Introduction

Cassava (Manihot esculenta) is an important crop grown for its starchy roots and is a staple food for millions of people in tropical regions. Cassava can not only be used as food but also can serve as a raw material for bioethanol production. However, cassava production can be affected by several pathogens, including Xanthomonas axonopodis pv. manihotis (Xam), the pathogen causing cassava bacterial blight (CBB). Xam is a gram-negative bacteria that infects plants via stomata and wounds1 and CBB is one of the most devastating diseases affecting cassava production, leading to significant yield losses2. This disease is characterized by leaf wilting, chlorosis, and necrosis, resulting in plant death. Inside plant tissues, colonization of Xanthomonas spp. occurs in the intercellular space or xylem vessels before spreading systemically within the plant3. The dispersal of Xam is stimulated by wind-driven rain splashes4causing widespread distribution in cassava fields. Moreover, Xam has an epiphytic stage that causes persistence of the pathogen between cropping cycles5. Host-pathogen interactions exist between Xanthomonas spp. and host plants, as bacterial effectors can induce the expression of some host genes6. These effectors can specifically bind to and induce a gene in a host in a so-called gene-for-gene interaction. The induction of bacterial effectors can affect the degree of resistance in various cultivars of the host plant species7. At present, the differential expression profiles of microRNA (miRNAs), small RNA (sRNA) and genes in cassava in response to Xam exposure have been reported8,9,10,11 but protein expression profiles remain largely unexplored, particularly with respect to the evaluation of proteome wide changes. Proteomic studies have become essential for investigating complex biological processes involved in plant-pathogen interactions. In the context of CBB, proteomic analysis can help elucidate the changes in protein expression that occur in response to Xam infection, providing insights into the molecular mechanisms underlying the plant defense response. Additionally, intracellular oxidative stress occurs after plants are infected with pathogens, producing reactive oxygen species (ROS)12. Polyunsaturated fatty acids (PUFAs) can be oxidized by ROS to generate malondialdehyde (MDA), an indicator of oxidative stress13. This study compared the proteome dynamics of two cassava cultivars, the tolerant industrial cultivar Rayong 72 (R72) and the susceptible edible cultivar Hanatee (HN), in response to Xam infection. Specifically, this study investigated changes in protein expression related to MDA levels in infected plants due to the involvement of ROS in pathogen defense response. Through this comparison, the study aimed to identify key proteins and pathways involved in cassava’s resistance to CBB, with the goal of developing more effective disease management strategies and enhancing cassava production.

Materials and methods

Plant materials

Cassava cultivars approved by the Department of Agriculture, Thailand including R72, a CBB-tolerant cultivar, and HN, a CBB-susceptible cultivar were used in this study. The R72 and HN stems were provided by the Nakhon Ratchasima Agricultural Research and Development Center, Department of Agriculture, Thailand. The stems were cut into lengths of 10–15 cm and planted in a 6 × 4-inch soil bag in the greenhouse with temperatures ranging from 36 to 24 °C.

Inoculum preparation

Xam isolate C (XamC) was cultured on YPGA solid medium containing 0.5% yeast extract, 0.5% peptone, 0.5% glucose, and 1.5% agar for single-colony selection at 28 °C for two days. The selected single colony was grown further on YPGA medium again for two days at 28 °C. These colonies on the plate were resuspended in YPG liquid medium and calibrated to OD600 nm = 0.2 or 108 colony forming units (CFU) per mL using a spectrophotometer. Stem puncture inoculation was performed on eight-week-old cassava plants. First, an incision point was made on the stem between the third (L3) and fourth (L4) fully expanded leaves using a sterile needle. Five microliters of calibrated inoculum of XamC were placed at the puncture point. Leaves L3, L4, and stems with incision points were collected at 0, 3, 7, 14, and 28 days after inoculation (DAI), and five plants per DAI were collected. Sample leaves and stems were immediately placed in liquid nitrogen and stored at −80 °C until use.

Symptom evaluation

The cassava response to XamC infection was determined by evaluating the area under the disease progression curve (AUDPC). Tolerant and susceptible cassava cultivars were compared at 0, 3, 7, 14, and 28 DAI on a severity scale of 0–5 as established by Lozano and Laberry, 198214. Infected plants were scored as follows: 0 as healthy plants, 1 as a dark area or necrosis around the inoculation point, 2 as gum exudates on the stem, 3 as wilting of one or two leaves and exudates, 4 as > 2 wilted leaves, and 5 as complete wilting and dieback. Five biological replicates were analyzed for each cultivar.

Shotgun proteomic analysis

The fourth leaves (L4) from infected R72 and HN cassava cultivars were collected, immediately snap-frozen in liquid nitrogen, and stored at − 80 °C until used. Leaf samples were subjected to protein extraction by the trichloroacetic acid (TCA)/acetone precipitation method15. Briefly, about 100 mg of L4 was ground to a fine powder in liquid nitrogen and homogenized with extraction buffer containing 1% sodium dodecyl sulfate, 0.1 M tris hydrochloride (pH 6.8), 2 mM EDTA, 20 mM dithiothreitol (DTT), and 1x protease inhibitor cocktail (Merck, Rahway, NJ). The mixture was centrifuged at 12,000 g for 5 min. The supernatant was transferred to a fresh tube, and the proteins were precipitated with a 1:1 ratio of 20% TCA in acetone and samples were incubated on ice for 30 min. The protein pellets were washed with acetone to obtain a white color and finally dissolved in a buffer containing 7 M urea, 2 M thiourea, and 20 mM DTT. The protein concentration was measured by the Bradford assay using a Bio-Rad protein assay kit (Bio-Rad Laboratories, Hercules, CA) which included bovine serum albumin (BSA) as a protein standard. Protein samples (5 µg total) for liquid chromatography with tandem mass spectrometry (LC-MS/MS) were prepared by incubation with 10 mM DTT in 10 mM NH4HCO3 at 56 °C for 1 h to reduce disulfide bonds. The protein samples were added to 10 mM iodoacetamide in 10 mM NH4HCO3 for alkylation in the dark at 28 °C for 45 min. Proteins were dehydrated using 100% acetonitrile before protease digestion. Protein samples were digested by sequencing grade trypsin at 37 °C overnight. Trypsin-digested peptides were extracted from the samples by incubating with 50% acetonitrile in 0.1% formic acid at 28 °C for 10 min, then dried at 45 °C for 4 h, and kept at −80 °C until the LC-MS/MS analysis. LC-MS/MS was performed using an Ultimate 3000/Capillary LC system (Thermo Fisher Scientific, Waltham, MA) coupled to a quadrupole Q-TOF Impact II (Bruker Daltonics Ltd., Hamburg, Germany) equipped with a Nano-captive spray ion source. The peptide enrichments were conducted on a p-Precolumn 300 pm I.D. × 5 mm C18 Pepmap 100, 5 pm, 100 A and separated on a 75 μm I.D. × 15 cm. Samples were then packed with Acclaim PepMap RSLC C18, 2 μm, 100 A, nanoViper (Thermo Fisher Scientific, Waltham, MA). Subsequently, solvents A (0.1% formic acid in water) and B (0.1% formic acid in 80% acetonitrile) were supplied to the analytical column. A gradient was maintained using a 5–55% solvent B linear concentration to elute the peptides at a constant flow rate of 0.30 µL/min for 30 min. Electrospray ionization was performed at 1.6 kV using a CaptiveSpray. Mass spectra (MS) and MS/MS spectra were obtained in positive-ion mode over the range of 150–2200 m/z using the Compass 1.9 software (Bruker, Billerica, MA) with BSA as a standard.

Protein identification and data analysis

MS/MS spectra were identified using MaxQuant software version 1.6.6.016. Protein identification was performed by searching the Manihot esculenta reference sequence database of UNIPROT (Manihot esculenta, UP000091857). Mass spectra data in this study were deposited to ProteomeXchange repository: PXD062377 (https://proteomecentral.proteomexchange.org/cgi/GetDataset?ID=PXD062377) and jPOST repository: JPST003734 (https://repository.jpostdb.org/entry/JPST003734)17. Proteins with a q-value greater than 0.95 were selected for further analysis. Missing intensity values were imputed using the MissForest package in R, which employs a random forest-based approach. Protein intensities for individual samples were normalized against the total intensity of each respective sample and expressed as percentage expression ratios.

Differentially abundant proteins between the tolerant (R72) and susceptible (HN) cultivars were calculated as the log2 fold change of R72 relative to HN. Statistical significance was assessed using the false discovery rate (FDR), and the results were visualized as volcano plots generated with the ggplot2 package19 in R. K-means clustering (KMC) was performed to explore expression trends of significantly differentially abundant proteins in R72 over time. Clustering was supervised by specifying 7 clusters and 100 iterations, with the optimal number of clusters determined using Gap-statistic analysis20,21 via the factoextra package21 in R. Correlations between the mean expression of each cluster and observed symptoms were analyzed using the corrplot package in R22.

The subcellular localization of proteins was predicted by submitting protein sequences to the Plant-mSubP website, which identifies target localization within plant cells23. Protein sequences were analyzed using the PseAAC-NCC-DIPEP feature set, which incorporates pseudo-amino acid composition, N-terminal and C-terminal amino acid composition, and dipeptide composition.

Functional annotation and enrichment were calculated using the Mercator4 website24. The identified proteins were subjected to functional prediction and enrichment analysis in cassava using the MapMan BIN ontology with 28 main biological categories.

Determination of leaf malondialdehyde

Malondialdehyde (MDA) content was determined using 2-thiobarbituric acid (TBA) following the method described by Sarker et al., 201825. Fresh weights of ground leaves (L3 and L4) and stems (including the inoculation point) were recorded prior to homogenization with 800 µL of 0.6% TBA in 10% trichloroacetic acid (TCA). The mixture was heated at 95 °C for 15 min and then cooled on ice. After cooling, it was centrifuged at 12,000 × g for 5 min, and the absorbance of the supernatant was measured at 450, 532, and 600 nm. MDA content was calculated based on fresh weight (FW) and expressed as nmol/g FW using the following equation: MDA = 6.45 × (A₅₃₂ − A₆₀₀) − 0.56 × A₄₅₀.

Quantitative reverse transcription polymerase chain reaction (qRT-PCR) analysis

To validate the candidate proteins involved in resistance to CBB, qRT-PCR was performed to assess the expression levels of the corresponding RNA. Fourth leaves from infected two cassava cultivars were immediately snap frozen in liquid nitrogen and stored at −80 °C until used. Total RNA was extracted from the fourth leaves of R72 and HN, with 5 samples collected from each time point (0, 3, 7, 14, and 28 DAI) using Fruit-mate™ (Takara, Kusatsu, Japan) and TriReagent (Molecular Research Center, Cincinnati, OH). Contaminating DNA was removed by treatment with RQ1 RNase-Free DNase (Promega, USA). Then cDNA was synthesized using a Viva cDNA synthesis kit (Vivantis, Shah Alam, Malaysia) primed with oligo (dT) as recommended by the manufacturer. For gene expression analysis, cDNA templates were amplified using KAPA SYBR® FAST qPCR kit (Thermo Fisher Scientific, Waltham, MA) by Mastercycle ep Realplex (Eppendorf) with selected primers (Table S2) that showed efficiency, slope, and R2 values according to acceptable ranges as the values of slope between − 3.6 and − 3.1, %efficiency between 90 and 105, and correlation coefficient (R2) between 0.98 and 1.00. The expression level of each gene in each sample was measured in triplicate and calculated as 2-ΔCT where CT is the cycle threshold of the gene of interest normalized against actin-7 gene (XM_021777064.2). Statistical analysis was conducted by ANOVA pair test and Duncan’s test for multiple tests.

Results

Symptom evaluation

To assess the resistance levels between the tolerant (R72) and susceptible (HN) cassava cultivars, disease symptoms were evaluated at the inoculation site, located between the third and fourth fully expanded leaves of the infected plants. Visible symptoms were scored 7 days after inoculation (DAI) in both cultivars, with HN showing a significantly higher symptom score (Fig. 1a). Disease progression in HN was further evidenced by leaf wilting observed at 28 DAI. Symptom scores between the two cassava cultivars began to diverge significantly at 7 days after inoculation (DAI). R72 did not exhibit necrosis at the inoculation site with a score exceeding 1. Disease progression over time was assessed by calculating AUDPC values, as shown in Fig. 1b. A significant difference (p < 0.01) in AUDPC was observed between the infected cassava cultivars. The lower AUDPC value indicates superior tolerance of R72. However, Xam-infected stem tissues in R72 showed some damage, confirming that R72 is tolerant while HN remains a susceptible cultivar.

Fig. 1
figure 1

(a) Disease progression in Rayong 72 (R72) and Hanatee (HN) cassava cultivars after XamC infection via stem inoculation described by symptom score. The symptom scores were statistically analyzed using Duncan’s test to compare days after inoculation (DAI) of R72 (presented as small letters) and HN (presented as capital letters). The same letters were not significantly different (p > 0.05). Asterisks indicate statistical significance based on ANOVA between cultivars with the same DAI: *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001. (b) The area under the disease progression curve (AUDPC) is the mean value ± standard deviation of five biological replications. Statistical analysis was performed using ANOVA to compare the differences between cultivars, and values indicated by the same letter are not significantly different (p > 0.05).

Full size image

Proteomic data analysis

Total protein was extracted from the fourth fully expanded leaf under the inoculation point at each DAI, with five biological replicates per DAI. The mass spectra were matched to Manihot esculenta in UNIPROT database. The datasets analyzed during the current study are available in the jPOST repository and ProteomeXchange repository, JPST003734 (https://repository.jpostdb.org/entry/JPST003734) and PXD062377 (https://proteomecentral.proteomexchange.org/cgi/GetDataset?ID=PXD062377). The intensity of each protein was normalized to the total intensity of each individual sample to obtain the percentage expression ratio explaining the protein expression. Sample variation was visualized using principal component analysis (PCA) (Fig. 2) and showed a distinct separation between R72 and HN, with R72 samples located on the negative x-axis and most HN samples on the positive x-axis. Additionally, samples with the same DAI were clustered together in the PCA plot. PCA provided an overview of the protein expression in our samples related to DAI and cultivars, highlighting the different protein expression profiles among the two cultivars that were induced by bacterial infection. To identify proteins responsive to XamC infection, the log2 fold changes in protein expression between the R72 and HN cultivars were calculated at each day after inoculation (DAI) and visualized using volcano plots (Fig. 3). Proteins with an FDR < 0.01 and a fold change > 2-fold change (log2 fold change ≥ 1 or ≤ −1) were considered as proteins with significantly changed expression. High protein counts were found at the start of the sample collection date (0 DAI), indicating a background difference between the two cassava cultivars, which should be subtracted from other DAIs. R72 exhibited the highest number of upregulated proteins (1,425) at 7 DAI. This time point, which corresponds to the initial appearance of disease symptoms, may represent a critical phase in the protein response to XamC in the tolerant cultivar. In contrast, HN showed the highest number of upregulated proteins (2,979) at 28 DAI. Differentially abundant proteins in R72 were further analyzed to identify protein groups potentially involved in the XamC response.

Fig. 2
figure 2

Principal component analysis plot of every sample. Squares indicate R72, and circles indicate HN. Color labels are distinguished by DAI: blue (0 DAI), green (3 DAI), yellow (7 DAI), red (14 DAI), and purple (28 DAI).

Full size image
Fig. 3
figure 3

Volcano plot of differential protein expression of every DAI protein, with log2 fold changes as the x-axis and -log10 (FDR) of significat value as the y-axis.

Full size image

Differentially abundant proteins clustering

Differentially abundant proteins in the tolerant cultivar were studied to understand the defense mechanisms against infection of Xam in cassava. First, significantly differentially abundant proteins at 0 DAI were subtracted from the list of proteins after 3 DAI, which could clarify the dynamics of protein changes in response to Xam infection in the tolerant cultivar. The trend of protein expression for significantly differentially abundant proteins at each DAI was classified into seven clusters using KMC, based on Gap-statistics analysis. Seven clusters were represented, and the mean expression values are shown in Fig. 4. The expression trends differed, with some peaking after infection at 3 DAI (clusters 1, 2, and 6), at 7 DAI (clusters 4 and 7), and at 14 DAI (cluster 5). However, clusters 4 and 7 were particularly interesting, as disease symptoms were first observed at 7 DAI in R72. The subcellular localization of each protein in the seven clusters is presented (as a percentage) in Figure S1. Most subcellular localization predictions for each cluster indicated a high proportion of nucleus-expressed proteins, with the highest proportion in cluster 4. However, the distinct results of cluster 7 showed that the majority of proteins were predicted to be in the plastid. Based on this information, it can be inferred that cluster 7 showed distinct behavior compared to the other clusters and this might be the most responsive cluster in the tolerant cultivar related to symptom progression.

Positive values in x-axis indicate upregulated protein in R72 (tolerant) and negative value in x-axis indicate upregulated protein in susceptible HN. Colored dots represent proteins that are upregulated proteins in a particular cultivar (fold change > 2) and significant value (FDR < 0.01).

Fig. 4
figure 4

K-means clustering of upregulated R72 proteins.

Full size image

Oxidative stress marker determination

ROS can induce plant stress during pathogen infection. In this study, MDA levels, a ROS indicator, were measured in the L3, L4 leaves and the stems of infected plants, and similar levels were observed in the two cultivars (Fig. 5). The only significant difference in R72 in MDA level was observed at 14 DAI in L4 (Fig. 5a). Although MDA levels in the tolerant cultivar were generally comparable to those in the susceptible cultivar, disease progression remained limited, as leaf wilting in the susceptible cultivar only reached scale 2 in the late stage. In contrast, stem MDA levels were higher in HN (Fig. 5c), accompanied by necrosis around the inoculation site.

Fig. 5
figure 5

Malondialdehyde (MDA) level in L4 (a), L3 (b), and stem (c) of R72 and HN at five time points. Statistical analysis was performed using Duncan’s test to compare the DAI of R72 (presented as small letters) and HN (presented as capital letters), and the same letters were not significantly different (p < 0.05). Asterisks indicate level of statistical significance between cultivars at the same DAI based on ANOVA: * p≤ 0.05, ** p≤ 0.01, ***p ≤ 0.001.

Full size image

Differences between the two cultivars were observed at 7, 14, and 28 days after inoculation (DAI). In the susceptible cultivar, elevated stem MDA levels correlated with significantly higher symptom scores, while in the tolerant cultivar, they were associated with slower disease progression. These findings suggest that MDA levels reflect differences in bacteria-induced oxidative stress both at the infection site and in distal tissues of cassava.

Correlation between proteomic and phenotypic results

The relationship between protein dynamics and observed symptoms using a correlation matrix between the protein expression trend in each cluster and observed symptoms (disease score, AUDPC, and MDA) in L3, L4, and the stem of R72 to provide evidence for candidate tolerant protein selection is displayed in Fig. 6. A positive correlation was found between the AUDPC and MDA of L4, which describes the concordance of the observed score and ROS-induced symptoms. In addition, MDA in L4 showed a positive correlation with both L3 and the stem. Regarding the correlation between protein expression trends and AUDPC, only significant negative correlations (p < 0.05) were detected in clusters 1, 2, and 7. Moreover, cluster 7 showed a negative correlation with L4 and L3, which may reflect the ROS management function of this cluster. Proteins in cluster 7 were of particular interest due to their involvement in oxidative stress, the expression trends observed in the 7 DAI clusters, and the high number of plastid proteins predicted based on subcellular localization. As a result, the plastid-expressed proteins in cluster 7 were selected as candidate proteins for further investigation. To visualize the function of each protein cluster, functional enrichment was classified and analyzed using MapMan BIN (Fig. 7). The defense-related protein group emerged as a significant category of proteins in the tolerant cultivar. In cluster 7, which correlated with AUDPC and MDA, coenzyme metabolism proteins were significantly enriched (p < 0.05). This group of proteins was selected as candidates from cluster 7 to obtain more information on oxidative stress related to the pathogen response in tolerant cassava cultivars.

Fig. 6
figure 6

Correlation matrix of protein expression and phenotypic trend. R-square values are displayed in the right-hand legend. The size of circles in the matrix indicates the significance level (p-value), presented as *p < 0.05, **p < 0.01, and ***p < 0.001.

Full size image
Fig. 7
figure 7

Functional enrichment and p-value in each K-means clustering cluster.

Full size image

qPCR validation

The selection of candidate proteins from cluster 7 in the KMC was based on the functional annotation of NAD/NADP biosynthesis, including nudix hydrolase 2-like (NUD2) and glutamine-dependent NAD(+) synthetase-like (QNS). NAD/NADP is associated with various mechanisms, such as signaling, plant immunity, and ROS metabolism, which can be explained by the MDA accumulation seen in the samples. Hence, proteins involved with those mechanisms and with plastid subcellular localization were selected from the list of top 10 functional enrichments as shown in Fig. 8. The candidate proteins in cell signaling were selected including those involved with calcium signaling (solute transport) and kinase (multi-process regulation) which were CBL-interacting protein kinase 18-like (CIPK18), CSC1-like protein (CSC1), and mitogen-activated protein kinase kinase kinase 18-like (MAPKKK18). In addition, the significant protein list in cluster 7 contained two proteins involved in the oxidative stress response, namely glutathione synthetase (GSS, redox regulation) and lycopene epsilon cyclase (LCYE, carotenoid biosynthesis).

Total RNA was extracted from the same leaf tissues used in the proteomic analysis, and the expression levels of the candidate genes at each DAI were quantified by qRT-PCR, using actin as the reference gene for comparison (Fig. 8). The proteomic results showed that five genes had significant differences ( p< 0.05) between the two cultivars at 7 DAI, when symptoms were initially observed, including NUD2, QNS, CIPK18, CSC1, and GSS. Some genes, including CIPK18 and GSS, exhibited the highest significant peak at 7 DAI in the R72 sample. At 0 DAI, the expression levels of NUD2, CIPK18, CSC1, GSS, and LCYE were distinct from that of the proteomic results, which showed no differences between R72 and HN. However, concordance between proteomic and gene expression results were explained by using log2 fold change (R72/HN) correlations (Table S1). Highly similar patterns of proteomic and qPCR results were observed for QNS, MAPKKK18, and GSS, which showed positive correlations of 0.67, 0.84, and 0.76, respectively. The consistency between the two investigations of these three proteins suggests they are suitable as tolerant protein markers.

Fig. 8
figure 8

Expression level of candidate genes including (a) nudix hydrolase 2-like (NUD2), (b) glutamine-dependent NAD(+) synthetase-like (QNS), (c) CBL-interacting protein kinase 18-like (CIPK18), (d) CSC1-like protein (CSC1), (e) glutathione synthetase (GSS), (g) mitogen-activated protein kinase kinase kinase 18-like (MAPKKK18). Statistical analysis was performed using Duncan’s test to compare the DAI of R72 (presented as small letters) and HN (presented as capital letters), and the same letters were not significantly different (p < 0.05). Asterisks indicate the level of statistical significance between cultivars at the same DAI based on ANOVA: *p ≤ 0.05, ** p≤ 0.01, *** p≤ 0.001.

Full size image

 Discussion

Cassava is an important agricultural crop grown in West Africa, tropical South America and South and Southeast Asia. Cassava is both used as a food source, as well as an important raw material for industrial processes26. There are many varieties of cassava, which are generally classified as either sweet or bitter based on their cyanide content27. Sweet (low-cyanide) varieties are typically used as a food source, whereas bitter (high-cyanide) varieties are mainly cultivated for industrial purposes.

Different cassava varieties exhibit varying levels of tolerance or susceptibility to a range of abiotic stressors, including aridity, salinity, and heat28,29,30. Like many industrial-scale crops, cassava is exposed to numerous biotic stressors, including pathogens such as cassava bacterial blight, caused by X. axonopodis pv. manihotis (Xam). Resistance or susceptibility to these pathogens varies among cassava varieties. Such diseases can lead to devastating outbreaks, wiping out crops and causing substantial economic losses31. Therefore, understanding the genetic and physiological basis of disease tolerance and susceptibility is crucial for the effective propagation of tolerant varieties, particularly in regions where these pathogens are prevalent.

One of the cassava varieties used in this study, HN, is classified as a sweet variety, while the other, R72, is considered a bitter variety. Notably, the bitter variety (R72) exhibited the highest tolerance to Xam infection, which is unexpected, as sweet varieties are generally reported to show greater tolerance to pathogens32,33. Previous studies have documented that HN is susceptible to both Xam34 and Colletotrichum gloeosporioides f. sp. manihotis35,36 following stem inoculation, findings that are consistent with the results observed in this study.

In terms of oxidative stress levels related to disease progression, MDA levels at infected sites in various plant species changed after pathogens infection, including the Capsicum annum response to Xanthomonas campestris37 and Arabidopsis thaliana response to Botrytis cinerea38 which supports the association of higher MDA levels with susceptibility in this study. The MDA levels in the stems of HN showed marked differences compared to R72, whereas in the distal organs, specifically L3 and L4 leaves where the levels were similar in both cultivars. Moreover, the differences in leaf symptoms suggest that the greater MDA tolerance observed in R72 may result from more effective MDA utilization during the biotic stress response, as previously reported in Arabidopsis³⁹. Reactive carbonyl species, such as MDA, are involved in ROS signaling and antioxidant defenses. Systemic acquired resistance can be induced by ROS acting as signaling molecules40,41. Salicylic acid (SA), the key plant growth regulator in systemic acquired resistance (SAR), can induce ROS production, and a higher level of MDA was found in a tolerant cultivar at the late stage of infection42.

In the study of MebHLH149 transcription factor silencing in cassava which interacts with resistance genes in response to XamC, an increase in susceptibility and a reduction in reactive oxygen species (ROS) were observed43. Lower levels of MDA or ROS in susceptible cultivars may be the result of bacterial suppression of ROS production, allowing the pathogen to detoxify the surrounding environment and enhance its survival44. These findings suggest that ROS not only contribute to stress in plants but also play a role as immune signaling molecules when maintained at non-toxic levels.

In the proteomic response, bacterial infection differentially modulated the protein expression profiles. However, a large number of background proteins showed no significant differences between the two cassava cultivars. The differentially abundant proteins observed after induction with Xam may be associated with MDA, as suggested by the correlation analysis. This association could be linked to defense mechanisms, both directly and indirectly. In terms of the response period, 7 DAI may be a critical point for the pathogen to induce a response in R72. In addition to defense-related protein function, subcellular localization of proteins is strongly related to pathogen response, especially in chloroplasts, as a battlefield between plants and pathogens45. Therefore, the differentially abundant proteins in cluster 7 of R72 are of particular interest, as they show an enrichment of NAD/NADP biosynthesis proteins. NAD/NADP are utilized in the bacterial response not only for photosynthesis and energy metabolism but also for ROS signaling in cells46. The plausibility of tolerance-related candidate proteins was evaluated by analyzing the correlation between log2 fold changes from proteomic and qPCR analyses in tolerant and susceptible cultivars. Some genes exhibited inconsequential correlations, which may be attributed to factors such as translational efficiency47 protein turnover rates under stress conditions48 and post-transcriptional regulation49. Nevertheless, certain candidates showed positive correlations such as those involved in NAD metabolism (QNS), signaling (MAPKKK18), and glutathione-mediated antioxidant activity (GSS) suggesting their potential roles as key players in tolerance mechanisms, particularly in relation to MDA.

Tolerant R72 may enhance resistance through the regulation of NAD/NADP metabolism, particularly by increasing NAD production via QNS to maintain the cellular NAD pool. In citrus plants responding to Xanthomonas citri, exogenous application of NAD has been shown to boost resistance50. Similarly, in Arabidopsis, phosphorylated NAD (NADP) has been reported to activate pathogenesis-related genes and trigger salicylic acid (SA)-mediated responses, in conjunction with calcium signaling, ultimately leading to increased resistance against Pseudomonas syringae51. The interaction of NAD with other signaling pathways, such as the induction of the cytosolic calcium ion pool by NAD in Arabidopsis52 leads to calcium ions that can sense pathogen infection and promote the signal cascade to induce SAR53. However, in this study, the calcium signaling proteins CSC1 and CIPK18 did not appear to be reliable markers of tolerance, due to the low correlation observed between their protein and gene expression levels. In addition to its role in the immune response, calcium signaling is known to interact with other defense pathways. Crosstalk between calcium signaling and MAPK pathways54 as well as between calcium signaling and ROS signaling55 has been reported and may contribute to the broader cascade of defense mechanisms. Furthermore, NAD is related to abscisic acid (ABA) synthesis and signaling56 in which ABA interacts with MAPKKK1857 and affects resistance to Ralstonia solanacearum in tomatoes58. Therefore, the increased expression of NAD synthesis-related proteins and genes observed in R72 may be associated with resistance mechanisms in cassava’s response to XamC.

Maintenance of ROS at harmless levels is essential to maintain the stimulating signal in plants because of the increase in glutathione and carotenoid levels after infection59. However, in this study, carotenoids may not serve as a reliable marker due to the weak correlation between LCYE protein and gene expression. Additionally, glutathione mediated by increased GSS production in tolerant cultivars may play an effective protective role, as indicated by symptom scores that remained below 1 and showed a negative correlation with MDA accumulation in the leaves. Furthermore, crosstalk between glutathione and salicylic acid (SA) was observed in Arabidopsis during ROS signal transmission-induced pathogen responses60.

In conclusion, the varying resistance levels of Xam-infected cassava cultivars were associated with MDA accumulation at the infection site. In contrast, MDA levels in the leaves may reflect the involvement of ROS in systemic acquired resistance (SAR) signaling. Notably, the proteomic profile of the tolerant cultivar R72 suggests that NAD metabolism and glutathione-mediated oxidative stress regulation play key roles in its tolerance to Xam infection. These findings provide valuable insights into the molecular mechanisms underlying disease resistance in cassava and may inform future strategies for breeding or engineering cassava varieties with enhanced tolerance.