Susceptibility to South African cassava mosaic virus is Associated with a RING Finger E3 Ubiquitin Ligase

Background The ubiquitylation of proteins is reprogrammed by plant geminiviruses which alter the ubiquitin proteasome system (UPS) to fully infect the host. A RING Finger E3 Ubiquitin Ligase (MeE3L) is located on a major cassava mosaic disease resistance-associated quantitative trait locus. Here, we examine the genetic structure and relative expression of MeE3L (native and gene-edited mutant), and determine how MeE3L affects geminivirus South African cassava mosaic virus (SACMV) DNA accumulation. analysed ΔΔ Genetic and sequences were analysed for and for construction of a phylogenetic using the Maximum Likelihood method and

, and one such process is ubiquitination (Kumar, 2019). They interfere with host UPSs to redirect protein metabolism, alter defense processes (Ramesh et al., 2017) and even suppress the plant's response to insect vectors (Li et al., 2019).
Geminiviruses have been known to inhibit proteasomal degradation of the viral silencing suppressor C2 protein, simultaneously target multiple E3 ligases, alter SUMOylation when the Rep protein interacts with SUMO-conjugating enzyme E1, and degrade of the movement protein (MP) (Alcaide-Loridan and Jupin, 2012).
One of the important geminivirus-induced diseases is Cassava Mosaic Disease (CMD), which is caused by a number of cassava mosaic begomovirus species, and produces foliar symptoms that include reduced leaf size, causing production of few or no tubers by the cassava (Manihot esculenta Crantz) plant (Alabi et al., 2011). CMD can result in yield losses of up to 100% in Africa alone (FAO, 2013), with major regional pandemics causing major food security destabilisation such as occurred in East and Central Africa in the 1990s (Legg and Thresh, 2000;Legg et al., 2006). Certain cassava genotypes exhibit resistance or tolerance to cassava mosaic begomoviruses, such as tropical M. esculenta 3 (TME3) and Tropical Manihot Series (TMS) 96/0023, whereas others (T200, TMS 8017) are susceptible to the virus and do not recover from infection (Rogans et al., 2016;Fondong and Rey, 2018). CMD resistance can be polygenic as is the case for CMD1 in wild cassava (Manihot glaziovii), reportedly monogenic dominant as in CMD2 in certain West African cassava landraces (Akano et al., 2002), or a cross of CMD1 and CMD2 to produce CMD3 (Lokko et al., 2006). Traditionally, CMD-resistant cultivars have been bred by introgression and marker-assisted breeding of resistant landraces with preferred cassava cultivars (Jennings, 1994;Legg and Thresh, 2000;Okogbenin et al., 2007;Ceballos et al., 2015;Nzuki et al., 2017).
Genetic modi cation of cassava for introduction of resistance traits has been hampered by the plant's recalcitrance to transformation (Zainuddin et al., 2012;Lentz et al., 2018). It has been posited that the use of 21st century genetic engineering (GE) or editing techniques such as arti cial miRNA, transactivating siRNA, and Clustered Regularly Interspaced Short Palindromic Repeats (CRISPR), to improve native antiviral mechanisms, can confer virus resistance without adverse effects on the plant (Vanderschuren et al., 2006;Fondong & Rey 2018). However, this will depend on an intimate understanding of the molecular processes of existing innate antiviral mechanisms, such as the degradation of viral proteins that is directed by the UPS. In this respect, CRISPR has the upper hand over other modern GE techniques because of its e ciency in silencing genes, which simpli es the process of identifying key genes involved in resistance/tolerance by analysing gene expression during virus infection (Doudna and Charpentier, 2014;Mao et al., 2018).
Transcriptome studies during infection have shown that the geminivirus, South African cassava mosaic virus, and Cassava brown streak virus (ssRNA ipomovirus) induce differential expression of UPS proteins, including four E3 ligases (Allie et al., 2014;Maruthi et al., 2014). A genome-wide association study found one E3 ligase (Manes.12G069400/cassava4.1_026906 m.g) among 105 gene candidates associated with the CMD2 locus (Wolfe et al., 2016). This E3 ligase (hereafter designated MeE3L) is located on a major CMD resistance-associated QTL (Lozano et al., 2015) and has been implicated in cassava's response to drought conditions (Liao et al., 2016;Patanun et al., 2016). To test whether MeE3L is involved in the response of cassava to South African cassava mosaic virus (SACMV), we set out to determine how the transient expression of MeE3L and SACMV replication in cassava protoplasts from model cultivar (cv.60444), susceptible (T200) and tolerant (TME3) landraces (Kuria et al., 2016) are affected by CRISPRmediated gene editing of MeE3L.
Methods And Materials 1. CRISPR-Cas9 vector construction Two genomic gRNA targets were identi ed using the CRISPOR version 4.7 web-based program (www.crispor.tefor.net;Haeussler et al., 2016) and used in designing duplex sgRNA that included an Arabidopsis thaliana U6-26 promoter, the sgRNA scaffold and terminator for each gRNA sequence. Synthesis of the duplex sgRNA was outsourced to Inqaba Biotech (Pretoria, South Africa) and the duplex was cloned into the pCambia1380 vector. A Cas9 insert (Tobacco mosaic virus promoter + Cas9 gene + eGFP gene + Hsp terminator) from the pl1m-f2-p35s-cas9-egfp-nucleo-thsp Golden Gate vector was cloned into the pCambia1380-gRNA construct. The construct was con rmed by restriction digestion and sequencing.

Protoplast transfection
Approximately 15 µg of the CRISPR construct and 4 µg each of pBIN19-SACMV-DNA-A and pBIN19-SACMV-DNA-B infectious clones (Berrie et al., 2001) were mixed with 1 mL of protoplasts and 25% polyethylene glycol 4000 (PEG 4000), and incubated at room temperature for 20 min. The mixture was gently diluted with 3 volumes of W5 solution (154 mM NaCl, 125 mM CaCl 2 , 5 mM KCl, 2 mM MES, pH 5.8) and centrifuged twice at 100 g for 2 min. Protoplasts were resuspended in 300 µL of WI solution (4 mM MES,0.5 M mannitol,20 mM KCl,pH 5.8) (Yoo et al., 2007) and incubated overnight in the dark at room temperature to induce gene expression. The expression of eGFP was checked using uorescence microscopy to con rm protoplast transformation. Protoplasts were washed with CPW9M medium at 24 h post-transfection. 4. Mutagenesis, viral load and gene expression assays DNA was extracted from the transformed protoplasts 24 h post-transfection (hpt) using QIAzol Lysis Reagent according to a user-developed protocol (www.qiagen.com/it/resources/) (Qiagen, Maryland, USA). Quantitative PCR (qPCR) for SACMV relative viral load quantitation using DpnI-digested (ThermoFisher Scienti c, Massachusetts, USA) DNA as template was performed in triplicate using forward (5GGCTAG ⊤ ℂCGGA ⊤ ACAT3) and reverse (5GAC ∀ GGACGGAGACAℂ3) primers, and 18S rRNA as the reference gene. The exonic region of MeE3L was ampli ed using the Phusion U Green Hot Start DNA Polymerase (ThermoFisher Scienti c) with forward (5CGCGCAGA ⊤ C ∀ GC3) and reverse (5TGTℂACATGG ∀ TG ∀ AG3) primers. Sequencing of amplicons on the ABI 3500XL Genetic Analyzer was outsourced to Inqaba Biotec (Pretoria, South Africa), and sequencing results were analysed for variation and alignment using TIDE web tool (https://tide.deskgen.com/; Brinkman et al., 2014) and MAFTT version 7 (https://mafft.cbrc.jp/alignment/software/; Rozewicki et al., 2019) before being employed as query terms for protein structure and binding prediction using the I-TASSER On-line Server (https://zhanglab.ccmb.med.umich.edu/I-TASSER/; Yang and Zhang, 2015) and protein similarity search in the Protein Data Bank (Berman et al., 2000). Analysed sequences were used as query terms for BLAST (https://blast.ncbi.nlm.nih.gov/; Zhang et al., 2000) searches of the SACMV genome (AF155806/7) and to construct a phylogenetic tree using the MEGA X software based on the Maximum Likelihood method and Jones-Taylor-Thornton (JTT) matrix-based model (Kumar et al., 2018).
RNA was extracted from the transformed protoplasts 24 hpt using QIAzol and according to the manufacturer's (Qiagen, Maryland, USA) protocol. First strand cDNA synthesis using RNA as template was performed using the RevertAid H Minus First Strand cDNA Synthesis Kit (ThermoFisher Scienti c, Massachusetts, USA). Reverse transcriptase qPCR for MeE3L relative expression quantitation was performed in triplicate with cDNA as template using the Maxima SYBR Green/ROX qPCR Master Mix (2X) (ThermoFisher Scienti c, Massachusetts, USA) with forward (5CGCGCAGA ⊤ C ∀ GC3) and reverse (5 TGTℂACATGG ∀ TG ∀ AG3) primers according to the manufacturer's protocol using 18S rRNA was the reference gene for qPCR. Relative expression and viral load of MeE3L and SACMV, respectively, were determined using the 2 −ΔΔT method (Livak and Schmittgen, 2001).

Isolation and transformation of cassava protoplasts
The in vitro growth of the cassava plantlets was conducted under controlled, sterile conditions and only young, expanded leaves were used as donors (Fig. 1A). Different sizes (~ 15-35 µm) of cassava protoplasts of round and irregular shape, containing one vacuole and chloroplasts, were observed ( Fig. 1B-D). The viability of protoplasts was at least 85% as shown by staining with Evans' Blue Dye ( Fig. 2A-C). Protoplast yields were 4.90-6.36 × 10 6 /g fresh weight (FW) (Additional File 1). The integrity of isolated protoplasts was analysed by ow cytometry prior to transfection (Fig. 2D-F). The presence of protoplasts as well as irregularly-shaped debris outside the gated area was detected. Approximately 10 4 protoplasts of each cassava genotype were transfected with 15 µg eGFP-tagged CRISPR construct and/or 4 µg SACMV infectious clones using PEG-mediated transformation. Stability of transient expression was veri ed by detection of eGFP expression 24 hpt, showing that at least 90% of protoplasts had been successfully transformed ( Fig. 2G-H). 2. Structure and phylogenetic analysis ofMeE3Lin model, susceptible and tolerant cassava genotypes PCR ampli cation and sequencing of the MeE3L partial transcript from leaves revealed that the T200 landrace homolog is slightly longer than the TME3 homolog (Fig. 3B). Sequencing of the exonic region of T200 MeE3L showed that this is due to a 53 bp insertion mutation of 9 TGAGAA nucleotide repeats that are absent in cv.60444 and TME3 (Fig. 3A). The resulting frameshift introduces a stop codon at amino acid 141 (Fig C-D). Computational analysis revealed that the truncated T200 MeE3L homolog is structurally distinct from the reference (AM560-2), cv.60444 and TME3 MeE3L homologs (Fig. 3E). It is also not signi cantly similar in structure to any protein in the Protein Data Bank (Berman et al., 2000), and it is not con dently predicted to bind any ligands. The other three (AM560-2, cv.60444 and TME3) homologs are structurally closest to the cIAP1 inhibitor of apoptosis protein which contains a RING domain with E3 ligase activity for autoubiquitination and modulates cell death (Dueber et al., 2011). An analysis of the phylogeny of E3 ligase homologs in plants ( Fig. 3F) reveals that although the MeE3Ls share Hevea brasiliensis as a common ancestor, the T200 homolog is signi cantly more evolutionarily distant from AM560-2 than the TME3 and cv.60444 homologs.
Primary structure, secondary structure and phylogenetic analysis ofMeE3Land/or its protein product. Sequence alignment and agarose gel resolution ofMeE3Lpartial gene and partial transcript respectively show a 53 bp insertion mutation in the lengthier T200 homolog that is absent in the AM560-2, cv.60444 and TME3 (Kumar et al., 2018). Bootstrap support was calculated from 1000 replicates. The tree is drawn to scale, with branch lengths measured in the number of substitutions per site.

Relative accumulation of SACMV DNA and relativeMeE3Lexpression in transformed cassava protoplasts
Relative SACMV DNA accumulation in wild-type protoplasts was highest in T200 at 8-fold, compared to 3fold in cv.60444 and TME3 at 24 hpt. Gene editing of MeE3L by CRISPR (Additional File 1) resulted in a substantial increase in accumulation of SACMV in T200 and TME3 (12-fold and 4-fold respectively) whereas only a 1-fold reduction was seen in cv.60444 (Fig. 4A). MeE3L expression is signi cantly upregulated with a log fold change of 1 and 9 in T200 and TME3 SACMV-infected protoplasts, respectively. Gene editing of MeE3L signi cantly upregulated expression of the MeE3L variant in TME3 but had no signi cant effect on T200 MeE3L expression under the different transformation conditions. No signi cant change in MeE3L expression was observed under all conditions in cv.60444 (Fig. 4B).

Discussion
1. African cassava landraces susceptible and tolerant to SACMV are amenable to enzymatic protoplast isolation and PEG-mediated transformation Protoplasts were chosen for this transient gene expression study because they can conveniently and e ciently be transformed with several DNA constructs simultaneously, and they allow higher resolution imaging compared to cells in intact tissue (Faraco et al., 2011). Additionally, they can be used for high throughput e cient screening of many candidate genes, and those that show an effect can then be silenced by CRISPR or virus induced gene silencing (VIGS) in planta, which are considerably longer (3-4 months) and more complicated procedures for the non-model host cassava. Leaf mesophyll was used as the source for protoplasts because it provides functional information about proteins expressed in other plant organs (Faraco et al., 2011). Round and irregularly-shaped protoplasts of different sizes were observed ( Fig.1B-D), although generally leaf mesophyll protoplasts should be uniformly spherical (Wu et al., 2017).
A previously determined enzyme concentration (1.6% cellulase, 0.8% macerozyme) that is suitable for obtaining the optimum number of viable protoplasts was used for leaf digestion (Wu et al., 2017). The viability of protoplasts in this study was at least 85% although cassava protoplast viability of up to 95% has been reported (International Plant Research Institute, 1984). The long digestion period (16 h) was ruled out as the cause of death for ~15% of protoplasts as this reportedly does not induce serious damage in protoplasts (Tang, 1982). It has been reported that micro-propagated plants grown in vitro lack epicuticular wax and thus allow rapid enzyme penetration (Kumar and Rao, 2018). Cassava, however, has a thick epicuticular layer which necessitates the long digestion period of 16h compared to 0.3-1h for Arabidopsis (Wu et al., 2009). Macerating enzymes such as macerase are known to cause wound reactions in protoplasts because of their degradation of the cell wall, which may lead to necrosis (Ishii, 1988), while cellulase is known to exert inadequate enzymatic activity at low concentrations and higher concentrations have no bene t or detriment (Uchimiya and Murashige, 1974). Therefore, a balance between digestion enzyme concentration and viability is essential in order to obtain the optimum yield of viable good quality protoplasts. The difference in cassava protoplast viability from previously reported percentages possibly may vary according to cassava variety, and in this case could be due to the particular physiological characteristics of the African cassava landraces from which protoplasts were derived in this study.
Although lower than the previously reported yields of 4.4x10 7 protoplasts/g FW leaves from M. esculenta cv. South China 8 (Wu et al., 2017) and 1.9x10 7 protoplasts/g FW leaves from M. esculenta cv. M. Thai 8 (Anthony et al., 1995), the protoplast yields in this present study (4.90-6.36x10 6 /g FW) were su cient to provide the recommended number of protoplasts (10 4 -10 7 ) required for each transfection (Yoo et al., 2007). Both pre-treatment of leaves in the dark (24-72 h) (Shahin and Shepard, 1980) and vacuum in ltration before enzyme digestion (Nanjareddy, 2016) were shown to intensify enzyme penetration in bean leaves. We found that pre-treatment resulted in release of much undesirable plant debris alongside protoplasts, and that vacuum in ltration did indeed help increase protoplast yield. Flow cytometry indicated a high concentration of protoplasts compared to irregularly-shaped debris. Based on microscopy images of puri ed protoplasts (Fig.1B-D), the irregular debris outside the gates was deemed to be free chloroplasts, plasmolysed cells, undigested cell wall fragments and other aggregates arising from the long digestion period of leaf material.
PEG-mediated plant protoplast transfection with plasmid DNA is a well-established procedure (Hayashimoto et al., 1990;Locatelli et al., 2003) and a popular protocol uses 10 μg DNA to transfect 2x10 4 Arabidopsis protoplasts (Yoo et al., 2007). At least 5 µg of plasmid DNA have previously been used to transform 10 6 tobacco protoplasts with African cassava mosaic virus (ACMV) (Ermak et al., 1993) and Cowpea mosaic comovirus (Wellink et al., 1993). Highly e cient co-expression of multiple constructs has been reported (Chen et al., 2006;Walter et al., 2004). Virus infectious clones have also been used in conjunction with other plasmid constructs for co-inoculation of plant protoplasts. Nicotiana tabacum protoplasts have been co-transformed with 5 µg eGFP construct, 3 μg siRNA and 4 μg each of ACMV or East African cassava mosaic virus (EACMV) with DNA/RNA extraction at 36 and 48 hpt (Vanitharani et al., 2003). To our knowledge, the present study is the rst report of co-transformation of cassava protoplasts with a CRISPR construct and geminivirus infectious clones.
2. The M. esculenta T200 MeE3L encodes a truncated RING-less protein due to a nonsense mutation The frameshift resulting from the insertion mutation in T200 MeE3L introduces a stop codon upstream of the RING domain, thus encoding a truncated protein in which the C3HC4-type RING nger motif is absent. Essentially, the E3 ligase domain in the T200 MeE3L would not be translated because of this mutation, making its potential protein product non-functional with respect to this E3 ligase activity. There is evidence that intronic GAA repeats impede transcription via the assembly of an inaccessible chromatin structure (Punga and Buhler, 2010). However, we ampli ed the T200 MeE3L exon from cDNA, showing that this gene is transcribed. It is more likely that the premature translation termination instruction makes the T200 MeE3L a possible candidate for nonsense-mediated mRNA decay (NMD) whereby mRNAs with long 3′-untranslated regions (UTRs) are targeted as substrates for degradation (Hori and Watanabe, 2005). In Arabidopsis, for example, evolutionarily conserved protein factors are deployed by the NMD system where the 3′-UTR is less than 300 nucleotides (Hori and Watanabe, 2007).
It is proposed that the loss of a substantial portion of the T200 MeE3L C-terminal region would alter the spatial chemical conformation necessary for any other interactions (such as binding the ubiquitinconjugated E2 and substrate) to occur. Furthermore, the possible non-functionality of the T200 MeE3L due to the nonsense mutation would suggest that the susceptibility of T200 to SACMV might be linked the loss of the RING domain since this is where the E3 ligase activity resides. However, susceptible cv.60444 encodes a functional RING domain in its MeE3L, suggesting that susceptibility to the virus is subject to at least another pathway besides that in which MeE3L is involved, or is under polygenic control.
The structural similarity between between MeE3L and cIAP1 is signi cant because IAPs are known for contributing to disease initiation, maintenance or progression by controlling protein stability (Silke and Meier, 2013). Binding of a substrate to cIAP1 leads to activation of the E3 ligase, leading to activation of downstream caspases which set cell death in motion (Berger et al., 2006). Viruses tend to inhibit host apoptotic responses in a bid to prolong cell viability and facilitate replication by targeting caspases for suppression. However, because caspases also play a role in cell proliferation and differentiation, some viruses stabilise caspase activity to maintain cell survival of the cell at a particular differentiation stage, or to facilitate replication by suppressing early apoptosis (Lee, Liao and Lin, 2005;Best, 2008). Reduced viral yield has been reported in a case where speci c caspases were inhibited, suggesting that these caspases expedite viral replication (Best, Wol nbarger and Bloom, 2002). It is suggested that silencing of the MeE3L RING domain would promote caspase activity in the susceptible T200 genotype, thereby enhancing pathogenicity by aiding SACMV replication.
The AM560-2, TME3, and cv.60444 cassava genotypes have previously been shown to cluster together under their nearest ancestor, Hevea brasiliensis (Bredeson et al., 2016). The evolutionary history of T200, a southern African landrace, is unknown and the phylogenetic analysis suggests that the T200 MeE3L evolved after the cv.60444 and TME3 variants. While it is known that wild plants in natural ecosystems may co-evolve with their virus partners, information regarding how viruses apply selective pressure to increase plant susceptibility is not known. A study of Drosophila and its host-speci c viruses found that coevolution may cause sustained genetic variation in susceptibility. This may explain why a southern African cassava landrace is highly susceptible to SACMV that appears to have migrated south from its origin, suspected to be in east Africa or the south-west Indian Ocean islands such as Madagascar (De Bruyn et al., 2016) that were once geographically attached to the African continent (Lefeuvre et al. 2007). South African cassava mosaic virus is a recombinant between East African cassava mosaic virus and two other unknown geminiviruses which contributed the AC4 and IR regions (Berrie et al., 2001), and moved southwards into Mozambique, Zimbabwe and South Africa where it may have encountered the T200 landrace. Subsequent to its rst discovery in South Africa, it has been reported in Zimbabwe (Briddon et al. 2004) and Madagascar (De Bruyn et al., 2016). It is suggested that the T200 landrace and SACMV may still be in the process of co-adaption, which would be why T200 exhibits extreme susceptibility to SACMV. It is known that infection with a new recombinant begomovirus requires hosts to adjust to minor or major differences in virus-host interactions (Montes et al., 2019). We speculate that maybe the MeE3L is either a paralog in T200 or it was introgressed from a wild relative in southern Africa.

MeE3L affects SACMV DNA accumulation in a genotypedependent manner
Quantitative PCR is a well-established method for precise quantitation of viral DNA amount in infected tissue and it requires a host reference gene with stable expression patterns under experimental conditions as the internal control for correct data normalisation (Moreno et al., 2011). Quantitative detection of African cassava mosaic virus and East African cassava mosaic virus using real time -qPCR has been reported (Otti et al., 2013) and SACMV load, in particular, has been assayed in planta in Arabidopsis (Pierce and Rey, 2013) and cassava (Allie, 2014). Replication of the geminivirus, Cassava brown streak virus, in cassava leaf mesophyll protoplasts has been assayed at 6 hpt (Anjanappa et al., 2016) and it has been reported that there was signi cant viral DNA accumulation in tobacco BY-2 protoplasts 36 and 48 hpt by co-inoculating with infectious ACMV and EACMV clones and siRNA (Vanitharani et al., 2003). Our nding that SACMV DNA accumulates in cassava protoplasts correlates well with these reports.
It was expected that viral DNA accumulation would be signi cantly lower in TME3 than in the susceptible cv.60444 and T200 if MeE3L was involved in tolerance since the TME3 landrace is known to have the CMD2 locus. Evidently, CRISPR-induced mutations in MeE3L in cassava protoplasts differentially affect viral DNA accumulation based on the cassava genotype. CRISPR-associated modi cation of the MeE3L structure and subsequent disruption of its activity may enhance SACMV DNA accumulation and pathogenicity in T200 and TME3 by interfering with the UPS-dependent tolerance/resistance response mechanisms of cassava.

Viral activity and gene editing of MeE3L affect the expression of MeE3L
Geminiviruses elude plant defense mechanisms by hijacking and redirecting ubiquitination, and interfering with responses regulated by ubiquitin E3 ligases (including responses to jasmonates, auxins, gibberellins, ethylene,abscisic acid) (Lozano-Duran et al., 2011). It follows then that alterations to E3 ligase genomic sequences may alter E3 ligase expression patterns during viral infection, as viruses are known to modulate RNA levels to enhance infection (Verchot, 2016). Both plant viruses and CRISPR systems are known to induce mutations in the genome (Machida et al., 2004;Cougot et al., 2005;Sander and Joung, 2014), and the employment of both against the MeE3L would provide an indication of how this gene is involved in the plant's response to SACMV.
Results indicate that in TME3 protoplasts, MeE3L expression is upregulated during SACMV infection. The concurrent CRISPR-mediated gene editing of MeE3L and infection with SACMV appears to induce increased expression of the MeE3L, suggesting that the interaction between the virus and MeE3L is based on recognition of the MeE3L's speci c base sequence. Previously, plant E3 ligases have been shown to be induced by viral infection (Lai et al., 2009;Czosnek et al., 2013;Chen et al., 2018) and plant defence elicitors (Libault et al., 2007;Sadanandom et al., 2012). The muted response of the T200 MeE3L to all treatments was expected given its nonsense mutation which silences the RING domain responsible for E3 ligase activity. However, the muted response of susceptible cv.60444 MeE3L was unexpected and suggests that this MeE3L sequence variant is not responsive to SACMV infection. It is possible that downstream plant hormone responses may be linked to MeE3L activity. It is known that geminiviruses interact with plant E3 ligases and induce their up-or down-regulation to promote infection or undergo degradation. Beet severe curly top virus (BSCTV) C4 induces activity of RING E3 ligase, RKP, to interfere with the cell cycle, increase susceptibility and promote infection (Lai, 2009). The C2 protein interacts with CSN5, a component of the SCF E3 ligase complex, to alter CUL1's regulation of plant hormones (Lozano-Duran et al., 2011). An F-box protein adaptor is triggered to overexpress by Tomato yellow leaf curl Sardinia virus (TYLCSV) and hence the virus co-opts SCF E3 ligase activity (Lozano-Duran and Bejarano, 2011). The geminiviral βC1 protein is ubiquitinated by E3 ligase, NtRFP1, for degradation and resultant attenuation of symptoms. Conversely, βC1 is able to interact with NbSKP1 to impair the SCF complex integrity and hence enhance symptom induction by interfering with jasmonic acid and gibberellic acid signalling. Evidently, geminiviruses can usurp the UPS by preventing E3 ligases from targeting viral proteins; or they can redirect the UPS to degrade host resistance-related proteins.
The MeE3L homolog sequences in SACMV-infected protoplasts show that SACMV is actively inducing silencing of the RING domain in order to achieve full infection of the host. The concomitant increase in SACMV DNA accumulation and CRISPR-mediated editing of MeE3L in TME3 shows that the interaction of MeE3L with SACMV is more directed at advancing susceptibility. One other plant E3 ligase, RHF2A, is silenced by TYLCSV to promote infection (Lozano-Duran et al., 2011). It has also been shown that inhibition of a RING E3 ligase impairs plant defence during Cabbage leaf curl virus (CaLCuV) infection (Sahu et al., 2013) and that silencing of tobacco RING E3 ligase, NtRFP1, leads to development of severe symptoms during Tomato yellow leaf curl China virus (TYLCCV) (Shen et al., 2016). The present study provides further evidence that geminiviruses interfere with expression of a plant E3 ligase in order to establish infection, and demonstrates for the rst time that MeE3L in cassava is responsive to a cassavainfecting geminivirus, SACMV.

SACMV's interaction with a tolerant cassava genotype induces numerous mutations in MeE3L
Functions of E3 ligases in regulating immunity systems are orchestrated at the interface of host-virus interactions (Zhang et al., 2018) and some of these interactions occur in the nucleus (Kushwaha et al., 2017). Sequencing of MeE3L from SACMV-infected TME3 protoplasts revealed multiple random single base mutations along the length of MeE3L, which translate to amino acid substitution (Fig. 4E). While these mutations do not alter the reading frame, they are apparently aimed at silencing the whole protein and not just the RING domain. The resulting disordered protein would presumably not only lack RING E3 ligase activity, but also the E2 and substrate binding activity.
Several viral proteins reportedly hinder E3 ligase function by encoding these E3 ligases in their viral genomes (Lai et al., 2009;Rosa-Diaz et al., 2013;Wang et al., 2013;Chen et al., 2018;Lozano-Duran and Bejarano, 2018). However, a blastn search for MeE3L homologs in the SACMV genome did not return any hits, thus ruling out the possibility that SACMV encodes a protein with E3 ligase activity as a strategy to target host proteins for degradation via the RNA silencing pathway. Since E3 ligases are frequently hijacked by different viruses (Jackson and Xiong, 2009), we suggest, based on the MeE3L mutations induced during SACMV infection, that MeE3L is one of the targets of SACMV-induced host reprogramming and plays a crucial strategic role in enhancing susceptibility. This is not a case of viral arrogation of MeE3L by SACMV to redirect the UPS to new targets but rather to inhibit its function. Interestingly, the discovery of these mutations in genomic DNA possibly point to a yet unknown geminivirus-induced host mechanism for genome editing. Some E3 ligases and viral proteins can localise to the nucleus, such as the tobacco E3 ligase, NtHUB1 which has a nuclear localisation sequence, is recruited by geminiviral Rep protein, and co-localises and interacts with the Rep protein to monoubiquitinate cellular chromatin and thus enable infection (Kushwaha et al., 2017). The viral coat protein, CP, also has a nuclear localisation signal, can localise in the nucleolus and nucleoplasm, and facilitates entry of ssDNA into the nucleus (Wang et al., 2017;Kumar, 2019). However, the mechanisms for these processes are yet to be investigated. Geminiviruses are known to induce the expression of genes related to repair of double-stranded breaks (DSBs) and DNA synthesis (Lozano et al., 2015), and to promote somatic homologous recombination (Richter et al., 2014). In spite of this evidence for geminiviral activity at the host genomic level, the mechanisms for these processes have not been fully investigated. However, this evidence provides a measure of support for our nding of gene mutation induction by SACMV in TME3.

The response of MeE3L to SACMV is virus-and host-speci c
Ubiquitin ligases are abundant in plants and provide substrate speci city to target particular proteins. In Arabidopsis alone, RING E3 ligases make up 499 out of over 1,500 E3 ligases (Mazucotelli et al., 2006). A comparison of E3 ligase and E3 ligase complex-associated gene expression during other plant geminivirus infection studies (Additional File 2) was conducted to determine whether MeE3L's response to SACMV is geminivirus-speci c or host-dependent.
In susceptible cassava, E3 ligase expression is downregulated during SACMV infection at early, middle and late time points (12, 32 and 64 days post infection (dpi)) but there is no differential expression of E3 ligases in tolerant cassava at any time point (Allie et al., 2014). However, no differential expression of E3 ligases is recorded during SACMV infection of Arabidopsis which is susceptible (Pierce and Rey, 2013). A study of transcriptomic responses to geminivirus Tomato leaf curl New Delhi virus (ToLCNDV) infection in potato found that ve E3 ligases in the susceptible cultivar and two in the tolerant cultivar are upregulated at 30 dpi (Jeevalatha et al., 2017) while the geminivirus Tomato yellow leaf curl virus (TYLCSV) has been shown to induce upregulation of E3 ligases in susceptible tomato at 42 dpi, except in the case of a CUL1 which is downregulated (Miozzi et al., 2014). A transcriptome study of Arabidopsis during geminivirus CaLCuV infection found that, out of 1570 E3 ligases, 149 were up-regulated and 23 were downregulated (Ascencio-Ibanez et al., 2008;Lozano-Duran et al., 2013). The CaLCuV AC2 protein, in particular, induces downregulation of two E3 ligases in Arabidopsis (Liu et al., 2014). These ndings, together with the current study, prove that plant E3 ligase responses to geminivirus infection are neither uniform nor similar, but they vary according to the speci c geminivirus and host involved in the interaction.
Responses of cassava to the ssRNA potyviruses Cassava brown streak virus (CBSV) and Ugandan cassava brown streak virus (UCBSV) with respect to E3 ligase expression variably show both downregulation and upregulation in the susceptible varieties. Interestingly, there is no differential expression of E3 ligases in resistant cassava varieties except in Kaleso where a CUL1 is upregulated and a RZPF34 is downregulated (Maruthi et al, 2014;Anjanappa et al., 2016;Amuge et al., 2017). This variable expression of E3 ligases with respect to the virus in the same host suggests that while responses to viral infection are host-dependent, they are also modulated according to the particular virus infecting the plant.
It appears that MeE3L, like other plant E3 ligases, is reprogrammed by SACMV in a manner that is virusspeci c and host-dependent. Moreover, current evidence indicates that E3 ligase responses are more pronounced in the virus-susceptible plant varieties. SACMV infection in planta is associated with occurrence of severe symptoms leading to persistent severe infection in susceptible T200 and mild symptoms with recovery at 67 days post infection (dpi) in tolerant TME3 (Allie et al., 2014). The SACMVinduced genetic mutations and differential expression of MeE3L post-infection in TME3 and T200 indicate that it is one of the signi cant genes involved in the plant's response to the virus, and the data shown herein suggest that the MeE3L-SACMV interaction is aimed at silencing the gene to promote viral activity. Although the mechanism for silencing of MeE3L by SACMV is not clear, we propose that the silencing of MeE3L is part of the SACMV-induced reprogramming of the UPS to inhibit ubiquitination and trigger a decrease in global ubiquitinated proteins. The subsequent developmental disturbance linked to reduced caspase activity would contribute to increased symptom severity, which is visible at 32 dpi. At 67 dpi in TME3, the homeostasis of ubiquitinated versus non-ubiquitinated proteins would be restored and the effects of SACMV on the UPS would be counteracted, leading to symptom decrease and plant recovery. Alternatively, it is possible that SACMV suppressors of RNA silencing (VSRs) such as C2 and C4 target E3 ligases, such as MeE3L in T200 and TME3, for silencing, thereby usurping the UPS and contributing to disease symptoms. In TME3, other defence mechanisms such as RNA silencing at 67 dpi (Rogans et al., 2016) and resistance (NLR) proteins (Louis and Rey, 2015) persist, which interfere and reduce the activity and replication of SACMV.

Conclusions
We have developed a simple and fast protocol for CRISPR-mediated transient gene expression assay in cassava protoplasts. While gene editing/knockout in cassava plants takes 6-8 months Availability of data and materials The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.    [Predictions were run on the I-TASSER On-line Server (https://zhanglab.ccmb.med.umich.edu/I-TASSER/; Yang and Zhang, 2015)] (F) Evolutionary analysis of plant MeE3L homologs using Maximum Likelihood method and Jones-Taylor-Thornton (JTT) matrix-based model in MEGA X (Kumar et al., 2018). Bootstrap support was calculated from 1000 replicates. The tree is drawn to scale, with branch lengths measured in the number of substitutions per site.