The Cassava NBS-LRR Genes Confer Resistance to Cassava Bacterial Blight

Cassava bacterial blight (CBB) caused by Xanthomonas axonopodis pv. manihotis (Xam) seriously affects cassava yield. Genes encoding nucleotide-binding site (NBS) and leucine-rich repeat (LRR) domains are among the most important disease resistance genes in plants that are specifically involved in the response to diverse pathogens. However, the in vivo roles of NBS-LRR remain unclear in cassava (Manihot esculenta). In this study, we isolated four MeLRR genes and assessed their expression under salicylic acid (SA) treatment and Xam inoculation. Four MeLRR genes positively regulate cassava disease general resistance against Xam via virus-induced gene silencing (VIGS) and transient overexpression. During cassava-Xam interaction, MeLRRs positively regulated endogenous SA and reactive oxygen species (ROS) accumulation and pathogenesis-related gene 1 (PR1) transcripts. Additionally, we revealed that MeLRRs positively regulated disease resistance in Arabidopsis. These pathogenic microorganisms include Pseudomonas syringae pv. tomato, Alternaria brassicicola, and Botrytis cinerea. Our findings shed light on the molecular mechanism underlying the regulation of cassava resistance against Xam inoculation.

Cassava is a widely grown drought-tolerant crop that can be cultivated as an annual crop in marginal soils in tropical and subtropical regions of the world (Lozano et al., 2015;Bredeson et al., 2016). However, as a clonally propagated crop, cassava is especially vulnerable to pathogens, especially cassava bacterial blight (X. axonopodis pv. manihotis = X. phaseoli pv. manihotis) (Bredeson et al., 2016;Constantin et al., 2016;Zárate-Chaves et al., 2021), cassava brown streak disease (Cassava brown streak virus, CBSV) and anthracnose disease (Colletotrichum gloeosporioides) (Utsumi et al., 2016). Therefore, it is best to identify the NBS-LRR proteins in cassava. Results presented by Utsumi et al. (2016) indicated that the transcript level of NBS-LRRs was inducted by C. gloeosporioides infection. Similar results were obtained under plants infected by viruses (Louis and Rey, 2015;Lozano et al., 2015;Amuge et al., 2017;Masumba et al., 2017). A cluster of NBS-LRR genes on chromosome 11 of cassava genome was associated with resistance to cassava brown streak disease via genome-wide associated mapping and genomic selection (Kayondo et al., 2018). However, the mechanisms remain unclear, particularly in experimental investigation and verification.
In this study, we analyzed the published transcriptome databases of cassava-pathogens interaction (Lozano et al., 2015;Utsumi et al., 2016). Within the database, four NBS-LRR genes that showed high transcription level after pathogen infection attracted our attention. The expression levels of four chosen MeLRRs were significantly induced by exogenous application of SA treatment and Xam inoculation. Moreover, these genes positively regulated cassava resistance to Xam inoculation. The functional analysis of MeLRR genes will offer potential roles in genetic breeding for disease-resistant cassava.

Subcellular Localization of the MeLRR Proteins
To investigate the subcellular localization of the MeLRR proteins, the coding sequences (CDSs) of MeLRRs were cloned and inserted into the poly-cloning sites of the fusion expression vector pEGAD and fused upstream to a green florescence protein (GFP) fusion partner by the constitutive CaMV35S promoter. The Agrobacterium tumefaciens strain GV3101 cell culture harboring the pEGAD empty vector containing 35S:GFP was used as a control, and tobacco (Nicotiana benthamiana) leaves were infected with 35S:GFP or 35S:GFP-MeLRR1,−2,−3,−4 plasmid as described by Sparkes et al. (2006). The fluorescence of transiently expressing MeLRR proteins in tobacco leaf epidermal cells was detected in the nucleus, cytoplasm and cytomembrane, similar to that of 35S:GFP (Figure 1).

Expression Level of MeLRR Genes in Response to SA Treatment and Xam Inoculation
The expression profile of MeLRRs in response to SA treatment and Xam inoculation were analyzed by qRT-PCR (real-time quantitative reverse transcription PCR). Under SA treatment, the expressions of MeLRR1, MeLRR3, and MeLRR4 were induced and FIGURE 1 | Subcellular localization of MeLRR proteins in N. benthamiana leaves. Transient expression of Agrobacterium GV3101 with 35S:GFP and 35S:GFP-MeLRRs plasmids in N. benthamiana leaves. After 2 dpi, the fluorescence was scanned by a Leica confocal microscopy system (Leica TCS SP8, Solms, Germany) with an excitation wavelength of 488 nm and a 505-530 nm bandpass emission filter. The empty vector 35S:GFP was used as a control. Nuclei were stained using DAPI (4',6-diamidino-2-phenylindole). Scale bar = 25 µm.
peaked at 1 h post treatment (hpt), while MeLRR2 showed the highest level at 3 hpt (Figure 2). Following infection by Xam, the expression level of MeLRRs at 1-24 hpt hpi was significantly higher than that at 0 hpi (Figure 2). Moreover, the expression of MeLRR1, MeLRR2, and MeLRR3 were induced and peaked at 3 hpt, while the expression of MeLRR4 reached the peak at 12 hpt (Figure 2).

Transient Overexpression of MeLRR Genes
To further verify the function of MeLRRs, 35S:GFP-MeLRR recombinant plasmids were constructed and introduced into Agrobacterium strain GV3101. Cassava leaves were infected with Agrobacterium containing the recombinant plasmids or empty vector for 3 days. The transcript level of the target MeLRR-1,-2,-3,-4 genes were significantly higher than that in the 35S:GFP empty vector ( Figure 4A). It is similar in silenced plant, overexpressing MeLRR1 plant did not affected the transcription of MeLRR2,-3,-4, and the same as in MeLRR2-, MeLRR3-, and MeLRR4-overexpressed plants (Supplementary Figure 4). However, the transcript levels of the four target genes were significantly enhanced in co-overexpression MeLRR-1,-2,-3,-4 plants (Supplementary Figure 5). On the contrary, the bacteria number was significantly lower than that in the control ( Figure 4B). However, the transcript levels of MePR1 in 35S:GFP-MeLRR1,−2,−3,−4 cassava were increased by 3. 77-, 23. 73-, 10. 70-, and 1.39-fold, respectively, compared to those in the control at 3 dpi ( Figure 4C). Similarly, the transcript level of MePR1 in co-overexpression MeLRR-1,−2,−3,−4 lines was significantly increased by 24.03-fold (Supplementary Figure 5). Interestingly, overexpression of MeLRRs conferred improved disease resistance in cassava leaves ( Figure 4D). Moreover, cassava leaves that overexpressed MeLRRs exhibited significantly higher ROS burst than 35S:GFP control during flg22 treatment (Figures 4E,F). These results suggest that MeLRRs positively regulated cassava resistance to Xam. In addition, trypan blue staining showed no cell death phenotype at 2 dpi at transient expression of MeLRRs in cassava and N. benthamiana leaves (Supplementary Figure 6).

MeLRR-Mediated Cassava Immune Responses via SA Accumulation
To further analyze the mechanism of MeLRRs in response to Xam inoculation, the SA content was measured. As shown in Figure 5, the SA level in MeLRR1,−2,−3,−4-silencing was significantly decreased compared with that in pTRV control cassava leaves ( Figure 5A). By contrast, the SA level in MeLRRs overexpression was significantly increased compared with the control cassava leaves ( Figure 5B). These results suggested that MeLRR1,−2,−3,−4 positively participated in cassava immune responses via SA accumulation.

Overexpression of MeLRR Genes in Arabidopsis Enhances Resistance to Plant Pathogens
To further confirm the MeLRR function, MeLRRs were overexpressed in Arabidopsis. Quantification of endogenous SA levels indicated that MeLRR-overexpressing lines accumulated significantly higher levels than WT leaves (Supplementary  Figure 7). The MeLRRs overexpression plants displayed slight symptoms of wilting in response to P. syringae pv. tomato, A. brassicicola, and B. cinerea infection support the hypothesis that MeLRRs functions in a pathogen response pathway. A difference was already observed in the WT, suggesting that restricted bacterial entry into the leaves may underlie part of the apparent resistance ( Figure 6A). Unlike P. syringae pv. tomato, A. brassicicola, and B. cinerea can enter hosts by penetrating the cuticle. Consistently, there was less fungal growth in leaves overexpressing these factors than WT plants by analyzing the transcript levels of the A. brassicicola AbAct (JQ671669.1) gene and B. cinerea BcActA (XM_024697950.1) gene (Liao et al., 2016)    with the Arabidopsis AtAct2 gene as an internal control at 2 and 4 dpi, respectively (Figures 6B,C).
To determine whether the enhanced resistance to plant pathogens was related to changing the defense response genes expression level, we used qRT-PCR to analyze the expression levels of AtICS1, AtPDF1.2, AtPR1, AtPR2, AtPR5, and AtTGA3 in WT and MeLRR overexpression lines upon A. brassicicola, B. cinerea, and P. syringae pv. tomato DC3000 infection (Supplementary Figure 8). Particularly, the relative expression levels of genes involved in the SA synthesis pathway and pathogen resistance showed higher level in overexpression

DISCUSSION
NBS-LRR proteins play important roles in pathogen recognition and defense response signal transduction (Urbach and Ausubel, 2017). An increasing number of NBS-LRR proteins that conferred resistance to pathogens have been cloned from higher plants , such as TaRCR1 (Zhu et al., 2017), ZmNBS25 (Xu et al., 2018), GbaNA1 (Li et al., 2018a,b), GhDSC1 , and OsRLR1 (Du et al., 2021). In this study, we found that MeLRR1,-2,-3,-4 expression could be induced by Xam inoculation. Similar expression patterns have been observed in other plant NBS-LRR genes, such as AhRRS5 (Zhang et al., 2017) and SacMi (Zhou et al., 2018). NBS-LRRs mainly participate in plant resistance against pathogen infection, and we speculated that the up-regulation of MeLRRs could help cassava successfully evade Xam inoculation.
SA is a secondary messenger for systemic acquired resistance (SAR), and its production in plants represents the successful recognition of pathogen infection and pathogen-associated molecular pattern (PAMP)-triggered immunity (PTI) and effector-triggered immunity (ETI) (Divi et al., 2010;Peng et al., 2021). In cassava, SA also plays an important role in the regulation of cassava resistance to CBB Chang et al., 2020;Wei et al., 2021a,b) and to whitefly (Irigoyen et al., 2020). Wei et al. (2018) found that MeHsf3 regulates cassava resistance to cassava bacterial blight through modulation of SA accumulation. Cassava co-chaperones MeHSP90.9 interacts with MeSRS1 and MeWRKY20 to activate SA biosynthesis, accumulation of SA, and thus improve resistance to CBB (Wei et al., 2021b). Therefore, endogenous SA accumulation levels are an indicator of resistance to CBB. We found that the expression levels of MeLRR were significantly increased by SA treatment, which showed the similar expression pattern of NPR1 in Arabidopsis, ZmNBS25 in maize, and GhDSC1 in cotton. In response to pathogen infection, plant endogenous SA is quickly and strongly induced.
AtPDF1.2, AtPR1, AtPR2, and AtPR5 are widely known as marker genes for innate immune response (Wang et al., 2017;Xu et al., 2018). AtICS1 is a key enzyme for SA biosynthesis (Macaulay et al., 2017). AtTGA3 showed strong affinity for the NPR1 protein (Zhou et al., 2000;Yuan et al., 2009). In pathogenic microorganism infection, the SAR defense response is triggered by elevated SA through an SA-NPR1-TGA-PR1 signaling pathway (Zhang, 2003). Further analysis of gene expression in overexpression of MeLRR1,−2,−3,−4 at Arabidopsis leaves suggested that these genes might exert their function through SA biosynthesis and immune responses. This is similar to the function of MeHsf3 MebZIP3, -5 (Li et al., 2017), which were confirmed to regulate cassava resistance against cassava bacterial blight. Hence, we conclude that MeLRR1,−2,−3,−4 may regulate the plant immune response through SA and ROS accumulation, and the transcription of disease resistance genes. Taken together, the MeLRR genes encode a class of NBS-LRR proteins, which controls immunity to Xanthomonas axonopodis pv. manihotis in cassava. Further investigation of the role of the MeLRRs will build an important foundation for future development of resistant cultivars, which may be the most effective means of controlling this devastating disease.

Comprehensive Characterization and Bioinformatics Analysis of MeLRR Genes
The sequences of MeLRR genes were searched and obtained from the cassava genome database, M. esculenta v6.1 (Phytozome v13 1 ) (Muñoz-Bodnar et al., 2014;Lozano et al., 2015;Bredeson et al., 2016). The ProtParam tool 2 was used to predict the number of amino acids, relative molecular mass of protein, isoelectric point, total average hydrophilicity stability index, fat coefficient, and instability index (Gasteiger et al., 2003). Alignments between MeLRRs and other NBS-LRR proteins were performed used DNAMAN 6.0, and the phylogenetic tree was constructed by the neighbor-joining method based on the whole protein sequences and considering 1,000 bootstrap replicates using ClustalW tool and MEGA 7 (Kumar et al., 2016). The 24 NBS-LRR protein amino acid sequences in 13 species were screened based on the principles of encoding nucleotidebinding site (NBS) and leucine-rich repeat (LRR) domains, and were validated through comparisons of the protein basic local alignment search tool (BLASTP) with the National Center for Biotechnology Information (NCBI

RNA Extraction, cDNA Synthesis, and Quantitative Real-Time PCR
Total RNA was extracted from three independent pools, and DNA contamination was removed using the Tiangen RNA prep pure plant plus kit (Tiangen Biotech, Beijing, China, Cat# DP441). cDNA synthesis was performed using the Tiangen FastQuant RT kit (Tiangen Biotech,Beijing,China,Cat# KR116) with 20-µl reaction mixture. qRT-PCR analysis was performed using UltraSYBR Mixture (low ROX) (CoWin Biosciences, Beijing, China, Cat# CW0956) in an ABI QuantStudio TM 6 flex Real-Time PCR System (ABI, CA, United States). The PCR cycling conditions were 95 • C for 10 min, followed by 40 cycles at 95 • C for 15 s and 60 • C for 1 min. The Arabidopsis and cassava gene transcripts were normalized to the AtAct2 gene (AT3G18780) and elongation factor 1α (EF1α, Me.15G054800) using the comparative 2 − Ct method, respectively (Livak and Schmittgen, 2001). The qRT-PCR primers of MeEF1a, MePR1 were obtained from Wei et al. (2018), AtPR1, AtPR2, AtPR5, AtPDF1.2, AtICS1, AtAct2, and BcActA were obtained from Mhamdi and Noctor (2016), and AtTGA3 was obtained from Ndamukong et al. (2017), respectively. The qRT-PCR primers of MeLRRs, and AbAct (JQ671669.1) of A. brassicicola were designed by Primer3Plus 3 to find optimal primers (Untergasser et al., 2007), and then the specificity of the melt curve analyzed performed to determine. In addtion, the qRT-PCR fragments and VIGS fragments are different CDS regions of MeLRRs. The primers used are listed in Supplementary Table 2.

Plasmid Construction and Transient Expression in Plant Leaves
For overexpression, the full-length coding regions of MeLRR1,-2,-3,-4 were amplified and cloned into the pEGAD vector (Promoter CaMV35S:GFP) via appropriate restriction enzyme digestion and T4 DNA ligase. The recombinant plasmids and empty vector were transformed into Agrobacterium GV3101. Then, the A. tumefaciens suspension was used to infect the leaves of cassava or tobacco as described by Sparkes et al. (2006) and Zeng et al. (2019). Tobacco leaves injected with Agrobacterium GV3101 for 2 days, the GFP fluorescence and DAPI (4' ,6diamidino-2-phenylindole, Thermo Fisher Scientific, Shanghai, China)-stained cell nuclei were imaged under a fluorescence microscope (Leica TCS SP8, Solms, Germany), with an excitation wavelength of 488 nm and a 505-530-nm bandpass emission filter. Cassava leaves inject with recombinant pEGAD plasmids or empty vector of Agrobacterium GV3101. Then, 3 days later, the cassava leaves were syringe infiltrated with 4 × 10 8 cfu/mL of pathogenic bacteria Xam used for disease resistance assay, include number of Xam populations, MePR1 transcript level, and symptoms of cassava bacterial blight at 0 and 1 dpi, respectively.
VIGS constructs are usually prepared using 300-500 bp partial CDS regions of MeLRRs and the online siDirect 2.0 4 tools (Naito et al., 2009) are available for predicting regions with high siRNA generating capability (Naito et al., 2009;Ui-Tei and Naito, 2013). Zeng et al. (2019) constructs the method about Agrobacterium-mediated Tobacco Rattle Virus (TRV)based gene silencing in cassava. For VIGS in cassava, the specific CDS fragments of MeLRR1,−2,−3,−4 were amplified and cloned into the pTRV2 vector through appropriate restriction enzyme digestion and T4 DNA ligase. The recombinant plasmids and empty vectors were transformed into Agrobacterium GV3101. Then, the Agrobacterium suspension, as well as pTRV1, was used to infect the leaves of cassava as previously described (Zeng et al., 2019). At 14 dpi, the new leaves were syringe infiltrated with 4 × 10 8 cfu/mL of pathogenic bacteria Xam used for disease resistance assay. The sequences of primers used for vector construction in this study are listed in Supplementary Table 2.

Arabidopsis Transformation
Arabidopsis thaliana ecotype Col-0 was used as wildtype. Overexpressing lines were transformed by floral dip transformation method of 35S:GFP-MeLRR recombinant plasmids constructs with Agrobacterium GV3101 (Bechtold and Pelletier, 1998). The overexpressing lines were selected by 100 mg/L kanamycin and 20 mg/L glufosinate (Basta; Sangon Biotech. Shanghai, China) resistance and further confirmed by PCR. Single insertion transgenic lines were chosen for further analysis in transgenic third generations (T3).

Quantification of Endogenous SA Contents
The endogenous SA content in leaves was determined as previously described . Briefly, leaves were flash-frozen in liquid nitrogen and ground to a very fine powder. SA was extracted from 0.1 g powder using phosphatebuffered solution (PBS, pH 7.4, 0.15 M) on ice. Then, the supernatant was used for SA quantification using a plant SA ELISA (enzyme-linked immunosorbent assay) kit (Jiangsu Meimian Industrial, Jiangsu, China, Cat#HLE01901) according to the manufacturer's instructions.

Reactive Oxygen Species Burst Measurements
The ROS burst in leaves was determined as described previously (de Torres Zabala et al., 2015;Chang et al., 2020;Yan et al., 2021). In tomato, flg22 was used to instead of Xanthomonas to measure the ROS burst (Bhattarai et al., 2016). Similar methods were applied to study the cassava resistance to Xam, such as MeCAMTA3 (Chang et al., 2020), MeRAV5 (Yan et al., 2021). Herein, to measure the ROS burst, 48 leaf discs (5 mm in diameter) of cassava were placed in 48 single wells of 96-well black plates and placed in the dark for 12 h in 100 µL double-distilled water. After 12 h, the 48 leaf discs were divided into two groups. In one group, the water was replaced with 100 µL incubation solution containing 0.2 µmol/L luminol (AppliChem, Darmstadt, Germany) and 10 µg/mL horseradish peroxidase (AppliChem, Darmstadt, Germany). In the other group, the water was then replaced with 100 µL incubation solution containing 0.2 µmol/L luminol, 10 µg/mL horseradish peroxidase and 1 µmol/L flg22 (Phyto Technology Laboratories, Lenexa, KS, United States). Luminescence was measured immediately for 30 min using a GloMax 96 Microplate Luminometer (Promega, Madison, WI, United States). Luminescence readout is given in relative light emitting units (RLU).

Trypan Blue Staining
The cassava or N. benthamiana leaves were boiled for 1 min in the trypan blue working solution (100 mL lactic acid, 100 mL glycerol, 100 g phenol, and 0.2 g trypan blue, dissolved in 100 mL distilled water) for 24 h at room temperature (Luo et al., 2017). The leaves were transferred into a chloral hydrate solution (2.5 g/mL) and repeatedly reduced until the background was gone (Luo et al., 2017).

Pathogen Culture and Disease Assays
The pathogenic bacterium P. syringae pv. tomato (Pst) DC3000 was streaked on LB medium with 50 mg/L of rifampicin at 28 • C and shaken to OD 600 reached 0.6. Thereafter, a fresh bacterial culture of Pst DC3000 was diluted to 4 × 10 8 cfu/mL in 10 mmol/L MgCl 2 and 0.05% Silwet L-77 and then sprayed on 24-day-old Arabidopsis leaves. The A. brassicicola and B. cinerea strains were cultured on potato dextrose agar (PDA) medium with 2% (w/v) sucrose at 28 • C. Conidia were suspended in distilled water for plant infection. Spore suspensions (about 4 × 10 6 spores/mL) of A. brassicicola and B. cinerea were sprayed on Arabidopsis leaves. The infected plants were grown in an incubator at 90% RH and 22 • C. At 0, 2, and 4 dpi, the number of Pst DC3000 bacteria was determined, as well as the fungal actin gene transcript in leaves of Col-0 and mutants infected with B. cinerea and A. brassicicola (Veronese et al., 2006;Mhamdi and Noctor, 2016).

Analysis of Experimental Data
Mean and standard deviations are displayed as representative values for data in the figures. Analysis of variance (ANOVA) with Duncan's test and Student's t-test were applied to the obtained data with the help of IBM SPSS v20. Statistical significance ( * ) was set at p < 0.05. Each assay contained three independent replicates.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author.