AtPDS overexpression in tomato: exposing unique patterns of carotenoid self‐regulation and an alternative strategy for the enhancement of fruit carotenoid content

Summary The regulation of plant carotenogenesis is an active research area for both biological discovery and practical implementation. In tomato (Solanum lycopersicum), we demonstrate additional bottlenecks exist in the poly‐cis‐transformation of phytoene to lycopene in the context of ripening‐induced PSY1 expression and activity and reveal phytoene desaturase (PDS), as a target for manipulation towards elevated lycopene content in maturing tomato fruit. Overexpression of Arabidopsis PDS, AtPDS, elevated PDS transcript abundance in all aerial tissues resulting in both altered carotenoid accumulation and associated pathway gene expression in a tissue‐specific manner. Significant increases in downstream carotenoids (all‐trans‐lycopene and β‐carotene) and minimal changes in carotenogenic gene expression (carotenoid isomerase‐like 1, CRTIL1) suggest overexpression of heterologous AtPDS in tomato circumvents endogenous regulatory mechanism observed with previous strategies. In transgenic leaves, depletion of the PDS substrate, phytoene, was accompanied by minor, but significant increases in xanthophyll production. Alterations in the leaf carotenogenic transcript profile, including the upstream MEP pathway, were observed revealing unique feedback and feedforward regulatory mechanisms in response to AtPDS overexpression. AtPDS overexpression in the background of the tangerine (carotenoid isomerase, CRTISO) mutant exposes its potential in elevating downstream cis‐lycopene accumulation in ripe tomato fruit, as cis‐lycopene is more bioavailable yet less abundant than all‐trans‐lycopene in the wild‐type control. In summary, we demonstrate the limitation of PDS in ripening fruit, its utility in modifying carotenoid profiles towards improved quality, and reveal novel carotenoid pathway feedback regulation.


Introduction
Carotenoids (e.g. lutein, b-carotene and lycopene) and their catabolites (i.e. vitamin A and other apocarotenoids) are integral in the human diet as they ensure proper development and aid in the prevention of numerous age-related diseases (Ip et al., 2013;Lian and Wang, 2008;Rhinn and Doll e, 2012;Sharoni et al., 2012). Humans' inability to synthesize carotenoids de novo makes them dependent on plants as their primary source of dietary carotenoids, the acquisition of which is dependent upon both content in specific plant tissues and their ability to be absorbed in the digestive tract (bioavailability). The capacity of many fruits to accumulate elevated levels of carotenoids as they ripen makes fruits an important source of dietary carotenoids in many diets. Lycopene, the red carotenoid predominant in ripe tomatoes, has received increased attention as a target for manipulation due to evidence that lycopene-derived apocarotenoid signals aid in cancer prevention (Ip et al., 2013;Lian and Wang, 2008).
In plants, carotenoids are essential for light absorption in photosynthetic tissues and as provisional colorants in flowers and fruit facilitating the attraction of pollinators and seed dispersing organisms (reviewed in Nisar et al., 2015). The enhanced accumulation of carotenoids in ripening fruit is dependent on the switch from a photosynthetic to a nonphotosynthetic-state, which embodies a remarkable transition of chloroplast to chromoplast-rich cells, concurrent with strategic adjustments to the carotenoid profile within (Barry et al., 2008(Barry et al., , 2012. Ripeningassociated carotenogenic gene expression in tomato is regulated by a complex symphony of transcription factors, the gaseous hormone ethylene and epigenome dynamics driving the flux of the pathway to all-trans-lycopene (and inhibition of its subsequent metabolism) conferring the typical red colour of tomato (Barry et al., 2005;Eriksson et al., 2004;Gallusci et al., 2016;Giovannoni et al., 2017;Martel et al., 2011;Vrebalov et al., 2002Vrebalov et al., , 2009Zhong et al., 2013). In the mature fruit, phytoene synthase (the first committed step in carotenogenesis) is encoded by the highly regulated PSY1 gene and is the major limiting activity for carotenoid flux (Figure 1a) (Fraser et al., 2002). Once synthesized by PSY1, phytoene undergoes a poly-cis-transformation conferred by four desaturation reactions catalysed by phytoene desaturase (PDS) and f-carotene desaturase (ZDS) and two isomerizations facilitated by f-carotene isomerase (ZISO) and carotene isomerase (CRTISO) ultimately producing all-trans-lycopene ( Figure 1a) (Alba et al., 2005;Fantini et al., 2013;Isaacson et al., 2002;Yu et al., 2011). Subsequent processing of all-trans-lycopene is prevented through downstream repression of lycopene eand b-cyclases involved in synthesizing a-carotene (e,b-rings), the precursor to lutein and the provitamin A, bcarotene (b,b-rings) ( Figure 1a) (Ronen et al., 1999(Ronen et al., , 2000. These combined activities, tightly regulated to push carotenoid flux towards specific products in the ripe fruit, demonstrate finely tuned pathway regulation. Multiple strategies have been pursued to enhance lycopene content in tomato deploying either plant or bacterial transgenes (reviewed in Fraser et al., 2009). In carotenoid deficient staple crops (e.g. rice and potato), the use of the bacterial carotenoid desaturase, CRTI, has proven effective due to its ability to catalyse all six reactions required to convert phytoene to all-trans-lycopene (Diretto et al., 2007;Misawa et al., 1990;Ye et al., 2000). Due to mechanisms yet to be defined, CRTI introduction in planta drives carotenogenesis beyond all-trans-lycopene to b-carotene via induced lycopene b-cyclase expression (Enfissi et al., 2017;Nogueira et al., 2013;R€ omer et al., 2000;Ye et al., 2000). In contrast to the successful applications in low carotenoid tissues, ectopic expression of CRTI was counterproductive for the enhancement of lycopene content in tomato fruit.
CRTI's structure has been solved facilitating characterization of its function as an FAD-dependent oxidase/isomerase (Schaub et al., 2012). Two subsequent studies provided insight on the causal processes leading to b-carotene enhancement in the tomato including CRTI protein localization (Nogueira et al., 2013) in the plastid and lycopene-mediated pathway regulation (Enfissi et al., 2017). Remaining unclear is the potential for an uncontrolled elevation of a foreign carotenogenic protein (i.e. CRTI) to disrupt one or more of the many protein interactions involved in plant carotenogenesis, thus negatively affecting carotenoid gene 15-cis-phytoene 15,9′-di-cis-phytofluene 9,15,9′-tri-cis-ζ-carotene 9,9′-di-cis-ζ-carotene 7,9,7′,9′-tetra-cis-lycopene all-trans-lycopene 7,9,9′-tri-cis-neurosporene  Shumskaya and Wurtzel, 2013). Targeted manipulation of one or more of the four native plant enzymes carrying out limiting CRTI reactions in ripening fruit may facilitate enhanced all-trans-lycopene accumulation in an already high carotenoid tissue and provide insight into aspects of endogenous pathway regulation.
Here, we (i) identify PDS within the poly-cis-transformation of phytoene to lycopene as a useful target for manipulation towards carotenoid enhancement in ripening tomato fruit, (ii) fully characterize the effect of PDS overexpression on carotenoid content and carotenoid gene expression in aerial organs and (iii) demonstrate the utility of PDS overexpression combined with CRTISO repression via the tangerine (t 3183 ) mutation as a means to both elevate carotenoid levels and modify carotenoid accumulation profiles towards more bioavailable forms compared to wild-type control tomatoes (Cooperstone et al., 2015).

PDS becomes limiting once PSY1 is elevated during ripening
There is overwhelming evidence that PSY1 represents the major rate-limiting step in ripening-associated carotenogenesis (McQuinn et al., 2015;Nisar et al., 2015). The prevalence of intermediates in the poly-cis-transformation of 15-cis-phytoene to all-trans-lycopene as the fruit ripens suggests additional points of pathway limitation providing targets to develop alternate strategies for enhancing flux to desired downstream carotenoids. In terms of amino acid sequence identity and conserved domains, PDS and ZDS, enzymes are predicted to have evolved from CRTI, with 41% and 30% identity, respectively, and both contain the same functional domains, NAD-binding domain and amine oxidase domain (NCBI Conserved Domains: http://www.ncbi.nlm.nih.gov/Structure/cdd/ cdd.shtml) ( Figure S1). In contrast, ZISO is considerably more distant and unrelated lacking both the NAD-binding and amine oxidase domains and instead contains a nitric acid domain (NCBI Conserved Domains: http://www.ncbi.nlm.nih.gov/Struc ture/cdd/cdd.shtml) ( Figure S1). Furthermore, analysis of protein structure (TOPCONS.net) indicates ZISO is the only enzyme in this part of the pathway predicted as a transmembrane protein with multiple transmembrane domains, consistent with previous reports (Beltr an et al., 2015). For these reasons and to gain a better understanding of the previously demonstrated metabolic constraints associated with CRTI overexpression, both PDS and ZDS were the focus of the efforts reported here, while ZISO, a more distant and distinct protein at the amino acid sequence and structural level (Beltr an et al., 2015), was left for later examination.
In an effort to identify the optimal target for increasing carotenoid pathway flux post-PSY1 activity, expression patterns of PDS and ZDS were measured through four stages of fruit ripening (Mature Green, MG; breaker, Br; and 3 and 7 days postbreaker, 3DPB and 7DPB) using Quantitative Real-Time PCR. PDS was the lowest expressed gene in the fruit as compared to ZDS and CRTISO (Figure 1b). While transcription does not determine enzyme activity, the elevated prevalence of PDS substrates (phytoene and phytofluene) compared to that of ZDS (9,9 0 di-cis-f-carotene and neurosporene) in ripe fruit, as well as trace levels of PDS substrates observed in leaves (Table 1) suggest PDS activity, may be limiting during carotenogenesis. PDS shares the substrate, phytoene, with CRTI allowing for the additional assessment of potential feedback or feedforward effects from depletion of a specific substrate (i.e. phytoene) or over accumulation of a specific downstream product (i.e. all-trans-lycopene) as speculated in previous reports (Enfissi et al., 2017;Fantini et al., 2013;Kachanovsky et al., 2012). Together these observations suggested PDS as a high value target within the poly-cistransformation for manipulation with the goal of elevating carotenoid accumulation.

Heterologous overexpression of Arabidopsis PDS in tomato enhances downstream carotenoid content in ripe fruit
The potential for activating an endogenous targeted gene repression system (e.g. siRNA-mediated gene silencing) in response to the introduction of a transgene can be exacerbated due to high nucleotide sequence similarity between the endogenous gene and the transgene. The coding sequence of AtPDS shares only 72% of sequence identity with that of the endogenous SlPDS, reducing the potential for activating DNA sequence homology-based repression systems. For this reason, the Arabidopsis homolog of PDS, AtPDS (GenBank accession no. NM_202816) was deployed for overexpression rather than SlPDS, under the control of the constitutive CaMV 35S promoter in the wild-type tomato background cv. Ailsa Craig (AC++). Four independent homozygous transgenic lines (i.e. AtPDS1A.1; AtPDS3.2; AtPDS4.2; AtPDS6.2) displaying elevated levels of AtPDS transcripts in the three aerial plant organs analysed (i.e. leaves, flowers and ripe fruits) were propagated to the T 2 generation ( Figure 2a). The observed minimal effect on endogenous PDS transcript levels ( Figure 2a) indicated that any operating suppression systems had been bypassed. Ripe fruit from all four AtPDS overexpressing lines presented noticeably deeper red pigmentation compared to their wild-type counterparts (Figure 2b).
Ripe fruit displayed the greatest increase in carotenoids downstream of PDS. Flowers and leaves had reduced and elevated lutein levels, respectively (   Figure S3). Total lycopene and bcarotene were increased up to 31.1% and 42.8%, respectively (Table 2 and Figure 3b,c). With regard to the elevated total lycopene content, 31.1% equals an increase of 21.3 lg/g fresh weight in this already high lycopene tissue ( Figure 3b). The observed increases in total lycopene and b-carotene content in AtPDS overexpressing ripe are not a consequence of reduced fruit size ( Figure S4). The largest metabolic shifts were observed in the immediate products of the PDS enzymatic reaction where total fcarotene levels were >160% higher than in wild-type fruit ( Table 2 and Figure 3a).

AtPDS overexpression in ripening fruit resulted in altered transcription of CRTIL1
Considering the previously observed deregulation of transcripts associated with carotenoid biosynthesis in CRTI overexpressing fruit (R€ omer et al., 2000), it was necessary to quantify transcript abundance of genes throughout the biosynthetic pathway in ripe fruit of AtPDS.OE lines. A comprehensive analysis of the resulting carotenogenic transcript profile in AtPDS.OE ripe fruit was carried out via qRT-PCR with gene-specific primers (Table S2) and compared to transcript abundance observed in the WT ripe fruit. Contrary to the observed off-target changes in carotenogenic gene expression in CRTI overexpressing fruit first described in R€ omer et al. (2000), no consistent or statistically significant changes in PSY1, CRTL-B1 and CRTL-B2 gene expression were identified ( Figure 4a). Analysis of relative transcript levels for the remaining carotenoid biosynthetic genes (DXS; DXR; IDI1; PSY2, ZISO; ZDS; CRTISO; CRTIL1; CRTR-B1; CRTR-B2; and CRTL-E) further shows limited impact of elevated PDS on carotenoid pathway self-regulation. A solitary significant change was observed in the reduction of CRTIL1 transcript levels, a gene suggested to be involved in the isomerization of f-carotenes (Fantini et al., 2013) (Figure 4b). These results indicate that PDS overexpression can provide a means for elevating carotenoid content in ripe fruit with limited feedback or feedforward effects on carotenoid gene expression and furthermore that CRTIL1 responds to changes in flux through early steps in the pathway during tomato ripening.

AtPDS overexpression in tomato leaves and flowers alters gene expression throughout the pathway
AtPDS overexpression in leaves enhanced carotenoid production in downstream xanthophylls, lutein, b-cryptoxanthin and zeaxanthin to small but measurable degrees (Figure 5a,b). Analysis of transcript levels of carotenoid biosynthetic genes by qRT-PCR revealed that AtPDS overexpression in tomato leaves had broad regulatory consequences and influenced multiple steps throughout the carotenogenic pathway. The gene representing the ratelimiting step of the MEP pathway (Rodr ıguez-Concepci on and Boronat, 2002), deoxyxylulose 5-phosphate (DXP) synthase (DXS) was induced approximately twofold in the AtPDS.OE lines compared to the wild-type control ( Figure 5c). PSY1 was also elevated in leaves, but to a lesser amount than DXS in AtPDS.OE leaves ( Figure 5c). Lycopene b-cyclase 1 (CRTL-B1) and bcarotene-hydroxylase 1 (CRTR-B1) were both expressed at higher levels in the AtPDS.OE leaves ( Figure 5c). CRTR-B1 increased the most, (threefold to fourfold), and in combination with other changes in carotenoid gene expression provides insight into the mechanism underlying the elevated lutein, b-cryptoxanthin and zeaxanthin content ( Figure 5). b-Carotene content remained unchanged ( Figure 5). Carotenoid biosynthetic gene expression analysis of the chromoplast-rich flowers displayed limited alterations in response to heterologous PDS expression, and this was again even more limited in the maturing fruit as noted above. The gene encoding the step immediately following PDS, f-carotene isomerase (ZISO), was elevated approximately twofold compared to wild-type flowers ( Figure 6). CRTR-B1 was reduced slightly in the AtPDS.OE flowers, potentially limiting the ability to increase xanthophyll content of the petals and anthers as confirmed by HPLC ( Figure 6).

AtPDS overexpression enhances cis-lycopene accumulation in tangerine (CRTISO) mutant ripe fruit
In an effort to further amplify the value of the metabolic shift towards cis-carotenoids in tangerine (t 3183 ) mutant fruit, both AtPDS3.2 and the high pigment 1 (hp1) mutations were crossed independently into the t 3183 background. The hp1 mutation in the tomato DAMAGED DNA BINDING PROTEIN1 (DDB1) gene was selected as it confers an increase in plastid number resulting in elevated accumulation of carotenoids in ripe fruit (Liu et al., 2004), a means of carotenoid elevation distinct from the direct

AtPDS-Ripe Fruit
Rel. mRNA levels (Log Plants homozygous for the tangerine (t 3183 ) mutation and positive for the AtPDS transgene or hp1 were selected in F2 populations. All mutations and transgenes were nearly isogenic in the inbred Ailsa Craig background to minimize additional genetic variation in double-mutant lines. Phenotypic and HPLC analyses were carried out on the ripe fruit of the AtPDS3.2;t 3183 double mutant, the t 3183 parent line and the hp1;t 3183 double mutant. Compared to both the parental t 3183 and hp1;t 3183 ripe fruit, AtPDS3.2;t 3183 fruit were darker in colour ( Figure 7a). The enhanced pigmentation associated with AtPDS3.2;t 3813 was observed throughout the pericarp of cut fruit (Figure 7b).
HPLC analysis revealed that AtPDS overexpression reduced steady-state levels of phytoene and phytofluene by 64.8% and 61.8%, respectively, consistent with reductions observed when overexpressed in the wild-type background ( Figure 7c and Table 2; Figure S5). AtPDS efficiency appears to be influenced by tissue type as AtPDS3.2;t 3183 anthesis flowers only displayed a 30% reduction relative to t 3183 phytoene and phytofluene levels (Figures 7c and S6). Both AtPDS overexpression and the hp1 mutation led to a similar increase in 9,9 0 -di-cis-f-carotene content when combined with t 3183 (Figures 7c and S5), while AtPDS3.2; t 3183 alone displayed significant increases in products downstream of 9,9 0 -di-cis-f-carotene when compared to both t 3183 and hp1;t 3183 (Figures 7c and S5). Total neurosporene accumulates up to 84.5% higher in AtPDS3.2;t 3183 compared to t 3183 and 44.5% higher than hp1;t 3183 (Figures 7c and S5). Additionally, tetra-cis-lycopene (prolycopene) levels are 30.3% and 42.2% higher than in t 3183 and hp1;t 3183 , respectively (Figures 7c and  S5).
Total lycopene levels followed the same trend as prolycopene in the three genotypes with AtPDS3.2;t 3183 displaying the highest levels (Figures 7d and S5). Subsequent lycopene isomer identification and quantification in the ripe fruit demonstrates genotypespecific effects on particular lycopene isomers (Figures 7e and S7; Table S2). Introduction of AtPDS overexpression and the hp1 mutation in t 3183 conferred elevations in two unidentified lycopene isomers and 5-cis-lycopene relative to the t 3183 parental line ( Figure 7e). AtPDS3.2;t 3183 tomatoes accumulate 2 and 7 times more 5,9(5 0 ,9 0 ) di-cis-lycopene and all-trans-lycopene, respectively, compared to levels found in t 3183 tomatoes, while levels remain relatively unchanged in hp1;t 3183 ( Figure 7e). The presence of genotype-specific changes in cis-lycopene isomers and elevated total lycopene content observed in AtPDS3.2;t 3183 demonstrates the potential for targeting PDS in a means that bypasses endogenous regulation in concert with CRTISO to enhance bioavailable cis-lycopene isomers in tomato.

Discussion
AtPDS overexpression effectively enhanced carotenoid accumulation in fruit and significantly alters the carotenoid gene expression pattern in a tissue-specific manner Carotenogenesis is heavily regulated regardless of plant organ, with PSY1 activity the major rate-limiting step in ripeningassociated carotenoid biosynthesis (McQuinn et al., 2015;Nisar et al., 2015). This study demonstrates that in the context of normal ripening induction and function of PSY1, PDS activity becomes limiting and when elevated through deployment of a heterologous transgene, can significantly increase downstream carotenoid (i.e. f-carotene, lycopene and b-carotene)  Figure S2). and f-carotene content (lg/g FW) in ripe fruit of stable, T 2 generation AtPDS.OE lines compared to wild type (n = 5, error bars AESEM) (*P < 0.05; **P < 0.01; ***P < 0.001). (b) Lycopene content (lg/g FW) in ripe fruit of stable, T 2 generation AtPDS.OE lines compared to wild type (n = 5, error bars AESEM) (*P < 0.05; **P < 0.01; ***P < 0.001). (c) b-Carotene content (lg/g FW) in ripe fruit of stable, T 2 generation AtPDS.OE lines compared to wild type (n = 5, error bars AESEM) (*P < 0.05; **P < 0.01; ***P < 0.001). (d) Total carotenoid content (lg/g FW) in ripe fruit of stable, T 2 generation AtPDS.OE lines compared to wild type (n = 5, error bars AESEM) (*P < 0.05; **P < 0.01; ***P < 0.001). suggesting that as PDS activity was elevated, ZISO and ZDS now became limiting. These two genes thus present interesting targets for further carotenoid enhancement in ripening fruit with elevated PDS. Importantly and in contrast to previous observations with transgenic tomatoes expressing the bacterial CRTI (Enfissi et al., 2017;Nogueira et al., 2013;R€ omer et al., 2000), total carotenoid content was significantly elevated in AtPDS.OE ripe fruit (Table 2 and Figure 3d). Furthermore, PSY1 transcripts did not decrease, nor did lycopene b-cyclase transcripts (CRTL-B1 and CRTL-B2) increase across the four independent transgenic lines ( Figure 4). The lack of feedback regulation within the carotenogenic pathway may rely on the fact that an enzyme native to plants, PDS, was introduced rather than a more distant bacterial enzyme, CRTI. Given that both PDS and CRTI considerably deplete the phytoene pool in the plastids (based on observed steady-state levels), the observed feedback regulation does not appear directly attributable to a response to lower substrate availability. It has recently been suggested that this response may be due to an elevation in lycopene content (Enfissi et al., 2017). PDS increases the accumulation of all-translycopene in ripe tomato plastids more than with CRTI overexpression (Table 2 and   (c) Cis-carotenoid content (lg/g FW) from the poly-cis-transformation of phytoene to all-trans-lycopene in t 3183 , AtPDS3.2; t 3183 and hp1; t 3183 7DPB fruit (n = 5; error bars, AESEM) (*P < 0.05; **P < 0.01; ***P < 0.001). (d) Total lycopene content in 7DPB fruit of t 3183 , AtPDS3.2; t 3183 and hp1; t 3183 (n = 5; error bars, AESEM) (*P < 0.05; **P < 0.01; ***P < 0.001). (e) Cis-and all-trans-lycopene content (lg/g FW) accumulating in 7DPB fruit of t 3183 , AtPDS3.2; t 3183 and hp1; t 3183 (n = 5; error bars, AESEM) (*P < 0.05; **P < 0.01; ***P < 0.001). All carotenoid names identify peaks with similar characteristics and retention timing as those identified in Melendez-Martinez et al. (2013). be through means other than or in addition to elevated lycopene.
While PDS evolved from CRTI, PDS and CRTI function in contrast to each other with regard to maintenance of the FAD redox state and may differentially influence carotenoid pathway regulation in the ripening tomato fruit. It has been established that the plastid terminal oxidase, PTOX and the plastoquinone (PQ) pool controls the redox state and function of PDS in plants (Carol et al., 1999;Jo€ et et al., 2002;Josse et al., 2003;Norris et al., 1995;Yu et al., 2014). Moreover, while in photosynthetic chloroplasts, alternative mechanisms appear to aid in the maintenance of PDS's redox state, in nonphotosynthetic chromoplasts, PDS is solely dependent on PTOX and the PQ pool (Shahbazi et al., 2007). Norflurazon inhibits PDS activity by maintaining the FAD within PDS in a reduced state once it is lodged in the active site (Gemmecker et al., 2015). In contrast, CRTI cannot be inhibited by norflurazon (Sandmann and Fraser, 1993), and as suggested by Schaub et al. (2012), it remains unclear whether CRTI can be oxidized similarly to PDS by PQ in planta. Therefore, in contrast to PDS overexpression, ripe fruit expressed CRTI protein may have the potential to disrupt the redox state of the PQ pool providing an alternative explanation for the observed carotenoid pathway misregulation.
AtPDS overexpression successfully circumvents the negative feedback mechanisms observed in response to bacterial CRTI. Indeed, AtPDS overexpression associated elevation of lycopene and b-carotene in the T 2 generation ripe fruit is improved compared to values observed with other strategies (Enfissi et al., 2004;Fraser et al., 2007). Lycopene content resulting from AtPDS overexpression is higher than reported in DXS overexpression lines in the T 1 generation, while total carotenoid content stays comparable (Enfissi et al., 2004). Observations by Fraser et al. (2007) demonstrated that PSY1 overexpression increases phytoene content in ripe transgenic fruit, but has no overall effect of the steady-state levels of lycopene when compared to wildtype fruit. In contrast, AtPDS overexpression overcomes the apparent limitation in the pathway at the PDS step and increases the steady-state levels of lycopene in the ripe transgenic tomato fruit.
Many examples of pathway self-regulation have been reported including, but not limited to, the feedback regulation observed when CRTI is overexpressed in tomato fruit (Enfissi et al., 2017;Nogueira et al., 2013;R€ omer et al., 2000) and epistatic regulation of PSY1 in the spontaneous tangerine (CRTISO) mutation (Kachanovsky et al., 2012;and reviewed in Cazzonelli and Pogson, 2010). Therefore, it was not surprising to observe tissue-specific transcriptional perturbations of carotenogenic genes in response to AtPDS overexpression (Figure 8). Interestingly, artificially increasing flux via elevated PDS triggered major alterations in gene expression throughout the carotenoid biosynthetic pathway in leaves (Figures 5 and 8), while modifications in transgenic ripe fruit across all four independent AtPDS.OE lines were limited to a small change in CRTIL1 (Figures 4b and 8). Additionally, minor unique changes in pathway gene expression were also detected in the flowers but with no significant alterations in carotenoid content (Figures 6 and 8).
Considering the important protective roles of carotenoids in chloroplast stress tolerance, it is understandable that the carotenoid pathway would be more responsive to perturbations within leaf chloroplasts. The transcript induction resulting from AtPDS overexpression was similar, yet distinct from gene expression changes in tomato leaves expressing CRTI reported in Nogueira et al. (2013). In the AtPDS.OE leaves, induced DXS, PSY1, CRTL-B1 and CRTR-B1 transcripts and subsequent elevation of xanthophylls (i.e. lutein, b-cryptoxanthin and zeaxanthin) are consistent with previous reports of elevated zeaxanthin levels in cyanobacteria from elevated PDS protein (Chamovitz et al., 1993). However, the minimal changes observed in xanthophyll accumulation in response to the elevation in gene expression throughout the pathway are intriguing. Similar observations have been observed when manipulating PSY in photosynthetic tissues of Arabidopsis (L€ atari et al., 2015;Maass et al., 2009;Zhou et al., 2015). Whether via overexpression of PSY or the PSY activating ORANGE protein, carotenoid enrichment was limited to nonphotosynthetic tissues (Maass et al., 2009;Zhou et al., 2015). L€ atari et al. (2015) demonstrate that overexpression of PSY in the ccd4 mutant background results in xanthophyll accumulation to toxic levels, suggesting a role for CCD4 in controlling xanthophyll accumulation. PDS-associated enrichment of xanthophylls in tomato leaves may be similarly restricted via CCD4.

A strategy for enhancing lycopene content and bioavailability in tomato fruit
Bioavailability as it is most commonly defined (Macrae et al., 1993) equally depends on both the accessibility of the nutrient to the human body and its absorption properties therein. Impaired accessibility of carotenoids can negatively influence the overall effectiveness of carotenoid enhanced crops in addressing food security and nutritional concerns. Recent reports have suggested chromoplast ultrastructure and carotenoid sequestration effect the overall bioavailability of carotenoids, where bioavailability is improved via sequestration of carotenoids in a lipid-dissolved liquid-crystalline state (Schweiggert et al., 2012(Schweiggert et al., , 2014. However, in tomatoes and other all-trans-lycopene-rich fruits, all-trans-lycopene forms long crystalline aggregates of approximately 15 lm in length in the chromoplasts inhibiting its accessibility (Cooperstone et al., 2015;Schweiggert et al., 2012Schweiggert et al., , 2014.
Chromoplast ultrastructure has been observed to be dependent on carotenoid composition (Harris and Spurr, 1969). Furthermore, an abundance of cis-carotenoids in chromoplastrich tissues has been suggested to promote lipid-associated sequestration of carotenoids (Ben-Amotz et al., 1988). Microscopic analysis of chromoplasts in tangerine mutant (t 3183 ) fruit demonstrates that the abundance of cis-lycopene isomers promotes carotenoid sequestration in globular lipid-rich structures improving the accessibility of all lycopene isomers (Ben-Amotz et al., 1988;Cooperstone et al., 2015). Therefore, overexpression of AtPDS in the tangerine mutant (t 3183 ) of tomato provides a unique and effective strategy capable of elevating total lycopene content without complementing the tangerine mutation, thereby accumulating higher levels of the bioavailable cislycopene isomers (e.g. tetra-cis-lycopene; 5,9-di-cis-lycopene; 5cis-lycopene, etc.) and maintaining enhanced accessibility of carotenoids, including all-trans-lycopene, sequestered in a lipiddissolved liquid crystalline state (Ben-Amotz et al., 1988;Cooperstone et al., 2015;Schweiggert et al., 2012Schweiggert et al., , 2014. The strategy overexpressing AtPDS in the t 3183 background reported here provides an initial step towards the enhancement of bioavailable carotenoids in fruit crops enabled in part by regulatory insights revealed through heterologous expression of PDS in tomato fruit.

Development of overexpression constructs and plant transformation
The AtPDS overexpression (AtPDS.OE) construct was generated as described previously in Gleave (1992) (Table S1). The resulting AtPDS.OE construct was sequence-verified and transformed into S. lycopersicum cv Ailsa Craig by Agrobacterium tumafaciens (strain LBA-4404) as previously described (Van Eck et al., 2006).

DNA isolation and zygosity and copy number analysis
Genomic DNA was isolated from fresh meristematic leaf tissue as previously described (Barry et al., 2005). Verification of insertion events in T0 plants was confirmed via PCR using primers specific for the 35S promoter in the pHELLSGATE 2 vector, 35S-for and 35S-rev (Table S1). Zygosity and copy number were determined in the T1 generation via Quantitative PCR relative to the single copy polygalacturonase 2a gene (PG2a, Accession No. X04583) using a modified protocol described in Haurogn e et al. (2007) Quantitative PCR was performed in 5 lL reactions containing 2.5 lL SYBR â Green PCR Master Mix (Cat No. 4309155; Applied Biosystems, Foster City, CA); 0.75 lL of 10 lM of each primer; and 1 lL of gDNA using an ABI PRISM TM 7900HT Sequence Detection System (Applied Biosystems, Foster City, CA) under the following reaction conditions: 48°C for 2 min; 95°C for 10 min; and 40 cycles of 95°C for 15 s and 60°C for 1 min. Amplification was followed by a dissociation curve to verify specificity of the 35S promoter-specific and PG gDNA-specific primers used (Table S1). Identification of positive double mutants/transgenics was performed using a forward and reverse primers in the interior of the 35S promoter and AtPDS ORF (35S-internal-FOR and AtPDS-internal-REV, respectively) and primers spanning the deletion in the tangerine mutant (i.e. tangerine.mut-for and tangerine.mut-rev) ( Table S1). All positive double mutants/transgenics were further verified by gene expression via qRT-PCR and identification of appropriate flower and fruit phenotypes (Isaacson et al., 2002).

RNA isolation and quantitative RT-PCR analysis
Total RNA was isolated from all previously described tissue using a modified protocol from the RNeasy Minikit (Cat. No 74106, Qiagen Sciences, Germantown, MD). RNA was extracted from 200 to 300 mg of frozen powdered tissue via the addition of 800 lL modified RLT buffer (4 M guanidine isothiocyanate; 0.2 M sodium acetate, pH 5.2; 25 mM EDTA; 2.5% (w/v) PVP-40) containing 1% (v/v) b-mercaptoethanol and 2% sarcosine and separated upon the addition of 800 lL chloroform. Total RNA in the upper aqueous phase was collected on the Econospin TM minispin column (Cat. No. 1920250; Epoch Life Sciences, Missouri City, TX). The column was washed twice with 500 lL of RPE buffer (Cat. No. 1018013; Qiagen Sciences; MD), and RNA was eluted with 360 lL of nuclease-free water. DNase treatment of total RNA was achieved by adding 40 lL DNase reaction buffer (60 mM MgCl2; 400 mM Tris, pH 7.5) and 3 lL of RQ1 DNase enzyme (Cat. No. M6101; Promega, Fitchburg, WI) and allowed to incubate for 45 min at 37°C. Finally, RNA samples were washed with an equal volume of phenol: chloroform solution (500 g phenol crystals; 500 mL chloroform; 20 mL isoamyl alcohol; 0.5 g 8-hydroxy-quinoline; equilibrated with 100 mM Tris, pH 8.0) and lastly with an equal volume of chloroform. Total RNA was then precipitated overnight at À20°C and resuspended in 50-100 lL of DEPC-treated water.
Quantitative Real-time PCR was performed using the Power SYBR â Green RNA-to-C T TM 1-Step Kit (Cat No. 4309169; Applied Biosystems, Foster City, CA) in a 5 lL reaction volume (2.5 lL 2X Master Mix; 1 lM forward and reverse primers; 1 lL of total RNA; 0.46 lL DEPC-treated water; 0.04 lL RT enzyme mix). All tissue samples were represented by three biological replicates, each being carried out in triplicate. Gene-specific primers were checked for efficiency using wild-type RNA (for primer sequence see Table S1). To be able to apply the standard curve method described in Applied Biosystems User Bulletin # 2 (http://www3.a ppliedbiosystems.com/cms/groups/mcb_support/documents/gene raldocuments/cms_040980.pdf), a standard curve was included on each plate for the specific gene being analysed using wild-type RNA (serial dilutions: 50 ng; 5 ng; 0.5 ng; 0.05 ng) in triplicates. For each gene analysis, template-free and negative-RT controls were included. Real-time PCRs were carried out using an ABI PRISM TM 7900HT Sequence Detection System (Applied Biosystems, CA) under the following reaction conditions: reverse transcription at 48°C for 30 min; enzyme activation at 95°C for 10 min; followed by 40 cycles of 95°C for 15 s and 60°C for 1 min. A dissociation curve was added at the end of the run for verification of primer specificity.
ABI PRISM TM SDS version 2.3 software (Applied Biosystems, CA) was used to determine gene-specific threshold cycles (C T ) using the endogenous reference (18S rRNA) for every sample. C T s were extracted, and the standard curve method (Applied Biosystems, 1997) was applied to calculate relative mRNA levels in comparison with the wild-type control or the reference sample (equal volume and concentration of RNA from each tissue combined), which was employed for the wild-type differential expression profile.

Carotenoid extraction and analysis
Carotenoids were extracted from 200 mg of frozen tomato pericarp using a modified protocol from Alba et al. (2005). Carotenoid content of anthesis flowers was extracted from frozen petals and anthers of a single flower. The frozen tissue was homogenized in a Savant FP120 Fast Prep machine first with 15 mg Mg-carbonate and 450 lL of tetrahydrofuran (THF) twice and then a third time with 450 lL of methanol containing 2,6-Di-tert-butyl-4-methylphenol (MeOH/BHT). The homogenate was then transferred and filtered through Spin-X centrifuge filters (0.45-mm nylon filter; Corning/Costar #8170; Corning Inc., Corning, NY). Tissue debris was re-extracted with an additional 500 lL of THF to ensure complete extraction of carotenoids. The carotenoid/nonpolar phase was separated from the aqueous phase through two separation steps, first with 375 lL of petroleum ether and 150 lL of 25% NaCl and next with 500 lL of petroleum ether. The two upper phase aliquots were combined and dried down in a vacufuge (Eppendorf). Prior to drying the flower extracts, xanthophylls went through a saponification as described in Galpaz et al. (2006). The dried extract from fruit was resuspended in 250 lL of ethyl acetate, while dried flower extracts were resuspended in 500 lL. All solvents used were HPLC grade.

Supporting information
Additional Supporting Information may be found online in the supporting information tab for this article: Figure S1 Assessment of protein structure of tomato PDS; ZDS; and ZISO compared to CRTI from bacteria. Figure S2 Representative chromatograph (285 nm-best wavelength to show all peaks) of carotenoids detected in the wild type ripe tomato. Figure S3 Absence of phytoene (lg/g FW) accumulation in young leaves of AtPDS.OE.1A, AtPDS.OE.3, AtPDS.OE.4, and AtPD-S.OE.6 (n = 5, Error bars AESEM). Figure S4 Fruit size in terms of mass (g) of AtPDS.OE.1A, AtPDS.OE.3, AtPDS.OE.4, and AtPDS.OE.6 ripe fruit compared to wild type ripe fruit (n > 10, Error bars AESEM). Figure S5 Chromatographs with carotenoids detected in the tangerine (t 3183 ); AtPDS3.2;t 3183 ; and hp1;t 3183 ripe tomatoes. Figure S6 Tissue type negatively affects the extent of AtPDS overexpression in tangerine mutant flower. Figure S7 Representative chromatograph (454 nm) of stereoisomers of lycopene red, all-trans-lycopene standard (4°C) and blue, stereomutated all-trans-lycopene (80°C for 1 h). Table S1 Primer sequences Table S2 Identification of lycopene stereoisomers detected in the tangerine (t 3183 ) ripe fruit based on artificial stereomutation of All-trans-lycopene standard according to Melendez et al. (2013)