Characterization of Sirnas Derived from Two Species of Cassava-Infecting Geminiviruses in Nicotiana Benthamiana

The infection dynamics of cassava mosaic geminiviruses was determined in Nicotiana benthamiana. Following inoculation, symptoms were observed at 5 and 10 dpi for African cassava mosaic virus (ACMV-CM) and the Kenyan strain K201 of East African cassava mosaic virus (EACMV KE2 [K201]). Northern and Southern blots indicated higher accumulation of ACMV-CM at 10 dpi compared with EACMV KE2 (K201) correlating with more severe symptoms at this time point. However, unlike for EACMV KE2 (K201), symptom remission characterized by appearance of non-asymptomatic leaves was observed after 20 dpi in plants infected with ACMV-CM. Deep sequencing identified between 969,015 and 1,307,689 total reads of which >80% mapped to the host genome. For both virus species, 22 nt virus derived sRNAs (vsRNAs) reads were the most abundant (34% - 53%) followed by 21 nt (27% - 27%) and 24 nt (9 – 23%) vsRNA reads. EACMV KE2 (K201) vsRNAs were proportionally higher (12.31% and 4.51%) than those derived from ACMV-CM (5.70% and 0.45%) for DNA A and DNA B respectively. vsRNAs were covered the entire virus genome with dense accumulation at regions where the viral ORFs overlapped. Key words; RNA interference, Cassava mosaic disease, Geminiviruses DOI: 10.7176/JNSR/10-8-06 Publication date: April 30 th 2020


Introduction
Eukaryotes utilizes a conserved RNA silencing as a potent defense against invading foreign nucleic acids among other regulatory mechanisms (Yang et al., 2014;Martínez de Alba et al., 2013;Csorba et al., 2009). RNA silencing is induced by recognition and processing of double stranded RNAs (dsRNAs) molecules or hairpin-like RNA secondary structures into 21 -24 nucleotides (nt) primary small interfering RNAs duplexes (siRNAs) . Biogenesis of siRNAs is coordinated by ribonuclease III-like enzymes, termed DICER-like proteins (DCLs) in plants (Pooggin, 2016;. The siRNAs associate with effector molecules to guide specific degradation or translation inhibition of cognate messenger RNA (mRNA) and DNA/histone modification (Carbonell and Carrington, 2015). RNA silencing is amplified by host RNA-dependent RNA-polymerases (RDRs) through generation of secondary siRNA that mediate systemic silencing throughout the plant (Molnar et al., Wang et al., 2010).
The core protein components of RNA silencing have been described and shown to be diverse in different plants species. Four DCLs, ten ARGONAUTE (AGO) family of proteins and six RDRs have been reported in Arabidopsis thaliana (Margis et al., 2006), rice genome encodes eight DCLs, five RDRs, and nineteen AGO proteins (Kapoor et al., 2008), in tomato seven DCL, 15 AGO and six RDR genes have been reported (Bai et al., 2012) whereas maize genome encodes five DCLs, five RDRs, and 18 AGOs (Qian et al., 2011). In A. thaliana antiviral response is mediated through biogenesis of 21 nt, 22 nt and 24 nt siRNAs conditioned by DCL4, DCL2, and DCL3 in a hierarchical manner (Garcia-Ruiz et al., 2010). Redundancy of functions in processing of 21 and 22 nt siRNAs has been reported for DCL4 and DCL2 (Axtell, 2013;Cao et al., 2014). Plants infected with RNA viruses have been shown to accumulate abundant amounts of 21 nt and 22 nt virus-derived small RNAs (Ogwok et al., 2016; whereas plants infected with DNA viruses mostly accumulate 24 nt siRNAs derived from virus genome that direct antiviral RNA silencing (Pooggin et al., 2016;Incarbone and Dunoyer, 2013). However, diverse plant species systemically infected with geminiviruses accumulate different profiles of virus derived sRNAs (Kuria et al., 2017;Rogans et al., 2016;Aregger et al., 2012) therefore, germplasm displaying different genetic backgrounds respond differently to virus infection. defense in Arabidopsis have been reported to involve diverse AGO families including, AGO1, AGO2, AGO5, and AGO10 that act in PTGS targeting RNA viruses Carbonell and Carrington, 2015;Garcia-Ruiz et al., 2015;Ma et al., 2015). On the other hand AGO4, AGO6, and AGO9 preferentially associate with 24 nt siRNAs to mediated methylation viral DNA genome (Havecker et al., 2010). RDR1, RDR2 and RDR6 have been implicated in amplification of virus derived siRNAs to induce antiviral defense (Wang et al., 2010).
Cassava-infecting geminiviruses (Family Geminiviridae, Genus Begomovirus) are majorly transmitted by whiteflies (Legg et al., 2015) and are the causal agent of cassava mosaic disease (CMD). The disease is endemic to cassava production areas of sub-Saharan Africa (Patil and Fauquet, 2009) and has been ranked as the seventh most important viral disease worldwide (Rybicki, 2015). CMD is further exacerbated by recombination and pseudo-recombination of different species leading to evolution of more virulent strains (Legg, and Thresh, 2000;Olsen et al., 1999). Emergence of recombinant strains such as East African cassava mosaic virus-Uganda Variant (EACMV-UG) have been associated with severe CMD pandemic in the 1990s and 2000s (Legg, and Thresh, 2000;Olsen et al., 1999;Zhou et al., 1997). However, the interaction between different cassava-infecting geminiviruses and host plant silencing machinery has not been precisely characterized. Cassava-infecting geminiviruses possess two similar sized but independent single stranded DNA (ssDNA) genomic components designated DNA A and DNA B (Brown et al., 2015;Hanley-Bowdoin et al., 2013). Both viral genomic components are essential for full disease establishment. DNA A encodes four genes in complementary sense orientation involved in replication and transcription and two genes in virion sense orientation involved in virus encapsidation and pathogenicity determinant. DNA B encodes two genes each in complementary and virion sense orientation that are involved in virus trafficking (Hanley-Bowdoin et al., 2013). The two viral genomic components share homologous sequence of 200 nucleotide in length known as common region (CR) that contain promoters involved in initiation of replication.
In order to develop effective management strategies of cassava-infecting geminiviruses, better understanding hostvirus interaction is essential. Therefore, this study reports response of Nicotiana benthamiana to infection by two species of cassava-infecting geminiviruses.

Material and methods
Plant material and Agro-inoculation Ten Nicotiana benthamiana plants were inoculated with Agrobacterium tumefaciens (strain GV3103) transformed with AKK1420 binary vector carrying infectious clones of ACMV-CM (AF112352 and AF112353) and EACMV KE2 (K201) (AJ717541 and AJ704953), respectively, cloned as head to tail partial dimers (Patil and Fauquet, 2015). Agrobacterium cultures were established at 28°C for 12h with shaking in Luria-Bertani (LB) broth supplemented with appropriate antibiotics. At an optical density of 1.0, the culture was spun at 10,000 g, for 2 min. After discarding the supernatant, the resulting pellet was re-suspended in 25 ml of infiltration buffer (0.5 strength Murashige and Skoog (MS) media, supplemented with 5 mM MES, 5 mM MgSO4, pH 5.7) and the culture incubated without shaking for three hours at room temperature. The culture was then diluted to an OD of 0.5 using the infiltration solution supplemented with 200uM acetosyringone. The Agrobacterium culture mixtures were infiltrated on the abaxial side of three week old N. benthamiana leaves as described by Leuzinger et al. (2013).

Nucleic acid extraction
Young leaves of N. benthamiana plants showing typical CMD symptoms were collected at 20 days post inoculation (dpi). The leaf samples were derived from four independent biological replicates for ACMV-CM and five biological replicates for EACMV KE2 (K201) infected plants. Total nucleic acids were extracted as previously described (Patil et al., 2015) and apportion into two. One of the samples aliquots was subjected to TURBO™ DNase (Ambion, Carlsbad, CA, USA) treatment as per kit instructions. The other aliquot was treated 4 μl DNase free ribonuclease (Roche, Indianapolis, IN, USA) for 2h at 37°C to remove RNA. The RNA and DNA samples were quantified on a NanoDrop 2000c spectrophotometer (Thermo Scientific, DE, USA).

Northern blot analysis
Virus derived small RNAs were detected using Northern blotting. Forty micrograms total RNA were fractionated on 15% Criterion TBE-Urea polyacrylamide gel (Bio-Rad, Hercules, CA, USA) at 100V for 2 h. RNA was electrotransferred onto positively charged Hybond nylon membrane (Amersham, UK) using trans-Blot Turbo Transfer System (Bio-Rad, Hercules, CA, USA) at 25V for 30 min and immobilized by crosslinking to the membrane twice at 120,000 microjoules/cm2 using a Stratalinker UV crosslinker 1800 (Stratagene, La Jolla, CA, USA). RNAs probes for virus derived siRNAs detection were PCR amplified from AC2/AC3 region of each virus genome using primers listed on Table 1. In vitro transcription was performed using DIG RNA labelling kit SP6/T7 (Roche, Indianapolis, IN, USA) following manufacturer's instructions. The ensuing steps were performed as described by Kuria et al. (2017). Blots were exposed to Amersham high-performance chemiluminescence film (GE Healthcare, Pittsburgh, PA, USA) for 15 min and processed on an automated developer (Konica Minolta-SRX-101A). The autoradiographs were scanned on Epson Perfection V700 photo scanned (Epson, CA, USA) and the signal quantified on image J.

Southern blot analysis
Southern blot analysis was performed to detect viral DNA titer using 5 μg genomic DNA derived from systemically infected N benthamiana plants at 20 dpi. Samples were fractionated on 1 % (w/v) agarose gel at 30V for 12 h. The DNA was depurinated in a solution of 0.2 N HCl for 15 min followed by denaturation in 0.5 N NaOH and 1.5 M NaCl for 30 min. Prior to transfer the pH of the gel was neutralized with 0.5 M Tris-HCl and 1.5 M NaCl for 25 min. The DNA was transferred overnight (12 h) onto a positively charged nylon membrane (Amersham, NJ, USA) using 20X SSC. The DNA was immobilized on the membrane through exposure to UV at 120,000 microjoules/cm2 using a Stratalinker UV crosslinker 1800 (Stratagene, La Jolla, CA, USA). Probes for virus detection were PCR amplified using primers listed on Table 1 targeting replicase and movement protein gene. Probes were PCR labelled using digoxigenin (DIG) DNA labelling kit (Roche, Indianapolis, IN, USA) as recommended by the manufacturer. Membranes were hybridized overnight at 55°C. Subsequent steps are previously described (Patil et al., 2015).
Small RNA library preparation Small RNA libraries were prepared using NEBNext® Multiplex Small RNA Library Prep Set for Illumina. The libraries were submitted to the Genome Technology Access Center (GTAC), Washington University in St. Louis, Missouri, USA followed by quality control on Agilent 2100 Bioanalyzer (Agilent Technologies, Inc., Santa Clara, CA). The small RNA libraries were subjected to Illumina HiSeq 2,500 using 1 × 50 single-end read protocol. After sequencing raw reads were downloaded from GTAC website and de-multiplexed by QIIME (Caporaso et al., 2010) and reads with quality score below 19 were discarded. Adapter sequences were trimmed using Cutadapt (Martin, 2011). Small RNA sequences in the size range of 21-24 nt were selected and mapped to the virus reference genome with zero mismatches. Statistical analysis was performed using BEDTools (Quinlan and Hall, 2010) and all outputs were graphically presented by Shell scripts (Fahlgren et al., 2009).

Response of N benthamiana to inoculation with ACMV-CM and EACMV KE2 (K201)
The infection dynamics of two species of cassava-infecting geminiviruses were studied in N benthamiana through agro-inoculation. The onset of CMD symptoms on N benthamiana leaves was observed at 5 and 10 dpi for ACMV-CM and EACMV KE2 (K201), respectively (Fig. 1). Co-inoculation with both ACMV-CM and EACMV KE2 (K201) produced the most severe symptoms starting from 5 dpi resulting into leaf curling, yellow mosaic and plant stunting (Fig. 1). Based on the intensity of the Northern and Southern blots ACMV-CM seem to have accumulated to higher amounts at 10 dpi compared with EACMV KE2 (K201) correlating with more severe symptoms at this time point. However, after 20 dpi plants infected with ACMV-CM displayed symptoms remission characterized by appearance of non-asymptomatic leaves (Fig. 1). Contrastingly EACMV KE2 (K201) induced symptoms persisted throughout the experimental period (Fig. 1).  Accumulation of virus mRNAs, siRNAs and DNA was detected in plants showing systemic CMD symptoms at 10 dpi (Fig. 2). At 10 dpi, there was direct correlation between virus titer, virus derived siRNAs, mRNAs and symptom expression (Figs. 1, and 2). The DNA conformations of the two virus species differed whereby plants systemically infected with ACMV-CM accumulated more single stranded DNA (ssDNA) forms than those infected with EACMV KE2 (K201) (Fig. 2C) indicating active replication and encapsidation of ACMV-CM virus. However, there was no obvious differences in the accumulation of double stranded DNA (dsDNA) forms for the two virus species (Fig. 2C). The two viral species were seen to accumulate differently whereby ACMV-CM accumulated more viral mRNAs, siRNAs and DNA compared with EACMV KE2 (K201) producing severe symptoms (Fig. 2). Deep sequencing of small RNAs in geminiviruses infected Nicotiana benthamiana Sequence data showing the average total small RNAs reads obtained from N benthamiana leaves systemically infected with ACMV-CM and EACMV KE2 (K201) respectively is presented in Table 1. After adaptor removal the average total of 21 -24 nt reads with Phred Quality score of above 20 were between 969,015 and 1,307,689 for libraries constructed from EACMV KE2 (K201) and ACMV-CM infected leaf tissues. Of the total small RNAs reads >80% mapped to the host genome. In N benthamiana systemically infected with EACMV KE2 (K201), 12.31% and 4.51% reads were mapped to the viral DNA A and DNA B (Table 1). On the other hand N benthamiana systemically infected with ACMV-CM DNA accumulated 5.70% and 0.45% vsRNAs reads mapping to ACMV-CM DNA A and DNA B components (Table 1). For both virus species, majority of vsRNAs were mapping to DNA A component. However, EACMV KE2 (K201) derived small RNAs were proportionally higher than those derived from ACMV-CM (Table 1).

Characterization of small RNAs derived from ACMV-CM and EACMV KE2 (K201) respectively
The total sRNAs reads of 21 -24 nt in length from libraries prepared from EAMCV KE2 (K201) and ACMV-CM infected leaf samples were mapped to the respective viral genome with zero mismatches. Different classes of vsRNAs showed a differential accumulation for both virus species with 22 nt being the most abundant (Fig. 3).
The total mapped reads of 22 nt vsRNAs varied between the two viruses, DNA components and polarity. EACMV K201 derived sRNAs ranged between 9888 and 42501 reads compared with 2828 and 34045 reads for ACMV (Fig. 3). Overall for both viruses DNA A accumulated the most vsRNAs compared with DNA B but for both components there was no apparent differences in polarity for all populations of vsRNAs (Fig. 3).

Figure 3 Illumina deep sequencing of small RNAs in N benthamiana infected with EACMV KE2 (K201) and
ACMV-CM respectively. The graphs shows total vsRNAs of size-classes 21-24 nt vsRNAs mapped to the virus genome with zero mismatches.

DISCUSSION
The interaction between cassava-infecting geminiviruses and host plant N benthamiana was studied through agroinoculation. Differential response was observed between the two viral species. N benthamiana plants infected with ACMV-CM expressed symptoms after 5 dpi while those infected with EACMV KE2 (K201) developed symptoms starting at 10 dpi. Previous studies have demonstrated different symptom dynamics in N benthamiana plants with ACMV like viruses inducing symptoms at 4 dpi and EACMV like viruses symptoms appearing after 8 dpi (Patil and Fauquet, 2015). Studies by Shen et al. (2014) demonstrated that phosphorylation of AC2 proteins from Cabbage leaf curl virus (CaLCuV) and Tomato mottle virus (ToMoV) by SnRK, a signaling kinase involved in regulating sugar metabolism in plants resulting in delayed symptom development in Arabidopsis thaliana plants.
Data presented here showed recovery from ACMV-CM infection beginning 20 dpi whereas EACMV KE2 (K201) symptoms were persistent throughout the experimental period. A range of symptom phenotypes have been previously reported for various cassava-infecting geminiviruses in different host plants (Patil and Fauquet, 2015). Past studies have associated viral load and symptoms severity with virus pathogenicity (Sun et al., 2015). The results described here showed that EACMV KE2 (K201) induced more severe persistent symptoms compared with ACMV-CM. EACMV-like virus have been shown to encode suppressors of gene silencing that target both post transcriptional gene silencing (PTGS) and transcriptional gene silencing (TGS) pathways whereas ACMV-CM encoded suppressors only subvert PTGS (Vanitharani et al., 2004).
In N benthamiana plants infected with ACMV-CM and EACMV KE2 respectively, the most frequent class of vsRNAs were 22 nt, followed by 21 nt and 24. Similar Genome wide resolution maps revealed heterogeneous distribution of vsRNAs throughout the virus genome in sense and antisense polarities. This collaborate well with formation of long dsRNA covering the entire virus genome in sense and antisense direction as a result of POI II-mediated bidirectional transcription (Poogin et al., 2016;2013). Readthrough transcripts and/or their degradation products are processed by DCLs into distinct classes of vsRNAs covering the entire virus genome (Seguin et al., 2014). Several hotspot for were identified within the coding regions and were more pronounced within the overlapping regions of DNA A components for both virus species as previously described (Bai et al., 2016;Miozzi et al. 2013;Yang et al. 2011). However, for DNA B components hotspots were identified in non-overlapping BC1 and BV1 regions ( Fig. 5B and D). One prominent 24 nt vsRNAs peak was identified within the promoter region of EACMV KE2 (K201) sense strand. The 24 nt vsRNAs potentially direct transcriptional silencing of virus BC1 as previously demonstrated (Pooggin et al., 2016). The distinct size classes of vsRNAs were localized in the same hotspots indicating targeting of same regions along viral genome by different DCLs. Comparison of ACMV-CM and EACMV KE2 (K201) vsRNAs distribution along the viral genome revealed that certain regions in ACMV-CM genome such as AC1, AC4 and non-coding regions of DNA B were poorly saturated with vsRNAs. Interaction between sRNAs and AGO proteins is mostly determined by 5' terminal nucleotide (Fang and Qi 2016;Carbonell and Carrington, 2015). For both virus species, vsRNAs exhibited a strong bias to uridine 21 nt (52.5%), 22 nt (57.6%) and 24 nt (36.9%) for EACMV KE2 (K201) and 21 nt (47.9%), 22 nt (58.4%) and adenosine 24 nt (44.3%) for ACMV-CM. This result support the participation of AGO1, AGO2 and AGO4 for geminiviruses defense in N. benthamiana and is consistent with previous studies (Piedra-Aguilera et al., 2019;Wu et al., 2019;Margaria et al., 2016). Loading of vsiRNAs into diverse AGOs mediated by the 5′-terminal U and A may play a role in targeting expression of multiple host genes involved in different processes and pathways that enhance virus infection. Recently, Wu et al. (2019) identified 19 vsiRNAs from Tobacco curly shoot virus (TbCSV) that inhibit expression of complementary cellular transcripts involved in molecular functions and biological processes and enhances expression of RNA-dependent RNA polymerase 1 (RDR1) resulting in ideal conditions for viral infection. Furthermore, vsiRNAs derived potato spindle tuber viroid (PSTVd)-derived vsiRNAs degrade two callose synthase genes exacerbating disease severity and accumulation of viroids (Adkar-Purushothama et al., 2015). Similarly, peach latent mosaic viroid (PLMVd)-derived siRNAs target the chloroplastic heat shock protein 90 gene in peach to induce albinism and potentially create a favorable host environment for viroid infection (Navarro et al., 2012). To further understand geminiviruses and host interaction we suggest AGO vsRNAs immunoprecipitation studies.

Conclusion
This study demonstrate infection dynamics of important viruses that inhibit potential yields in cassava. The profiles of EACMV KE2 (K201) and ACMV-CM derived small RNAs were different in N benthamiana whereby more vsRNAs were derived from EACMV KE2 (K201) indicating differences in virus pathogenicity. The outcome of this study highlight possible role of RNA silencing pathway in geminiviruses pathogenicity.