Cucumber mosaic virus-induced gene silencing in banana

Banana (Musa spp.) is one of the world’s most important staple and cash crops. Despite accumulating genetic and transcriptomic data, low transformation efficiency in agronomically important Musa spp. render translational researches in banana difficult by using conventional knockout approaches. To develop tools for translational research in bananas, we developed a virus induced-gene silencing (VIGS) system based on a banana-infecting cucumber mosaic virus (CMV) isolate, CMV 20. CMV 20 genomic RNA 1, 2, and 3, were separately cloned in Agrobacterium pJL89 binary vectors, and a cloning site was introduced on RNA 2 immediately after the 2a open reading frame to insert the gene targeted for silencing. An efficient Agrobacterium inoculation method was developed for banana, which enabled the CMV 20 VIGS vector infection rate to reach 95% in our experiments. CMV 20-based silencing of Musa acuminata cv. Cavendish (AAA group) glutamate 1-semialdehyde aminotransferase (MaGSA) produced a typical chlorotic phenotype and silencing of M. acuminata phytoene desaturase (MaPDS) produced a photobleachnig phenotype. We show this approach efficiently reduced GSA and PDS transcripts to 10% and 18% of the control, respectively. The high infection rate and extended silencing of this VIGS system will provide an invaluable tool to accelerate functional genomic studies in banana.

Bananas belong to the family Musaceae and represent one of the world's most important cash and staple crops 1 . This perennial crop is cultivated throughout the tropics and sub-tropics, and over 400 million people in developing regions are dependent on bananas as a vital food nutrient. In 2017, worldwide production of bananas was estimated to be over 113 million tons 2 . The vast majority of cultivated bananas are triploid, and are derived from inter-and intra-specific crosses between two diploid wild species, Musa acuminata (A genome) and M. balbisiana (B genome) 3 . Currently, commercialized banana is dominated by the Cavendish (AAA) cultivar group 4 .
The genome of M. acuminata ssp. malaccensis, a progenitor of the current cultivated banana, was the first Musa genome sequenced 5 , followed by a draft release of the genome of another banana progenitor, M. balbisiana 6 , and a close relative, M. itinerans 7 . In addition, a number of gene expression profiles of different banana cultivars, tissues, and conditions have also been investigated [8][9][10][11][12][13][14] . Despite the success of banana genomic and transcriptomic sequencing projects, reverse genetic approaches for identifying and characterizing gene functions in banana remain a challenge.
To date, characterizing target gene(s) of interest in banana has mainly relied on transgenic banana plants overexpressing the exogenous target gene [15][16][17] . More recently, a novel transgenic approach based on clustered regularly interspaced short palindromic repeats (CRISPR)/CRISPR-associated protein 9 (Cas9) are being reported for plants, and the use of CRISPR-Cas9 has been reported in banana cv. Rasthali 18 . However, low transformation efficiency and the time-exhaustive regeneration processes involved in this transgenic approach are disadvantageous in functional characterization of one or several candidate genes using CRISPR-Cas9 in banana. Moreover, polypoidy especially in the triploid Cavendish banana presents additional challenges for genetic knockout analysis 19 , as knockout of the target gene in all sets of chromosomes would be needed for loss-of-function analysis.
The availability of genomic and transcriptomic data have provided the basis for reverse genetic studies [5][6][7][8][9]13,[20][21][22] . Virus-induced gene silencing (VIGS) has been used extensively as a tool for reverse genetic study in a wide range of plant species to characterize gene functions or for high throughput phenotypic screening of genes [23][24][25][26][27][28][29][30][31][32][33][34] . Instead of complete knockout of a gene, the VIGS system knocks down a gene at the RNA level thereby overcoming polyploidy concerns 30 . Such a system offers an attractively rapid method for loss-of-function assay 23,35 in banana by circumventing the laborious transformation method. The ease of the VIGS system has led to the development of several VIGS vectors being used for gene function analysis or for high throughput phenotypic screening of genes in dicotyledonous plants. However, less vectors are available for studies in monocotyledonous plants 26,29,32,36-39 . Cucumber mosaic virus (CMV) is known to infect a wide host range of more than 1200 species in over 100 plant families, including banana [37][38][39] . CMV is the type member of the genus Cucumovirus and consists of three positive sense single-stranded genomic RNAs (RNA1, 2, and 3) that expresses five proteins (see Supplementary  Fig. S1) [40][41][42] . RNA1 and RNA2 encode the 1a and 2a proteins, respectively, involved in viral genome replication 43 . Additionally, RNA2 contains the 2b gene, which encodes the suppressor of RNA silencing 41,44 . The 2b has been reported to suppress RNA silencing through binding to short-interfering RNAs or double-stranded RNAs 45,46 , and impairing Argonaute 1 and 4 protein activities by direct interaction [47][48][49][50][51][52] . Previous studies indicate that without functional 2b gene, CMV systemically infects plants with mild symptoms 53,54 . RNA3 encodes the 3a movement protein and coat protein (CP). The CMV CP is a multifunctional factor that has roles in viral packaging as well as in systemic movement 42 . CMV has been successfully deployed as a VIGS in dicotyledonous plants such as Nicotiana benthamiana 55 , Glycine max 56 , Antirrhinum majus 57 , Solanum lycopersicum, and Capsicum annuum 58 , as well as monocotyledonous maize 59 ; whether CMV can induce silencing in the monocotyledonous banana remains unknown.
It has been demonstrated that the successful development of VIGS vector largely depends on the virus strain 26,55,59 . Previously, we reported isolation of number of CMV isolates from bananas grown in the field 60 . Subsequently, by selecting a naturally banana-infecting CMV isolate and improving plant infection rates we successfully established a robust VIGS-based system in bananas. The developed tool will greatly expedite gene validation and selection prior to use in transgenic crops.

Development of efficient CMV 20 vector inoculation methods on banana by
pseudostem-rhizome junction injection of Agrobacterium. Despite successfully agro-infiltrating CMV 20 vectors in N. benthamiana, we were not able to transmit CMV 20 vectors in banana using a regular agro-infiltration method. Therefore, we optimized an agroinfection method for banana. We used inoculation buffer that has been reported to enable high inoculation efficiency 62,63 , and used needled injection instead of needleless infiltration of Agrobacterium-mixture to minimize the wounding as we previously demonstrated on orchids 32,64 . In addition, we tested agro-injection or CMV 20 vector on different banana tissues (Fig. 2d). In our repeated sets of experiments, 95% (41 of 43 plants) of bananas inoculated with CMV20-R2E at the pseudostem-rhizome junction showed severe viral symptoms, similar to the slender leaves previously described 60 , starting at 25 dpi to 36 dpi (Fig. 2e). In addition to symptom observation, qRT-PCR was used to detect CMV in ten randomly selected plants that had been inoculated with CMV20-R2E, which confirmed consistent CMV infection.

CMV 20-vectors induce generation of siRNA.
To further confirm that reduced expression of GSA and PDS RNA was due to RNA interference caused by VIGS, small RNA-seq was conducted using mock, CMV20-35S, CMV20-MaGSA, and CMV20-MaPDS agro-inoculum inoculated plants. Short sense and antisense reads of 21-24 nt were mapped to the cDNA sequences of MaGSA and MaPDS (Fig. 4). Although short sense and antisense reads that mapped to either MaGSA or MaPDS can be identified in all plant groups (mock, CMV20-35S, CMV20-MaGSA, and CMV20-MaPDS agro-inoculum inoculated plants), the relative abundance of sense and antisense reads of MaGSA and MaPDS was significantly higher in CMV20-MaGSA and CMV20-MaPDS bananas compared to control plants (mock and CMV20-35S plants), respectively. In CMV20-MaGSA agroinfected bananas, the small RNA (21-24 nt) abundance mapped to MaGSA was 1052 fold higher than the mock and 446 fold higher than the CMV20-35S agroinfected bananas (see Supplementary Table S1). In CMV20-MaPDS agroinfected bananas, the small RNA (21-24 nt) abundance mapped to MaPDS was 1195 fold higher compared to mock and 451 fold higher compared to CMV20-35S agroinfected bananas (see Supplementary Table S1). These results indicate the target specificity of the established VIGS using CMV 20 as viral vector.

Discussion
Here, we report the development of a robust and efficient CMV-VIGS system for bananas and demonstrated that this approach enabled efficient silencing of genes using PDS and GSA. The developed CMV 20 vector allows silencing of targeted genes of interest with high infection rate and mild disease symptoms. The success of our VIGS system in banana may be partly attributed to using a naturally banana infecting CMV isolate for the construction of banana VIGS system, as well as overcoming the difficulties of CMV 20 inoculation on banana using a pseudostem-rhizome junction needled agroinjection system (Fig. 2d).
In our study, we tested a number of methods to inoculate CMV 20 empty vector or CMV 20 vector with a targeted silence gene insert. While agro-inoculation systems have been widely adopted for virus delivery into plants [65][66][67] , inoculation of plant viruses with an agrobacterium-assisted system on Musa plants has not been successfully established prior to this study. Compared to inoculating CMV 20 on N. benthamiana, the thicker banana leaves, latex, and other inhibitory substances present in bananas add to the challenge of establishing an efficient viral inoculation system 68,69 . Of several inoculation methods, we were able to obtain a desirable infection rate (Table 1) by replacing regular needleless agro-infiltration methods with a syringe and fine needle for inoculum injection. Such an approach may allow minimal wounding on banana, and injecting an Agrobacterium mixture carrying CMV 20 to the center of banana pseudostem-rhizome juncture, likely allowed virus infection to take place on the meristem or yet-to-emerge leaves before the true banana leaf is developed. Interestingly, it has been previously reported that 2b protein-defective mutant is unable to invade the shoot apical meristem 70 . This may partly explain why CMV 20 accumulation is significantly higher in CMV20-R2E agroinfected plants compared to CMV20-GSA or CMV20-PDS.
The GSA and PDS genes are commonly used as phenotypic markers for VIGS vector development in many plant species 59 . Using CMV 20-based VIGS vector, silencing MaGSA and MaPDS in banana resulted in a chlorotic or photobleaching phenotype, respectively, that was observed starting 30 dpi and persisted until 63 dpi. We speculate that the meristem tissue is vulnerable to transformation of agrobacterium, and serves as the initial infection site for CMV. As no vascular system in the meristem is available for virus movement, virus infection is confined to the meristem, and VIGS phenotype is only visualized with the development of the leaves. Moreover, the visual phenotype was observed in 77% or 85% of the infected bananas inoculated with CMV20-MaGSA agro-inoculum or CMV20-MaPDS agro-inoculum, respectively (Experiment 3, Table 1). We found initial temperature set at www.nature.com/scientificreports www.nature.com/scientificreports/ 25 °C for two weeks followed by constant 28 °C provided the best VIGS results compared to constant temperature maintained at 28 °C. It has been reported that lower temperature during inoculation may increase the transformation efficiency of Agrobacterium 71 . The temperature control allowed us to achieve a balance for plant growth and VIGS efficacy.
Small RNA sequencing from mock, CMV20-35S agro-inoculum, CMV20-MaGSA agro-inoculum, and CMV20-MaPDS agro-inoculum inoculated banana revealed that higher levels of small RNAs were mapped to the targeted gene and corresponded to the reduction in the mRNA transcript level of the targeted gene, and this higher level of MaGSA and MaPDS small RNAs are found in both the sense and anti-sense orientation indicating that double stranded small RNA is formed and induced post-transcriptional silencing response in this VIGS system.
In this study, the CMV 20-based VIGS vector developed effectively silenced endogenous genes in both N. benthamiana and banana, demonstrating that CMV 20 is able to induce silencing in both dicots and monocots, respectively. We demonstrated that the combination of the developed CMV 20 vector and inoculation strategy enabled silencing of the targeted gene of interest in banana with high infection rate and silencing efficacy. Although VIGS may be limited by the duration of silencing or inherently having to observe plant phenotype under virus infection, the availability of the system developed here may enable previously unavailable rapid functional analysis of genes in banana and expedite a more complete understanding of different biological questions in this important staple and cash crop.  Supplementary   Table S2. The full-length cucumber mosaic virus (CMV20) cDNA of RNAs 1, 2, and 3 (NCBI accession number: MH709142-MH709144) were first cloned into pGEM-T vectors (Promega, Madison, WI, USA) and designated pCMV20-R1, pCMV20-R2, and pCMV20-R3. Total RNA extracted from CMV20 infected N. benthamiana using TRIzol reagent (Thermo Fisher Scientific, Waltham, MA, USA) according to the manufacturer's instructions, was used as template for synthesis of first strand cDNA of CMV 20 RNA1 by use of CMV 3′ primer and M-MLV Reverse Transcriptase (Promega). Two overlapping CMV20 RNA 1 fragments were PCR amplified using primer pairs R125 (with T 3 promoter)/YR1 1646RE and YR1 1584FR/CMV3′. The two amplified fragments were cloned into pGEM-T (Promega). The two plasmids were digested with EcoRV and NotI and separated on 1% agarose gel. The 4.6 and 1.7 kb fragments were gel purified and ligated to generate pCMV20-R1.

Construction of infectious CMV20-derived vectors. All primers used are listed in
The construction of pCMV20-R2 is similar to construction of pCMV20-R1 except primer pairs R125 (with T 3 promoter)/YR2 1720RE and YR2 1536FR/CMV3′ were used instead to amplify two overlapping CMV fragments for cloning to pGEM-T (Promega), and both cloned plasmids were digested with XhoI and SpeI then, separated on 1% agarose gel. The 4.5 kb and 1.5 kb fragments were gel purified and ligated to generate pCMV20-R2.
An AfeI restriction enzyme site was added on the 2b ORF to generate a gene insertion site similar to the method previously described 55 . Two fragments were amplified by PCR with pCMV20-R2 used as template and primer pairs CMV2-2225F/CMV2-2663AfeIR and CMV2-2663AfeIF/CMV2-294R. The amplified fragments were gel extracted and mixed as templates for overlapping extension PCR reaction by use of primer pairs CMV2-2225F/CMV2-2949R. The amplified fragments were gel extracted and digested with SfoI and AvrII. The digested fragments were ligated into pCMV20-R2 that had been digested with SfoI and AvrII to generate pCMV20-R2E.
For bananas, preparation of the inoculum was similar to that previously described. For inoculation of banana, aliquots of 400 μl of Agrobacterium mixture was injected into pseudostem-rhizome junction of banana plants (1-month after transplanting tissue culture plantlets to soil pots) using a 0.5-ml sterile syringe.

RT-PCR and qRT-PCR analysis.
For N. benthamiana, total RNAs were extracted using TRIzol Reagent (Thermo Fisher Scientific) from 1 mg of leaf sample, and treated with Turbo DNAse (Thermo Fisher Scientific) to remove genomic DNA. cDNA for qRT-PCR was synthesized from 1 μg of DNA-free RNA and oligo (dT) using PrimeScript RT Reagent Kit (Takara Bio), following the manufacturer's instructions. For banana samples, total RNAs were extracted using Purelink Plant Reagent (Thermo Fisher Scientific), and the synthesis of cDNA followed the same instructions as N. benthamiana.
For detection and quantification of CMV20 in both N. benthamiana and banana, primers pairs CMV20CP-F/R that targeted CMV20 coat protein were used. All qRT-PCR was carried out using SYBR Premix EX Taq II Kit (Takara Bio) with ABI Prism 7500 sequence detection system (Applied Biosystems). Actin of N. benthamiana or banana was used as a control to calculate the relative CMV20 level.
For qRT-PCR assay of silencing experiments, gene-specific primers were designed to prime outside the region targeted for silencing. For qRT-PCR of N. benthamiana PDS or GSA, primer pairs NbPDS-qF/qR or NbGSA-qF/qR were used, respectively. N. benthamiana actin (NCBI accession number: JQ256516.1) was used as an internal control using primer pairs NbActin-qF/qR. For banana PDS or GSA, primer pairs MusaPDS2-qF/ qR or MusaGSA-qF/qR were used, respectively. Banana actin (NCBI accession number: XM_009412781.2) was used as in internal control using primer pairs MusaActin-qF/qR. RNA derived from mock-or 35S control vector-inoculated plants, as specified in the figures, was set as 1 for relative quantitative analysis.
Small RnA sequencing and small RnA-Seq raw read processing. Total RNA of mock, CMV20-35S, CMV20-MaGSA, and CMV20-MaPDS were extracted from banana using PureLink Plant Reagent (Thermo Fisher Scientific) and sent underwent small RNA sequencing conducted by Biotools (Taipei, Taiwan). In brief, the small RNA was size excised from agarose gel and used as input material for a small RNA library construction using NEBNext Multiplex Small RNA Library Prep Set for Illumina (New England Biolabs, Ipswich, MA, USA) according to the manufacturer's protocol. The libraries were pooled and sequenced on NextSeq 500 (Illumina, San Diego, CA, USA), and the sequenced data were uploaded into the NCBI Sequence Read Archive (Accession number: SRP159135). For raw read processing, the FastX-Toolkit v.0.0.14 was used to trim the adapter and for quality control. The trimmed reads of 18-28 nt were quality-filtered with patterns [-q 30 -p 90]. Next, the trimmed reads were mapped to the banana genome and cDNA sequences retrieved from the banana genome hub (Musa acuminata DH-Pahang version 2; banana-genome-hub.southgreen.fr) by BOWTIE 1.2.1.1 with 1 mismatch. The reads with >20 genomic hits or mapped to the mitochondria genome, chloroplast genome from GenBank (HF677508.1), non-coding RNA from Rfam 13.0, such as ribosomal RNAs, transfer RNAs, small nuclear RNAs and small nucleolar RNAs, were discarded. Reads derived from the transcript of MaPDS and MaGSA sequences were identified using BOWTIE 1.2.1.1 with 1 mismatch.