Enzyme protein turnover accounts for about half the maintenance energy budget in plants. Slowing turnover─i.e., extending the effective working life (Catalytic Cycles till Replacement, CCR)─of short-lived enzymes is thus a rational strategy to conserve energy and carbon and raise crop productivity. Arabidopsis histidinol dehydrogenase (HDH) is a short-lived enzyme that can sustain life-shortening damage from its aminoaldehyde reaction intermediate. We used the yeast OrthoRep continuous directed evolution system in a his4Δ strain to raise cumulative HDH function and, by proxy, lifespan as functional enzymes, by selecting for growth rate while tapering histidinol concentration and escalating that of the inhibitor histamine. Improved HDH variants carried diverse nonsynonymous mutations and ranged 20-fold in level. Improved HDH performance was associated with higher HDH abundance in some cases and with greater catalytic efficiency or histamine resistance in others. These findings indicate that OrthoRep-based directed evolution can extend enzyme working life in vivo in addition to, as expected, altering kinetic properties.

Enzyme protein turnover accounts for about half the maintenance energy budget in plants. Slowing turnover – i.e., extending lifespan – of short-lived enzymes is thus a rational strategy to conserve energy and carbon, and raise crop productivity. Arabidopsis histidinol dehydrogenase (HDH) is a short-lived enzyme that can sustain life-shortening damage from its aminoaldehyde reaction intermediate. We used the yeast OrthoRep continuous directed evolution system in a his4 Δ strain to raise HDH protein abundance (a proxy for lifespan) by selecting for growth rate while tapering histidinol concentration and escalating that of the inhibitor histamine. Improved HDHs carried diverse nonsynonymous mutations and ranged 20-fold in level. Improved HDH performance was associated with higher HDH abundance in some cases and with greater catalytic efficiency or histamine resistance in others. These findings indicate that OrthoRep-based directed evolution can extend enzyme lifespan in vivo in addition to, as expected, altering kinetic properties. ### Competing Interest Statement The authors have declared no competing interest. National Institute of Food and Agriculture, https://ror.org/05qx3fv49, FLA-HOS-005796 C.V. Griffin, Sr. Foundation Australian Research Council, https://ror.org/05mmh0f86, FL200100057 Bioplatforms Australia, https://ror.org/042gz1a70

Continuous directed evolution (CDE) is a powerful tool for enzyme engineering due to the depth and scale of evolutionary search that it enables. If suitably controlled and calibrated, CDE could be widely applied in plant breeding and biotechnology to improve plant enzymes ex planta. We tested this concept by evolving Arabidopsis arogenate dehydratase (AtADT2) for resistance to feedback inhibition. We used an Escherichia coli platform with a phenylalanine biosynthesis pathway reconfigured ("plantized") to mimic the plant pathway, a T7RNA polymerase-base deaminase hypermutation system (eMutaT7), and 4-fluorophenylalanine as selective agent. Selection schemes were prevalidated using a known feedback-resistant AtADT2 variant. We obtained variants that had 4-fluorophenylalanine resistance at least matching the known variant and that carried mutations in the ACT domain responsible for feedback inhibition. We conclude that ex planta CDE of plant enzymes in a microbial platform is a viable way to tailor characteristics that involve interaction with small molecules.

Nonphotosynthetic plant metabolic processes are powered by respiratory energy, a limited resource that metabolic engineers-like plants themselves-must manage prudently.

Like angiosperms from several other families, the leguminous shrub Gastrolobium bilobum R.Br. produces and accumulates fluoroacetate, indicating that it performs the difficult chemistry needed to make a C–F bond. Bioinformatic analyses indicate that plants lack homologs of the only enzymes known to make a C–F bond, i.e., the Actinomycete flurorinases that form 5′-fluoro-5′-deoxyadenosine from S-adenosylmethionine and fluoride ion. To probe the origin of fluoroacetate in G. bilobum we first showed that fluoroacetate accumulates to millimolar levels in young leaves but not older leaves, stems or roots, that leaf fluoroacetate levels vary >20-fold between individual plants and are not markedly raised by sodium fluoride treatment. Young leaves were fed adenosine-13C-ribose, 13C-serine, or 13C-acetate to test plausible biosynthetic routes to fluoroacetate from S-adenosylmethionine, a C3-pyridoxal phosphate complex, or acetyl-CoA, respectively. Incorporation of 13C into expected metabolites confirmed that all three precursors were taken up and metabolized. Consistent with the bioinformatic evidence against an Actinomycete-type pathway, no adenosine-13C-ribose was converted to 13C-fluoroacetate; nor was the characteristic 4-fluorothreonine product of the Actinomycete pathway detected. Similarly, no 13C from acetate or serine was incorporated into fluoroacetate. While not fully excluding the hypothetical pathways that were tested, these negative labeling data imply that G. bilobum creates the C–F bond by an unprecedented biochemical reaction. Enzyme(s) that mediate such a reaction could be of great value in pharmaceutical and agrochemical manufacturing.

Plants make a variety of specialized metabolites that can mediate interactions with animals, microbes and competitor plants. Understanding how plants synthesize these compounds enables studies of their biological roles by manipulating their synthesis in vivo as well as producing them in vitro . Acylsugars are a group of protective metabolites that accumulate in the trichomes of many Solanaceae family plants. Acylinositol biosynthesis is of interest because it appears to be restricted to a subgroup of species within the Solanum genus. Previous work characterized a triacylinositol acetyltransferase involved in acylinositol biosynthesis in the Andean fruit plant Solanum quitoense (lulo or naranjillo). We characterized three additional S. quitoense trichome expressed enzymes, and found that virus induced gene silencing of each caused changes in acylinositol accumulation. Surprisingly, the in vitro triacylinositol products of these enzymes are distinct from those that accumulate in planta . These enzymes, nonetheless, provide an opportunities to test the biological impact and properties of these triacylinositols in vitro .

Continuous directed evolution of enzymes and other proteins in microbial hosts is capable of outperforming classical directed evolution by executing hypermutation and selection concurrently in vivo, at scale, with minimal manual input. Provided that a target enzyme’s activity can be coupled to growth of the host cells, the activity can be improved simply by selecting for growth. Like all directed evolution, the continuous version requires no prior mechanistic knowledge of the target. Continuous directed evolution is thus a powerful new way to modify plant or non-plant enzymes for use in plant metabolic research and engineering. Here, we first describe the basic features of the Saccharomyces cerevisiae OrthoRep system for continuous directed evolution and compare it briefly with other systems. We then give a step-by-step account of three ways in which OrthoRep can be deployed to evolve primary metabolic enzymes, using a THI4 thiazole synthase as an example and illustrating the mutational outcomes obtained. We close by outlining applications of OrthoRep that serve growing demands (i) to change the characteristics of plant enzymes destined for return to plants, and (ii) to adapt (‘plantize’) enzymes from prokaryotes – especially exotic prokaryotes – to function well in mild, plant-like conditions. One-sentence summary Continuous directed evolution using the yeast OrthoRep system is a powerful new way to improve enzymes for use in plant engineering as illustrated by ‘plantizing’ a bacterial thiamin synthesis enzyme..

Plant and fungal THI4 thiazole synthases produce the thiamin thiazole moiety in aerobic conditions via a single-turnover suicide reaction that uses an active-site Cys residue as sulfur donor. Multiple-turnover (i.e. catalytic) THI4s lacking an active-site Cys (non-Cys THI4s) that use sulfide as sulfur donor have been biochemically characterized -- but only from archaeal methanogens that are anaerobic, O2-sensitive hyperthermophiles from sulfide-rich habitats. These THI4s prefer iron as cofactor. A survey of prokaryote genomes uncovered non-Cys THI4s in aerobic mesophiles from sulfide-poor habitats, suggesting that multiple-turnover THI4 operation is possible in aerobic, mild, low-sulfide conditions. This was confirmed by testing 23 representative non-Cys THI4s for complementation of an Escherichia coli ΔthiG thiazole auxotroph in aerobic conditions. Sixteen were clearly active, and more so when intracellular sulfide level was raised by supplying Cys, demonstrating catalytic function in the presence of O2 at mild temperatures and indicating use of sulfide or a sulfide metabolite as sulfur donor. Comparative genomic evidence linked non-Cys THI4s with proteins from families that bind, transport, or metabolize cobalt or other heavy metals. The crystal structure of the aerotolerant bacterial Thermovibrio ammonificans THI4 was determined to probe the molecular basis of aerotolerance. The structure suggested no large deviations compared with the structures of THI4s from O2-sensitive methanogens, but is consistent with an alternative catalytic metal. Together with complementation data, use of cobalt rather than iron was supported. We conclude that catalytic THI4s can indeed operate aerobically and that the metal cofactor inserted is a likely natural determinant of aerotolerance.

Enzymes have in vivo life spans. Analysis of life spans, i.e., lifetime totals of catalytic turnovers, suggests that nonsurvivable collateral chemical damage from the very reactions that enzymes catalyze is a common but underdiagnosed cause of enzyme death. Analysis also implies that many enzymes are moderately deficient in that their active-site regions are not naturally as hardened against such collateral damage as they could be, leaving room for improvement by rational design or directed evolution. Enzyme life span might also be improved by engineering systems that repair otherwise fatal active-site damage, of which a handful are known and more are inferred to exist. Unfortunately, the data needed to design and execute such improvements are lacking: there are too few measurements of in vivo life span, and existing information about the extent, nature, and mechanisms of active-site damage and repair during normal enzyme operation is too scarce, anecdotal, and speculative to act on. Fortunately, advances in proteomics, metabolomics, cheminformatics, comparative genomics, and structural biochemistry now empower a systematic, data-driven approach for identifying, predicting, and validating instances of active-site damage and its repair. These capabilities would be practically useful in enzyme redesign and improvement of in-use stability and could change our thinking about which enzymes die young in vivo, and why.

Plants make many biologically active, specialized metabolites, which vary in structure, biosynthesis, and the processes they influence. An increasing number of these compounds are documented to protect plants from insects, pathogens, or herbivores or to mediate interactions with beneficial organisms, including pollinators and nitrogen-fixing microbes. Acylsugars, one class of protective compounds, are made in glandular trichomes of plants across the Solanaceae family. While most described acylsugars are acylsucroses, published examples also include acylsugars with hexose cores. The South American fruit crop naranjilla (lulo; Solanum quitoense) produces acylsugars containing a myoinositol core. We identified an enzyme that acetylates triacylinositols, a function homologous to the last step in the acylsucrose biosynthetic pathway of tomato (Solanum lycopersicum). Our analysis reveals parallels between S. lycopersicum acylsucrose and S. quitoense acylinositol biosynthesis, suggesting a common evolutionary origin.

Article Figures and data Abstract eLife digest Introduction Results Discussion Materials and methods Data availability References Decision letter Author response Article and author information Metrics Abstract Plants produce phylogenetically and spatially restricted, as well as structurally diverse specialized metabolites via multistep metabolic pathways. Hallmarks of specialized metabolic evolution include enzymatic promiscuity and recruitment of primary metabolic enzymes and examples of genomic clustering of pathway genes. Solanaceae glandular trichomes produce defensive acylsugars, with sidechains that vary in length across the family. We describe a tomato gene cluster on chromosome 7 involved in medium chain acylsugar accumulation due to trichome specific acyl-CoA synthetase and enoyl-CoA hydratase genes. This cluster co-localizes with a tomato steroidal alkaloid gene cluster and is syntenic to a chromosome 12 region containing another acylsugar pathway gene. We reconstructed the evolutionary events leading to this gene cluster and found that its phylogenetic distribution correlates with medium chain acylsugar accumulation across the Solanaceae. This work reveals insights into the dynamics behind gene cluster evolution and cell-type specific metabolite diversity. eLife digest Plants produce a vast variety of different molecules known as secondary or specialized metabolites to attract pollinating insects, such as bees, or protect themselves against herbivores and pests. The secondary metabolites are made from simple building blocks that are readily available in plants, including amino acids, fatty acids and sugars. Different species of plant, and even different parts of the same plant, produce their own sets of secondary metabolites. For example, the hairs on the surface of tomatoes and other members of the nightshade family of plants make metabolites known as acylsugars. These chemicals deter herbivores and pests from damaging the plants. To make acylsugars, the plants attach long chains known as fatty acyl groups to molecules of sugar, such as sucrose. Some members of the nightshade family produce acylsugars with longer chains than others. In particular, acylsugars with long chains are only found in tomatoes and other closely-related species. It remained unclear how the nightshade family evolved to produce acylsugars with chains of different lengths. To address this question, Fan et al. used genetic and biochemical approaches to study tomato plants and other members of the nightshade family. The experiments identified two genes known as AACS and AECH in tomatoes that produce acylsugars with long chains. These two genes originated from the genes of older enzymes that metabolize fatty acids – the building blocks of fats – in plant cells. Unlike the older genes, AACS and AECH were only active at the tips of the hairs on the plant’s surface. Fan et al. then investigated the evolutionary relationship between 11 members of the nightshade family and two other plant species. This revealed that AACS and AECH emerged in the nightshade family around the same time that longer chains of acylsugars started appearing. These findings provide insights into how plants evolved to be able to produce a variety of secondary metabolites that may protect them from a broader range of pests. The gene cluster identified in this work could be used to engineer other species of crop plants to start producing acylsugars as natural pesticides. Introduction Despite the enormous structural diversity of plant specialized metabolites, they are derived from a relatively small number of primary metabolites, such as sugars, amino acids, nucleotides, and fatty acids (Maeda, 2019). These lineage-, tissue- or cell- type specific specialized metabolites mediate environmental interactions, such as herbivore and pathogen deterrence or pollinator and symbiont attraction (Mithöfer and Boland, 2012; Pichersky and Lewinsohn, 2011). Specialized metabolism evolution is primarily driven by gene duplication (Moghe and Last, 2015; Panchy et al., 2016), and relaxed selection of the resulting gene pairs allows modification of cell- and tissue-specific gene expression and changes in enzymatic activity. This results in expanded substrate recognition and/or diversified product formation (Khersonsky and Tawfik, 2010; Leong and Last, 2017). The neofunctionalized enzymes can prime the origin and diversification of specialized metabolic pathways (Schenck and Last, 2020; Weng et al., 2012; Weng, 2014). There are many examples of mechanisms that lead to novel enzymatic activities in specialized cell- or tissue-types, however, the principles that govern assembly of multi-enzyme specialized metabolic pathways are less well established. One appealing hypothesis involves the stepwise recruitment of pathway enzymes (Noda-Garcia et al., 2018). In rare cases, non-homologous specialized metabolic enzyme genes occur in proximity to each other in a genomic region, forming a biosynthetic gene cluster (Nützmann et al., 2016; Nützmann and Osbourn, 2014; Rokas et al., 2018). In recent years, an increasing number of specialized metabolic gene clusters (SMGCs) were experimentally identified or bioinformatically predicted in plants (Boutanaev et al., 2015; Castillo et al., 2013; Schläpfer et al., 2017). However, although most experimentally characterized plant SMGCs are co-expressed, the majority of the bioinformatically predicted ones do not show coexpression under global network analysis (Wisecaver et al., 2017). While examples of SMGCs are still relatively rare in plants, experimentally validated cases were reported for a surprisingly diverse group of pathways. These include terpenes (Chae et al., 2014; Prisic et al., 2004; Qi et al., 2004; Wilderman et al., 2004), cyclic hydroxamic acids (Frey et al., 1997), biosynthetically unrelated alkaloids (Itkin et al., 2013; Winzer et al., 2012), polyketides (Schneider et al., 2016), cyanogenic glucosides (Takos et al., 2011), and modified fatty acids (Jeon et al., 2020). However, whereas each cluster encodes multiple non-homologous enzymes of a biosynthetic pathway, evolution of their assembly is not well understood. Acylsugars are a group of insecticidal (Leckie et al., 2016) and anti-inflammatory (Herrera-Salgado et al., 2005) chemicals mainly observed in glandular trichomes of Solanaceae species (Fan et al., 2019; Schuurink and Tissier, 2020). These specialized metabolites are sugar aliphatic esters with three levels of structural diversity across the Solanaceae family: acyl chain length, acylation position, and sugar core (Fan et al., 2019). The primary metabolites sucrose and aliphatic acyl-CoAs are the biosynthetic precursors of acylsucroses in plants as evolutionarily divergent as the cultivated tomato Solanum lycopersicum (Fan et al., 2016; Figure 1), Petunia axillaris (Nadakuduti et al., 2017) and Salpiglossis sinuata (Moghe et al., 2017). The core tomato acylsucrose biosynthetic pathway involves four BAHD [BEAT, AHCT, HCBT, DAT (D'Auria, 2006) family acylsucrose acyltransferases (Sl-ASAT1 through Sl-ASAT4), which are specifically expressed in the type I/IV trichome tip cells (Fan et al., 2016; Schilmiller et al., 2015; Schilmiller et al., 2012). These enzymes catalyze consecutive reactions utilizing sucrose and acyl-CoA substrates to produce the full set of cultivated tomato acylsucroses in vitro (Fan et al., 2016). Figure 1 Download asset Open asset Primary metabolites are biosynthetic precursors of tomato trichome acylsugars. In cultivated tomatoes, the trichome acylsucroses are synthesized by four Sl-ASATs using the primary metabolites – sucrose and different types of acyl-CoAs – as substrates. In this study we provide evidence that medium chain fatty acids are converted to acyl-CoAs by an acyl-CoA synthetase for medium chain acylsugar biosynthesis. Co-option of primary metabolic enzymes contributed to the evolution of acylsugar biosynthesis and led to interspecific structural diversification across the Solanum tomato clade. One example is an invertase-like enzyme originating from carbohydrate metabolism that generates acylglucoses in the wild tomato S. pennellii through cleavage of the acylsucrose glycosidic bond (Leong et al., 2019). In another case, allelic variation of a truncated isopropylmalate synthase-like enzyme (IPMS3) – from branched chain amino acid metabolism – leads to acylsugar iC4/iC5 (2-methylpropanoic/3-methylbutanoic acid) acyl chain diversity in S. pennellii and S. lycopersicum (Ning et al., 2015). Acylsugar structural diversity is even more striking across the family. Previous studies revealed variation in acyl chain length (Ghosh et al., 2014; Liu et al., 2017; Moghe et al., 2017): Nicotiana, Petunia and Salpiglossis species were reported to accumulate acylsugars containing only short acyl chains (carbon number, C ≤ 8). In contrast, some species in Solanum and other closely related genera produce acylsugars with medium acyl chains (C ≥ 10). These results are consistent with the hypothesis that the capability to produce medium chain acylsugars varies across the Solanaceae family. In this study, we identify a metabolic gene cluster on tomato chromosome 7 containing two non-homologous genes – acylsugar acyl-CoA synthetase (AACS) and acylsugar enoyl-CoA hydratase (AECH) – affecting medium chain acylsugar biosynthesis. Genetic and biochemical results show that the trichome enriched AACS and AECH are involved in generating medium chain acyl-CoAs, which are donor substrates for acylsugar biosynthesis. Genomic analysis revealed a syntenic region on chromosome 12, where the acylsucrose biosynthetic Sl-ASAT1 is located (Fan et al., 2016). Phylogenetic analysis of the syntenic regions in Solanaceae and beyond led to evolutionary reconstruction of the origin of the acylsugar gene cluster. We infer that sequential gene insertion facilitated emergence of this gene cluster in tomato. These results provide insights into specialized metabolic evolution through emergence of cell-type specific gene expression, the formation of metabolic gene clusters and illuminates additional examples of primary metabolic enzymes being co-opted into specialized metabolism. Results Identification of a metabolic gene cluster that affects tomato trichome medium chain acylsugar biosynthesis S. pennellii natural accessions (Mandal et al., 2020), as well as the S. lycopersicum M82 ×S. pennellii LA0716 chromosomal substitution introgression lines (ILs) (Eshed and Zamir, 1995), offer convenient resources to investigate interspecific genetic variation that affects acylsugar metabolic diversity (Mandal et al., 2020; Schilmiller et al., 2010). In a rescreen of ILs for S. pennellii genetic regions that alter trichome acylsugar profiles (Schilmiller et al., 2010), IL7-4 was found to accumulate increased C10 medium chain containing acylsugars compared with M82 (Figure 2, A and B). The genetic locus that contributes to the acylsugar phenotype was narrowed down to a 685 kb region through screening selected backcross inbred lines (BILs) (Ofner et al., 2016) that have recombination breakpoints on chromosome 7 (Figure 2C). Because tomato acylsucrose biosynthesis occurs in trichomes, candidate genes in this region were filtered based on their trichome-specific expression patterns. This analysis identified a locus containing multiple tandemly duplicated genes of three families – an acyl-CoA synthetase (ACS), enoyl-CoA hydratase (ECH), and BAHD acyltransferase. Our analysis (Moore et al., 2020) revealed co-expression of four Sl-ASATs (Fan et al., 2019) and three genes at the locus – Solyc07g043630, Solyc07g043660, and Solyc07g043680 (Supplementary file 1 and Figure 2—figure supplement 1). Expression of these three genes was trichome enriched (Figure 2D), and thus they were selected for further analysis. Figure 2 with 1 supplement see all Download asset Open asset Mapping of a genetic locus related to acylsugar variations in tomato interspecific introgression lines. (A) Electrospray ionization negative (ESI-) mode, base-peak intensity (BPI) LC/MS chromatogram of trichome metabolites from cultivated tomato S. lycopersicum M82 and introgression line IL7-4. The orange bars highlight two acylsugars that have higher abundance in IL7-4 than in M82. For the acylsucrose nomenclature, ‘S’ refers to a sucrose backbone, ‘3:22’ means three acyl chains with twenty-two carbons in total. The length of each acyl chain is shown in the parentheses. (B) Peak area percentage of seven major trichome acylsugars in M82 and IL7-4. The sum of the peak area percentage of each acylsugar is equal to 100% in each sample. The data is shown for three plants ± SEM. **p<0.01, Welch two-sample t test. Figure 2—source data 1 includes values for the analysis. (C) Mapping the genetic locus contributing to the IL7-4 acylsugar phenotype using selected backcross inbred lines (BILs) that have recombination break points within the introgression region of IL7-4. (D) Narrowing down candidate genes in the locus using trichome/stem RNA-seq datasets generated from previous study (Ning et al., 2015). A region with duplicated genes of three types – acyl-CoA synthetase (ACS), BAHD acyltransferase, and enoyl-CoA hydratase (ECH) – is shown. The red-blue color gradient provides a visual marker to rank the expression levels represented by Fragments Per Kilobase of transcript per Million mapped reads (FPKM). Coexpression analysis of tomato ACS, ECH, and BAHD acyltransferase family genes is shown in Figure 2—figure supplement 1. Figure 2—source data 1 Data used to make Figure 2B. Peak area percentage of seven major trichome acylsugars in M82 and IL7-4. https://cdn.elifesciences.org/articles/56717/elife-56717-fig2-data1-v2.xlsx Download elife-56717-fig2-data1-v2.xlsx The three candidate genes were tested for involvement in tomato acylsugar biosynthesis by making loss of function mutations using the CRISPR-Cas9 gene editing system. Two guide RNAs (gRNAs) were designed to target one or two exons of each gene to assist site-specific DNA cleavage by hCas (Brooks et al., 2014; Figure 3—figure supplement 1, A–C). In the self-crossed T1 progeny of stably transformed M82 plants, at least two homozygous mutants were obtained in Solyc07g043630, Solyc07g043660, and Solyc07g043680 (Figure 3—figure supplement 1, A–C), and these were analyzed for leaf trichome acylsugar changes. Altered acylsugar profiles were observed in the ACS-annotated Solyc07g043630 or ECH-annotated Solyc07g043680 mutants (Figure 3, A and B), but not in the ACS-annotated Solyc07g043660 mutant (Figure 3—figure supplement 1D). Despite carrying mutations in distinctly annotated genes (ACS or ECH), the two mutants exhibited the same phenotype – no detectable medium acyl chain (C10 or C12) containing acylsugars (Figure 3, A and B). We renamed Solyc07g043630 as acylsugar acyl-CoA synthetase 1 (Sl-AACS1) and Solyc07g043680 as acylsugar enoyl-CoA hydratase 1 (Sl-AECH1) based on this analysis. Figure 3 with 2 supplements see all Download asset Open asset CRISPR/Cas9-mediated gene knockout of tomato Sl-AACS1 or Sl-AECH1 eliminates detectable medium chain containing acylsugars. (A) Combined LC/MS extracted ion chromatograms of trichome metabolites from CRISPR mutants sl-aacs1 and sl-aech1. The medium chain acylsugars that are not detected in the two mutants are denoted by pairs of vertical dotted lines. Figure 3—figure supplement 1 describes the design of the gRNAs and details of the gene edits. (B) Quantification of seven major trichome acylsugars in sl-aacs1 and sl-aech1 mutants. Two independent T2 generation transgenic lines for each mutant were used for analysis. The peak area/internal standard (IS) normalized by leaf dry weight (DW) is shown from six plants ± SEM. Figure 3—source data 1 includes values for the analysis. (C) Confocal fluorescence images showing that GFP fluorescence driven by Sl-AACS1 or Sl-AECH1 is located in the tip cells of type I/IV trichomes. Their tissue specific expressions are similar to Sl-ASAT1 (Fan et al., 2016), which locates in a chromosome 12 region that is syntenic to the locus containing Sl-AACS1 and Sl-AECH1. Figure 3—figure supplement 2 provides the detailed information of the syntenic region. Sl-AACS1, Sl-AECH1, and Sl-ASAT1 are the only gene models with demonstrated functions in acylsugar biosynthesis. Figure 3—source data 1 Data used to make Figure 3B. Quantification of seven major trichome acylsugars in the CRISPR mutants sl-aacs1 and sl-aech1, as well as the parent M82. https://cdn.elifesciences.org/articles/56717/elife-56717-fig3-data1-v2.xlsx Download elife-56717-fig3-data1-v2.xlsx Further genomic analysis revealed that Sl-AACS1 and Sl-AECH1 belong to a syntenic region shared with a locus on chromosome 12, where Sl-ASAT1 is located (Figure 3C and Figure 3—figure supplement 2). Sl-ASAT1 is specifically expressed in trichome tip cells and encodes the enzyme catalyzing the first step of tomato acylsucrose biosynthesis (Fan et al., 2016). This led us to test the cell-type expression pattern of Sl-AACS1 and Sl-AECH1. Like Sl-ASAT1, the promoters of both genes drove GFP expression in the trichome tip cells of stably transformed M82 plants (Figure 3C). This supports our hypothesis that Sl-AACS1 and Sl-AECH1 are involved in tomato trichome acylsugar biosynthesis. Taken together, we identified a metabolic gene cluster involved in medium chain acylsugar biosynthesis, which is composed of two cell-type specific genes. In vitro analysis of Sl-AACS1 and Sl-AECH1 implicates their roles in medium chain acyl-CoA metabolism ACS and ECH are established to function in multiple cell compartments for the metabolism of acyl-CoA (Buchanan et al., 2015), the acyl donor substrates for ASAT enzymes. We sought to understand the organelle targeting of Sl-AACS1 and Sl-AECH1, to advance our knowledge of acylsugar machinery at the subcellular level. We constructed expression cassettes of Sl-AACS1, Sl-AECH1 and Solyc07g043660 with C-terminal cyan fluorescent protein (CFP), hypothesizing that the targeting peptides reside at the N-terminus of precursor proteins. When co-expressed in tobacco leaf epidermal cells, three CFP-tagged recombinant proteins co-localized with the mitochondrial marker MT-RFP (Nelson et al., 2007; Figure 4A and Figure 4—figure supplement 1A). To rule out the possibility of peroxisomal localization, we fused Sl-AACS1, Sl-AECH1, or Solyc07g043660 with N-terminus fused yellow fluorescent protein (YFP), considering that potential peroxisomal targeting peptides are usually located on the C-terminus (Brocard and Hartig, 2006). The expressed YFP-recombinant proteins were not co-localized with the peroxisomal marker RFP-PTS (Nelson et al., 2007; Figure 4—figure supplement 1B). Instead, they appeared distributed in the cytosol (Figure 4—figure supplement 1B), presumably because the N-terminal YFP blocked the mitochondria targeting signal. Taken together, protein expression and co-localization analyses suggest that Sl-AACS1, Sl-AECH1, and Solyc07g043660 encode enzymes targeted to mitochondria. Figure 4 with 2 supplements see all Download asset Open asset Functional analysis of Sl-AACS1 and Sl-AECH1 in N. benthamiana and recombinant Sl-AACS1 enzyme analysis. (A) Confocal images of co-expression analysis in tobacco leaf epidermal cells using C-terminal CFP-tagged either Sl-AACS1 or Sl-AECH1 and the mitochondrial marker MT-RFP. Arrowheads point to mitochondria that are indicated by MT-RFP fluorescent signals. Scale bar equals 10 μm. Figure 4—figure supplement 1B describes that the expressed YFP-recombinant proteins were not co-localized with the peroxisomal marker RFP-PTS (B) Aliphatic fatty acids of different chain lengths were used as the substrates to test Sl-AACS1 acyl-CoA synthetase activity. Mean amount of acyl-CoAs generated (nmol min−1 mg−1 proteins) was used to represent enzyme activities. The results are from three measurements ± SEM. Figure 4—source data 1 includes values for the measurements. (C) Enzyme activity of Sl-AACS1 for six fatty acid substrates. (D) Identification of membrane lipid phosphatidylcholine (PC), which contains medium acyl chains, following transient expression of Sl-AECH1 in N. benthamiana leaves. The results from expressing Sl-AACS1 and co-expressing both Sl-AECH1 and Sl-AACS1 are also shown. Mole percentage (Mol %) of the acyl chains from membrane lipids with carbon number 12, 14, 16, and 18 are shown for three biological replicates ± SEM. *p<0.05, **p<0.01. Welch two-sample t test was performed comparing with the empty vector control. Figure 4—source data 2 includes values for the lipid analysis. Acyl groups of the same chain lengths with saturated and unsaturated bonds were combined in the calculation. Figure 4—figure supplement 2 shows that the putative Sl-AECH1 orthologs from S. pennellii and S. quitoense generated medium chain lipids in the infiltrated leaves. Figure 4—source data 1 Data used to make Figure 4B. Aliphatic fatty acids of different chain lengths were used as the substrates to test Sl-AACS1 acyl-CoA synthetase activity. https://cdn.elifesciences.org/articles/56717/elife-56717-fig4-data1-v2.xlsx Download elife-56717-fig4-data1-v2.xlsx Figure 4—source data 2 Data used to make Figure 4D and Figure 4—figure supplement 1C. N. benthamiana leaf membrane lipid acyl chain composition. https://cdn.elifesciences.org/articles/56717/elife-56717-fig4-data2-v2.xlsx Download elife-56717-fig4-data2-v2.xlsx Sl-AACS1 belongs to a group of enzymes that activate diverse carboxylic acid substrates to produce acyl-CoAs. We hypothesized that Sl-AACS1 uses medium chain fatty acids as substrates, because ablation of Sl-AACS1 eliminated acylsugars with medium acyl chains. To characterize the in vitro activity of Sl-AACS1, we purified recombinant His-tagged proteins from Escherichia coli. Enzyme assays were performed by supplying fatty acid substrates with even carbon numbers from C2 through C18 (Figure 4B). The results showed that Sl-AACS1 utilized fatty acid substrates with lengths ranging from C6 to C12, including those with a terminal branched carbon (iC10:0) or an unsaturated bond (trans-2-decenoic acid, C10:1) (Figure 4, B and C). However, no activity was observed with the 3-hydroxylated C12 and C14 fatty acids as substrates (Figure 4B). These results support our hypothesis that Sl-AACS1 produces medium chain acyl-CoAs, which are in vivo substrates for acylsugar biosynthesis. To test whether Sl-AACS1 and Sl-AECH1 can produce medium chain acyl-CoAs in planta, we transiently expressed these genes in Nicotiana benthamiana leaves using Agrobacterium-mediated infiltration (Sainsbury et al., 2009). It is challenging to directly measure plant acyl-CoAs, due to their low concentration and separate organellar pools. We used an alternative approach and characterized membrane lipids, which are produced from acyl-CoA intermediates. We took advantage of the observation that N. benthamiana membrane lipids do not accumulate detectable acyl chains of 12 carbons or shorter. N. benthamiana leaves were infiltrated with constructs containing Sl-AACS1 or Sl-AECH1 individually, or together (Figure 4D). In contrast to the empty vector control, infiltration of Sl-AECH1 led to detectable levels of C12 acyl chains in the leaf membrane lipid phosphatidylcholine (PC) (Figure 4D). We also observed increased C14 acyl chains in PC, phosphatidylglycerol (PG), sulfoquinovosyl diacylglycerol (SQDG), and digalactosyldiacylglycerol (DGDG) in Sl-AECH1 infiltrated plants (Figure 4D and Figure 4—figure supplement 1C). These results suggest that Sl-AECH1 participates in generation of medium chain acyl-CoAs in planta, which are channeled into lipid biosynthesis. No medium chain acylsugars were detected, presumably due to the lack of core acylsugar biosynthetic machinery in N. benthamiana mesophyll cells. We asked whether the closest known homologs of Sl-AECH1 from Solanum species can generate medium chain lipids when transiently expressed in N. benthamiana. Two SQDGs with C12 chains were monitored by LC/MS as peaks diagnostic of lipids containing medium chain fatty acids (Figure 4—figure supplement 2, A and B). The results showed that only the putative Sl-AECH1 orthologs Sopen07g023250 (Sp-AECH1) and Sq_c37194 (Sq-AECH1) – from S. pennellii and S. quitoense respectively – generated medium chain lipids in the infiltrated leaves (Figure 4—figure supplement 2C). This confirms that not all ECHs can produce medium chain lipids and suggests that the function of Sl-AECH1 evolved recently, presumably as a result of neofunctionalization after gene duplication (Figure 4—figure supplement 2C). AACS1 and AECH1 are evolutionarily conserved in the Solanum Medium chain acylsugars were documented in Solanum species besides cultivated tomato, including S. pennellii (Leong et al., 2019), S. nigrum (Moghe et al., 2017), as well as the more distantly related S. quitoense (Leong et al., 2020; Hurney, 2018) and S. lanceolatum (Herrera-Salgado et al., 2005). We hypothesized that evolution of AACS1 and AECH1 contributed to medium chain acylsugar biosynthesis in Solanum. As a test, we analyzed the genomes of Solanum species other than cultivated tomato. Indeed, the acylsugar related synteny containing ACS and ECH was found in both S. pennellii and S. melongena (eggplant), suggesting that the cluster assembly evolved before divergence of the tomato and eggplant lineage (Figure 5A). Figure 5 with 2 supplements see all Download asset Open asset AACS1 and AECH1 are evolutionarily conserved in Solanum plants. (A) A conserved syntenic genomic region containing AACS1 and AECH1 was found in three selected Solanum species. Nodes representing estimated dates since the last common ancestors (Särkinen et al., 2013) shown on the left. The closest homologs of AACS1 and AECH1 in Solanum quitoense are shown without genomic context because the genes were identified from RNA-seq and genome sequences are not available. The lines connect genes representing putative orthologs across the four species. The trichome/stem RNA-seq data of two biological S. pennellii replicates are summarized (Supplementary file 2) for genes in the syntenic region. The red-blue color gradient provides a visual marker to rank the expression levels in FPKM. Structures of representative medium chain acylsugars from S. quitoense (acylinositol, I4:26) (Hurney, 2018) and S. pennellii (acylglucose, G3:19) (Leong et al., 2019) are on the right. Figure 5—figure supplement 1 shows that stable Sp-AACS1 transformation of the M82 CRISPR mutant sl-aacs1 restores C12 containing acylsugars (B) CRISPR/Cas9-mediated gene knockout of Sp-AACS1 or Sp-AECH1 in S. pennellii produce no detectable medium chain containing acylsugars. The ESI+ mode LC/MS extracted ion chromatograms of trichome metabolites are shown for each mutant. The m/z 127.01 (left panel) corresponds to the glucopyranose ring fragment that both acylsucroses and acylglucoses generate under high collision energy positive-ion mode. The m/z 155.14 (center panel) and 183.17 (right panel) correspond to the acylium ions from acylsugars with chain length of C10 and C12, respectively. Figure 5—figure supplement 2A–C describes the design of the gRNAs and the detailed information of gene edits. (C) Silencing Sq-AACS1 (Sq-c34025) or Sq-AECH1 (Sq-c37194) in S. quitoense using VIGS leads to reduction of total acylsugars. The peak area/internal standard (IS) normalized by leaf dry weight was shown from sixteen plants ± SEM. ***p<0.001, Welch two-sample t test. Figure 5—figure supplement 2E and F describes the VIGS experimental design and the representative LC/MS extracted ion chromatograms of S. quitoense major acylsugars. (D) Reduced gene expression of Sq-AACS1 or Sq-AECH1 correlates with decreased acylsugar levels in S. quitoense. The qRT-PCR gene expression data are plotted with acylsugar levels of the same leaf as described in Figure 5—figure supplement 2E. Figure 5—source data 1 includes raw data for the S. quitoense VIGS experiments. Figure 5—source data 1 Data used to make Figure 5C and D. Silencing Sq-AACS1 or Sq-AECH1 in S. quitoense using VIGS leads to reduction of total acylsugars. https://cdn.elifesciences.org/articles/56717/elife-56717-fig5-data1-v2.xlsx Download elife-56717-fig5-data1-v2.xlsx We applied gene expression and genetic approaches to test the in vivo functions of ACS and ECH in selected Solanum species. To explore the expression pattern of S. pennellii ACS and ECH cluster genes, we performed RNA-seq analysis on trichomes and shaved stems to identify acylsugar biosynthetic candidates (Supplementary file 2). The expression pattern of S. pennellii cluster genes is strikingly similar to S. lycopersicum: one ECH and two ACS genes are highly enriched in trichomes, including the orthologs of Sl-AACS1 and Sl-AECH1. Sp-AACS1 function (Sopen07g023200) was first tested by asking whether it can reverse the cultivated tomato sl-aacs1 mutant acylsugar phenotype. Indeed, Sp-AACS1 restored C12 containing acylsugars in the stably transformed sl-aacs1 plants (Figure 5—figure supplement 1). To directly test Sp-AACS1 and Sp-AECH1 function, we used CRISPR-Cas9 to make single mutants in S. pennellii LA0716. No medium chain acylsugars were detected in T0 generation mutants with edits for each gene (Figure 5B and Figure 5—figure supplement 2, A and C). Similar to the ACS-annotated Solyc07g043660 cultivated tomato mutant (Figure 3—figure supplement 1D), deletion of S. pennellii ortholog Sopen07g023220 has no observed effects on S. pennellii trichome acylsugars (Figure 5—figure supplement 2D). The medium chain acylsugar producer S. quitoense (Hurney, 2018) was used for AACS1 and AECH1 fun

Plants make hundreds of thousands of biologically active specialized metabolites varying widely in structure, biosynthesis and the processes that they influence. An increasing number of these compounds are documented to protect plants from harmful insects, pathogens, or herbivores, or mediate interactions with beneficial organisms including pollinators and nitrogen fixing microbes. Acylsugars – one class of protective compounds – are made in glandular trichomes of plants across the Solanaceae family. While most described acylsugars are acylsucroses, published examples also include acylsugars with hexose cores. The South American fruit crop Solanum quitoense (Naranjilla) produces acylsugars that contain a myo-inositol core. We identified an enzyme that acetylates triacylinositols, a function homologous to the last step in the Solanum lycopersicum acylsucrose biosynthetic pathway. Our analysis reveals parallels between S. lycopersicum acylsucrose and S. quitoense acylinositol biosynthesis, suggesting a common evolutionary origin. Material availability The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (www.plantphysiol.org) is: Robert L. Last (lastr@msu.edu). One sentence summary Evidence that the final step in Solanum quitoense acylinositol biosynthesis evolved from an acylsucrose acetyltransferase enzyme.

Plants produce phylogenetically and spatially restricted, as well as structurally diverse specialized metabolites via multistep metabolic pathways. Hallmarks of specialized metabolic evolution include enzymatic promiscuity and recruitment of primary metabolic enzymes and examples of genomic clustering of pathway genes. Solanaceae glandular trichomes produce defensive acylsugars, with sidechains that vary in length across the family. We describe a tomato gene cluster on chromosome 7 involved in medium chain acylsugar accumulation due to trichome specific acyl-CoA synthetase and enoyl-CoA hydratase genes. This cluster co-localizes with a tomato steroidal alkaloid gene cluster and is syntenic to a chromosome 12 region containing another acylsugar pathway gene. We reconstructed the evolutionary events leading to this gene cluster and found that its phylogenetic distribution correlates with medium chain acylsugar accumulation across the Solanaceae. This work reveals insights into the dynamics behind gene cluster evolution and cell-type specific metabolite diversity.

Plant specialized metabolites mediate interactions between plants and the environment and have significant agronomical/pharmaceutical value. Most genes involved in specialized metabolism (SM) are unknown because of the large number of metabolites and the challenge in differentiating SM genes from general metabolism (GM) genes. Plant models like Arabidopsis thaliana have extensive, experimentally derived annotations, whereas many non-model species do not. Here we employed a machine learning strategy, transfer learning, where knowledge from A. thaliana is transferred to predict gene functions in cultivated tomato with fewer experimentally annotated genes. The first tomato SM/GM prediction model using only tomato data performs well (F-measure = 0.74, compared with 0.5 for random and 1.0 for perfect predictions), but from manually curating 88 SM/GM genes, we found many mis-predicted entries were likely mis-annotated. When the SM/GM prediction models built with A. thaliana data were used to filter out genes where the A. thaliana-based model predictions disagreed with tomato annotations, the new tomato model trained with filtered data improved significantly (F-measure = 0.92). Our study demonstrates that SM/GM genes can be better predicted by leveraging cross-species information. Additionally, our findings provide an example for transfer learning in genomics where knowledge can be transferred from an information-rich species to an information-poor one.

Acylsugars are insecticidal plant specialized metabolites produced in the Solanaceae (nightshade family). Despite having simple constituents, these compounds are unusually structurally diverse. Their structural variations in phylogenetically closely related species enable comparative biochemical approaches to understand acylsugar biosynthesis and pathway diversification. Thus far, varied, enzyme classes contributing to their synthesis were characterized in cultivated and wild tomatoes, including from core metabolism - isopropylmalate synthase (Leu) and invertase (carbon) - and a group of evolutionarily related BAND acyltransferases known as acylsucrose acyltransferases. Gene duplication and neofunctionalization of these enzymes drove acylsugar diversification both within and beyond tomato. The broad set of evolutionary mechanisms underlying acylsugar diversity in Solanaceae make this metabolic network an exemplar for detailed understanding of the evolution of metabolic form and function.

Plant specialized metabolism (SM) enzymes produce lineage-specific metabolites with important ecological, evolutionary, and biotechnological implications. Using Arabidopsis thaliana as a model, we identified distinguishing characteristics of SM and GM (general metabolism, traditionally referred to as primary metabolism) genes through a detailed study of features including duplication pattern, sequence conservation, transcription, protein domain content, and gene network properties. Analysis of multiple sets of benchmark genes revealed that SM genes tend to be tandemly duplicated, coexpressed with their paralogs, narrowly expressed at lower levels, less conserved, and less well connected in gene networks relative to GM genes. Although the values of each of these features significantly differed between SM and GM genes, any single feature was ineffective at predicting SM from GM genes. Using machine learning methods to integrate all features, a prediction model was established with a true positive rate of 87% and a true negative rate of 71%. In addition, 86% of known SM genes not used to create the machine learning model were predicted. We also demonstrated that the model could be further improved when we distinguished between SM, GM, and junction genes responsible for reactions shared by SM and GM pathways, indicating that topological considerations may further improve the SM prediction model. Application of the prediction model led to the identification of 1,220 A. thaliana genes with previously unknown functions, each assigned a confidence measure called an SM score, providing a global estimate of SM gene content in a plant genome.

Plants produce a myriad of taxonomically restricted specialized metabolites. This diversity-and our ability to correlate genotype with phenotype-makes the evolution of these ecologically and medicinally important compounds interesting and experimentally tractable. Trichomes of tomato and other nightshade family plants produce structurally diverse protective compounds termed acylsugars. While cultivated tomato (Solanum lycopersicum) strictly accumulates acylsucroses, the South American wild relative Solanum pennellii produces copious amounts of acylglucoses. Genetic, transgenic, and biochemical dissection of the S. pennellii acylglucose biosynthetic pathway identified a trichome gland cell-expressed invertase-like enzyme that hydrolyzes acylsucroses (Sopen03g040490). This enzyme acts on the pyranose ring-acylated acylsucroses found in the wild tomato but not on the furanose ring-decorated acylsucroses of cultivated tomato. These results show that modification of the core acylsucrose biosynthetic pathway leading to loss of furanose ring acylation set the stage for co-option of a general metabolic enzyme to produce a new class of protective compounds.

The diversity of life on Earth is a result of continual innovations in molecular networks influencing morphology and physiology. Plant specialized metabolism produces hundreds of thousands of compounds, offering striking examples of these innovations. To understand how this novelty is generated, we investigated the evolution of the Solanaceae family-specific, trichome-localized acylsugar biosynthetic pathway using a combination of mass spectrometry, RNA-seq, enzyme assays, RNAi and phylogenetics in non-model species. Our results reveal that hundreds of acylsugars are produced across the Solanaceae family and even within a single plant, revealing this phenotype to be hyper-diverse. The relatively short biosynthetic pathway experienced repeated cycles of innovation over the last 100 million years that include gene duplication and divergence, gene loss, evolution of substrate preference and promiscuity. This study provides mechanistic insights into the emergence of plant chemical novelty, and offers a template for investigating the ∼300,000 non-model plant species that remain underexplored.

Specialized metabolic enzymes and metabolite diversity evolve through a variety of mechanisms including promiscuity, changes in substrate specificity, modifications of gene expression and gene duplication. For example, gene duplication and substrate binding site changes led to the evolution of the glucosinolate biosynthetic enzyme, AtIPMDH1, from a Leu biosynthetic enzyme. BAHD acyltransferases illustrate how enzymatic promiscuity leads to metabolite diversity. The examples 4-coumarate:CoA ligase and aromatic acid methyltransferases illustrate how promiscuity can potentiate the evolution of these specialized metabolic enzymes.

The diversity of life on Earth is a result of continual innovations in molecular networks influencing morphology and physiology. Plant specialized metabolism produces hundreds of thousands of compounds, offering striking examples of these innovations. To understand how this novelty is generated, we investigated the evolution of the Solanaceae family-specific, trichome-localized acylsugar biosynthetic pathway using a combination of mass spectrometry, RNA-seq, enzyme assays, RNAi and phylogenomics in different non-model species. Our results reveal hundreds of acylsugars produced across the Solanaceae family and even within a single plant, built on simple sugar cores. The relatively short biosynthetic pathway experienced repeated cycles of innovation over the last 100 million years that include gene duplication and divergence, gene loss, evolution of substrate preference and promiscuity. This study provides mechanistic insights into the emergence of plant chemical novelty, and offers a template for investigating the ~300,000 non-model plant species that remain underexplored.