Introduction

Modern citriculture is currently threatened by the disease Huanglongbing (HLB), also known as "citrus greening". Due to its rapid dispersal, severity, and fast progression, HLB is currently considered the most devastating bacterial disease affecting citrus production and quality worldwide. HLB has been reported in Asia, the Middle East, Africa, and the Americas but it is expected to continue spreading. Hence, the European Commission qualified HLB as a quarantine priority pest in 2019 (Gabriel et al., 2020). The disease is associated with at least three uncultured alpha-proteobacteria species of the genus Candidatus Liberibacter, Ca. L. asiaticus (CLas), Ca. L. africanus (CLaf), and Ca. L. americanus (CLam), which are vectored by two psyllid species into the phloem cells. CLas and CLam are spread in citrus by the Asian psyllid Diaphorina citri Kuwayama, whereas CLaf is transmitted by Trioza erytreae (Bové, 2006; Gottwald, 2010). The three bacterial species are phloem-limited but differ in their thermosensitivity, the severity of the induced symptoms, and each appears to have independently evolved on the continent indicated in its name (Nelson et al., 2013; Wang et al., 2017).

To date, there are no methods for long-term control of HLB, making it challenging to curb the spread of this disease. Given the absence of a cure, the majority of citrus producers have opted to coexist with infected trees rather than completely removing them. Within the range of disease control strategies that have been assessed to reduce the impact of the disease are antimicrobial compounds potential resistance inducers, phytohormones, thermotherapy, improved nutritional programming (ENP) and symptomatic branch trimming (Gottwald et al., 2012; Zhang et al., 2014; Doud et al., 2017; Killiny et al., 2020). Unfortunately, implementing all these approaches would be excessively costly. Hence, there is a requirement to conduct further assessment of disease control strategies aimed at addressing pathogen spread on three distinct levels; pathogen, host and vector.

Bacterial diseases of plants are typically very challenging to control. Unluckily, the HLB pathosystem is problematic for laboratory studies. Advances in understanding the pathogenesis of CLs have been hindered because the bacteria had not been successfully cultured long-term (Ha et al., 2019; Zuñiga et al., 2020), and pure cultures have yet to be maintained. Additionally, the specific characteristics of the HLB infection, such as the uneven distribution of the bacteria, the low CLas titer during the initial phase of the latent period, and the influence of the environment on CLas replication, not only hampered the development of detection methods and treatments, but also led to confusing results in resistance analysis (Alves et al., 2021).

Consequently, there is a need for unequivocal resistant citrus cultivars, which can be utilized in breeding programs or directly as potential rootstocks or interstocks, to mitigate the harmful effects caused by this disease. A wide variety of HLB resistance levels, including citrus relatives described as full-resistant, has been reported in the family Rutaceae, subfamily Aurantioideae (Cifuentes-Arenas et al., 2019; Folimonova et al., 2009; Wang, 2019). However, they are phylogenetically distant from commercial citrus and breeding hybrids are unviable. Taking all of the above into consideration, Alves et al., (2021, 2022) tested a diverse collection of close cross- and/or graft-compatible citrus relatives belonging to the subtribe Citrinae for CLas resistance. Utilizing an accurate and consistent evaluation system, they have demonstrated full resistance to CLas in a small group of the analyzed species (Alves et al., 2021, 2022).

Omics-based approaches constitute a powerful tool to understand the citrus response to the pathogen, and for performing comparative analyses among citrus cultivars with different levels of resistance to HLB. In particular, metabolomics has been previously utilized to study the mechanisms supporting this resistance/tolerance. For example, Killiny (2016), reported that the phloem sap of HLB-resistant curry leaf tree contains lower amounts of most metabolites compared to the partially resistant citrus orange jasmine and the susceptible Valencia sweet orange. Different studies on others citrus cultivars have shown a positive correlation between tolerance to HLB and most amino acids and leaf volatiles, while a negative correlation was observed for several organic acids (Hijaz et al., 2016; Killiny & Hijaz, 2016; Killiny et al., 2017). Valim and Killiny (2017) analyzed the fatty acid composition of the phloem sap of 12 citrus varieties with different tolerance levels toward HLB and found that capric acid was higher in the “Sugar Belle” and Carrizo citrange hybrids which show tolerance to HLB. Interestingly, Albrecht et al. (2020) demonstrated that rootstock influences the metabolic response to CLas in grafted sweet orange trees. Other comparative metabolomics studies have suggested that tolerance is linked to preserving plant growth and phloem formation rather than activating plant defence mechanisms to overcome the disease (Suh et al., 2021).

Herein, we have taken advantage of the strict evaluation of the resistance to CLas of different Citrinae genotypes carried out by Alves et al., (2021, 2022). We selected three genotypes from each of the defined categories: susceptible, partially resistant, and resistant. Notably, the latter two categories included cultivars formerly classified as Microcitrus spp. and Eremocitrus glauca, which are now in the genus Citrus and are used in some breeding programs. However, the basis of HLB resistance in these citrus relatives remains enigmatic. To determine whether the level of resistance depends on the metabolic background of the different cultivars before infection, we carried out a metabolomics analysis of healthy phloem plants.

Materials and methods

Plant material and plant growing conditions

The nine healthy citrus varieties used in the present study were maintained in a greenhouse with a temperature of 30 ºC and a relative humidity of 35%. These Citrinae species were previously assessed and classified into three categories; susceptible, partially resistant, and resistant by Alves et al. (2021). They include Citrus x sinensis 'Pera', Poncirus trifoliata 'Rubidoux' and Citrus x sinensis ´Valencia Midknight' (susceptible), Microcitrus australasica hybrid, Microcitrus virgata hybrid and Microcitrus garrawayae (partially resistant), Microcitrus australis hybrid, Eremocitrus glauca and Microcitrus warburgiana hybrid (resistant) (Table 1) (Alves et al., 2022). All of these citrus varieties were 3-year-old scions, approximately 50–60 cm in height and grafted on Citrus macrophylla with the exception of the Microcitrus australis hybrid which was propagated on Citrus volkameriana. At least three independent mature trees were used for each species and hybrid. All scions were propagated using buds from a single donor mother plant per genotype and were kept at an insect-proof climate-controlled greenhouse at the Institute for Plant Molecular and Cell Biology (UPV-CSIC, Valencia, Spain) with temperatures maintained between 25-27ºC. The plants were grown in 8 L pots filled with coir, irrigated and fertilized twice a week, and sprayed monthly with preventive insecticides.

Table 1 Citrinae genotypes/accessions used in this study
Full size table

Phloem sap collection

Stems measuring 10–20 cm in length with a diameter of 0.2–0.3 cm were collected from greenhouse grown healthy plants. Phloem sap was collected using the centrifugation method described by Hijaz and Killiny, (2014). Briefly, the bark was manually removed from the sprig and washed with deionized water to eliminate any potential contamination from the xylem sap. To collect the phloem sap, five small pieces of the bark were placed in a 0.5-mL microcentrifuge tube with a small hole at the bottom, and this tube was then nested into a 2-mL microcentrifuge tube. The sample was centrifuged at 12,000 rpm for 15 min at room temperature, and the collected phloem sap was stored at -80 ºC until further analysis.

Primary metabolite analysis

Primary metabolite analysis was conducted at the Metabolomics Platform of the Institute for Plant Molecular and Cell Biology (UPV-CSIC, Valencia, Spain). The method used was a modification of Roessner et al. (2000). For the analysis, one µL of each sample of phloem sap was dried. The dry residues were redissolved in 40 µL of 20 mg/mL methoxyamine hydrochloride in pyridine and incubated for 90 min at 37 ºC. Then, 70 µL MSTFA (N-methyl-N-[trimethylsilyl]trifluoroacetamide) and 6 µL of a retention time standard mixture (3.7% [w/v] mix of fatty acid methyl esters ranging from 8 to 24C) were added and the samples were incubated for 30 min at 37 ºC.

Sample volumes of 2 µL were injected in split1:30 and splitless mode in a 6890 N gas chromatograph (Agilent Technologies Inc. Santa Clara, CA) coupled to a Pegasus4D TOF mass spectrometer (LECO, St. Joseph, MI). Gas chromatography was performed on a BPX35 (30 m × 0.32 mm × 0.25 μm) column (SGE Analytical Science Pty Ltd., Australia) with helium as the carrier gas, constant flow 2 mL/min. The liner was set at 250 °C. The oven program was 85 °C for 2 min, 8 °C/min ramp until 360 °C. Mass spectra were collected at 6.25 spectra s − 1 in the m/z range 35–900 and ionization energy of 70 eV. Chromatograms and mass spectra were evaluated using the CHROMATOF program (LECO, St. Joseph, MI). Metabolites were identified by comparison with a custom library made with commercial standards.

Data analyses were performed using Metaboanalyst, version 5.0 (https://www.metaboanalyst.ca). Logarithmic transformation and autoscaling scaling were employed as normalization to perform the principal component analysis and the hierarchical clustering and also to generate the heatmap. In the context of hierarchical clustering, the Euclidean distance metric and the Ward clustering algorithm were utilized as parameters to identify features with significantly different accumulation.

Real-time quantitative reverse transcription PCR

Total RNA was extracted from a pool of three leaves of each Citrinae healthy species using RIBOzol Reagent. Remnant genomic DNA was removed by DNase I treatment. First-strand cDNA was synthesized from 0.5 µg of total RNA using RevertAid H Minus Reverse Transcriptase and oligo(dT) (Thermo Fisher Scientific, Carlsbad, CA, USA). Real-time quantitative PCR (qPCR) was carried out using QuantStudio 3 Real-Time PCR machine (Applied Biosystems, Waltham, MA, USA) and PyroTaq EvaGreen qPCR Supermix (Solis BioDyne, Tartu, Estonia), specific oligonucleotides primers, and recommended qPCR cycles as follows: initial denaturation for 12 min at 95 °C, followed by 50 cycles of 15 s at 95 °C and 60 s at 60 °C. Specific oligonucleotides primers were designed using Primer3web version 4.1.0 (https://bioinfo.ut.ee/primer3). Oligonucleotide efficiencies were tested by qRT-PCR using ten-fold serial dilutions of the corresponding cDNA. Each technical replicate was run in triplicate. SAND (SAND family protein) and GAPC2 (Glyceraldehyde-3-phosphate dehydrogenase C2) genes were used as endogenous controls (Mafra et al., 2012; Máximo et al., 2017; Shi et al., 2018; Wang et al., 2016). The primer sequences of both target and reference genes are shown in Supplementary Table S1.

Statistical analysis

The comparison of transcript abundance of the different genes analyzed across the different samples was carried out using GraphPad Prism version 6.00 for Windows, GraphPad Software, La Jolla California USA, (www.graphpad.com). A non-parametric Kruskal–Wallis one-way ANOVA was performed. Statistically significant differences were considered for p-values less than 0.05. A post hoc test, the Dunn test for multiple comparissons, was conducted to identify which specific groups differ from each other.

Results

Metabolic variations in the phloem of HLB-resistant, partially resistant, and susceptible citrus accessions

Considering that the citrus phloem sap is the plant tissue in which CLas must survive and grow, we performed a comparative metabolomics analysis of the phloem sap from nine healthy Citrinae genotypes exhibiting different levels of resistance to CLas (Table 1). Partial resistance was assigned to species/hybrids showing CLas infection in a fraction of tested scions and/or presenting a delayed infection, without ever reaching the bacterial titre found in susceptible genotypes (Alves et al., 2021, 2022). A diverse range of molecules was detected after trimethylsilyl derivatization and classified for subsequent analysis. They included 20 amino acids, 12 organic and inorganic acids, 15 sugars and sugar acids, 2 amines, 2 nitrogenated bases, and 2 unknown compounds (Supplementary Table S2A and S2B). Their retention time, quantification ion and injection mode (split 1:10/ splitless) are shown in Supplementary Table S3. The higher percentage of detectable metabolites were sugars, except for the HLB-resistant R2 Eremocitrus glauca, in which organic acids were more abundant (Fig. 1a).

Fig. 1
figure 1

(a) Percentage distribution of the main groups of metabolites found in the phloem sap of Citrus x sinensis ´Pera´ (S1), Poncirus trifoliata Rubidoux (S2), Citrus sinensis valencia Midknight (S3), Microcitrus australasica (PR1), Microcitrus virgata (PR2), Microcitrus garrawayae (PR3), Microcitrus australis (R1), Emocitrus glauca (R2) and Microcitrus warburgiana (R3). Proteinogenic amino-acids (PAAs), Non-proteinogenic amino-acids (NPAAs). (b) Two-way hierarchical cluster analysis and the heatmap showing the distribution of different citrus varieties using all of the detected compounds in phloem saps. Rows represents the different cultivars; Susceptible in green (S1, S2 and S3), Partially Resistant in red (PR1, PR2 and PR3) and Resistant in blue (R1, R2 and R3). Columns represent compound concentrations. Red and blue indicate high and low concentration, respectively. Colour density indicating levels of fold change was displayed

Full size image

We conducted a principal component analysis (PCA) and a total of 7 principal components (PCs) were obtained, accounting for 88.8% of the total variance. The total variation explained by the first two principal components was 56%, with PC1 contributing 40% and PC2 contributing 16% (Supplementary Fig. S1a). Notably, the partially resistant Microcitrus australasica (PR1) appeared as an outlier in PC1, and this separation was primarily attributed to the elevated levels of many compounds in PR1, as confirmed by the loading plot (see Supplementary Fig. S1b). This distinct profile was also evident in the heatmap, which displayed the fold change for each compound identified through gas chromatography and mass spectrometry analysis (Fig. 1b). To mitigate the influence of PC1 on the variance, we opted to exclude this outlier from the subsequent PCA analysis. As shown in Supplementary Fig. S1c, the PCA analysis without PR1 revealed that susceptible samples clustered in the right lower quadrant, indicating their similarity in terms of metabolite profiles.

Organic and inorganic acid accumulate in higher amounts in both partially resistant and resistant citrus samples

A total of eleven organic acids and one inorganic acid (phosphoric acid) were detected in the phloem sap samples. Remarkably, the average percentage of the total organic and inorganic acids was higher in both partially resistant and resistant samples when compared to the susceptible ones (Fig. 2a and Supplementary Table S4). Among the detected organic acids were dicarboxylic acids such as fumaric, malic, malonic, oxalic, and succinic.

Fig. 2
figure 2

(a) Bar graph represents the average percentage of total organic acids calculated from the Total Ion current Chromatogram (TIC) vs. the three analyzed categories: Susceptible (S), Partially Resistant (PR), and Resistant (R) (no significant difference in one-way ANOVA; α 0.05, P 0.28). (b) Scot plot of principal component analysis (PCA) visualizing recognizable differences in the composition of all detected phloem organic and inorganic acids between the resistant (R1, R2 and R3 in blue) and susceptible (S1, S2 and S3 in green) samples after derivatization with Trimethylsilyl (TMS)

Full size image

Malic and quinic acid were generally the most abundant organic acids, however, only quinic acid was found to be over-represented in PR and R samples, with the exception of the R3 sample (Supplementary Fig. S2). While the percentage of malic acid was similar in susceptible and resistant samples, we observed differences in quinic acid accumulation between both categories. Quinic acid was the most abundant organic acid in 83% of the resistant and partially resistant samples, as well as in one susceptible sample. On average, the percentage of this organic acid was higher in resistant cultivars. Additionally, the average of each organic acid percentage within the three categories revealed that oxalic acid, fumaric acid, threonic acid, lactic acid, and quininic acid were elevated in susceptible samples. By contrast, citric acid and anthranilic acid were higher in resistant cultivars (Supplementary Figs. S3a y b). The distribution of individual organic acid within each accession did not exhibit any differential pattern among the three categories (Supplementary Fig. 2).

PCA analysis generated using the organic and inorganic acids did not reveal any separation among the groups with different resistance levels (data not shown). However, when PCA analysis was conducted by excluding PR samples, it distinguished resistant and susceptible samples in two distinct groups, with PC1 contributing 25.4% of the variation (Fig. 2b). Resistant varieties clustered on the left side of the PCA plot, indicating a distinct organic acid composition in the phloem sap compared to susceptible varieties. The corresponding loading plot showed that the majority of organic acid levels in resistant cultivars were higher than in the susceptible ones, although a high variability is patent (Supplementary Fig. S4).

In conclusion, the percentage of total organic acids was higher in the R and PR samples. Quinic acid was in general the most abundant organic acid and displayed a tendency to have higher content in PR and R compared to S samples.

Sugars showed a higher accumulation in the susceptible citrus genotypes

The average percentage of total sugars showed a noticeable increasing trend in the susceptible group although differences were not statistically significant (Fig. 3a and Supplementary Table S5). The identified sugars included monosaccharides such as fructose, glucose, and rhamnose, disaccharides such as sucrose, and trisaccharides such as raffinose. Additionally, several sugar alcohols, like galactinol and myo-inositol, as well as sugar acids were also detected. Interestingly, there was a trend toward an average percentage increase of sugar acids in PR and R accessions, being more pronounced in the R group (Fig. 3b). This finding agrees with that previously reported in Sugar Belle mandarin (Killiny et al., 2017), and the fact that sugar acids are induced after CLas infection in Cleopatra mandarin (Albrecht et al., 2016) and sweet orange (Hijaz et al., 2013).

Fig. 3
figure 3

Bar graph represents the average percentage of total sugars and sugar acids (no significant difference in one-way ANOVA; α 0.05, P.0.43) (a) and sugar acids (no significant difference in one-way ANOVA; α 0.05, P.0.39) (b) vs the three analyzed categories; Susceptible (S), Partially Resistant (PR), and Resistant (R). Graphics (c) and (d) represent the average of each detected sugar in percentage vs each category. Different scale depicts the most abundant (C) and the minority sugars (D) detected in the phloem sap. The percentages were calculated from the Total Ion current Chromatogram (TIC)

Full size image

The PCA analysis based on sugars and sugar derivatives is shown in Supplementary Fig. S5a. The total variation explained by the first two principal components was 49.7% with PC1 contributing 32.2% and PC2 contributing 17.5%. The loading plot analysis revealed that the sugar composition of susceptible samples was different from the rest of the samples. However, within the susceptible category, there appeared to be higher variability compared to the resistant varieties (Supplementary Fig. S5b). The most abundant sugars in the citrus phloem sap were fructose, sucrose, glucose, and the sugar alcohol myoinositol (Fig. 3c). While glucose exhibited a trend of higher levels in the susceptible samples, the differences compared to the other categories were not statistically significant. By contrast, myoinositol was found at higher levels in the resistant samples (Fig. 3c and Supplementary Fig. S6). The rest of the minority sugar and sugar derivatives clustered to the left of the loading plot together with the resistant and partially resistant samples (Supplementary Fig. S5b). Some of them, such as glycerol, and the detected three unidentified monosaccharides showed an upward trend in the resistant cultivars (Fig. 3d).

Thus, we observed an upward trend in the total sugar levels of the susceptible genotypes. It's worth noting that there is an observable trend in the average percentage increase of sugar acids in PR and R accessions, with a more pronounced effect in the R group.

Low amount of lysine in phloem sap as a possible marker for HLB resistance

Seventeen proteinogenic amino acids and three non-proteinogenic amino acids (GABA, 4-hydroxyproline, and ornithine) were detected in the phloem sap. The percentage of non-proteinogenic amino acids accounted for approximately 25% of the total amino acids in all the samples, regardless of their resistance levels. In terms of proteinogenic amino acids, the quantity of essential amino acids was considerably lesser than non-essential amino acids, with no significant differences among the samples (Supplementary Fig. S7). Neither significant differences in the percentage of the total amino acids nor a trend bias was observed comparing the three categories (Fig. 4a and Supplementary Table S6). Non-essential amino acid distribution was more homogeneous among the susceptible species, with proline being the most abundant amino acid in 78% of the samples (Supplementary Fig. S8a). On the other hand, the distribution of most of the essential amino acid was very similar among the three categories, with the exception of lysine levels, which were significantly higher in susceptible samples (Supplementary Fig. S9). Although the levels of other amino acid did not show statistically significant differences among the three categories, the average percentage showed changes that could be potentially associated with the level of tolerance. The levels of other three amino acids from the aspartate family, threonine, asparagine, and aspartic acid were present at higher levels in the resistant cultivars (Fig. 4b and c). Four amino acids of the glutamate family, including glutamic acid and 4-hydroxyproline were also abundant in resistant varieties, in contrast to proline and ornithine, which had lower representation in this category (Fig. 4c and d and Supplementary Figs. S8a and S8b). Finally, the level of alanine from the pyruvate family was elevated in resistant and partially resistant cultivars.

Fig. 4
figure 4

(a) Bar graph represents the average percentage of total detected amino acids in the three analyzed categories; Susceptible (S), Partially Resistant (PR), and Resistant (R). The percentages were calculated from the Total Ion current Chromatogram (TIC). (No significant difference in one-way ANOVA; α 0.05, P.0.54). Graphics represent the average of each detected essential amino acid over the total essential amino acids (b), non-essential amino acid over the total non-essential amino acids (c) and non-proteinogenic amino acid over the total non-proteinogenic amino acids (d) in percentage within each analyzed categories. The percentage of each component was calculated from the Total Ion current Chromatogram (TIC). The asterisk (*) indicating lysine, means the only amino acid in which we observed significant differences between susceptible (S) and resistant (R) samples (t student analysis for lysine, S vs R (α 0.05, P 0.0035)). Colours in (A), (B) and (C) represent the different amino acid families; Aspartate family (red); Glutamate family (green); Pyruvate family (blue); Aromatic amino acid family (yellow) and Serine family (purple)

Full size image

The PCA analysis that emerged from amino acid data is presented in Supplementary Fig. S10a. The total variation explained by the first two principal components was 59.7%, with PC1 contributing 39.4% and PC2 contributing 20.3%. Resistant and susceptible samples clustered into two well-defined groups. Additionally, partially resistant sample PR2, clustered with the susceptible samples, while PR3 clustered with the resistant ones. The corresponding loading plot showed that susceptible samples were high in proline, lysine, and the non-proteinogenic amino acid ornithine. In contrast, resistant samples were high in 4-hydroxyproline, glutamic acid, threonine, and histidine, although many of the amino acid levels detected in PR2 were elevated (Supplementary Fig. S10b).

Therefore, lysine decrease in phloem sap could be considered as a potential marker of HLB resistance.

Expression of genes of lysine catabolism is affected in resistant citrus samples.

Based on our data, lysine was the only amino acid present at significantly higher levels in the phloem sap of susceptible samples in comparison with the resistant ones. Lysine and its catabolic intermediates have been involved in plant responses to abiotic and biotic stresses (Arruda & Barreto, 2020). Thus, we conducted RT-qPCR analyses to investigate the potential involvement of the lysine catabolic pathway in determining the tolerance levels of the accessions analyzed. We evaluated the expression of five genes including two genes of the saccharopine pathway (lysine-ketoglutarate reductase/saccharopine dehydrogenase (LKR/SDH) and aldehyde dehydrogenase 7B4 (ALDH7B4)), two genes associated with a pathway recently link to plant systemic acquired resistance (SAR) (SAR-deficient 4 (SARD4)) and flavin-dependent monooxygenase 1 (FMO 1) and, one gene implicated in the connection between these two branches of lysine catabolism (pyrroline-5-carboxylate reductase (P5CR). As shown in Fig. 5a, the transcript abundance of the FMO1 gene was significantly higher in two resistant samples, Eremocitrus glauca (R2), and Microcitrus warburgiana hybrid (R3), in comparison to all the susceptible samples. However, in the resistant Microcitrus australis (R1), the level of this transcript was only significantly higher compared with one of the susceptible genotypes, Poncirus trifoliata Rubidoux (S2). On the other hand, the transcript of ALDH7B4 was downregulated in the resistant cultivars. All the observed differences were significant except between Citrus x sinensis ´Pera´ (S1) and Microcitrus australis hybrid (R1) (Fig. 5b). The expression of the rest of the analyzed genes did not exhibit significant differences among the three categories.

Fig. 5
figure 5

Relative Quantification (RQ) plot of the target genes flavin-dependent monooxygenase 1 (FMO1) (a) and aldehyde dehydrogenase 7B4 (ALDH7B4) (b) in the leaf samples from the nine Citrinae genotypes. Expression was normalized to Glyceraldehyde-3-phosphate dehydrogenase C2 (GAPC2) and a SAND family protein (SAND). A non-parametric Kruskal–Wallis one-way ANOVA was performed. Statistically significant differences were considered for p-values less than 0.05. A post hoc test, the Dunn test for multiple comparissons, was conducted to identify which specific groups differ from each other. (α 0,05; * P < 0,05; ** P < 0,01; *** P < 0,001; **** P < 0,0001)

Full size image

Our findings suggest that lysine catabolism is influenced in the resistant plants. Among the five genes we analyzed, FMO1 and ALDH7B4 displayed significant alterations in the accumulation of their transcripts when compared to susceptible cultivars. These changes in gene expression might be indicative of the role of lysine catabolism in the resistance of these citrus accessions to CLas infection.

Discussion

While significant efforts have been dedicated to study the bases of tolerance/resistance to the devastating HLB disease, we are still a long way from a comprehensive understanding of this process. To gain further insights into the underlying mechanisms of the disease, omics technologies constitute an indispensable research tool to understand plant tolerance to this pathogen. In this study, we use metabolomics to try to gain insight into the mechanism underpinning the resistance, by analysing the composition of phloem sap in healthy citrus varieties with different level of susceptibility to HLB (Alves et al., 2021, 2022).

Regarding organic acids, we observed a global upward trend in both the resistant and partially resistant cultivars. This aligns with the finding of Killiny et al. (2017), who conducted a metabolite profile analysis of the “Sugar Belle” mandarin hybrid to investigate its relative tolerance to HLB in comparison to some of its ancestors. Their research revealed that “Sugar Belle” exhibited elevated levels of phosphoric and some organic acids, including malic and threonic acid. Furthermore, previous studies have also reported an increase in the presence of various organic acids in resistant varieties (Killiny, 2017), and CLas-infected leaves from susceptible cultivars (Albrecht et al., 2016). Therefore, our data, in line with these previous studies, suggest that organic acids may have a role in nutrient uptake from the soil and serve as a priming strategy for those cultivars to be more tolerant.

In our analysis, we found that malic and quinic acid were the most abundant organic acids, consistent with prior research (Jones et al., 2012). Notably, only quinic acid showed a trend of higher content in PR and R samples compared to S samples. This metabolite has previously been link to defence responses in citrus leaves and has been detected in fruits of semi-tolerant varieties to CLas (Jones et al., 2012; Killiny, 2017), although its specific role in countering this pathogen remains unknown. Additionally, the average percentage of each organic acid within the three categories revealed trends that may be associated with the level of tolerance. Among them, citric acid and anthranilic acid, found at higher accumulation levels in resistant cultivars in our analysis, have been reported as metabolites involved in stress tolerance and plant protection (Köllner et al., 2010; Zahan et al., 2021).

Higher levels of sugars have been previously associated with HLB susceptibility (Albrecht et al., 2016; Killiny, 2016). Although there was not a consistent pattern in sugar composition among the different categories in our analysis, we did observe an upward trend in total sugar levels in the susceptible genotypes. However, these metabolites are unlikely to be the limiting factor for the HLB vector, since no clear correlation between glucose, fructose, sucrose, and citrus susceptibility to CLas was found (Killiny, 2016, 2017). It´s worth mentioning that effects on carbohydrate metabolism have been previously described in different phloem-related plant-pathosystems, including HLB (Albrecht et al., 2016).

If there is one group of metabolites that we aim to emphasize based on our findings, it is the amino acids. Irrespective of their role in protein biosynthesis, numerous amino acids have been reported to participate in plant response to different stresses (Trovato et al., 2021). In our study, we have observed distinct trends in the levels of certain amino acids that could be coupled to the degree of HLB resistance, with a majority of them belonging to the glutamate and aspartate families. Higher constitutive concentrations of amino acids from the glutamate family have previously been associated with tolerance and resistance to CLas. In our analysis, glutamate and 4-hydroxyproline showed an upward tendency in resistant cultivars. Both of these amino acids have been formerly correlated with defence responses and resistance to plant pathogens (Albrecht et al., 2020; Deepak et al., 2007, 2010; Kim et al., 2021; Qiu et al., 2020). In contrast, two other members of this family, ornithine and proline, displayed a downward trend in the resistant samples. Supporting this, ornithine has been associated with susceptibility in various pathosystems (Dhodary et al., 2022; Jiménez-Bremont et al., 2014). Proline upregulation, on the other hand, has been connected to CLas tolerance, as well as other biotic and abiotic stresses, and has been observed in infected plants compared to controls (Albrecht et al., 2020; Chin et al., 2020). However, in a metabolic profiling of the phloem sap from fourteen varieties with different levels of tolerance to CLas, only one tolerant variety exhibited higher proline levels (Killiny, 2017). The variability in the metabolic composition of the different varieties and the conditions used in the different studies make it challenging to draw consistent conclusions. Nonetheless, it is worth noting that CLas lacks the ability to synthesize proline, phenylalanine, tryptophan, cysteine, tyrosine, and histidine and other essential translation components. These components are crucial for the bacterium's replication and metabolic activity, and it must acquire them from the host (Mendonça et al., 2017; Zuñiga et al., 2020).

The aspartate family pathway encompasses the synthesis of amino acids such as lysine, asparagine, and threonine (Yang & Ludewig, 2014). Our metabolomics analysis revealed higher levels of threonine in the resistant cultivars as well as increased levels of asparagine in both the resistant and partially resistant cultivars. The essential role of threonine as a sensor of different metabolic and environmental signals and translator of these signals into specific functional outputs, including stress tolerance, has been previously reported (Killiny, 2017; Killiny et al., 2017; Muthuramalingam et al., 2018). However, it's worth noting that lower amounts of threonine in resistant Rutaceae genotypes in comparison to susceptible cultivars, have also been described (Killiny, 2016). Pepper asparagine synthase 1, the enzyme required for the production of asparagine from aspartate, was reported as essential for stress response (Hwang et al., 2011).

According to our findings, we must emphasize that the only amino acid displaying significant differences across the different categories was lysine. Lysine is a limiting essential amino acid in plants, and its biosynthesis has been a matter of study aimed at enhancing the nutritional value of crops. As mentioned above, lysine catabolism research has recently focused on tolerance to biotic and abiotic stresses. Lysine can be catabolized through several metabolic pathways (Fig. 6). In some plants, lysine is converted into the alkaloid cadaverine (Arruda & Barreto, 2020; Hartmann & Zeier, 2018). A second ubiquitous catabolic pathway in plants is the saccharopine pathway rendering acetyl-CoA and glutamate. Furthermore, the third pathway, with a central role in plant immunity, leads to the generation of N-hydroxypipecolic acid (NHP), a metabolite recently involved in plant SAR (Hartmann et al., 2018). Contradictory results have been reported describing the lysine accumulation levels in response to pathogens (e.g. Albrecht et al., 2016, 2020; Killiny, 2016; Killiny & Hijaz, 2016; Návarová et al., 2013). According to our results, among the essential amino acids, lysine represented on average 19.26%, 12.66%, and 11.40% in susceptible, partially resistant, and resistant samples, respectively. The significant difference in lysine levels between the resistant and susceptible cultivars, suggests a potential key role in plant defense.

Fig. 6
figure 6

Lysine catabolism including three metabolic routes. Cadaverine pathway (yellow), Sacchaporine pathway (green) and the N-hydroxyl L-pipecolic acid route (red). Lysine decarboxylase (LDC), lysine-ketoglutarate reductase/ saccharopine dehydrogenase (LKR/SDH), aldehyde dehydrogenase 7B4 (ALDH7B4), AGD2-like defense response protein 1 (ALD1), SAR-deficient 4 (SARD4), flavin-dependent monooxygenase 1 (FMO 1), Pirroline-carboxylate reductase (P5CR). α-amino-adipic semialdehyde (AAS), α-amino adipic acid (Aad), Δ1-piperideine-6-carboxylic acid (1,6-DP). (SAR) systemic acquired resistance

Full size image

Regarding the synthesis of NHP (N-hydroxypipecolic acid), several enzymes have been associated with its production. The aminotransferase ALD1 (AGD2-like defense response protein 1), (Hartmann & Zeier, 2018; Holmes et al., 2019), a reductase, SARD4 (Ding et al., 2016); and FMO1, an NHP-synthesizing pipecolate N-hydroxylase (Hartmann & Zeier, 2018; Holmes et al., 2019). Additionally, the saccharopine pathway, which is associated with both abiotic and biotic stress responses, can produce pipecolate through two enzymatic reactions catalyzed by lysine-ketoglutarate reductase/saccharopine dehydrogenase (LKR/SDH) and pyrroline-5-carboxylate reductase (P5CR) (Arruda & Barreto, 2020). These authors explored whether both pathways contribute to SAR activation by comparing the transcriptional response of key genes, including LKR/SDH, ALD1, SARD4, and FMO1 in Arabidopsis thaliana under various biotic and abiotic conditions. Under most biotic stresses, all the analyzed genes were upregulated. Conversely, only the saccharopine pathway was upregulated under abiotic conditions.

The observed differences in lysine levels between susceptible and resistant categories led us to study the potential involvement of both the saccharopine and NHP pathways in HLB tolerance. We monitored transcriptional response of key genes, including LKR/SDH, ALDH7B4, P5CR, SARD 4, and FMO1, in the leaves of citrus healthy plants. Our results revealed significant upregulation of FMO1 in two of the resistant samples compared to the susceptible ones. As previously mentioned, FMO1 is the final enzyme of a pathogen-inducible L-lysine catabolic pathway in Arabidopsis thaliana. Koch et al. (2006) reported that the overexpression of FMO1 under the control of the 35S promoter increases the basal resistance to Pseudomonas syringae. Additionally, Mishina and Zeier (2006) demonstrated the crucial role of this enzyme in the establishing of SAR. Afterward, supporting this, it was reported that FMO1 catalyzes the conversion of pipecolic acid to NHP, which is a critical amino acid with a central role in the establishment of SAR (Hartmann et al., 2018). Interestingly, these authors did not detect this metabolite in unstressed Arabidopsis thaliana; however, NHP strongly accumulated in the infected leaves. This metabolite was also not detected in any healthy phloem sap samples we analysed. Concerning the pipecolic acid, we did not obtain any conclusive result indicating significant differences among categories.

Another lysine-catabolite is the α-amino adipic acid (Fig. 6). Its biosynthesis is dependent on the LKR/SDH and ALDH7B4 saccharopine enzymes. In our analysis, ALDH7B4 exhibited a decreased transcriptional response in the resistant genotypes. It's interesting to note that Návarová et al. (2013) previously suggested that the saccharopine pathway leading to α-amino adipic acid has no critical role in the resistance and SAR establishment after Pseudomonas syringae infection. Upstream of ALDH7B4, LKR/SDH converts lysine to α-aminoadipate semialdehyde and glutamate, which is a precursor for several metabolites related to stress including pipecolic acid (Arruda & Barreto, 2020).

Based on our results, lysine catabolism might be related to varying levels of tolerance to HLB. Although it represents a limited picture particularly due to the number of samples analysed, it allows us to speculate about the potential roles of the upregulation of FMO1 and the downregulation of ALDH7B4 transcriptional responses. This differential regulation could create a more favourable scenario to combat the bacteria, potentially leading to increase resistance in certain cultivars. Somehow, it could be a plant strategy to anticipate defence mechanisms, thereby conferring an advantage over the pathogen. By identifying specific metabolic steps and defining the pathways involved in tolerance and resistance, we can develop powerful tools not only for discovering potential antimicrobial compounds against CLas but also for expediting the development of new citrus cultivars with enhanced resistance to this devastating disease.