Molecular and Physiological Plant PathologyFree Access icon

Metabolomic Patterns of Septoria Canker Resistant and Susceptible Populus trichocarpa Genotypes 24 Hours Postinoculation

    Affiliations
    Authors and Affiliations
    • Ryan R. Lenz1
    • Katherine B. Louie2
    • Kelsey L. Søndreli1
    • Stephanie S. Galanie3
    • Jin-Gui Chen3
    • Wellington Muchero3
    • Benjamin P. Bowen2
    • Trent R. Northen2
    • Jared M. LeBoldus1 4
    1. 1Botany and Plant Pathology, Oregon State University, Corvallis, OR 97331
    2. 2Metabolomics Technology, DOE Joint Genome Institute, Lawrence Berkeley National Laboratory, Berkeley, CA 94720
    3. 3Biosciences Division, Oak Ridge National Laboratory, Oak Ridge, TN 37831
    4. 4Forest Resources, Engineering, and Management Department, Oregon State University, Corvallis, OR 97331

    Published Online:https://doi.org/10.1094/PHYTO-02-21-0053-R

    Abstract

    Sphaerulina musiva is an economically and ecologically important fungal pathogen that causes Septoria stem canker and leaf spot disease of Populus species. To bridge the gap between genetic markers and structural barriers previously found to be linked to Septoria canker disease resistance in poplar, we used hydrophilic interaction liquid chromatography and tandem mass spectrometry to identify and quantify metabolites involved with signaling and cell wall remodeling. Fluctuations in signaling molecules, organic acids, amino acids, sterols, phenolics, and saccharides in resistant and susceptible P. trichocarpa inoculated with S. musiva were observed. The patterns of 222 metabolites in the resistant host implicate systemic acquired resistance (SAR), cell wall apposition, and lignin deposition as modes of resistance to this hemibiotrophic pathogen. This pattern is consistent with the expected response to the biotrophic phase of S. musiva colonization during the first 24 h postinoculation. The fungal pathogen metabolized key regulatory signals of SAR, other phenolics, and precursors of lignin biosynthesis that were depleted in the susceptible host. This is the first study to characterize metabolites associated with the response to initial colonization by S. musiva between resistant and susceptible hosts.

    Populus species and their hybrids (i.e., poplars) are valuable trees commonly grown in plantation forestry because of their rapid growth, ease of propagation, and adaptation to a range of edaphic conditions (Ward and Ostry 2005). However, poplar productivity is limited by the pathogenic fungus Sphaerulina musiva, which causes Septoria stem canker and leaf spot disease (Bier 1939; Feau et al. 2010; Waterman 1955). Multiple poplar species and their hybrids are susceptible to Septoria stem canker including black cottonwood (P. trichocarpa Torr. & A. Gray), balsam poplar (P. balsamifera L.), black poplar (P. nigra L.), Japanese poplar (P. maximowiczii Henry), and laurel poplar (P. laurifolia Ledebour) (Bergdahl and Hill 2016). S. musiva is native to eastern North America but has recently expanded its range to the Pacific Northwest, threatening forestland, riparian areas, and plantations (Callan et al. 2007; Søndreli et al. 2020). To combat this disease, we need a better understanding of the interaction between the poplar hosts and S. musiva.

    Many metabolites act as signaling components leading to the production of pathogenesis-related (PR) proteins and phytoalexins mediating plant–microbe interactions. These include small molecules that are part of the induced responses initiated when a pathogen penetrates the physical barriers of the cuticle, bark, and plant cell walls (Kovalchuk et al. 2013). Certain pathogen-associated metabolites are perceived by cell surface receptors, which induce a signal for stress responses (Couto and Zipfel 2016). Downstream plant host cellular responses include an influx of calcium ions, a burst of reactive oxygen species (ROS), transcriptional reprogramming (often via WRKY transcription factors), and cell signaling via salicylic acid (SA), jasmonic acid (JA), and other plant growth regulators and small molecules (Bigeard et al. 2015). The culmination of these signals leads to defense responses such as cell wall apposition, translation of PR proteins, and production of phytoalexins to inhibit pathogen colonization. In Populus species, genetic and transcriptomic evidence suggests that an early defense response contributes to resistance to Septoria canker and leaf spot disease, in contrast to susceptible genotypes, which have a reduced or delayed defense response (Abraham et al. 2019; Liang et al. 2001; Muchero et al. 2018).

    S. musiva is a hemibiotroph (Ohm et al. 2012) with an initial biotrophic phase of infection in which the host’s immune response is actively suppressed or evaded to facilitate colonization (Horbach et al. 2011). After the initial colonization, hemibiotrophs switch to a necrotrophic phase in which the fungus can consume dead and dying host tissue. S. musiva infects at wounds and natural openings of stem and leaf tissue (Abraham et al. 2019; Krupinsky 1989; Qin and LeBoldus 2014). In a compatible interaction, disease resistance gene expression is suppressed during the first 72 h of infection (Abraham et al. 2019; Muchero et al. 2018). This inhibition allows the fungus to colonize the plant, eventually leading to necrosis 2 to 3 weeks after initial infection (Qin and LeBoldus 2014; Weiland and Stanosz 2007). In response to the pathogen, resistant poplar trees upregulate expression of cell surface receptors and other genes commonly associated with resistance to biotrophic pathogens including those encoding PR proteins (e.g., chitinases) and the phenylpropanoid pathway (Abraham et al. 2019; Foster et al. 2015; Muchero et al. 2018). In addition, resistant trees develop thick layers of necrophylactic periderm (NP) localized at infection sites (Qin and LeBoldus 2014). This NP is characterized by the lignification and suberization of cells immediately surrounding affected tissue (Biggs et al. 1984; Mullick 1977). Overall, resistance to S. musiva requires rapid disease response signaling during the initial stages of the interaction to induce the development of structural barriers that inhibit colonization of S. musiva; however, there is limited information about the metabolic patterns and signaling molecules involved in this response.

    Metabolic patterns correlated with plant colonization by microbes including blast and blight fungi of cereals (Balmer et al. 2013; Parker et al. 2009), phytoplasma diseases of mulberry and grape (Gai et al. 2014; Prezelj et al. 2016), herbivory of oak (Kersten et al. 2013), rust disease in the Rosaceae (Lee et al. 2016), mycorrhizal mutualism in poplar (Tschaplinski et al. 2014), and plant–pest interactions in Arabidopsis and other model species (Heuberger et al. 2014) have been characterized. To test the hypothesis that resistance to S. musiva is facilitated by cell wall remodeling and cellular signaling cascades, we used hydrophilic interaction liquid chromatography (HILIC) and tandem mass spectrometry (MS/MS) in a targeted metabolomics approach to analyze the accumulation and depletion patterns of small molecules in Septoria canker resistant and susceptible poplar stems after inoculation. The specific objectives of the study were to identify key signaling molecules and physiological markers associated with resistance and susceptibility and characterize the consumption and secretion of metabolites by the fungus while growing in liquid media.

    MATERIALS AND METHODS

    Plant propagation.

    Dormant cuttings of P. trichocarpa genotypes SKWE 24-5 (resistant) and BESC-148 (susceptible) (Muchero et al. 2018) were collected from the Oregon State University Research Farm near Corvallis, Oregon. Cuttings measuring 10 cm in length were soaked in distilled water for 48 h and then planted in 3.8 × 21 cm cones (Ray Leach SC10 Super Cone-Tainers; Stuewe and Sons, Inc., Tangent, OR) containing SunGro Professional Mix #8 (SunGro Horticulture Ltd., Agawam, MA). The medium was amended with 38 g of Osmocote slow-release fertilizer (15–9–12, N–P–K) with micronutrients (7.0% NH3-N, 8.0% NO3-N, 9.0% P2O5, 12.0% K2O, 1.0% Mg, 2.3% S, 0.02% B, 0.05% Cu, 0.45% Fe, 0.23% chelated Fe, 0.06% Mn, 0.02% Mo, and 0.05% Zn; Scotts Osmocote Plus; Scotts Company Ltd., Marysville, OH). Plants were grown in the greenhouse with an 18-h photoperiod under 600-W high-pressure sodium lamps and a temperature regime of 20/16°C (day/night). Trees were watered as needed and fertilized bimonthly with liquid fertilizer (20–20–20, N–P–K; Scotts Peters Professional; Scotts Company Ltd., Marysville, OH).

    Isolate propagation and inoculation.

    Frozen plugs of S. musiva, isolate MN-14, collected from cankered trees (Dunnell and LeBoldus 2017) were poured onto Petri plates containing sterile KV-8 medium (2 g of calcium carbonate, 20 g of agar, 820 ml of deionized water, and 180 ml of V-8 vegetable juice [Campbell Soup Company, Camden, NJ]) amended with 240 mg/liter of chloramphenicol (chloramphenicol USP, Amresco, Solon, OH) and 100 mg/liter of streptomycin (streptomycin sulfate USP, Amresco) (Stanosz and Stanosz 2002). After 7 days, plugs of sporulating fungal mycelium were subcultured onto new KV-8 plates. For the two separate inoculation experiments (i.e., resistant poplar experiment and susceptible poplar experiment), 7 days after subculturing, plugs of sporulating mycelium (about 5 mm in diameter) were placed over lenticels at two positions, approximately 15 cm from the soil surface. Mock inoculations were conducted in a similar manner with the sole exception that sterile KV-8 agar was used. Preparation of fungal cultures for the fungal growth experiment was conducted in an identical manner to that described previously.

    Experimental design.

    Resistant poplar experiment.

    P trichocarpa genotype SKWE 24-5 was inoculated with S. musiva isolate MN-14. The experimental design was completely randomized with four replicates. There were four inoculated and four mock-inoculated trees. Each tree was inoculated at two positions as described previously. Twenty-four h postinoculation (hpi), a piece of stem tissue from each tree including bark, vascular cambium, and phloem measuring approximately 1 cm2 was collected, flash frozen, and stored at −80°C.

    Susceptible poplar experiment.

    P. trichocarpa genotype BESC-148 was inoculated and sampled as described previously. The experimental design was completely randomized with inoculated (n = 3) and mock-inoculated trees (n = 3). Three reps instead of four were used because of the limited amount of plant material.

    Fungal growth experiment.

    The experimental design was completely randomized with inoculated (n = 4) and control (n = 4) treatments. MN-14 conidia were harvested from the surface of a KV-8 plate, described previously. Briefly, 100 µl of sterile distilled water was added to a plate with sporulating colonies of S. musiva, the surface was gently rubbed to dislodge the spores, and the spore suspension was used to inoculate four Erlenmeyer flasks containing 100 ml of clarified liquid KV-8 media. The flasks were incubated at room temperature on a rotating shaker (130 rpm). Four control Erlenmeyer flasks, containing 100 ml of sterile KV-8 media, were incubated under the same conditions. Four days after inoculation, the fungal biomass was separated from the inoculated media by centrifugation, the liquid was decanted, and the mycelial pellet was washed twice with 1 ml of phosphate buffered solution. Approximately 30 ml of the decanted media was collected from the inoculated flasks (spent media) and the control flasks. A total of 50 mg of mycelium (fresh weight) was collected in a 2-ml tube. All samples were flash frozen, lyophilized (FreeZone 2.5 Plus, Labconco), and stored at −80°C.

    Sample preparation.

    Poplar tissue.

    Equal weights from the 1-cm2 lyophilized poplar stems from each experiment were disrupted in 2-ml lysing matrix tubes with MP Biomedicals beads (#116918050-CF) with a bead beater (Mini-Beadbeater-96, Biospec Products) for 5 s, twice, at room temperature. To extract polar metabolites, we added 600 µl of methanol to each pulverized plant sample, and the samples were briefly vortexed, sonicated in a water bath for 10 min at room temperature, and centrifuged for 5 min at 5,000 rpm. For each sample, 200 µl of supernatant was transferred to a new tube. Extracts were dried in a SpeedVac (SPD111V, Thermo Scientific) and stored at −20°C.

    Fungal and media samples.

    The lyophilized fungal pellets and media samples were pulverized in 2-ml Eppendorf tubes containing 3.2-mm stainless steel beads with the same bead beater and settings described previously. Polar metabolites were extracted with 350 µl of methanol. Samples were vortexed, sonicated, and centrifuged as described previously. Then 200 µl of supernatant was transferred to a new tube for each sample. Extracts were dried in a SpeedVac and stored at −20°C. Extraction controls for all experiments prepared via the same procedure, on empty microcentrifuge tubes, were included.

    HILIC liquid chromatography MS/MS.

    Samples from all three experiments were processed in a similar manner. The dried extracts were resuspended in 150 µl of 100% methanol containing 10 µM labeled internal standards (5 to 50 µM of 13C-15N Cell Free Amino Acid Mixture, #767964, Sigma). The samples were briefly vortexed and sonicated for 5 min. Subsequently, 150 µl of each sample was centrifuge filtered (0.22 µM polyvinylidene difluoride membrane; Millipore, Ultrafree-CL GV, #UFC40GV0S) and transferred to glass liquid chromatography-mass spectrometry (LC-MS) vials. Liquid chromatography MS/MS was performed via normal phase chromatography with an Agilent 1290 LC, with MS and MS/MS data collected with a Q Exactive Orbitrap Mass Spectrometer (Thermo Scientific, San Jose, CA). Full MS spectra were collected in both positive and negative ion modes, with an m/z ratio of 70 to 1,050 at 70,000 full width at half maximum resolution. MS/MS fragmentation data were acquired using stepped then averaged 10-, 20-, and 40-eV collision energies at 17,500 full width at half maximum resolution. Orbitrap instrument parameters included a sheath gas flow rate of 50 arbitrary units (au), auxiliary gas flow rate of 20 (au), sweep gas flow rate of 2 (au), 3-kV spray voltage, and 400°C capillary temperature. Chromatography was performed in an HILIC column (Millipore SeQuant ZIC-HILIC, 150 × 2.1 mm, 5 µm, catalog no. 50454) at 40°C with a 2-µl injection volume for each sample. Gradient chromatography was performed at a flow rate of 0.45 ml/min beginning with 100% buffer B (95:5 ACN/H2O with 5 mM ammonium acetate) for 1.5 min, a gradient to 65% with buffer B and 35% with buffer A (H2O with 5 mM ammonium acetate) over 13.5 min, a further gradient to 0% B over 3 min while increasing flow to 0.6 ml/min, and finally an isocratic elution in 100% buffer A for 5 min. Sample injection order was randomized with methanol blanks run between each sample. Quality control samples containing pooled aliquots of each biological sample per experiment were used for the duration of analyses to monitor feature variance.

    Feature selection and identification.

    For all experiments, metabolites were identified from a list of MS features based on exact mass, comparing retention time (RT), and MS/MS fragmentation spectra compared to that of the standards. LC-MS data were analyzed via custom Python code (Yao et al. 2015). Each detected feature (unique m/z coupled with RT) was assigned a score (0 to 3) representing the level of confidence in the compound identification and subsequently rated according to the Metabolomics Standards Initiative (MSI) confidence levels (Sumner et al. 2007). Compounds given a positive identification were detected at m/z ≤ 5 ppm or 0.001 Da from theoretical as well as RT ≤ 0.5 min compared with a pure standard run under the same LC-MS method. A compound with the highest level of positive identification (score of 3) also had matching MS/MS fragmentation spectra compared with either an outside database (METLIN) or internal database generated from standards run and collected on a Q Exactive Orbitrap Mass Spectrometer. Any MS/MS mismatches were listed as invalidated as an identification. All scores are reported in Supplementary File Metadata and the MS/MS spectra and authentic standards are provided in Supplementary File MSMS_mirrorplots.

    Data filtering and normalization.

    All sample ion intensities (peak heights) from positive and negative modes of each feature were combined into one Excel file (Metadata). Before analysis, prefiltering for each experiment included the following steps: The median of the blank sample runs for each feature was compared with the maximum value of the plant tissue. Any percentage (blank median/sample maximum) >50% was removed from the list of features and was not analyzed. Subsequently, all features were sorted by name and MSI level. All features that were invalidated by MS/MS were removed. For metabolites detected in both positive and negative ionization modes, one mode was removed based on having a lower MSI score or lower abundance. Metabolites with opposing patterns of abundance (i.e., accumulated and depleted) between its duplicates were completely omitted from the dataset. The remaining features were organized by treatment groups for statistical analysis in MetaboAnalyst version 4.0 (https://www.metaboanalyst.ca; Chong et al. 2019; Xia et al. 2015). The default missing value estimation parameters were used to remove features with >80% missing values and to impute values for those with <80% missing data. The replacement values were half of the minimum positive values in the original data assumed to be the limit of detection. The prefiltered dataset was normalized via the cubed root of each measurement and mean centering for each feature. Sample and feature medians were validated for alignment as a metric of successful data normalization. This normalization process was contrasted to manual centered log ratio transformation to confirm consistent patterns in the fold change of each feature. The normalized datasets for each experiment were analyzed independently with the statistical analysis module in MetaboAnalyst version 4.0 (Chong et al. 2019).

    Statistical analyses.

    Resistant and susceptible poplar experiments.

    For both poplar experiments, data were visualized via principal component analysis. The top 25 features differentiating the treatments in each experiment were selected according to variable importance in projection (VIP) scores estimated from partial least squares discriminant analysis (PLSDA). The predictive power of PLSDA was cross-validated via leave-one-out cross-validation in MetaboAnalyst (Worley and Powers 2013). The fold change between inoculated and mock-inoculated treatments and corresponding P values (t test) for each feature were reported for each experiment (Vinaixa et al. 2012). The top 25 metabolites from each experiment according to fold change and P values were visualized as a heatmap. Overall, metabolites with a VIP score >0.90 (Cocchi et al. 2018) or a P < 0.10 were considered significantly different between treatments.

    Metabolic pathway enrichment analysis was also conducted in MetaboAnalyst with GlobalTest and GlobalAncova (Hummel et al. 2008) for enrichment and betweenness centrality to estimate node importance to the pathway. This analysis linked compounds to physiological function based on the Kyoto Encyclopedia of Genes and Genomes (https://www.genome.jp/kegg/) dataset for Arabidopsis thaliana. Metabolic pathways were considered significantly enriched if the −log10(P) was >2 or the pathway impact (i.e., ratio of represented metabolites and node impact) was >0.50. All identified metabolites were verified with MAGI (Erbilgin et al. 2019) to confirm the likelihood of being produced or consumed by the fungus or plant used in the experiments.

    Fungal growth experiment.

    The relative abundance of all metabolites among fungal biomass, spent media, and control media were compared via one-way analysis of variance (false discovery rate [FDR] <0.05) and Fisher’s least significant difference (α < 0.05). Hierarchical clustering was used to categorize the pattern of fungal accumulation or secretion across the three treatments (Supplementary Fig. S1). Categories included Biosynthesized & secreted, low abundance in control media and fungal biomass, abundant in spent media (may include waste products); Biosynthesized & accumulated, low abundance in both media types, abundant in the fungal biomass; Catabolized, depleted in spent media; Consumed, depleted in spent media and also accumulated in fungal biomass; and Not metabolized, no change between control and spent media and not accumulated in fungal biomass. PLSDA was used to identify the top metabolites differentiating treatments as described previously but was not used to compare treatment means.

    RESULTS

    Global patterns.

    A total of 456 putatively identified features were detected from positive and negative ionization via HILIC-MS/MS. From 320 authentic standards, 206 features were confidently identified as metabolites or compounds in the resistant poplar experiment after filtering, with an additional 16 that remained putatively identified (Fig. 1A; Supplementary File Metadata.xlsx). In a similar approach, 160 metabolites from a list of 280 features were identified in the susceptible poplar experiment. A total of 135 matched those identified in the resistant poplar experiment (Fig. 1A). For the fungal growth experiment, 247 metabolites were identified with 219 matching the resistant poplar experiment (Fig. 1A; Supplementary File Metadata.xlsx). Downstream analysis was conducted on the metabolites initially identified in the resistant poplar experiment and subsequently identified in the other two experiments.

    Fig. 1.

    Fig. 1. Total metabolites identified and overall data differentiation between mock-inoculated and Sphaerulina musiva-inoculated stem tissue. A, Venn diagram of the number of metabolites identified and analyzed among the three experimental groups: resistant poplar (res.), susceptible poplar (sus.), and fungal (fun.) growth experiment. B, Principal component analysis (PCA) of the resistant plant samples. C, PCA of the susceptible plant samples. D, PCA of the fungal biomass and media samples. The circles in the PCA plots represent 95% confidence regions.

    Download as PowerPoint

    Principal component analysis was used to summarize the data for each experiment (Fig. 1B, 1C, 1D). In all three experiments the variables clustered according to treatment class. In the resistant poplar experiment, principal component 1 (PC1) explained 44.9% of the variation between treatments. In the susceptible poplar experiment, the majority of the separation of the treatments was along PC2 (31.7%), and in the fungal growth experiment all treatments were separated by PC1 (63.9%) and PC2 (34.3%).

    Resistant poplar experiment.

    The 25 metabolites with the greatest contribution to the separation of treatment groups were identified (Fig. 2A and B). Eight of the metabolites with the greatest contribution were carbohydrates or sugar alcohols, six were related to lipid metabolism, three were associated with nucleic acid metabolism, 16 were associated with amino acid metabolism, five were plant growth regulators, and seven were involved with stress signaling. All of these metabolites had a P < 0.10 or a VIP score of >0.90. All VIP values ranged from zero to 7.20, with an average score of 0.43. Pathway enrichment analysis highlighted zeatin biosynthesis, glycerophospholipid metabolism, indole alkaloid biosynthesis, purine metabolism, phenylalanine and tryptophan metabolism, carbohydrate metabolism, the tricarboxylic acid cycle, phenylpropanoid biosynthesis, amino acid metabolism, stress signaling, and biosynthesis of other secondary metabolites (Supplementary Fig. S2).

    Fig. 2.

    Fig. 2. Top 25 significant features identified from A and C, partial least squares discriminant analysis and B and D, t tests with heatmap representation of the relative log2-fold change of abundance between treatments. A and B, Resistant poplar experiment; C and D, susceptible poplar experiment. In A and C, circles indicate a relative increase in abundance in inoculated samples relative to the control, and blue triangles represent a relative decrease in abundance. Metabolites are categorized with superscripts as follows: 1, carbohydrate metabolism; 2, lipid metabolism; 3, nucleic acid metabolism; 4, amino acid metabolism; 5, plant growth regulator; 6, stress signaling molecule; 7, phenolic/aromatic acid.

    Download as PowerPoint

    Susceptible poplar experiment.

    Of the top 25 metabolites identified by PLSDA and t tests, 10 were carbohydrate-associated metabolites, seven were lipid-associated metabolites, four were metabolites associated with nucleic acids, seven were metabolites associated with amino acid metabolism, two were plant growth regulators, seven were signaling molecules, and four were aromatic acids (Fig. 2C and D). Of those, 15 had P < 0.10 or VIP > 0.90. All VIP values ranged from zero to 6.53, with an average score of 0.47. Pathway enrichment highlighted the lysine degradation pathway, amino acid metabolism, glycerolipid metabolism, vitamin B6 degradation, isoquinoloid biosynthesis, and other secondary metabolic pathways (Supplementary Fig. S3).

    Fungal growth experiment.

    The majority of the metabolites (244 of 247) had significant changes among the three fungal growth treatments. Hierarchical clustering analysis revealed five main metabolic patterns of the identified metabolites (Supplementary Fig. S1). The metabolites that were not significantly consumed or secreted were either not metabolized by the fungus or not present in the control media and instead accumulated in the fungal biomass or spent media (Supplementary Fig. S1). A total of 14 compounds were not metabolized by the fungus, and 43 were biosynthesized and accumulated in the fungal biomass. Of the rest, the fungus biosynthesized and secreted 41 metabolites, consumed 94 metabolites, and catabolized 56 metabolites without accumulation or absorption (Supplementary Fig. S1). PLSDA revealed that the top 25 metabolites differentiating the treatments were involved in carbohydrate, lipid, nucleotide, and amino acid metabolism (Supplementary Fig. S4). Likewise, pathway enrichment indicated that multiple amino acid and carbohydrate metabolism pathways were represented in the fungal growth experiment (Supplementary Fig. S5).

    Carbohydrates and sugar alcohol metabolism.

    Resistant poplar experiment.

    Approximately 12% (27/223) of the identified metabolites were carbohydrates or sugar alcohols. Eight of these metabolites exhibited significant fluctuations between the inoculated and mock-inoculated controls (Fig. 2A and B). Sucrose (VIP = 1.86, P = 0.16) and mannitol/sorbitol (VIP = 4.33, P = 1.40E-4) were the carbohydrates having the largest influence on treatment separation according to PLSDA. Both accumulated in the inoculated tissues (Fig. 2A). A significant accumulation of glucose (VIP = 0.83, P = 0.047), trehalose (VIP = 0.33, P = 0.01), and gluconolactone (VIP = 0.84, P = 3.88E-4) were observed. The sugar alcohols arabitol (VIP = 0.35, P = 0.02) and erythritol (VIP = 0.14, P = 0.02) (Fig. 2A and B) were accumulated. In contrast, rhamnose (VIP = 0.22, P = 0.07) had a significant decrease in abundance in inoculated tissue (Supplementary Table S1).

    Susceptible poplar experiment.

    Approximately 13% (17/135) of the metabolites were associated with carbohydrate metabolism, and 10 of those were identified as the top 25 differentiating features (Fig. 2C and D). PLSDA determined that sucrose (VIP = 1.80, P = 0.39), lactic acid (VIP = 1.02, P = 0.29), and pyridoxine (VIP = 0.70, P = 0.13) were depleted relative to the mock-inoculated controls. In the heatmap, erythritol (VIP = 0.25, P = 0.27), stachyose (VIP = 0.37, P = 0.12), D-glucose-6-phosphate (VIP = 0.45, P = 0.11), uridine diphosphate (UDP) glucose (VIP = 0.37; P = 0.20), and D-galacturonic acid (VIP = 0.27, P = 0.24) were all in the top 25 fluctuating metabolites but were not significantly depleted (Fig. 2D; Supplementary Table S1). Mannitol/sorbitol (VIP = 1.40; P = 0.09) was the only carbohydrate metabolite to increase in abundance after pathogen inoculation (Fig. 2C and D).

    Fungal growth experiment.

    Sucrose, D-glucose-6-phosphate, glucose, 2-deoxy-D-glucose, arabinose, xylose, fructose, ribose, maltose, glucosamine, gluconolactone, erythritol, and mannitol/sorbitol were all significantly depleted (FDR <0.05) in the spent media relative to the control (Supplementary Table S1). Most of those metabolites accumulated in the fungal biomass. The fungus biosynthesized and secreted (FDR <0.05) rhamnose, stachyose, and raffinose while accumulating trehalose, ribose-5-phosphate, arabitol, and UDP glucose and did not metabolize xanthosine and pyridoxine.

    Lipid and fatty acid metabolism.

    Resistant poplar experiment.

    Choline-O-sulfate (VIP = 6.54; P < 0.01), L-acetylcarnitine (VIP = 5.31; P < 0.01), L-carnitine (VIP = 7.20; P < 0.01), deoxycarnitine (VIP = 1.07; P = 0.01), mevalonic acid (VIP = 0.17; P = 0.12), and glycerol-3-phosphate (G3P) (VIP = 0.71; P = 0.04) all increased in the inoculated sample relative to the control (Fig. 2A and B, Supplementary Table S1). Phosphorylcholine was the only lipid-associated metabolite with a significant decrease in abundance (VIP = 1.49; P = 0.05).

    Susceptible poplar experiment.

    Increases in choline-O-sulfate (VIP = 6.30; P = 0.17), L-carnitine (VIP = 5.50; P = 0.03), deoxycarnitine (VIP = 1.79; P = 0.03), G3P (VIP = 0.42; P = 0.13), and phosphorylcholine (VIP = 1.76; P = 0.07) were observed in inoculated stem tissue relative to controls (Fig. 2C and D, Supplementary Table S1).

    Fungal growth experiment.

    The fungus biosynthesizes, accumulates, and secretes most of the lipid metabolites mentioned previously except G3P, deoxycarnitine, or phosphorylcholine, which did not accumulate in the spent media. Mevalonic acid (FDR <0.05) had the greatest fold change increase (54 times greater) in the spent media compared with all other metabolites (Supplementary Table S1).

    Nucleotide metabolism.

    Resistant poplar experiment.

    Adenosine monophosphate (AMP) (VIP = 0.32, P = 0.01), a precursor used in zeatin biosynthesis, was the only nucleotide metabolite with a significant increase in abundance after inoculation (Fig. 2B). Riboflavin (VIP = 0.23, P = 0.02), a vitamin involved in nucleotide metabolism, was also significantly more abundant in the inoculated tissue than in the control (Fig. 2B). Thymine (VIP = 0.10, P = 0.04) was the only nucleotide metabolite that had a significant decrease in abundance after inoculation (Fig. 2B).

    Susceptible poplar experiment.

    There were increases in allantoin (VIP = 1.62; P = 0.02), thymine (VIP = 1.22; P = 0.07), adenosine (VIP = 1.22; P = 0.59), and uridine (VIP = 0.94; P = 0.42) in inoculated stems relative to controls (Fig. 2C and D, Supplementary Table S1).

    Fungal growth experiment.

    The fungus appears to consume almost all metabolites associated with nucleotide metabolism (FDR <0.05) except for xanthine, S-adenosylmethionine, nicotinamide adenine dinucleotide, allantoin, deoxycytidine, and uric acid, which were biosynthesized and significantly accumulated in the fungal biomass (Supplementary Table S1). Thymine and riboflavin were biosynthesized and significantly accumulated in the spent media (Supplementary Table S1).

    Amino acid metabolism.

    Resistant poplar experiment.

    There was an overall reduction in available amino acids (Supplementary Table S1). This amino acid perturbation was accompanied by an accumulation of amino acid degradation products such as creatine (VIP = 0.45; P = 0.03), creatinine (VIP = 0.52; P = 0.06), 2-oxovaleric acid (VIP = 1.07; P = 0.01), betaine (VIP = 4.68; P = 0.01), S-adenosylhomocysteine (VIP = 0.27; P = 0.1), and xanthurenic acid/zeanic acid (VIP = 3.10; P < 0.01) (Fig. 2A and B). The amino acids with the greatest reduction in abundance are aromatic and branched chained amino acids including the secondary metabolite precursors tyrosine (VIP = 0.90; P= 0.06) and tryptophan (VIP 1.14; P = 0.03) and the acetyl-CoA precursors valine (VIP = 1.20; P = 0.12), isoleucine (VIP = 1.70; P = 0.07), and leucine (VIP = 1.43; P = 0.11). The lysine degradation product and potential antimicrobial precursor aminoadipic acid (VIP = 0.46; P = 0.10) also decreased in abundance (Supplementary Table S1).

    Susceptible poplar experiment.

    Most of the amino acids were unchanged after inoculation with the exception of lysine (VIP = 0.18; P = 0.09), arginine (VIP = 0.68; P = 0.32), and tyrosine (VIP = 0.31; P = 0.22), which were more abundant in the inoculated versus mock-inoculated control (Fig. 2C and D; Supplementary Table S1). Histidinol (VIP = 0.11; P = 0.08) was the only metabolite related to amino acid metabolism that was significantly lower in the susceptible poplar experiment. There was no change in aminoadipic acid (VIP = 0.09; P = 0.33).

    Fungal biomass experiment.

    All identified amino acids except for L-glutamine were consumed from the media and accumulated in the fungal biomass. L-glutamine (FDR <0.05) was biosynthesized and accumulated by the fungus and also accumulated in the spent media. In addition, xanthurenic/zeanic acid, creatinine, creatine, carnosine, pyruvate, p-octopamine, pyroglutamic acid, kynurenic acid, and S-adenosylmethionine were all accumulated in the spent media. Metabolites associated with antibiotic production including (R/S)-mandelic acid and aminoadipic acid were biosynthesized by the fungus (Supplementary Table S1).

    Plant growth regulators and stress signaling molecules.

    Resistant poplar experiment.

    Concentrations of the auxin indole acetic acid (IAA) did not fluctuate on pathogen inoculation (VIP = 0.06; P = 0.68), nor did its precursor, indole-3-pyruvic acid (IPA) (VIP = 0.01; P = 0.93) (Supplementary Table S1). The only metabolite related to auxin metabolism with an increase in abundance was the putatively identified isomer labeled xanthurenic/zeanic acid (VIP = 3.10; P < 0.01). In contrast, the cytokinin pathway for zeatin biosynthesis was significantly enriched (Supplementary Fig. S2). The zeatin precursors AMP (VIP = 0.32; P = 0.01) and N6-(delta2-isopentenyl)-adenine (VIP = 0.55; P = 0.01) were more abundant in inoculated stems than in the control (Fig. 2A and B; Supplementary Table S1).

    Signaling molecules of systemic acquired resistance (SAR) were accumulated after pathogen inoculation (Table 1; Fig. 3). SA (VIP = 1.83; P = 0.13) and its precursors and derivatives, 2,3-dihydroxybenzoic acid (VIP = 1.17; P = 0.18), gentisic acid (VIP = 1.04, P = 0.22), and 3-aminosalicylic acid (VIP = 0.18; P = 0.17) increased in abundance (Fig. 3; Supplementary Table S1). G3P (VIP = 0.71; P = 0.04) also increased in abundance after pathogen challenge (Table 1). The SAR antagonistic metabolites abscisic acid (ABA) (VIP = 1.21; P = 0.04) and trigonelline (VIP = 1.88; P = 0.08) and the ethylene precursor 1-aminocyclopropane-carboxylic acid (VIP = 0.41; P = 0.09) were less abundant in the inoculated stems than in the control (Table 1). There was no significant change in the abundance of allantoin (VIP = 0.15; P = 0.31) (Table 1).

    TABLE 1. Fold change and raw P values (t test) of small molecules associated with plant disease response signaling between inoculated and mock-inoculated Populus trichocarpa and Sphaerulina musiva metabolism patterns

    Fig. 3.

    Fig. 3. Shikimate/phenylalanine ammonia lyase pathway for salicylic acid and polyphenol biosynthesis. Colors represent the relative log2 (fold change) of each metabolite in inoculated resistant stem tissue compared with control. Gray boxes indicate metabolites that were not identified from the mass spectrometry data.

    Download as PowerPoint

    Susceptible poplar experiment.

    The auxin IAA (VIP = 0.17; P = 0.48) did not change in abundance, whereas the zeatin precursors described previously were not successfully identified. In contrast, SA (VIP = 1.65; P = 0.04) was significantly less concentrated after pathogen inoculation (Table 1). The SAR antagonistic metabolite allantoin (VIP = 1.62, P = 0.02) was more abundant after pathogen challenge (Table 1). There was no significant change in the abundance of ABA (VIP = 0.24; P = 0.47) (Table 1).

    Fungal growth experiment.

    IAA was biosynthesized and secreted by the fungus while catabolizing its precursor IPA during growth in liquid media (FDR <0.05). Conversely, the cytokinin precursors AMP and N6-(delta2-isopentenyl)-adenine were consumed by the fungus (FDR <0.05) (Supplementary Table S1). The fungus also biosynthesized and secreted ABA while catabolizing SA and other SAR-related compounds from the media (FDR <0.05) (Table 1; Supplementary Table S1). The fungus biosynthesized and accumulated the SAR antagonist allantoin while biosynthesizing and secreting the growth inhibitors butanal and nicotinic acid (FDR <0.05).

    Phenolic and other aromatic acids.

    Resistant poplar experiment.

    The shikimate pathway precursor 3-dehydroshikimic acid (VIP = 0.73; P = 0.04) was accumulated after inoculation (Fig. 3). In conjunction, phenolics associated with lignin biosynthesis were also accumulated (Fig. 3). These metabolites included p-coumaric acid (VIP = 0.55; P = 0.19), caffeic acid (VIP = 0.79; P = 0.14), and quinic acid (VIP = 0.56; P = 0.16). In contrast, phenolics associated with herbivory and microbial mutualism including vanillin (VIP = 0.40; P = 0.49), vanillic acid (VIP = 0.04; P = 0.82), and syringic acid (VIP = 0.06; P = 0.80) did not change in abundance (Supplementary Table S1).

    Susceptible poplar experiment.

    The aromatic acids 3-dehydroshikimic acid (VIP = 0.42; P = 0.02) and 4-hydroxybenzoic acid (VIP = 0.89; P = 0.06) decreased in abundance after inoculation (Fig. 2C and D). The lignin precursor chlorogenic acid (VIP = 2.68; P = 0.27) also decreased in abundance. All other phenolics had no significant change between inoculated and control treatments.

    Fungal growth experiment.

    Lignin-associated phenolic and aromatic acids including coumaric acid, caffeic acid, ferulic acid, quinic acid, and chlorogenic acid were catabolized by S. musiva (FDR < 0.05), whereas the allelochemical benzoic acid was secreted (FDR < 0.05) (Supplementary Table S1).

    DISCUSSION

    Cell wall remodeling.

    Genomic and transcriptomic evidence suggests that rapid perception and subsequent changes in gene expression lead to Septoria canker resistance in P. trichocarpa (Muchero et al. 2018). These early changes in gene expression result in anatomical differences between resistant and susceptible poplar trees after inoculation (Qin and LeBoldus 2014; Weiland and Stanosz 2007), including NP development within 1 week of infection in resistant genotypes. NP is produced as a response to mechanical damage caused by pathogen colonization and characterized by the lignification and suberization of cells immediately below the affected tissue (Mullick 1977). This process probably involves the accumulation of cell wall precursors such as UDP glucose, arabinose, D-galacturonic acid, xylose, D-glucose, and phenolic acids (Benedetti et al. 2015; Lee et al. 2016; Rodrigues Mota et al. 2018). The accumulation of lignin precursors such as transcinnamic acid, p-coumaric acid, caffeic acid, ferulic acid, quinic acid, and chlorogenic acid could support the development of the NP layer. The contrasting patterns of accumulation and depletion in the two poplar experiments (Fig. 4; Supplementary Table S1) indicate that the resistant genotype is better able to marshal the resources necessary for cell wall remodeling in response to pathogen infection. Studies by Zhang et al. (2018) and Abraham et al. (2019) showed that genes involved with lignification are upregulated in resistant poplar compared with susceptible plants after S. musiva inoculation. In addition, the fungus catabolized these lignin precursors (Fig. 4; Supplementary Table S1), supporting the hypothesis that S. musiva metabolism could inhibit the production of structural barriers in the host.

    Fig. 4.

    Fig. 4. Cell wall remodeling and homeostasis-related metabolites with varying trends of accumulation within each experiment: inoculated resistant poplar, inoculated susceptible poplar, and spent media after Sphaerulina musiva growth. Relative fold change of each metabolite was calculated via log2 (treatment/control). Statistical significance is denoted with asterisks (P < 0.1 or partial least squares discriminant analysis variable importance in projection >1.0). All spent media metabolites were significant except for those labeled “ns.”

    Download as PowerPoint

    Not all sugars putatively involved in cell wall remodeling accumulated in the resistant genotype after infection. For example, rhamnose was depleted in the resistant genotype relative to the control. Rhamnose, a key building block of pectic rhamnogalacturonan I and rhamnogalacturonan II, can be a structural component of plant cell walls (Engelsdorf et al. 2017). However, it is also used in the production of secondary metabolites including anthocyanins, flavonoids, and triterpenoids (Saffer and Irish 2018; Sloothaak et al. 2016). This finding may suggest that depletion of this structural monosaccharide in the resistant poplar experiment precedes the production of antifungal compounds rather than cell wall remodeling. Rhamnose accumulated in the spent media, indicating that the fungus did not consume this structural monosaccharide.

    Osmoprotectants and antioxidants.

    Sugar alcohols (i.e., polyols), osmoprotectants, and antioxidant metabolites (Lewis and Smith 1967) were greatly enriched in comparisons of inoculated and mock-inoculated tissue regardless of the hosts resistance (Figs. 2 and 4; Supplementary Table S1). Generally, these compounds promote tolerance to abiotic stress by stabilizing intracellular proteins and neutralizing ROS (Slama et al. 2015; Street et al. 2006). However, their role in biotic stress is less clear. The protective polyol mannitol/sorbitol was accumulated in both poplar experiments, although the increase was approximately five times greater in the resistant relative to the susceptible poplar experiment (2.65 versus 12.07) (Fig. 4; Supplementary Table S1). Metabolites associated with fatty acid metabolism such as choline-o-sulfate (Nissen and Benson 1961; Slama et al. 2015), carnitine, and acetylcarnitine also accumulated in both poplar experiments (Supplementary Table S1). The main function of carnitine involves transport of fatty acids for energy production (Jacques et al. 2018; Oney-Birol 2019), but it also participates in cell signaling as an ABA inhibitor (Jacques et al. 2018). The related metabolite acetylcarnitine plays a role in plastid biogenesis and fatty acid transport for synthesis of glycerolipids (Jacques et al. 2018). One potential explanation for the accumulation of these metabolites regardless of the resistance status of the host tree is their generalized role in response to cellular damage. Consequently, it is consistent with the hypothesis that the resistant genotype is better able to mount a comprehensive and sustained defense response.

    Disease signaling by carbohydrate and amino acid metabolism.

    In the resistant poplar experiment, sucrose and glucose abundance increased in the inoculated tissue relative to the control (Figs. 2 and 5). The opposite pattern was apparent in the susceptible poplar experiment, where sucrose and glucose were depleted (Figs. 2 and 5). Sucrose, the main energy source translocated from source to sink tissue, can be hydrolyzed to initiate defense signaling directly (Chaliha et al. 2018; Lemoine et al. 2013). Host cells adjacent to infected tissue could use the energy and defense priming from sucrose accumulation to limit pathogen colonization. This pattern has been observed in the response to hemibiotrophic pathogens in rice and Arabidopsis (Engelsdorf et al. 2013; Gómez-Ariza et al. 2007) and is consistent with the differences in cell wall remodeling described previously.

    Several other carbohydrate and carbohydrate-related pathways followed similar patterns of accumulation. For example, trehalose is a simple carbohydrate that mediates plant defense responses in other plant–microbe interactions (Figueroa and Lunn 2016; Govind et al. 2016; Lunn et al. 2014) and was accumulated in inoculated resistant poplar tissue relative to controls (Figs. 2 and 5). This accumulation was not mirrored in the susceptible poplar experiment. In wheat, exogenous treatments of trehalose upregulated chitinase genes, PR protein 1, and oxalate oxidase, which promoted resistance to powdery mildew (Tayeh et al. 2013). The oxalate oxidase gene has been shown to increase disease resistance of transgenic poplar to S. musiva (Liang et al. 2001). Other carbohydrates can influence disease resistance in plants via the pentose phosphate pathway (PPP). Gluconolactone, a key component of the PPP, was accumulated in resistant poplar after inoculation (Figs. 2 and 5), and PPP was one of the most significantly enriched pathways in the resistant poplar experiment (Supplementary Fig. S2). The PPP supplies cells with NADPH, amino acid precursors, and vitamin B6 (pyridoxine) precursors (Stincone et al. 2015). A potential explanation for the enrichment of this pathway may be to replenish the depleted nucleotides, vitamins, and amino acids used by the resistant poplar. Alternatively, it could increase NADPH production, an essential cofactor in PAMP-induced ROS and disease signaling (Zeier 2013) or a combination of both.

    Amino acid fluctuations play critical roles in stress tolerance by reallocating nitrogen, supporting secondary metabolism for the production of disease resistance compounds and proteins (Dulermo et al. 2009; Hildebrandt et al. 2015; van Andel 1966). In the resistant poplar experiment, there was a general decline in all amino acids (Supplementary Table S1). The amino acids with the greatest reductions were the aromatic and branched chained amino acids. Branched chain amino acids (Val, Ile, Leu) are catabolized to produce acetyl-CoA for energy and to facilitate crosstalk between the SA and jasmonic acid pathways (Zeier 2013). The aromatic amino acids (Phe, Tyr, Trp) are secondary metabolite precursors involved in the production of flavonoids, plant hormones, and other plant compounds such as lignin and suberin (Fig. 3) (Bahadur et al. 2012; Grey et al. 1997; Parthasarathy et al. 2018). It is also likely that the amino acids are depleted for the production of proteins associated with changes in gene expression. For example, Muchero et al. (2018) noted a large difference in gene expression between resistant and susceptible genotypes, with 4,686 versus 76 differentially expressed genes 24 hpi, respectively (Muchero et al. 2018). In addition, the abundance of amino acids did not appear to change in the susceptible genotype except for the increase seen for lysine and tyrosine (Supplementary Table S1).

    Disease signaling by plant growth regulators.

    Plant growth regulators (PGRs) and hormones modulate plant metabolism and stress tolerance pathways (Denancé et al. 2013; Verma et al. 2016). Auxins including IAA are a major group of PGRs that elongate cells and promote root initiation but are also linked to ethylene in abiotic stress tolerance (Verma et al. 2016). Pathogens can also generate IAA to stimulate cell wall modifications in plant hosts to weaken physical barriers through the release of cell wall saccharides (Fu and Wang 2011). The main precursor to IAA is tryptophan via the IPA pathway (Sardar and Kempken 2018); however, tryptophan is also used for the production of phytoalexins and glucosinolates in disease resistance (Bednarek et al. 2009; Hiruma et al. 2013; Ishihara et al. 2008;). In the resistant poplar experiment, IPA and IAA concentrations did not change despite tryptophan being the most depleted amino acid after inoculation (Supplementary Table S1). In fact, the resistant genotype accumulated the IPA pathway inhibitor, L-kynurenine (Supplementary Table S1) (He et al. 2011). This indicates that tryptophan catabolism is not contributing to auxin accumulation in the resistant host but may be implicated in the production of antifungal compounds mentioned previously.

    Tryptophan metabolism has also been linked to the xanthurenic acid pathway. Xanthurenic acid was putatively coidentified as an isomer with zeanic acid in the resistant poplar dataset, whereas it was confidently identified independent of zeanic acid in the susceptible poplar dataset based on MS/MS confirmation. There was an accumulation of the putative xanthurenic/zeanic acid in the resistant poplar experiment, although they did not change in the susceptible poplar experiment (Supplementary Table S1). This observation could be explained by the comeasurement of its isomer, zeanic acid, in the resistant dataset, which would indicate an increase in zeanic acid in response to the pathogen. Interestingly, zeanic acid is a growth promoter that is probably derived from the oxidation of indole metabolites such as IAA (Matsushima et al. 1973; Sebanek et al. 2012; Takahashi 2018). Oxidation of the indole ring of IAA into zeanic acid may be used by the resistant host to combat excessive auxin accumulation during plant–pathogen interactions (Gao and Zhao 2014); however, it is difficult to draw any conclusions because it is unknown how auxin levels change through time as the fungus colonizes plant tissue, and we did not measure zeanic acid levels in the susceptible host. Additionally, the growth-promoting properties of zeanic acid have not been well characterized. These unknowns are interesting hypotheses to test in future research.

    Unlike auxins, cytokinins are a group of plant hormones that promote plant growth and resistance to pathogens. Cytokinins induce physiological changes such as stomatal closure, lignification, callose deposition, and defense gene upregulation (Albrecht and Argueso 2017; Didi et al. 2015). Zeatin biosynthesis, a plant cytokinin pathway, was enriched in the resistant poplar experiment, and multiple zeatin-associated metabolites were accumulated after inoculation (Fig. 2; Supplementary Fig. S2). Zeatin has been demonstrated to increase disease resistance against hemibiotrophic pathogens in Arabidopsis and tobacco (reviewed in Schäfer et al. 2015). It is likely that the resistant poplar uses cytokinins as signaling compounds in its disease response because these hormones are associated with SA and SAR (Verma et al. 2016). SA is also associated with disease resistance to rust pathogens in other poplar species (Ullah et al. 2019; Wei et al. 2019) and is known to be involved in poplar resistance to S. musiva (Abraham et al. 2019).

    ABA is a PGR associated with abiotic stress tolerance and also has inhibitory effects on plant growth and SAR (Lievens et al. 2017; Pieterse et al. 2009; Verma et al. 2016). The PLSDA identified ABA as one of the 25 metabolites with reduced abundance in the resistant poplar experiment and an increased abundance in the susceptible poplar experiment (Figs. 2 and 5; Table 1). In addition, the fungal growth experiment indicated that ABA was biosynthesized and secreted (Table 1). A potential explanation for this pattern would be that the fungus is able to suppress the SA and cytokinin signaling observed in the resistant poplar by producing and secreting ABA. Ethylene (ET) is another PGR that can inhibit SA signaling and is quantified by its precursor, 1-aminocyclopropanecarboxylic acid. Although there was no increase in the abundance of ET/1-aminocyclopropanecarboxylic acid in the susceptible genotype, its abundance was reduced after inoculation in the resistant poplar experiment. This pattern indicates that resistance to S. musiva appears to be mediated more strongly by cytokinin and SA signaling during the initial phase of infection rather than the JA–ET–ABA pathways typical of resistance to necrotrophic pathogens and herbivorous pests (Pieterse et al. 2009; Smith et al. 2009).

    Trigonelline is another growth regulator that works by halting the cell cycle in response to DNA damaging conditions (e.g., ultraviolet stress), salt stress, oxidative stress, and biotic interactions (Ashihara 2008; Minorsky 2002). Trigonelline was significantly depleted in the resistant genotype and was identified as one of the top 25 features differentiating inoculated and control plant tissues (Figs. 2 and 5). The resistant poplar may manipulate the abundance of trigonelline to limit G2 arrest, facilitating cell division for physical barriers and cell wall remodeling (e.g., NP). The role of trigonelline in resistance versus susceptibility is further illustrated by the accumulation in the susceptible poplar after fungal inoculation (Fig. 5).

    Fig. 5.

    Fig. 5. Signaling metabolites of interest grouped by physiological role with varying trends of accumulation within each experiment: inoculated resistant poplar, inoculated susceptible poplar, and spent media after Sphaerulina musiva growth. Relative fold change of each metabolite was calculated via log2 (treatment/control). Statistical significance is denoted with asterisks (P < 0.1 or partial least squares discriminant analysis variable importance in projection >1.0). All spent media metabolites were significant except for those labeled “ns.”

    Download as PowerPoint

    Metabolites of SAR.

    SAR is the distribution of disease response signals to neighboring plant cells and distal plant tissue, priming disease resistance pathways in those tissues (Ryals et al. 1996; Shah et al. 2014; Wenig et al. 2019). SAR is both SA dependent and SA independent. The results from the resistant poplar experiment demonstrated a strong induction of SA, SAR metabolites, and PAL pathway products after fungal inoculation (Fig. 3). Conversely, these metabolites were depleted in the susceptible genotype and catabolized by the fungus (Table 1; Fig. 5). These SA- and SAR-inducing metabolites are well-known initiators of plant defenses against biotrophic pathogens (Shah et al. 2014) and promote the production of receptors and PR proteins such as beta-1,3 glucanases and chitinases that inhibit fungal colonization (Ryals et al. 1996). These SAR-associated genes are upregulated in resistant poplar genotypes after S. musiva inoculation (Abraham et al. 2019; Foster et al. 2015; Muchero et al. 2018). Clearly, SA-mediated SAR plays a critical role in the initial defense response of poplar to S. musiva.

    Additional SAR-related compounds identified in the poplar experiments included azelaic acid (AzA), G3P, and pipecolic acid (Pip). AzA and G3P were more abundant in the resistant poplar experiment than in the control (Table 1; Fig. 5). AzA is a local response compound that can prime plants for SA and G3P accumulation and SAR signaling (Chanda et al. 2011; Jung et al. 2009; Mandal et al. 2011; Shah et al. 2014). G3P acts as a positive feedback stimulator for the production of SAR-signaling intermediates and other small molecules such as monoterpenes (Wenig et al. 2019). Pip induces SAR via SA-dependent and SA-independent pathways (Bernsdorff et al. 2016; Shah et al. 2014); however, it was depleted in both poplar experiments after inoculation (Table 1). This finding could be explained by the recent discovery that Pip-induced SAR depends on the hydroxylation of Pip to N-hydroxypipecolic acid (Hartmann et al. 2018; Shan and He 2018). It is possible that Pip was depleted because of its conversion to N-hydroxypipecolic acid. It would be beneficial to conduct a timepoint experiment to see whether Pip levels reaccumulate in resistant trees after the initial signaling event.

    In contrast to SAR, induced systemic resistance (ISR) is initiated by JA, ET, and ABA signaling. This signaling is typically induced by insect wounding, necrotrophic pathogens, beneficial microbes, and abiotic stress (Pieterse et al. 2009). Despite the divergence of these pathways, ISR (i.e., JA, ET, ABA signaling) and SAR (i.e., SA signaling) outcomes depend on intrapathway crosstalk to facilitate effective defense responses (Derksen et al. 2013; Li et al. 2019; Pieterse et al. 2009). In other poplar species, genes involved with both ISR and SAR signaling are expressed at different timepoints, probably corresponding to the switch from the biotrophic to necrotrophic phase of S. musiva (Foster et al. 2015). In the case of the P. trichocarpaS. musiva interaction, ISR-associated metabolites were secreted by S. musiva and increased in abundance in the susceptible genotype 24 hpi (Table 1; Fig. 5). These metabolites included the purine derivative allantoin, which can promote JA signaling and ABA production (Takagi et al. 2016; Watanabe et al. 2014; Witte and Herde 2020). These same metabolites were reduced in the resistant poplar, suggesting that a divergence from abiotic stress signaling is necessary for resistance during this stage of infection (Table 1; Fig. 5). In contrast, SAR-associated metabolites were consumed by the fungus, were less abundant in the susceptible poplar postinoculation, and accumulated in the resistant poplar after inoculation (Table 1; Figs. 5 and 6). These metabolic patterns in the susceptible poplar host infected with S. musiva coincide with the reduced expression of disease responsive genes (Abraham et al. 2019; Muchero et al. 2018).

    Stress signaling from JA, ET, and ABA can also lead to the accumulation of phosphorylcholine (i.e., phospholipid head group of cell membranes), which is a biomarker of autophagy in plants (Lenz et al. 2011; Liao and Bassham 2020; Mitou et al. 2009). This biomarker significantly accumulated in the susceptible poplar after inoculation, but the resistant poplar had the opposite reaction (Fig. 2). Interestingly, it has been shown that autophagy-deficient Arabidopsis mutants accumulate SA, defense-related proteins, and phytoalexins, resulting in increased resistance to biotrophic pathogens (Lenz et al. 2011). Biotrophic pathogens can manipulate plant autophagy for their benefit (Leary et al. 2018; Üstün et al. 2018), and S. musiva may be another example.

    Suppression of defense responses and weakened cellular structure in poplar trees are prerequisites for fungal mutualism and mycorrhizal colonization (Tschaplinski et al. 2014). Hemibiotrophic and biotrophic pathogens can also manipulate these mutualism pathways to promote disease (Plett and Martin 2018). S. musiva appears to mimic signals produced during abiotic stress and beneficial plant–microbe interactions (Fig. 6). Examples include benzoic acid and anthranilate (phenolic precursors), mevalonic acid and 1-hydroxy-2-naphthoate (terpene precursors), fumaric acid and 6-hydroxynicotinic acid (alkaloid precursors), S-adenosylmethionine (an ethylene precursor), oxalic acid (a nonspecific phytotoxin), and aminoadipic acid (a broad antibiotic precursor) (Supplementary Table S2). Although the cellular functions of these metabolites vary, they are all implicated in abiotic stress, senescence, or weakening cell wall apposition to promote wood colonization (Dhillon et al. 2015; Lehner et al. 2008; Lopes and Pupo 2011; Mandal et al. 2010; Usha Rani and Pratyusha 2014; Wang et al. 2002; Yasmin et al. 2012).

    Fig. 6.

    Fig. 6. Fungal metabolism of secondary metabolites associated with plant cell structure, cell signaling, and allelopathy.

    Download as PowerPoint

    The objectives of this study were to identify key signaling molecules and physiological markers associated with resistance and susceptibility to the hemibiotrophic fungal pathogen S. musiva. The resistant poplar trees perceive S. musiva and mount a rapid and sustained defense response that is characterized in this study by the accumulation of growth promoters, SAR metabolites, and elements of cell wall apposition. This is a prototypical response for resistance to biotrophic or hemibiotrophic pathogens. The response of the susceptible genotype appears to be manipulated by the fungus causing it to favor the ISR and senescence pathways commonly associated with beneficial microbes or resistance to necrotrophic pathogens, insect herbivores, and abiotic stress. These pathways inhibit plant growth and SAR. In addition, S. musiva appears to limit the defense response of susceptible P. trichocarpa by promoting abiotic stress responses and mutualism pathways while metabolizing or inhibiting compounds associated with SAR, cell wall apposition, and growth promotion.

    The author(s) declare no conflict of interest.

    LITERATURE CITED

    Funding: This study was supported by the U.S. Department of Energy (DOE) Office of Science, Office of Biological and Environmental Research (BER), grant DE-SC0018196 to J. M. LeBoldus; the DOE Office of Science, BER, grant DE-SC0018196; the Plant–Microbe Interfaces Scientific Focus Area in the Genomic Science Program; and the DOE BioEnergy Science Center project. The BioEnergy Science Center is a U.S. DOE Bioenergy Research Center supported by the Office of BER in the DOE Office of Science. Oak Ridge National Laboratory is managed by UT-Battelle, LLC for the U.S. DOE under contract number DE-AC05-00OR22725. S. S. Galanie recognizes support from the Center for Bioenergy Innovation, a U.S. DOE Bioenergy Research Center supported by the Office of Biological and Environmental Research in the DOE Office of Science. The work conducted by the U.S. DOE Joint Genome Institute, a DOE Office of Science User Facility, is supported by the Office of Science of the U.S. DOE under contract DE-AC02-05CH11231.

    The author(s) declare no conflict of interest.