Preserved and variable spatial‐chemical changes of lipids across tomato leaves in response to central vein wounding reveals potential origin of linolenic acid in signal transduction cascade

Abstract Membrane lipids serve as substrates for the generation of numerous signaling lipids when plants are exposed to environmental stresses, and jasmonic acid, an oxidized product of 18‐carbon unsaturated fatty acids (e.g., linolenic acid), has been recognized as the essential signal in wound‐induced gene expression. Yet, the contribution of individual membrane lipids in linolenic acid generation is ill‐defined. In this work, we performed spatial lipidomic experiments to track lipid changes that occur locally at the sight of leaf injury to better understand the potential origin of linolenic and linoleic acids from individual membrane lipids. The central veins of tomato leaflets were crushed using surgical forceps, leaves were cryosectioned and analyzed by two orthogonal matrix‐assisted laser desorption/ionization mass spectrometry imaging platforms for insight into lipid spatial distribution. Significant changes in lipid composition are only observed 30 min after wounding, while after 60 min lipidome homeostasis has been re‐established. Phosphatidylcholines exhibit a variable pattern of spatial behavior in individual plants. Among lysolipids, lysophosphatidylcholines strongly co‐localize with the injured zone of wounded leaflets, while, for example, lysophosphatidylglycerol (LPG) (16:1) accumulated preferentially toward the apex in the injured zone of wounded leaflets. In contrast, two other LPGs (LPG [18:3] and LPG [18:2]) are depleted in the injured zone. Our high‐resolution co‐localization imaging analyses suggest that linolenic acids are predominantly released from PCs with 16_18 fatty acid composition along the entire leaf, while it seems that in the apex zone PG (16:1_18:3) significantly contributes to the linolenic acid pool. These results also indicate distinct localization and/or substrate preferences of phospholipase isoforms in leaf tissue.


| INTRODUC TI ON
Wounding is a ubiquitous experience of plants where its impacts on crop productivity range from herbivory to weather damage.
The plant response to wounding extends from rapid vasculature occlusion (analogous to clotting; Ernst et al., 2012), to systemic protection (Delano-Frier et al., 2013) that reaches beyond the plant through volatiles to warn other plants, and even attract allies for protection (Unsicker et al., 2009). Mechanical wounding in tomato leaves induces differential expression of more than 200 genes (Scranton et al., 2013). The first molecular signal that is released is an 18-amino acid polypeptide hormone systemin which triggers the synthesis of powerful inducers of defense gene transcription, jasmonic acid (JA), and 12-oxophytodienoic acid (PDA; Ali & Baek, 2020). In this lipid-mediated signaling cascade, phospholipase enzymes play a key role in the production of JA and PDA by releasing linolenic acid from membrane phospholipids (Ali & Baek, 2020;Ryu, 2004;Vu et al., 2014). Other modifications of membrane lipids, like oxidation of fatty acids on galactolipids and head group acylation of monogalactosyldiacylglycerols (MGDG), are also reported as part of the wounding response in plant leaves (Vu et al., 2014).
Despite extensive study of the cascade of molecular events in response to plant wounding, there is very limited knowledge on the contribution and fate of individual membrane lipids (Hou et al., 2016;Vu et al., 2014;, especially their spatial distribution. Bulk lipidomic mass spectrometry (MS) analyses have been employed to reveal the changes in levels of membrane lipids during the wounding process and to show that individual plants vary in lipid composition despite application of comparable stress (Vu et al., 2014;. In general, it was observed that levels of structural lipids in Arabidopsis leaves decrease, whereas monoacyl molecular species (e.g. lysophospholipids), galactolipids, and phosphatidylglycerols increase (Vu et al., 2014). The significant limitation of these bulk analyses is that spatial information is lost, as molecular signatures are averaged across the sample, hence highly localized and low abundance lipids can be obscured in the data. The ability to observe lipid changes that occur locally at the site of injury can not only help resolve localized versus systemic lipid responses, but also provide lipid markers of leaf wounding that are mechanistically linked to plant recovery (Leon et al., 2001). Matrix-assisted laser desorption/ionization (MALDI) mass spectrometry imaging (MSI) is a robust chemical imaging technology that uses a focused laser beam to ablate and ionize material into the mass spectrometer at high spatial resolution. Combined with spatial probing, a MALDI imaging experiment enables simultaneous visualization of hundreds of molecules mapped to tissue morphology. This approach has been widely used to localize injury-induced lipid changes in the brain (Hankin et al., 2011;Mallah et al., 2018; and other mammalian organs (Quanico et al., 2018;Rao et al., 2016), as well as map lipid profiles in plant tissues affected by biotic or abiotic stresses (Sarabia et al., 2018).
Here, we applied global lipidomic and multimodal MALDI-MSI analyses to characterize spatial changes in membrane lipids of tomato leaves harvested 30 min after mechanical wounding of the primary vein. We used three biological replicates: three pairs of tomato leaflets harvested from the three individual plants. Each pair is composed of a wounded and a control leaflet and each leaflet represented with a cross-section from the apex (tip), middle, and base portion of the organ.
Micrometer scale resolution of biological replicates enabled visualization of individual plant response to wounding, to reveal consistent patterns of spatial-chemical lipid changes during the wounding process.

| Sample preparation
The central vein of 18 primary leaflets from six tomato plants (Solanum lycopersicum, purchased at the local market) were crushed using surgical forceps, and 9 leaflets were harvested 30 min (T30) and nine leaflets 60 min (T60) after the injury. Before wounding, 18 control primary leaflets were harvested on the opposite side of the rachis ( Figure S1).

| Liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis
For global lipidomic analysis, 15 ml Falcon tubes were pre-weighted and an apex, middle or base section of the leaflet was dissected and snap-frozen in liquid nitrogen. In total, there were 108 samples (18 control × 3 segments + 9 T30 × 3 segments + 9 T60 × 3 segments).
Lipids were extracted using chloroform-methanol and analyzed on a Linear Trap Quadrupole-Orbitrap Velos (Thermo Fisher Scientific; Methods S1). Two samples were removed from our analysis (one control sample and one 30 min after wounding sample due to poor data quality), resulting in 106 analyzed samples. LIQUID software (Kyle et al., 2017)

| MALDI MS imaging
For MALDI MSI, the three control and three T30 wounded leaflets were harvested whole and snap-frozen in 50 ml Falcon tubes filled with 2.5% carboxymethyl cellulose to preserve leaflet orientation.  interaction of treatment and leaf segment, and nested random effects of leaf within plant to account for samples coming from the same leaves and plants. Models were fit in R (version 3.6.1; R Core Team, 2020) using the lmer() function from the lme4 package (Bates et al., 2015). A Wald test for a significant interaction term, using Type III sums of squares, was first investigated to determine if the effect of treatment differed between leaf segments. If the interaction term was not significant, the pairwise comparisons of T30 versus control and T60 versus control were run across all segments. If the interaction term was significant, the pairwise comparisons of T30 versus control and T60 versus control were run within each leaf segment. In both cases, a Tukey multiple comparison adjustment (Kutner, 2005) was included to correct for multiple testing. SCiLS Lab software (version 2020a, Bremen, Germany) was used for image processing and creating m/z co-localization (based on Pearson's correlation analysis) plots. Imaging experiments (datasets) of wounded and control leaf segments were combined into single imaging SCiLS (.slx) file so that relative intensities of the individual ion image over control and wounded leaf segment can be compared.

LC-MS/MS lipidomic experiments performed (MS/MS spectra and
lipid identification can be found in Spectra S1) on extracts from control leaflets and leaflets harvested 30 and 60 min after wounding, were statistically compared across nine biological replicates for each condition. Two samples were removed from our analysis (one control sample and one 30 min after wounding sample due to poor data quality), resulting in 106 analyzed samples. A total of 71 and 126 lipids were observed in at least one of the 106 samples for the negative and positive ionization data, respectively. Significant changes in lipid composition are only observed 30 min (T30) after wounding, while after 60 min (T60) lipidome homeostasis has been re-established, "recovering" has occurred, and only a few lipids still show elevated or decreased content compared to control leaflets (all lipids that showed statistically significant (p < .05) changes in abundance in response to wounding are listed in Table S1).    (Steuer et al., 2003;Vu et al., 2014). Notably, the ability to achieve experimental consistency, in precisely controlled incubators, is not indicative of the biological variation in a more realistic plant growth environment (with variations in watering, shading, etc.). Accumulation of lysophospholipids at the site of injury indicates that the production of the lysophospholipids is a direct result of cellular damage. Accumulation of these monoacyl lipids in the injured zone results from phospholipid hydrolysis by phospholipases A (PLAs) and is responsible for systemic wound signal transduction in the tomato leaf (Conconi et al., 1996;Narvaez-Vasquez et al., 1999) and plants in general (Vu et al., 2014;. Our co- The complexity of plant lipidomic characterization presents challenges that require replication and independent, orthogonal methodologies, so that accurate assessments of lipid dynamics can be applied to our understanding of plants under stress (Vu et al., 2014).

| D ISCUSS I ON
Our results highlight the importance of including high-resolution image analysis in the interpretation of localized responses that occur during localized stress. Our multimodal MALDI-MSI experiments characterized the lipidomic response at the site of wounding and surrounding regions by creating snapshots detailing how individual lipid species were distributed 30 min after injury. Replicate imaging revealed co-localization of some lipid classes, and dynamic response of other lipids that while consistent with bulk lipid measurements, reveal tremendous spatial variability in response to wounding.
More specifically, the co-localization analysis provided biomolecular evidence for the pathway origin of linolenic acid-an important substrate for signal transducer biosynthesis during wounding. This unique ultra-high-resolution multi-component biomolecular characterization can assist in a better understanding of PLAs activation, specificity, and selectivity in the initial steps of stress signaling biochemistry. Understanding of these signal transduction cascades at a spatial-molecular level can aid in developing future breeding and genetic engineering strategies for plant protection from stresses due to herbivore-attacks, pathogen infections, and environmental stresses such as drought.

ACK N OWLED G M ENTS
This work was funded in part by the U.S. Department of Energy