In Silico Identification of Banana High-Confidence MicroRNA Binding Sites Targeting Banana Streak GF Virus

1. Introduction

The “Cavendish” banana (Musa acuminata, AAA) is an economically important popular respiratory climacteric tropical, most widely consumed fresh fruit and appreciated around the worldwide [1]. “Cavendish” banana is a triploid (AAA) cultivar banana. “Cavendish” banana variety is developed by intraspecific hybridization in Musa acuminata (A-genome, 2n = 2x = 22) [2]. Banana streak disease (BSD) is the most destructive disease of banana, threaten food security and caused by viruses of the genus Badnavirus (Caulimoviridae) that are approved by the ICTV [3]. Badnaviruses are transmitted by the most prevalent sap-feeding mealybug species such as Planococcus citri (Pseudococcidae: Hemiptera)[4].

The Cavendish banana cultivars are most widely grown in China lack resistance to banana streak badnavirus infection [5,6,7]. The Banana streak GF virus (BSGFV) monopartite, non-enveloped, bacilliform, circular double-stranded (ds) DNA is 7325 bp in length[5]. The Badnavirus genome usually contains a minimum of three well-defined, computationally-predicted open reading frames, referred to as ORFs 1-3[8,9,10]. The ORF1 encodes a hypothetical protein. The ORFII encodes a DNA binding protein and a polyprotein was encoded by ORFIII. The intergenic region (IR), which is separated by ORF3 and ORF1, contains cis-regulatory motifs. The polyprotein is cleaved by a virus-encoded aspartic protease to multiple functional subunits. There are capsid protein, movement protein, aspartate protease, reverse transcriptase, and RNase H [5,9,11,12].

RNA interference (RNAi) is a powerful tool and reliable method for high-throughput gene functional screening in biology laboratories worldwide [13,14]. RNAi is a conserved intracellular process that is being increasingly utilized for crop protection [13]. MicroRNAs (miRNAs) are 18–24 nucleotide, endogenous, small single stranded (ss), non-coding RNA biomolecules, transcribed by RNA polymerase II, capped and polyadenylated using pre-miRNAs [15,16,17,18,19]. The miRNAs are involved to regulate host–virus interactions, cell growth, and appear as key cytoplasmic regulators of gene expression[20,21]. Artificial microRNA (amiRNA) technology is based on the design, synthesis and characterization of endogenous miRNA precursor. The amiRNA has revolutionized as a highly effective, safe and co-friendly antiviral defense approach to develop resistance in crop plants against viral infection [22,23,24,25,26].

In silico identification, annotation and expression analysis of mature miRNAs in banana has opened new ways to evaluate biotic and abiotic stresses [27,28,29]. Recently, 32 conserved mature banana mac-miRNAs were annotated, identified and further experimentally validated [30]. Bioinformatics technology accelerates to design amiRNA constructs, based on banana locus-derived miRNA-binding sites in the dsDNA-encoded genome of the BSGFV. In this study, current therapeutic strategies based on computational approach, was designed for combatting BSGFV infection in banana cultivars through in silico tools. The predicted banana miRNAs can lead to the construction of the amiRNA-based expression cassettes for transformation in the banana plants.

2. Materials and Methods

2.1. Biologicl Data

In a previous study, 32 mature banana microRNAs (Musa acuminata-microRNA) (mac-miR156-mac-5538) and (Musa ABB Group-microRNA) (mbg-miR397-mbg-399) have been computationally predicted (Supplementary File S1)[30]. In our previous study, the full-length genomic transcript (7325 bases) of BSGFV-YN (accession no. MN296502) was identified, purified and characterised from infected banana (Cavendish Musa AAA group)[5]. The full-length BSGFV dsDNA genome sequence was downloaded from the (NCBI) database [31].

2.2. Target Prediction in BSGFV Genome

Reliable target predictions were based on publically available ‘three computational algorithms’ cited in the literature (Figure 1) ― RNA22, RNAhybrid and TAPIR―were selected to analyze banana-encoded miRNAs for targeting the BSGFV dsDNA-encoded mRNAs. Analysis was carried out for the banana miRNA sequences and full-length genomic transcript sequence of BSGFV, processed in plain FASTA text format.

2.3. RNA22 Algorithm

The RNA22 relies on a web-based prediction algorithm that implements a non-seed based pattern recognition approach for detecting and analyzing miRNA-binding target sites. The RNA22 is based on an efficient, user-friendly diverse web server (http://cm.jefferson.edu/rna22v1.0/) (accessed on 9 November 2022) for predicting non-seed based biologically significant miRNA-mRNA interaction[32,33]. Default parameters were considered to find binding strength of banana mature miRNAs in the BSGFV genome. The MFE was set as −15.00 Kcal/mol.

2.4. RNAhybrid Algorithm

The RNAhybrid algorithm is an easy-to-use, online flexible tool. RNAhybrid was used to detect miRNA target binding sites in the target genome with a high stringency, based on intermolecular hybridization. RNAhybrid is used to calculate the minimum hybridization energy using the MFE model (−20.00 Kcal/mol). Free energy calculation is an initial and key step for prediction [34].

2.5. TAPIR Algorithm

The TAPIR is a state-of-the art plant-based algorithm to identify seed-based miRNA-mRNA interaction. Additionally, target binding sites were studied employing the TAPIR server. It is a web-based, rapid, and precise algorithm. TAPIR is a small RNA target analysis server. TAPIR is used to identify highly specific miRNA–mRNA duplexes. TAPIR is based on the MFE ratio in plants[35]. TAPIR was run with the standard (default) parameters.

2.6. RNAcofold Algorithm

RNAcofold is a newly develop web-based algorithm that was used to evaluate miRNA–mRNA interactions. The free energy ΔG of miRNA–mRNA duplex hetero-dimer binding is estimated. RNAcofold was run with default settings.

2.7. Discovering Banana Genome-Encoded miRNAs-Target Interaction

The miRNA-Target interaction was mapped between banana mature miRNAs and ORFs of the BSGFV genome using the circos algorithm [36].

2.8. Graphical Representation

The predicted miRNA-binding sites were processed and interpreted to generate graphical representation using the open source software package R (version 3.1)[37].

2.9. BSGFV Genome Analysis

The dsDNA genome of the BSGFV was annotated to generate ORFs using the pDRW32 (AcaClone software) (v 1.1.152).

3. Results

3.1. Banana miRNA’s Loci n BSGFV Genome

The presented multiple computational framework combines biological data and miRNA prediction algorithms to detect banana-encoded miRNAs with potential to target the BSGFV dsDNA genome (Figure 2).

In our previous study, we assessed the sequencing of BSGFV ORF genome to reveals infection in banana cultivars. We accessed the BSGFV genome form NCBI Genbank database and annotation of protein coding ORFs was performed. The BSGFV genome carries the largest double-stranded (ds) DNA with plus strand having nucleotide content 7325 bp, was drawn that exhibits significant genetic shift from different isolates. The BSGFV ORF genome consists of three ORFs (Figure 3). Multiple and highly promiscuous banana-encoded miRNA-binding sites were identified in the BSGFV ds DNA genome.

The computational approach to identifying binding affinity and significance of the 32 banana-encoded miRNAs to BSGFV dsDNA genome, was designed using an integrated in-silico predictive approach involving “three algorithms”.

In total, 12 banana-encoded miRNAs targeting 22 loci were detected by the RNA22 algorithm. The RNAhybrid algorithm identified 32 banana-genome encoded locus-derived mac-miRNA/mbg-miRNA-target pairs. Suitable candidate miRNA from banana (mac-miR172b) was observed to target BSGFV dsDNA genome at a single loci nucleotide position (nt). The TAPIR algorithm predicted potential target site of mac-miR172b at nt position (7-26) (Figure 4) (Supplementary Table S1 and File S2).

3.2. Viral ORF1-Encoding Hypothetical Protein

The badnaviral ORF1 484–1011 (527 nt) encodes the hypothetical protein and incudes a domain of unknown function [38]. The ORF1 was targeted by three predicted banana locus-derived miRNAs: mbg-miR159 (locus 971), mbg-miR399a (locus 603), and mac-miR4995 (locus 656), based on the RNA22 algorithm (Figure 5A). Suitable candidate miRNA from banana was observed to target ORFI at a single loci nucleotide position. The RNAhybrid algorithm identified binding affinity of mac-miR156g to target ORF1 at nt position 653 (Figure 5B). No banana miRNA was predicted to target the ORF1 gene with the TAPIR tool (Figure 5C and Table 1) (Table S2).

3.3. Viral ORF11-Encoding DNA Binding Protein

The badnavirus ORFII (1008–1346 nt) of 338 bases in size encodes a nucleic acid (DNA) - binding protein [38]. Among the BSGFV ORFs targeted, ORFII had very few hybridization binding sites. RNA22 predicted binding of mbg-miR159 to target ORFII at nt position 1027 (Figure 5A). Similarly, RNAhybrid identified two mac-miRNAs to target the ORFII: mac-miR156a-3p and mac-miR156h-3p at common nucleotide position 1173-1193 (Figure 5B). The miRNA prediction results revealed that no banana-encoded miRNA was detected to target the ORFII region with the TAPIR tool (Figure 5C and Table 1).

3.4. Viral ORFIII-Encoding Polyprotein

The badnavirus ORFIII (1343–6841), consisting of 5498 bases, encodes a polyprotein that is cleaved to functional sub-unit proteins. These include viral capsid and movement protein domains, aspartic protease, reverse transcriptase, and RNAse H. The polyprotein functions as pathogenicity determinant [39].

Eleven miRNAs were predicted for silencing of the polyprotein (ORFIII) by RNA22: mac-miR156a-3p (locus 2502), mac-miR156h-3p (locus 2502), mbg-miR159 (locus 1027), mac-miR319m (locus 5961), mac-miR160a (locus 1929, 4433), mac-miR160g-5p (locus 1929,4433), mac-miR162a (locus 3714, 4648, 5499), mac-miR164e (locus 1795, 4894), mac-miR166b (locus 5281), mbg-miR399a1 (locus 3567), mac-miR4995 (locus 1392, 3878, 4013) (Figure 5A) (Table S2).

Several “potentially efficient” miRNAs in the banana genome were detected to silence the ORFIII by the RNAhybrid algorithm: mac-miR156a-5p (locus 4849), mac-miR157b (locus 1873), mac-miR159 (locus 1354), mac-miR319c (locus 2698), mac-miR160a (locus 2683), mac-miR160a-5g (locus 2683), mac-miR160g (locus 1777), mac-miR162a (locus 5502), mac-miR164e (locus 1455), mac-miR166 (locus 5437), mac-miR166b (locus 6783), mac-miR166c-5p (locus 3959), mac-miR166c-3p (locus 5757), mac-miR167c (locus 2681), mac-miR167d (locus 2680), mac-miR397a (locus 2140), mac-miR399a (locus 5484), mac-miR399a1 (locus 4718) and mac-miR4995 (locus 4826) (Figure 5B). No banana locus-derived miRNA was detected by the TAPIR (Figure 5C and Table 1)

3.5. Banana miRNAs Targetig Intergenic Region of BSGFV genome

An intergenic region (IR) of Badnaviruses is located between ORF1 and ORV3, have the potential to be used as promoters. Predicted candidate miRNA from banana (mac-miR4995) was detected that targeted the BSGFV IR at nt positions 7209–7230, based on the RNA22 algorithm (Figure 5A).

The RNAhybrid algorithm analysis predicted that the BSGFV ORFIII was targeted by nine miRNAs: mac-miR156, mac-miR156d, mac-miR157b-5p, mac-miR162, mac-miR162b, mac-miR169h, mac-miR172b, mac-miR172c and mac-miR5538 at nucleotides 6945, 68, 68, 6844, 6844, 6988, 9, 9 and 55, respectively (Figure 5B). The TAPIR algorithm identified a hybridization binding site of mac-miR172b at locus position 7 (Figure 5C and Table 1).

3.6. Identification of Unique Banana-Encoded miRNAs

Based on the prediction of high-confidence banana-encoded miRNA binding sites, twelve miRNAs (mac-miR156a-3p, mac-miR156h-3p, mac-miR159, mac-miR319m, mac-miR160a, mac-miR160g-5p, mac-miR162a, mac-miR164e, mac-miR166b, mbg-miR399, mbg-miR399a1 and mac-miR4995 were detected by at least two algorithms, considered in this study (Figure 4 and Table 2).

3.7. Predicting Consensual Banana miRNAs and Silencing BSGFV Genome

Three distinct algorithms RNA22, RNAhybrid and TAPIR were used to identify high-affinity banana-miRNA binding sites in the BSGFV gnome. The consensus is expected to show robustness for predicting conserved antiviral miRNAs in the banana genome. Out of 32 banana locus-derived miRNAs investigated, only two conserved banana genome-encoded miRNAs: mac-miR162a and mac-miR172b at unique nucleotide positions 5502 and 9, respectively were detected as high-confidence hybridization binding sites located in the BSGFV genome consensus, based on combined results of two algorithms, making it the only unique miRNAs identified in this study (Figure 6).

The RNAhybrid algorithm was used to estimate the thermodynamically stable hybridization MFE of the consensus miRNA-mRNA pairs: mac-miR162a (−24Kcal/mole) and mac-miR172b (−20.5Kcal/mole) and mode of target inhibition [40] (Table 2).

3.8. Association of Banana miRNA-Target Interaction

A Circos (“Circos” software) map was generated to identify host-miRNA targets. To validate a comprehensive visualization of the host–virus interaction, the predicted circos map represents biologically credible information that integrated mapped banana miRNAs interacting with BSGFV genomic target genes (ORFs) (Figure 7).

3.9. Evaluation of Free Energy (ΔG)

Free hybridization Energy (ΔG) of the consensual banana miRNAs and their corresponding viral targets were estimated for a perfectly complementary to validate miRNA-mRNA interactions (Table 3).

4. Discussion

The BSGFV is a monopartite Badnavirus that spread to China and exhibit a general reduction in productivity and fruit quality. The BSGFV contributed to the epidemic of BSD in China [5,6,41]. A multitude of studies have investigated the expression of potential endogenous host locus-derived mature miRNAs that may target host-plant DNA or RNA viruses based on in silico criteria [42,43,44]. We observed the computational efficiency of multiple in silico tools used here, to identify binding strength of banana-encoded miRNAs for screening of large number of false-positive results. The limitations and bottlenecks of existing in silico tools were evaluated at the individual, intersection and union levels of predictions. This strategy results the ‘most effective approach’ for analyzing banana-encoded miRNA prediction findings. The miRNA targeting is based on perfect base-pairing pattern of a miRNA–mRNA seed region [45].

Also, this study examined, mature banana-encoded miRNAs were hybridized with the BSGFV dsDNA-encoded mRNAs, to evaluate promising ‘seed binding sites’ and to elucidate potential interactions with the ORF1, ORFII, and ORFII of BSGFV. MicroRNA-based target prediction remains hotspot in computational biology research. The thermal stability plays the former role in prediction analysis. RNAhybrid is used to calculate ‘free energy’ as first step for evaluation and includes ‘seed region’ as a key biological element for miRNA–target prediction [34]. We calculated the MFE of consensual miRNA-mRNA target pairs—mac-miR162a (−24.0 Kcal/mol) and mac-miR172b (−20.5 Kcal/mol), with high stringency using RNAhybrid algorithm (Table S1). These results indicate support for effective interaction of miRNA-mRNA duplexes as ‘true target’[46,47]. Further, ‘mode of target inhibition’ of consensual target pairs was determined as recommended by Broderson [40]. Based on BSGFV dsDNA genetic sequence and mature banana genome-encoded miRNAs (mac-miR162a, mac-miR172b) that are predicted to target BSGFV toward developing streak resistance in banana (Figure 6). Our study aimed to identify consensual banana genome-encoded miRNAs to interact with the ORFIII and IR of BSGFV genome. We first present a general framework of association of banana-encoded miRNA targets to capture the complexities of interactions (Figure 7). To model the system-wide virus-host interaction, computational approaches have been employed to identify the binding affinity of host miRNAs for targeted pathogens such as SCMV[48], CLCuKoV[42], RTBV[49], ACLSV[44], ICMV[43], GBNV[50], RTV1[51], SCYLV[52], ZYMV[53] and PhCMoV[54].

In our previous studies, we undertook whole-genome sequencing (WGS) of BSGFV [5], in this study; banana locus-derived potential high-confidence miRNA-binding sites were predicted from the BSGFV genome to validate miRNA–mRNA hybridization (Table 1). Our studies indicate that banana miRNAs are predicted to bind directly with high-confidence sites on BSGFV dsDNA-encoded mRNAs (Figure 5). Further, in previous studies, we find no evidence of banana genome-encoded mac-miRNAs that have predicted potential of targeting BSV genome. Moreover, based on predicted miRNA binding sites, we found that two consensus endogenous miRNAs, mac-miR162a and mac-172b were identified as highly promising biomolecules, have an even higher chance to participate in virus replication. The predicted consensual binding site of mac-miR172b is located in vital regulatory region such as IR of the BSGFV genome (Figure 5B and Figure 5C). Notably, the replication of the BSV Badnavirus relies on the duplication of ORFIII [11]. The mac-miR162a was detected top effective interacting mac-miRNAs targeting the ORFIII for silencing of polyprotein of the BSGFV genome (Figure 6). The mac-miR162a was involved to exhibit the highest overall expression in banana fruit [29]. The miR162 regulates stomatal conductance in tomato [55]. The over-expression of osa-miR162 fine-tunes rice immunity against fungal infection [56].

The host-delivered RNAi was employed to develop banana varieties to combat viral infection offers an extremely effective strategy to enhance banana production [57]. The over-expression profile of miR172 has been reported to reduce red coloration in multiple apple tissues and inhibits flavonoid biosynthesis [58]. The miR172 might be involved in carotenoid biosynthesis and involved in regulating biological clock during cold stress in wild banana [59]. The miR172b is involved in salt-stress response in rice and wheat crops [60]. Previous study show that over-expression of miR172 may lead to inhibition of SIAP2a for high production of ethylene. This process may cause fruit ripening [61]. Over expression of miR172 suppresses the brassinosteroid signalling [62]. The miR172 is involved to regulate both vegetative and reproductive development in Jatropha [63].

While interactions between miRNA-mRNA target pairs have been depicted, the construction of amiRNA-based expression cassettes and further transformations in banana to combat BSGFV infection are not fully understood. As amiRNA expression cassette relies on the potential complementary of base pairing in order to reduce off-targets toxicity [64,65]. The present study highlights the strengths and weaknesses of the first genome-based in silico evaluation of mature banana-encoded miRNAs to detect high affinity miRNA-binding sites, are based on the energetics for miRNA–target binding. By applying in silico analysis, experimental work was designed to test whether these predicted consensual miRNAs could impart resistance against BSGFV in banana plants. Furthermore, our earlier research work provides large-sale in silico support for the control of Badnaviruses [66,67]. Future work will be focused on understanding the mechanisms of how important host–virus-related interactions.

5. Conclusions

The BSGFV, which infects banana plants worldwide. BSGFV appeared as a major problem in China, associated with globally emerging ongoing infectious endogenous BSV epidemic that diminishes quantitative yield. This research reports in silico identification and characterization of banana genome-encoded mac-miRNAs/mbg-miRNA, predicted to show efficacy against BSGFV ORF genome. Prior to cloning, potential banana locus-derived high-confidence miRNA-target binding sites were identified in the targeted BSGFV ORF genome. Three “algorithms” including RNA22, RNAhybrid and TAPIR were used to detect high-confidence banana genome-encoded miRNA-binding sites in the BSGFV genome. Among the 32 banana miRNAs investigated, 2 consensus banana locus-derived mac-miRNAs (mac-miR162a and mac-miR172b) were identified as highly promising, naturally occurring biomolecules with predicted potential to impede BSGFV-YN replication. This study develops an innovative approach to focus on the development of amiRNA-based mac-miRNA therapeutics to target BSGFV ORF genome. Hence, comprehensive miRNA structure and interactome mapping banana miRNA-mRNA target interaction is critical to understand infection and disease progression.