Beyond the limits of photoperception: constitutively active PHYTOCHROME B2 overexpression as a means of improving fruit nutritional quality in tomato

Summary Photoreceptor engineering has recently emerged as a means for improving agronomically beneficial traits in crop species. Despite the central role played by the red/far‐red photoreceptor phytochromes (PHYs) in controlling fruit physiology, the applicability of PHY engineering for increasing fleshy fruit nutritional content remains poorly exploited. In this study, we demonstrated that the fruit‐specific overexpression of a constitutively active GAF domain Tyr252‐to‐His PHYB2 mutant version (PHYB2Y252H) significantly enhances the accumulation of multiple health‐promoting antioxidants in tomato fruits, without negative collateral consequences on vegetative development. Compared with the native PHYB2 overexpression, PHYB2Y252H‐overexpressing lines exhibited more extensive increments in transcript abundance of genes associated with fruit plastid development, chlorophyll biosynthesis and metabolic pathways responsible for the accumulation of antioxidant compounds. Accordingly, PHYB2Y252H‐overexpressing fruits developed more chloroplasts containing voluminous grana at the green stage and overaccumulated carotenoids, tocopherols, flavonoids and ascorbate in ripe fruits compared with both wild‐type and PHYB2‐overexpressing lines. The impacts of PHYB2 or PHYB2Y252H overexpression on fruit primary metabolism were limited to a slight promotion in lipid biosynthesis and reduction in sugar accumulation. Altogether, these findings indicate that mutation‐based adjustments in PHY properties represent a valuable photobiotechnological tool for tomato biofortification, highlighting the potential of photoreceptor engineering for improving quality traits in fleshy fruits.


Introduction
Tomato (Solanum lycopersicum L.) is one of the most consumed vegetable crops in the world, both fresh or as processed juices and sauces (Bergougnoux, 2014). It is an important source of health-promoting substances in the human diet, including carotenoids (e.g. lycopene and b-carotene), flavonoids, ascorbic acid (i.e. vitamin C) and tocopherols (i.e. vitamin E) (Dorais et al., 2008;Frusciante et al., 2007). Accordingly, increasing attention has been devoted worldwide to improve tomato productivity and fruit quality traits both via traditional breeding or transgenicbased approaches Liu et al., 2004;Martin et al., 2011).
Most attempts to improve fruit nutritional composition have focused on altering the expression of specific genes directly involved in the production of carotenoids, flavonoids and other health-promoting substances (Gerszberg et al., 2015). In contrast, the manipulation of key players of light perception and signalling transduction pathway has emerged as an alternative to promote more extensively tomato fruit biofortification by simultaneously altering the accumulation of multiple nutraceutical compounds (Azari et al., 2010a;Davuluri et al., 2005;Ganesan et al., 2017;Gururani et al., 2015).
Responsible for the red/far-red light wavelength perception, phytochromes (PHYs) regulate a wide range of photomorphogenic responses throughout plant life cycle (Demotes-Mainard et al., 2016). Once photoactivated by red light exposure, PHYs allosterically change their conformation, triggering their translocation from cytosol to nucleus (Smith, 2000). In the nucleus, active PHYs interact with light signal transduction components, initiating highly complex and extensively interconnected signalling cascades that ultimately lead to the differential expression of photomorphogenesis-related genes and proteasome-dependent protein degradation events (Chen and Chory, 2011;Shin et al., 2016). Upon far-red light or dark exposure, active PHYs are converted back to the biologically inactive form (Seluzicki et al., 2017). In tomato, five PHY-encoding genes, namely PHYA, PHYB1, PHYB2, PHYE and PHYF, have been identified (Alba et al., 2000a). PHYB2 is the most expressed in tomato fruits (Bianchetti et al., 2017;Hauser et al., 1997;Pratt et al., 1995) and a major regulator of fruit chloroplast maturation and carotenoid accumulation Gupta et al., 2014).
Besides the mRNA levels, PHY protein activity can also be manipulated. In Arabidopsis thaliana and Avena sativa, specific amino acid changes in the photosensor module domains result in a light-independent constitutive biological activity (Jeong et al., 2016;Su and Lagarias, 2007). This is the case of the mutation Tyr 276 -to-His in the conserved GAF domain of Arabidopsis PHYB that renders constitutive PHY-dependent photomorphogenic responses, even in the darkness (Hu et al., 2009;Oka et al., 2011;Su and Lagarias, 2007).
Here, we investigated whether the overexpression of the constitutively active GAF domain Tyr-to-His PHYB2 mutant version (PHYB2 Y252H ) is suitable to improve PHY-dependent developmental and metabolic responses associated with tomato fruit nutritional composition, such as plastid development (Bianchetti et al., 2017, carotenogenesis (Alba et al., 2000b;Bianchetti et al., 2018;Gupta et al., 2014) and tocopherol biosynthesis (Gramegna et al., 2018). To circumvent adverse collateral effects on vegetative growth (e.g. reduced plant height, apical dominance and leaf size) normally caused by whole-plant PHY overexpression (Garg et al., 2006;Husaineid et al., 2007;Thiele et al., 1999), a fruit-specific overexpression strategy was employed. Our data indicated that the expression of the constitutively active PHYB2 Y252H promotes the biosynthesis and accumulation of carotenoids, flavonoids and vitamins C and E in tomato fruits to significantly higher levels than those detected when the native PHYB2 was overexpressed. Fruit plastid biogenesis and differentiation were particularly promoted by PHY-B2 Y252H overexpression, accompanied by an overall up-regulation of genes encoding chloroplast components and photosynthesisrelated proteins. Fruit primary metabolism was only marginally affected by the fruit-specific PHYB2 or PHYB2 Y252H overexpression, which further highlights the potential application of this genetic manipulation to enhance tomato fruit nutritional quality.

Results
Fruit-specific PHYB2 and PHYB2 Y252H overexpression in tomato plants To identify the conserved tyrosine residue to generate hyperactive GAF domain Tyr-to-His PHYB2 mutants, the coding sequence of Solanum lycopersicum PHYB2 was aligned and compared with Arabidopsis thaliana PHYB. The conserved Y 276 residue of the Arabidopsis PHYB GAF domain corresponds to Y 252 in tomato PHYB2 ( Figure S1). Next, transgenic lines of tomato (Micro-Tom cultivar) expressing the native PHYB2 gene or its mutant hyperactive form PHYB2 Y252H under the control of fruit-specific PHOSPHOENOLPYRUVATE CARBOXYLASE 2 (PPC2) promoter (Fernandez et al., 2009) were generated. Overexpression of the transgenes was confirmed by RT-qPCR in two PPC2::PHYB2 (PPC:: B2) lines (namely L1 and L3) and three PPC2::PHYB2 Y252H (PPC:: B2 Y252H ) lines (namely L5, L6 and L15) (Figure 1a). PHYB2 or PHYB2 Y252H transcript levels were between twofold and sevenfold higher in the transgenic lines compared with PHYB2 mRNA levels detected in WT fruits throughout initial fruit development (immature green, IMG; mature green, MG) and ripening stages (breaker, BK; red ripe, RR) ( Figure 1a). Transcript abundance of other PHY genes remained virtually indistinguishable between the WT and transgenic lines throughout fruit development and ripening (Table S1).
The fruit-specific overexpression of either PHYB2 or PHY-B2 Y252H resulted in no apparent phenotypical alterations on plant vegetative growth ( Figure S2). In contrast, green fruits (IMG and MG) from the PPC::B2 Y252H lines exhibited significantly darker green coloration at the fruit pedicellar (shoulder) region than WT counterparts (Figure 1b). At the BK stage, the green shoulder phenotype persisted in both PPC::B2 and PPC::B2 Y252H fruits, whereas it was significantly attenuated in WT fruits. The green shoulder phenotype was subsequently lost in all genotypes at the RR stage (Figure 1b).
RNA-seq transcriptomic profiling performed in BK fruits of representative PPC::B2 (L1) and PPC::B2 Y252H (L6) lines revealed that up to 5.4% of the 23 685 transcriptionally active genes at this developmental stage was affected either by the PHYB2 or PHYB2 Y252H overexpression. Compared with the WT, 906 and 719 genes were differentially expressed exclusively in PPC::B2 and PPC::B2 Y252H fruits, respectively. Only 379 differentially expressed genes (DEGs) were common in both transgenics when compared to the WT, with a predominance of up-regulated (331) over down-regulated (48) genes ( Figure 1c, Table S2). RNA-seq results were further validated by RT-qPCR analysis, demonstrating consistency (r 2 correlation above 0.9) between both methods ( Figure S3).
According to the Gene Set Enrichment Analysis (GSEA), the exclusive PPC::B2 DEGs were predominantly associated with global cell functioning and diverse metabolic processes, grouping around GO terms such as sequence-specific DNA binding, cell wall organization or biogenesis, oxoacid metabolic process and small molecule biosynthetic process ( Figure 1c, Table S3). In contrast, the majority of DEGs exclusively detected in PPC:: B2 Y252H fruits were associated with chloroplast components and related processes, including GO terms as photosynthesis, thylakoid, photosynthetic membrane and generation of precursor metabolites and energy ( Figure 1c, Table S3).
Either PHYB2 or PHYB2 Y252H overexpression promotes fruit plastid biogenesis but only PHYB2 Y252H enhances chlorophyll accumulation The greener fruit phenotype and up-regulation of plastid-related genes in the transgenic lines prompted us to investigate the impacts of either PHYB2 or PHYB2 Y252H overexpression on fruit chloroplast biogenesis and development. Microscopy analysis revealed increments of up to 50% in plastid abundance in pericarp cells of PPC::B2 and PPC::B2 Y252H green fruits compared with WT counterparts (Figure 2a,b). The increased number of plastids detected in PPC::B2 and PPC::B2 Y252H fruits was associated with up-regulation of PLASTID DIVISION 2 (PDV2) (Figure 2c), which encodes a vital component of the plastid division machinery (Okazaki et al., 2009).
Total chlorophyll content was about threefold higher in PPC:: B2 Y252H green fruits than in the WT (Figure 2d, Table S5), which was associated with increments of approximately twofold in transcript levels of PROTOCHLOROPHYLLIDE OXIDOREDUCTASE 1 (POR1) (Figure 2c), which encodes an essential light-triggered chlorophyll biosynthesis-related enzyme (Heyes and Hunter, 2005). Moreover, transcriptome analysis carried out in MG fruits revealed enrichment of transcripts associated with photosynthesis, thylakoid and plastid membranes in PPC::B2 Y252H compared with WT fruits (Figure 2e). Most chlorophyll biosynthetic genes were significantly up-regulated in PPC::B2 Y252H MG fruits compared with the WT (Figure 2f, Table S6), which agrees with the higher chlorophyll content detected in the transgenic fruits. In contrast, PPC::B2 green fruits showed no significant changes in total chlorophyll levels nor in POR1 mRNA levels compared with the WT (Figure 2c,d, Tables S4 and S5).

PHYB2 Y252H overexpression promotes fruit isoprenoid metabolism
Besides photosynthesis, chloroplasts are essential organelles for the synthesis and storage of secondary metabolites (Armbruster et al., 2011). Therefore, increments in fruit plastid abundance and size have consistently been associated with increased tomato fruit nutritional composition (Bianchetti et al., 2017;Bino et al., 2005;Cocaliadis et al., 2014;Cruz et al., 2018;Enfissi et al., 2010;Galpaz et al., 2008;Kolotilin et al., 2007;Lupi et al., 2019;Nguyen et al., 2014). Besides, light signalling can also directly influence the expression of genes involved in the production of  Tables S2 and S3, respectively. IMG, immature green; MG, mature green; BK, breaker; RR, red ripe.  Table S3. (f) Simplified chlorophyll biosynthetic pathway. Intermediate reactions are omitted. Up-regulated chlorophyll biosynthesis-related genes in PPC::B2 Y252H MG fruits according to RNA-seq analysis are highlighted in red. Gene abbreviations, as well as logFC and FDR values, are detailed in Table S6. In (b) and (d), data are mean AE SE, dots represent individual values, and statistical differences within each stage are given by different letters (Tukey's Test, a = 0.05). BK, breaker; ALA, 5-aminolevulinic acid.
Carotenoid and tocopherol profiling revealed that PHYB2 overexpression promoted exclusively b-carotene and lutein accumulation in RR fruits (Figures 3a and S6, Table S5). In contrast, PPC::B2 Y252H RR fruits exhibited significant increments in virtually all individual carotenoids analysed, with increments of about 50% in phytoene, phytofluene and lycopene levels and between 100% and 200% in b-carotene and lutein levels, respectively. Total tocopherol content in red fruits exhibited a similar trend, with increments of up to twofold in PPC::B2 Y252H lines (Figures 3a and S6, Table S5). The levels of a-tocopherol, the dominant tocopherol form in tomato fruits, as well as b-carotene and lutein contents were significantly higher in PHYB2 Y252H -overexpressing lines than in the WT at all sampled stages (Table S5). This indicates that the PHYB2 Y252H -triggered intensification of fruit isoprenoid accumulation was not restricted to the ripening stages.
The activities of the carotenoid desaturase enzymes PHYTOENE DESATURASE (PDS) and f-CAROTENE DESATURASE (ZDS) are especially sensitive to the cellular redox state, requiring plastoquinones (PQs) as co-factors to accept electrons produced during the desaturation steps of GGDP to lycopene (Fanciullino et al., 2014;Norris et al., 1995). The transcript levels of the PQ biosynthetic enzyme SOLANESYL DIPHOSPHATE SYNTHASE (SPS) were up-regulated in PPC::B2 Y252H lines, likely boosting PQ biosynthesis in the transgenic fruits. Moreover, ORANGE RIPEN-ING (ORR), a gene encoding a NADH dehydrogenase (NdH) complex subunit responsible for PQ reduction (Endo et al., 2008), was also up-regulated in PPC::B2 Y252H lines ( Figure 3b).
GERANYLGERANYL DIPHOSPHATE REDUCTASE (GGDR), which encodes the enzyme responsible for the de novo synthesis of tocopherol precursor phytyl diphosphate (PDP), was markedly upregulated exclusively in PPC::B2 Y252H fruits at all sampling stages, suggesting increased conversion of GGDP to PDP in these transgenic lines. During green stages of fruit development, GGDR up-regulation presumably promoted PDP production, feeding the enhanced chlorophyll biosynthetic pathway mentioned above ( Figure 3b). From the onset of ripening onwards, chlorophyll biosynthesis ceases and the precursor GGDP are channelled towards carotenoid production by the transcriptional inhibition of GGDR in WT (Table S7, Quadrana et al., 2013), whereas chlorophyll degradation releases phytol, which is, in part, incorporated into tocopherol (Zhang et al., 2014), leading to its increment during ripening (Table S5). Interestingly, at the BK stage in PPC::B2 Y252H lines, PHYTOL KINASE (VTE5), which is responsible for the first step of phytol phosphorylation (Valentin et al., 2006), was up-regulated compared with WT counterparts (Figure 3b), probably enhancing the production of PDP and contributing to the increase in total tocopherol content observed all along fruit development ( Figure 3b). Accordingly, tocopherol biosynthetic genes, particularly 2,3-DIMETHYL-5-PHYTYLQUINOL METHYLTRANSFERASE b (VTE3b), TOCOPHEROL CYCLASE (VTE1) and TOCOPHEROL C-METHYL TRANSFERASE (VTE4), were also predominantly up-regulated in PPC::B2 Y252H lines from the IMG to the BK stage ( Figure 3b).
Both flavonoid and ascorbate metabolisms are upregulated in PHYB2 Y252H ripe fruits In our study, the fruit-specific PHYB2 Y252H overexpression significantly promoted both flavonoid and ascorbate contents in RR fruits, whereas the levels of these antioxidants were slightly, but not significantly, increased in PPC::B2 fruits. Increments between twofold and threefold in rutin, kaempferol rutinoside, naringenin chalcone and naringenin glucoside contents were registered in PPC::B2 Y252H RR fruits compared with WT counterparts (Figures 4a and S7, Table S8). Both ascorbate and dehydroascorbate levels were overaccumulated in PPC::B2 Y252H fruits ( Figure S7), resulting in increments of approximately 20% in the total pool of this antioxidant (Figure 4b).
On the other hand, the PPC::B2 Y252H -induced increments in rutin, which is a major flavonoid accumulated in tomato fruits (Slimestad and Verheul, 2009), as well as kaempferol rutinoside, were associated with significantly higher transcript abundance of FLAVONOL SYNTHASE (FLS) in the transgenic fruits at the BK stage ( Figure 4c). The higher ascorbate content was associated with increased transcript levels of GDP-D-MANNOSE-3,5-EPIMER-ASE 1 (GME1) (Figure 4d), which encodes an enzyme responsible for an initial step in the ascorbate biosynthetic pathway (Wolucka and Van Montagu, 2007). RNA-seq data revealed that GME1 was the only core ascorbate biosynthetic gene modulated by PHYB2 Y252H overexpression ( Figure S8), with up to twofold increments in mRNA levels at the BK and RR stages as confirmed by qPCR analysis (Figure 4d). Consistent with the differential effect of PHYB2 and PHY-B2 Y252H overexpression on isoprenoid flavonoids and ascorbate metabolism in RR fruits, only PPC::B2 Y252H lines consistently exhibited higher total antioxidant activity, as estimated by the DPPH assay, reaching increments of up to 25% compared with WT counterparts (Figure 4e, Table S8).
Lipid and sugar metabolisms are the main primary metabolic pathways affected in the transgenic fruits Mass spectrometry (MS)-based metabolite profiling and highperformance liquid chromatography (HPLC)-based soluble carbohydrate (sucrose, glucose, fructose) and organic acid (citric acid) analysis were used to investigate the impacts of PHYB2 or PHYB2 Y252H overexpression on primary metabolism of ripe fruits. About 25% of the metabolites identified via GC-MS profiling was consistently altered in the transgenic lines compared with WT counterparts (Table S10). Hierarchical clustering analysis supported the differences between the clades composed by the PPC::B2 and PPC::B2 Y252H lines based on polar and apolar metabolite profiles, evidencing that fruit metabolome was differentially affected by PHYB2 or PHYB2 Y252H overexpression (Figure 5a-b).
Minor organic acids such as acetic, benzoic, lactic and succinic acids were under-accumulated in all transgenic lines, whereas oxalic acid was overaccumulated in PPC::B2 lines. Citric acid, which is the most abundant organic acid in tomato (Morgan et al., 2013), was not significantly altered in transgenic fruits throughout fruit development and ripening (Table S11). Except for aspartic acid, aminoacids were not consistently altered in the transgenic fruits compared with WT counterparts (Figure 5a).
Glucose, fructose and sucrose contents were indistinguishable in WT and transgenic green fruits (Table S11). However, glucose and fructose levels were predominantly lower in the RR transgenic fruits than in the WT counterparts, whereas sucrose remained at similar levels in all genotypes (Figure 5c and S9, Table S11). Accordingly, total soluble solids content (°Brix) values in transgenic ripe fruits were approximately 5% lower than in the WT counterparts ( Figure S9). The amount of soluble sugars in ripe tomato fruits is determined by several factors, including the plant photosynthetic capacity, dynamics of starch biosynthesis and breakdown, and sucrose import from vegetative tissues and conversion into hexoses (Patrick et al., 2013). To investigate the possible causes of the reduced levels of soluble sugars in the transgenic lines, we first estimated the plant photosynthetic capacity by measuring gas exchange, fluorescence parameters, chlorophyll content and starch levels in source leaves of all genotypes. Overall, net photosynthesis rate, stomatal conductance, photosystem II efficiency and chlorophyll and starch content in leaves of the transgenic lines were indistinguishable from the WT counterparts (Table S12), which is in accordance with the fruit-specific genetic manipulation.
As the transient starch accumulation during early tomato fruit development also directly influences total soluble sugar content at ripe stage (Davies and Cocking, 1965;Yin et al., 2010), we next characterized the dynamics of starch accumulation in green WT and transgenic fruits. Both PHYB2 and PHYB2 Y252H overexpression negatively impacted starch accumulation, reducing the levels of this carbohydrate at IMG stage in 27% and 44%, respectively ( Figure 5d, Table S11). The lower starch content in the transgenic lines at IMG stage was associated with a significant reduction in mRNA levels of the gene responsible for encoding the subunit L1 of ADP-GLUCOSE PYROPHOSPHORYLASE (AGPa-seL1) (Figure 5e), which is the AGPase large subunit-encoding gene most expressed in tomato sink tissues Petreikov et al., 2010).
Although reduced starch synthesis in immature tomato fruits can also be caused by reduced sink capacity by the fruit tissues (Osorio et al., 2014), the transcript abundance of key genes associated with tomato fruit sugar import and breakdown, such as LYCOPERSICUM INVERTASE 5 and 6 (LIN5 and LIN6) and ACID INVERTASE 1 (TIV1), was not consistently altered in the transgenic lines compared with the WT during early fruit development (Table S13).
Taken together, our data suggest that sugar and lipid metabolisms are the major fruit primary metabolic pathways affected by fruit-specific PHYB2 or PHYB2 Y252H overexpression. Rather than associated with changes in plant photosynthetic or fruit sink capacities, the reduced levels of soluble sugars in red ripe transgenic fruits seem to be associated with lower starch synthesis and accumulation during early fruit development.

PHYB2 Y252H overexpression leads to phytonutrient overaccumulation in fruits of distinct tomato genetic backgrounds
To evaluate whether fruit-specific PHYB2 Y252H overexpression can promote fruit quality regardless of the tomato genetic background, this phenomenon was also investigated in the wellcharacterized tomato cultivar Ailsa Craig (AC). Wild-type AC plants were crossed with Micro-Tom (MT) wild-type or homozygous PHYB2-or PHYB2 Y252H -overexpressing plants generating F 1 plants (AC-WT, AC-PPC::B2 and AC-PPC::B2 Y252H , respectively) that were cultivated under greenhouse, semi-controlled conditions until fruit harvesting.
As observed for the MT background, MG fruits of AC plants overexpressing PHYB2 Y252H exhibited darker green fruit coloration than AC-WT, a phenotype explained by the higher chlorophyll accumulation and increased POR1 transcript levels of the transgenic fruits (Figure 6a-c and S10). At the ripe stage, fruits from both AC-PPC::B2 and AC-PPC::B2 Y252H genotypes displayed higher lycopene accumulation and more intense red coloration than the AC-WT counterparts (Figure 6a,d).
Compared with AC-WT, ripe AC-PPC::B2 Y252H fruits overaccumulated b-carotene, lutein and all four (a, b, d and c) tocopherol forms (Figure 6d and S10). PHYB2 Y252H overexpression in AC cultivar also resulted in the up-regulation of key tocopherol biosynthetic genes as GGDR and VTE3b (Figure 6e). The levels of flavonoids, such as rutin and quercetin derivatives, and ascorbate were also increased in response to PHYB2 Y252H overexpression, and reductions of around 10% in°Brix were reported in red fruits of both AC-PPC::B2 and AC-PPC::B2 Y252H genotypes compared with the AC-WT counterparts ( Figure S10). Comparing data obtained from transgenic lines of MT and AC cultivars reveals that PHYB2 Y252H overexpression can promote fruit nutritional quality in different tomato genetic backgrounds as well as distinct growth conditions (i.e. laboratory and greenhouse conditions).

Discussion
Manipulation of light signalling pathway components has been increasingly investigated as a means to improve desired quality traits in many crops, including adjustments in plant architecture,  Table S10. Dendrograms indicate hierarchical clustering relationships between lines. (c) Soluble sugars (sucrose, glucose and fructose) content in RR fruits. (d) Starch content in immature green (IMG) fruits. (e) AGPaseL1 mRNA levels in IMG fruits. In (d,e), data are mean AE SE and dots represent individual values. Metabolites, gene abbreviations and relative transcript values are detailed in Tables S11 and S13. Statistical differences are given by different letters (Tukey's test, a = 0.05).
shade avoidance responses, abiotic stresses tolerance and flowering time (Ganesan et al., 2017;Gururani et al., 2015). In tomato, accumulating genetic evidence indicates that alterations in downstream components of photoreceptor signal transduction pathways can impact fruit chemical composition by simultaneously affecting multiple metabolic pathways (Azari et al., 2010b;Davuluri et al., 2005;Liu et al., 2004;Llorente et al., 2016;Wang et al., 2008). In contrast, attempts to promote tomato biofortification via increments in photoreceptor protein abundance (Giliberto et al., 2005) or by genetic engineering photoreceptor properties (e.g. light sensitivity, protein stability) are currently scarce or missing, respectively. Here, we demonstrated that the fruit-specific overexpression of the constitutively active Tyr 252 -to-His PHYB2 mutant version leads to the overaccumulation of multiple health-promoting antioxidants in tomato fruits, regardless of the genetic background (i.e. Micro-Tom or Ailsa Craig cultivars), without causing adverse collateral effects on vegetative development either when plants were grown under laboratory (i.e. Micro-Tom) or greenhouse (i.e. Ailsa Craig) conditions. In contrast, overexpression of native PHYB2 under the control of the same fruit-specific promoter (pPPC2) resulted in less prominent, and sometimes non-significative, increases in bioactive compounds (e.g. carotenoids, flavonoids, vitamin C and vitamin E) in ripe tomato fruits. These findings confirm that engineered PHY forms, particularly light-independent constitutively active versions (Jeong et al., 2016;Oka et al., 2011;Su and Lagarias, 2007), represent a powerful photobiotechnological tool to increment the nutritional quality of fruits.
Making room for more: PHYB2-positive impacts on fruit plastid abundance and ultrastructure The most striking visual phenotype of PPC::B2 and particularly PPC::B2 Y252H transgenic lines was the production of greener immature fruits and the persistence of the green shoulder during initial ripening, which were attributed to a higher abundance of chloroplasts in the pedicular region compared with the WT (Figure 1). In agreement, PHY-mediated light perception has long been associated with chloroplast development and thylakoid formation from early seedling de-etiolation throughout adult plant life (Girnth et al., 1978;Mohr, 1977). In tomato fruits, PHY deficiency leads to reduced plastid size and density per cell (Bianchetti et al., 2017), whereas the opposite phenotype has been observed in the light-hyperresponsive tomato high-pigment (hp) mutants hp1 and hp2 (Davuluri et al., 2005;Kendrick et al., 1997;Wang et al., 2008). Fruit-localized PHY deficiency has also been demonstrated to regulate mRNA levels of multiple tomato genes encoding key components of the plastid division machinery . Among these genes, PDV2 was particularly up-regulated in the PHYB2-or PHYB2 Y252H -overexpressing lines (Figure 2), which corroborates the proposed central role of this gene in determining plastid division rates in immature tomato fruits  and Arabidopsis leaves (Okazaki et al., 2009).
In line with the identification of CURT1a protein accumulation as requisite for grana stacking in Arabidopsis leaf chloroplasts (Armbruster et al., 2013;Pribil et al., 2018), the up-regulation of CURT1a1 and CURT1a2 in PPC::B2 Y252H IMG fruits may be linked to the conspicuously incremented grana size registered in these transgenic lines (Figure 2). Also, genes associated with photosynthesis and chloroplast structure and functioning predominate among those up-regulated in response to PHYB2 or PHYB2 Y252H overexpression (Figure 1), which agrees with the well-reported positive influence of PHY signalling pathway on these gene categories in de-etiolating seedlings and developing leaves (Dubreuil et al., 2017;Hu et al., 2009;Oh and Montgomery, 2014;Yoo et al., 2019).
Increments in plastid ultrastructure positively affected isoprenoid accumulation once many active biosynthetic enzymes and their hydrophobic products are physically associated with plastid membranes (Llorente et al., 2017;Yuan et al., 2015). In tomato, such positive correlation is evident in several mutants and transgenic lines that produce plastid-rich, dark green immature fruits, such as hp1 and hp2 mutants (Kolotilin et al., 2007) and GLK2-overexpressing lines (Lupi et al., 2019;Powell et al., 2012). Therefore, the increments in plastid abundance and ultrastructure observed in PPC::B2 and PPC::B2 Y252H fruits can be interpreted as a physical facilitator of isoprenoid biosynthesis and accumulation in the pericarp cells. However, many isoprenoid biosynthetic genes are also transcriptionally regulated by components of PHY signalling pathway (Gramegna et al., 2018;Inagaki et al., 2015;Llorente et al., 2016), which possibly explains the differential impacts of PHYB2 and PHYB2 Y252H overexpression on the abundance of isoprenoids in green (chlorophylls) and ripening fruits (carotenoids and tocopherols). In fact, only PPC::B2 Y252H fruits presented higher chlorophyll content than the WT, a differential response associated with the exclusive up-regulation of chlorophyll biosynthetic genes in the PPC::B2 Y252H lines (Figures 2, 3 and 6).

PHYB2 Y252H -induced overaccumulation of bioactive compounds mainly relies on the up-regulation of target biosynthetic genes
Tomato fruits are a major source of essential antioxidants in the human diet, such as carotenoids, tocopherols, flavonoids and ascorbate, which are associated with reducing the risks of cancer and cardiovascular diseases (Dorais et al., 2008;Frusciante et al., 2007). Many of these antioxidant classes accumulate or display altered composition as tomato fruit ripens via complex metabolic pathways tightly coordinated at the transcriptional level (Bovy et al., 2007;Li et al., 2019;Quadrana et al., 2013;Verhoeyen et al., 2002;Yuan et al., 2015).
Carotenoids and tocopherols have as a precursor the MEP pathway-derived GGDP (Pulido et al., 2012). Transcriptional analysis revealed that DXS1 and GGPS2 were particularly upregulated in ripening PPC::B2 and PPC::B2 Y252H fruits, possibly boosting the production of GGDP to sustain higher carotenoid and tocopherol biosynthesis fluxes (Figure 3). Although PHY-B2 Y252H overexpression resulted in significantly higher levels of lycopene, b-carotene and lutein, the most significant bioactive carotenoids in ripe tomatoes (Raiola et al., 2014), core carotenoid biosynthetic genes (e.g. PSY1, PDS) were not consistently modulated at the transcriptional level in these transgenic lines (Figure 3). This seemingly contradictory result suggests that the PHYB2 Y252H -triggered up-regulation of GGDP-biosynthetic genes and/or the increased plastid abundance with more developed grana may suffice to sustain the higher carotenoid synthesis and accumulation detected in the ripe transgenic fruits. Alternatively, the presence of the constitutively active PHYB2 Y252H molecules may also have affected fruit carotenogenesis via additional regulatory levels.
Carotenoid-related enzymes can also be post-transcriptionally regulated, primarily relying on PQs as electron acceptors during the desaturation steps of phytoene and subsequent metabolites, driving lycopene biosynthesis (Norris et al., 1995). SPS, a PQ biosynthetic gene, was up-regulated in PPC::B2 Y252H fruits, suggesting that PQ pool was increased in the transgenic fruits. Accordingly, silencing of tomato SPS reduces the levels of PQs, modifying carotenoid composition and leading to phytoene accumulation, possibly due to the impairment of desaturases activity (Jones et al., 2013). The positive relation between the levels of PQs and carotenoid biosynthesis has also been reported in Arabidopsis (Kim et al., 2015) and tomato hp2 leaves (Jones et al., 2013). The reduction of PQs is catalysed by the Ndh complex, essential for the control of carotenoid accumulation (Endo et al., 2008). In tomato fruits, mutations in the ORR gene, encoding a Ndh subunit, led to decreased carotenoid content (Nashilevitz et al., 2010), highlighting the importance of proper functioning of this redox chain to sustain tomato carotenogenesis. In line with this and in agreement with the enhanced carotenoids accumulation, ORR was up-regulated in PPC::B2 Y252H fruits.
The marked increase in total tocopherol content in the PPC:: B2 Y252H lines can be attributed to the up-regulation of genes related to bottlenecks of PDP biosynthesis. VTE5, which catalyses the first and limiting step of channelling phytol chain into tocopherol biosynthetic pathway (Almeida et al., 2016), was transcriptionally up-regulated in PPC::B2 Y252H BK fruits, providing PDP from chlorophyll turnover. Also, GGDR levels were upregulated in PPC::B2 Y252H BK and RR fruits, possibly increasing PDP supply via MEP pathway. In line with our findings, the upregulation of GGDR in MG stages of hp1 tomato mutants was also associated with increased tocopherol content in this mutant (Enfissi et al., 2010). It has also been shown that GGDR transcription is negatively regulated by PIF3 in tomato fruits (Gramegna et al., 2018). Thus, the intensification of PHYB2 Y252H -PIF3 interaction due to the constant presence of PHYB2 Y252H in pericarp cell nuclei, and consequent reduction in PIF3 protein accumulation, explains the markedly higher GGDR expression in PPC::B2 Y252H from early stages of fruit development until the completion of ripening. Besides GGDR, PHYB2 Y252H overexpression promoted the expression of VTE1, VTE4 and VTE3b, the latter two already described to be positively correlated with fruit vitamin E content in tomato genotypes (Fritsche et al., 2017;Quadrana et al., 2013).
Light signals have been widely recognized as major environmental factors controlling flavonoid and vitamin C accumulation in tomato (Løvdal et al., 2010;Massot et al., 2012). In agreement, our findings indicate that the higher content of naringenin chalcone, rutin and kaempferol registered in the PPC:: B2 Y252H lines can be partially explained by the marked upregulation of FLS mRNA levels detected in the transgenic fruits at BK stage, when this gene is at peak expression (Figure 4c). Similarly, among all core ascorbate biosynthetic genes, only GME1 was positively correlated with the vitamin C overaccumulation of PPC::B2 Y252H fruits, corroborating the proposition of GME1 expression as a major determinant of ascorbate levels in tomato fruits (Gilbert et al., 2009;Li et al., 2019;Massot et al., 2012).

Fruit primary metabolism is marginally affected by PHYB2 manipulation
Besides playing vital roles in primary metabolism, soluble sugars, organic acids and some aminoacids are also essential determinants of tomato fruit flavour (Carli et al., 2009). Different from the overall PHYB2 Y252H -mediated changes in secondary metabolism, more limited differences in the abundance of primary metabolites were observed between the WT and transgenic ripe fruits, except for the over-and under-accumulation of some lipids and soluble sugars, respectively ( Figure 5). Such reduction in soluble sugars agrees with the slightly lower o Brix values detected in ripe transgenic fruits, and the use of alternative promoters or further engineering in PHY sequence may be necessary to circumvent this negative collateral effect.
Accumulation of sugars in ripe tomato is directly influenced by the pool of starch synthesized at the early stages of fruit development, which is under the influence of multiple endogenous and environmental factors (Sagar et al., 2013;Schaffer et al., 2000;Yin et al., 2010). After confirming that the fruitspecific transgene overexpression had no impact on leaf photosynthetic capacity nor in the expression of predominant invertase genes, we identified a selective under-accumulation of transcripts encoding a specific subunit of AGPase, an enzyme responsible for catalysing the first and rate-limiting step of starch accumulation (Petreikov et al., 2006;Stark et al., 1992). In tomato, AGPaseL1 is the predominant large AGPase subunit in immature tomato fruits Petreikov et al., 2006), whose gene transcript accumulation is affected by fruit-localized PHYs . Interestingly, PHYB2 and PHYB2 Y252H overexpression significantly reduced AGPaseL1 mRNA levels specifically during early fruit development, when starch biosynthetic rates are maximum, suggesting a possible restriction in AGPase-dependent production of the starch precursor ADPglucose.
As carbohydrate and fatty acids biosynthesis compete for the same carbon precursors (Rawsthorne, 2002;Yu et al., 2018), it seems plausible to suggest a possible link between the lower sugar and higher lipid abundance in PPC::B2 Y252H ripe fruits compared with the WT ( Figure 5). Additionally, it has been demonstrated that tocopherol deficiency affects fatty acid metabolism, with consequences in sugar metabolism in tomato. By two distinct genetic interventions, tocopherol-deficient plants showed higher amounts of starch and enhanced unsaturated fatty acid accumulation than WT. The changes in lipid profiles correlated with the reduction in FADs expression (Almeida et al., 2016;Berm udez et al., 2018). Interestingly, an inverse scenario was described by our results. PPC::B2 Y252H fruits showed higher tocopherol content, reduced amount of starch and up-regulation of FADs. How tocopherol influences sugar metabolism has not yet been precisely addressed but it might be the result of the interaction between lipid and sugar metabolism machineries as observed here.

Conclusion remarks
Our findings indicate that fruit-specific overexpression of native PHYB2 sequence is not sufficient to drive significant increments in health-promoting secondary metabolites, a limitation resolved with the use of a light-independent constitutively active version of this photoreceptor. The presence of continuously active PHY-B2 Y252H molecules all over the fruit cells, regardless of the surrounding light conditions, probably leads to saturated PHYB2dependent light signalling. As a consequence, PHYB2 signallingdependent gene transcription is intensified, including many plastid-related genes as well as biosynthetic enzymes responsible for the production of antioxidants. Experiments conducted using two genetic backgrounds (i.e. Micro-Tom and Ailsa Craig) and two cultivation conditions (i.e. laboratory and greenhouse) allowed us to infer that PHYB2 Y252H overexpression is a valid tool for promoting fruit nutritional content across tomato cultivars and at distinct growth conditions. Moreover, as phytochrome GAF domain Tyr residues are well conserved across plant species, the gain-of-function engineering of PHYs employed in this work may also be applicable as a means to promote biofortification in other fleshy fruits.

Plasmid constructs and tomato transformation
Solanum lycopersicum PHYB2 full-length coding sequence was amplified with Platinum SuperFi Green PCR Master Mix (Thermo Fischer Scientific, Waltham, USA) using tomato fruit cDNA as a template and the oligonucleotide sequences detailed in Table S14. The fragment was inserted into entry vectors pDON221 (Thermo Scientific) and recombined with pK7m24GW,3 and pEN-L4-PPC-R1 plasmids (Fernandez et al., 2009) via LR clonase II Plus (Thermo Scientific) reaction to generate the final construct PPC2::PHYB2. The single T-to-C base change required to modify the amino acid translation from Tyr 252 to His 252 was inserted using the QuikChange Lightning Site-Directed Mutagenesis Kit (Agilent), resulting in the PPC2:: PHYB2 Y252H construct. The Agrobacterium-mediated transformation of tomato (Solanum lycopersicum L.) plants cv. Micro-Tom, which harbours the wild-type GLK2 allele (Carvalho et al., 2011), was conducted as described by Bianchetti et al. (2018). Oligonucleotides used for plasmid construction and selection of transgenic plants are listed in Table S14. The homozygous Micro-Tom (MT) lines were obtained from the T 2 or T 3 progeny by seeking 100% kanamycin resistance in the T 3 or T 4 seed population, respectively. Experiments were performed in homozygous T 4 (or later) generations.
Homozygous MT PHYB2 or PHYB2 Y252H -overexpressing lines were used to generate F 1 plants in Ailsa Craig (AC) background. Flowers from wild-type AC plants were emasculated at anthesis and hand-pollinated with MT-WT, MT-PPC2::PHYB2 or MT-PPC2:: PHYB2 Y252H pollen. F 1 seedlings were selected via kanamycin resistance assay, selected using oligonucleotides listed in Table S14.

Growth conditions
Micro-Tom wild-type and transgenic lines were grown in a chamber under controlled conditions: 250 µmol m À2 s À1 , 12-h photoperiod, air temperature of 27°C day/22°C night and 60% day/80% night relative air humidity. Fruits were collected at immature green (IMG, on average 13 days post-anthesis), mature green (MG, when green fruits are fully grown presenting locular gel), breaker (BK, when first signs of yellowing appear on the fruit bottom) and red ripe (RR, 12 days after BK) stages.
Ailsa Craig wild-type and transgenic lines were grown under greenhouse, semi-controlled conditions: average mean temperature of 25 AE 5°C, 11.5/13 light hours (winter/summer) and 250-350 mmol m -2 s -1 of incident photoirradiance. Fruits were harvested at MG and red ripe (RR, 12 days after BK) stages.
In all cases, fruits were harvested at the same time of the day (between the 4th and 6th hour of light period) in at least four biological replicates, with each replicate composed by at least four fruits from different plants. Seeds were removed, and the remaining tissues were immediately frozen in liquid nitrogen, powdered and stored at À80°C until use.

RNA sequencing and data analysis
RNA extraction was performed as described in Bianchetti et al. (2018), and RNA quality was assessed using the Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, USA). Library preparation and sequencing using the Illumina HiSeq2500 system, as well as RNA-seq assembly, annotation and differential expression analysis are detailed in Methods S1. Gene Set Enrichment Analysis (GSEA) was performed on Blast2GO software (version 5.2.5) (http://www.blast2go.org/) with the following modifications on default parameters: Gene Set Minimal Size and FDR filter value were set to 10 and 0.25, respectively.

RT-qPCR analysis
RNA extraction, cDNA synthesis and RT-qPCR analyses were performed as described by Bianchetti et al. (2018). Oligonucleotide sequences used in the study are detailed in Table S14.

Plastid ultrastructure and abundance
Plastid ultrastructure was assessed following methods described by Bianchetti et al. (2018). Pericarp sections from three immature fruits harvested from different plants were analysed per genotype. Plastid abundance was determined as described in Bianchetti et al. (2017). At least 15 individual cells were analysed per sample.

Chlorophyll, carotenoid and tocopherol profiling
Chlorophyll a, chlorophyll b, phytoene, phytofluene, lycopene, bcarotene and lutein levels were determined by high-performance liquid chromatography with diode-array detection (HPLC-DAD) 1100 system (Agilent Technologies) as described in Cruz et al. (2018). Tocopherol extraction and quantification were performed as described in Lira et al. (2017). The endogenous metabolite concentration was obtained by comparing the peak areas of the chromatograms with commercial standards.

Antioxidant activity, flavonoids and ascorbate quantification
For the antioxidant activity of polar extracts and flavonoid content analysis, approximately 100 mg fresh weight (FW) of powdered fruit pericarp samples was extracted with 1 mL of 80% (v:v) methanol for 30 min in an ultrasonic bath at room temperature followed by the collection of the supernatants by centrifugation (12 000 g, 2 min, 25°C). DPPH (2,2-Diphenyl-1-picrylhydrazyl) radical scavenging activity assay followed the protocol described by Furlan et al. (2015). Flavonoids were identified and quantified by HPLC systems according to the protocols and configurations described in Methods S1. Ascorbate extraction and detection were performed as described in Methods S1.

Starch, soluble sugars and citrate quantification
Starch and soluble sugar extractions were performed as described by Bianchetti et al. (2017). Starch levels were determined from the dried pellet as described in Suguiyama et al. (2014). Citrate quantification followed the protocol described in Am oros et al. .

Leaf gas exchange and fluorescence parameters
Gas exchange parameters were measured between the 2nd and 4th hour of light period on the second fully expanded leaves from shoot to apex of approximately 2-month-old plants. Analyses were performed using a portable LI-6400XTR infrared gas analyzer (LI-COR Biosciences, Lincoln, USA) adjusted to a constant chamber temperature of 25°C, reference CO 2 concentration of 400 ppm and photosynthetic photon flux density of 1000 µmol photons m À2 s À1 . Chlorophyll fluorescence parameters were measured on the third fully expanded leaves from shoot to apex using a portable fluorometer MINI-PAM (Heinz Walz GmbH, Effeltrich, Germany) following the protocol and derived calculations described in Alves et al. (2016).

Metabolite profiling
Mass spectrometry (MS)-based metabolite profiling followed the protocols described by Lisec et al. (2006) for polar compounds whereas that described in Bligh and Dyer (1959), Fiehn et al. (2000) and Ichihara and Fukubayashi (2010) was used for apolar compounds, with modifications described in Methods S1. Metabolite profile analyses were carried out in a gas chromatograph (Agilent Technologies 7890B) equipped with autoinjector (CombiPAL, CG sampler 80) and a mass-selective filter (Agilent Technologies 5977A). GC-MS configurations are described in Methods S1.

Statistical analyses
Experimental design was completely randomized. Statistical differences between groups were determined by ANOVA followed by Dunnett's test for transcript abundance analyses (with WT as a control group, a = 0.05) or Tukey's HSD test (a = 0.05) for metabolites and remaining variables. Statistical analyses were performed using JMP statistical software package version 14 (https://www.jmp.com).

Supporting information
Additional supporting information may be found online in the Supporting Information section at the end of the article. Figure S1 Construct designed for the generation of transgenic lines. Figure S2 Representative individual tomato plants from wild-type and transgenic lines. Figure S3 Validation of RNASeq analysis via RT-qPCR. Figure S4 TEM images of plastids of wild-type and transgenic fruits. Figure S5 Phylogenetic reconstruction of the CURT protein family. Figure S6 Carotenoid and tocopherol profiles in wild-type and transgenic fruits. Figure S7 Flavonoid and ascorbate contents in wild-type and transgenic fruits. Figure S8 Simplified ascorbate biosynthetic pathway. Figure S9 Brix and soluble sugar contents in wild-type and transgenic fruits. Figure S10 Impacts of PHYB2 and PHYB2 Y252H -overexpression on Brix, isoprenoid, flavonoid, ascorbate composition in Ailsa Craig tomato cultivar. Method S1 Extended Materials and Methods. Table S1 Transcript abundance of phytochrome-encoding genes in wild-type and transgenic fruits. Table S2 DEGs between transgenic and wild-type fruits. Table S3 Gene set enrichment analysis. Table S4 Transcript abundance of plastid-related genes in wildtype and transgenic fruits. Table S5 Chlorophyll, carotenoid and tocopherol content in wildtype and transgenic fruits. Table S6 Up-regulated enzyme-encoding DEGs involved in chlorophyll biosynthesis in tomato PPC2::PHYB2 Y252H mature green (MG) fruits compared to wild-type (WT). Table S7 Transcript abundance of isoprenoid-related genes in wild-type and transgenic fruits. Table S8 Antioxidant capacity and flavonoid contents in wildtype and transgenic fruits. Table S9 Transcript abundance of flavonoid and ascorbaterelated genes in wild-type and transgenic fruits. Table S10 Relative metabolite abundance registered in red ripe transgenic fruits. Table S11 Starch, soluble sugar and citric acid contents in wildtype and transgenic fruits. Table S12 Photosynthetic parameters, chlorophyll and starch contents in leaves of wild-type and transgenic plants. Table S13 Transcript abundance of sugar-related genes in wildtype and transgenic fruits. Table S14 Oligonucleotides used in this study.