cover 100%; E-value = 4e−66). Low identity levels (more than 20% of both nucleotide and amino acid differences) indicated that this sequence could belong to a new pararetrovirus, which we provisionally named alfalfa endogenous pararetrovirus (AePV).

To confirm an identity of the discovered AePV EPRVs, we performed a phylogenetic analysis with the representative viruses. Phylogenetic trees were deduced from ClustalW alignments of the AePV sequence and the complete nucleotide sequences of the representative species from the family Caulimoviridae and built using MEGA 7 software [16] with Maximum Likelihood method based on the Tamura-Nei model and bootstrap analysis of 1000 replicates. Phylogenetic analysis clustered AePV together with SbCMV, thus confirming their relationship (Fig. 1). The group branched to another species in the genus Soymovirus, Blueberry red ringspot virus. All other members of the family grouped correspondingly to their respective genera.

Screening the AePV fragment against the genome of Medicago truncatula, a close relative of alfalfa, using the Plant Genomic Resource Phytozome 12 (https://​phytozome.​jgi.​doe.​gov/​pz/​portal.​html), resulted in the identification of orthologous sequences located in all eight chromosomes of the species as well as in several scaffolds (Table 2). Representative orthologous sequence from the chromosome 3 of M. truncatula clustered together with the AePV, supporting their close relationship (Fig. 1).

When the translated portion of the AePV or the orthologous sequence of M. truncatula were searched against the genome of M. truncatula (taxid 3880), a first BLASTp hit corresponded to the “viral movement protein” (AES67244.1) that was 97% identical (100% query cover; E-value = 3e-111) to the AePV. This indicates a presence of the AePV EPRVs in both species.

Short FMV-like EPRVs were found in most chromosomes of the tetraploid genome and scaffold assemblies of the diploid genome (Table 1). The sequences overlapped with each other, forming a 93 nt consensus sequence with a minimum/maximum percent match 85%/100% (Additional file 3: Figure S1). In BLASTn search optimized for high similarity, the consensus sequence had only one hit—with the FMV capsid protein (query cover 97%; E-value = 2e−09; identity 80.43%; X06166.1). When BLASTn was optimized to search for the somewhat similar sequences, the best hits also included portions of the capsid proteins of other species in the family Caulimoviridae.

Phylogenetic trees, deduced from the alignment of the FMV-like consensus sequence and complete genomes of the representative members of the family Caulimoviridae, placed FMV-like consensus in the same cluster with the reference sequence of figwort mosaic virus (NC_003554.1; Fig. 2). Thus, while the FMV-like EPRVs certainly belong to Caulimoviridae and are in all probability resembling FMV, the sequences are too short to make any definite conclusions on their exact identity. The search of the FMV-like consensus against genome of M. truncatula resulted in multiple hits (Table 2), implying that the EPRVs are present in the genome of this species as well.

To ensure that identified EPRVs are not artifacts or contaminants in sequence databases, they were randomly examined by PCR following by Sanger sequencing of the PCR products. DNA was extracted using DNeasy Plant kit (Qiagen) from one-week-old seedlings of alfalfa (M. sativa L) cv. Regency SY and common pea (Pisum sativum) cv. Lincoln, germinated in Petri dishes on moistened sterile filter paper. Prior to germination, alfalfa seeds were scarified with H₂SO₄, surface-sterilized with 70% ethanol for 1 min and rinsed with distilled water. PCR was performed using AmpliTaq Gold 360 DNA Polymerase under conditions recommended by the manufacturer (Thermo Fisher Scientific, USA), in two–three technical replicates. PCR primers used in the assays were specific to the alfalfa sequences located outside of the selected endogenous elements (Additional file 2: Table S1). DNA from common pea (P. sativum cv. Lincoln) served as a negative control. PCR products were purified with Qiagen PCR Purification kit and sequenced at Macrogen USA facility (Macrogen, MD USA). In all cases, PCR led to the amplification of the products of the expected size (Fig. 3). Sequencing of the purified reactions confirmed that they contained targeted regions incorporating EPRVs. These experiments verified computational findings.

In conclusion, we report the identification of two types of EPRVs in publicly available genomes of tetraploid and diploid alfalfa: SbCMV-like endogenous segments, presumably representing a novel virus in the family Caulimoviridae, tentatively named AePV; and FMV-like elements, which could not be classified assuredly due to the small size, although their homology to FMV and other members of the family Caulimoviridae is evident.

While great many EVEs have been found in the genomes of various plant species, as far as we know, none have been reported in alfalfa prior to this work [5, 7, 8, 12, 13]. This is likely due to the absence of the alfalfa genome, which only recently became available [14].

Pararetroviruses of the family Caulimoviridae identified in this study represent the most common group of EVEs in plants. They have been proposed to be major components of various plant genomes, possibly contributing to evolution of their hosts as sources of novel genetic material [12]. Despite that two viruses, SbCMV and FMV, which are similar to the EPRVs found in this study, have not currently been reported to infect alfalfa, these infections might occur in nature but so far had gone unnoticed. As a matter of fact, our unpublished results indicate that SbCMV and FMV are present among other viral genomes detected by high-throughput sequencing in field alfalfa samples (Nemchinov et al. unpublished).

Based on the BLASTn results and phylogenetic analysis, it can be assumed that the integration events recorded here are unlikely of recent nature and most probably occurred at some point during the species evolution. The diploid genome used in this work originated from CADL alfalfa (Cultivated Alfalfa at the Diploid Level), developed from cultivated tetraploids using haploidy, with breeding and selection [17]. It is, therefore, plausible that EPRVs identified in the CADL genome could be traced back to the tetraploid germplasm. This implies that they have reached fixation in the alfalfa genome, were inherited as alleles, and may presumably carry a functional load [12, 18].

Of two classes of EPRVs, only FMV-like sequences were found in most chromosomal sets of the tetraploid genome (Table 1). This may suggest, although speculatively due to the small size of the FMV inserts, that the integration events occurred before the whole genome duplication and both diploid progenitors of the tetraploid genome [19] contained EPRVs.

Hypothetically, uncovering the orthologous viral sequences in the genome of M. truncatula may further postpone the timetable of the integration events to at least ~ 5.3 million years ago, before alfalfa divergence from M. truncatula [14]. Therefore, it appears that EPRVs are stable constituents of the host genome. As a result, they could potentially acquire functional roles in alfalfa’s normal growth, organ development, metabolism, and response to environmental stresses [12]. It cannot be ruled out that EPRVs in alfalfa may also represent a source of infection.