In silico identification of sugarcane (Saccharum officinarum L.) genome encoded microRNAs targeting sugarcane bacilliform virus

Sugarcane bacilliform virus (SCBV) is considered one of the most economically damaging pathogens for sugarcane production worldwide. Three open reading frames (ORFs) are characterized in the circular, ds-DNA genome of the SCBV; these encode for a hypothetical protein (ORF1), a DNA binding protein (ORF2), and a polyprotein (ORF3). A comprehensive evaluation of sugarcane (Saccharum officinarum L.) miRNAs for the silencing of the SCBV genome using in silico algorithms were carried out in the present study using mature sugarcane miRNAs. miRNAs of sugarcane are retrieved from the miRBase database and assessed in terms of hybridization with the SCBV genome. A total of 14 potential candidate miRNAs from sugarcane were screened out by all used algorithms used for the silencing of SCBV. The consensus of three algorithms predicted the hybridization site of sof-miR159e at common locus 5534. miRNA–mRNA interactions were estimated by computing the free-energy of the miRNA–mRNA duplex using the RNAcofold algorithm. A regulatory network of predicted candidate miRNAs of sugarcane with SCBV—ORFs, generated using Circos—is used to identify novel targets. The predicted data provide useful information for the development of SCBV-resistant sugarcane plants.


Introduction
Sugarcane bacilliform viruses (SCBVs) are classified into the Badnavirus genus of the Caulimoviridae family. These viruses are composed of monopartite, circular, non-enveloped bacilliforms that are (30 × 120-150 nm) in size, with a double-stranded DNA (ds-DNA)-genome of approximately 7.2-9.2 Kbp in size [1]. The genome of SCBV constitutes three major open reading frames (ORFs) that are located on the 'plus DNA strand' with a single discontinuity [2]. ORF1 encodes a small hypothetical protein. interactions. The novel computational approach here supports the idea of generating SCBVresistant sugarcane plants through genetic engineering.

SCBV genome retrieval and annotation
The full-length transcript of the SCBV-BRU genome was isolated from the S. officinarum cultivar and then published, and available via accession no. JN377537 [31]. The expected sizes and abundances of the ORFs along nucleotide distributions of the above mentioned NCBI retrieved SCBV-BRU genome were estimated using the pDRAW32 DNA analysis software (version 1.1.129) (AcaClone software). The SCBV-BRU genome annotation represents ORFs of varying lengths.

Target prediction in SCBV genome
Target prediction is considered a key feature towards the identification of credible miRNA-mRNA interaction hybridization. At present, many target prediction algorithms have been designed to predict and identify the best miRNA target candidate. Each tool uses specific criteria and methods for miRNA target prediction. We used four target prediction algorithms cited in the literature (miRanda, RNA22, RNAhybrid and psRNATarget) to find the most relevant sugarcane miRNAs for silencing of the SCBV genome (Table 1). These computational tools compute the complementarity-based attachment of miRNA-mRNA. This attachment is divided into seed and mid regions. The mismatch in the seed region is more damaging than that of a mismatch in the middle region of miRNA-mRNA attachment. This provides the basis for over-sensitivity for the computation. We can set higher penalty of a mismatch in seed region which will make the prediction more sensitive. We designed an effective computational approach to analyze miRNA targets at three different prediction levels namely the individual, union, and intersection levels. A detailed workflow pipeline is presented in (Fig 1) below.

RNA22
RNA22 is a user-friendly, web-based (http://cm.jefferson.edu/rna22v1.0/) novel pattern-recognition algorithm that is used for predicting target sites with corresponding hetero-duplexes. Non-seed-based interaction, pattern recognition, site complementarity, and folding energy are the key parameters of the RNA22 algorithm [34]. Final scoring removes the need to use a cross-species conservation sequence filter [35].

RNAhybrid
RNAhybrid is an easy-to-use, fast, flexible, web-based (http://bibiserv.techfak.uni-bielefeld.de/ rnahybrid) intermolecular hybridization algorithm that is used to estimate mi RNA-mRNA interaction as well perform target prediction based on MFE hybridization. A p-value is assigned to assess RNA-RNA interaction-based hybridization sites in the 3 0 UTR sequence [36]. RNAhybrid is widely used to estimate the MFE of the consensual mi RNA-target pair and the mode of target inhibition as suggested [37].

psRNATarget
psRNATarget is a new web server (http://plantgrn.noble.org/psRNATarget/) that is used to identify the target genes of plant miRNAs based on a complementary matching scoring schema. It has been used to discover validated mi RNA-mRNA interactions [38]. The plant psRNATarget was designed to integrate a key function for miRNA target prediction using complementarity scoring and secondary structure prediction [39]. Target site accessibility was evaluated by estimating the unpaired energy (UPE) to unfold a secondary structure [37].

Genome Organization of SCBV
SCBV is a plant pararetrovirus that is, classified in the genus Badnavirus of the family Caulimoviridae. The genomic ds-DNA molecule of SCBV is comprised of three ORFs, separated by an intergenic region (IR). ORF1 is composed of 557 nucleotides (618-1175 nt), encoding a hypothetical protein (P1) with 185 amino acids (aa),while ORF2 is composed of 370 nucleotides (1176-1546 nt) codes for a virion-associated DNA binding protein (P2) with 123 aa. The precise functional capabilities of these proteins (encoded by ORF1 and ORF2) have not been explored. A large polyprotein (1977 amino acids) is encoded by ORF3 (1547-7479 nt) to cleave by a viral aspartic protease. The resulting proteins obtained are named as movement, capsid protein, aspartyl proteinase, reverse transcriptase and ribonuclease H. The IR is composed of 1022 nucleotides (7479-618) and is located between 3'-ORF3 to 5'-ORF1. The intergenic region (IR) works as a promoter and controls the transcription and regulation of the SCBV genome. The genome organization of the SCBV with three ORFs is shown in (Fig 2).

ORF2 encoding DNA binding protein
A nucleic acid (DNA)-binding protein of the SCBV genome is encoded by ORF2 [6,45]. RNAhybrid and miRanda predicted potential target binding site of ssp-miR166 at locus position 1449-1470 ( Fig 3A and 3C). Suitable candidate miRNAs from sugarcane (ssp-miR444 (a, b, 3p) were observed to target ORF2 at a single loci nucleotide position (1301-1326) as determined by the miRanda algorithm ( Fig 3A). No sugarcane miRNAs were predicted to target the ORF2 gene with the RNA22 tool ( Fig 3B). Similarly, RNAhybrid predicted the binding of ssp-miR166 at locus 1450 ( Fig 3C). The miRNA prediction results revealed that no candidate miRNA was identified to have a potential genome binding site in the ORF2 region, as predicted by psRNATarget ( Fig 3D).

ORF3 encoding polyprotein (CP, AP, RT, and RNase H)
The poly proteins constitute the largest portion of the SCBV genome encoded by ORF3 [2,6]. Potential candidate miRNAs from sugarcane were identified by the miRanda algorithm to

Visualization and analysis of miRNA-target interaction network
Initially, the Circos plotting tool was designed to analyze mutations with comparative metagenomics and transcriptomic biological data [46]. To study a comprehensive visualization of host-virus interaction, we created a Circos plot to integrate biological data from sugarcane miRNAs and their predicted SCBV genomic target genes (ORFs) (Fig 5). In order to reduce visual graphical complexity and permit improved readability, we only used selected sugarcane miRNAs and their SCBV targets obtained from miRanda analysis. The miRanda algorithm considers seed-based interactions and the conservation level [47,48]. The results suggest that biological data visualization of candidate miRNAs from sugarcane, with SCBV-encoded ORFs determines credible information of desirable preferred targets of SCBV ORFs using consensual miRNAs. We have combined sugarcane miRNA data and their predicted SCBV targets simultaneously in this manner.

Predicting consensual sugarcane miRNAs for silencing the SCBV genome
Out of 28 sugarcane miRNAs, only six sugarcane miRNA (sof-miR159 (a, b, d and e) at common locus position 5535 and ssp-miR444 (a, b) at locus 6797) were predicted at the common locus by at least three of the algorithms used (Fig 7 and Table 2). Out of 14 consensual miR-NAs, only one miRNA of S. officinarum (sof-miR159e at locus 5535), with a MFE of -26.7 Kcal/mol, was considered as the top effective candidate in terms of support more efficient silencing of the SCBV genome. The efficacy of the sof-miR159e target against SCBV was validated by the suppression of RNAi-mediated viral combat through the cleavage of viral mRNA or translational inhibition [43]. Multiple loci interactions were observed for sof-miR159e at nucleotide positions 5534-5552 (consensus of three algorithms, namely, miRanda, RNA22, and RNAhybrid) and 2647 (psRNATarget) of ORF3.

Prediction of consensus secondary structures
The validation of consensual sugarcane miRNAs was confirmed by the prediction of their stable secondary structures using the RNAfold algorithm. Precursors of mature sugarcane miR-NAs were manually curated. The MFE is the key factor to determine the stable secondary structures of precursors. All the predicted consensual sugarcane miRNA precursors were observed to possess lower MFE values (ranging from −57.70 to −114.70 kcal/mol) ( Table 3).
The predicted secondary structures of six precursors of pre-miRNAs are shown in (Fig 8), as predicted by the intersection of three consensual algorithms at the same locus. The top stable secondary structure of the sof-MIR159e precursor was predicted with standard features (MFE: 107.50 Kcal/mol, MFEI: 1.06 Kcal/mol). The predicted secondary structures of 14 consensual sugarcane miRNAs passed the aforementioned standard criteria. We have determined

Tissue preferential expression analysis of sugarcane miRNAs
We used the "PmiRExAt" database to search for the expression analysis of the predicted sugarcane miRNAs. Homologous miRNAs were present in all three plant species, i.e., maize, rice, and wheat (S1-S3 Figs). The expression of these microRNAs was identified in all tissue types

PLOS ONE
Computational prediction of sugarcane (Saccharum officinarum L.)-encoded microRNA targets against SCBV in each species. Therefore, the expression of sugarcane miRNAs was confirmed in other plant species, i.e., maize, rice, and wheat. Evidence of the existence of the same miRNAs in sugarcane is also provided. Most of the stated miRNAs have also been confirmed, in multiple studies, for their expression and roles in plant cellular pathways [49,50].

Discussion
For the filtering of false positive results, we studied the effectiveness of the computational algorithms considered here to validate the miRNA target prediction data. We designed an effective approach for the validation of miRNA target prediction results at individual, union, and intersection levels. Computational prediction algorithms offer rapid methods to identify potential host-derived miRNA targets in virus genomes. Default parameters represent optimized specifications for each miRNA to its respective target site in the viral genome. This varies with respect to each algorithm/tool and can be modified for fine-tuning the settings or increasing the level of sensitivity for predicted sites. Default parameters are effective for screening out false-positive attachment sites for miRNAs using multiple prediction tools. miRanda is a widely used algorithm that includes the main aspects of miRNA-target prediction, such as the conservation level and miRNA 3'UTR site [51]. The RNA22 algorithm is a novel alternative option for exploring new miRNA-mRNA interactions because of its unique capabilities-although it has a high likelihood of generating false-positive results [47]. We calculated the MFE and determined the target inhibition as recommended by Broderson by using RNAhybrid [37]. Several potential sugarcane miRNA targets and miRNA-mRNA interactions could be consensually predicted by all of the algorithms (Fig 7). Plant miRNAs are responsible for inducing the degradation of the target genes using perfect or imperfect complementarity base pairing [52]. The current study demonstrates that SCBV genome components (ORF1, ORF2, and ORF3) are susceptible to targeting by a set of consensual sugarcane miRNAs. In addition, sof-miR159 (a, b, d, and e) was found to target ORF3 at a consensual hybridization site by at least three algorithms (Fig 8). Free energy assessment is a dynamic feature of miRNA and target binding. Previous studies have revealed a significant correlation of free energy between the translational repression and the hybridization binding of the seed region [53]. The thermodynamic stability of the miRNA-mRNA duplex was estimated by the assessment of free energy to monitor site accessibility for the determination of the secondary structure duplex [27]. In order to validate miRNA-mRNA interaction, the free energy of a duplex was assessed ( Table 2). Our prediction results show high stability for the sugarcane-encoded miRNA-

PLOS ONE
Computational prediction of sugarcane (Saccharum officinarum L.)-encoded microRNA targets against SCBV SCBV-mRNA duplex at a low free energy level (Table 3 and Fig 8). The RNA duplex is considered to be more stable due to the stronger binding of miRNA to mRNA [54,55]. We used union and intersection approaches to reduce false positive prediction. Union approaches rely on combining more than one target prediction tool when finding true and false targets. The sensitivity level for a predicted target increases due to a decrease in specificity.

PLOS ONE
Computational prediction of sugarcane (Saccharum officinarum L.)-encoded microRNA targets against SCBV An intersection approach is entirely different and depends upon the combination of two or more computational tools and enhances the specificity level of predicted targets due to a decrease in sensitivity [56]. Our target prediction results revealed that both computational approaches achieved the best outcomes with maximum performance for predicting and estimating the best targets (Figs 6 and 7). Previous studies have also reported the silencing of plant viruses using host-derived miRNAs when applying a set of computational algorithms. The identification and evaluation of best-fit candidate miRNA targets for different plants has been concluded successfully with potato virus Y (PVY) [57], maize chlorotic mottle virus (MCMV) [58], CLCuKoV-Bu [59], rice yellow mottle virus (RYMV), [60] and SCBGAV to find miRNA-target interaction [61]. We have designed an equal novel bioinformatics approach for target prediction in the SCBV genome to control the emerging presence of Badnavirus in sugarcane cultivars.
In our previous study, we identified the most ideal consensual sugarcane miRNA (sof-miR396) to target ORF3 of the SCBGAV genome using multiple computational algorithms [61]. The quantity of false positive miRNA-target interaction estimated by multiple algorithms depends upon the mode of miRNA-target recognition. MFE is also another important factor that affects miRNA-target interaction in result validation [62]. To set a lower MFE value will give rise to a higher probability of miRNA-target complex formation [63]. In the current study, for miRanda analysis, a stringent cut-off point of −15 kcal/mol was set for narrowing down the miRNA candidates. Similarly, to validate host-virus interaction, a MFE cut-off point of -20 kcal/mol applied for RNAhybrid analysis [32].
Although MFE has a considerable role for development of miRNA-mRNA complexes, it does not certify that interactions will lead to functional changes. In the current study, we have identified six potential miRNA hybridization binding sites that have exhibited low MFEs and free energy for duplex formation. These predicted miRNAs not only have potential targets for the SCBV genome at the transgenic level but also have a stronger probability to develop miRNA-viral mRNA complex formation. These miRNAs also have chance to participate in a SCBV replication mechanism, where a consensus sugarcane miRNA (sof-miR396) has a binding site within the SCBV large intergenic region (LIR) at locus 79 as predicted by the miRanda and RNAhybrid algorithms. In the previous study, we predicted that sof-miRNA396 is an effective candidate to target the SCBGAV genome [61]. Notably, sof-miR159e was predicted by all the algorithms. Additionally, miR159 was explored and was found to present a strong role for silencing GAMYB to enable normal growth [64]. Phe-MIR159 involved in regulating the gene responsible for secondary thickening in Phyllostachys edulis [65]. It is important to assess the function of predicted potential consensual miRNAs for the identification of Badnavirus replication to demonstrate SCBV replication experimentally. A hypothetical model was designed to show that sugarcane-derived miRNAs can inhibit SCBV mRNA and sugarcane genes against SCBV virus (Fig 9). It facilitates plant-encoded miRNAs in the cleavage of SCBV miRNA.

PLOS ONE
Computational prediction of sugarcane (Saccharum officinarum L.)-encoded microRNA targets against SCBV RNAi screening is a novel technology for discovering various cellular functions and identifying host-derived factors of viruses [66]. Here, we selected 28 experimentally validated sugarcane miRNAs with annotated targets that are part of SCBV. amiRNA-based silencing technology has been successfully validated in many crop plants for controlling emerging plant viruses [23,24,26]. In summary, our computational work for SCBV genome silencing could offer a new approach for the production of antiviral agents. Furthermore, we demonstrated a method to minimize the novel antiviral effects of host-derived miRNAs against SCBV.

Conclusions
SCBV has appeared as a major problem in China. SCBV diminishes quantitative yields in all sugarcane cultivars. In the current study, prior to cloning, we have applied computational tools to predict and comprehensively analzse candidate miRNA from sugarcane against SCBV. Among them, sof-miR159e was predicted as the top effective candidate that could target the vital gene (ORF3) of the SCBV genome. Our results demonstrate an alternative strategy to existing molecular approaches that could be repurposed to control badnaviral infections. The current findings provide in silico evidence of a novel scheme to construct miRNA-mediated gene silencing therapeutics to combat SCBV.