Differential Regulation of Phytoene Synthase PSY1 During Fruit Carotenogenesis in Cultivated and Wild Tomato Species (Solanum section Lycopersicon)

In plants, carotenoids define fruit pigmentation and are involved in the processes of photo-oxidative stress defense and phytohormone production; a key enzyme responsible for carotene synthesis in fruit is phytoene synthase 1 (PSY1). Tomatoes (Solanum section Lycopersicon) comprise cultivated (Solanum lycopersicum) as well as wild species with different fruit color and are a good model to study carotenogenesis in fleshy fruit. In this study, we identified homologous PSY1 genes in five Solanum section Lycopersicon species, including domesticated red-fruited S. lycopersicum and wild yellow-fruited S. cheesmaniae and green-fruited S. chilense, S. habrochaites and S. pennellii. PSY1 homologs had a highly conserved structure, including key motifs in the active and catalytic sites, suggesting that PSY1 enzymatic function is similar in green-fruited wild tomato species and preserved in red-fruited S. lycopersicum. PSY1 mRNA expression directly correlated with carotenoid content in ripe fruit of the analyzed tomato species, indicating differential transcriptional regulation. Analysis of the PSY1 promoter and 5′-UTR sequence revealed over 30 regulatory elements involved in response to light, abiotic stresses, plant hormones, and parasites, suggesting that the regulation of PSY1 expression may affect the processes of fruit senescence, seed maturation and dormancy, and pathogen resistance. The revealed differences between green-fruited and red-fruited Solanum species in the structure of the PSY1 promoter/5′-UTR, such as the acquisition of ethylene-responsive element by S. lycopersicum, could reflect the effects of domestication on the transcriptional mechanisms regulating PSY1 expression, including induction of carotenogenesis during fruit ripening, which would contribute to red coloration in mature fruit.


Introduction
Tomato (Solanum lycopersicum L.) fruit represents a fleshy berry, which during ripening, changes the color from green to red (most common) or yellow, orange, pink, brown, or purple, primarily because of chlorophyll degradation and intensification of carotenoid synthesis and accumulation [1,2]. S. lycopersicum belongs to the Solanum section Lycopersicon, which also includes 12 related wild tomato species with different evolutionary ages. Most wild tomatoes produce green or dark-green fruit, with the exception of three species, which have yellow (Solanum cheesmaniae [L. Riley] Fosberg and Solanum galapagense S.C. Darwin and Peralta) or red (Solanum pimpinellifolium B. Juss.) fruit color depending on the lycopene to β-carotene ratio [3,4]. All tomato species have flower petals of bright yellow-orange color due to the accumulation of carotenoids neoxanthin, violaxanthin, and lutein [5]. Carotenoids not only color flower organs and fruit to attract insects and animals for pollination and seed dispersion but are also involved in photosynthesis, photo-oxidative stress defense, and phytohormone production during plant vegetative growth [6][7][8][9][10].  Compared to the PSY1 gene of S. lycopersicum cv. Heinz 1706, the average variability of the identified PSY1 genes was 12.8% for complete sequences (641 SNPs in the aligned 4995-bp gene portion): 8.5% for exons (105 SNPs in 1239 bp), 12.2% for introns (298 SNPs in 2117 bp), and 14.6% for 5 -UTR (238 SNPs in 1639 bp); 5 -UTR and exon I were the most variable parts. In total, 51 (48.57%) exonic SNPs were non-synonymous and found mostly in PSY1 homologs of GF tomatoes ( Figure 1).

Structural Analysis of Tomato PSY1 Homologs
The size of the translated PSY1 proteins was the same for all studied tomato species: 412 amino acids (aa) ( Table 1). Similar to S. lycopersicum PSY1, the putative PSY1 protein homologs contained a conserved phytoene synthase domain head-to-head (HH)-IPPS (75-405 aa, according to the NCBI-CDD) characteristic for the isoprenoid biosynthesis C1 superfamily. The N-terminal transit peptide (TP) was determined according to a previous study [12]; its predicted cleavage site V61/R62 was the same for the analyzed PSY1 homologs, excluding V61/W62 in PSY1 of S. cheesmaniae LA0421. However, according to Resource UniProtKB (Mitofates, Predotar, and TargetP tools) the PSY1 phytoene synthase domain was mapped to 130-412 aa and TP-to 1-129 aa, which is consistent with the plastidic localization of PSY1; in this case, the TP cleavage site was located at A129/E130 and was the same for all analyzed PSY1 proteins.
No species or fruit color-specific structural features were observed in TP or the HH-IPPS domain. Considering that reliable modeling covered the 114-401-aa region of PSY1, which included a domain conserved in dehydrosqualene synthases (also belonging to the isoprenoid biosynthesis C1 superfamily), the obtained data may support the predicted 1-129-aa location of TP in PSY1, which could extend as far as the end of the fourth α-helix (L117-E129), i.e., up to the active site (130-YAKTF-134) ( Figure 4). Although the TP sequence varied among plant PSY1 enzymes, its C-terminus (α-helix with motif LSEAYDRCGEVCAE) was modeled with more than 90% confidence, suggesting its conservation as the region preceding the cleavage site.
conserved in dehydrosqualene synthases (also belonging to the isoprenoid biosynthesis C1 superfamily), the obtained data may support the predicted 1-129-aa location of TP in PSY1, which could extend as far as the end of the fourth α-helix (L117-E129), i.e., up to the active site (130-YAKTF-134) ( Figure 4). Although the TP sequence varied among plant PSY1 enzymes, its C-terminus (α-helix with motif LSEAYDRCGEVCAE) was modeled with more than 90% confidence, suggesting its conservation as the region preceding the cleavage site.

PSY1-Based Phylogeny of Tomato Species
The coding sequences of the identified and already known PSY1 genes were used to analyze the phylogeny of the Solanaceae family. The foundation of the Solanaceae phylogenetic tree was occupied by P. axillaris, followed by N. tabacum, L. barbarum, and C. annuum ( Figure 5). The Solanum species were the most recent; eggplant and the related species were grouped in the basic sub-cluster, whereas sister clades of the Petota and Lycopersicon species were positioned at the top of the tree ( Figure 5). Tomatoes were divided into two main clades containing RF and GF species, respectively. The GF clade was rooted by the most ancient species S. habrochaites and S. pennellii grouped together, and included YF S. cheesmaniae ( Figure  5).

PSY1-Based Phylogeny of Tomato Species
The coding sequences of the identified and already known PSY1 genes were used to analyze the phylogeny of the Solanaceae family. The foundation of the Solanaceae phylogenetic tree was occupied by P. axillaris, followed by N. tabacum, L. barbarum, and C. annuum ( Figure 5). The Solanum species were the most recent; eggplant and the related species were grouped in the basic sub-cluster, whereas sister clades of the Petota and Lycopersicon species were positioned at the top of the tree ( Figure 5). Tomatoes were divided into two main clades containing RF and GF species, respectively. The GF clade was rooted by the most ancient species S. habrochaites and S. pennellii grouped together, and included YF S. cheesmaniae ( Figure 5). Arabidopsis thaliana (Brassicaceae) PSY was used as an outgroup. Analysis was performed using the neighbor-joining method. The optimal tree with the sum of branch length = 2.35980228 is shown. Percentages of replicate trees in which the associated taxa clustered together in the bootstrap test (1000 replicates) are shown next to the branches. The tree is drawn to scale, with branch lengths in the same units as those of the evolutionary distances used to infer the phylogenetic tree. The evolutionary distances were computed using the Jukes-Cantor method and are in the units of the number of base substitutions per site. For each species, the color of the ripe fruit is indicated to the right of the name.

PSY1 Expression Pattern
PSY1 expression was tested in young leaves, flower buds with green petals, yellow petals, and mature green (MG) and ripe fruit of four tomato species differing in fruit color-GF S. habrochaites LA2144 and S. pennellii LA0716, YF S. cheesmaniae LA0421, and RF S. lycopersicum (cv. Heinz 1706-BG). The aim was to analyze a possible correlation between the level of PSY1 expression and the content of carotenoids in fruits, and the choice of species, in addition to the ripe fruit color, was determined by the species phylogeny: RF S. lycopersicum as the most recent species, YF S. cheesmaniae as an evolutionary older species, and also the two most ancient GF species S. pennellii and S. habrochaites. S. chilense was not included in the comparative analysis; this species is another GF tomato, evolutionary located between the recent RF and YF and the most ancient GF species, and its fruits contained the same amount of carotenoids as other GF species.
In YF S. cheesmaniae and RF S. lycopersicum, PSY1 mRNA was expressed in all tested tissues, except for the leaves of S. cheesmaniae; PSY1 expression was consistently upregulated from a very low level in leaves to a high level in ripe fruit in both species; however, it was about five times higher in S. lycopersicum than in the S. cheesmaniae fruit ( Figure 6). Arabidopsis thaliana (Brassicaceae) PSY was used as an outgroup. Analysis was performed using the neighbor-joining method. The optimal tree with the sum of branch length = 2.35980228 is shown. Percentages of replicate trees in which the associated taxa clustered together in the bootstrap test (1000 replicates) are shown next to the branches. The tree is drawn to scale, with branch lengths in the same units as those of the evolutionary distances used to infer the phylogenetic tree. The evolutionary distances were computed using the Jukes-Cantor method and are in the units of the number of base substitutions per site. For each species, the color of the ripe fruit is indicated to the right of the name.

PSY1 Expression Pattern
PSY1 expression was tested in young leaves, flower buds with green petals, yellow petals, and mature green (MG) and ripe fruit of four tomato species differing in fruit color-GF S. habrochaites LA2144 and S. pennellii LA0716, YF S. cheesmaniae LA0421, and RF S. lycopersicum (cv. Heinz 1706-BG). The aim was to analyze a possible correlation between the level of PSY1 expression and the content of carotenoids in fruits, and the choice of species, in addition to the ripe fruit color, was determined by the species phylogeny: RF S. lycopersicum as the most recent species, YF S. cheesmaniae as an evolutionary older species, and also the two most ancient GF species S. pennellii and S. habrochaites. S. chilense was not included in the comparative analysis; this species is another GF tomato, evolutionary located between the recent RF and YF and the most ancient GF species, and its fruits contained the same amount of carotenoids as other GF species.
In YF S. cheesmaniae and RF S. lycopersicum, PSY1 mRNA was expressed in all tested tissues, except for the leaves of S. cheesmaniae; PSY1 expression was consistently upregulated from a very low level in leaves to a high level in ripe fruit in both species; however, it was about five times higher in S. lycopersicum than in the S. cheesmaniae fruit ( Figure 6). For example, PSY1 expression in S. lycopersicum leaves differed significantly from those in S. cheesmaniae, S. habrochaites and S. pennellii leaves, which is, respectively, denoted by three letters "bcd" above the L-bar.
In GF S. pennellii, PSY1 transcription was also detected in all tested tissues; however, in contrast to YF and RF species, it was maximal in yellow petals and minimal in MG and ripe fruit, where its level was approximately 5-13 times lower than in S. cheesmaniae, and 65 times lower than in S. lycopersicum. In another GF species S. habrochaites, PSY1 mRNA was detected, at a very low level, only in buds and petals ( Figure  6). PSY1 expression was minimal or absent in the leaves of all four tomato species and very low in buds of all wild accessions. The highest PSY1 mRNA level in yellow petals was observed in S. lycopersicum, where it exceeded that in S. cheesmaniae/S. pennellii and S. habrochaites by two and 10 times, respectively ( Figure 6).

Carotenoid and Chlorophyll Content
Accumulation of total (x + c) and specific carotenoids (lycopene and β-carotene), as well as colorless carotenoid precursors (phytoene and phytofluene) and chlorophylls a and b, was assessed in ripe fruit and leaf of GF species S. habrochaites (LA2144), S. pennellii (LA0716) and S. chilense (LA1963), YF S. cheesmaniae (LA0421) and RF S. lycopersicum (cv. Heinz 1706-BG). The results indicated that the fruit of wild tomatoes lacked lycopene and contained 10-20 times less of total carotenoids than those of S. lycopersicum; however, fruit β-carotene content was the same for all analyzed species (Table 2, Figure 7). At the same time, the For example, PSY1 expression in S. lycopersicum leaves differed significantly from those in S. cheesmaniae, S. habrochaites and S. pennellii leaves, which is, respectively, denoted by three letters "bcd" above the L-bar.
In GF S. pennellii, PSY1 transcription was also detected in all tested tissues; however, in contrast to YF and RF species, it was maximal in yellow petals and minimal in MG and ripe fruit, where its level was approximately 5-13 times lower than in S. cheesmaniae, and 65 times lower than in S. lycopersicum. In another GF species S. habrochaites, PSY1 mRNA was detected, at a very low level, only in buds and petals ( Figure 6). PSY1 expression was minimal or absent in the leaves of all four tomato species and very low in buds of all wild accessions. The highest PSY1 mRNA level in yellow petals was observed in S. lycopersicum, where it exceeded that in S. cheesmaniae/S. pennellii and S. habrochaites by two and 10 times, respectively ( Figure 6).

Carotenoid and Chlorophyll Content
Accumulation of total (x + c) and specific carotenoids (lycopene and β-carotene), as well as colorless carotenoid precursors (phytoene and phytofluene) and chlorophylls a and b, was assessed in ripe fruit and leaf of GF species S. habrochaites (LA2144), S. pennellii (LA0716) and S. chilense (LA1963), YF S. cheesmaniae (LA0421) and RF S. lycopersicum (cv. Heinz 1706-BG). The results indicated that the fruit of wild tomatoes lacked lycopene and contained 10-20 times less of total carotenoids than those of S. lycopersicum; however, fruit β-carotene content was the same for all analyzed species (Table 2, Figure 7). At the same time, the content of total carotenoids in fruit was significantly lower than in leaf (Table 2). However, leaf-specific carotenogenesis is mainly mediated by another phytoene synthase isoform PSY2; thus, a comparison of the carotenoid content in fruit and leaf could be correct if the ripe green fruit consisted only of photosynthetic tissues, which is obviously not the case. Chlorophyll content in the fruit of GF tomatoes was considerably reduced compared to the leaves, whereas in the fruit of YF S. cheesmaniae and RF S. lycopersicum it was absent at all ( Table 2).
Plants 2020, 9, x FOR PEER REVIEW 12 of 25 content of total carotenoids in fruit was significantly lower than in leaf (Table 2). However, leaf-specific carotenogenesis is mainly mediated by another phytoene synthase isoform PSY2; thus, a comparison of the carotenoid content in fruit and leaf could be correct if the ripe green fruit consisted only of photosynthetic tissues, which is obviously not the case. Chlorophyll content in the fruit of GF tomatoes was considerably reduced compared to the leaves, whereas in the fruit of YF S. cheesmaniae and RF S. lycopersicum it was absent at all (Table 2).   In the fruit of GF species, the content of total carotenoids was almost equal to the β-carotene content ( Figure 7, Table 2), which means the lowest possible amount of other types of colorful carotenoids. In the fruit of YF S. cheesmaniae, β-carotene accounted for about 2/3 of the total carotenoids ( Figure 7, Table 2), which may indicate the accumulation of α-carotene and lutein, as was previously shown for yellow-fruited tomato 1 mutant fruit [37]. We assume that the fruit of S. cheesmaniae produce so little lycopene that it could be fully converted into αand β-carotene by lycopene cyclases.
The potential accumulation of phytoene and phytofluene was assessed by the absorption spectra of fruit extracts. The characteristic spectral features of colorless carotenoid precursors were revealed only in the red ripe fruit of S. lycopersicum cv. Heinz, while in the spectra of ripe fruit of GF and YF accessions, these spectral features were not recorded, which indicates a minor content of carotenoid precursors (Supplementary File). The sizes of the promoter/5 -UTR were 906/1535 bp in S. lycopersicum and 919/1550 bp in S. pennellii. Compared to the S. lycopersicum PSY1 promoter and 5 -UTR, the corresponding S. pennellii PSY1 regions contained 56 and 63 SNPs, indicating 6% and 3.98% variability, respectively; these variability levels were significantly higher than those in the whole gene (from start to stop codon), cDNA and protein: 2.54%, 0.8% and 0.95% (86, 10 and 3 substitutions), respectively.

Promoter and 5 -UTR Analysis
Search for sites important for PSY1 transcription revealed that the PSY1 promoter and 5 -UTR contained 37 types of regulatory elements related to tissue-specific expression and response to phytohormones and stress factors. Among them, 32 were common for PSY1 of both species, including core promoter elements and 11 were species-specific; among the latter, there were light-and ethylene-responsive elements, sites associated with reactions to salicylic acid, cold and wounding and binding sites of the MYB transcription factor implicated in drought response (Table 3). Compared to the ancient S. pennellii, the S. lycopersicum PSY1 promoter/5 -UTR region lost light-(chs-Unit 1 m1 and TCT motif), salicylic acid-(two TCA sites) and drought-(MBS) responsive elements and an AAGAA motif with unknown function, while acquiring other light-responsive elements (3-AF3 binding site and GATA motif) as well as ethylene-(ERE) and wound-(WUN-motif) responsive elements and an AACCTAACCT motif with unknown function (Table 3).   The elements that are different (present or not) between the two species are highlighted in bold and pink. The elements that are different (present or not) between the two species are highlighted in bold and pink.

Discussion
The existing wide variety of tomato cultivars, which differ in fruit color and other economically important traits, are derived from the most evolutionary recent species S. lycopersicum. It is proposed that tomato domestication began in South America, when wild RF species S. pimpinellifolium gave rise to S. lycopersicum var. cerasiforme, further evolution of which in Mesoamerica resulted in S. lycopersicum var. lycopersicum [38,39] that, together with its 12 wild relatives, now form Solanum section Lycopersicon [3].
Tomato diversification due to domestication is distinct from the natural species divergence as it passed through various genetic bottlenecks caused by the selection of a limited set of traits valuable for humans, which significantly limited genomic diversity among modern S. lycopersicum cultivars [36,40,41]. As a result, S. lycopersicum differs from its wild ancestors in a wide range of acquired morphophysiological characteristics (so-called domestication syndrome) and in genetic diversity, which constitutes no more than 5% of that existing among wild tomatoes [39]. Fruit color and carotenoid content are included in the domestication syndrome; these features are controlled by qualitative trait genes, including PSY1 encoding phytoene synthase, the key enzyme of carotenogenesis in fruit, which has a conserved structure but also contains polymorphisms associated with the evolution of plant species [42,43].
In the present study, PSY1 homologous genes were identified and characterized in one cultivated (S. lycopersicum cv. Heinz) and four wild (S. pennellii, S. habrochaites, S. chilense, and S. cheesmaniae) tomato species that differ in ripe fruit color ( Table 1). All identified homologs were highly conserved and had the same functionally important residues, indicating that the enzymes should have the same catalytic mechanism underlain by the formation of the C-terminal catalytic site-a large central cavity composed of two anti-parallel alpha-helices and motifs DELVD and DVGED at the opposite sides of the cavity (Figures 2 and 4) [44]. These data indicate that the catalytic function of the PSY1 homologs in carotenogenesis is conserved in wild and domesticated tomato species.
The N-terminal TP is responsible for PSY translocation into plastids [10,12], and its polymorphism may be one of the main species-specific signatures in plants [45]. Indeed, in the analyzed tomato species, the N-terminal TP was found to be the most polymorphic PSY1 region, which distinguished Lycopersicon species from the other Solanaceae members (Figures 3 and 4). Furthermore, it appears that in the course of evolution, TPs of Capsicum and Solanum PSY1 enzymes have lost the FSCLGGSRTKNGRIFSVRSAIVATPAGEM motif and acquired the CNERIKRG motif instead, which was later deleted in Petota and Lycopersicon species (Figure 3).
It has been suggested that carotenogenic enzymes can be used to shed light on the molecular evolution of different plant species [43]. Phylogenetic analysis based on whole-genome sequence data or trait-encoding genes produces similar results, to which exons contribute more evolution-related information than introns as they are under higher selective pressure [42]. Therefore, in this study, we examined the possibility of using PSY1 coding sequences to evaluate Solanaceae phylogeny ( Figure 5). The results indicated that the analyzed tomato species were grouped according to the generally accepted evolution history of nightshades, which suggests major splits between tomato and potato (eight million years ago [Mya]), eggplant and tomato/potato (14 Mya), Solanum and Capsicum (19 Mya), and Solanum and Nicotiana (24 Mya) [46]. Fleshy fruit developed during the latter split had the ability to acquire color because of carotenogenesis. In this study, the analyzed tomatoes could be further divided into two main clades of RF and GF species based on the PSY1 sequence ( Figure 5), indicating that PSY1 may be a good phylogenetic marker to investigate the recent evolution of tomatoes.
In the GF clade, the most distant S. habrochaites and S. pennellii were grouped together, which is consistent with the incompletely resolved dichotomy between the two species belonging to different informal groups, Eriopersicon and Neolycopersicon, respectively [47]; sometimes, S. pennellii is considered to be an intermediate species between S. habrochaites and S. arcanum [42]. The other two informal groups are Lycopersicon, which includes all the YF-RF tomato species, and Arcanum [47]. Carotenoid accumulation is the main feature of Lycopersicon species: the red color of S. lycopersicum and S. pimpinellifolium fruit is caused by the accumulation of lycopene (up to 500-fold during ripening [40]), which in S. cheesmaniae and S. galapagense is further processed into β-carotene accounting for yellow-to-orange fruit color [48,49]. In Neolycopersicon, Eriopersicon, and Arcanum species, the level of carotenoid biosynthesis in fruit is similar to that in photosynthetic tissues [49].
Considering the conserved structure of PSY1 as the key carotenogenic enzyme in all tomato species, the question is why the fruit of S. lycopersicum turns red, whereas those of wild tomatoes remain green or only slightly yellow. The answer may lie in differential regulation of PSY1 expression, the level of which positively correlates with carotenoid content. Thus, in maize, PSY1 is transcribed in yellow endosperm but not in the carotenoid-deficient white endosperm [17,50] and in rice, PSY1 overexpression makes white endosperm turn yellow because of β-carotene accumulation [51]. In RF S. lycopersicum, PSY1 expression is low in photosynthetic tissues and high in petals and ripe fruit [11,12], whereas in GF species (S. peruvianum, S. chilense, S. pennellii and S. chmielewskii) PSY1 is not expressed in ripe fruit, which do not accumulate lycopene and contain 100 times less β-carotene than those of RF tomatoes [31,32].
However, the question of why wild tomato ripe fruits remain green cannot be answered simply by the PSY1 expression level. For instance, overexpression of PSY in A. thaliana does not turn photosynthetic leaves into orange or red, and the content of carotenoids there remains unchanged, while β-carotene accumulation in non-photosynthetic tissues sharply increases [52]. The green color of the ripe fruit in GF tomatoes may be due to the lack of special globular structures for carotenoid accumulation, which typically present in the chromoplasts of non-photosynthetic tissues, or to the impaired chromoplast formation [33,35,53,54]. In this regard, it is interesting that PSY1 expression and PSY1 enzymatic activity are quite significant in the fruit of GF S. habrochaites (accession LA1777); however, there is no carotenoid accumulation because of impaired chromoplast formation [33] due to a mutation in the ORANGE protein [35,54].
Consistent with these observations, in the current study, we detected PSY1 transcription in yellow-colored petals and red-and yellow-colored fruits but not in leaves (Figure 6), which confirms chromoplast specificity of PSY1 expression in tomato species [12].
The absence of yellow-to-red pigments in the ripe green fruit may result from a metabolic bottleneck and blockage of reactions following PSY1-catalyzed phytoene biosynthesis. Then the ripe green fruit could accumulate an excess of colorless precursors, which in RF tomatoes are rapidly processed to form colorful carotenoids. In this study, significant amounts of phytoene and phytofluene were observed in the ripe fruit of only RF species (and these amounts directly correlated with the content of colorful carotenoids), but not in the GF or YF tomatoes (Supplementary file). Apparently, there is no blocking of the reactions following the synthesis of phytoene and phytofluene, and the low content of colorful carotenoids in the ripe fruit of GF and YF tomatoes may be explained by low amounts of carotenoid precursors. Thus, the final content of carotenoids in the ripe fruit is largely controlled by the level of PSY1 expression, determining the amount of colorless precursors.
Overall, in this study, the content of colorful carotenoids in ripe fruit directly correlated with the level of PSY1 expression-highest in S. lycopersicum, moderate in S. cheesmaniae, and lowest in S. habrochaites and S. pennellii (Figures 6 and 7; Table 2).
PSY1 of S. habrochaites LA2144 behaved differently from that of S. habrochaites LA1777 [33] and similar to the other analyzed GF species in that it was weakly expressed in ripe fruit and produced small amounts of carotenoids (Figures 6 and 7). The latter may be the result of β-carotene biosynthesis in remaining chloroplasts of green fruit tissues due to the expression of both PSY1 and PSY2 [12]. Ripe fruit of YF S. cheesmaniae had five times lower PSY1 expression than those of RF S. lycopersicum cv. Heinz and accumulated β-carotene instead of lycopene (Figures 6 and 7), which was probably due to increased activity of chromoplast-specific lycopene beta-cyclase caused by polymorphisms [42,48].
The expression level of PSY1 in RF tomatoes was much higher than in RF C. annuum [55], which was thought to compensate for low PSY1 activity in tomato caused by Y136N mutation [10] that was detected in all PSY1 homologs identified in the current study. Two other known substitutions that can affect the level of carotenoid biosynthesis (the C-terminal P to L substitution in grass PSY-E1, found as a result of white endosperm selection, and capable of reducing the biosynthesis of carotenoids in grains [56], and A191D, leading to an increase in PSY enzymatic activity [57]) were not detected in the identified tomato PSY1 homologs.
In wheat, the phenotypic variation of grain yellow pigments is significantly associated with three Psy1-A1 alleles, differing in the effect on carotenoid content [58]. In the fourth intron of Psy1-A1a, related to a large reduction in grain pigment, the 676-bp INDEL was detected, which could be linked to mutations in a regulatory region of the gene that alters its expression [58].
It may be proposed that species-specific differences in PSY1 expression may depend on variations in the promoter/5 -UTR sequence, in particular, the presence and location of regulatory elements and transcription factor-binding sites. Thus, S. lycopersicum PSY1 has been shown to contain 5 -UTR with MBS, ABRE, and TC-rich repeats [59]. In the 2.5-kb PSY1 region comprising the promoter and 5 -UTR, we found 37 types of regulatory elements ( Table 3), indicating that PSY1 transcription may be regulated in response to light, abiotic stresses, and hormones such as methyl jasmonate (MeJA), auxins, abscisic acid (ABA), and ethylene. There were variations in the presence of regulatory elements between RF S. lycopersicum and GF S. pennellii, such as the loss of some and acquisition of the other light-responsive elements, and the lack of drought-responsive elements in S. lycopersicum, which may indicate lower resistance of tomato cultivars to abiotic stresses compared to their wild relatives. ABA is an apo-carotenoid involved in plant developmental processes, including seed maturation and dormancy and stress tolerance [60], whereas JA and auxins together promote resistance to necrotrophic pathogens [61]. The presence of hormone-responsive elements in PSY1 5 -UTR/promoter of both RF and GF species indicate the involvement of PSY1 in the regulation of hormone-mediated signaling in tomato. Gibberellin (GAs)-responsive elements have been previously reported only in 5 -UTR of the S. lycopersicum PSY2 gene [60] but not in the PSY1 gene, despite the fact that GAs participate in fruit senescence and have the same precursor (GGPP) as carotenoids [61]. Although we also did not identify GAs-responsive elements in the PSY1 5 -UTR/promoter, we found them in PSY1 intron and exon sequences-P-box (CCTTTTG) in introns IV and V of all GF species and GARE (TCTGTTG) in exon I and intron II of both RF and GF species and in intron III of S. habrochaites LA2144.
The fruit ripening process, including ethylene production, was shown to be independent of carotenogenesis [62]; nevertheless, we observed that S. lycopersicum acquired ethylene-responsive elements in the PSY1 promoter/5 -UTR (Table 3), which could account for stronger induction of carotenogenesis during fruit ripening compared to S. pennellii, and, consequently, contribute to red pigmentation in fruit.
However, despite these differences, the overall arrangement of regulatory elements in the PSY1 5 -UTR/promoter is similar in RF S. lycopersicum and GF S. pennellii, suggesting a significant degree of conservation of the PSY1 transcriptional regulatory mechanism between RF and GF species.

Plant Material
Accessions of tomato species (Solanum section Lycopersicon) ( Table 1) were kindly provided by the Tomato Genetics Resource Center (USA; https://tgrc.ucdavis.edu/). Plants were grown in a greenhouse with temperature kept at 28 • C/23 • C during a 16-h/8-h day/night light cycle (light intensity, 300-400 µmol m −2 s −1 ). Young leaves, whole flower buds (closed flowers in the pre-anthesis stage with green petals), colored (yellow) petals, and MG and ripe fruit were collected in two biological replicates and homogenized in liquid nitrogen. The MG stage was defined as firm green fruit of a final (maximal) size, and ripe fruit in GF species was defined by softness and in YF or RF species-by a change of color from green to yellow or red, respectively, due to chlorophyll degradation and carotenoid accumulation [36,63,64].

Gene Identification
To amplify the full-length PSY1 genes from tomato species, gene-specific primers (Table 4) were designed based on PSY1 genomic sequences of S. lycopersicum (cv. Heinz 1706) available in NCBI GenBank (Gene ID: 543988, NC_015440.3 [4350836 . . . 4355976], http://www.ncbi.nlm.nih.gov/Genbank, genome annotation releases), and Sol Genomics Network (Solyc03g031860.2.1, https://solgenomics.net/); manual revision of sequence polymorphisms and additional evaluation were performed using Primer3 (http://frodo.wi.mit.edu/primer3/). Genomic DNA was isolated from young leaves of a single plant of each species accession, as previously described [65], and used as a template (100 ng) for PCR amplification at the following conditions: initial denaturation at 94 • C for 10 min, 35 cycles of denaturation at 94 • C for 30 s, primer annealing at 55 • C for 30 s, and extension at 65 • C for 4.5 min, and a final extension at 65 • C for 10 min. The amplified PCR products of the expected size were purified by using the QIAEX ® II Gel Extraction kit (QIAGEN, Hilden, Germany), cloned in the pGEM ® -T Easy (Promega, Madison, WI, USA), and sequenced (3-5 clones for each accession) on ABI Prism 377 DNA Sequencer (Applied Biosystems, Waltham, MA, USA) using the designed primers (Table 4).

Structural and Phylogenetic Analysis
Multiple sequence alignments and structural and phylogenetic analyses of PSY1 genes and encoded proteins were conducted with MEGA 7.0 [66]. Phylogenetic dendrograms were constructed based on complete gene, cDNA, and protein sequences using the neighbor-joining [67] and Jukes-Cantor [68] methods; confidence for tree topologies was estimated by bootstrap values of 1000 replicates.
For comparative structural analysis, the complete sequences and/or transcripts of PSY1 gene homologs were extracted from the NCBI database. We used PSY1 from the model plant Arabidopsis thaliana (Gene ID:

Gene Expression
Total RNA was isolated from individual samples of young leaves, flower buds with green petals, yellow petals, and fruit using the RNeasy Plant Mini Kit (QIAGEN, Hilden, Germany), qualified by gel electrophoresis, and used for first-strand cDNA synthesis (Reverse Transcription System; Promega, Madison, WI, USA) with an oligo-dT primer. The RNA and cDNA concentrations were quantified by fluorimetry (Qubit ® Fluorometer, Thermo Fisher Scientific, Waltham, MA, USA). Quantitative real-time (qRT)-PCR was performed with 2.5 ng of cDNA, SYBR Green RT-PCR mixture (Syntol, Moscow, Russia), and specific primers (Table 4) at the following cycling conditions: initial denaturation at 95 • C for 5 min and 40 cycles of denaturation at 95 • C for 15 s and annealing/extension at 60 • C for 40 s. To normalize the levels of PSY1 expression, two reference tomato genes Expressed (SGN-U346908) and actin 2/7 (NM_001330119.1) [73,74] were used.
The qRT-PCR results were statistically analyzed with Graph Pad Prism version 7.02 (GraphPad Software Inc., San Diego, CA, USA; https://www.graphpad.com/scientific-software/prism/). The data were expressed as the mean ± standard deviation (SD) based on three technical replicates of two biological replicates. The unequal variance (Welch's) t-test was applied to assess statistical significance (p-value < 0.05) of differences in gene expression between tissues within the same species and between the same tissues of different tomato species.

Carotenoid and Chlorophyll Content
Total carotenoid content was measured by spectrophotometry in two biological and three technical replicates using a modified Folch method [75,76]. Briefly, 0.2 g of plant tissue was homogenized in Folch solution (2:1 chloroform-methanol [v/v]) in the presence of trace Mg 2 CO 3 amounts [76], incubated at 4 • C for 1 h, and centrifuged at 4000 rpm for 10 min at 4 • C. The lower chloroform phase was collected and used for spectrophotometric assay of chlorophyll, lycopene, β-carotene, and total carotenoid contents were measured in acetone-hexane solutions, as previously described [77] using a spectrophotometer (Basic, Eppendorf, Hamburg, Germany) and the following equations:

Conclusions
In this study, we identified PSY1 homologous genes in RF, YF, and GF tomato species of Solanum section Lycopersicon. PSY1 homologs shared high similarity of structure, conserved motifs, and functionally important sites. PSY1 transcription levels were species-and plant organ-specific, and directly correlated with carotenoid content in ripe fruit. Analysis of the PSY1 promoter and 5 -UTR sequences revealed differences between GF and RF tomatoes, which could be attributed to domestication. Our results provide valuable data for further functional and evolutionary characterization of the carotenogenesis pathway in fleshy fruit in the Solanaceae family.