
==== Front
Sci Rep
Sci Rep
Scientific Reports
2045-2322
Nature Publishing Group UK London

70499
10.1038/s41598-024-70499-z
Article
Spatial chemistry of citrus reveals molecules bactericidal to Candidatus Liberibacter asiaticus
A. Aksenov Alexander aaksenov@uconn.edu

2811
Blacutt Alex 3
Ginnan Nichole 312
Rolshausen Philippe E. 10
V. Melnik Alexey 128
Lotfi Ali 1
C. Gentry Emily 113
Ramasamy Manikandan 4
Zuniga Cristal 614
Zengler Karsten 567
Mandadi Kranthi K. 49
Dorrestein Pieter C. 27
Roper M. Caroline 3
1 https://ror.org/02der9h97 grid.63054.34 0000 0001 0860 4915 Department of Chemistry, University of Connecticut, Storrs, CT USA
2 grid.266100.3 0000 0001 2107 4242 Collaborative Mass Spectrometry Innovation Center, Skaggs School of Pharmacy & Pharmaceutical Sciences, University of California – San Diego, La Jolla, CA USA
3 grid.266097.c 0000 0001 2222 1582 Department of Microbiology and Plant Pathology, University of California, Riverside, CA USA
4 Department of Plant Pathology and Microbiology, Texas A&M AgriLife Research and Extension Center, Weslaco, TX USA
5 grid.266100.3 0000 0001 2107 4242 Department of Pediatrics, University of California, San Diego, La Jolla, CA USA
6 grid.266100.3 0000 0001 2107 4242 Department of Bioengineering, University of California, San Diego, La Jolla, CA USA
7 grid.266100.3 0000 0001 2107 4242 Center for Microbiome Innovation, University of California, San Diego, La Jolla, CA USA
8 Arome Science Inc., Farmington, CT USA
9 grid.264756.4 0000 0004 4687 2082 Institute for Advancing Health Through Agriculture, Texas A&M AgriLife, College Station, TX USA
10 grid.266097.c 0000 0001 2222 1582 Department of Botany and Plant Sciences, University of California, Riverside, CA USA
11 https://ror.org/02der9h97 grid.63054.34 0000 0001 0860 4915 Present Address: Department of Chemistry, University of Connecticut, Storrs, CT USA
12 https://ror.org/04p491231 grid.29857.31 0000 0001 2097 4281 Present Address: One Health Microbiome Center, Huck Institutes of the Life Sciences, Pennsylvania State University, University Park, PA USA
13 https://ror.org/02smfhw86 grid.438526.e 0000 0001 0694 4940 Present Address: Department of Chemistry, Virginia Tech, Blacksburg, VA USA
14 https://ror.org/0264fdx42 grid.263081.e 0000 0001 0790 1491 Present Address: Department of Biology, San Diego State University, San Diego, CA USA
2 9 2024
2 9 2024
2024
14 203065 4 2024
16 8 2024
© The Author(s) 2024
2024
https://creativecommons.org/licenses/by-nc-nd/4.0/ Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
Huanglongbing (HLB), associated with the psyllid-vectored phloem-limited bacterium, Candidatus Liberibacter asiaticus (CLas), is a disease threat to all citrus production worldwide. Currently, there are no sustainable curative or prophylactic treatments available. In this study, we utilized mass spectrometry (MS)-based metabolomics in combination with 3D molecular mapping to visualize complex chemistries within plant tissues to explore how these chemistries change in vivo in HLB-infected trees. We demonstrate how spatial information from molecular maps of branches and single leaves yields insight into the biology not accessible otherwise. In particular, we found evidence that flavonoid biosynthesis is disrupted in HLB-infected trees, and an increase in the polyamine, feruloylputrescine, is highly correlated with an increase in disease severity. Based on mechanistic details revealed by these molecular maps, followed by metabolic modeling, we formulated and tested the hypothesis that CLas infection either directly or indirectly converts the precursor compound, ferulic acid, to feruloylputrescine to suppress the antimicrobial effects of ferulic acid and biosynthetically downstream flavonoids. Using in vitro bioassays, we demonstrated that ferulic acid and bioflavonoids are indeed highly bactericidal to CLas, with the activity on par with a reference antibiotic, oxytetracycline, recently approved for HLB management. We propose these compounds should be evaluated as therapeutics alternatives to the antibiotics for HLB treatment. Overall, the utilized 3D metabolic mapping approach provides a promising methodological framework to identify pathogen-specific inhibitory compounds in planta for potential prophylactic or therapeutic applications.

Subject terms

Analytical chemistry
Environmental chemistry
Plant sciences
Analytical biochemistry
Mass spectrometry
Metabolomics
http://dx.doi.org/10.13039/100005825 National Institute of Food and Agriculture 2021-70029-36056 issue-copyright-statement© Springer Nature Limited 2024
==== Body
pmcIntroduction

Huanglongbing (HLB) is a highly destructive citrus disease exhibiting complex symptomatology. As one of the most serious plant diseases worldwide, HLB poses a major threat to US citrus production, which is primarily concentrated in Florida, California, Arizona, and Texas1. All commercial citrus varieties are HLB-susceptible to various degrees and HLB has been identified in 51 out of 140 countries that produce citrus2. HLB is caused by the bacterium Candidatus Liberibacter asiaticus (CLas) and is spread by the insect vector Diaphorina citri, commonly known as the Asian Citrus Psyllid (ACP). Early HLB symptoms include leaf mottling and yellow shoots, which leads to root dieback and tree death3. Moreover, the effect on fruit development has major downstream consequences on fresh fruit and juice quality4. Fruit from HLB-infected trees are small, misshapen, with irregular maturation patterns that yield unpalatable flavor profiles. The disease has a long asymptomatic stage, making early detection and management particularly challenging4,5.

The ACP, first reported in Florida in 1998, caused rapid regional HLB outbreaks by 2005, resulting in endemic establishment and billions of dollars in damages in industry5,1. ACP has since spread east to west across the US citrus belt, and HLB is now detected in thousands of Californian urban citrus plants, threatening commercial groves6. This rapid spread of HLB follows a consistent pattern: once the ACP vector establishes itself in a new area, HLB typically appears soon after. Disease management has mainly focused on insect vector control with repeated sprays of synthetic insecticides7, and no sustainable bacterium-targeting HLB interventions yet exist. The uneven distribution of CLas within infected trees further complicates reliable detection using PCR and assessment of chemical and biological perturbations caused by the bacterium8. While HLB-specific microbial and molecular biomarkers9, and HLB-induced proteome changes have been identified10,11, these advances have yet to translate into effective disease management strategies.

CLas is a phloem limited bacterium, that can not be axenically cultured8. The phloem is a high pressure vascular tissue that transports photosynthates (metabolic products) and sugars throughout the plant. These features of the pathosystem make it particularly difficult to study and manage. Currently, oxytetracycline antibiotic trunk injection is the primary approach to reduce and control CLas12–14. However, antibiotic deployment is economically unsustainable for growers and imposes risk to human and environmental health by potentially increasing antimicrobial resistance in bacterial populations. A “precautionary tale”—widespread antibiotic administration has led to the rapid emergence of resistant Candidatus Liberibacter asiaticus strains in China15. Thus, alternative bactericidal agents need to be explored for viable and sustainable approaches to CLas control.

Metabolomics, both nuclear magnetic resonance (NMR) and mass spectrometry (MS)-based, gives valuable insights into molecular distributions, including those underlying pathogenesis, and thus may provide leads on potential therapeutic molecules. These techniques have repeatedly demonstrated their value in providing important insights into pathogenesis and the mechanistic details of underlying biology16,17. Past investigations employing NMR and MS established associations between CLas infection, such as terpenoid/sugar disruption underlying symptomatic flavor decline18. However, uneven HLB distribution hinders generalizing systemic changes19,20. As symptoms manifest locally, pinpointing locally perturbed metabolites is essential to understand the disease and propose targeted therapies.

Recent advances in metabolomics enable exploring spatial chemical distributions21–24. This includes mass spectral molecular networking, which supports annotation propagation within a network of compounds clustered based on their structural similarities25. This method can capture metabolomic transformations and cluster related compounds, partially addressing the challenge of sparse annotation due to limited compound libraries. We hypothesized that detailed infection mapping would expedite discovery of compounds active against CLas in planta, to provide candidates for novel HLB therapeutics. In this work, we visualized the HLB effect on the metabolome of citrus plants using 2D and 3D molecular maps to gain pathological insights, illuminating possible bacterium-mediated host metabolite exploitation. While we focused on citrus, this approach, in principle, should be generally applicable to a wide range of plant-pathogen systems.

Results

Metabolome shifts in field and greenhouse trees with HLB

We carried out an untargeted metabolomics analysis of citrus tree tissues across a spectrum of disease severity collected from seven HLB-infected Florida orchards, where the disease is endemic. Tissues throughout the citrus tree: stems, roots, and leaves were collected and analyzed. Because of the complex metabolome of citrus trees across different tissues, we used molecular networking25,26 to map and explore the detected chemistries. The resulting network grouped molecules that fragmented in a similar fashion, thus capturing their structural similarity and aiding in understanding the chemical distributions26,27. The network of the detected metabolome across grove trees is shown in Fig. 1a,b. The metabolome varied as HLB progressed (Fig. 1c), with molecules associated with disease severity forming distinct clusters in the network. This suggests that metabolic disruption in HLB progresses not through individual molecular changes, but shifts across entire biochemical families. Clusters of compounds—dilinolenins, fatty amides like octadecenamide, glycerophosphocholines, terpenoids, and, prominently, flavonoids—showed changes linked to symptom severity across tissues. To control for confounding factors such as length of infection, local conditions etc., samples were also collected from a controlled greenhouse environment, including infected and uninfected healthy trees (confirmed using qPCR). Metabolome differences related to HLB symptomatology were much more pronounced in greenhouse samples as compared to field samples, even though symptoms were significantly less pronounced. The most discerning diagnostic markers, identified via partial least squares discriminant analysis (PLSDA) and Random Forest, were found to be various flavonoids. This implicates the flavonoid pathway as the main metabolomic “barometer” of infection19,28–33.Fig. 1 Global network analysis of metabolomic shifts in HLB-infected trees. Citrus trees (leaves, stems and roots) were sampled across multiple groves in Florida. (a) Global network of the metabolome detected in samples collected from grove trees in Florida, which included trees exhibiting symptoms across a range of disease severity (1—appears healthy to 5—severe HLB symptoms). (b) A close-up of a network cluster showing distributions of molecular abundances in trees with different disease ratings (higher values correspond to higher symptom severity); node size is related to total compound abundance. The compounds in the cluster were present in higher amounts when the disease was more severe. The selected example shows perturbation in amounts of tryptophan, indicative of altered metabolism, and a metabolite known to be of microbial origin, indole-3-lactic acid (ILA), a tryptophan metabolite that is known to play a role in microbe-host interactions; ILA is associated with increased disease severity. (c) A supervised analysis (partial least squares discriminant analysis, PLSDA) of tissues from trees in orchards across Florida showed metabolome stratification according to disease severity (Q2 0.1775). (d) Unsupervised analysis (principal component analysis, PCA) of tissues from greenhouse-reared trees indicated drastic differences in the metabolome of healthy and infected symptomatic tissues.

3D metabolome mapping

To explore the spatial patterns in distribution of metabolites, molecular maps were generated using the ‘ili tool21. We further elucidated these patterns for HLB-discriminating molecules. This mapping was conducted on tissues from greenhouse-reared trees of the same age and length of infection. 3D metabolome maps were rendered by collecting individual leaves along an infected branch and mapping specific metabolite concentrations, such as the flavonoid scutellarein tetramethyl ether, back to branch images (Fig. S1). This allowed for visualization of metabolite concentration along the same branch where apical leaves exhibited HLB symptoms and basal leaves are asymptomatic (Fig. S1a). Notably, the distribution of the flavonoid scutellarein tetramethyl ether may be related to leaf age rather than symptomology. Therefore, the symptom-related patterns were investigated at single-leaf resolution, where leaves are age-matched. At this scale, symptom-related patterns were noticeable to an expert, with various flavonoids sharing similar spatial propagation pattern; depleted in visibly chlorotic and mottle signatures of infection (Fig. S2). At the same time, oxidized flavonoids tended to exhibit opposite spatial patterns, suggestive of mutual interconversion (Figs. 2, S3).Fig. 2 3D mapping of the distribution of the flavonoids, didymin and hesperidin in infected and healthy citrus branches: (a) Didymin (m/z 595.2026 (rt 208.16)), (b) Hesperidin m/z 611.1973 (rt 131.25)) and infected plant: (c) Didymin and (d) Hesperidin. Didymin was present at higher abundance in younger leaves of healthy plants and was depleted in the infected plant, while hesperidin increased in abundance in the infected plant and didymin was depleted, indicating possible formation of hesperidin due to oxidation of didymin.

Other than flavonoids, we noted feruloylputrescine, a conjugate of ferulic acid (part of the flavonoid biosynthesis pathway) and putrescine as molecules descriptive of HLB symptoms. Feruloylputrescine is preferentially accumulated in apical symptomatic leaves, exactly opposite the non-oxidized flavonoid depletion patterns (Fig. S4a). The ferulic acid abundance decreased concomitantly with putrescine conjugation, indicative of the possible conversion into feruloylputrescine (Fig. S4b). The same trends were evident on the single-leaf scale, with a clear correspondence of spatial distributions to the chlorotic spots phenotype (Fig. 3; note that in greenhouse-grown plants the visual differences may not be stark, particularly in early or mild infection). Another related phenolic acid, hydroxycinnamic acid (p-coumaric acid), showed a similar distribution, enriched in symptomatic areas (Figs. S5, 3C).Fig. 3 2D mapping of polyamine derivatives within single leaves. (A) Unmodified images of leaves with different disease severity; the symptoms are subtle, with chlorotic spots appearing as slightly lighter areas. (B) The leaves shown in (A) in a color scheme that more clearly shows chlorotic spots (chlorotic spots appear as dark shade areas). (C) Mapped distributions of (top to bottom): ferulic acid—H2O (only water loss ion was observed), hydroxycinnamic acid and feruloylputrescine. 2D maps: healthy, infected asymptomatic, infected symptomatic (from left to right). The mapped symptomatic and asymptomatic leaves were sampled from two branches of a tree that was confirmed by qPCR to be infected with CLas (Ct = 19.6). Increases in hydroxycinnamic acid and feruloylputrescine were concomitant with increased disease severity, while ferulic acid exhibited the opposite trend. (D) Plot showing Liberibacter crescens (a culturable surrogate for CLas) growth inhibition assay, when challenged with 35 mg/mL solution by ferulic acid, feruloylputrescine or putrescine as quantified by the diameter of the zone of inhibition. Ferulic acid had a pronounced inhibition effect.

Metabolic modeling of CLas uptake rates of plant metabolites

A possible rationale for the observed spatial patterns in the metabolome (in particular, the low abundance of ferulic acid in symptomatic tissue) may be the metabolism of ferulic acid by the bacterium that would protect itself from antimicrobial activity. A supporting argument for this would be the observation that phenolic acids, like ferulic acid, reduce the growth rate of CLas, while in conjugation with putrescine, this effect is reduced due to the synthesis of feruloylputrescine. Therefore, we tested this hypothesis by conducting model simulations using the previously established genome-scale metabolic network of CLas15. We simulated uptake rates of plant metabolites by CLas for compounds that are part of arginine and proline metabolism, as well as those present in the biosynthesis of phenylpropanoids. The model predicted that uptake rates of ferulic acid between 1 × 10–10 and 1 × 10–8 mmol/gDW/h would be sufficient to reduce the growth rate of CLas up to 90%. These predictions were experimentally confirmed and putrescine uptake rates in this range did not predict a reduction of CLas growth. Interestingly, when these ferulic acid uptake rates (1 × 10–10 and 1 × 10–8 mmol/gDW/h) were combined with putrescine uptake rates between 1 × 10–2 and 10 × 10–1 mmol/gDW/h, the model predicted the highest CLas proliferation (Fig. 4).Fig. 4 Genome-scale metabolic network simulations about the effect of feruloylputrescine biosynthesis on the growth of CLas. Contour plot shows the predicted growth rates of CLas, while varying putrescine and ferulate uptake rates. Individual scatter plots show the response of CLas growth to changes in the individual uptake rates of metabolites of interest (mmol/gDW/h). 2D maps of leaves show from left to right: healthy, infected symptomatic, infected asymptomatic samples of citrus plants. 2D maps of plants show from left to right: healthy and infected citrus plants.

In vitro verification of model predictions with disc assay

In order to further confirm in silico predictions, we assayed putrescine, ferulic acid, and feruloylputrescine for potential antibacterial activity using a previously developed disc diffusion assay based on Liberibacter crescens, a culturable surrogate of the unculturable CLas34. Metabolic modeling of the pathogen described in the “Metabolic modeling of CLas uptake rates of plant metabolites” section above predicted putrescine and feruloylputrescine to be mildly beneficial and ferulic acid to be inhibitory to CLas growth. The model predictions were consistent with the metabolites’ spatial distributions in greenhouse citrus tree leaves. Disc assay results showed that putrescine and ferulic acid were indeed moderately and highly inhibitory to L. crescens, respectively. The conjugated metabolite, feruloylputrescine, was not inhibitory to L. crescens growth as predicted by our model (Fig. 3D). These in vitro results, in conjunction with the observed interconversion between ferulic acid, putrescine, and feruloylputrescine in HLB symptomatic trees, indicate that CLas likely directly, or through manipulation of host metabolic pathways, conjugate ferulic acid to putrescine producing the non-toxic feruloylputrescine, which increasing its survivability in the host.

Ex vivo verification of model predictions with CLas-citrus hairy root assay

To further validate the toxicity of ferulic acid and flavonoids, whose biosynthesis is disrupted due to conjugation, against the target pathogen CLas, we conducted a CLas-citrus hairy root cultures assay35. As a possible low-cost, accessible therapeutic intervention, we tested over-the-counter citrus peel extracts as a widely available source of native citrus bioflavonoids. CLas-containing hairy roots were treated with bioflavonoids and ferulic acid for 72 h (Fig. 5a). Untreated and ethanol (0.2% v/v) samples were used as negative controls. A reference antibiotic, oxytetracycline (OXY), reported as an inhibitor of Clas14,36,37, was used as a positive control. After treatment, all tissue samples were exposed to PMAxx dye (propidium monoazide, Biotium, Fremont, CA) that allows measurement of only live CLas bacterial DNA, following DNA extraction and molecular diagnostics. The relative titers of CLas were estimated using qPCR. We found that bioflavonoids significantly inhibited CLas (p ≤ 0.05) in a dose-dependent manner (125, 250, and 500 ppm), whereas ferulic acid showed inhibition at 125 and 250 ppm (p ≤ 0.05), on par with oxytetracycline when compared to untreated controls (Fig. 5b).Fig. 5 Bioflavonoids and ferulic acid inhibit the growth of Candidatus Liberibacter asiaticus (CLas) in CLas-citrus hairy roots. (a) The CLas-citrus hairy roots were treated for 72 h with 125, 250, 500, and 1000 ppm of bioflavonoids and ferulic acid. Oxytetracycline hydrochloride (OXY)-treated hairy roots (250 and 500 ppm) were used as a positive control, and untreated (UT), ethanol (0.2%) used to dissolve the compounds was used as a negative control. Relative titers of CLas (b) were estimated after 72 h of treatment, followed by qPCR analysis. Error bars represent the standard error of the mean (n = 6), and p values were calculated by Student t-test relative to untreated samples.

Discussion

We explored HLB disease metabolome dynamics using spatial mapping, predictive modeling, in vitro, and ex vivo assays to identify endogenous citrus compounds with plant therapeutic potential. Metabolites altered by HLB symptoms were consistent among field and greenhouse trees, with flavonoids comprising the most discriminating “biomarkers”, supporting previous observation19,37–39. The discriminant flavonoids structures did not comport to a notable trend, and molecules differing by various backbone substitutions, e.g. addition of a glycan (which increases the compound’s solubility), were found to be predominantly declining in infected symptomatic tissues. This implies upstream flavonoid pathway suppression that potentially aids pathogen establishment, fitting the known antimicrobial and defense roles of these compounds. However, this information alone is insufficient to understand disease etiology9,37,38,40. For example, certain flavonoids were instead found to increase with HLB symptom development.

Overlaying 3D distribution onto molecular networking provided further insights into functional interconnections in molecular distributions, not accessible otherwise. For example, the association with HLB symptoms is notably different for oxidized versions of some flavonoids. As shown in Fig. 2, hesperidin accumulation mirrored declines in the flavonoid didymin, possibly reflecting oxidative conversion of the latter into the former during infection. Such localized oxidative shifts were further evident across flavonoid families (Fig. S3), aligning with oxidative stress reports previously described41,42. In another example, 3D maps revealed that leaf age was the predominant driver of flavonoid abundance, rather than disease severity. Younger leaves possess elevated flavonoids, necessitating age-matching to correctly determine infection biomarkers.

Exploring the HLB-discriminating ability and corresponding 3D maps for various metabolites allowed us to identify other molecules of interest in addition to flavonoids. In particular, feruloylputrescine was noted to be highly associated with symptom severity. Distribution patterns implicated the conversion of ferulic acid into feruloylputrescine, mirrored by ferulate disappearance and putrescine buildup (Fig. 3). This shift was further echoed by hydroxycinnamate rises, likely reflecting the oxidation of cinnamic acid precursors. Taken together, spatial patterns indicate metabolic channeling away from phenylpropanoid biosynthesis toward polyamine conjugation due to infection.

Exploring upstream pathways points to a branch point: while general phenylalanine supply appeared unaffected, p-coumaroyl-CoA intermediates depleting from the flavonoid pathway were instead visible as hydroxycinnamate en route to feruloylputrescine. This suggests that CLas has evolved to counteract both ferulic acid and flavonoids formation, ostensibly as a self-defense against their toxicity.

To confirm that these molecules are indeed toxic to CLas, we conducted the disc and hairy roots assays that confirmed their bactericidal effects toward CLas. Although the disc assay exploited a surrogate organism, Liberibacter crescens, the hairy roots assay is a way to directly assess the toxicity, both genetic- and chemical-based to CLas itself35,43,44. In this approach, CLas is maintained in citrus hairy root tissues, and the antimicrobials are directly infiltrated into the roots by vacuum, thus overcoming the delivery bottleneck that exists with conventional whole tree assays13,36,45. Additionally, because citrus hairy roots have intact vasculature, they are much closer to the in planta citrus environment where CLas resides, thus providing reliability over testing against surrogate bacteria or heterologous systems46–49. The assay has demonstrated a clear bactericidal effect of the flavonoids on par with the anti-CLas activity of oxytetracycline. In the case of ferulic acid, the lower dosages (125 to 250 ppm) are highly effective, while increasing the dosage to 500 or 1000 ppm did not seem to be effective, possibly due to the toxicity of the compounds to the hairy roots themselves50. Both flavonoids and ferulic acid bacterial inhibitory abilities have been well characterized in other systems, but the mode of action is not clear51,52. These phytochemicals may disrupt bacterial cell-to-cell communication53, modify biofilm production and inhibit swimming motility54, impact bacterial membranes55, or even create pores or rupture cell members56. Nevertheless, these experiments confirmed the robust anti-CLas activity of the flavonoids55,56. It is not known how these compounds may impact other members of the citrus microbiome. Indeed, both bacterial and fungi communities drastically shift as HLB symptoms advance9, and citrus-associated fungi can produce Liberibacter spp. inhibitor compounds34. Future studies should consider the interactions of these compounds with the microbiome, and vice versa, to increase efficacy of disease management applications.

With the efficacy against CLas confirmed, the ferulic acid and bioflavonoids present an opportunity for field application as novel Huanglongbing therapies. These compounds are plant-derived, non-toxic, biodegradable, inexpensive to manufacture at scale, and thus would be a very appealing class of therapeutics. They may help to bypass negative environmental and economic externalities compared to current methods, such as antibiotic use and synthetic chemical pesticide applications for ACP control12–15. While optimal therapy delivery systems for citrus plants remain an active topic of research and development, multiple practical options such as trunk injection already exist9,13,14,37,57–60.

The impact of HLB has been devastating. Deployed alongside psyllid population suppression, the proposed “botanical” solutions harnessing natural plant defenses, when deployed at scale, may offer a viable path towards stable HLB mitigation and control. Finally, although we demonstrate the described approach for the citrus and CLas, the same approach of mapping the metabolic distributions coupled with toxicity assays may present opportunities to generalize the development of a therapy agent discovery pipeline that could be used for a wide range of pathosystems.

Materials and methods

Tissue collection and processing

Branch and single leaf assay

For the branch and single leaf assays, Valencia sweet orange (Citrus sinensis) grafted onto Swingle rootstock was propagated in a greenhouse and inoculated with CLas using infectious ACP. Five ACP adults enclosed in a nylon mesh drawstring bag were applied and confined to newly emerged leaves for 14 days. Trees were then treated with insecticides, imidacloprid, and carbaryl, to eliminate ACP, and the trees were kept free of ACP until the time of sampling. For branch analyses, all leaves for a given time point were detached via razor blades, flash-frozen, and held in liquid nitrogen during sampling. Petioles were removed and stored at − 80 °C for CLas quantitation via qPCR. For metabolome analysis, 1/8 in (3.175 mm) diameter punches were dispensed into the wells of 96 well plates containing 500 µL 50% ethanol. Tissues were lysed by repeat freeze–thaw cycles between − 80 °C and room temperature. The resulting extracts were filtered through 0.22 um filter plates via centrifugation before analysis via LC–MS. Leaf punches were collected in the same manner for the single-leaf assay.

Field collected tissues

Stems, roots, and leaves from 50 trees (n = 150) were collected from 5 different citrus Florida citrus groves in 20169. In 2017, stems, roots, and leaves were collected from 80 trees located in 7 different orchards (n = 240). The stem collection included both xylem and phloem.

Each tree was divided into 4 quadrants (North, South, East, and West), and stems with attached leaves were collected from each of the quadrants and pooled. Topsoil from two sides of the tree and approximately 1.5 feet away from the base of the trunk near the irrigation line was removed, and the feeder roots near the irrigation line were sampled, shaken to remove soil, and sealed in a plastic bag. Gloves were changed, and clippers and shovels were sterilized with 30% household bleach between each tree that was sampled. All samples were immediately placed on ice for transit to the laboratory, where they were placed at 4˚C and processed within 24 h.

MS data acquisition

The tissue extracts were prepared in 100% ethanol, spiked with 1 µM sulfadimethoxine internal standard, and analyzed with UltiMate 3000 UPLC system (Thermo Scientific) using a Kinetex™ 1.7 µm C18 reversed-phase UHPLC column (50 X 2.1 mm) and Maxis Q-TOF mass spectrometer (Bruker Daltonics) equipped with ESI source. The column was equilibrated with 2% solvent B (98% acetonitrile, 0.1% formic acid in LC–MS grade water with solvent A as 0.1% formic acid in water), followed by a linear gradient from 2 B to 10% B in 0.2 min and then to 100% B at 12 min, held at 100% B for 2 min. Following each run, the column was equilibrated at 2% B for 1 min at a flow rate of 0.5 mL/min. MS spectra were acquired in positive ion mode in the range of 80–2000 m/z. A mixture of 10 µg/mL of each sulfamethazine, sulfamethizole, sulfachloropyridazine sulfadimethoxine, amitriptyline, and coumarin-314 was run at the beginning and the end of each batch (one 96-well plate). An external calibration with ESI-L Low Concentration Tuning Mix (Agilent Technologies) was performed prior to data collection, and internal calibrant Hexakis(1H,1H,3H-tertrafluoropropoxy)phosphazene was used throughout the runs. The capillary voltage of 4500 V, nebulizer gas pressure (nitrogen) of 1.4 bar, ion source temperature of 180 °C, and dry gas flow of 4 L/min, were used. For acquiring MS/MS fragmentation, the 7 most intense ions per MS1 were selected. A stepping function was used to fragment ions at 50%, 100%, 150%, and 200% of the CID, with a timing of 25% for each step. Similarly, basic stepping of collision RF of 250 to 1500 Vpp with a timing of 25% for each step and transfer time stepping of 50, 75, 100, and 150 µs with a timing of 25% for each step was employed. MS/MS active exclusion parameter was set to 5 and released after 30 s. The mass of internal calibrant was excluded from the MS/MS list using a mass range of m/z 921.5–924.5. The data were deposited in the MassIVE online repository and are available under the IDs: MSV000082967 (2D leaf mapping); MSV000082962 (3D branch mapping); MSV000082963 and MSV000085416 (field study).

MS data analysis

The collected HPLC–MS raw data files were first converted from Bruker’s d to mzXML format and then processed with the open-source MZmine2 software [https://www.ncbi.nlm.nih.gov/pubmed/20650010?dopt=Abstract]. crop filtering with a retention time (RT) range of 0 to 14 min chromatograms. Mass detection was performed with a signal threshold of 1E3 and a 0.04-s minimum peak width. The mass tolerance was set to 20 ppm, and the maximum allowed retention time deviation was set to 5 s. For chromatographic deconvolution, the local minimum search algorithm with a 30% chromatographic threshold, minimum RT range of 0.6 s, minimum relative height of 1%, minimum absolute height of 5E2, the minimum ratio of peak top/edge 2, and peak duration range of 0.04—min was used. After isotope peak removal, the peak lists of all samples were aligned within the corresponding retention time and mass tolerances. Gap filling was performed on the aligned peak list using the peak finder module with 1% intensity, 10-ppm m/z tolerance, and 0.05-min RT tolerance, respectively. After the creation and export of a feature matrix containing the feature retention times, exact mass, and peak areas of the corresponding extracted ion chromatograms, the sample metadata was added to the feature matrix metadata of the samples.

All of the peaks that were present in any of the blanks with a signal-to-noise ratio (S/N) below 3:1 were removed from the final feature table.

Data pretreatment and statistical analysis

The data pretreatment and following statistical analysis were carried out with the MetaboAnalyst platform61. The feature tables generated with MZmine were filtered to remove features with near-constant, very small values and values with low repeatability using the interquartile range (IQR) estimate. A detailed description of the methodology is given in Ref.62. The samples were normalized using quantile normalization. The data were further scaled by mean centering and divided by standard deviation for each feature.

Principal component analysis (PCA) and partial least-squares discriminant analysis (PLS-DA)63 were used to explore and visualize variance within data and differences among experimental categories. Random forests (RF)64 supervised analysis was used to further verify the validity of determined discriminating features.

Molecular networking

The molecular network was created using the online workflow at GNPS platform (gnps.ucsd.edu)25,26. The data were clustered with MS-Cluster with a parent mass tolerance of 0.1 Da and an MS/MS fragment ion tolerance of 0.1 Da to create consensus spectra. The consensus spectra that contained less than 3 spectra were discarded. A network was then created where edges were filtered to have a cosine score above 0.65 and more than 4 matched peaks. The edges between two nodes were kept in the network if and only if each of the nodes appeared in each other's respective top 10 most similar nodes. The spectra in the network were then searched against GNPS's spectral libraries. All matches kept between network spectra and library spectra were required to have a score above 0.7 and at least 5 matched peaks. The molecular networks and the parameters used are available at the links below: 2D leaf mapping: https://gnps.ucsd.edu/ProteoSAFe/status.jsp?task=105598d2d782412ca0c988bfe933c032, 3D branch mapping: https://gnps.ucsd.edu/ProteoSAFe/status.jsp?task=e2f1a1e367a7450fb940fe7cda04dc19, Field study: https://gnps.ucsd.edu/ProteoSAFe/status.jsp?task=2b0b1e554bc14d1fba321257b0ffc827, Field study, feature-based molecular network65: https://gnps.ucsd.edu/ProteoSAFe/status.jsp?task=4ac8aff08af5433292143047c8cc5e90.

2D/3D visualization

The procedure for the creation and visualization of 3D models is described in detail previously21. Briefly, the 2D images and 3D model of the sampled plants were created and the coordinates for sampled spots were selected according to described protocol21. The abundances of detected metabolites were normalized and autoscaled61, and the coordinates for spots corresponding to each sample were inserted into the feature tables along with the spot size. The 2D or 3D models were drag-and-dropped into the ‘ili website in the browser (https://ili.embl.de/), followed by the feature table with coordinates. All figures were generated with the “Jet” color map. Spot size, opacity, and border opacity were adjusted for optimal visualization.

Synthesis of feruloyl putrescine hydrochloride

As the feruloyl putrescine was not available commercially, we have synthesized this compound for further compound identification validation and disc assay testing. N-Boc putrescine (1 equiv, 4.25 mmol, 0.91 mL) and ferulic acid (1.06 equiv, 4.5 mmol, 880 mg) were combined in anhydrous CH2Cl2 (32.5 mL) with stirring and the solution was cooled to 0 °C. Then, a solution of N,N′-dicyclohexylcarbodiimide (DCC, 1.7 equiv, 7.2 mmol, 1.49 g) in anhydrous CH2Cl2 was added dropwise. The reaction mixture was allowed to warm to room temperature and stirred at rt for 2 days. Then, the mixture was filtered to remove precipitated dicyclohexylurea and the filtrate was concentrated in vacuo. The crude residue was purified over silica gel using 2–4% MeOH in CH2Cl2 to afford N-Boc feruloyl putrescine as a light yellow solid in 78% yield (1.29 g). For Boc deprotection, trifluoroacetic acid (TFA, 15 mL) was added to a stirred solution of N-Boc feruloyl putrescine in CH2Cl2 (75 mL) under an inert atmosphere of Ar. The reaction was allowed to stir at room temperature for 40 min, then the solvent was removed in vacuo. The residue was dissolved in methanol (15 mL) with HCl (25 mL) and evaporated in vacuo to yield feruloyl putrescine hydrochloride as a light yellow solid (75% overall yield). 1H and 13C NMR data were consistent with those previously reported66. HRMS (ESI) exact mass calculated for [M + H]+ (C14H21N2O3) is m/z 265.1547.

Simulations

Simulations were performed using the CLas Ishi-1 M15 metabolic model15. Standard biomass constraints were maintained to predict the overall CLas growth rate. All model simulations were performed using the Gurobi Optimizer v.5.6.3 solver (Gurobi Optimization) in the COBRA toolbox67 for MATLAB (MathWorks). We simulated the maximal growth rate of CLas using flux-balance analysis. The main metabolic compounds affecting CLas growth were identified using the metabolome data, that was anthranilate, ferulate, glutamate, N-acetyl-L-ornithine, ornithine, putrescine, tyrosine, and vanillin and sensitivity analysis, looking for metabolite-specific growth responses was performed by varying uptake rates among 1 × 10–12, 1 × 10–10, 1 × 10–8, 1 × 10–6, 1 × 10–4, 1 × 10–2, 1 × 10–1, 1 × 100, 1 × 101, 1 × 102, 1 × 103. Additionally, we performed a sensitivity analysis, which deployed a phenotypic phase plane that facilitates the observation of effects on the CLas growth by varying a particular constraint, in this case putrescine and ferulic acid. Predicted growth rates were compared with experimental results.

Microbial culture assays

Citrus metabolites were assayed using a previously developed disc-diffusion assay for L. crescens34. Briefly, L. crescens liquid cultures were incorporated to a soft agar (0.8%) overlay and applied to a 20 mL solid agar (1.5%). L. crescens strain BT-1] was maintained and grown exclusively on the previously described bBM7 + 1.0 methyl-β-cyclodextrin. To these overlaid plates were applied autoclave-sterilized 6 mm paper discs (Whatman, NJ, USA), previously loaded with 35 μL of a given metabolite solution and dried in a sterile biosafety cabinet. Once discs were applied, plates were sealed and stored upside down in a 28 ℃ incubator for 6 days to allow a clear zone of inhibition development for measurement. 35 mg/mL solutions of putrescine (Fisher Scientific, Waltham, MA, USA), and feruloylputrescine (synthesis described above) were prepared using sterile water.

The efficacy of bioflavonoids and ferulic acid in vitro CLas-hairy root assay

The in vitro CLas-hairy roots assay was performed according to the previously described protocol35. CLas-citrus hairy roots were generated using CLas-infected citrus plant tissues, and the diagnosis of CLas was confirmed by quantitative PCR (qPCR). CLas-hairy roots were surface sterilized, and ~ 100 mg was transferred into multi-well plates containing Gamborg’s B-5 medium with 1% sucrose. Different concentrations of bioflavonoids (Horbaach, https://horbaach.com/products/citrus-bioflavonoids-complex-1500mg-300-vegetarian-caplets) and ferulic acid (Fisher Scientific, Pittsburgh, PA, Catalog No. ICN10168505): 125, 250, 500, and 1000 ppm/mL, were added, vacuum infiltrated and incubated on a rotator shaker at 50 rpm in the dark at 25 °C for 72 h. The experiments were carried out with six biological replicates, positive control of oxytetracycline hydrochloride (Sigma Aldrich, Burlington, MA), untreated CLas hairy roots, and an equal concentration of ethanol solvent used to dissolve the bioflavonoids and ferulic acid as negative controls. After the treatments, tissue samples were treated with PMAxx dye (propidium monoazide, Biotium, Fremont, CA) to inactivate dead CLas bacterial DNA. Further, total DNA was extracted, and viable bacterial titer was estimated by qPCR analysis using primers specific to the CLas gene encoding the Ribonucleotide reductase β-subunit (nrdB, RNR-F/RNR-R)68 and the relative CLas titers were estimated and plotted relative to untreated using the 2 − ΔΔCt method. After normalization of target Ct with an endogenous reference gene (Ct') glyceraldehyde3-phosphate dehydrogenase 2 (GAPC2)69 to correct for DNA template concentration differences among the samples, it was plotted relative to untreated controls.

Primers used in this study

RNR-F	CATGCTCCATGAAGCTACCC	68	
RNR-R	GGAGCATTTAACCCCACGAA	
GAPC2-F	GAGGAGATCCCATGGGCAAA	69	
GAPC2-R	AAGAGGAGCTAGGCAGTTGG	

Supplementary Information

Supplementary Figures.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-024-70499-z.

Acknowledgements

This work was partly supported by USDA NIFA grant no. 2017-70016-26053; USDA NIFA grants 2021-70029-36056, and Texas A&M AgriLife Insect-vector disease seed grants to K.M. AAA, AL were supported by the USDA NIFA GRANT 13665683. The Authors are grateful to Dr. Greg McCollum.

Author contributions

CR, PR, PCD created the idea for the work. NG, AB, GM, PR collected plant tissues NG, AB extracted plant tissues for acquisition of mass spectrometry data AA acquired mass spectrometry data AA, AM analyzed mass spectrometry data GM conducted plants growth, infection by CLas, PCR diagnostics AM created 3D models AB, AM picked coordinates on plant models AA performed statistical analysis EG performed organic synthesis CZ, KZ performed and analyzed model simulations AB performed disk assays MR, KM conducted hairy roots assay AA wrote the manuscript CR, AB, PR, NG, AL edited the manuscript.

Data availability

The data were deposited in the MassIVE online repository and are available below links: https://massive.ucsd.edu/ProteoSAFe/dataset.jsp?task=bc1261c22e6c49d1b4f6490c7414845b, https://massive.ucsd.edu/ProteoSAFe/dataset.jsp?task=2008873282ac4beda2a008ada2917fa4, https://massive.ucsd.edu/ProteoSAFe/dataset.jsp?task=d4416036d13141b9b3008c7712395534, https://massive.ucsd.edu/ProteoSAFe/dataset.jsp?task=7db2bd0c5a0941f28286cabdc407a17f, https://massive.ucsd.edu/ProteoSAFe/dataset.jsp?task=25598a865760458fbc45a3f579c8479c.

Competing interests

AAA and AVM are founders of Arome Science, Inc. PCD is an advisor of and holds equity in Cybele, consulted for MSD Animal Health in 2023 and is a cofounder of, holds equity in and is scientific advisor for Ometa Labs, Arome and Enveda with prior approval by the University of California San Diego. The remaining authors declare no competing interests.

Publisher's note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
==== Refs
References

1. Liu Y Heying E Tanumihardjo SA History, global distribution, and nutritional importance of citrus fruits Compr. Rev. Food Sci. Food Saf. 2012 11 530 545
Liu, Y., Heying, E. & Tanumihardjo, S. A. History, global distribution, and nutritional importance of citrus fruits. Compr. Rev. Food Sci. Food Saf. 11, 530–545 (2012).
2. Alvarez S Rohrig E Solís D Thomas MH Citrus greening disease (Huanglongbing) in Florida: Economic impact, management and the potential for biological control Agric. Res. 2016 5 109 118
Alvarez, S., Rohrig, E., Solís, D. & Thomas, M. H. Citrus greening disease (Huanglongbing) in Florida: Economic impact, management and the potential for biological control. Agric. Res. 5, 109–118 (2016).
3. Bové JM Huanglongbing: A destructive, newly-emerging, century-old disease of citrus J. Plant Pathol. 2006 88 7 37
Bové, J. M. Huanglongbing: A destructive, newly-emerging, century-old disease of citrus. J. Plant Pathol. 88, 7–37 (2006).
4. Lee JA Asymptomatic spread of Huanglongbing and implications for disease control Proc. Natl. Acad. Sci. U. S. A. 2015 112 7605 7610 26034273
Lee, J. A. et al. Asymptomatic spread of Huanglongbing and implications for disease control. Proc. Natl. Acad. Sci. U. S. A. 112, 7605–7610 (2015).26034273
5. Manjunath KL Halbert SE Ramadugu C Webb S Lee RF Detection of ‘Candidatus Liberibacter asiaticus’ in Diaphorina citri and its importance in the management of citrus Huanglongbing in Florida Phytopathology® 2008 98 387 396 18944186
Manjunath, K. L., Halbert, S. E., Ramadugu, C., Webb, S. & Lee, R. F. Detection of ‘Candidatus Liberibacter asiaticus’ in Diaphorina citri and its importance in the management of citrus Huanglongbing in Florida. Phytopathology® 98, 387–396 (2008).18944186
6. Gottwald TR Current epidemiological understanding of citrus Huanglongbing Annu. Rev. Phytopathol. 2010 48 119 139 20415578
Gottwald, T. R. Current epidemiological understanding of citrus Huanglongbing. Annu. Rev. Phytopathol. 48, 119–139 (2010).20415578
7. Boina DR Bloomquist JR Chemical control of the Asian citrus psyllid and of Huanglongbing disease in citrus Pest Manag. Sci. 2015 71 808 823 25491482
Boina, D. R. & Bloomquist, J. R. Chemical control of the Asian citrus psyllid and of Huanglongbing disease in citrus. Pest Manag. Sci. 71, 808–823 (2015).25491482
8. Ghosh D Huanglongbing pandemic: Current challenges and emerging management strategies Plants 2022 12 160 36616289
Ghosh, D. et al. Huanglongbing pandemic: Current challenges and emerging management strategies. Plants 12, 160 (2022).36616289
9. Ginnan NA Disease-induced microbial shifts in citrus indicate microbiome-derived responses to Huanglongbing across the disease severity spectrum Phytobiomes J. 2020 4 375 387
Ginnan, N. A. et al. Disease-induced microbial shifts in citrus indicate microbiome-derived responses to Huanglongbing across the disease severity spectrum. Phytobiomes J. 4, 375–387 (2020).
10. Zhong Y Comparative transcriptome and iTRAQ proteome analyses of citrus root responses to Candidatus Liberibacter asiaticus infection PLoS One 2015 10 e0126973 26046530
Zhong, Y. et al. Comparative transcriptome and iTRAQ proteome analyses of citrus root responses to Candidatus Liberibacter asiaticus infection. PLoS One 10, e0126973 (2015).26046530
11. Li B Zhang Y Qiu D Francis F Wang S Comparative proteomic analysis of sweet orange petiole provides insights into the development of Huanglongbing symptoms Front. Plant Sci. 2021 12 656997 33953735
Li, B., Zhang, Y., Qiu, D., Francis, F. & Wang, S. Comparative proteomic analysis of sweet orange petiole provides insights into the development of Huanglongbing symptoms. Front. Plant Sci. 12, 656997 (2021).33953735
12. Yang C Metagenomic analysis reveals the mechanism for the observed increase in antibacterial activity of penicillin against uncultured bacteria Liberibacter asiaticus relative to oxytetracycline in planta Antibiotics (Basel) 2020 9 874 33291469
Yang, C. et al. Metagenomic analysis reveals the mechanism for the observed increase in antibacterial activity of penicillin against uncultured bacteria Liberibacter asiaticus relative to oxytetracycline in planta. Antibiotics (Basel) 9, 874 (2020).33291469
13. Zhang M Effective antibiotics against ‘Candidatus Liberibacter asiaticus’ in HLB-affected citrus plants identified via the graft-based evaluation PLoS One 2014 9 e111032 25372135
Zhang, M. et al. Effective antibiotics against ‘Candidatus Liberibacter asiaticus’ in HLB-affected citrus plants identified via the graft-based evaluation. PLoS One 9, e111032 (2014).25372135
14. Hu J Wang N Evaluation of the spatiotemporal dynamics of oxytetracycline and its control effect against citrus Huanglongbing via trunk injection Phytopathology 2016 106 1495 1503 27482624
Hu, J. & Wang, N. Evaluation of the spatiotemporal dynamics of oxytetracycline and its control effect against citrus Huanglongbing via trunk injection. Phytopathology 106, 1495–1503 (2016).27482624
15. Zuñiga C Linking metabolic phenotypes to pathogenic traits among ‘Candidatus Liberibacter asiaticus’ and its hosts NPJ Syst. Biol. Appl. 2020 6 24 32753656
Zuñiga, C. et al. Linking metabolic phenotypes to pathogenic traits among ‘Candidatus Liberibacter asiaticus’ and its hosts. NPJ Syst. Biol. Appl. 6, 24 (2020).32753656
16. Schauer N Fernie AR Plant metabolomics: Towards biological function and mechanism Trends Plant Sci. 2006 11 508 516 16949327
Schauer, N. & Fernie, A. R. Plant metabolomics: Towards biological function and mechanism. Trends Plant Sci. 11, 508–516 (2006).16949327
17. Macel M Van Dam NM Keurentjes JJB Metabolomics: The chemistry between ecology and genetics Mol. Ecol. Resour. 2010 10 583 593 21565063
Macel, M., Van Dam, N. M. & Keurentjes, J. J. B. Metabolomics: The chemistry between ecology and genetics. Mol. Ecol. Resour. 10, 583–593 (2010).21565063
18. Yao L Proteomic and metabolomic analyses provide insight into the off-flavour of fruits from citrus trees infected with ‘Candidatus Liberibacter asiaticus’ Hortic. Res. 2019 6 31 30792870
Yao, L. et al. Proteomic and metabolomic analyses provide insight into the off-flavour of fruits from citrus trees infected with ‘Candidatus Liberibacter asiaticus’. Hortic. Res. 6, 31 (2019).30792870
19. Hijaz FM An HPLC-MS characterization of the changes in sweet orange leaf metabolite profile following infection by the bacterial pathogen Candidatus Liberibacter asiaticus PLoS One 2013 8 e79485 24223954
Hijaz, F. M. et al. An HPLC-MS characterization of the changes in sweet orange leaf metabolite profile following infection by the bacterial pathogen Candidatus Liberibacter asiaticus. PLoS One 8, e79485 (2013).24223954
20. Aksenov AA Detection of Huanglongbing disease using differential mobility spectrometry Anal. Chem. 2014 86 2481 2488 24484549
Aksenov, A. A. et al. Detection of Huanglongbing disease using differential mobility spectrometry. Anal. Chem. 86, 2481–2488 (2014).24484549
21. Protsyuk I 3D molecular cartography using LC-MS facilitated by Optimus and ’ili software Nat. Protoc. 2018 13 134 154 29266099
Protsyuk, I. et al. 3D molecular cartography using LC-MS facilitated by Optimus and ’ili software. Nat. Protoc. 13, 134–154 (2018).29266099
22. Kapono CA Creating a 3D microbial and chemical snapshot of a human habitat Sci. Rep. 2018 8 3669 29487294
Kapono, C. A. et al. Creating a 3D microbial and chemical snapshot of a human habitat. Sci. Rep. 8, 3669 (2018).29487294
23. Floros DJ Mass spectrometry based molecular 3D-cartography of plant metabolites Front. Plant Sci. 2017 8 429 28405197
Floros, D. J. et al. Mass spectrometry based molecular 3D-cartography of plant metabolites. Front. Plant Sci. 8, 429 (2017).28405197
24. Bouslimani A Molecular cartography of the human skin surface in 3D Proc. Natl. Acad. Sci. U. S. A. 2015 112 E2120 E2129 25825778
Bouslimani, A. et al. Molecular cartography of the human skin surface in 3D. Proc. Natl. Acad. Sci. U. S. A. 112, E2120–E2129 (2015).25825778
25. Wang M Sharing and community curation of mass spectrometry data with global natural products social molecular networking Nat. Biotechnol. 2016 34 828 837 27504778
Wang, M. et al. Sharing and community curation of mass spectrometry data with global natural products social molecular networking. Nat. Biotechnol. 34, 828–837 (2016).27504778
26. Aron AT Reproducible molecular networking of untargeted mass spectrometry data using GNPS Nat. Protoc. 2020 10.1038/s41596-020-0317-5 32405051
Aron, A. T. et al. Reproducible molecular networking of untargeted mass spectrometry data using GNPS. Nat. Protoc.10.1038/s41596-020-0317-5 (2020).32405051 10.1038/s41596-020-0317-5
27. Watrous J Mass spectral molecular networking of living microbial colonies Proc. Natl. Acad. Sci. U. S. A. 2012 109 E1743 E1752 22586093
Watrous, J. et al. Mass spectral molecular networking of living microbial colonies. Proc. Natl. Acad. Sci. U. S. A. 109, E1743–E1752 (2012).22586093
28. KEGG PATHWAY: Phenylpropanoid biosynthesis - Citrus sinensis (Valencia orange). https://www.genome.jp/kegg-bin/show_pathway?cit00940.
29. Feng G Ai X Yi H Guo W Wu J Genomic and transcriptomic analyses of Citrus sinensis varieties provide insights into Valencia orange fruit mastication trait formation Hortic. Res. 2021 8 218 34593784
Feng, G., Ai, X., Yi, H., Guo, W. & Wu, J. Genomic and transcriptomic analyses of Citrus sinensis varieties provide insights into Valencia orange fruit mastication trait formation. Hortic. Res. 8, 218 (2021).34593784
30. Chen Q Metabolomic analysis revealed distinct physiological responses of leaves and roots to Huanglongbing in a citrus rootstock Int. J. Mol. Sci. 2022 23 9242 36012507
Chen, Q. et al. Metabolomic analysis revealed distinct physiological responses of leaves and roots to Huanglongbing in a citrus rootstock. Int. J. Mol. Sci. 23, 9242 (2022).36012507
31. Chin EL Mishchuk DO Breksa AP Slupsky CM Metabolite signature of Candidatus Liberibacter asiaticus infection in two citrus varieties J. Agric. Food Chem. 2014 62 6585 6591 24959841
Chin, E. L., Mishchuk, D. O., Breksa, A. P. & Slupsky, C. M. Metabolite signature of Candidatus Liberibacter asiaticus infection in two citrus varieties. J. Agric. Food Chem. 62, 6585–6591 (2014).24959841
32. Deng H Comparative leaf volatile profiles of two contrasting mandarin cultivars against Liberibacter asiaticus infection illustrate Huanglongbing tolerance mechanisms J. Agric. Food Chem. 2021 69 10869 10884 34499509
Deng, H. et al. Comparative leaf volatile profiles of two contrasting mandarin cultivars against Liberibacter asiaticus infection illustrate Huanglongbing tolerance mechanisms. J. Agric. Food Chem. 69, 10869–10884 (2021).34499509
33. Padhi EMT Metabolome and microbiome signatures in the roots of citrus affected by Huanglongbing Phytopathology 2019 109 2022 2032 31433274
Padhi, E. M. T. et al. Metabolome and microbiome signatures in the roots of citrus affected by Huanglongbing. Phytopathology 109, 2022–2032 (2019).31433274
34. Blacutt A An in vitro pipeline for screening and selection of citrus-associated microbiota with potential anti-‘Candidatus Liberibacter asiaticus’ properties Appl. Environ. Microbiol. 2020 10.1128/AEM.02883-19 32086307
Blacutt, A. et al. An in vitro pipeline for screening and selection of citrus-associated microbiota with potential anti-‘Candidatus Liberibacter asiaticus’ properties. Appl. Environ. Microbiol.10.1128/AEM.02883-19 (2020).32086307 10.1128/AEM.02883-19
35. Irigoyen S Plant hairy roots enable high throughput identification of antimicrobials against Candidatus Liberibacter spp Nat. Commun. 2020 11 5802 33199718
Irigoyen, S. et al. Plant hairy roots enable high throughput identification of antimicrobials against Candidatus Liberibacter spp. Nat. Commun. 11, 5802 (2020).33199718
36. Li J The in planta effective concentration of oxytetracycline against ’ Liberibacter asiaticus’ for suppression of citrus Huanglongbing Phytopathology 2019 109 2046 2054 31369360
Li, J. et al. The in planta effective concentration of oxytetracycline against ’ Liberibacter asiaticus’ for suppression of citrus Huanglongbing. Phytopathology 109, 2046–2054 (2019).31369360
37. Hu J Jiang J Wang N Control of CITRUS Huanglongbing via trunk injection of plant defense activators and antibiotics Phytopathology 2018 108 186 195 28945516
Hu, J., Jiang, J. & Wang, N. Control of CITRUS Huanglongbing via trunk injection of plant defense activators and antibiotics. Phytopathology 108, 186–195 (2018).28945516
38. Munir S Unraveling the metabolite signature of citrus showing defense response towards Candidatus Liberibacter asiaticus after application of endophyte Bacillus subtilis L1–21 Microbiol. Res. 2020 234 126425 32035248
Munir, S. et al. Unraveling the metabolite signature of citrus showing defense response towards Candidatus Liberibacter asiaticus after application of endophyte Bacillus subtilis L1–21. Microbiol. Res. 234, 126425 (2020).32035248
39. Hung W-L Wang Y A targeted mass spectrometry-based metabolomics approach toward the understanding of host responses to Huanglongbing disease J. Agric. Food Chem. 2018 66 10651 10661 30220206
Hung, W.-L. & Wang, Y. A targeted mass spectrometry-based metabolomics approach toward the understanding of host responses to Huanglongbing disease. J. Agric. Food Chem. 66, 10651–10661 (2018).30220206
40. Suh JH Tang X Zhang Y Gmitter FG Jr Wang Y Metabolomic analysis provides new insight into tolerance of Huanglongbing in Citrus Front. Plant Sci. 2021 12 710598 34421957
Suh, J. H., Tang, X., Zhang, Y., Gmitter, F. G. Jr. & Wang, Y. Metabolomic analysis provides new insight into tolerance of Huanglongbing in Citrus. Front. Plant Sci. 12, 710598 (2021).34421957
41. Nuutila AM Kammiovirta K Oksman-Caldentey K-M Comparison of methods for the hydrolysis of flavonoids and phenolic acids from onion and spinach for HPLC analysis Food Chem. 2002 76 519 525 10.1016/s0308-8146(01)00305-3
Nuutila, A. M., Kammiovirta, K. & Oksman-Caldentey, K.-M. Comparison of methods for the hydrolysis of flavonoids and phenolic acids from onion and spinach for HPLC analysis. Food Chem. 76, 519–525. 10.1016/s0308-8146(01)00305-3 (2002).10.1016/s0308-8146(01)00305-3
42. Pourcel L Routaboul J Cheynier V Lepiniec L Debeaujon I Flavonoid oxidation in plants: from biochemical properties to physiological functions Trends Plant Sci. 2007 12 29 36 10.1016/j.tplants.2006.11.006 17161643
Pourcel, L., Routaboul, J., Cheynier, V., Lepiniec, L. & Debeaujon, I. Flavonoid oxidation in plants: from biochemical properties to physiological functions. Trends Plant Sci. 12, 29–36. 10.1016/j.tplants.2006.11.006 (2007).17161643 10.1016/j.tplants.2006.11.006
43. Mandadi K Irigoyen S Ancona V Sétamou M Coaker G Borneman J Irey M Hairy roots to the rescue: Speeding up discovery for HLB Management Citrograph 2020 11 20 22
Mandadi, K. et al. Hairy roots to the rescue: Speeding up discovery for HLB Management. Citrograph 11, 20–22 (2020).
44. Kennedy JP A perspective on current therapeutic molecule screening methods against ’ Liberibacter asiaticus’, the presumed causative agent of citrus Huanglongbing Phytopathology 2023 113 1171 1179 36750555
Kennedy, J. P. et al. A perspective on current therapeutic molecule screening methods against ’ Liberibacter asiaticus’, the presumed causative agent of citrus Huanglongbing. Phytopathology 113, 1171–1179 (2023).36750555
45. Zhang M Field evaluation of integrated management for mitigating citrus Huanglongbing in Florida Front. Plant Sci. 2018 9 1890 30766544
Zhang, M. et al. Field evaluation of integrated management for mitigating citrus Huanglongbing in Florida. Front. Plant Sci. 9, 1890 (2018).30766544
46. Davis MJ Mondal SN Chen H Rogers ME Brlansky RH Co-cultivation of ‘Candidatus Liberibacter asiaticus’ with actinobacteria from citrus with Huanglongbing Plant Dis. 2008 92 1547 1550 30764444
Davis, M. J., Mondal, S. N., Chen, H., Rogers, M. E. & Brlansky, R. H. Co-cultivation of ‘Candidatus Liberibacter asiaticus’ with actinobacteria from citrus with Huanglongbing. Plant Dis. 92, 1547–1550 (2008).30764444
47. Leonard MT Fagen JR Davis-Richardson AG Davis MJ Triplett EW Complete genome sequence of Liberibacter crescens BT-1 Stand. Genom. Sci. 2012 7 271 283
Leonard, M. T., Fagen, J. R., Davis-Richardson, A. G., Davis, M. J. & Triplett, E. W. Complete genome sequence of Liberibacter crescens BT-1. Stand. Genom. Sci. 7, 271–283 (2012).
48. Barnett MJ Solow-Cordero DE Long SR A high-throughput system to identify inhibitors of Liberibacter asiaticus transcription regulators Proc. Natl. Acad. Sci. U. S. A. 2019 116 18009 18014 31427509
Barnett, M. J., Solow-Cordero, D. E. & Long, S. R. A high-throughput system to identify inhibitors of Liberibacter asiaticus transcription regulators. Proc. Natl. Acad. Sci. U. S. A. 116, 18009–18014 (2019).31427509
49. Jain M Is a cultured surrogate for functional genomics of uncultured pathogenic ‘Liberibacter’ spp. and is naturally competent for transformation Phytopathology 2019 109 1811 1819 31090497
Jain, M. et al. Is a cultured surrogate for functional genomics of uncultured pathogenic ‘Liberibacter’ spp. and is naturally competent for transformation. Phytopathology 109, 1811–1819 (2019).31090497
50. dos Santos WD Soybean (Glycine max) root lignification induced by ferulic acid. The possible mode of action J. Chem. Ecol. 2008 34 1230 1241 18626717
dos Santos, W. D. et al. Soybean (Glycine max) root lignification induced by ferulic acid. The possible mode of action. J. Chem. Ecol. 34, 1230–1241 (2008).18626717
51. Cushnie TPT Lamb AJ Antimicrobial activity of flavonoids Int. J. Antimicrob. Agents 2005 26 343 356 16323269
Cushnie, T. P. T. & Lamb, A. J. Antimicrobial activity of flavonoids. Int. J. Antimicrob. Agents 26, 343–356 (2005).16323269
52. Ou S Kwok K-C Ferulic acid: Pharmaceutical functions, preparation and applications in foods J. Sci. Food Agric. 2004 84 1261 1269
Ou, S. & Kwok, K.-C. Ferulic acid: Pharmaceutical functions, preparation and applications in foods. J. Sci. Food Agric. 84, 1261–1269 (2004).
53. Paczkowski JE Flavonoids suppress virulence through allosteric inhibition of quorum-sensing receptors J. Biol. Chem. 2017 292 4064 4076 28119451
Paczkowski, J. E. et al. Flavonoids suppress virulence through allosteric inhibition of quorum-sensing receptors. J. Biol. Chem. 292, 4064–4076 (2017).28119451
54. Borges A Saavedra MJ Simões M The activity of ferulic and gallic acids in biofilm prevention and control of pathogenic bacteria Biofouling 2012 28 755 767 22823343
Borges, A., Saavedra, M. J. & Simões, M. The activity of ferulic and gallic acids in biofilm prevention and control of pathogenic bacteria. Biofouling 28, 755–767 (2012).22823343
55. Li A-P He Y-H Zhang S-Y Shi Y-P Antibacterial activity and action mechanism of flavonoids against phytopathogenic bacteria Pestic. Biochem. Physiol. 2022 188 105221 36464329
Li, A.-P., He, Y.-H., Zhang, S.-Y. & Shi, Y.-P. Antibacterial activity and action mechanism of flavonoids against phytopathogenic bacteria. Pestic. Biochem. Physiol. 188, 105221 (2022).36464329
56. Borges A Ferreira C Saavedra MJ Simões M Antibacterial activity and mode of action of ferulic and gallic acids against pathogenic bacteria Microb. Drug Resist. 2013 19 256 265 23480526
Borges, A., Ferreira, C., Saavedra, M. J. & Simões, M. Antibacterial activity and mode of action of ferulic and gallic acids against pathogenic bacteria. Microb. Drug Resist. 19, 256–265 (2013).23480526
57. Archer L Kunwar S Alferez F Batuman O Albrecht U Trunk injection of oxytetracycline for Huanglongbing management in mature grapefruit and sweet orange trees Phytopathology 2023 113 1010 1021 36474420
Archer, L., Kunwar, S., Alferez, F., Batuman, O. & Albrecht, U. Trunk injection of oxytetracycline for Huanglongbing management in mature grapefruit and sweet orange trees. Phytopathology 113, 1010–1021 (2023).36474420
58. Blaustein RA Lorca GL Teplitski M Challenges for managing Candidatus Liberibacter spp. (Huanglongbing disease pathogen): Current control measures and future directions Phytopathology 2018 108 424 435 28990481
Blaustein, R. A., Lorca, G. L. & Teplitski, M. Challenges for managing Candidatus Liberibacter spp. (Huanglongbing disease pathogen): Current control measures and future directions. Phytopathology 108, 424–435 (2018).28990481
59. Gardner CL Assessment of unconventional antimicrobial compounds for the control of ‘Candidatus Liberibacter asiaticus’, the causative agent of citrus greening disease Sci. Rep. 2020 10 5395 32214166
Gardner, C. L. et al. Assessment of unconventional antimicrobial compounds for the control of ‘Candidatus Liberibacter asiaticus’, the causative agent of citrus greening disease. Sci. Rep. 10, 5395 (2020).32214166
60. Archer L Qureshi J Albrecht U Efficacy of trunk injected imidacloprid and oxytetracycline in managing huanglongbing and Asian citrus psyllid in infected sweet orange (Citrus sinensis) trees Collect. FAO Agric. 2022 12 1592
Archer, L., Qureshi, J. & Albrecht, U. Efficacy of trunk injected imidacloprid and oxytetracycline in managing huanglongbing and Asian citrus psyllid in infected sweet orange (Citrus sinensis) trees. Collect. FAO Agric. 12, 1592 (2022).
61. Xia J Wishart DS Using MetaboAnalyst 3.0 for comprehensive metabolomics data analysis Curr. Protoc. Bioinform. 2016 55 14.10.1 14.10.91
Xia, J. & Wishart, D. S. Using MetaboAnalyst 3.0 for comprehensive metabolomics data analysis. Curr. Protoc. Bioinform. 55, 14.10.1-14.10.91 (2016).
62. Hackstadt AJ Hess AM Filtering for increased power for microarray data analysis BMC Bioinform. 2009 10.1186/1471-2105-10-11
Hackstadt, A. J. & Hess, A. M. Filtering for increased power for microarray data analysis. BMC Bioinform.10.1186/1471-2105-10-11 (2009).10.1186/1471-2105-10-11
63. Wold S Sjöström M Eriksson L PLS-regression: A basic tool of chemometrics Chemom. Intell. Lab. Syst. 2001 58 109 130
Wold, S., Sjöström, M. & Eriksson, L. PLS-regression: A basic tool of chemometrics. Chemom. Intell. Lab. Syst. 58, 109–130 (2001).
64. Breiman L Random forests Mach. Learn. 2001 45 5 32
Breiman, L. Random forests. Mach. Learn. 45, 5–32 (2001).
65. Nothias LF Feature-based molecular networking in the GNPS analysis environment Nat. Methods 2019 10.1101/812404 31686038
Nothias, L. F. et al. Feature-based molecular networking in the GNPS analysis environment. Nat. Methods10.1101/812404 (2019).31686038 10.1101/812404
66. Kyselka J Antifungal polyamides of hydroxycinnamic acids from sunflower bee pollen J. Agric. Food Chem. 2018 66 11018 11026 10.1021/acs.jafc.8b03976 30296072
Kyselka, J. et al. Antifungal polyamides of hydroxycinnamic acids from sunflower bee pollen. J. Agric. Food Chem. 66, 11018–11026. 10.1021/acs.jafc.8b03976 (2018).30296072 10.1021/acs.jafc.8b03976
67. Schellenberger J Quantitative prediction of cellular metabolism with constraint-based models: The COBRA Toolbox v2.0 Nat. Protoc. 2011 6 1290 1307 21886097
Schellenberger, J. et al. Quantitative prediction of cellular metabolism with constraint-based models: The COBRA Toolbox v2.0. Nat. Protoc. 6, 1290–1307 (2011).21886097
68. Zheng Z Unusual five copies and dual forms of nrdB in ‘Candidatus Liberibacter asiaticus’: Biological implications and PCR detection application Sci. Rep. 2016 6 39020 27958354
Zheng, Z. et al. Unusual five copies and dual forms of nrdB in ‘Candidatus Liberibacter asiaticus’: Biological implications and PCR detection application. Sci. Rep. 6, 39020 (2016).27958354
69. Mafra V Reference genes for accurate transcript normalization in citrus genotypes under different experimental conditions PLoS One 2012 7 e31263 22347455
Mafra, V. et al. Reference genes for accurate transcript normalization in citrus genotypes under different experimental conditions. PLoS One 7, e31263 (2012).22347455
