There are 228 NBS-LRRs in cassava, including both TIR-NBS-LRR and CC-NBS-LRR. Their transcript levels were analyzed through RNA-seq in response to CBSV and C. gloeosporioides infection (Lozano et al., 2015; Utsumi et al., 2016). Among these, four MeLRRs (MeLRR1, MeLRR2, MeLRR3, and MeLRR4) were both induced under CBSV and C. gloeosporioides infection and selected for further analysis. The four MeLRR proteins have typical leucine-rich repeats, which are named MeLRR1 (Manes. 11G053000.1), MeLRR2 (Manes. 03G071700.1), MeLRR3 (Manes. 13G036800.1), and MeLRR4 (Manes. 07G107800.1), located on chromosomes 11, 3, 13, and 7, respectively. MeLRR1, MeLRR3, and MeLRR4 belong to CC-NBS-LRR protein, while MeLRR2 is one of the TIR-NBS-LRR protein. Bioinformatics predicted that the MeLRR proteins were unstable and hydrophilic (Supplementary Table 1). The phylogenetic analysis showed that MeLRR1 clustered with XP_012073222.1 of Jatropha curcas, MeLRR2 clustered with XP_021684995.1 of Hevea brasiliensis, MeLRR3 clustered with XP_020535356.1 of J. curcas, and MeLRR4 clustered with KAF2295929.1 of H. brasiliensis based on whole protein sequences (Supplementary Figure 1).

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).

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 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).

To analyze the function of MeLRRs, we constructed MeLRR-silenced cassava plants by virus-induced gene silencing (VIGS). The partial sequences of MeLRR1 (453 bp), MeLRR2 (441 bp), MeLRR3 (433 bp) and MeLRR4 (423 bp) were individually inserted into pTRV2 plasmid to construct VIGS vector. At 14 days post-infection (dpi) in cassava infected with Agrobacterium GV3101 carrying the pTRV-MeLRR plasmids, qRT-PCR was performed to detect the target gene transcript level. The transcript level of the target MeLRR-1,-2,-3,-4 genes were significantly decreased in the MeLRR-silenced cassava leaves compared to the pTRV empty vector. The silencing efficiency of MeLRR-1,-2,-3,-4 was 46.33 (± 2.31)%, 15.28 (± 0.49)%, 30.22 (± 2.28)%, and 17.45 (± 0.87)% (Mean ± SD, n = 3), respectively (Figure 3A). It was noteworthy that the silenced of MeLRR1 did not affected the transcription of MeLRR2,-3,-4. Similar results were verified in MeLRR2-, MeLRR3-, and MeLRR4-silenced plants (Supplementary Figure 2). When co-silenced four target genes (MeLRR-1,-2,-3,-4) in one VIGS line, the transcript levels of all four target genes were significantly decreased (Supplementary Figure 3). On the contrary, the bacteria number was significantly higher than that in the pTRV empty vector-infected cassava leaves (Figure 3B). Moreover, the cassava MePR1 transcript level was significantly decreased (Figure 3C). And the transcript level of MePR1 in pTRV-MeLRR1,−2,−3,−4 cassava was reduced on average to 84, 7, 58, and 69%, respectively, of the transcript level in the pTRV control at 14 dpi (Figure 3C). Additionally, the transcript level of MePR1 in MeLRR-1,-2,-3,-4-silenced plant was significantly reduced to 66% (Supplementary Figure 3). Silencing of MeLRRs conferred increased disease susceptibility in cassava leaves (Figure 3D). Moreover, MeLRRs-silenced cassava leaves showed significantly lower ROS burst measurements than the empty vector (Figures 3E,F). These results indicate that silencing of MeLRRs impairs cassava resistance to Xam.

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).

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.

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 MeLRR1 and MeLRR2 in Arabidopsis plants than in control plants without A. brassicicola, B. cinerea, and P. syringae pv. tomato DC3000 infection. Similar results were observed in plant pathogen-infected overexpression of MeLRR3 in Arabidopsis plants compared with control plants. However, AtPDF1.2 and AtPR1 were significantly down-regulated in overexpression of MeLRR3 in Arabidopsis plants than in control plants without P. syringae pv. tomato DC3000 infection. On the other hand, the expression levels of AtPR2 and AtTGA3 were significantly up-regulated in overexpression of MeLRR4 in Arabidopsis plants than in control plants. AtICS1, AtPDF1.2, AtPR1, and AtPR5 genes were up-regulated or down-regulated under different plant pathogen infections. These results indicate that overexpression of MeLRRs resulted in enhanced resistance simultaneously against pathogenic bacteria and pathogenic fungi, demonstrating the requirement of MeLRRs for resistance to plant pathogens.

Cassava (M. esculenta), variety South China 124 (SC124), and N. benthamiana were cultivated in mixed soil (vermiculite/nutritional soil = 2:1, v.v.) in a greenhouse with 16/8 h light/dark at 28/22°C, 60–70% relative humidity with irradiance of 130–150 uE.m^–2.s^–1. A. thaliana ecotype Col-0 (Columbia-0) seedlings were cultivated in the mixed soil under fluorescent light (130–150 uE.m^–2.s^–1) and were grown under 16/8 h light/dark at 22°C. For axenic growth, N. benthamiana, and A. thaliana seeds were sterilized (10% NaClO for 1 min, washed five times with sterile water) and sown on half-strength MS (Murashige and Skoog) medium (PhytoTechnology Laboratories, Kansas, United States) with 0.4% agar powder and 2% (w/v) sucrose. The seeds were grown in chambers under 16/8 h light/dark at 22°C and 130–150 uE.m^–2.s^–1. For expression analysis, 4-week-old cutting seedlings of cassava leaves were sprayed with 5 mmol/L salicylic acid or Xam suspension for 0, 1, 3, 6, 12, or 24 h, and the bacterial solution was diluted to 4 × 10⁸ colony-forming units/mL (cfu/mL) using 10 mmol/L MgCl₂ with 0.05% Silwet L-77.

The sequences of MeLRR genes were searched and obtained from the cassava genome database, M. esculenta v6.1 (Phytozome v13¹) (Muñoz-Bodnar et al., 2014; Lozano et al., 2015; Bredeson et al., 2016). The ProtParam tool² 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 nucleotide-binding 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). The 24 NBS-LRR proteins were derived from A. thaliana (CAA0374684.1, CAD5320387.1, CAE6029947.1, NP_181039.1, OAP10808.1, VYS54481.1), Durio zibethinus (XP_022746274.1), H. brasiliensis (XP_021646775.1, XP_021652057.1, XP_021646749.1, XP_021684995.1, KAF229 5929.1), J. curcas (XP_012073222.1, KDP37136. 1. XP_02053 5356.1), Populus alba (XP_034892116.1, XP_034896332.1), P. euphratica (XP_011001622.1), P. trichocarpa (RQO87881.1), Ricinus communis (EEF44774.1), Theobroma cacao (XP_01796 9995.1), Vernicia montana (AMM43068.1), V. vinifera (XP_010657.1), and Ziziphus jujuba (XP_024924720.1), respectively.

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™ 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^{–Δ Δ 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 Primer3Plus³ 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.

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’,6-diamidino-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 × 10⁸ 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 2.0⁴ 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 × 10⁸ 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 thaliana ecotype Col-0 was used as wild-type. 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).

The endogenous SA content in leaves was determined as previously described (Wei et al., 2018). Briefly, leaves were flash-frozen in liquid nitrogen and ground to a very fine powder. SA was extracted from 0.1 g powder using phosphate-buffered 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.

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).

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).

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 OD₆₀₀ reached 0.6. Thereafter, a fresh bacterial culture of Pst DC3000 was diluted to 4 × 10⁸ cfu/mL in 10 mmol/L MgCl₂ 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 × 10⁶ 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).

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.