A Pathogen-Responsive Gene Cluster for Highly Modified Fatty Acids in Tomato

Abstract

In response to biotic stress, plants produce suites of highly modified fatty acids that bear unusual chemical functionalities. Despite their chemical complexity and proposed roles in pathogen defense, little is known about the biosynthesis of decorated fatty acids in plants. Falcarindiol is a prototypical acetylenic lipid present in carrot, tomato, and celery that inhibits growth of fungi and human cancer cell lines. Using a combination of untargeted metabolomics and RNA sequencing, we discovered a biosynthetic gene cluster in tomato (Solanum lycopersicum) required for falcarindiol production. By reconstituting initial biosynthetic steps in a heterologous host and generating transgenic pathway mutants in tomato, we demonstrate a direct role of the cluster in falcarindiol biosynthesis and resistance to fungal and bacterial pathogens in tomato leaves. This work reveals a mechanism by which plants sculpt their lipid pool in response to pathogens and provides critical insight into the complex biochemistry of alkynyl lipid production.

Keywords: acetylenase; acetylenic lipids; antifungal oxylipin; falcarindiol; plant lipid gene cluster; plant natural product biosynthesis; transcriptomics; untargeted metabolomics.

Figure 1.. Representative acetylenic natural products found in edible plants

Linoleic acid produced in primary metabolism is released from plant membranes and is used in secondary metabolism to produce the hormone jasmonic acid. Linoleic acid is also a possible precursor for the synthesis highly modified acetylenic lipids. Falcarindiol is a representative acetylenic lipid found in tomato, carrot and ginseng, for which no biosynthetic genes have been described.

Figure 2.. Identification of putative non-heme diiron enzymes including desaturases and decarbonylases to discover biosynthetic genes involved in falcarindiol production (See also Figures S1–2 and Table S1–3)

(A) Correlation of falcarindiol levels (bar graph shows extracted ion chromatogram (EIC) peak integration (m/z 283.16, [M+Na]⁺) using LC-MS) to transcript abundance of six candidate acetylenase genes (heatmap illustrates Z-score determined using counts per million base pairs values obtained using RNA-Seq) in tomato leaves under various elicitation conditions. Enumeration of transcripts was performed by mapping reads to the SL3.0 version of the Heniz tomato genome. Transcripts are sorted by their Pearson’s correlation coefficient (PCC) values against falcarindiol production (Table S2). Colors represent elicitor types (green: fungal elicitor, orange: bacterial elicitor, blue: Mock, water). Experiments were conducted in two separate batches: Mock 1 for MAMPs and human-associated microbes; Mock 2 for plant-associated microbes. (B) PCC values for 40 candidate tomato desaturases’ transcript abundance (CPM) against falcarindiol levels (red: falcarindiol, blue: six candidate genes in Figure 2A, black: remaining 34 desaturases detected in RNA-Seq analysis). (C) Genomic organization of the candidate metabolic gene cluster on chromosome 12 in Heinz SL3.0 genome compared to the VF36 genome sequence obtained by Nanopore and Sanger DNA sequencing. Genes related to the falcarindiol pathway are highlighted. Genes are denoted without ‘Solyc’ and chromosome number.

Figure 3.. Proposed falcarindiol biosynthetic pathway and candidate enzyme classes for each step

Dotted arrows represent proposed reactions and the solid arrow represents a transformation characterized previously in plants.

Figure 4.. Reconstitution of tomato biosynthetic genes for highly modified fatty acid in Nicotiana benthamiana (See also Figures S3–5 and Table S4)

(A) Proposed pathway from linoleic acid to falcarindiol and role for candidate genes from identified gene cluster. Genes are denoted without ‘Solyc’ and chromosome number. Proposed chemical structures of octadecene-diynoic acid and octadecadien-diynoic acid based on LC-MS data and acetylene derivatization using click chemistry. (B) Metabolite accumulation in N. benthamiana leaves determined by integration of EIC for indicated compound (ion abundance plotted as mean ± SD (n = 3). Boxes represent transiently expressed enzymes (red = 250 cDNA for 100250, blue = 240/260 cDNA for 100240 and 100260, grey = Crep1 from Crepis alpina or green fluorescent protein (GFP) as negative control).

Figure 5.. CRISPR/Cas9-induced targeted mutagenesis and gene complementation in tomato (See also Figures S6 and Table S4).

(A) Two CRISPR/Cas9 targeted regions in ACET1a and in ACET1b (ΔACET1), one in 100250 (Δ250), and one in 100270 (Δ270) are marked with red boxes. (B) Falcarindiol production in wild type (WT, n=6) and independent T1 generation CRISPR knock-out mutants of tomato (ΔACET1, n=5; Δ270, n=5) after Mock or Cladosporium fulvum treatment. Genotype of each mutant plant can be found in Fig. S6 and the plant ID is # 33, 39, 42, 50, 53 (ΔACET1) and #1, 3, 4, 13, and 20 (Δ270), respectively). (C) Falcarindiol production in WT (n=4) and mutant lines (ΔACET1, n=3; Δ270, n=5) after transient overexpression (ox) of genes via Agrobacterium-mediated infiltration (GFPox, ACET1ox, or 10027ox) as assessed by LC-qTOF-MS. The y axis represents falcarindiol content as determined by integration of EICs from LC-MS data. (D) Falcarindiol production in wild type (WT, n=3) and independent T1 generation CRISPR knock-out mutants of tomato (Δ250, n=4) after transient overexpression of GFP (GFPox) via Agrobacterium (C58C1 or 1D1249)-mediated infiltration. Genotype of each mutant plant can be found in Fig. S6 and the four Δ250 plant ID is # 9, 16, 23, and 41.

Figure 6.. Phenotype of infected tomato wild type, ΔACET1, Δ270, and Δ250 leaves (See also Figures S7 and Table S4)

(A) Disease symptoms of Botrytis cinerea-infected WT and mutant (ΔACET1 and Δ270) leaves 3 days post-infection (dpi). Plant lines (T2 generation) used for analysis: ΔACET1 plant 50 (−1bp, Cas9-) and Δ270 plant 20 (−4bp, Cas9+) (See Figure S6). Representative photos. Experiment was performed 3 times. (B) Relative lesion size for B. cinerea-infected leaves (n = 12) using ImageJ. Error bars represent SE. (C) Relative fungal biomass in B. cinerea-infected leaves (n = 12) measured by qPCR. Error bars represent SE. (D) Number of bacteria (log(cfu/cm²)) in infected leaves at 0 and 4 dpi. Pseudomonas syringae pathovar tomato strain DC3000 (Pst DC3000) and ΔavrPtoΔavrPtoB mutant (ΔavrPtoΔavrPtoB). n = 3 plants. Plant lines used for analysis: ΔACET1 plant 50 (−1bp, Cas9-, T2), Δ270 plant 20 (−4bp, Cas9+, T2), and Δ250 plant 9 (−77bp, Cas9-, T1) (See Figure S6). Error bars indicate SD. Experiment was performed 3 times. Different letters indicate the statistically significant (one-way analysis of variance and Tukey’s HSD test, P <0.05) differences between the samples.

Figure 7.. Comparative genomics study with other plants in Solanaceae family and carrot

(A) Blastp top hit for each enzyme in the cluster in other plants (three Solanaceae family plants and D. carota: carrot). Percent in parenthesis indicates the percent homology with the query. (B) Genomic organization of the metabolic gene clusters in Solanaceae family plants (S. lycopersicum: tomato, S. pennellii: wild tomato, S. tuberosum: potato, C. annuum: pepper). The genes are indicated by arrows. The numbers on arrows represent the query gene names in a shorten form (e.g. Solyc12g100250 = 5). If multiple homologs are present, the one with the highest sequence identity is marked with an asterisk.