This is the accepted version of the article: Barja, M. Victoria; Ezquerro, Miguel; Beretta, Stefano; [et al.]. «Several geranylgeranyl diphosphate synthase isoforms supply metabolic substrates for carotenoid biosynthesis in tomato». New Phytologist, (March 2021). DOI 10.1111/nph.17283 This version is available at https://ddd.uab.cat/record/238219 under the terms of the license Accepted Article DR ALISDAIR FERNIE (Orcid ID : 0000-0001-9000-335X) PROF. MANUEL RODRIGUEZ-CONCEPCION (Orcid ID : 0000-0002-1280-2305) Article type : - Regular Manuscript Several geranylgeranyl diphosphate synthase isoforms supply metabolic substrates for carotenoid biosynthesis in tomato M. Victoria BARJA1,a, Miguel EZQUERRO1,a, Stefano BERETTA1, Gianfranco DIRETTO2, Igor FLOREZ-SARASA1, Elisenda FEIXES1, Alessia FIORE2, Rumyana KARLOVA3, Alisdair R. FERNIE4, Jules BEEKWILDER5, Manuel RODRIGUEZ-CONCEPCION1,6,* 1 Centre for Research in Agricultural Genomics (CRAG) CSIC-IRTA-UAB-UB, Campus UAB Bellaterra, 08193 Barcelona, Spain. 2 Italian National Agency for New Technologies, Energy, and Sustainable Development, Casaccia Research Centre, 00123 Rome, Italy. 3 Laboratory of Plant Physiology, Wageningen University and Research, 6700AA Wageningen, The Netherlands. 4 Max-Planck-Institut für Molekulare Pflanzenphysiologie, 14476 Potsdam-Golm, Germany. 5 BU Bioscience, Wageningen University and Research, 6700AA Wageningen, The Netherlands 6 Institute for Plant Molecular and Cell Biology (IBMCP), CSIC-Universitat Politècnica de València, 46022 Valencia, Spain. a, These authors equally contributed to the work. (*) Corresponding author: Manuel RODRIGUEZ-CONCEPCION, Email: manuelrc@ibmcp.upv.es This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the Version of Record. Please cite this article as doi: 10.1111/NPH.17283 This article is protected by copyright. All rights reserved ORCID numbers: Accepted Article M. Victoria BARJA: 0000-0002-3846-4885 Miguel EZQUERRO: 0000-0002-3051-5502 Stefano BERETTA: none yet Gianfranco DIRETTO: 0000-0002-1441-0233 Igor FLOREZ-SARASA: 0000-0002-1862-7931 Elisenda FEIXES: 0000-0003-0285-1979 Alessia FIORE: none yet Rumyana KARLOVA: 0000-0003-0230-6428 Alisdair R. FERNIE: 0000-0001-9000-335X Jules BEEKWILDER: 0000-0003-3238-4427 Manuel RODRIGUEZ-CONCEPCION: 0000-0002-1280-2305 Received: 22 July 2020 Accepted: 8 February 2021 This article is protected by copyright. All rights reserved SUMMARY Accepted Article  Geranylgeranyl diphosphate (GGPP) produced by GGPP synthase (GGPPS) serves as a precursor for many plastidial isoprenoids, including carotenoids. Phytoene synthase (PSY) converts GGPP into phytoene, the first committed intermediate of the carotenoid pathway.  Here we used biochemical, molecular, and genetic tools to characterize the plastidial members of the GGPPS family in tomato (Solanum lycopersicum) and their interaction with PSY isoforms.  The three tomato GGPPS isoforms found to localize in plastids (SlG1, 2 and 3) exhibit similar kinetic parameters. Gene expression analyses showed a preferential association of individual GGPPS and PSY isoforms when carotenoid biosynthesis was induced during root mycorrhization, seedling deetiolation and fruit ripening. SlG2, but not SlG3, physically interacts with PSY proteins. By contrast, CRISPR-Cas9 mutants defective in SlG3 showed a stronger impact on carotenoid levels and derived metabolic, physiological and developmental phenotypes than those impaired in SlG2. Double mutants defective in both genes could not be rescued.  Our work demonstrates that the bulk of GGPP production in tomato chloroplasts and chromoplasts relies on two cooperating GGPPS paralogs, unlike other plant species such as Arabidopsis thaliana, rice or pepper, which produce their essential plastidial isoprenoids using a single GGPPS isoform. KEYWORDS carotenoids, geranylgeranyl diphosphate (GGPP), prenyltransferase, ripening, synthase, tomato This article is protected by copyright. All rights reserved INTRODUCTION Accepted Article Isoprenoids are essential biological molecules in all living organisms. In particular, plants are the main source of the enormous structural and functional variety that characterizes this family of compounds (Pulido et al., 2012; Tholl, 2015). The building blocks for the biosynthesis of all isoprenoids are isopentenyl diphosphate (IPP) and its double-bond isomer dimethylallyl diphosphate (DMAPP). These five-carbon (C5) universal isoprenoid units are produced in plants by the mevalonic acid (MVA) pathway in the cytosol and the methylerythritol 4-phosphate (MEP) pathway in plastids (Vranová et al., 2013; Rodríguez-Concepción and Boronat, 2015). Short- chain prenyltransferases subsequently condense one or more molecules of IPP to one molecule of DMAPP giving rise to C10, C15, C20 and C25 prenyl diphosphates known as geranyl diphosphate (GPP), farnesyl diphosphate (FPP), geranylgeranyl diphosphate (GGPP), and geranylfarnesyl diphosphate (GFPP), respectively. These molecules are the immediate precursors for downstream pathways leading to the production of the main groups of isoprenoids. Carotenoids are one of the most studied groups of plant isoprenoids. These C40 tetraterpenes are greatly demanded by cosmetic and agro-food industries as natural red to yellow pigments and provide benefits for human health, e.g. as precursors of vitamin A and other biologically active molecules (Sandmann, 2015; Rodriguez-Concepcion et al., 2018). In plants, carotenoids have different functions. In photosynthetic tissues, they are required for the assembly of the photosynthetic apparatus, contribute to light harvesting and are essential for photoprotection by dissipating excess light energy as heat and by scavenging reactive oxygen species. They are also fundamental in growth regulation, since they are the precursors of retrograde signals and phytohormones such as abscisic acid (ABA) and strigolactones. As a secondary role, carotenoids provide distinctive colors to flowers and fruits to attract pollinators and seed dispersal animals (Nisar et al., 2015; Yuan et al., 2015). In plants, carotenoids are produced and stored in plastids, including chloroplasts and chromoplasts (Ruiz-Sola and Rodríguez-Concepción, 2012; Sun et al., 2018). MEP-derived IPP and DMAPP are converted into GGPP by plastidial GGPP synthase (GGPPS) isoforms and then GGPP is transformed into phytoene by phytoene synthase (PSY) enzymes. The production of phytoene, the first committed intermediate of the carotenoid pathway, is considered to be a major rate-determining step regulating the metabolic flux through this pathway (Fraser et al., 2002). In tomato (Solanum lycopersicum), three PSY-encoding genes control carotenoid biosynthesis in different tissues. PSY1 expression is boosted during ripening to produce carotenoids involved in the pigmentation of the fruit (Bartley et al., 1992; Fray and Grierson, 1993; Giorio et al., 2008; Kachanovsky et al., 2012). PSY2 is expressed in all tissues, including fruits, but transcript levels are much higher than those of PSY1 in photosynthetic This article is protected by copyright. All rights reserved tissues, where carotenoids are required for photosynthesis and photoprotection (Bartley and Accepted Article Scolnik, 1993; Giorio et al., 2008). Lastly, PSY3 is mainly expressed in roots and it is induced during mycorrhization (Walter et al., 2015; Stauder et al., 2018), when carotenoid biosynthesis is up-regulated to produce strigolactones and apocarotenoid molecules essential for the establishment of the symbiosis (Fester et al., 2002, 2005; Baslam et al., 2013; Ruiz-Lozano et al., 2016; Stauder et al., 2018). Whether the corresponding PSY isoforms use GGPP supplied by different GGPPS isoforms remains unknown. Several GGPP synthase (GGPPS) paralogs have been retained in plants during evolution (Beck et al., 2013; Zhang et al., 2015; Ruiz-Sola et al., 2016a, 2016b; Zhou et al., 2017; Wang et al., 2018). However, a single GGPPS isoform appears to produce the GGPP substrate needed for the production of carotenoids and other plastidial isoprenoids in the three plant species whose GGPPS families have been best characterized to date: Arabidopsis thaliana, rice (Oryza sativa) and pepper (Capsicum annuum) (Ruiz-Sola et al., 2016a, 2016b; Zhou et al., 2017; Wang et al., 2018). While tomato has become one of the best plant systems to study the biosynthesis of carotenoids and its regulation, we still have an incomplete picture of the GGPPS family in this plant. Recent work has determined that five genes encoding GGPPS homologs exist in the tomato genome, three of which were confirmed to produce GGPP in vitro and localize in plastids (Zhou and Pichersky, 2020). Which of these plastidial GGPPS isoforms are required for the production of carotenoids in photosynthetic tissues (e.g. for photoprotection), fruits (e.g. for pigmentation) or roots (e.g. for mycorrhization) remains unknown. Here we characterized the in vivo role of these plastidial GGPPS enzymes and provide clues to understand how the supply of plastidial GGPP for the synthesis of carotenoids with different biological functions in particular tomato tissues is regulated in this important crop plant. MATERIALS AND METHODS Plant material Tomato (Solanum lycopersicum var. MicroTom) plants were used for most experiments. Seed germination, plant growth and sample collection were carried out as described (Method S1). A. tumefaciens GV3101 strain was used to stably transform tomato MicroTom cotyledons with plasmids harboring two sgRNAs to disrupt SlG2 and SlG3 genomic sequences as described (Fernandez et al., 2009). The sgRNAs were designed for each gene to create short deletions using the CRISPR P 2.0 online tool (http://crispr.hzau.edu.cn/CRISPR2/) (Liu et al., 2017). The cloning of the sgRNA sequences was performed as described (Schiml et al., 2016) using a pDE- This article is protected by copyright. All rights reserved Cas9 plasmid providing kanamycin resistance (Method S2). Primers and cloning steps are Accepted Article detailed in Tables S1 and S2, respectively. In vitro regenerated T1 lines were identified based on kanamycin resistance (100 µg/ml), PCR genotyping and restriction analyses. Homozygous T2 lines lacking Cas9 were obtained after segregation. Stable T3 offspring was used for further experiments. Method S2 and Tables S1 and S2 describe the generation of the rest of the constructs. Nicotiana benthamiana plants were grown and used for transient expression assays (agroinfiltration) as previously described (Llorente et al., 2020). Gene Co-expression Network (GCN) analyses GCN analyses were performed as previously described (Ahrazem et al., 2018). Pairwise Pearson correlations between each GGPPS gene and each selected isoprenoid biosynthetic input gene were computed for leaf and fruit tissues throughout their development and Fisher’s Z- transformation was used to test their statistical significance. RNA analyses RNA isolation, cDNA synthesis, and RT-qPCR analyses were carried out as described (Method S3). Normalized transcript abundances were calculated as described previously (Simon, 2003) using tomato ACT4 (Solyc04g011500) or EXP (Solyc07g025390) as endogenous reference genes. Three biological replicates of cDNA samples from roots of non-mycorrhized and mycorrhized tomato plants (Ruiz-Lozano et al., 2016) were kindly provided by Dr. Juan Antonio López-Ráez. Protein analyses In vitro GGPPS activity determination was performed as described (Method S4). Purified enzymes were used to calculate kinetic parameters as described (Barja and Rodríguez- Concepción, 2020). Protein concentration was determined according to Bradford method (Bradford, 1976). GGPPS activity assays in E. coli were carried out as described (Beck et al., 2013). Subcellular localization assays were performed by Agrobacterium tumefaciens-mediated transient expression in N. benthamiana leaves (Sparkes et al., 2006). Leaves were co-infiltrated with strains carrying appropriate constructs (Method S2) and a HC-Pro silencing suppressor (Goytia et al., 2006) as described (Method S5). Subcellular localization of GFP fusion proteins was determined three days post-infiltration with an Olympus FV 1000 confocal laser-scanning This article is protected by copyright. All rights reserved microscope (Method S5). Co-immunoprecipitation (Co-IP) assays were performed in N. Accepted Article benthamiana leaves as described (Muñoz and Castellano, 2018) (Method S6). Immunoblot analyses was performed as described (Pulido et al., 2013). Metabolite analysis Detection of prenyl diphosphates was carried out as described (Ruiz-Sola et al., 2016b). Carotenoids, chlorophylls and tocopherols were extracted as described (Method S7). Separation and detection was next performed using an Agilent 1200 series HPLC system (Agilent Technologies) as previously reported (Fraser et al., 2000). ABA levels were determined as described (Diretto et al., 2020). Primary metabolites were extracted, annotated and quantified as described (Llorente et al., 2020). RESULTS SlG1, SlG2 and SlG3 are GGPP-producing plastidial enzymes with similar kinetic properties. Several genes encoding proteins with homology to GGPPS enzymes are found in the tomato genome (Ament et al., 2006; Fraser et al., 2007; Stauder et al., 2018; Zhou and Pichersky, 2020). From these, three have been found to localize in plastids and produce GGPP in vitro, namely GGPPS1 (Solyc11g011240), GGPPS2 (Solyc04g079960) and GGPPS3 (Solyc02g085700), herein referred to as SlG1, SlG2 and SlG3 (Table S3). We confirmed the plastidial targeting of these three isoforms by expressing constructs encoding GFP fusions of the full-length SlG1-3 proteins in agroinfiltrated tobacco (Nicotiana benthamiana) leaves. In all three cases, fluorescence corresponding to the GFP fusion proteins co-localized with chlorophyll autofluorescence (Figure S1), supporting the conclusion that they are all efficiently targeted to chloroplasts. We also experimentally confirmed the ability of purified SlG1-3 proteins to produce GGPP in vitro. The three tomato isoforms were expressed in Escherichia coli cells without their predicted plastid-targeting sequences (Figure S2) and whole-cell protein extracts were directly used for activity assays in the presence of IPP and DMAPP followed by the analysis of the reaction products by LC-MS (Figure S3). As positive and negative controls, we used the Arabidopsis AtG11 (active) and AtG11s (inactive) proteins (Ruiz-Sola et al., 2016b). This experiment confirmed that SlG1, SlG2, SlG3 and AtG11 (but no AtG11s) produced only GGPP (Figure S3A), in agreement with recently reported data (Zhou and Pichersky, 2020). To gain new knowledge on the biochemical properties of these enzymes, we used purified proteins to This article is protected by copyright. All rights reserved calculate their kinetic parameters. Enzymatic assays performed as described (Barja and Accepted Article Rodríguez-Concepción, 2020) showed that all tested GGPPS proteins exhibited a similar optimal pH around 7.5 (Figure S3B), as expected for stromal enzymes (Höhner et al., 2016). The parameters Km (an estimator of the apparent affinity for the IPP and DMAPP substrates) and Vmax exhibited very similar values among the three tomato enzymes (Table 1). They are also similar to those obtained for AtG11 here and elsewhere (Wang and Dixon, 2009; Camagna et al., 2019). We therefore conclude that tomato SlG1, SlG2 and SlG3 and Arabidopsis AtG11 are plastidial GGPPS enzymes with very similar kinetic properties. Gene expression profiles suggest a major role of SlG2 and SlG3 in chloroplasts and chromoplasts Analysis of public gene expression databases showed that the genes encoding SlG1-3 enzymes were expressed in roots, leaves and flowers (Figure S4). Of these, the most highly expressed gene is SlG3 followed by SlG2, while SlG1 transcripts are present at very low levels. SlG2 and SlG3, but not SlG1, are also expressed at high levels in fruit pericarp and seed tissues (Figure S4). As an initial approach to get an insight on the possible functions of these individual isoforms, we performed a gene co-expression network (GCN) analysis. This is a powerful tool to infer biological functions that we previously used to identify AtG11 as the main GGPPS isoform for plastidial isoprenoid production in Arabidopsis (Ruiz-Sola et al., 2016a). By using publicly available databases for plant comparative genomics (PLAZA 4.0, Phytozome), we searched for tomato homologs of the plastidial pathways that supply GGPPS substrates (MEP pathway) and consume GGPP to produce carotenoids, chlorophylls, tocopherols, phylloquinone, plastoquinone, gibberellins, strigolactones and ABA (Table S4). We retrieved their expression data from TomExpress (Zouine et al., 2017) experiments carried out using either leaf or fruit samples at different developmental stages (Table S5). Then, we calculated their correlation with SlG1, SlG2 and SlG3 expression using pairwise Pearson correlations. The results of the GCN analyses are shown in Figure 1 and Figure S5, and correlations are listed in Table S6. It was not possible to obtain correlation data for tomato roots since only two experiments using root samples are deposited in the TomExpress database. In leaves and fruits, SlG1 was poorly co-expressed with the query genes. By contrast, and similar to that observed with AtG11 (Ruiz-Sola et al., 2016a), SlG2 and, to a lower extent, SlG3 were highly connected to plastidial isoprenoid biosynthetic genes in leaf tissues. Connectivity was lower in fruit, and in this case it was a bit higher for SlG3 (Figure 3). These results suggest that SlG2 and SlG3 might be the main GGPP-producing isoforms in leaf chloroplasts and fruit chromoplasts. This article is protected by copyright. All rights reserved In tomato, carotenoids contribute to mycorrhizal associations, photoprotection, and fruit Accepted Article pigmentation and, hence, the levels of these GGPP-derived metabolites increase during root mycorrhization, seedling de-etiolation, and fruit ripening. In agreement with the rate-determining role of PSY for carotenoid synthesis (Fraser et al., 2002), the expression levels of PSY-encoding genes also increase during such carotenoid-demanding developmental processes. By using real- time quantitative PCR (qPCR) analysis, we experimentally confirmed the up-regulation of PSY1 during fruit ripening and PSY3 in mycorrhized roots (Figure 2). Furthermore, we found that the PSY2 gene was more strongly upregulated than PSY1 during tomato seedling de-etiolation (Figure 2). Using the same samples, we observed that only SlG1 was upregulated during root mycorrhization, similarly to the expression pattern observed for PSY3 (Figure 2). During fruit ripening, SlG2 and, to a lower extent, SlG3 were up-regulated, but not as much as PSY1 (Figure 2). SlG2 was also the most strongly upregulated GGPPS-encoding gene during seedling de- etiolation, paralleling PSY2 induction. Interestingly, SlG3 and PSY1 were also induced with a similar profile during this process, even though induction levels were much lower than those observed for SlG2 and PSY2 (Figure 2). Together, these data suggest that SlG1 might provide GGPP for PSY3 to produce carotenoids in roots, particularly when needed during mycorrhization, whereas both SlG2 and SlG3 would be required in leaves and fruits to support carotenoid production for photosynthesis (mostly via PSY2) and fruit pigmentation (via PSY1). SlG2, but not SlG3, can interact with PSY1 and PSY2 A coordinated role for SlG1 and PSY3 in mycorrhization has been already proposed (Stauder et al., 2018) but the possible connection between the other plastidial GGPPS and PSY isoforms remains unclear. GGPPS proteins can physically interact with PSY and other enzymes catalyzing both upstream and downstream biosynthetic steps in the plastids of different plant species (Dogbo and Camara, 1987; Camara, 1993; Maudinas et al., 1977; Fraser et al., 2000; Ruiz-Sola et al., 2016a; Zhou et al., 2017; Camagna et al., 2019; Wang et al., 2018). This mechanism may facilitate channeling of precursors towards specific groups of plastidial isoprenoids. Protein complexes containing both GGPPS and PSY enzymes were isolated from tomato chloroplasts and fruit chromoplasts (Maudinas et al., 1977; Fraser et al., 2000), but the specific isoforms forming these protein complexes were never identified. Given the co-regulation of SlG2 and SlG3 with PSY1 and PSY2 genes in chloroplasts (i.e. photosynthetic tissues) and chromoplasts (i.e. fruits), we decided to test possible interactions of these isoforms in co-immunoprecipitation assays (Figure 3). Constructs harboring C-terminal Myc-tagged GGPPS and HA-tagged PSY sequences were combined and transiently co-expressed in N. benthamiana leaves. As negative This article is protected by copyright. All rights reserved control, we used a Myc-tagged version of Arabidopsis phosphoribulokinase (PRK-Myc), a stromal Accepted Article enzyme of the Calvin cycle. Both PSY1-HA and PSY2-HA could be co-immunoprecipitated with SlG2-Myc, suggesting that they are present in the same complexes in vivo (Figure 3). By contrast, none of these PSY isoforms could be detected in the samples co-immunoprecipitated with either SlG3-Myc or PRK-Myc. The same Myc-tagged SlG2 and SlG3 proteins used in these experiments were able to co-immunoprecipitate their HA-tagged counterparts (Figure 3). This result, consistent with the ability of GGPPS proteins to form homodimers and also heterodimers, confirms that the observed lack of interaction of SlG3 with PSY enzymes was not due to SlG3- Myc having lost its capacity to interact with other proteins. Loss of function mutants defective in SlG3, but not those impaired in SlG2, show lower levels of photosynthetic pigments and activity To further explore the biological roles of SlG2 and SlG3, we generated CRISPR-Cas9 mutants defective in these enzymes (Figure 4). We designed two single guide RNAs (sgRNA) for each gene with the aim of creating deletions encompassing unique restriction sites for rapid screening (Figure 4A). Two independent deletion alleles that created premature translation stop codons were selected for each gene and named slg2-1, slg2-2, slg3-1 and slg3-2 (Figure 4A) (Figures S6-S8). To confirm that the truncated proteins lacked GGPPS activity, we tested them in E. coli strains that synthesize the red carotenoid lycopene only when a source of GGPP is supplied (Ruiz-Sola et al., 2016b). Transformation with constructs harboring the mutant enzymes did not produce more lycopene than empty plasmid controls, indicating that they lack GGPPS activity. (Figure 4B). Once confirmed that the selected mutant alleles produced non-functional proteins, homozygous lines without Cas9 were obtained and used for further experiments. The most obvious phenotype among the selected lines was the pale color of slg3 mutants compared to slg2 alleles or azygous (WT) plants (Figure 5). This phenotype was clear in emerging and young leaves but it weakened as leaves grew and became mature (Figure 5A). The pale color correlated with significantly reduced levels of carotenoids and chlorophylls in young leaves of sgl3-1 and sgl3-2 lines compared to those of WT plants (Figure 5B) (Table S7). The differences were less clear in the case of tocopherols, another group of GGPP-derived plastidial isoprenoids (Figure 5B). Similar levels of carotenoids, chlorophylls and tocopherols were detected in mature leaves of WT, slg2 and slg3 plants (Figure 5B) (Table S7). To test whether the reduced accumulation of photosynthesis-related isoprenoids in slg3 lines had an impact on photosynthesis, we quantified effective quantum yield of photosystem II (ɸPSII) in both This article is protected by copyright. All rights reserved young and mature leaves (Figure 5C). A 30% reduction in ɸPSII was observed in young leaves Accepted Article from slg3 plants compared to those of WT or slg2 lines, consistent with the slg3-specific reduction of GGPP-derived metabolites. Despite similar levels of photosynthetic pigments accumulated in the mature leaves of all genotypes tested, ɸPSII was slightly reduced in some mutants relative to WT lines (Figure 5C). We further explored possible effects that the loss of SlG2 or SlG3 function might have on other metabolic pathways using the same samples of young leaves used for isoprenoid and ɸPSII determination (Figure 6). GC-MS metabolite profiling showed strongly decreased levels of sucrose, glucose and fructose in SlG3-defective leaves, likely due to photosynthetic impairment. Mutant slg3 leaves also displayed increased levels of amino acids derived from glycerate (Ser and Gly), shikimate (Phe, Trp and Tyr), pyruvate (Val, Ile and Ala), 2-oxoglutarate (Glu, Orn, His and GABA) and malate (Asp, Asn, Lys, Thr, Met, homoserine and beta-alanine). In line with some of these amino acid changes, SlG3-defective leaves displayed altered accumulation of tricarboxylic acid cycle-related intermediates (citrate and 2-oxoglutarate). Only a few common changes were detected in both slg2 and slg3 leaves. They included a decrease in putrescine and ascorbate levels (more pronounced in slg3 leaves), as well as an altered accumulation of metabolites produced by the plastidial shikimate pathway, including the above mentioned aromatic amino acids and phenylpropanoid derivatives such as caffeate and 3-caffeoyl-quinate (Figure 6). The levels of the carotenoid-derived hormone ABA were similar in WT and mutant samples (Figure 6 and Table 2). Ripening-associated fruit pigmentation is altered in slg2 and slg3 mutants in correlation with their carotenoid profile Lines with reduced levels of plastidial GGPPS activity also showed alterations in reproductive development (Figure 7). Flowering time was similar in WT, slg2 and slg3 plants (Figure 7A). However, pigmentation changes associated to fruit ripening were visually delayed in mutant fruits (Figure 7B). Tomato fruits reach their final size at the mature green (MG) stage and then start to ripe. The first visual symptoms of ripening define the breaker (B) stage, when chlorophyll degradation and carotenoid biosynthesis change the fruit color from green to yellow (Figure 7C). As ripening advances, accumulation of orange and red carotenoids (β-carotene and lycopene, respectively) progressively change the fruit color and define the orange (O) and eventually red (R) stages (Figure 7C). The time from anthesis to B was similar in WT and SlG2-defective fruits but it was longer in the slg3 mutants (Figure 7B) (Figure S9). Fruits from lines defective in SlG3, This article is protected by copyright. All rights reserved but also those defective in SlG2, showed a pigmentation delay in the transition from B to O. The Accepted Article delay was observed both on-vine (i.e. in fruits attached to the plant) and off-vine (i.e. in fruits detached from the plant at the B stage) (Figure 7B) (Figure S9). Both on-vine and off-vine measurements revealed that slg2 mutants also took longer to reach the R stage compared to WT fruits (Figure S9), whereas slg3 mutants did not reach a proper R stage as they developed a dark-orange color when ripe and never turned fully red (Figure 7C). WT and mutant fruits showed similar levels of carotenoids, chlorophylls and tocopherols at the MG stage (Figure S10), but clear differences were detected in ripe fruits at B+10, i.e. 10 days after B (Figures 6 and 7D) (Table S7). Phytoene and lycopene were decreased in all mutants, although the impact was higher in the case of slg3 fruits. No significant differences were found for -carotene, although the levels of this orange carotenoid tended to be higher in slg3 mutants. This, together with the lower levels of the red carotenoid lycopene may explain the dark orange color of B+10 slg3 fruits (Figure 7C). Tocopherols also showed a trend towards higher abundance in SlG3-deficient fruits, a change that was statistically significant in the slg3-1 allele (Figure 7D) or when slg3-1 and slg3-2 samples were considered together (Figure 6). Unlike that observed in young leaves, ABA levels were reduced in B+10 fruits of slg2 and, most strongly, slg3 mutants compared to WT controls (Figure 6 and Table 2). At the level of primary metabolites, B+10 fruits from both slg2 and slg3 mutants exhibited increased levels of raffinose, galacturonate, pyruvate and Asp, and lower levels of Ser, Gly, Tyr, Val, Ala, Glu and GABA than WT controls (Figure 6). The changes in these metabolites were typically stronger in the case of slg3 fruits, paralleling that observed for carotenoids and derived ABA levels. Double mutants defective in both SlG2 and SlG3 are not viable To assess the impact of simultaneous disruption of both SlG2 and SlG3 genes, alleles slg2-2 and slg3-1 were crossed using the former as female parent and the latter as male parent or vice versa. Double heterozygous F1 plants from each cross were allowed to self-pollinate and the resulting seeds were used to screen the F2 population for double homozygous plants, which were expected to occur at a Mendelian frequency of 6.25% (1 in 16). We performed two rounds of screening. In the first one, 200 seeds (100 from each cross) were plated and all of them germinated and produced green seedlings. In the second round, carried out with older seeds, 80 seeds were plated and 76 (95%) germinated (Table 3). The seeds that failed to germinate (4) were manually open and found to contain either albino/pale (3) or green (1) embryos (Figure S11). PCR genotyping of these embryos (Figure S11) and of the remaining 276 seedling did not This article is protected by copyright. All rights reserved identify double homozygous mutants (Table 3). A chi-squared goodness of fit test performed with Accepted Article 8 degrees of freedom and 95% interval of confidence confirmed that the observed genotype frequencies did not follow the expected Mendelian segregation in any of the two experiments or when considering all data together (Table 3). Besides the absence of double slg2-2 slg3-1 mutants (herein referred to as g2g2 g3g3), lines with one of the two genes in homozygosis and the second one in heterozygosis (i.e g2g2 G3g3 and G2g2 g3g3) were found at lower frequencies than predicted (Table 3), suggesting a gene dosage effect. Our interpretation of these results is that the absence of both SlG2 and SlG3 results in a lethal phenotype that is partially rescued by incorporating one copy of any of these two genes (as in g2g2 G3g3 or G2g2 g3g3 plants) and fully rescued when two copies are present in the genome (as in double heterozygous or single homozygous mutants). These results, together with the similar expression levels of both genes in developing tomato seeds (Figure S4), suggest that SlG2 and SlG3 contribute similarly and additively to embryo or/and seed development. The phenotypes of single slg3 mutants are exacerbated in lines with the SlG2 gene in heterozygosis. Plants segregating from double heterozygous F1 plants (G2g2 G3g3) that showed a single mutant genotype (i.e g2g2 G3G3 and G2G2 g3g3) or one of the two genes in homozygosis and the second one in heterozygosis (i.e g2g2 G3g3 and G2g2 g3g3) were transferred to soil and used to carefully examine their phenotype. Consistent with that described for the slg2-2 and slg3- 1 parentals (Figure 5), young leaves of g2g2 G3G3 plants showed unchanged pigmentation and WT levels of photosynthetic pigments (chlorophylls and carotenoids) and photosynthetic activity (ɸPSII), whereas those of G2G2 g3g3 plants were paler and displayed a reduction of photosynthetic pigments and activity (Figure 8). Most interestingly, the phenotypes of the slg3 mutants were intensified when one of the two genomic copies of SlG2 was inactivated in the G2g2 g3g3 line (Figure 8). Loss of a SlG3 gene copy in the slg2 mutant background, however, was not sufficient to trigger statistically significant changes in young leaves compared to WT or slg2 lines. This result indicates that a single copy of the SlG3 gene is sufficient to provide GGPP for the production of photosynthetic pigments in chloroplasts, even when no SlG2 activity is available. In the case of mature leaves, no significant differences were observed between WT and any of the mutant lines (Figure 8). At the level of fruit ripening, quantification of fruit color using the TomatoAnalyzer 4.0 tool (Gonzalo et al., 2009) confirmed the pigmentation delay previously observed in single mutants This article is protected by copyright. All rights reserved defective in SlG2 or, to a higher extent, SlG3 (Figure 7) and further showed a stronger effect Accepted Article when one of the two genomic copies of SlG2 was additionally inactivated in the slg3 background (Figure 9A). Analysis of the expression of ripening marker genes such as E8 and ACS2 (Estornell et al., 2009; Llorente et al., 2016; D’Andrea et al., 2018) showed that the peak of E8 and ACS2 expression observed at the onset of ripening (Figure S4) was reduced in the mutants (Figure 9B). Again, the stronger effect was observed in lines without SlG3 activity and tended to be higher in G2g2 g3g3 compared to G2G2 g3g3 lines (Figure 9B). DISCUSSION The fundamental basis for our knowledge of the regulation of GGPP biosynthesis in plants mainly comes from the characterization of the Arabidopsis GGPPS family (Zhu et al., 1997a, 1997b; Okada et al., 2000; Beck et al., 2013; Nagel et al., 2015; Ruiz-Sola et al., 2016a, 2016b; Wang et al., 2016). In this model plant, there are two plastid-targeted GGPPS paralogs (AtG2 and AtG11) but only AtG11 appears to be required for the production of plastidial isoprenoids (Beck et al., 2013; Nagel et al., 2015; Ruiz-Sola et al., 2016a, 2016b). The gene encoding AtG11 is ubiquitously expressed at high levels and can generate long transcripts encoding the plastid- targeted isoform but also short transcripts encoding a cytosolic enzyme that retains enzymatic activity and is essential for embryo development (Ruiz-Sola et al., 2016b). The production of GGPP has also been studied in a few crop plants (Wang and Dixon, 2009; Zhang et al., 2015; Zhou et al., 2017; Wang et al., 2018, 2019). Similar to Arabidopsis, rice and pepper contain only one enzymatically active GGPPS isoform localized in plastids, named OsGGPPS1 (OsG1 in short) and CaGGPPS1 (CaG1), respectively (Zhou et al., 2017; Wang et al., 2018). Strikingly, only scattered information was available to date on the tomato GGPPS family despite this species being a well-established model plant that accumulates high amounts of GGPP-derived metabolites of human interest such as carotenoids in fruits. Here we demonstrate that, in tomato, two plastidial isoforms (SlG2 and SlG3) coordinately supply GGPP to produce carotenoids and other isoprenoids essential for photosynthesis, fruit pigmentation, and seed viability. Subfunctionalization of plastidial GGPPS paralogs in tomato might involve several mechanism with a major role for differential gene expression. The three plastid-targeted GGPPS homologs present in tomato (SlG1-3) produce GGPP with similar kinetic parameters and an optimal pH around 7.5 (Figure S3 and Table 1). Several mechanisms might allow enzymatically similar GGPPS isoforms to acquire new functions, This article is protected by copyright. All rights reserved including (a) localization in distinct subcellular compartments, (b) specific interactions with other Accepted Article protein, and (c) diversification of spatio-temporal gene expression patterns. Despite the clear plastidial localization observed here (Figure S1) and elsewhere (Zhou and Pichersky, 2020) for GFP fusions of the SlG1-3 isoforms, we cannot exclude that shorter extraplastidial versions of these proteins could also be produced in vivo, paralleling that observed in the case of AtG11 (Ruiz-Sola et al., 2016b). Indeed, several M residues can be found in the N-terminal region of both SlG2 and SlG3 enzymes (Figure S8); they could be used as alternative translation start sites to produce catalytically active GGPPS enzymes with an absent or shorter (i.e. dysfunctional) plastid-targeting domain. Besides localization in distinct subcellular compartments, subfunctionalization of GGPPS paralogs might also involve isoform-specific interactions with other proteins. The enzymatic properties of GGPPS proteins change to produce GPP upon heterodimerization with members of the GPP synthase small subunit type I (SSU-I) subfamily (Orlova et al., 2009; Wang and Dixon, 2009). This is what happens upon interaction of SlG1-3 enzymes with the tomato SSU-I protein (Solyc07g064660) (Zhou and Pichersky, 2020). Multienzymatic complexes appear to be particularly important for metabolic channeling of GGPP. In particular, PSY cannot access freely diffusible GGPP or time-displaced GGPP supply by GGPPS (Camagna et al., 2019). Arabidopsis AtG11 and pepper CaG1 can directly interact with PSY proteins (Ruiz-Sola et al., 2016a; Camagna et al., 2019; Wang et al., 2018). We found that tomato SlG2, but not SlG3, is able to interact with PSY1 and PSY2 in planta (Figure 3). However, tomato SlG3 might deliver GGPP to PSY enzymes via heterodimerization with PSY-interacting SlG2 (Figure 3). An alternative possibility involves interaction with members of another catalytically-inactive SSU subfamily named type II (SSU-II). Similar to AtG11 and CaG1, OsG1 is the only GGPPS enzyme producing GGPP for carotenoid biosynthesis in rice. Strikingly, OsG1 does not interact with PSY but heterodimerizes with a SSU-II homolog, resulting in its delivery to a large protein complex in thylakoid membranes (Zhou et al., 2017). Interaction with SSU-II proteins was also shown to enhance the GGPP-producing activity of rice OsG1 but also of pepper CaG1 (Wang et al., 2018) and tomato SlG1-3 isoforms (Zhou and Pichersky, 2020). Interestingly, the pepper SSU-II protein also interacts with PSY, suggesting that binding of CaG1 to SSU-II might stimulate both its GGPPS activity and its interaction with PSY (Wang et al., 2018). It is therefore possible that heterodimerization with tomato SSU-II (Solyc09g008920) might also deliver SlG3 to PSY- containing protein complexes and enhance interaction of SlG2 with PSY isoforms. Regardless of other possible mechanisms discussed above, it appears that a major determinant defining the biological roles of plastidial GGPPS isoforms in tomato is their distinct expression This article is protected by copyright. All rights reserved profiles. Mining of public tomato gene expression databases, GCN analyses and qPCR assays Accepted Article led us to conclude that SlG1 is likely contributing to carotenoid biosynthesis in roots together with PSY3. This conclusion is supported by a recent study showing that the expression of PSY3 and SlG1 coordinately responds to tomato root mycorrhization and phosphate starvation (Stauder et al., 2018). The SlG1-PSY3 tandem might be channeling the flux of MEP-derived precursors towards the synthesis of carotenoid-derived molecules that are crucial for the establishment of symbiosis, such as strigolactones and apocarotenoids (Stauder et al., 2018). Unlike SlG1, SlG2 and SlG3 are constitutively expressed, with SlG3 being the paralog with the highest expression level in all plant tissues (Figure S4). In leaves, SlG2 is more strongly co-expressed than SlG3 with genes from photosynthesis-related isoprenoid pathways (Figure 1). This suggests that the expression of the SlG2 gene changes more than that of SlG3 to adapt to conditions requiring a readjustment of the gene expression network regulating the metabolism of isoprenoids such as carotenoids. In agreement, SlG2 was much more upregulated than SlG3 during seedling de- etiolation (Figure 2) and leaf development (Figure S4C), when an enhanced production of carotenoids and other photosynthesis-related isoprenoids contributes to assemble a functional photosynthetic machinery. SlG2 was also much more induced than SlG3 during fruit ripening, when carotenoid biosynthesis is boosted thanks to the up-regulation of the PSY1 isoform. PSY1 and SlG2, but not SlG3, are coordinately regulated by FUL and RIN transcription factors that control the expression of ripening-related genes, including many of the MEP and carotenoid pathways (Fujisawa et al., 2013, 2014). All these expression data show that SlG2 expression is more responsive to sudden demands of precursors for the production of isoprenoids, including carotenoids. In contrast, SlG3 expression is higher and does not change so much, suggesting a house-keeping role to maintain a continuous supply of GGPP in plastids for basal production of carotenoids and other isoprenoids. According to this model, SlG1 and SlG2 would help SlG3 to supply GGPP when a boost in carotenoid production is needed. The very low and restricted expression level of SlG1, however, strongly suggests that SlG2 is the main helper isoform for SlG3 in chloroplasts of cotyledons and expanding leaves and chromoplasts of ripening fruit. GGPPS isoforms SlG2 and SlG3 have functionally interchangeable roles in chloroplasts and chromoplasts. Analysis of tomato mutants defective in gene copies for SlG2 or/and SlG3 further suggested that these are functionally exchangeable isoforms that participate in the same biological processes. This article is protected by copyright. All rights reserved This might not be obvious when analyzing leaves, as only slg3 alleles were found to display Accepted Article reduced levels of GGPP-derived isoprenoids and subsequent inhibition of photosynthesis (Figures 5 and 8). However, the effects of reduced isoprenoid synthesis could be indirectly detected in slg2 leaves too. Thus, our GC-MS analysis showed higher levels of all aromatic amino acids derived from the shikimate pathway (Trp, Tyr and Phe) as well as Phe-derived phenylpropanoids caffeate (caffeic acid) and 3-caffeoyl-quinate (chlorogenic acid) in both slg2 and slg3 mutant lines (Figure 6). This might be a physiological response to cope with photooxidative stress caused by lower levels of carotenoids in the mutants, as phenylpropanoids (including Phe-derived flavonoids and anthocyanins) can also function as photoprotective metabolites (Muñoz and Munné-Bosch, 2018). Reduced levels of well-known metabolites associated with oxidative stress such as ascorbate and putrescine in leaves from both mutant lines would also support this view. Loss of one SlG3 gene copy in the slg2 mutant background failed to cause a statistically significant decrease in the levels of photosynthetic pigments or activity, even though a trend towards reduction of chlorophyll and carotenoid levels was observed (Figure 8). However, complete loss of SlG3 activity in lines with one or two functional SlG2 copies was sufficient to reduce levels of GGPP-derived photoprotective isoprenoids such as carotenoids and tocopherols to an extent that became detectable and impacted photosynthesis (Figure 5), causing sugar starvation and the subsequent metabolic changes observed only in the slg3 mutant (Figure 6). In agreement, the increased accumulation of most amino acids in slg3 leaves suggested a high proteolytic activity to generate an alternative respiratory source, a likely response to sugar starvation derived from reduced photosynthesis and/or photooxidative stress (Araújo et al., 2011; Obata and Fernie, 2012; Galili et al., 2016). The absence of any of the two individual enzymes also decreases plastidial GGPP production in fruit, as deduced from the levels of the main GGPP-derived metabolites (Figure 7D) (Table S7). Tocopherol levels did not decrease in mutant fruit, perhaps because they are mostly produced by recycling the phytyl chain released from the chlorophylls degraded during fruit ripening. By contrast, lycopene (by far the most abundant carotenoid in ripe fruit) and, to a lower extent, phytoene, showed reduced levels in both mutants (Figure 7D) (Table S7). Similar to that observed in leaves, the impact is stronger in slg3 mutants, consistent with the higher expression levels of the SlG3 compared to SlG2 in young leaves and MG fruits (Figure S4). While altered levels of 3-caffeoyl-quinate and citrate were detected only in fruit of the slg3 mutant, the rest of metabolic changes were similar in slg2 and slg3 lines (Figure 6), again supporting the conclusion that these enzymes are redundant and interchangeable. In particular, both slg2 and slg3 fruit This article is protected by copyright. All rights reserved showed pigmentation defects associated with a decreased carotenoid accumulation (Figures 7 Accepted Article and 9A). Because ABA is synthesized from carotenoids, its reduced levels in GGPPS-defective ripe fruits but not in leaves (Table 2) may be the result of a more substantial reduction in carotenoid contents in mutant fruit (Figure 7) compared to leaves (Figure 5) (Table S7). A role for ABA in promoting tomato fruit ripening has been proposed based on the analysis of mutants or external application of hormones and inhibitors. This, together with the observed down-regulation of ethylene-related ripening marker genes (E8 and ACS2) in GGPPS-defective fruit (Figure 9B) allows to speculate that reduced ABA levels in the mutant fruit may contribute to delay ripening either directly or indirectly via ethylene (Zhang et al., 2009; McQuinn et al., 2020). Additionally, metabolic roles of SlG2 and SlG3 besides their GGPPS activity in plastids might play a role in fruits but also in developing seeds, hence explaining why we could not isolate a double slg2 slg3 mutant (Table 3). The observation that the lethal phenotype is dose-dependent in an isoform- independent fashion (i.e. can be rescued by a single genomic copy of either SlG2 or SlG3) reinforces our conclusion that SlG2 and SlG3 have functionally interchangeable roles. Concluding remarks. Retention of multiple gene copies after duplication events may allow the acquisition of new functions (neofunctionalization) or partitioning the ancestral functions between duplicate partners (subfunctionalization), via evolution of coding sequence and/or regulatory regions. The work reported here demonstrates that the bulk of GGPP production in tomato leaf chloroplasts and fruit chromoplasts relies on two redundant but cooperating GGPPS paralogs, SlG2 and SlG3. Additionally, the SlG1 isoform might contribute to GGPP synthesis in root plastids. This subfunctionalization scenario contrasts with that described to date in other plant species such as Arabidopsis, rice or pepper, which produce their essential plastidial isoprenoids using a single GGPPS isoform. However, it is likely that tomato is not an exception. Examples of gene families encoding enzyme isoforms located in the same cell compartment but differing in gene expression profiles abound in the literature. They include deoxyxylulose 5-phosphate synthase (DXS) and PSY, the rate-determining enzymes of the MEP and carotenoid pathways, respectively (Walter et al., 2015). Both DXS and PSY are encoded by single genes in Arabidopsis but several differentially expressed genes in tomato. Subfunctionalization is also widespread beyond the isoprenoid pathway, contributing to the huge diversity of specialized metabolism in plants (Moghe and Last, 2015). Deciphering how different plants regulate plastidial GGPP production and channeling will be useful for future metabolic engineering approaches targeted to manipulate the This article is protected by copyright. All rights reserved accumulation of specific groups of GGPP-derived isoprenoids without negatively impacting the Accepted Article levels of others. ACKNOWLEDGEMENTS We greatly thank Juan Antonio López-Ráez for providing cDNA samples of non-mycorrhized and mycorrhized tomato roots; Ernesto Llamas for providing the pGWB417_AtPRK construct, and Albert Ferrer and Laura Gutiérrez for the pDE-Cas9 (with kanamycin resistance) plasmid. The technical support of M. Rosa Rodríguez-Goberna and all CRAG services is also appreciated. This work was funded by the European Regional Development Fund (FEDER) and the Spanish Agencia Estatal de Investigación (grants BIO2017-84041-P and, BIO2017-90877-REDT) and Generalitat de Catalunya (2017SGR-710) to MRC. Support by the collaborative European Union’s Horizon 2020 (EU-H2020) ERA-IB-2 (Industrial Biotechnology) BioProMo project to MRC (PCIN-2015-103), RK and JB (053-80-725) is also acknowledged. CRAG is financially supported by the Severo Ochoa Programme for Centres of Excellence in R&D 2016-2019 (SEV-2015-0533) and the Generalitat de Catalunya CERCA Programme. MVB was funded with a Spanish Ministry of Education, Culture and Sports PhD fellowship (FPU14/05142) and a EU-H2020 COST Action CA15136 (EuroCaroten) short-stay fellowship. ME is supported by a Spanish Agencia Estatal de Investigación (BES-2017-080652) PhD fellowship. IFS is supported by the EU-H2020 Marie S. Curie Action 753301 (Arcatom). AUTHOR CONTRIBUTIONS MVB, ME and MRC designed the research; MVB, ME, SB, GD, IFS, EF, and AF performed research; RK, ARF and JB contributed analytic tools; MVB, ME, SB, GD, IFS, EF, AF, RK, ARF, JB and MRC analyzed data; MVB and MRC wrote the paper. MVB and ME contributed equally to this work. REFERENCES Ahrazem, O., Argandoña, J., Fiore, A., Aguado, C., Luján, R., Rubio-Moraga, Á., Marro, M., Araujo-Andrade, C., Loza-Alvarez, P., Diretto, G., and Gómez-Gómez, L. (2018). Transcriptome analysis in tissue sectors with contrasting crocins accumulation provides novel insights into apocarotenoid biosynthesis and regulation during chromoplast biogenesis. Sci. Rep. 8: 1–17. This article is protected by copyright. All rights reserved Ament, K., Van Schie, C.C., Bouwmeester, H.J., Haring, M.A., and Schuurink, R.C. (2006). Accepted Article Induction of a leaf specific geranylgeranyl pyrophosphate synthase and emission of (E,E)- 4,8,12-trimethyltrideca-1,3,7,11-tetraene in tomato are dependent on both jasmonic acid and salicylic acid signaling pathways. Planta 224: 1197–1208. Araújo, W.L., Tohge, T., Ishizaki, K., Leaver, C.J., and Fernie, A.R. (2011). Protein degradation - an alternative respiratory substrate for stressed plants. Trends Plant Sci. 16: 489–498. Barja, M.V. and Rodríguez-Concepción, M. (2020). A Simple In Vitro Assay to Measure the Activity of Geranylgeranyl Diphosphate Synthase and Other Short-Chain Prenyltransferases. Methods Mol. Biol. 2083: 27–38. Bartley, G. and Scolnik, P. (1993). cDNA Cloning, Expression during Development, and Genome Mapping of a second phytoenen synthase. Biochemistry 268: 25718–25721. Bartley, G.E., Viitanen, P. V, Bacot, K.O., and Scolnik, P.A. (1992). A tomato gene expressed during fruit ripening encodes an enzyme of the carotenoid biosynthesis pathway. J. Biol. Chem. 267: 5036–9. Baslam, M., Esteban, R., García-Plazaola, J.I., and Goicoechea, N. (2013). Effectiveness of arbuscular mycorrhizal fungi (AMF) for inducing the accumulation of major carotenoids, chlorophylls and tocopherol in green and red leaf lettuces. Appl. Microbiol. Biotechnol. 97: 3119–3128. Beck, G., Coman, D., Herren, E., Ruiz-Sola, M.A., Rodríguez-Concepción, M., Gruissem, W., and Vranová, E. (2013). Characterization of the GGPP synthase gene family in Arabidopsis thaliana. Plant Mol. Biol. 82: 393–416. Bradford, M.M. (1976). A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding. Anal. Biochem. 72: 248–54. Camagna, M., Grundmann, A., Bär, C., Koschmieder, J., Beyer, P., and Welsch, R. (2019). Enzyme Fusion Removes Competition for Geranylgeranyl Diphosphate in Carotenogenesis. Plant Physiol. 179: 1013–1027. Camara, B. (1993). Plant phytoene synthase complex: Component enzymes, Immunology and Biogenesis. Methods Enzymol. 214: 352–365. D’Andrea, L., Simon-Moya, M., Llorente, B., Llamas, E., Marro, M., Loza-Alvarez, P., Li, L., and Rodriguez-Concepcion, M. (2018). Interference with Clp protease impairs carotenoid accumulation during tomato fruit ripening. J. Exp. Bot. 69: 1557–1568. Diretto, G., Frusciante, S., Fabbri, C., Schauer, N., Busta, L., Wang, Z., Matas, A.J., Fiore, A., This article is protected by copyright. All rights reserved Rose J.K.C., Fernie, A.R. et al. (2020). Manipulation of β-carotene levels in tomato fruits Accepted Article results in increased ABA content and extended shelf life. Plant Biotechnol. J. 18: 1185–1199. Dogbo, O. and Camara, B. (1987). Purification of isopentenyl pyrophosphate isomerase and geranylgeranyl pyrophosphate synthase from Capsicum chromoplasts by affinity chromatography. Biochim. Biophys. Acta (BBA)/Lipids Lipid Metab. 920: 140–148. Estornell, L.H., Orzáez, D., López-Peña, L., Pineda, B., Antón, M.T., Moreno, V., and Granell, A. (2009). A multisite gateway-based toolkit for targeted gene expression and hairpin RNA silencing in tomato fruits. Plant Biotechnol. J. 7: 298–309. Fernandez, A.I., Viron, N., Alhagdow, M., Karimi, M., Jones, M., Amsellem, Z., Sicard, A., Czerednik, A., Angenent, G., Grierson, D. et al. (2009). Flexible Tools for Gene Expression and Silencing in Tomato. Plant Physiol. 151: 1729–1740. Fester, T., Schmidt, D., Lohse, S., Walter, M.H., Giuliano, G., Bramley, P.M., Fraser, P.D., Hause, B., and Strack, D. (2002). Stimulation of carotenoid metabolism in arbuscular mycorrhizal roots. Planta 216: 148–154. Fester, T., Wray, V., Nimtz, M., and Strack, D. (2005). Is stimulation of carotenoid biosynthesis in arbuscular mycorrhizal roots a general phenomenon? Phytochemistry 66: 1781–1786. Fraser, P.D., Enfissi, E.M.A., Halket, J.M., Truesdale, M.R., Yu, D., Gerrish, C., and Bramley, P.M. (2007). Manipulation of Phytoene Levels in Tomato Fruit: Effects on Isoprenoids, Plastids, and Intermediary Metabolism. Plant Cell 19: 3194–3211. Fraser, P.D., Romer, S., Shipton, C.A., Mills, P.B., Kiano, J.W., Misawa, N., Drake, R.G., Schuch, W., and Bramley, P.M. (2002). Evaluation of transgenic tomato plants expressing an additional phytoene synthase in a fruit-specific manner. Proc. Natl. Acad. Sci. U. S. A. 99: 1092–7. Fraser, P.D., Schuch, W., and Bramley, P.M. (2000). Phytoene synthase from tomato (Lycopersicon esculentum) chloroplasts--partial purification and biochemical properties. Planta 211: 361–369. Fray, R.G. and Grierson, D. (1993). Identification and genetic analysis of normal and mutant phytoene synthase genes of tomato by sequencing, complementation and co-suppression. Plant Mol. Biol. 22: 589–602. Fujisawa, M., Nakano, T., Shima, Y., and Ito, Y. (2013). A large-scale identification of direct targets of the tomato MADS box transcription factor RIPENING INHIBITOR reveals the regulation of fruit ripening. Plant Cell 25: 371–386. Fujisawa, M., Shima, Y., Nakagawa, H., Kitagawa, M., Kimbara, J., Nakano, T., Kasumi, T., and This article is protected by copyright. All rights reserved Ito, Y. (2014). Transcriptional regulation of fruit ripening by tomato FRUITFULL homologs and Accepted Article associated MADS box proteins. Plant Cell 26: 89–101. Galili, G., Amir, R., and Fernie, A.R. (2016). The Regulation of Essential Amino Acid Synthesis and Accumulation in Plants. Annu. Rev. Plant Biol. 67: 153–178. Giorio, G., Stigliani, A.L., and D’Ambrosio, C. (2008). Phytoene synthase genes in tomato (Solanumlycopersicum L.) - New data on the structures, the deduced amino acid sequences and the expression patterns. FEBS J. 275: 527–535. Gonzalo, M.J., Brewer, M.T., Anderson, C., Sullivan, D., Gray, S., and Van Der Knaap, E. (2009). Tomato fruit shape analysis using morphometric and morphology attributes implemented in tomato analyzer software program. J. Am. Soc. Hortic. Sci. 134: 77–87. Goytia, E., Fernández-Calvino, L., Martínez-García, B., López-Abella, D., and López-Moya, J.J. (2006). Production of plum pox virus HC-Pro functionally active for aphid transmission in a transient-expression system. J. Gen. Virol. 87: 3413–23. Höhner, R., Aboukila, A., Kunz, H.-H., and Venema, K. (2016). Proton Gradients and Proton- Dependent Transport Processes in the Chloroplast. Front. Plant Sci. 7: 1–7. Kachanovsky, D.E., Filler, S., Isaacson, T., and Hirschberg, J. (2012). Epistasis in tomato color mutations involves regulation of phytoene synthase 1 expression by cis-carotenoids. Proc. Natl. Acad. Sci. U. S. A. 109: 19021–6. Liu, H., Ding, Y., Zhou, Y., Jin, W., Xie, K., and Chen, L.L. (2017). CRISPR-P 2.0: An Improved CRISPR-Cas9 Tool for Genome Editing in Plants. Mol. Plant 10: 530–532. Llorente, B., D’Andrea, L., Ruiz-Sola, M.A., Botterweg, E., Pulido, P., Andilla, J., Loza-Alvarez, P., and Rodriguez-Concepcion, M. (2016). Tomato fruit carotenoid biosynthesis is adjusted to actual ripening progression by a light-dependent mechanism. Plant J. 85: 107–19. Llorente, B., Torres-Montilla, S., Morelli, L., Florez-Sarasa, I., Matus, J.T., Ezquerro, M., D'Andrea, L., Houhou, F., Majer, E., Picó, B. et al. (2020) Synthetic conversion of leaf chloroplasts into carotenoid-rich plastids reveals mechanistic basis of natural chromoplast development. Proc. Natl. Acad. Sci. USA 117: 21796-803. Maudinas, B., Bucholtz, M.L., Papastephanou, C., Katiyar, S.S., Briedis, A. V., and Porter, J.W. (1977). The partial purification and properties of a phytoene synthesizing enzyme system. Arch. Biochem. Biophys. 180: 354–362. McQuinn, R.P., Gapper, N.E., Gray, A.G., Zhong, S., Tohge, T., Fei, Z., Fernie, A.R., and Giovannoni, J.J. (2020). Manipulation of ZDS in tomato exposes carotenoid- and ABA-specific This article is protected by copyright. All rights reserved effects on fruit development and ripening. Plant Biotechnol. J. 18: 2210-2224. Accepted Article Moghe, G.D. and Last, R.L. (2015). Something Old, Something New: Conserved Enzymes and the Evolution of Novelty in Plant Specialized Metabolism. Plant Physiol. 169: 1512–1523. Muñoz, A. and Castellano, M.M. (2018). Coimmunoprecipitation of Interacting Proteins in Plants. Methods Mol. Biol. 1794: 279–287. Muñoz, P. and Munné-Bosch, S. (2018). Photo-oxidative stress during leaf, flower and fruit development. Plant Physiol. 176: 1004–1014. Nagel, R., Bernholz, C., Vranová, E., Košuth, J., Bergau, N., Ludwig, S., Wessjohann, L., Gershenzon, J., Tissier, A., and Schmidt, A. (2015). Arabidopsis thaliana isoprenyl diphosphate synthases produce the C 25 intermediate, geranylfarnesyl diphosphate. Plant J. 84: 847–859. Nisar, N., Li, L., Lu, S., Khin, N.C., and Pogson, B.J. (2015). Carotenoid metabolism in plants. Mol. Plant 8: 68–82. Obata, T. and Fernie, A.R. (2012). The use of metabolomics to dissect plant responses to abiotic stresses. Cell. Mol. Life Sci. 69: 3225–3243. Okada, K., Saito, T., Nakagawa, T., Kawamukai, M., and Kamiya, Y. (2000). Five geranylgeranyl diphosphate synthases expressed in different organs are localized into three subcellular compartments in Arabidopsis. Plant Physiol. 122: 1045–1056. Orlova, I., Nagegowda, D.A., Kish, C.M., Gutensohn, M., Maeda, H., Varbanova, M., Fridman, E., Yamaguchi, S., Hanada, A., Kamiya, Y. et al. (2009). The small subunit of snapdragon geranyl diphosphate synthase modifies the chain length specificity of tobacco geranylgeranyl diphosphate synthase in planta. Plant Cell 21: 4002–4017. Pulido, P., Perello, C., and Rodriguez-Concepcion, M. (2012). New Insights into Plant Isoprenoid Metabolism. Mol. Plant 5: 964–967. Pulido, P., Toledo-Ortiz, G., Phillips, M. a, Wright, L.P., and Rodríguez-Concepción, M. (2013). Arabidopsis J-protein J20 delivers the first enzyme of the plastidial isoprenoid pathway to protein quality control. Plant Cell 25: 4183–4194. Quinet, M., Angosto, T., Yuste-Lisbona, F.J., Blanchard-Gros, R., Bigot, S., Martinez, J.P., and Lutts, S. (2019). Tomato Fruit Development and Metabolism. Front. Plant Sci. 10: 1554. Rodriguez-Concepcion, M., Avalos, J., Bonet, M.L., Boronat, A., Gomez-Gomez, L., Hornero- Mendez, D., Limon, M.C., Meléndez-Martínez, A.J., Olmedilla-Alonso, B., Palou, A. et al. (2018). A global perspective on carotenoids: Metabolism, biotechnology, and benefits for This article is protected by copyright. All rights reserved nutrition and health. Prog. Lipid Res. 70: 62–93. Accepted Article Rodríguez-Concepción, M. and Boronat, A. (2015). Breaking new ground in the regulation of the early steps of plant isoprenoid biosynthesis. Curr. Opin. Plant Biol. 25: 17–22. Ruiz-Lozano, J.M., Aroca, R., Zamarreño, Á.M., Molina, S., Andreo-Jiménez, B., Porcel, R., García-Mina, J.M., Ruyter-Spira, C., and López-Ráez, J.A. (2016). Arbuscular mycorrhizal symbiosis induces strigolactone biosynthesis under drought and improves drought tolerance in lettuce and tomato. Plant, Cell Environ. 39: 441–452. Ruiz-Sola, M.Á., Coman, D., Beck, G., Barja, M.V., Colinas, M., Graf, A., Welsch, R., Rütimann, P., Bühlmann, P., Bigler, L. et al. (2016a). Arabidopsis GERANYLGERANYL DIPHOSPHATE SYNTHASE 11 is a hub isozyme required for the production of most photosynthesis-related isoprenoids. New Phytol. 209: 252–264. Ruiz-Sola, M.Á., Barja, M.V., Manzano, D., Llorente, B., Schipper, B., Beekwilder, J., and Rodriguez-Concepcion, M. (2016b). A Single Arabidopsis Gene Encodes Two Differentially Targeted Geranylgeranyl Diphosphate Synthase Isoforms. Plant Physiol. 172: 1393–1402. Ruiz-Sola, M.Á. and Rodríguez-Concepción, M. (2012). Carotenoid Biosynthesis in Arabidopsis: A Colorful Pathway. Arab. B. 10: e0158. Sandmann, G. (2015). Carotenoids of Biotechnological Importance. Adv. Biochem. Eng. / Biotechnol. 148: 449–467. Schiml, S., Fauser, F., and Puchta, H. (2016). CRISPR/Cas-Mediated Site-Specific Mutagenesis in Arabidopsis thaliana Using Cas9 Nucleases and Paired Nickases. Methods Mol. Biol. 1469: 111–122. Seymour, G.B., Chapman, N.H., Chew, B.L., and Rose, J.K.C. (2013). Regulation of ripening and opportunities for control in tomato and other fruits. Plant Biotechnol. J. 11: 269–278. Simon, P. (2003). Q-Gene: Processing quantitative real-time RT-PCR data. Bioinformatics 19: 1439–1440. Sparkes, I.A., Runions, J., Kearns, A., and Hawes, C. (2006). Rapid, transient expression of fluorescent fusion proteins in tobacco plants and generation of stably transformed plants. Nat. Protoc. 1: 2019–25. Stauder, R., Welsch, R., Camagna, M., Kohlen, W., Balcke, G.U., Tissier, A., and Walter, M.H. (2018). Strigolactone Levels in Dicot Roots Are Determined by an Ancestral Symbiosis- Regulated Clade of the PHYTOENE SYNTHASE Gene Family. Front. Plant Sci. 9: 255. Sun, T., Yuan, H., Cao, H., Yazdani, M., Tadmor, Y., and Li, L. (2018). Carotenoid Metabolism in This article is protected by copyright. All rights reserved Plants: The Role of Plastids. Mol. Plant 11: 58–74. Accepted Article Tholl, D. (2015). Biosynthesis and Biological Functions of Terpenoids in Plants. Adv. Biochem. Eng. Biotechnol. 148: 63–106. Vranová, E., Coman, D., and Gruissem, W. (2013). Network analysis of the MVA and MEP pathways for isoprenoid synthesis. Annu. Rev. Plant Biol. 64: 665–700. Walter, M.H., Stauder, R., and Tissier, A. (2015). Evolution of root-specific carotenoid precursor pathways for apocarotenoid signal biogenesis. Plant Sci. 233: 1–10. Wang, C., Chen, Q., Fan, D., Li, J., Wang, G., and Zhang, P. (2016). Structural Analyses of Short-Chain Prenyltransferases Identify an Evolutionarily Conserved GFPPS Clade in Brassicaceae Plants. Mol. Plant 9: 195–204. Wang, G. and Dixon, R. a (2009). Heterodimeric geranyl(geranyl)diphosphate synthase from hop (Humulus lupulus) and the evolution of monoterpene biosynthesis. Proc. Natl. Acad. Sci. U. S. A. 106: 9914–9919. Wang, J., Lin, H.-X., Su, P., Chen, T., Guo, J., Gao, W., and Huang, L.-Q. (2019). Molecular cloning and functional characterization of multiple geranylgeranyl pyrophosphate synthases (ApGGPPS) from Andrographis paniculata. Plant Cell Rep. 38: 117–128. Wang, Q., Huang, X.-Q., Cao, T.-J., Zhuang, Z., Wang, R., and Lu, S. (2018). Heteromeric Geranylgeranyl Diphosphate Synthase Contributes to Carotenoid Biosynthesis in Ripening Fruits of Red Pepper (Capsicum annuum var. conoides). J. Agric. Food Chem. 66: 11691– 11700. Yuan, H., Zhang, J., Nageswaran, D., and Li, L. (2015). Carotenoid metabolism and regulation in horticultural crops. Hortic. Res. 2: 15036. Zhang, M., Su, P., Zhou, Y.-J., Wang, X.-J., Zhao, Y.-J., Liu, Y.-J., Tong, Y.-R., Hu, T.-Y., Huang, L.-Q., and Gao, W. (2015). Identification of geranylgeranyl diphosphate synthase genes from Tripterygium wilfordii. Plant Cell Rep. 34: 2179–2188. Zhang, M., Yuan, B., and Leng, P. (2009). The role of ABA in triggering ethylene biosynthesis and ripening of tomato fruit. J. Exp. Bot. 60: 1579–1588. Zhou, F. and Pichersky, E. (2020). The complete functional characterisation of the terpene synthase family in tomato. New Phytol. 226: 1341-1360. Zhou, F., Wang, C.-Y., Gutensohn, M., Jiang, L., Zhang, P., Zhang, D., Dudareva, N., and Lu, S. (2017). A recruiting protein of geranylgeranyl diphosphate synthase controls metabolic flux toward chlorophyll biosynthesis in rice. Proc. Natl. Acad. Sci. U. S. A. 114: 6866–6871. This article is protected by copyright. All rights reserved Zhu, X.F., Suzuki, K., Okada, K., Tanaka, K., Nakagawa, T., Kawamukai, M., and Matsuda, K. Accepted Article (1997a). Cloning and functional expression of a novel geranylgeranyl pyrophosphate synthase gene from Arabidopsis thaliana in Escherichia coli. Plant Cell Physiol. 38: 357–61. Zhu, X.F., Suzuki, K., Saito, T., Okada, K., Tanaka, K., Nakagawa, T., Matsuda, H., and Kawamukai, M. (1997b). Geranylgeranyl pyrophosphate synthase encoded by the newly isolated gene GGPS6 from Arabidopsis thaliana is localized in mitochondria. Plant Mol. Biol. 35: 331–41. Zouine, M., Maza, E., Djari, A., Lauvernier, M., Frasse, P., Smouni, A., Pirrello, J., and Bouzayen, M. (2017). TomExpress, a unified tomato RNA-Seq platform for visualization of expression data, clustering and correlation networks. Plant J. 92: 727–735. This article is protected by copyright. All rights reserved SUPPORTING INFORMATION Accepted Article FIGURES Figure S1. Subcellular localization of tomato GGPPS proteins. Figure S2. Purification of recombinant GGPPS proteins for in vitro activity assays. Figure S3. Biochemical activity of purified recombinant GGPPS proteins. Figure S4. Transcript levels of tomato genes in different tissues. Figure S5. Gene co-expression network (GCN) analysis of tomato plastidial GGPPS genes in leaf and fruit tissues. Figure S6. DNA sequence alignment of SlG2 CRISPR mutants. Figure S7. DNA sequence alignment of SlG3 CRISPR mutants. Figure S8. Protein alignments of WT and mutant SlG2 (A) and SlG3 (B) sequences. Figure S9. Fruit ripening initiation and progression in WT and mutant plants. Figure S10. Relative levels of plastidial isoprenoids in mature green fruits from WT and mutant lines. Figure S11. PCR-based genotyping of non-germinating F2 seeds from the cross of slg2-2 and slg3-1 mutant plants. TABLES Table S1. Primers used in this work. Table S2. Constructs and cloning details. Table S3. List of tomato GGPPS-like sequences. Table S4. List of plastidial isoprenoid-related genes used for the tomato GGPPS GCN analyses. Table S5. List of expression datasets in TomExpress used for the generation of the tomato GGPPS GCN in leaf and fruit tissues. Table S6. Co-expression of tomato GGPPS paralogs (guide genes) with isoprenoid-related genes (query genes) in leaf and fruit tissues. Table S7. Levels of GGPP-derived metabolites detected by HPLC. Table S8. Relative levels of metabolites detected by GC-MS in samples from WT and mutant young leaves. Table S9. Parameters used for peak annotation in leaves. This article is protected by copyright. All rights reserved Table S10. Relative levels of metabolites detected by GC-MS in samples from WT and mutant Accepted Article B+10 fruit. Table S11. Parameters used for peak annotation in B+10 fruit. METHODS Method S1. Growth conditions, sample collection and phenotypic analyses. Method S2. Constructs. Method S3. RNA isolation and cDNA synthesis. Method S4. GGPPS activity determination. Method S5. Subcellular localization assays. Method S6. Co-immunoprecipitation assays. Method S7. Isoprenoid analysis. This article is protected by copyright. All rights reserved TABLES Accepted Article Table 1. Kinetic parameters of tomato plastidial GGPPS enzymes DMAPP IPP (+100 μM IPP) (+100 μM DMAPP) Km Vmax Km Vmax (μM) (nmol•min-1•mg-1) (μM) (nmol•min-1•mg-1) SlG1 31.82 ± 2.92 47.47 ± 1.40 74.18 ± 7.55 59.87 ± 2.73 SlG2 49.55 ± 5.31 38.87 ± 1.53 79.75 ± 8.33 36.73 ± 1.73 SlG3 45.75 ± 6.81 26.13 ± 1.40 45.92 ± 4.86 29.13 ± 1.13 AtG11 32.86 ± 4.86 21.53 ± 1.07 38.49 ± 4.94 24.13 ± 1.07 . Values correspond to the mean±SD of three independent experimental replicates (n=3). Table 2. ABA levels in GGPPS-defective tomato leaves and fruit. Young leaves B+10 fruit WT 1.67 ± 0.19 0.63 ± 0.13 slg2-1 1.69 ± 0.10 0.55 ± 0.12 slg2-2 1.98 ± 0.39 0.30 ± 0.08 slg3-1 1.96 ± 0.09 0.16 ± 0.04 slg3-2 1.61 ± 0.29 0.08 ± 0.01 Values (µg/g dry weight) correspond to the mean ± SD of four independent samples (n=4). Statistically significant changes in mutants compared to WT samples (t-test, p<0.01) are indicated in bold. This article is protected by copyright. All rights reserved Table 3. Expected and observed frequencies of the F2 population from the crosses of slg2- Accepted Article 2 and slg3-1 mutant tomato plants. Round 1 Round 2 Combined Genotypes Expected # % # % # % G2g2 G3g3 25% 52 26% 15 20% 67 24% G2g2 G3G3 12.5% 26 13% 18 24% 44 16% G2G2 G3g3 12.5% 35 17.5% 10 13% 45 16% g2g2 G3g3 12.5% 18 9% 6 8% 24 9% G2g2 g3g3 12.5% 16 8% 5 7% 21 8% g2g2 g3g3 6.25% 0 0% 0 0% 0 0% g2g2 G3G3 6.25% 17 8.5% 5 7% 22 8% G2G2 g3g3 6.25% 14 7% 8 11% 22 8% G2G2 G3G3 6.25% 22 11% 9 12% 31 11% Total plants (#) 200 76 276 Chi-square 30.84 22.68 45.17 p-value 0.0002 0.0038 <0.0001 Mutant alleles are marked in red. A chi-squared goodness of fit test was performed with 8 degrees of freedom and 95% interval of confidence to check the Mendelian segregation of the mutant alleles. #, number of plants. This article is protected by copyright. All rights reserved FIGURE LEGENDS Accepted Article Figure 1. Gene co-expression analysis of tomato genes encoding plastidial GGPPS isoforms in leaf and fruit tissues. Positive co-expression relationships (𝜌≥0.55) are depicted in tissue-specific networks as edges. SlG1, SlG2 and SlG3 are depicted as central green nodes. Surrounding smaller nodes represent genes from the indicated isoprenoid pathways. Red, green and black edges indicate positive co-expression with SlG1, SlG2 and SlG3 genes, respectively. See Table S4 for gene accessions, Table S5 for leaf and fruit datasets used, and Table S6 for 𝜌 values. Figure 2. Expression profiles of genes encoding tomato PSY and GGPPS paralogs during processes involving increased carotenoid production. First column corresponds to non- mycorrhized (-) and mycorrhized roots (+) at 6 weeks post-inoculation. Transcript levels were normalized using the tomato EXP gene and are shown relative to untreated root samples. Central column samples correspond to 7-day-old dark-grown seedlings at 0, 6 and 24 h after exposure to light and to seedlings continuously grown in the light (L). Transcript levels were normalized to the EXP gene and are represented relative to etiolated (0 h) samples. Third column depicts different fruit ripening stages: MG, mature green; B, breaker; O, orange; and R, red ripe. Levels were normalized using ACT4 and are shown relative to MG samples. Expression values represent the mean±SD of three independent biological replicates (n=3). Asterisks indicate statistically significant differences relative to untreated (-), etiolated (0 h) or MG samples (t-test or one-way ANOVA with Dunnett’s multiple comparisons test, *p<0.05, **p<0.01). Figure 3. Co-immunoprecititation analyses. Nicotiana benthamiana leaves were co- agroinfiltrated with the indicated proteins tagged with C-terminal Myc (in blue) or HA (in red) epitopes. Controls agroinfiltrated only with the HA-tagged protein as indicated as (-). A fraction of the protein extracts (INPUT) was used to test protein production by immunoblot analyses using antibodies against Myc (𝛂Myc) and HA (𝛂HA). After immunoprecipitation (IP) of the remaining protein extracts using 𝛂Myc, samples were used for immunoblot analyses with 𝛂Myc (to confirm successful IP) and 𝛂HA (to detect the presence of co-immunoprecipitated HA-tagged proteins). Figure 4. CRISPR-Cas9 mutagenesis of tomato SlG2 and SlG3 genes. (A) Scheme representing the designed strategy to generate deletions on SlG2 and SlG3 genes and the resulting proteins in selected mutant alleles (See Figures S6-S8 for further details). Green, pink and black boxes represent transit peptides, protein-protein interaction motifs, and catalytic domains (FARM and SARM), respectively. Blue arrowheads indicate the position of the designed sgRNAs encompassing specific restriction sites, and black arrows represent primer pairs used for This article is protected by copyright. All rights reserved genotyping. (B) Activity assays of WT and mutant GGPPS enzymes in E. coli strains expressing Accepted Article bacterial genes for lycopene biosynthesis (crtB and crtI) but lacking GGPPS activity. Lycopene production after transformation with an empty vector (labelled as “Control” in the plots) or plasmid constructs harboring the indicated sequences is represented relative to the levels obtained with the bona-fide GGPPS enzyme AtG11. Values represent the mean±SD of at least three independent transformants (n=3). Figure 5. Leaf phenotypes of mutant tomato lines defective in SlG2 or SlG3. (A) Representative images of 4-week-old plants of the indicated lines. (B) Relative levels of total carotenoids, chlorophylls and tocopherols in young and mature leaves of WT and mutant lines. Values are represented relative to WT levels and they correspond to the mean±SD of at least three independent biological replicates (n=3). See Table S7 for absolute values. (C) ɸPSII in young and mature leaves of the indicated lines. Values represent the mean±SD of four different leaf areas from three different plants. In all cases, different letters represent statistically significant differences (p<0.05) among means according to posthoc Tukey’s tests run when one-way ANOVA detected different means. Figure 6. Metabolic changes in slg2 and slg3 mutants. Colors represent statistically significant fold-change (FC) values (t-test, p<0.05) of metabolite levels in young leaves or ripe fruit (B+10) from mutant tomato plants relative to those in WT controls. Quantitative and technical data are detailed in Tables S8 and S9 for leaves and Tables S10 and S11 for fruit. Figure 7. Flowering and fruit phenotypes of mutant tomato lines defective in SlG2 or SlG3. (A) Flowering time measured as days after germination (left) or number of leaves (right). Values correspond to the mean±SD of at least n=4 independent biological replicates. (B) Number of days to reach the indicated ripening stages represented as days post-anthesis on-vine (DPA, left) and days post-breaker off-vine (DPB, right). In both box-plots, the lower boundary of the boxes indicates the 25th percentile, the black line within the boxes marks the median, and the upper boundary of the boxes indicates the 75th percentile. Dots mark data values and whiskers above and below the boxes indicate the minimum and maximum values. (C) Representative images of fruit from WT and mutant lines harvested from the plant at the breaker stage. (D) Relative levels of individual carotenoids (phytoene, lycopene and 𝛃-carotene) and total tocopherols in B+10 fruits of WT and mutant lines. Values are represented relative to those in WT samples and correspond to the mean±SD of n=3 independent biological replicates. See Table S7 for absolute values. In all plots, different letters represent statistically significant differences (one-way ANOVA followed by Tukey’s multiple comparisons test, p<0.05). This article is protected by copyright. All rights reserved Figure 8. Leaf phenotypes of tomato lines with different combinations of slg2 snd slg3 Accepted Article mutations. (A) Representative images of 4-week-old plants of the indicated lines. Mutant alleles are marked in red. (B) Total levels of photosynthetic pigments (carotenoids and chlorophylls) in young and mature leaves of WT and mutant lines. Values, mean and SD of n=3 independent biological replicates are represented. (C) ɸPSII in young and mature leaves of the indicated lines. Values, mean and SD of four different leaf areas from three different plants are shown. In all plots, different letters represent statistically significant differences (p<0.05) among means according to posthoc Tukey’s tests that were run once the existence of different means was established by one-way ANOVA. Figure 9. Ripening-associated pigmentation and marker gene expression in tomato fruits with different combinations of slg2 snd slg3 mutations. (A) Average red color quantification (arbitrary units) of on-vine fruit from WT and mutant lines at the indicated times. Values represent the mean±SD of three different fruits (n=3) for each point. (B) RT-qPCR analysis of ACS2 and E8 transcript levels in WT and mutant fruits collected at the indicated developmental stages. Expression values were normalized using ACT4 and represent the mean±SD of n=3 independent biological replicates. In all plots, asterisks indicate statistically significant differences among means relative to WT samples (t-test, *p<0.05, **p<0.01). Asterisk color represents the genotype. This article is protected by copyright. All rights reserved Accepted Article This article is protected by copyright. All rights reserved Accepted Article nph_17283_f2.eps This article is protected by copyright. All rights reserved Accepted Article nph_17283_f3.eps This article is protected by copyright. All rights reserved Accepted Article nph_17283_f4.eps This article is protected by copyright. All rights reserved Accepted Article nph_17283_f5.eps This article is protected by copyright. All rights reserved Accepted Article This article is protected by copyright. All rights reserved Accepted Article nph_17283_f7.eps This article is protected by copyright. All rights reserved Accepted Article nph_17283_f9.eps This article is protected by copyright. All rights reserved