Chorismate and isochorismate turnover reveal hidden dynamics of aromatic metabolism in Arabidopsis leaves Authors: Anya E Franks, Scott A Ford, Anya Hu, Christina De Melo, Sonia E Evans, Matthew E Bergman, Michael A Phillips Published: 2026 ABSTRACT AbstractThe shikimate pathway provides precursors for phenolic metabolites in plant primary and secondary metabolism. Carbon flux measurements in intact Arabidopsis leaves using an MS2 isotopologue methodology revealed unexpected dynamics between chorismate and isochorismate pools. 13CO2 labeling kinetics point to chloroplast-derived shikimate as a major bifurcation point, with approximately one third diverted from the chloroplast. In contrast, the small pool of chorismate was mostly chloroplast-localized and turned over rapidly, reaching more than 70% labeling within minutes. Isochorismate, a precursor to salicylic acid (SA) in the Brassicales, was present at 50-fold molar excess over chorismate but labeled more slowly and appeared primarily extrachloroplastic. Total isochorismate declined by 90% in the Arabidopsis “enhanced disease susceptibility” mutant, which lacks the isochorismate exporter. Nonaqueous fractionation further supported a primarily chloroplast-localized chorismate pool but extraplastidic isochorismate. Populus trichocarpa and Nicotiana benthamiana leaves contained chorismate but only trace isochorismate, consistent with use of the benzoyl-CoA route to SA. Carbon commitment calculations indicated that one third of the total chorismate pool in Arabidopsis leaves is diverted to isochorismate. Global leaf calculations based on elemental analysis-isotope ratio mass spectrometry and targeted isotope recovery indicate that only 0.03% of total assimilated carbon (5.27 pmol·mg−1 D.W.·min−1) enters the shikimate pathway in photosynthetic tissue of Arabidopsis thaliana. For comparison ∼0.05% (7.75 pmol·mg−1 D.W.·min−1) enters the 2-C-methyl-D-erythritol 4-phosphate pathway, which provides precursors for photosynthetic pigments and electron carriers. The compartmentalization and turnover dynamics of shikimate, chorismate, and isochorismate suggest continuous demand for aromatic precursors in mesophyll tissue is comparable to demand for 2-C-methyl-D-erythritol-4-phosphate pathway derived pigments. Introduction Chorismate is a central metabolic intermediate in plants, fungi, algae, and bacteria, linking primary metabolism to an extensive array of aromatic products (El-Azaz and Maeda 2025). As the product of the shikimate pathway (Maeda and Dudareva 2012), chorismate represents the last common precursor for aromatic amino acids (AAAs; Phe, Tyr, and Trp) and numerous downstream metabolites, including phenylpropanoids, lignin, salicylic acid (SA) (Rekhter et al. 2019), tetrahydrofolate, phylloquinone, and naphthoquinones (Fig. 1a). It is synthesized from D-erythrose-4-phosphate (E4P) and 2 equivalents of phosphoenolpyruvate (PEP) via sequential enzymatic steps of the shikimate pathway. In photosynthetic tissue, E4P derives from the Calvin–Benson–Bassham (CBB) cycle, while PEP is reimported from the cytosol by the PEP:phosphate translocator 1 (PPT1) (Streatfield et al. 1999) (Fig. 1b). The pathway begins with the condensation of E4P and PEP by 3-deoxy-D-arabino-heptulosonate-7-phosphate (DAHP) synthase (Yokoyama et al., 2022a, 2022b) followed by a rearrangement to dehydroquinate, dehydration to dehydroshikimate, and reduction leading to shikimate. This in turn is followed by phosphorylation to shikimate-3-phosphate and condensation with a second unit of PEP to 5-enolpyruvyl shikimate-3-phosphate (EPSP), whereupon chorismate synthase then dephosphorylates EPSP to yield chorismate (Rekhter et al. 2019). The dominant pathway to Phe in the plastid involves isomerization of chorismate to prephenate by chorismate mutase (CM), transamination to arogenate by prephenate aminotransferase, and dehydration to Phe by arogenate dehydratase. Cytosolic Phe is also thought to arise in flowering plants through a cytosolic CM isoform to generate prephenate, which is followed by prephenate dehydratase-mediated conversion to phenylpyruvate and transamination to Phe using Tyr as amine donor (Yoo et al. 2013; Qian et al. 2019). However, direct evidence supporting the cytosolic formation of a chorismate pool (Lynch 2022) has not been reported. Roles of chorismate in primary metabolism and analysis of its flux in plants by 13C-LC-MS/MS. a) Chorismate provides precursors for aromatic metabolites in primary plant metabolism. The contributions of phosphoenolpyruvate (PEP) (orange) and D-erythrose-4-phosphate (E4P) (green) to the carbon skeletons of downstream phenolic metabolites are highlighted. Hollow arrows signify multiple enzymatic steps. OP signifies a phosphate moiety. b) Schematic highlighting the trafficking and metabolism of phenolics and terpenoids in Arabidopsis mesophyll described in this study. Dashed lines indicate multiple steps. Abbreviations: RuBP, ribulose-1,5-bisphosphate; 3PGA, 3-phosphoglycerate; GAP, glyceraldehyde-3-phosphate; F6P, fructose-6-phosphate; S7P, sedoheptulose-7-phosphate; Xu5P, xylulose-5-phosphate; Pyr, pyruvate; DXP, 1-deoxyxylulose-5-phosphate; IDP, isopentenyl diphosphate; DMADP, dimethylallyl diphosphate; PhyQ, phylloquinone; PhytylDP, phytyl diphosphate; TPT, triose phosphate/phosphate translocator; PPT, phosphoenolpyruvate/phosphate translocator; GGDS, geranylgeranyl diphosphate synthase; GGDP, geranylgeranyl diphosphate; ICS, isochorismate synthase; IC, isochorismate; EDS5, enhanced disease susceptibility 5; PBS3, AvrPphB susceptible 3; IC-Glu, isochorismate-9-glutamate; EPS1, enhanced pseudomonas susceptibility; SA, salicylic acid; IP, isoprephenate; AAA, aromatic amino acids; cAAA, carboxy aromatic amino acids; MSB, meta-substituted benzoates. c-h) MS2 and MS3 spectra of chorismate and strategy for calculating fractional labeling in whole plant 13CO2 time-course experiments. All scans were performed using direct infusion of an authentic chorismate standard at 10 µg mL−1 (see Methods for details). The quasimolecular [M-H]− ion (m/z 225) is unstable and undergoes in-source dehydration. c) Product scan of the in-source dehydration product of chorismate (m/z 207) showing the major products (m/z 137, 179, 135, 163, 119). d) The 137 precursor scan confirmed that 207 is the exclusive precursor. e) The 207 → 179 MS3 scan, which implies loss of 28 Da from the 207 ion, is more likely the result of CO2 loss, O2 adduct formation, and loss of O. f) MS3 scans verified that the m/z 163 and 119 product ions are the result of successive CO2 losses from the m/z 207 precursor ion. g) The principal MS/MS fragmentation of chorismate, involving loss of the C3 enolpyruvate group (m/z 207 → 137) from the insource dehydration product of chorismate, was selected to quantify 13C label in 13CO2 labeled plant extracts. h) Isotopologue expansion series used to track chorismate labeling during CID fragmentation. This multiplexed MRM table was required to quantify the relative abundances of the chorismate isotopolog precursor ions. Because of the distribution of carbon atoms in (g) a total of 32 MRM transitions must be monitored simultaneously for exact calculation of isotopic labeling into chorismate and unambiguously account for all distributions of 13C atoms during CID. While no shikimate transporter has been identified to date, a portion is thought to be exported from the plastid to the cytosol, where it participates as a reusable esterification adaptor in monolignol biosynthesis. Hydroxycinnamoyl-CoA:shikimate/quinate hydroxycinnamoyltransferase esterifies p-coumaryl-CoA to p-coumaryl shikimate (Schoch et al. 2001; Hoffmann et al. 2003), which is hydroxylated to caffeoyl shikimate and subsequently re-esterified to CoA or hydrolyzed to caffeate (Vanholme et al. 2013). Phe exported from the plastid enters the phenylpropanoid pathway, initiated by phenylalanine ammonia lyase (PAL) (with contributions from tyrosine ammonia lyase in the Poaceae) (Barros and Dixon 2020). Aromatic plant natural products derived from the shikimate pathway passing through these enzymatic steps reportedly account for up to 30% of carbon assimilated in woody species (Razal et al. 1996; Boerjan et al. 2003; Maeda and Dudareva 2012). Although chorismate is a precursor to multiple downstream pathways, its predominant plastidial role is supplying Phe through the arogenate pathway (Vogt 2010). Much of this Phe is exported to the cytosol and converted to phenylpropanoids (Widhalm et al. 2015). In Petunia hybrida, chorismate conversion to Phe supports benzenoid and phenylpropene volatile biosynthesis (Qian et al. 2019). Cytosolic isoforms of CM have been identified in several plant species (d'Amato et al. 1984; Singh et al. 1986; Westfall et al. 2014), suggesting chorismate may be present outside of the chloroplast. Several other enzymes of the shikimate pathway have duplicated paralogs in the cytosol in one or more plant species, leading to the “dual pathway hypothesis” (Lynch 2022), which posits that portions of the shikimate pathway are active in both compartments. Labeling studies indicate a significant extraplastidic shikimate pool in Arabidopsis (Bergman et al. 2022), but analogous insights into chorismate trafficking remain lacking. A major downstream fate of chorismate is isochorismate, particularly in Brassicales, where it supplies SA, a central defense hormone. Chorismate is isomerized to isochorismate by isochorismate synthase (ICS) within the plastid, after which it is exported to the cytosol by the transporter ENHANCED DISEASE SUSCEPTIBILITY 5 (EDS5). There it is conjugated to Glu by AvrPphB Susceptible 3 (PBS3) and subsequently hydrolyzed to SA (Torrens-Spence et al. 2019). The ICS-EDS5-PBS3 pathway evolved specifically within a lineage of the Brassicales order (Hong et al. 2025). In plants outside of this lineage, the bulk of SA is produced through the PAL pathway through cinnamate-derived intermediates (Liu et al. 2025; Ma et al. 2025; Wang et al. 2025; Zhu et al. 2025). This route may also contribute to SA formation in Arabidopsis (Huang et al. 2010). However, feeding Arabidopsis with 13C-labeled benzoic acid produced labeled SA whereas 13C-labeled Phe feeding did not (Wu et al. 2023). Thus, the contribution of the PAL pathway to SA production in Arabidopsis remains unresolved. Despite this specialization of SA biosynthesis through the ICS pathway in the Brassicales, ICS homologs are conserved across land plants due to their essential role in phylloquinone biosynthesis in the chloroplast. Many species carry duplicated ICS paralogs, with varying functional divergence. In Arabidopsis ICS1 and ICS2 have overlapping functions, but ICS1 accounts for most SA production (Garcion et al. 2008). The ics1/ics2 double mutant lacks phylloquinone but retains trace SA, consistent with limited contribution from the PAL-derived pathway (Wu et al. 2023). Outside of the Brassicales, ICS isoforms have diverged functionally. In cotton, ICS1 produces isochorismate for phylloquinone while ICS2 supplies SA (Guo et al. 2023), whereas in rice (Wang et al. 2024) and barley (Qin et al. 2019), isochorismate is required only for phylloquinone biosynthesis, while SA is produced through the PAL route. The centrality of chorismate contrasts with how little is known about its turnover. Its instability, arising from a prearomatic ring structure and labile bonds, leads to rapid decomposition during MS analysis (Khera et al. 2017). Such fragmentation complicates quantitative assays of its metabolic role. Developing robust mass spectrometry methods would enable isotopic tracing, particularly with 13C, and thereby clarify flux through the shikimate pathway. Indeed, 13C labeling has become a standard tool in plant systems biology, improving models of primary and secondary metabolism (Ma et al. 2014; Wright et al. 2014; Xu et al. 2022). 13C labeling coupled to gas and liquid chromatography tandem mass spectrometry (GC-MS and LC-MS) has been instrumental in studying carbon flux in diverse metabolic networks (Szecowka et al. 2013; Ma et al. 2014; Xu et al. 2021; Bergman et al. 2022; Xu et al. 2024). Chorismate analysis by GC-MS is precluded by instability, unlike shikimate, whose isotopologues can be tracked with GC-MS using soft ionization (Bergman et al. 2022). Many central phosphorylated metabolites, such as intermediates of the 2-C-methyl-D-erythritol-4-phosphate (MEP) pathway, are more tractable in MS/MS because their fragmentation yields dominant phosphate (m/z 97) and phosphite (m/z 79) ions devoid of carbon, simplifying isotopologue calculations (Heise et al. 2014; Wright et al. 2014; Evans et al. 2022). Their isotopologue distribution can thus be quantified using multiple reaction monitoring (MRM) simply by varying precursor mass (Wright and Phillips 2014). In contrast, nonphosphorylated metabolites, including amino acids (Cocuron et al. 2017) and organic acids (Koubaa et al. 2014), often split carbon fragments during collision-induced dissociation (CID), complicating isotopic analysis. Chorismate, with its multiple labile bonds, exhibits diverse CID fragmentation (Khera et al. 2017). Systematic characterization of its MS/MS behavior would allow isotopologue deconvolution in 13C labeling experiments. We therefore analyzed its fragmentation by tandem MS to support flux studies into its synthesis and transport. In this report, we characterize the turnover and transport of chorismate and isochorismate in Arabidopsis by implementing a multiplexed LC-MS/MS MRM technique. We determined that a substantial portion of chorismate is diverted to isochorismate, which is exported from the chloroplast and remains largely metabolically inert in the absence of biotic stress. We further report that the allocation of carbon to the shikimate pathway is a small fraction of total assimilated carbon in photosynthetically active mesophyll tissue that is comparable in magnitude to carbon delivered to the MEP pathway for production of terpenoid pigments. Both precursor pathways combined consume <1% of total carbon, underscoring the metabolic specialization of photosynthetic cell types and the contrasting allocation of carbon resources needed to meet developmental and physiological demands. Results Resolving the cryptic fragmentation of chorismate enables flux measurements of the shikimate pathway We surveyed the labeling and turnover of shikimate and chorismate by 13C-LC-MS/MS to support flux and transport studies of the shikimate pathway in plant cells. Their fragmentation patterns were characterized by MS2 and MS3 to determine the best precursor/product pairs for monitoring label incorporation in 13CO2 treated plants. Shikimate undergoes a straightforward loss of CO2 and 2 water molecules during CID to form a resonance stabilized phenolate ion (m/z 173 → 93; Figure S1), and this fragmentation pathway has been previously reported to quantify both pool size and label incorporation (Jaini et al. 2017; El-Azaz and Maeda 2024). Chorismate, in contrast, displays more complex behavior. Under negative ESI conditions, the chorismate M-H− ion readily undergoes in-source dehydration to yield the aromatically stabilized 3-enolpyruvyl benzoate ion (m/z 225→207). During CID, the 207 precursor yields product ions of m/z 137, 179, 135, 163, and 119 in order of decreasing abundance (Fig. 1c). A precursor scan of the major product ion at m/z 137 confirmed the m/z 207 ion as its exclusive source (Fig. 1d), while the second largest (m/z 179) further fragmented to a product of m/z 135 (Fig. 1e). Additional MS2 and MS3 scans indicate that 2 successive CO2 losses from 3-enolpyruvyl benzoate account for the m/z 207→163→119 fragmentation sequence (Fig. 1f), ultimately forming a product whose identity is likely the phenoxyvinylate ion (Figure S1). The dominant transition (m/z 207→137) involves elimination of a 70 Da neutral fragment from 3-enolpyruvyl benzoate to yield 3-carboxyphenolate (Fig. 1g). The neutral loss may be explained through elimination of an acrylate zwitterion, which can rearrange to propiolic acid or, less likely, oxetone (Fig. 1g). The origin of the m/z 179 ion could, in principle, reflect concerted loss of CO2 and 2H· from the chorismate quasimolecular ion, [M-H-46]− (Figure S1e), leading to aromatization and retention of the phenol group. MS3 scans confirmed that the 225→179→153→109 sequence was readily detected, consistent with an initial loss of 46, corresponding to loss of CO2 and 2H·, followed by loss of acetylene and a second CO2, to produce 5-carboxyl-2-hydroxy phenolate (m/z 153) and 2-hydroxyl phenolate (m/z 109) (Figure S1). However, precursor scans confirmed that 3-enolpyruvyl benzoate (m/z 207), and not deprotonated chorismate (m/z 225), was the main precursor for the m/z 179 product ion (Figure S2a−c). A neutral loss of 28 implied by the 207 → 179 mass transition suggests loss of either carbon monoxide or ethylene, neither of which could be readily explained considering the likely neutral leaving groups of 3-enolpyruvyl benzoate. Based on similar observations in benzoic acid (Liu et al. 2024), we considered O2 adduct formation due to trace contaminants followed by loss of neutral O from the ensuing peroxide for a net loss of 28 Da (Figure S1d). This was confirmed with MS3 ion trapping scans, which suggested that the m/z 179 ion further loses CO2 (44 Da), followed by loss of ketene (42 Da) to form a phenolate ion (207→179→135→93; Figure S2e). Based on the relative signal intensities and structural diagnostic value of the various chorismate fragmentation pathways, the m/z 207→137 MRM transition was selected for isotope incorporation calculations to characterize flux and localization of the shikimate pathway in Arabidopsis leaves. Chorismate and shikimate turnover in Arabidopsis leaves suggest pathway branching and transport kinetics The dominant m/z 207→137 MS/MS fragmentation pathway of chorismate partitions carbon atoms between the neutral loss and product ions, complicating quantitative isotopologue analysis due to uncertainty in 13C distribution between fragments. To resolve this, we developed a multiplexed MRM strategy enabling complete isotopologue coverage for chorismate and its fragment ions. Using the full set of 32 MRM precursor/product labeling combinations derived from the monoisotopic 207→137 (C10→C7) model transition (Fig. 1h), we quantified 13C incorporation into chorismate and related polar metabolites in Arabidopsis leaves exposed to 13CO2 under controlled physiological conditions for 4.8 s—154.8 min prior to rapid freezing. The resulting time-resolved labeling curves exhibited hyperbolic kinetics for both PEP and chorismate, characterized by nearly identical apparent rate constants (k = 0.185 and 0.198 min−1, respectively), corresponding to half-lives of 5.41 and 5.06 min and plateau enrichments of 68.4% and 73.4%, respectively (Fig. 2 and Table S1). This similarity suggests a tight correlation between the labeled carbon entering chorismate and a rapidly equilibrating chloroplastic PEP pool. 13CO2 time-course labeling of whole Arabidopsis rosettes under physiological conditions. Fractional labeling curves for select intermediates in the shikimate and MEP phosphate pathways are shown. For rapid (hyperbolic) labeling metabolites, curve fitting was performed using the exponential equation, A × (1 – e[−k × t]), where A is the labelling plateau, k is the kinetic rate constant, and t is the labeling time. For slow (approximately linear) labeling metabolites, a linear regression model was applied. For additional curve fitting parameters, see Table 1. In contrast, shikimate displayed a low, linear rate of 13C incorporation under identical labeling conditions with a slope of 0.15% min−1 and a y-intercept of 1.05% (Fig. 2). When the same samples were analyzed by soft ionization GC-MS, the molecular ion cluster of the 4× TMS derivative of shikimate (m/z 480.2 to 487.2) gave nearly identical labeling results (Figure S3). This was consistent with previously reported values obtained with this method (Bergman et al. 2022) and validated our LC-MS/MS MRM method for quantifying isotopologues of aromatic intermediates. The intercept closely matched the expected natural 13C abundance of 1.07% for a C3 species, including 13C depletion (Bergman et al. 2021), validating both calibration and quantification procedures. The MRM transitions used to calculate isotopic enrichments in shikimate and other analyzed metabolites are listed in Table S2. Our analytical conditions were unable to quantify E4P in leaf extracts, the other central carbon intermediate entering the shikimate pathway, and we were only able to infer its uptake by comparing the labeling curves of PEP and shikimate. Control metabolites confirmed the sensitivity of this approach to biochemical turnover rate and compartmentation. Sucrose, representative of slowly exchanging cytosolic carbon pools, exhibited a linear and gradual incorporation of 13C (Fig. 2). Conversely, methyl-D-erythritol-2,4-cyclodiphosphate (MEcDP), an intermediate of the plastidic MEP pathway, showed rapid, pseudo first-order labeling kinetics consistent with its proximity to the primary site of 13CO2 assimilation (Evans et al. 2024). To account for instability of intermediates during pool-size calculations, we first conducted standard addition recovery assays by comparing peak intensities of authentic standards added to plant matrix either at the beginning of the extraction protocol or just before injection (Nasiri et al. 2021; Raposo and Barceló 2021; Williams et al. 2023). These values were then used to calculate recoveries and correct metabolite concentrations derived from external calibration curves. Recoveries for chorismate, MEcDP, and shikimate were calculated at 82.4%, 71.4%, and 63.1%, respectively, while PEP recovery was <4%, consistent with previously reported values (El-Azaz and Maeda 2024). The low recovery and instability of PEP precluded accurate assessment of its pool size by this approach, although this did not preclude determination of its labeling pattern. Labeling and pool-size data were next combined to estimate absolute carbon fluxes into shikimate and chorismate as a function of time (Fig. 3). Initial velocities for hyperbolically labeling metabolites were calculated from the linear portion of their exponential rise (0 to 8 min), while linear-labeling metabolites were analyzed directly by linear regression of their 13C incorporation slopes (Table 1). Comparing the initial velocity of carbon entering the shikimate pool (5.27 pmol 13C atoms·mg−1 D.W.·min−1) to the shikimate-derived portion (7 of 10 carbon atoms) entering chorismate (0.7 × 4.74 = 3.32 pmol 13C atoms·mg−1 D.W.·min−1; Table 1), we estimate that 63% of the carbon entering shikimate continues to chorismate. If we assume shikimate is not converted to chorismate in the cytosol, this implies that ∼37% of all shikimate exits the shikimate pathway prematurely, remains metabolically inactive, and likely resides outside the chloroplast, a figure that comports well with its slow, linear rate of labeling. Carbon commitment (CC) to central metabolic pathways compared with global carbon assimilation. For each metabolite, CC as nmol 13C atoms mg−1 DW was calculated by multiplying the fractional labeling at each time point by the average pool size and the number of carbon atoms. The overlay of all fractional labeling time-course curves (a) from Fig. 2 is provided as a reference. Initial slope for metabolites displaying nonlinear (b) or linear (c) labeling kinetics was obtained from the linear portions of the curves (0-8 min). Slopes of each metabolite were compared with the slope of total bulk tissue 13C enrichment, as determined by EA-IRMS (d), to obtain a dimensionless fraction estimating the portion of total carbon directed to a specific metabolite pool. e) A comparison of CC into shikimate and chorismate pools. CC into chorismate was derived from the linear portion of the CC curve (0-8 min) and multiplied by the fraction of its carbon atoms that are derived directly from shikimate (7/10). Table 1 13C incorporation rates into individual metabolite pools compared with total 13C in bulk leaf tissue following 13CO2 whole plant labeling.aPoolPool size (nmol ·mg−1)Initial slope (pmol 13C·mg−1 d.W.·min−1)% of total 13CNTotal 13C (EA-IRMS)n/a15,162.21 (714.07)10045MEcDP (MEP pathway)0.028 (0.0032)7.75 (0.43)0.05113Shikimate (E4P + PEP)0.491 (0.031)5.27 (0.21)0.03433Chorismate (E4P + 2PEP)0.005 (0.0004)4.74 (0.40)0.03116Isochorismate (E4P + 2PEP)0.134 (0.010)1.67 (0.07)0.01113Sucrose38.53 (4.13)1,181.88 (23.99)7.79528Initial slopes were determined from plants labeled for 0 to 7.82 min, during which 13CO2 incorporation was approximately linear for all metabolites surveyed. Percent of total 13C was determined by dividing the initial slope of each metabolite by the slope of 13C incorporation into bulk leaf tissue, as determined by EA-IRMS. Standard errors shown in parentheses. a Bergman et al. (2021) Plant Methods 17 (1), 32. 13C incorporation rates into individual metabolite pools compared with total 13C in bulk leaf tissue following 13CO2 whole plant labeling.a Initial slopes were determined from plants labeled for 0 to 7.82 min, during which 13CO2 incorporation was approximately linear for all metabolites surveyed. Percent of total 13C was determined by dividing the initial slope of each metabolite by the slope of 13C incorporation into bulk leaf tissue, as determined by EA-IRMS. Standard errors shown in parentheses. a Bergman et al. (2021) Plant Methods 17 (1), 32. When the incorporation rates derived from LC-MS/MS were normalized to the total 13C assimilation rate measured by EA-IRMS (Fig. 3c), the resulting ratios provided a measure of the fractional allocation of newly assimilated carbon through each metabolic pathway. This indicated that only ∼0.031% of total assimilated carbon completed the full shikimate pathway sequence to reach chorismate (Table 1), whereas 0.034% entered the shikimate pool, consistent with the inference that about a third of total shikimate enters a metabolically inactive or slowly recycled pool. Comparing this to the carbon allocation to an adjacent precursor pathway, analysis of the plastidial MEP pathway (using MEcDP as a representative intermediate) indicated that ∼0.05% of the total assimilated carbon was channeled into terpenoid precursor biosynthesis, or about 1.7 times the carbon observed entering the shikimate pathway. As a high-flux reference, the corresponding estimate for sucrose showed that during the middle portion of the day, about 8% of assimilated carbon (1,181.9 pmol 13C atoms·mg−1 D.W.·min−1) was diverted toward sucrose synthesis, consistent with its dominant role as the principal transport and storage carbohydrate (Figs. 2 and 3; Table 1). These results collectively imply that, in photosynthetically active leaf tissue, flux through the MEP pathway slightly exceeds that of the complete shikimate pathway but that the 2 precursor pathways are comparable in magnitude. This partitioning is physiologically plausible. The MEP pathway supports continuous synthesis and turnover of photosynthetic pigments derived from isopentenyl and dimethylallyl diphosphate on a timescale of hours (Beisel et al. 2010), whereas flux through the shikimate pathway must satisfy both protein production as well as vascular tissues (Ehlting et al. 2005) which prioritize Phe-derived lignification. A large, inactive isochorismate pool accumulates outside the chloroplast Chromatographic separation of polar metabolites revealed that the small chorismate peak detected in Arabidopsis leaf extracts was preceded by a much more abundant isobaric compound (Fig. 4a). It was ∼50-fold larger than that of chorismate and exhibited similar MS/MS fragmentation behavior (m/z 207→137). Although prephenate is isobaric with chorismate, it cannot undergo spontaneous aromatization by in-source dehydration. CID of a prephenate standard confirmed a distinct transition (m/z 225→163) (Figure S2f), corresponding to a concerted decarboxylation–dehydration event occurring after Q1 selection, consistent with observations by El-Azaz and Maeda (El-Azaz and Maeda 2024). This diagnostic transition was absent in the m/z 225 product ion scan of chorismate (Figure S2c), ruling out prephenate as the source of the major isobaric peak. We therefore hypothesized that the abundant isobar represented isochorismate, a positional isomer of chorismate and precursor of SA. Because a commercial standard was unavailable, isochorismate was synthesized in vitro using purified recombinant A. thaliana ISOCHORISMATE SYNTHASE 1 (ICS1, At1g74710) expressed heterologously and assayed with chorismate as substrate (Figure S4). The enzymatic product was analyzed by LC-MS/MS, and its retention time precisely matched that of the unknown isobaric peak in Arabidopsis leaf extracts, confirming the identity of the compound as isochorismate (Fig. 4g). Further validation was obtained using Arabidopsis ICS knockout lines. Chorismate pool size was significantly larger than wild type in the ics2 mutant (P < 0.005), consistent with the absence of the constitutively active ICS2. However, the decline in isochorismate levels was only statistically significant in the sid2 mutant (P < 0.01). In sid2-1 (ics1) and ics2 single mutant backgrounds, isochorismate molar excess over chorismate compared with wild type was reduced from 52.1 (±5.9) to 22.2 (±5.7) and 28.6 (±3.8), respectively (Fig. 4b, c, and j). This change in isochorismate:chorismate ratio was statistically significant (P < 0.0005 and 0.005 for ics2 and sid2-1, respectively, compared with wild type) (Fig. 4h−j). While the average isochorismate pool size decreased in each of the single knockout lines, it was not abolished entirely, consistent with the partial redundancy between the 2 ICS paralogs (Garcion et al. 2008). Detection of chorismate and isochorismate in whole leaf extracts by LC-MS/MS. Polar metabolite leaf extract of (a) wild-type Arabidopsis; b) sid2-1 mutant (SA INDUCTION DEFICIENT2); (c) ics2 mutant (ISOCHORISMATE SYNTHASE 2); d) P. trichocarpa leaf extract; e) N. benthamiana leaf extract; f) chorismate standard; g) enzymatically generated isochorismate standard (see Methods for additional details). The y axis represents the detector signal corresponding to the m/z 207→137 mass transition (counts per second) from the presumed m/z 225 pseudomolecular chorismate/isochorismate ion, which undergoes an ion source fragmentation. Orange triangle = chorismate; black triangle = isochorismate; h and i) chorismate and isochorismate pool size in ics2, sid2-1, and wild type (WT); j) pool-size ratios of isochorismate:chorismate in ICS mutants and wildtype. P-values represents a two-tailed Student’s t-test with n = 5 for ics2 and sid2, and n = 9 for WT. Retention time drift between batches was corrected with an authentic chorismate standard. We employed the same global isotope recovery and kinetic curve fitting approach to isochorismate, which exhibited a slow, linear-labeling pattern resembling that of shikimate (Figs. 2 and 3). This suggested that a large, metabolically stable isochorismate pool, likely outside the chloroplast, accumulates under basal conditions. We tested this idea with a time-course labeling series of the enhanced disease susceptibility 5 (eds5) mutant, which is defective in the isochorismate transporter (Nawrath et al. 2002). In the eds5-1 mutant line, isochorismate pool size was only ∼10% of the wild-type plant (Fig. 5a and b), consistent with a disruption of isochorismate export from chloroplasts. The linear-labeling rate of this highly reduced isochorismate pool in the eds5-1 mutant was nearly double that of the wild type (Fig. 5c), presumably reflecting the absence of the unlabeled isochorismate pool outside the chloroplast observed in wild-type plants. Metabolite profiling of the eds5-1 mutant and subcellular distribution of selected metabolic intermediates. a) Chorismate (yellow triangle) and isochorismate (black triangle) detection in eds5-1 and wild type (WT) leaf extracts by LC-MS/MS (y axis represents the detector signal corresponding to the m/z 207→137 mass transition (counts per second)). b) Isochorismate steady-state concentration in illuminated Arabidopsis rosette tissue from wild type and eds5-1 mutant plants. The latter is defective in the isochorismate translocator. c) 13C time-course labeling of the isochorismate pool in intact wild type and eds5-1 individuals. The trend line differences are highly significant (p = 3.8 × 10−12). Correlation matrices in (d) and (e) show Pearson correlation coefficients (r) for pairwise metabolite combinations in wild type (d) and eds5-1 (e) (see Materials and Methods for details). Correlations were determined from metabolite distributions across five fractions with densities ranging from <1.33 to 1.62 g/cm³. For consistency, only metabolites quantified in both wild type and eds5-1 mutants are shown. For an expanded list of metabolites, see Figure S5. The symmetric hierarchical clustered heatmap was generated using Euclidean distance and average linkage and depicts the averaged correlations of independent biological replicates (n = 3). Abbreviations: S7P; sedoheptulose-7-phosphate; DXP, 1-deoxy-D-xylulose 5-phosphate; MEcDP, 2-C-methyl-D-erythritol-2,4-cyclodiphosphate; UDP-glucose, uridine diphosphate glucose. To further investigate the subcellular distribution of shikimate, chorismate and isochorismate, we conducted nonaqueous fractionation (NAF) of wild-type and eds5-1 leaves (Fürtauer et al. 2016). We tested the hypothesis that chorismate is largely confined to the chloroplast while isochorismate is mostly stored outside that compartment following its export through EDS5. To evaluate performance of our NAF system, we quantified 30 metabolites across density fractions from wild-type plants (Table S3). These targets were verified with authentic standards and include 17 compartment markers metabolites previously validated by NAF studies employing protein or enzyme markers (Gerhardt and Heldt 1984; Benkeblia et al. 2007; Klie et al. 2011; Krueger et al. 2011; Szecowka et al. 2013; Arrivault et al. 2014) together with central metabolites of mixed or uncertain localization and the aromatic intermediates examined here (Figure S5). Fraction reproducibility was assessed using principal component analysis and hierarchical clustering (HC) (Krueger et al. 2011). HC of 3 biological replicates yielded a NMI score of 0.71 for the 5 expected density fractions (Figure S6a). This indicates high replicate reproducibility, comparable to other studies (Beshir et al. 2019). The complete metabolite panel revealed that several established marker metabolites exhibited overlapping distributions under our experimental conditions. For example, ADP-glucose and UDP-glucose, despite being associated predominantly with plastidial and cytosolic metabolism, respectively, displayed substantial correlation (Pearson r > 0.60), while thylakoid-associated pigments and prenylquinones formed a distinct cluster, likely reflecting separation from stromal material (Figure S5). Likewise, AAAs and glycolytic intermediates common to multiple compartments showed broadly distributed fractionation profiles, consistent with NAFalyzer predictions (Hernandez et al. 2023) (Table S4). To maximize compartment specificity for the metabolites central to this study, subsequent analyses were restricted to marker metabolites exhibiting robust and unambiguous fractionation behavior (Fig. 5d). Using this refined marker set, chorismate clustered strongly with MEP pathway and CBC cycle intermediates DXP, MEcDP, and S7P (r > 0.95). Linear regression analysis predicted that ∼60% of chorismate was plastid-associated. Conversely, isochorismate was predicted to reside primarily outside the chloroplast, with ∼82% assigned to vacuolar and cytosolic compartments (Fig. 5c and Table S4). Both metabolites showed weak or negative correlations with vacuolar organic acids succinate, fumarate, and malate (r 0.25 to −0.25), suggesting that neither accumulates substantially in the vacuole. Shikimate correlated strongly with isochorismate (r ∼ 0.93), moderately with MEP pathway intermediates (∼0.6), and was estimated to be ∼79% extraplastidial (Table S4), in agreement with its low, linear rate of 13C incorporation. Dual localization of shikimate is also consistent with previous assessments based on NAF and isotopic labeling (Szecowka et al. 2013; Bergman et al. 2022). Analysis of eds5-1 produced the same overall pattern for the principal metabolites despite greater variability between biological replicates (NMI = 0.62; Figures S6b and d). Chorismate remained strongly associated with the plastid markers MEcDP and DXP, whereas isochorismate showed increased correlation with chorismate (0.55 to 0.76), implying decreased presence outside the plastid when export via EDS5 was impaired (Fig. 5e). Interpretation of isochorismate was limited by its substantially reduced abundance (Fig. 5b). Overall, the NAF data are consistent with predominantly extraplastidial localization of isochorismate and chloroplastic localization of chorismate. Global CC reveals a large proportion of chorismate is converted to metabolically inert isochorismate To compare total CC into isochorismate and chorismate, we applied the external calibration curve constructed for chorismate to isochorismate based on the similarity in mass and fragmentation pattern. Comparison of total CC into chorismate and isochorismate indicated that approximately one third of the chorismate pool is diverted to isochorismate (Table 1), most of which appears exported out of plastids based on its labeling pattern and matching NAF results. These results are consistent with the interpretation that most extraplastidic isochorismate represents a pre-existing, slowly exchanging reservoir rather than a rapidly cycling intermediate. To assess whether such accumulation was unique to Arabidopsis, we compared isochorismate and chorismate levels in species known or suspected to synthesize SA via alternative routes, such as the benzoyl-CoA–mediated pathway (Liu et al. 2025; Ma et al. 2025; Wang et al. 2025; Zhu et al. 2025). In Populus trichocarpa and Nicotiana benthamiana leaves, isochorismate was detected only in trace amounts (Fig. 4d and e). These interspecific differences support the interpretation that Arabidopsis possesses an unexpectedly large extraplastidic isochorismate pool that potentially supplies SA production during pathogen challenge through the ICS pathway as well as other specialized routes of cytosolic aromatic metabolism (Scholten et al. 2024; Wu et al. 2024). Discussion Isotopic labeling in intact plants highlights branching and compartmentation in shikimate pathway metabolism The present work resolves several analytical and conceptual challenges surrounding the quantification of shikimate pathway flux and transport. As the product of the shikimate pathway, chorismate is the last common intermediate leading to diverse aromatic natural products in plants (El-Azaz and Maeda 2025). We implemented a multiplexed 13C-based LC-MS/MS framework to reconstruct its isotopologues unambiguously for fractional labeling calculations. Leveraging the characteristic fragmentations of chorismate, isochorismate, and shikimate, we were able to distinguish kinetic signatures of plastidial and extraplastidial carbon flow through this central aromatic pathway. Isotopic labeling strategies have previously been applied to studying the shikimate pathway and AAA biosynthesis in plants. El-Azaz et al. recently reported time-course labeling of intact plants to highlight differential regulation of the shikimate pathway between monocots and dicots (El-Azaz et al. 2023). Using multi-hour time-course labeling experiments, the authors demonstrated that allosteric regulation of DAHP synthase evolved in grasses to accommodate the inclusion of Tyr in addition to Phe for lignin biosynthesis in that clade. Tyr:Phe ratios were further explored with labeled AAAs in wheat, which found that upon fungal elicitation, Tyr was preferentially incorporated into downstream defense metabolites over Phe (Rypar et al. 2025). The novelty of our approach rests in the measurement of the rate of appearance of 13C label in any metabolite pool relative to the global 13C intake, providing a benchmark for comparing absolute CC by metabolite that is independent of its rate of consumption. The comparison of time-resolved, targeted flux measurements paired to EA-IRMS data further enables the identification of inactive metabolite stores and infers transport processes. Thus, these experiments point to not only the fraction of total assimilated carbon entering the shikimate pathway but also the proportion that completes the sequence to chorismate. Our calculations indicate that in Arabidopsis leaf about one third of shikimate carbon departs the canonical sequence to chorismate and presumably accumulates outside the chloroplast. In the cytosol, shikimate participates in monolignol biosynthesis by forming transient esters with p-coumarate during caffeoyl-CoA formation needed for guaiacyl and syringyl lignin formation (Schoch et al. 2001; Hoffmann et al. 2003; Hoffmann et al. 2004; Saleme et al. 2017), but is evidently not turned over further, resulting in slow exchange of isotopic label. This pre-existing, unlabeled shikimate ester pool, recycled during monolignol formation with minimal net flux (Vanholme et al. 2013), may account for the low, near linear rate of 13C incorporation observed in whole tissue homogenates, which combine this cytosolic pool with concurrent de novo shikimate synthesis in the chloroplast during 13CO2 labeling. The mechanism of shikimate export from the chloroplast is currently unknown. In contrast, chorismate appears largely localized to the chloroplast compartment based on its rapid, hyperbolic labeling pattern (half-life of ∼5 min) and high labeling plateau (∼70% enrichment in <10 min). This labeling pattern parallels that of CBB cycle intermediates observed during its early elucidation, which reached only 70% to 80% 14C enrichment when fed 100% 14CO2 (Sharkey 2019). This isotopic dilution was later shown to be due to unlabeled stores of pentose phosphates undergoing import during photosynthesis (Xu et al. 2022). The rapid formation and conversion of chorismate implies both formation and utilization close to the site of primary carbon fixation, which is inconsistent with the notion of a chorismate pool outside the chloroplast. While hyperbolic labeling and rapid turnover are characteristic of metabolites proximal to carbon fixation (Szecowka et al. 2013), these features alone are not diagnostic of chloroplast localization. For example, PEP labels rapidly despite being produced primarily in the cytosol from triose phosphate. Its fast labeling kinetics likely reflect reimport via PPT1 in mesophyll tissue (Streatfield et al. 1999), rapid consumption by DAHPS, and low steady-state abundance (Evans et al. 2024). However, the low recovery of PEP observed here suggests its pool size may be larger than originally estimated. Conversely, chloroplast-localized metabolites with high abundance and slow turnover, such as chlorophyll a/b and carotenoids may require hours to days for detectable labeling (Beisel et al. 2010; Banh et al. 2022), resulting in approximately linear label incorporation during short-term assays. Thus, labeling kinetics alone cannot resolve metabolite compartmentation. NAF remains the preferred methodology to determine subcellular metabolite localization. Although robust compartment assignment is typically supported by marker enzymes or proteins (Gerhardt and Heldt 1984; Arrivault et al. 2014), previous NAF studies and protocol papers have relied on starch as marker for the plastid compartment and nitrate and flavonoids for the vacuolar compartment (Geigenberger et al. 2011; Krueger et al. 2011; Beshir et al. 2019; Medeiros et al. 2019). In the present study, compartment assignment relied on a panel of well-studied reference metabolites whose distributions were previously established in landmark NAF studies using protein- and enzyme-validated compartment markers (Klie et al. 2011; Krueger et al. 2011; Szecowka et al. 2013). However, since our metabolite markers could not be corroborated by proteins in this study, we regard our NAF analysis as supportive rather than definitive evidence of compartmentation. Nevertheless, our initial marker set identified a subset of metabolites exhibiting robust compartment-specific fractionation, allowing confident discrimination between plastid-enriched and extraplastidial fractions. Within this framework, chorismate consistently associated with plastid marker metabolites, whereas isochorismate exhibited a contrasting distribution consistent with EDS5-dependent export from the chloroplast. Considered together with the independent evidence from 13C labeling kinetics and the eds5-1 mutant, these observations support predominantly chloroplastic localization of chorismate in leaves of a brassicaceous species such as Arabidopsis. This does not exclude that chorismate may play a functional role in the cytosol of plant species engaged in specialized shikimate-derived metabolism, such as benzenoid volatile production in petunia petals (Qian et al. 2019). While a parallel, cytosolic route to chorismate has been proposed (Lynch 2022), direct evidence in a model plant system remains lacking. In Arabidopsis, most shikimate pathway enzymes, including enolpyruvylshikimate phosphate synthase (AT2G45300 and At1g48860) and chorismate synthase (AT1G48850), are encoded by genes predicted to produce plastid-targeted proteins with no clear cytosolic paralogs. Although plastid-targeting does not preclude cytosolic localization through alternative transcriptional and translational start sites (Willems et al. 2021), such mechanisms, while well-established in terpenoid metabolism (Cunillera et al. 1997; Phillips et al. 2008; Ruiz-Sola et al. 2016), have not yet been demonstrated for shikimate pathway enzymes. Indeed, Willems et al. (2021) detected alternative translation products for terpenoid enzymes but not those of the shikimate pathway. Together, our data support predominantly chloroplast localization of chorismate in mesophyll tissue but dual compartmentation for shikimate, which partitions between plastidial and extraplastidic pools with distinct turnover kinetics. However, we cannot exclude that chorismate may play a role outside the plastid in heterotrophic tissues. For instance, various chorismate utilizing enzymes, some of which are predicted to localize to the cytosol, produce a series of related nonaromatic, isomeric compounds in the roots of Arabidopsis (Peng et al. 2026). Fluxes into aromatic and terpenoid precursor pathways are comparable in photosynthetic tissue The integrated labeling and flux analyses demonstrate that, in mature Arabidopsis leaves, the shikimate pathway operates at unexpectedly low flux, consuming only ∼0.05% of newly assimilated carbon in photosynthetic tissue. This challenges the notion that 30% of photosynthetically fixed carbon passes through Phe in plant systems generally (Haslam 1993; Razal et al. 1996; Boerjan et al. 2003; Qian et al. 2019) and suggests that carbon allocation into phenolic metabolic pathways depends on tissue type. While woody species (El-Azaz and Maeda 2025), lignin-rich vascular tissue, or tissue specializing in phenylpropanoid volatile production (Boatright et al. 2004; Maeda et al. 2010; Lynch et al. 2017; Qian et al. 2019) may indeed direct a substantial portion of fixed carbon to Phe biosynthesis, our results suggest that the fraction of total assimilated carbon dedicated to phenolic precursors in photosynthetic leaf tissues is well below this figure. This may reflect the diminished requirement for shikimate pathway products in mesophyll tissue, compared with the elevated need for Phe for lignin precursors in vascular tissue and stems (Ehlting et al. 2005), both of which consume large quantities of PEP for lignification (Evans et al. 2025). The large discrepancy in shikimate pathway flux between lignin-rich tissue and mesophyll tissue may depend on both biochemical factors and substrate supply. For instance, in Petunia x hybrida petals, downregulation of arogenate dehydratase, leading to Phe, unexpectedly resulted in an overall decrease in flux to shikimate (Maeda et al. 2010). Certain AAA isoforms are allosteric inhibitors of DAHPS, the first committed step of the pathway (Yokoyama et al. 2021). Elevated Tyr levels cause the cytosolic retention of Arabidopsis DAHP2 isoform through interaction with a 14-3-3 protein (Kanaris et al. 2022), possibly providing another means to reduce shikimate pathway flux in tissues where its requirement is reduced, although the details remain unclear. Recent progress toward understanding the regulation of the shikimate pathway may help explain the bifurcation of shikimate between compartments. Arabidopsis mutants defective in the TyrA gene encoding arogenate dehydrogenase display a stunted growth and reticulated leaf phenotype (de Oliveira et al. 2019). A genetic screening for TyrA suppressor mutants identified multiple lines mapping to DAHPS (Yokoyama et al., 2022a, 2022b). These suppressor of tyra (sota) mutants carry point mutations that abolish feedback inhibition by AAA, resulting in increased CO2 assimilation, elevated AAA levels, and higher flux through the shikimate pathway, as reflected by increased 13C labeling rates and expansion of the shikimate pool (Yokoyama et al., 2022a, 2022b). In terms of metabolic control theory (Fell 1997), DAHPS isoforms normally exhibit negative elasticities toward pathway end products, analogous to DXS in the MEP pathway, which is inhibited by IDP and DMADP (Banerjee et al. 2013; Wright et al. 2014; Mitra et al. 2021). Negative product elasticity at the first committed step is a common network feature that dampens fluctuations in metabolite pools while permitting modulation of pathway flux. Loss of this feedback in sota mutants removes a stabilizing constraint, increasing pathway flux and reducing concentration buffering, thereby allowing downstream metabolites such as shikimate to accumulate. These observations are consistent with a substantial, metabolically responsive chloroplast-localized pool of shikimate that can expand when flux constraints at DAHPS are relieved, even if a significant fraction of total cellular shikimate resides outside the chloroplast. Beyond regulation at the biochemical level, restriction of substrate supply may also directly impact flux. Reimport of PEP into chloroplasts is governed by 2 PEP:phosphate transporters (PPT1 and PPT2). PPT1 is broadly expressed in vascular tissues and roots, resulting in a distinct reticulated phenotype when mutated (Streatfield et al. 1999; Voll et al. 2003), whereas PPT2 is primarily expressed in mesophyll tissue but does not complement the PPT1 mutant phenotype (Knappe et al. 2003), also known as the chlorophyll a/b binding protein underexpressed (cue1) mutant. The cue1 phenotype has been speculated to derive from more than a simple lack of AAA (Voll et al. 2003), and the cell-specific requirement for high shikimate flux in vascular tissue for lignification, where PPT1 is highly expressed, is consistent with the notion that disruptions in phenylpropanoid availability for lignin biosynthesis, not AAA shortage for protein synthesis, is the source of the reticulated phenotype of cue1 (Staehr et al. 2014). Although the total CC to shikimate pathway is lower than expected, the total carbon investment into the MEP pathway is similar. Although we observed <1% of assimilated carbon entering the MEP pathway, our short-term labeling approach captures flux only through rapidly turning-over intermediates and does not resolve incorporation rates into different downstream end products, including photosynthetic pigments, prenylated lipids, and signaling metabolites (Ramel et al. 2012; Nisar et al. 2015). Isotopic labeling studies indicate that these compounds turnover on a timescale of hours (Beisel et al. 2010) to days (Banh et al. 2022), well beyond the timeframe of the precursor pathways described here. Future metabolic studies to extend the CC framework to functionalized terpenoid end products must overcome several physical limitations. Among them, their slow biosynthetic and turnover rates; the presence of large, pre-existing pools whose low natural abundance dilute early isotopic enrichment; and their structural complexity, particularly for chlorophyll, which obtains less than half its carbon from the MEP pathway. Consequently, quantitative studies to track flux into photosynthetic pigments using a similar framework will require both specialized labeling strategies and development of appropriate analytical methods. The cytosolic isochorismate pool may serve as a reservoir for defense-related carbon resources Shikimate and isochorismate, but not chorismate, highlight export of prearomatic intermediates in mesophyll cells. Extraplastidic isochorismate constitutes a large, metabolically inert reservoir (∼50-fold higher in magnitude than chorismate) with slow isotopic labeling kinetics, suggesting limited turnover under basal conditions. While the fraction remaining in the plastid participates in phylloquinone biosynthesis (Garcion et al. 2008), the extraplastidic pool may represent a previously unrecognized carbon reserve linking the shikimate pathway to inducible defense metabolism. High cytosolic isochorismate levels imply that pre-existing isochorismate could potentially provide an immediate substrate for SA biosynthesis during pathogen attack or abiotic stress. If accessible to cytosolic SA-forming enzymes such as PBS3 or other yet-unidentified N-glutamyl-isochorismate lyases, this pool could enable rapid SA accumulation within minutes, prior to de novo induction of biosynthetic enzymes. Moreover, SA accumulation is not PBS3/EPS1 dependent in Arabidopsis (Lefevere et al. 2020), highlighting previous observations that isochorismate-independent sources of SA are also important (Huang et al. 2010). However, ICS induction remains important to sustain and amplify SA production (Wildermuth et al. 2001). Transcriptional upregulation of ICS and EDS5 (Nawrath et al. 2002) would provide a secondary, sustained phase of SA biosynthesis over a timescale of hours to days (Yang et al. 2015; Zheng et al. 2015; Vlot et al. 2021). Mechanistically and evolutionarily, a 2-phase response (fast pool-driven production followed by transcriptionally reinforced synthesis) holds obvious advantages for the plant, and ICS induction alone does not exclude formation from pre-existing reserves of isochorismate in the cytosol. However, the question of whether SA biosynthesis from preformed precursors occurs independently of de novo production in Arabidopsis has not yet been addressed with the appropriate methodology (eg, isotopic labeling with pathogen induction at high temporal resolution). In a related species, rapid SA accumulation within minutes was observed in Brassica rapa in response to abiotic stress (Ang et al. 2024), and the kinetics suggest that formation from pre-existing substrate/enzyme pools is at least feasible. In addition to its role as precursor to SA, isochorismate serves as a substrate for cytosolic aromatic metabolism in Arabidopsis, including the formation of O-substituted benzoates involved in defense (Scholten et al. 2024), and alternative AAA pathways (Wu et al. 2024). In the latter, the cytosolic aromatic aminotransferase, REVERSAL OF SAV3 PHENOTYPE 1 (VAS1), channels isochorismate, following its conversion to isoprephenate, into carboxylated AAA intermediates (eg, 3-carboxy-4-hydroxyphenylpyruvate or 3-carboxylphenylpyruvate), which are further transaminated to 3-carboxy-Tyr and 3-carboxy-Phe (Wu et al. 2024). These pathways rely on EDS5-dependent export of isochorismate to the cytosol but diverge downstream of isoprephenate, illustrating their role as precursors in cytosolic metabolism of aromatic compounds. Together, these findings support an expanded function for cytosolic isochorismate in both defense and modulation of AAA biosynthesis. Comparative analyses across species further indicate that this feature may be specific to Arabidopsis and closely related taxa employing the ICS-dependent route to SA (Hong et al. 2025). Together, these findings provide a revised view of the shikimate–chorismate–isochorismate axis as a dynamically regulated interface between plastidial and extraplastidic branches of aromatic metabolism and defense signaling. Their distinct kinetic signatures and carbon allocation patterns furthermore highlight the previously unrecognized importance of isochorismate outside the chloroplast in Arabidopsis leaves. Tandem mass spectrometry combined with structurally diagnostic CID fragmentation enabled quantitative tracking of chorismate isotopologues and their phenolic precursors and derivatives, providing a robust framework for studying aromatic metabolism. Isotopic flux analyses in Arabidopsis plants exposed to 13CO2 revealed that chorismate synthesis and turnover are likely confined to the chloroplast, whereas shikimate and isochorismate are exported and maintain substantial extrachloroplastic pools. Notably, cytosolic isochorismate accumulates to high levels and turns over slowly, suggesting a role as a latent carbon reservoir linking primary metabolism with inducible defense pathways. Follow up studies will focus on the roles of these poorly understood aromatic precursors at the interface of primary and secondary metabolism. Materials and methods Plant cultivation and isotopic labeling A. thaliana (ecotype Columbia 0) wild type or mutant seeds were introduced to moist BX soil (Promix) in 5 cm pots and incubated for 3 d at 4 °C in the dark, before being transferred to short day conditions (8/16 h light/dark cycles at 150 μmol photons·m−2·s−1, 21 °C, and 60% to 70% relative humidity). Plants were watered as needed and fertilized once weekly with MiracleGro™ (20/20/20) according to manufacturer's instructions. Rosette stage Arabidopsis plants 6 to 8 wk old were subjected to gas exchange measurements and 13CO2 in vivo labeling. This was accomplished with a hybrid LI-COR 6800 photosynthesis system fitted with a custom cuvette adaptor and an environmentally controlled dynamic flow chamber (Evans et al. 2022). Photosynthetically active surface area of plants was determined in Photoshop by comparison to size standards. Air flow containing CO2 at 400 μL·L−1 was passed over the enclosed rosette at 680 μmol air·s−1 (∼1 L·min−1 at 21 °C) using the LI-COR console until the assimilation rate had stabilized, indicating a photosynthetic steady state (∼30 min). Labeling was initiated by rapidly switching from air supplied by the LI-COR 6800 to a pressurized gas tank (Linde Canada) containing synthetic air (80% N2, 20% O2) supplemented with 13CO2 at 400 μL·L−1. Relative humidity of the synthetic air was ensured by passage through a wash bottle prior to reaching the plant, and the exposure time of each time-course plant was monitored with a stopwatch until quenching by liquid nitrogen. The labeling time was corrected to the half-way point of the 2 atmospheres, as judged by the decay of the 12CO2 gas analyzer signal as 13CO2-containing air entered the chamber, usually in the range of 5 to 10 s. A total of 38 wild-type plants were labeled from times ranging from 5 s to 155 min. Arabidopsis mutant lines were obtained from the Arabidopsis Biological Resource Centre (https://abrc.osu.edu/). This includes mutant lines of ISOCHORISMATE SYNTHASE 2 (ics2) (SALK_084635; AT1G18870), ISOCHORISMATE SYNTHASE 1 (sid2-1; AT1G74710), (Wildermuth et al. 2001), and ENHANCED DISEASE SUSPECTIBILITY 5 (eds5-1; AT4G39030). Genotyping primer sequences are listed in Table S5. N. benthamiana were grown under long day conditions (16/8 h light/dark cycles) in environmentally controlled chambers with 300 μmol photons·m−2·s−1, 25 °C, and 60% to 70% relative humidity. P. trichocarpa leaves were harvested on the University of Toronto—Mississauga campus at midday in July 2025 and immediately flash frozen in liquid nitrogen for processing. Plant processing and extraction Flash frozen, labeled plant tissue (or unlabeled control) derived from a single labeling experiment was ground to a fine powder in liquid nitrogen and lyophilized to dryness against a vacuum of 25 μbar for 24 h. The resulting powder was stored in a 2 mL tube at −20 °C until extracted. All labeled plant samples were analyzed by elemental analysis-isotope ratio mass spectrometry (EA-IRMS) analysis as previously described to obtain the 13C/12C ratio and carbon content of bulk tissue (Bergman et al. 2021). To extract polar metabolites for LC-MS/MS analysis, the sample was kept at 4 °C for all steps of the procedure. A 15 mg aliquot of dry tissue was weighed out on an analytical balance, extracted twice with 50% acetonitrile containing 10 mM ammonium acetate pH 9.0, and centrifuged at 16,000 × g for 10 min after each extraction. An internal standard (200 ng deoxyglucose-6-phosphate) (Millipore-Sigma) was added at the beginning of the protocol. The extracts were lyophilized overnight, resuspended in 100 μL aqueous 10 mM ammonium acetate pH 9.0, and back extracted against 1 vol CHCl3. Following centrifugation at 16,000 × g, the upper aqueous phase was diluted with 1 vol acetonitrile, filtered through a 0.2 μm PVDF InnoSep™ Spin filter (Canadian Life Science), and transferred to a 2 mL glass LC vial with a polypropylene insert. For quantitation of MEP and shikimate pathway intermediates, standard curves were prepared using the protocol above with differing amounts of authentic standards in solvent only while maintaining internal standards constant (Figure S7). To infer matrix effects, standard curves were then repeated in the presence of 15 mg control tissue with standards added at the beginning of the protocol using the same levels. For comparison, a calibration series was prepared in parallel where standards were added just prior to injection (Figure S8). The results were compared to calculate recovery, matrix effects, and ion suppression using previously reported methods (Nasiri et al. 2021; Raposo and Barceló 2021; Williams et al. 2023). Sucrose analysis was carried out on 2 mg lyophilized tissue extracted with 500 μL 75% methanol (v/v) containing the same internal standards, centrifuged, and filtered as above. Concentration and isotopologue profile were analyzed by LC-MS/MS as described in Table S2. For GC-MS analysis of isotopically labeled plant tissue, 75% (v/v) methanolic extracts of a 5 mg tissue aliquot were dried on a nitrogen evaporator, resuspended in 100 μL pyridine, and converted to their methyloxime/trimethylsilyl (MeOX/TMS) derivatives in a 2-step sequence as described previously (González-Cabanelas et al. 2015). To calculate label incorporation into shikimate by GC-MS, MeOX/TMS derivatized extracts were analyzed by ammonia positive chemical ionization as described previously (Bergman et al. 2022). Liquid chromatography—mass spectrometry analysis LC-MS/MS targeted metabolite analysis reported was performed on an Agilent 1290 series II ultrahigh pressure LC, coupled to a Sciex 4500QTrap linear ion trap-triple quadrupole mass spectrometer. All metabolites described in this study were separated by hydrophilic interaction chromatography (HILIC) using an XBridge Premier BEH amide column (2.5 μm particle size, 2.1 mm × 150 mm; Waters Corporation) based on González-Cabanelas et al. (2016) with minor modifications. The LC mobile phases consisted of 20 mM ammonium bicarbonate, adjusted to pH 10.5 with NH4OH (A) and 80% acetonitrile with 20 mM ammonium bicarbonate pH 10.5 (B). The flow rate was constant at 0.5 mL·min−1. The solvent gradient was as follows: 0% to 16% A (0 to 5 min), 16% to 40% A (5 to 10 min), hold 40% A (10 to 15 min), and then re-equilibrate at 0% A (15 to 30 min). 13C-labeled isotopologues of intermediates from the MEP and shikimate pathway detected in plant extracts were quantified by MRM using electrospray ionization (ESI) in negative mode. Detailed data acquisition parameters, including mass transitions of isotopologue series for all analytes in this study, are listed in Tables S2 and S5. The detection of analytes was optimized by compound optimization with purified standards to establish optimal ionization and fragmentation energies, and product and precursor scans were performed to select the best Q1/Q3 pairs for overall sensitivity and linearity. CID of chorismate was investigated by direct infusion of a purified standard (Millipore-Sigma, ≥98% purity) dissolved in a mixture of solvents A and B. The analyte was ionized in negative ESI mode (−4,500 V, 200 °C temperature). Precursor and product ion scans were acquired for the deprotonated parent ion ([M–H]−, m/z 225) and for the dehydrated in-source fragmentation product (m/z 207). To confirm structural identity of product ions, fragmentation pathways were further characterized with MS2 and MS3 using the linear ion trap function of this instrument, focusing on major product ions (m/z 179, 163, 153, 137, and 135). Ion trap experiments using purified chorismate were performed using −16 V collision energy, dynamic fill time, and a mass range of m/z 50 to 200. Spectra were averaged over 3 replicate scans to improve the signal-to-noise ratio. All experiments were performed in triplicate to ensure reproducibility. These experiments defined the principal fragmentation routes of chorismate under negative CID, providing mechanistic insight into its MS behavior needed for its targeted detection in complex biological extracts. Calculation of fractional 13C enrichment and isotope recovery Fractional labeling (F) of PEP and MEcDP were carried out as previously reported (Evans et al. 2024). Fractional labeling of chorismate and shikimate was determined using a newly developed multiplex MRM approach that accounts for all possible isotopologues and their distributions into product and neutral loss fragments during CID. When carbon atoms are retained in both the product ion p and neutral loss l following CID, the total transitions that must be monitored for an exact calculation of fractional labeling (F) must take into consideration all possible product ions that can arise from a given isotopolog precursor, based on physically possible distributions of 13C atoms between the product ion and neutral loss fragment. Because both the fragment ion and the neutral loss retain subsets of the total carbon skeleton in the case of chorismate, each isotopologue of the precursor (M + 0 to M + 10) can produce multiple product ions with distinct isotopic compositions. For chorismate, the m/z 207 → 137 transition was selected to analyze label incorporation in Arabidopsis leaves. All 10 carbons of the original chorismate molecule are present in the m/z 207 ion, while the product contains 7 carbons. A total of 32 unique transitions were required to account for all possible distributions of 13C in the 11 chorismate isotopologues (Fig. 1h). By summing the integrated peak areas of the product ion isotopologues derived from each precursor isotopologue, we calculated relative abundances of each isotopologue. Following normalization to the monoisotopic peak, calculation of fractional 13C enrichment in the chorismate pool of 13CO2 labeled plants was accomplished using the following relationship: where N is the number of carbons of the compound of interest, and Mi is the fractional abundance of the ith isotopologue. To calculate global carbon commitment (CC) and pathway specific carbon allocation, we first computed the CC for each individual plant at each time point into each metabolite, defined as the nmol of 13C atoms in each metabolite pool and calculated as follows: F represents the fractional labeling of that individual sample and n is the number of carbons in the metabolite. We next established the linear regression for carbon assimilation into each metabolite pool surveyed here using time-course data. The initial slopes of hyperbolic labeling curves for chorismate, PEP, and MEcDP were inspected for linearity, and time points ranging from 0 to 8 min presented R2 values of 0.97, 0.88, and 0.91, respectively. By plotting CC against time for the linear portion, we obtained their initial slopes. Once all slopes had been converted to the common units of pmol 13C atoms·mg−1 D.W. ·min−1, we obtained percent-wise commitment of assimilated carbon to each metabolite pool by dividing each slope by the total 13C assimilation slope derived from EA-IRMS analysis of all time-course labeled plants. The resulting dimensionless fraction represents the proportion of all assimilated carbon entering that metabolite pool during time-course labeling experiments. Enzymatic production of IC To prepare an IC standard for LC-MS/MS analysis, the Arabidopsis ICS1 gene (AT1G74710) was PCR amplified without a chloroplast targeting peptide using Gateway attB primers (Figure S4 and Table S5). The PCR amplicon was cloned into the pDONR207 entry vector with BP Clonase II (InvitrogenTM) and then transferred to a GatewayTM modified pET28 vector (Yu and Liu 2006) for bacterial protein expression fused to a His9 affinity tag using LR Clonase II. Chemically competent BL21 (DE3) Escherichia coli cells were transformed with the expression vector containing the truncated ICS construct and plated on Luria-Bertani agar media supplemented with kanamycin (50 μg·mL−1) for overnight incubation at 37 °C. Positive colonies were selected by colony PCR, grown in 5 mL starter cultures overnight at 37 °C, 220 RPM, and each used to inoculate a separate 200 mL culture of Terrific Broth with the selective antibiotic, kanamycin (50 μg·mL−1). Cultures were incubated under previous conditions until an optical density at 600 nm (OD600) of 0.6 to 0.8 was achieved, after which cultures were induced for protein expression with 1 mM isopropyl β-D-1-thiogalactopyranoside, and the temperature reduced to 16 °C for an incubation period of 18 to 20 h at 230 RPM. Cells were harvested by centrifugation at 4 °C and 1,000 × g for 30 min, after which the cell pellets were resuspended in 10 mL of lysis buffer (phosphate buffered saline, 60 mM imidazole, 1 mg·mL−1 lysozyme) with a protease inhibitor cocktail mix (Sigma-Aldrich®) used according to the manufacturer's instructions. The resuspension was incubated on ice for 30 min, subjected to 2 freeze-thaw cycles, and sonicated at 60% power for 6 cycles of 10 s power and 10 s of an intermittent cooling period. The suspension was clarified by centrifugation for an hour at 4 °C, 4,000 × g, and the supernatant purified by Ni-NTA agarose (QIAGEN) chromatography as per the manufacturer's instructions. Eluent was desalted with desalting buffer (50 mM Tris-HCl pH 8.0, 10 mM MgCl2, 150 mM NaCl, 10% glycerol) using an Amicon® ultra centrifugal filter (10 kDa MWCO; Millipore), and the protein concentration estimated using a standard Bradford assay. Assays were performed in duplicate 100 μL reactions with 0.8 μg of ICS1 per assay in desalting buffer supplemented with 10 μM chorismate and incubated at 25 °C for an hour. Reactions were terminated by vortexing with 1 vol of CHCl3, centrifuged for 10 min at 10,000 × g; after which 50 μL of the aqueous supernatant was diluted with 50 μL of acetonitrile and transferred to a 2 mL LC vial with a polypropylene insert for analysis by LC-MS/MS. Nonaqueous fractionation Subcellular metabolite distribution was analyzed by nonaqueous fractionation using the benchtop protocol of Fürtauer et al. (2016). Approximately 25 mg of finely ground lyophilized tissue was resuspended in 2 mL chilled heptane:tetrachloroethylene (TCE) (0.484:1, v/v; ρ = 1.3 g/cm3) and further homogenized on ice using a Tenbroek homogenizer. The homogenate was transferred to a 50 mL tube, and the homogenizer was rinsed with 8 mL of the same solvent, which was pooled with the sample. The suspension was sonicated in an ice-water bath for 10 min, filtered through a 20 μm nylon mesh, and the filter washed with 10 mL cold heptane. The filtrate was centrifuged (20 min, 4,000 × g, 4 °C), the supernatant discarded, and the pellet resuspended in 1 mL TCE and transferred to a 2 mL tube. Following an additional 10 min cold sonication, samples were centrifuged (12.5 min, 16,000 × g) and the pellet retained to yield the densest fraction. The supernatant was sequentially diluted with heptane to decrease density, followed by sonication (10 min) and centrifugation (12.5 min, 16,000 × g) at each step to generate pellets of decreasing density relative to the initial fraction. This iterative process produced fractions corresponding to densities of approximately 1.33, 1.41, 1.49, and 1.61 g/cm3 (F2-F5; Figure S6). Each pellet was resuspended in 900 μL heptane:TCE adjusted to its respective density, split into 2 aliquots, washed with 1 mL heptane, and centrifuged (12.5 min; 16,000 × g). Pellets were dried under nitrogen and stored at −20 °C. After collecting the lowest-density pellet (F2; ρ = 1.33 g/cm3), the remaining supernatant, containing material less dense than 1.33 g/cm3 (“F1”) was similarly split into 2 equal aliquots, dried under nitrogen, and stored at −20 °C prior to metabolite analysis. For polar metabolite analysis of NAF samples, dried fractions were resuspended in 100 μL cold ammonium acetate (pH 9.5), extracted with an equal volume of chilled chloroform, vortexed (20 min, 4 °C), and centrifuged for phase separation. A 50 μL aliquot of the aqueous phase was mixed with 50 μL cold acetonitrile and analyzed by HILIC LC-MS/MS as described above. For organic acids, samples were diluted 1:100 in acetonitrile:ammonium acetate (1:1, v/v; pH 9). Undiluted fractions were also analyzed by reverse phase (RP) LC-MS/MS in ESI mode on an Eclipse XDB-C18 RRHT (1.8 μm particle size, 4.6 mm × 50 mm; Agilent Technologies) column. Reverse mobile phases consisted of 0.1% formic acid in water (A) and 0.1% formic acid in acetonitrile (B). At a flow rate of 0.8 mL·min−1, the separation gradient was as follows: 99.5% A (0 min), decreasing to 20% A over 10 min, then to 1% A from 10 to 12 min followed by a min hold at 1% A and finally, re-equilibration of the column (99.5% A) from 13 to 20 min. For pigment analysis, the chloroform fractions were saved and dried under nitrogen, resuspended with 100 uL of cold acetone, and analyzed on the 4500QTrap in atmospheric chemical ionization (APCI) mode following separation on a YMC C30 Carotenoid column (3 μm particle size, 3 mm × 150 mm; YMC America). Mobile phases involved 0.5% formic acid in water (A) and methanol:ethylacetate (65:35) (B), as previously described (Petibon and Wiesenberg 2022). The flow rate was 0.4 mL·min−1, and the solvent gradient was as follows: 15% A (0 min), decreasing to 0% A by 15 min, holding at 0% A until 28 min, then returning to 15% A from 28.01 min to 40 min. MS parameters for the RP-ESI and RP-APCI are further outlined in Table S6. Relative metabolite distributions (n = 3 biological replicates) were calculated as the percentage of total peak area across the 5 fractions. Statistical analyses of metabolite data were performed using MetaboAnalyst 6.0 and Python scripts utilizing the scikit-learn (v1.6.1), and scipy (v1.16.3) libraries. PCA (autoscaled data) and hierarchical cluster analysis (HCA; Euclidean distance, average linkage) were used to assess fraction resolution and reproducibility (Beshir et al. 2019). HCA dendrograms were generated using Seaborn (v0.13.2) and matplotlib (v3.10.0) in Python. Clustering performance was quantified using the normalized mutual information (NMI) score calculated for a fixed 5-cluster model (corresponding to the 5 fractions) using the metrics.normalized_mutual_info_score function in scikit-learn. Pairwise Pearson correlation coefficients (r) were calculated for metabolite distributions across the 5 density fractions (1.61, 1.49, 1.41, 1.33, and <1.33 g/cm3), averaged across replicates, and visualized as a symmetric hierarchical clustermap (Euclidean distance, single linkage). Correlation strength and clustering pattern were used to infer subcellular co-localization. Compartmental distributions (chloroplast, cytosol, and vacuole) were estimated using NAFalyzer (Hernandez et al. 2023). Plastidic, cytosolic, and vacuolar markers were defined as the mean relative distribution of selected compartment-specific metabolites (Table S3). The chloroplast markers included DXP, S7P, MEcDP, and 2-phosphoglycolate; the cytosolic markers consisted of UDP-glucose, gibberellin A3, HMG-COA, and SA; and the vacuolar markers included malate, fumarate, succinate, and citrate (Table S3). Outliers were handled as described by (Krueger et al. 2011). NAFalyzer output is provided in Table S4. Data analysis Sciex OS (v2.0) and Agilent MassHunter Qualitative Analysis (v10.0) were used to analyze LC-MS/MS and GC-MS data, respectively. Analyte peaks were identified by matching retention times and masses to authentic standards, then peak areas were determined by integrating extracted ion chromatograms. Absolute quantification of metabolites was determined by normalizing peak areas to internal standards and comparison to an external calibration curve of authentic standards. Recovery of internal standard was calculated by comparing DGP peak areas in nontissue controls to DGP recovered in Arabidopsis leaf extracts (15 mg wild-type Arabidopsis tissue) prepared under the same conditions. To assess statistical significance, a 2-tailed Student's t-test was used with a significance threshold set to P < 0.01. Time course-labeling curves were generated using 13C mass spectral data as described elsewhere (Evans et al. 2024). Supplementary Material Acknowledgments The authors acknowledge generous NSERC PGSD and CGSM graduate scholarships supporting AEF and SAF, respectively, and a CGSD graduate scholarship supporting MEB. The authors further recognize the contributions of Iggy for her many salient insights. Finally, the authors thank Keenan′s pub for their substantive and tasteful renovations. Contributor Information Anya E Franks, Department of Cell and Systems Biology, University of Toronto, 25 Harbord Street, Toronto, ON, Canada, M5S 3G5. Scott A Ford, Department of Cell and Systems Biology, University of Toronto, 25 Harbord Street, Toronto, ON, Canada, M5S 3G5. Anya Hu, Department of Biology, University of Toronto—Mississauga, 3359 Mississauga Rd, Mississauga, ON, Canada, L5L 1C6. Christina De Melo, Department of Biology, University of Toronto—Mississauga, 3359 Mississauga Rd, Mississauga, ON, Canada, L5L 1C6. Sonia E Evans, Department of Cell and Systems Biology, University of Toronto, 25 Harbord Street, Toronto, ON, Canada, M5S 3G5. Matthew E Bergman, Department of Cell and Systems Biology, University of Toronto, 25 Harbord Street, Toronto, ON, Canada, M5S 3G5. Michael A Phillips, Department of Cell and Systems Biology, University of Toronto, 25 Harbord Street, Toronto, ON, Canada, M5S 3G5; Department of Biology, University of Toronto—Mississauga, 3359 Mississauga Rd, Mississauga, ON, Canada, L5L 1C6. Author contributions S.A.F. and A.E.F. performed main experiments. A.H., C.D.M., and S.E.E. performed minor experiments. M.A.P. wrote the manuscript and directed the research. Supplementary material Supplementary material is available at The Plant Cell online. Funding This study was funded by a Discovery grant from the Natural Sciences and Engineering and Research Council (NSERC) of Canada (RGPIN-2024-05076) and a John Evans Leadership Fund grant from the Canadian Foundation for Innovation (36131). Data availability All data supporting the findings of this study are available within the article and its Supplemental Data files. References Associated Data Supplementary Materials Data Availability Statement All data supporting the findings of this study are available within the article and its Supplemental Data files.