Functional Characterization and in Silico Analysis of Phytoene Synthase Family Genes Responsible for Carotenoid Biosynthesis in Watermelon ( Citrullus lanatus L.)

: Carotenoids are the main pigments in watermelon ( Citrullus lanatus L.) fruit and contribute to its aesthetic and nutritional value. Phytoene synthase (PSY) is reported to be the ﬁrst rate-limiting enzyme in carotenogenesis and controls the carotenoid ﬂux. This study aimed to identify PSY genes responsible for carotenoid biosynthesis in the red-ﬂeshed watermelon cultivar LSW-177. The PSY gene members ClPSY1, ClPSY2 and ClPSY3 were characterized and their catalytic activities were displayed in the heterologous complementation assay. The transcript levels of ClPSY genes at the di ﬀ erent developmental stages of LSW-177 fruit and the promoter sequence of ClPSY1 were also analyzed. Transcription factors involved in regulating the ClPSY1 expression were scanned with previous RNA-seq data of the di ﬀ erent stages during fruit ripening. Results showed that the PSY proteins from watermelon LSW-177 contained the conserved PSY domains and exhibited the ability to condense GGPP into phytoene in E. coli . ClPSY1 is the dominant carotenogenic gene during fruit ripening; and can be induced by light and hormones. Furthermore, Cla013914 and Cla007950 that, respectively encode the transcription factors WD40-like protein and bZIP, likely upregulate ClPSY1 during fruit ripening. In conclusion, ClPSY1 play a dominant role in carotenoid biosynthesis during watermelon fruit ripening and is regulated by complex light and hormone-responsive networks.


Introduction
The watermelon is an important fruit crop with cultivars that differ in flesh pigmentation due to varying composition and levels of carotenoids [1]. The major carotenoid in the red-fleshed cultivars is lycopene, whereas β-carotene is the predominant carotenoid in cultivars with yellow flesh [1]. Carotenoids are a large group of natural and lipid-soluble pigments present in photosynthetic organisms some non-photosynthetic organisms, including plants, algae, cyanobacteria and some bacteria [2]. More than 1100 natural carotenoids had been identified so far, of which most are C 40 compounds (http://carotenoiddb.jp) [3]. Carotenoids capture light during photosynthesis and also protect plants from photodamage [4,5]. They are also precursors of vitamin A [6], phytohormones [7,8]

Plant Material and Lycopene Measurement
The lycopene-rich red fleshed watermelon (Citrullus lanatus L.) cultivars LSW-177 and W1-1 were grown in the greenhouse of Northeast Agricultural University (44 • 04 N, 125 • 42 E), Haerbin, China. The plants were placed under field management using standard horticultural procedures [49]. Samples were scooped from the center of three intact watermelon fruits at five different ripening stages (10,18,26,34 and 42 DAP or days after pollination) and stored at −80 • C until further analysis. The lycopene content was estimated in 3 g flesh per sample as previously described [48]. Three technical replicates were tested per sample.

Isolation of PSY cDNA Sequences and Corresponding Genomic Sequences
Total RNA was extracted using the RNAprep pure plant kit (Tiangen Biotech, Beijing, China) according to the manufacturer's instructions and 1 µg RNA was reverse transcribed to first-strand cDNA using the RevertAid TM -MuLV Kit (MBI, Vilnius, Lithuania) and an oligo dT primer. The open reading frames (ORFs) of watermelon PSY (ClPSY) gene were amplified using primers (Table S1) synthesized according to the sequence available in the Cucurbit genomics database (CuGenDB) (http://cucurbitgenomics.org/). The amplified PCR products were cloned into the pMD19-T Easy vector (TaKaRa, Kyoto, Japan) and transformed into Escherichia coli strain DH5α. The genomic DNA of watermelon samples was extracted using the PlantGen DNA Kit (CWBiotech, Beijing, China) and the promoter and coding regions of ClPSY1 gene were amplified and sequenced.

Protein Sequence Analysis
The signal peptide and transmembrane regions of ClPSY proteins were predicted using CBS prediction servers and the putative conserved domains were analyzed with the help of NCBI conserved domains. The resulting amino acid sequences were aligned using Clustal X v.2.0 [50] and a phylogenetic tree was constructed using MEGA 7.0 by neighbor-joining method and 1000 bootstrap replications [51]. The genomic and coding sequences of ClPSYs from watermelon and other species were compared using TBtools JRE1.6 [52]. The conserved motifs for all PSY protein sequences were detected by multiple expectation maximization for motif elicitation (MEME v.5.1.1) online tools (http://meme-suite.org/tools/ meme).

Real-Time PCR Analysis
Primers specific for ClPSYs exons were designed using the Primer 5.0 software to avoid genomic amplification (Table S1). First, strand cDNA was synthesized using random primers and RT-PCR was performed using SYBR green master mix (TOYOBO, Osaka, Japan) in the iQ5 Gradient Real Time PCR system (Bio-Rad, Berkeley, CA, USA) following the manufacturers' protocols. The reaction conditions were as follows: initial denaturation at 95 • C for 5 min, followed by 40 cycles of 95 • C for 15 s and 60 • C for 30 s. Relative gene expression was calculated using comparative C T method [53], with the watermelon ClYLS8 gene as the internal control [54]. Three technical replicates were tested for each sample.

Plasmids and Functional Complementation
To test whether the watermelon PSY genes encode functional enzymes, heterologous complementation was performed for the carotenoid biosynthetic genes. The pACCRT-E plasmid carrying the bacterial crtE gene that encodes geranylgeranyldiphosphate synthase (GGPS) was transformed into E. coli [55]. Fragments of ClPSY genes coding sequences were amplified and cloned into the pET28a vector (Novagen, Germany) to generate pETPSY1, pETPSY2 and pETPSY3 constructs. The bacteria carrying pACCRT-E plasmid was co-transformed with the respective ClPSY constructs or empty pET28a vector (negative control); (E) coli harboring the pACCRT-EB vector with phytoene-producing crtE and crtB genes was used as the positive control. The suitably transformed E. coli cells were cultured at 37 • C in Luria-Bertani (LB) liquid medium containing 100 µg mL −1 ampicillin and 35 µg mL −1 chloramphenicol till OD 600 reached 0.6-0.8. After inducing protein expression with 0.05 mM isopropyl-β-d-thiogalactopyranoside (IPTG), the cells were cultured at 28 • C for 2 days in the dark without shaking. The cells were harvested by centrifuging at 5000× g for 15 min at 4 • C and washed twice with sterile distilled water. The pellets were resuspended in 5 mL hexane/acetone (1:1) solution containing 1% butylated hydroxytoluene (BHT) and sonicated twice on ice. Samples were centrifuged at 4000× g to remove the lysed cells, and the supernatant was dried under nitrogen. The dried extracts were resuspended in acetone for high-performance liquid chromatography (HPLC) as previously described [56].

In Silico Analysis of ClPSY1 Promoter Elements and Related Transcription Factors in Watermelon
The ClPSY1 promoter sequence was amplified from the genomic DNA of both cultivars using specific primers (Table S1). The nucleotide start site of the ClPSY1 gene was determined using transcription start site (TSS) program of the Softberry software and the cis-regulatory elements were predicted with plantCARE [57] and PLACE [58,59]. The RNA-seq data of watermelon at different ripening stages are available in the NCBI SRA database under the numbers SRX2037189 for cultivar LSW-177 [49] and SUB4001164 for cultivar W1-1 [60]. The expression levels of different transcription factors and ClPSY1 were analyzed at the different stages of fruit ripening.

Characterization of Watermelon PSY Gene and Protein Sequences
Three watermelon PSY genes were searched in the CuGenDB and BLAST annotation indicated that each mapped to different chromosomes (1, 7 and 2). To determine their role in carotenogenesis, we amplified the ClPSY1, ClPSY2 and ClPSY3 coding sequences from the LSW-177 cultivar. The full-length 1266 bp ClPSY1 sequence encoded a 421 amino acid protein with a predicted isoelectric point (pI) of 6.86. The putative protein encoded by ClPSY2 consisted of 355 amino acids and a pI of 8.46, while the deduced ClPSY3 protein had 388 amino acids and a pI of 8.87. Multiple alignment of the predicted proteins indicated 74.5% homology between ClPSY1 and ClPSY2 and 67.6% between ClPSY3 and ClPSY2. The variable sequences were concentrated in the N-terminal regions and predicted as transit peptides. Both ClPSY2 and ClPSY3 contained a shorter N-terminal sequence compared to ClPSY1 ( Figure 1). According to the NCBI conserved domains database, watermelon PSYs have six motifs in common ( Figure 1 and Figure S1), including the substrate binding pocket, Mg 2+ binding site, active site lid residues, catalytic residues and two aspartate-rich sites (DXXXD). The catalytic domains are mainly concentrated between 144-193 aa and 299-348 aa (Figure 1, yellow and pink background). A transmembrane region was also predicted in all three proteins ( Figure 1, and Figure S2).
Comparison of the genomic and cDNA sequences revealed that the genomic sequence of ClPSY3 (3180 bp) was longer than that of both ClPSY1 (2886 bp) and ClPSY2 (1902 bp). Furthermore, the coding sequence of ClPSY1 consisted of 6 exons split by 5 introns similar to that of AtPSY and SlPSYs, compared to 5 exons in both ClPSY2 and ClPSY3 ( Figure 2). Analysis of the conserved motifs with MEME identified the same regions as described above. Phylogenetic analysis of ClPSY proteins with that of PSY proteins from other plants showed that the central region was conserved across the species, whereas the N-terminal sequences were diverse on account of the variations in plastid transit peptides. Proteins from watermelon are more closely related to the ones from melon and grouped, respectively. ClPSY1, ClPSY2 and ClPSY3 were classified into two distinct clades, indicating their different functionality ( Figure 2). Comparison of the genomic and cDNA sequences revealed that the genomic sequence of ClPSY3 (3180 bp) was longer than that of both ClPSY1 (2886 bp) and ClPSY2 (1902 bp). Furthermore, the coding sequence of ClPSY1 consisted of 6 exons split by 5 introns similar to that of AtPSY and SlPSYs, compared to 5 exons in both ClPSY2 and ClPSY3 ( Figure 2). Analysis of the conserved motifs with MEME identified the same regions as described above. Phylogenetic analysis of ClPSY proteins with that of PSY proteins from other plants showed that the central region was conserved across the species, whereas the N-terminal sequences were diverse on account of the variations in plastid transit peptides. Proteins from watermelon are more closely related to the ones from melon and grouped, respectively. ClPSY1, ClPSY2 and ClPSY3 were classified into two distinct clades, indicating their different functionality ( Figure 2).

ClPSY1
E L

ClPSY1
R The tree was constructed using MEGA7 with a bootstrap of 1000 by the neighbor-joining method [51]. The motif distribution is shown in the middle, wherein rectangles of different colors represent conserved motifs. Color coding is the same as in Figure 1. Exon-intron distribution is shown in the right panel, wherein the black line and blue boxes represent introns and exons, respectively. The database accession numbers are:

Functional Complementation
A heterologous complementation assay was performed to evaluate whether watermelon PSY genes encode functional enzymes. Briefly, the coding sequence of ClPSY1, ClPSY2 and ClPSY3 genes were each introduced into GGPP-expressing E. coli that harbored the pACCRT-E plasmid. phytoeneproducing bacteria that carried the pACCRT-EB plasmid was used as the positive control. Introduction of the ClPSY1, ClPSY2 and ClPSY3 construct in the pACCRT-E cells resulted in a peak with retention time and spectral qualities similar to that of phytoene ( Figure 3). These results indicated all the three PSY genes from watermelon encoded functional proteins and were able to catalyze the conversion of GGPP to phytoene in bacteria.

Functional Complementation
A heterologous complementation assay was performed to evaluate whether watermelon PSY genes encode functional enzymes. Briefly, the coding sequence of ClPSY1, ClPSY2 and ClPSY3 genes were each introduced into GGPP-expressing E. coli that harbored the pACCRT-E plasmid. phytoene-producing bacteria that carried the pACCRT-EB plasmid was used as the positive control.
Introduction of the ClPSY1, ClPSY2 and ClPSY3 construct in the pACCRT-E cells resulted in a peak with retention time and spectral qualities similar to that of phytoene ( Figure 3). These results indicated all the three PSY genes from watermelon encoded functional proteins and were able to catalyze the conversion of GGPP to phytoene in bacteria.

Carotenoid Synthesis and Accumulation in Watermelon Fruit
To further determine the role of ClPSY proteins in watermelon carotenogenesis during the ripening period, we measured the lycopene content in the LSW-177 fruit at the different stages. The flesh color of LSW-177 gradually deepened from white to red by 26 DAP due to accumulation of lycopene, which peaked at 42 DAP and corresponded to the deepest red color ( Figure 4A). We next detected the transcript levels of ClPSY1, ClPSY2 and ClPSY3 in the fruit flesh at different stages of ripening. ClPSY3 did not alter significantly during ripening, indicating that it does not regulate watermelon fruit maturation. While both ClPSY1 and ClPSY2 had similar transcript profiles till 18 DAP, the ClPSY1 levels were 3.3-fold higher than that of ClPSY2 at 26 DAP and peaked at 42 DAP, which was consistent with the pattern of lycopene accumulation ( Figure 4B). Based on these results and previous reports detailing lack of correlation between the ClPSY2 levels and lycopene content in watermelon of different flesh color [61], we hypothesize that ClPSY1 has a dominant role in the carotenoid biosynthetic pathway during watermelon fruit ripening.

Carotenoid Synthesis and Accumulation in Watermelon Fruit
To further determine the role of ClPSY proteins in watermelon carotenogenesis during the ripening period, we measured the lycopene content in the LSW-177 fruit at the different stages. The flesh color of LSW-177 gradually deepened from white to red by 26 DAP due to accumulation of lycopene, which peaked at 42 DAP and corresponded to the deepest red color ( Figure 4A). We next detected the transcript levels of ClPSY1, ClPSY2 and ClPSY3 in the fruit flesh at different stages of ripening. ClPSY3 did not alter significantly during ripening, indicating that it does not regulate watermelon fruit maturation. While both ClPSY1 and ClPSY2 had similar transcript profiles till 18 DAP, the ClPSY1 levels were 3.3-fold higher than that of ClPSY2 at 26 DAP and peaked at 42 DAP, which was consistent with the pattern of lycopene accumulation ( Figure 4B). Based on these results and previous reports detailing lack of correlation between the ClPSY2 levels and lycopene content in watermelon of different flesh color [61], we hypothesize that ClPSY1 has a dominant role in the carotenoid biosynthetic pathway during watermelon fruit ripening.

Cis-Regulatory Structures of ClPSY1 Promoter
The regulatory motifs of most plant genes lie within 2000 bp upstream of the start codon [62,63]. To identify the cis-regulatory elements of ClPSY1 and compare the motifs in different watermelon cultivars, we amplified their ClPSY1 genomic sequences 2000 bp upstream of the ATG codon. The reference sequence of red-flesh watermelon 97103 was obtained from Cucurbit genomics database (http://cucurbitgenomics.org/). The predicted TSS was located at about 500 bp upstream from ATG. Putative cis-regulatory elements in the ClPSY1 gene promoter were searched using plantCARE and PLACE programs. Most of the motifs were light-related, including Box 4, G-Box, GT1-motif, chs-CMA1a, AE-box, TCT-motif and Sp1 ( Figure 5 and Table S2). Furthermore, hormone-related elements such as ABA-responsive motif ABRE and MeJA-responsive TGACG-motif were also identified in the ClPSY1 gene promoter in all three cultivars. The presence of the ARE motif in the ClPSY1 gene of the three cultivars indicated that is likely induced by anaerobic conditions. Interestingly, the promoter sequence of ClPSY1 from W1-1 and 97103 harbored one G-box motif compared to two in LSW-177. The ATCTA motif, which is recognized by the transcription factor RAP2.2, was only found in the ClPSY1 promoter in LSW-177 and not in W1-1 and 97103.

Cis-Regulatory Structures of ClPSY1 Promoter
The regulatory motifs of most plant genes lie within 2000 bp upstream of the start codon [62,63]. To identify the cis-regulatory elements of ClPSY1 and compare the motifs in different watermelon cultivars, we amplified their ClPSY1 genomic sequences 2000 bp upstream of the ATG codon. The reference sequence of red-flesh watermelon 97103 was obtained from Cucurbit genomics database (http://cucurbitgenomics.org/). The predicted TSS was located at about 500 bp upstream from ATG. Putative cis-regulatory elements in the ClPSY1 gene promoter were searched using plantCARE and PLACE programs. Most of the motifs were light-related, including Box 4, G-Box, GT1-motif, chs-CMA1a, AE-box, TCT-motif and Sp1 ( Figure 5 and Table S2). Furthermore, hormone-related elements such as ABA-responsive motif ABRE and MeJA-responsive TGACG-motif were also identified in the ClPSY1 gene promoter in all three cultivars. The presence of the ARE motif in the ClPSY1 gene of the three cultivars indicated that is likely induced by anaerobic conditions. Interestingly, the promoter sequence of ClPSY1 from W1-1 and 97103 harbored one G-box motif compared to two in LSW-177. The ATCTA motif, which is recognized by the transcription factor RAP2.2, was only found in the ClPSY1 promoter in LSW-177 and not in W1-1 and 97103.

Cis-Regulatory Structures of ClPSY1 Promoter
The regulatory motifs of most plant genes lie within 2000 bp upstream of the start codon [62,63]. To identify the cis-regulatory elements of ClPSY1 and compare the motifs in different watermelon cultivars, we amplified their ClPSY1 genomic sequences 2000 bp upstream of the ATG codon. The reference sequence of red-flesh watermelon 97103 was obtained from Cucurbit genomics database (http://cucurbitgenomics.org/). The predicted TSS was located at about 500 bp upstream from ATG. Putative cis-regulatory elements in the ClPSY1 gene promoter were searched using plantCARE and PLACE programs. Most of the motifs were light-related, including Box 4, G-Box, GT1-motif, chs-CMA1a, AE-box, TCT-motif and Sp1 ( Figure 5 and Table S2). Furthermore, hormone-related elements such as ABA-responsive motif ABRE and MeJA-responsive TGACG-motif were also identified in the ClPSY1 gene promoter in all three cultivars. The presence of the ARE motif in the ClPSY1 gene of the three cultivars indicated that is likely induced by anaerobic conditions. Interestingly, the promoter sequence of ClPSY1 from W1-1 and 97103 harbored one G-box motif compared to two in LSW-177. The ATCTA motif, which is recognized by the transcription factor RAP2.2, was only found in the ClPSY1 promoter in LSW-177 and not in W1-1 and 97103. Putative cis-regulatory motifs were identified with plantCARE [57] and PLACE [58,59]. Motifs associated with light and hormonal response are highlighted in different colors. Motif positions are relative and not drawn to scale. Red arrows indicate differences between cultivars.

WD40-Like and bZIP Regulate ClPSY1 Gene Expression during Watermelon Fruit Ripening
We previously annotated and classified genes that are differentially expressed during watermelon fruit ripening [49]. To further identify transcription factors (TF) that potentially bind to the ClPSY1 promoter and regulate its expression during ripening, we analyzed the RNA-seq data from the different stages of the LSW-177 (10, 18, 26, 34 and 42 DAP) and W1-1 (WA, WB, WC, WD and WE) cultivars. Both transcriptomic and subsequent qPCR analysis indicated that ClPSY1 transcript levels increased rapidly in the first three stages, suggesting that the upstream TFs likely exert their effects in the initial stages of fruit maturation. We then screened for the differentially expressed TFs in these stages based on p-value <0.05, q value <0.05 and fold change >1. Twenty-two TFs were upregulated in the LSW-177 cultivar, including AP2, bHLH, bZIP, MYB, WD40-like and WRKY ( Figure 6A and Table S3). In the W1-1 fruits, 50 TFs were upregulated ( Figure 6B and Table S4). The Cla013914 and Cla007950 genes, respectively encoding the WD40-like and bZIP TFs were upregulated in both cultivars, indicating that they likely increase ClPSY1 expression levels during fruit ripening.

WD40-Like and bZIP Regulate ClPSY1 Gene Expression during Watermelon Fruit Ripening
We previously annotated and classified genes that are differentially expressed during watermelon fruit ripening [49]. To further identify transcription factors (TF) that potentially bind to the ClPSY1 promoter and regulate its expression during ripening, we analyzed the RNA-seq data from the different stages of the LSW-177 (10, 18, 26, 34 and 42 DAP) and W1-1 (WA, WB, WC, WD and WE) cultivars. Both transcriptomic and subsequent qPCR analysis indicated that ClPSY1 transcript levels increased rapidly in the first three stages, suggesting that the upstream TFs likely exert their effects in the initial stages of fruit maturation. We then screened for the differentially expressed TFs in these stages based on p-value <0.05, q value <0.05 and fold change >1. Twenty-two TFs were upregulated in the LSW-177 cultivar, including AP2, bHLH, bZIP, MYB, WD40-like and WRKY ( Figure 6A and Table S3). In the W1-1 fruits, 50 TFs were upregulated ( Figure 6B and Table  S4). The Cla013914 and Cla007950 genes, respectively encoding the WD40-like and bZIP TFs were upregulated in both cultivars, indicating that they likely increase ClPSY1 expression levels during fruit ripening.

Discussion
The content and composition of carotenoids in the fleshy part of watermelon are highly important traits for breeding new varieties since they contribute to the appearance and nutritional value [64,65]. Many transcriptome and QTL genetic linkage experiments had been performed to

Discussion
The content and composition of carotenoids in the fleshy part of watermelon are highly important traits for breeding new varieties since they contribute to the appearance and nutritional value [64,65]. Many transcriptome and QTL genetic linkage experiments had been performed to reveal the genes responsible for carotenoid biosynthesis in watermelon fruit ripening [46,49,65,66]. Phytoene synthase (PSY) has been confirmed to catalyzes the first committed step of carotenogenesis and therefore controls the carotenoid flux through the pathway in many species [67,68]. The aim of this study was to analyze the functionality of ClPSY proteins in order to determine which one was responsible for the carotenoid biosynthesis in the watermelon fruit. In this study, we cloned and identified three PSY-encoding genes from the red-flesh LSW-177 cultivar and designated them as ClPSY1, ClPSY2 and ClPSY3.
The PSY isozymes showed extensive homology in their amino acid sequences with the exception of the N-terminal region. The high diversity in the potential transit peptide at the N-terminal of ClPSYs possibly affect their localization in photosynthetic versus non-photosynthetic tissues, which is also previously observed in Brassica napus and apple PSY proteins [24][25][26]. In silico analysis further showed the presence of the active aspartate-rich motifs DELVD and DVGED, squalene/PSY signature one and two in each ClPSY protein at the same positions as in AtPSY [15], indicating that these motifs are highly conserved across different species and may origin from the same ancestor [24,69,70]. In addition, the watermelon PSY proteins were phylogenetically closer to that of other species than among themselves. For instance, ClPSY1 is organized into six exons similar to that of its homolog in melon (CmPSY1), while the five exon-four intron structure of ClPSY2 and ClPSY3 is comparable to that of MdPSY3/4 and MdPSY1/2 in apples [24]. Furthermore, the structure of ClPSY2 is more divergent with CmPSY2 gene. Taken together, the nonconserved intron/exon structure suggest duplication events and functional diversity.
Complementation assay confirmed that all ClPSY proteins can condense GGPP into phytoene in bacteria. However, they exhibited distinctive expression profiles at the different stages of fruit ripening. The ClPSY1 transcripts showed a rapid surge in the initial stages of ripening in LSW-177 cultivars, which corresponded to maximum lycopene accumulation, and both showed a subsequent decline in the fully ripened fruit. Guo et al. [66] reported a similar pattern of ClPSY1 expression in the red-flesh 97103 watermelon cultivar. In contrast, its paralog CmPSY1 is predominantly expressed in the ripe melon fruit [28]. Wang et al. [61] analyzed ClPSY2 transcripts in six watermelon cultivars with distinct fruit pigmentation and found that the expression profile of ClPSY2 was not consistent with lycopene accumulation. The red-fleshed MSW28 cultivar had the highest lycopene content, while ClPSY2 displayed lower levels. Finally, ClPSY3 expression levels did not alter significantly across the different development stages. Taken together, the ClPSY1 gene plays the dominant role in carotenogenesis during watermelon fruit ripening.
Various light-and hormone-responsive elements were identified in the ClPSY1 gene promoter regions from different watermelon cultivars, indicating complex regulatory mechanisms. Chs-CMA1a, Box 4, GT1-motif, AE-box and Sp1 motifs were the common light-responsive elements across LSW-177, W1-1 and 97103 cultivars. In Arabidopsis, PIF1 represses AtPSY transcription by directly binding to a G-box element in its promoter [31]. MabZIP4 directly binds to the BanAAT promoter in banana, while MabZIP5 binds to the promoters of MaMT1, MaACY1, MaAGT1 and BanAAT via the G-box motif, implicating diverse functional roles of MabZIPs during fruit ripening [71]. Fruit growth and ripening are also controlled by multiple phytohormones. For instance, ABA regulates fruit ripening in the watermelon 97103, W1-1 and Pi186490 cultivar [60,72]. We identified ABA and MeJA-responsive elements in the ClPSY1 promoter of LSW-177, W1-1 and 97103 cultivars, suggesting that multiple factors involved in hormone synthesis and signaling regulate ClPSY1 expression during fruit ripening. The ABA-responsive element (ABRE) ACGTG is the major cis-element for ABA-responsive gene expression and TFs containing basic-domain leucine zipper (bZIP) bind to the ABRE sequence in Arabidopsis [73,74]. Furthermore, the MeJA-responsive motif TGACG is a binding site for bZIP that is involved in the MeJA signal transduction pathway [75]. We hypothesize that bZIP regulates ClPSY1 expression via these motifs during watermelon fruit ripening. In addition, an RAP2.2 motif that is recognized by the ethylene response factors (ERF) RAP2.2 were only detected in LSW-177, indicating the distinct regulatory mechanisms between the different cultivars.
Further screening of the transcriptome data of LSW-177 and W1-1 revealed the TFs Cla013914 and Cla007950 as potential upstream activators of CIPSY1. Cla013914 encodes an unknown protein containing the WD40/YVTN repeat-like-containing domain (IPR015943). The WD40 TF typically contains 4-16 tandem repeat WD motifs that form a scaffold to regulate the assembly and binding of multiple protein complexes involved in plant growth and development [76,77]. A total of 54 putative WD40 genes were differentially expressed between the red-fleshed and white-fleshed peach fruits at five developmental stages [78]. In blueberries, high MYB-bHLH-WD40 expression correlated positively with anthocyanin accumulation and color development [79]. Similar expression patterns have also been detected in the Lycium fruit [80]. Cla007950 encodes a protein of the bZIP family that shows high homology with AtbZIP42. As one of the largest TF families in plants, the bZIP family controls multiple biologic processes, including development, plant growth, fruit ripening and stress responses. Based on these findings, it is highly likely that Cla013914 and Cla007950 activate ClPSY1 in watermelon fruits and promote carotenoid biosynthesis.

Conclusions
This work presents the structural and functional characterization of phytoene synthase proteins in red-fleshed watermelon LSW-177. Taken together, higher expression levels of ClPSY1 and the ability to generate phytoene in a heterologous system, suggest that ClPSY1 play a dominant role in the carotenoid biosynthesis of watermelon during fruit ripening. This provides a target for watermelon fruit-color breeding. We also analyzed the promoter sequences of ClPSY1 and selected Cla013914 and Cla007950 as the candidate genes that may positively regulate ClPSY1 expression in watermelon during fruit ripening. This study also provide important cues to unveil the underlying mechanisms of carotenoid biosynthesis in watermelon fruit.
Supplementary Materials: The following are available online at http://www.mdpi.com/2073-4395/10/8/1077/s1, Table S1: PCR primers used in the research; Table S2: Putative cis-regulatory elements related to specific function present in ClPSY1 promoter from red-fleshed watermelon cultivars LSW-177, W1-1 and 97103; Table S3: Excel file containing the upregulated transcription factors in LSW-177 at five developmental stages; Table S4: Excel file containing the upregulated transcription factors in W1-1 at five developmental stages; Figure S1: The motifs of watermelon PSY1 protein analyzed using NCBI conserved domains; Figure S2: The transmembrane sequence of watermelon ClPSY1 protein as predicted with CBS prediction servers.