Level Expression of a Novel Chromoplast Phosphate Transporter Clpht4;2 Is Required for ﬂesh Color Development in Watermelon

Research

High-level expression of a novel chromoplast phosphate transporter ClPHT4;2 is required for ﬂesh color development in watermelon

Jie Zhang, Shaogui Guo, Yi Ren, Haiying Zhang, Guoyi Gong, Ming Zhou, Guizhang Wang, Mei Zong, Hongju He, Fan Liu and Yong Xu National Engineering Research Center for Vegetables, Beijing Academy of Agriculture and Forestry Sciences, Key Laboratory of Biology and Genetic Improvement of Horticultural Crops (North China), Beijing Key Laboratory of Vegetable Germplasm Improvement, Beijing 100097, China

Summary Author for correspondence: Chromoplast development plays a crucial role in controlling carotenoid content in water- Yong Xu melon flesh. Modern cultivated watermelons with colorful flesh are believed to originate from Tel: +86 10 51503199 pale-colored and no-sweet progenitors. But the molecular basis of flesh color formation and Email: [email protected] regulation is poorly understood. Received: 23 May 2016 More chromoplasts and released carotenoid globules were observed in the red-fleshed fruit Accepted: 11 September 2016 of the 97103 cultivar than in the pale-colored fruits of the PI296341-FR line. Transcriptome profiles of these two materials identified Cla017962, predicted as ClPHT4;2, was dramatically New Phytologist (2016) up-regulated during flesh color formation. High ClPHT4;2 expression levels were closely cor- doi: 10.1111/nph.14257 related with increased flesh carotenoid contents among 198 representative watermelon accessions. Down-regulation of ClPHT4;2 expression in transgenic watermelons reduced the Key words: chromoplast, Citrullus lanatus, fruit carotenoid accumulation. flesh color, phosphate transporter, transcrip- ClPHT4;2 as a function of chromoplast-localized phosophate transporter was tested by tional regulation. heterologous expression into a yeast phosphate-uptake-defective mutant, western blotting, subcellular localization, and immunogold electron microscopy analysis. Two transcription factors, ClbZIP1 and ClbZIP2, were identified, which responded to ABA and sugar signaling to regulate ClPHT4;2 transcription only in cultivated watermelon species. Our findings suggest that elevated ClPHT4;2 gene expression is necessary for carotenoid accumulation, and may help to characterize the co-development of flesh color and sweetness during watermelon development and domestication.

Introduction semiwild watermelon accessions of C. lanatus ssp. mucosospermus produce a wide range of flesh colors and sugar contents (Jeffrey, Fruit flesh color is an important nutritional and sensory quality 2001; Guo et al., 2013). There are different opinions regarding that has received considerable attention from breeders and con- the origin of the genus Citrullus and the nomenclature used to sumers. Carotenogenesis in chromoplast contributes to the fruit describe them (Chomicki & Renner, 2015; Paris, 2015). How- flesh color formation (Li & Yuan, 2013). Watermelon (Citrullus ever, it is generally accepted that, during the domestication pro- lanatus), as one of the five most consumed fresh fruits grown cess, the pigmentation of watermelon flesh increased. Differences worldwide, possesses colorful flesh and provides abundant in fruit flesh color development between cultivated and wild carotenoids to worldwide consumers (Guo et al., 2013). Different watermelons make these species ideal for comparative genetic composition and concentration of carotenoids contribute to the analyses to reveal the underlying mechanisms of flesh color for- red, orange, canary yellow, salon yellow and white flesh color in mation in fruit development. cultivated watermelons (Henderson et al., 1998). As one of the Chromoplasts use unique mechanisms to synthesize and few species that accumulate a large amount of lycopene in fruits, deposit large amounts of carotenoids (Li & Yuan, 2013; Yuan red-fleshed watermelon contains more lycopene per unit FW et al., 2015). Previous studies have been devoted to clarifying the than fresh tomatoes (Solanum lycopersicum) (Collins et al., 2006). carotenoid biosynthetic pathway in watermelon (Bang et al., Compared with the colorful flesh of modern cultivated watermel- 2007; Kang et al., 2010). Amino acid substitutions of lycopene ons (C. lanatus ssp. vulgaris), fruits of the wild accessions b-cyclase were revealed to be responsible for the accumulation of C. lanatus ssp. lanatus, have pale-colored flesh that is not sweet. lycopene or b-carotene in chromoplasts, resulting in the red or They occur naturally in southern Africa. Additionally, the canary yellow color in flesh (Bang et al., 2007). In contrast to the

Ó 2016 The Authors New Phytologist (2016) 1 New Phytologist Ó 2016 New Phytologist Trust www.newphytologist.com New 2 Research Phytologist

significant advances made in understanding of carotenoid and stability (Wiese et al., 2004). Glucose can regulate ABA metabolism (Nisar et al., 2015), very little is known about the biosynthesis and activate some important transcriptional factors, biogenesis and regulation of chromoplasts. As nonphotosynthetic such as ABI3 and ABI5, to regulate plant growth (Cheng et al., plastids in ripening fruit, chromoplasts are bound by double 2002; Rolland et al., 2006). However, there is little information membranes. The inner membrane functions as the primary per- about the key regulators and the related cis-elements controlling meability barrier where several proteins help transport substrates watermelon flesh color development. between the plastids and the cytosol (Egea et al., 2010). In this study, we observed that high expression of the newly Researchers have recently focused on the transport processes characterized chromoplast Pi transporter gene ClPHT4;2 is nec- across both the inner and outer envelope membranes of plastids essary in all of the colored-flesh watermelon accessions. Two and their role in connecting the metabolism of plastids with that bZIP transcription factors, ClbZIP1 and ClbZIP2, associated of the other parts of a plant cell (Fischer, 2011; Flugge€ et al., with ABA and sugar signaling pathways regulate ClPHT4;2 tran- 2011). Proteome-level analyses of chromoplast revealed an abun- scription in the analyzed red-fleshed watermelon cultivar. dance of proteins related to ATP and hexose phosphate production and transport, suggesting that anabolic processes are Materials and Methods important for chromoplast development (Quick & Neuhaus, 1996; Wang et al., 2013). Activated anabolic processes in chro- Plant materials moplasts would lead to the excessive accumulation of inorganic phosphate (Pi) if not balanced by a Pi export activity. Many plant The red-fleshed Citrullus lanatus (Thunb.) Matsum. & Nakai ssp. Pi transporters have been identified and classified into distinct vulgaris cultivar 97103 is a typical early-maturing East Asian culti- families, including PHT1-5 and plastidic Pi translocator (pPT) var with sweet, red and crispy flesh. The pale-fleshed C. lanatus (Rausch & Bucher, 2002; Liu et al., 2016). The pPT, PHT2, ssp. lanatus watermelon line PI296341-FR (also called C. lanatus and PHT4 families are involved in plastid-localized Pi transport var. citroides) is a wild watermelon with a round shape, medium systems (Guo et al., 2008). In Arabidopsis thaliana, PHT4;2 is a size, thick and hard rind, and light-green striped fruit with non- sink-specific plastidic phosphate transporter (Irigoyen et al., sweet and pale-colored (white) flesh. Roots, leaves, stems, female 2011). However, which PHT family member is involved in chro- flowers, flowers, and fruits at four critical developmental stages moplast development and carotenoid accumulation is unclear. (immature white (10 d after pollination, DAP), white-pink (18 A transcriptome analysis of watermelon fruits at different devel- DAP), red (26 DAP) and overripe (34 DAP)) were examined in opmental stages may help researchers identify genes influencing 97103. The development of PI296341-FR fruit was divided into fruit color formation. We have published our findings regarding six stages (10, 18, 26, 34, 42 and 50 DAP) for subsequent analysis. global transcriptome profiles of fruit development in the red- Fruits from 96 recombinant inbred lines (RILs) derived from a fleshed cultivar 97103 (C. lanatus ssp. vulgaris) and in the pale- cross between 97103 and PI296341-FR and of 102 representative colored (white) line PI296341-FR (C. lanatus ssp. lanatus) (Guo watermelon accessions, including 61 C. lanatus ssp. vulgaris lines, et al., 2011, 2015). We reported that the carotenoid biosynthetic 12 C. lanatus ssp. lanatus accessions and 29 C. lanatus ssp. pathway is activated in red-fleshed fruits. Phytoene synthase (PSY) mucosospermus (also called C. lanatus var. eguis) accessions, were and phytoene desaturase (PDS) in this pathway are up-regulated analyzed at 38 DAP. The watermelon juice sugar content (soluble during red-fleshed watermelon fruit development. Additionally, solid content, °Brix) was measured using a pocket refractometer several of the differentially expressed genes in 97103 are associated PAL-1 (Atago Co. Ltd, Tokyo, Japan) from a sample of juice that with transporting ions and small molecules (Guo et al., 2015). was collected from the center of each watermelon. These differentially expressed genes are probably regulated by special transcription factors (TFs), and their upstream regulatory Carotenoid extraction and analysis regions are important for mediating transcriptional activities. Numerous studies have been conducted on phytohormone- Carotenoids from watermelon flesh (0.5 g FW) were extracted related TFs and cis-elements affecting fruit ripening and and analyzed using a Nexera high-performance liquid chro- carotenogenesis (Welsch et al., 2003, 2007; Lee et al., 2012; Su matography (HPLC) system (Shimadzu, Japan) following the et al., 2015). The ABRE motif, as ABA and osmotic stress method described by Bang et al. (2010). Relative concentrations response element with typical consensus sequence of T/G/ of individual carotenoids were determined by comparing the CACGTGG/TC, was identified to regulate high expression of individual peak areas with calibration curves constructed using PSY and PDS in A. thaliana (Welsch et al., 2003, 2007). By bind- commercial lycopene, a-carotene, b-carotene and lutein stan- ing to GCC-box-type cis-elements that are present in target pro- dards (Sigma-Aldrich). All samples were analyzed in triplicate. moters, ethylene response factors (ERFs) can regulate tomato fruit ripening and carotenoid biosynthesis (Lee et al., 2012; Su RNA extraction, quantitative reverse transcription et al., 2015). There is accumulating evidence that in addition to polymerase chain reaction (RT-PCR) and rapid plant hormones, soluble sugars act as signal molecules influencing amplification of cDNA ends the development of sink organs (Smeekens et al., 2010; Ruan, 2014). Sugars are reported to regulate gene expression (Koch, Total RNA was extracted using the Quick RNA isolation kit 1996), affect mRNA stability, and influence protein translation (Huayueyang Biotechnologies Co. Ltd, Beijing, China), and

New Phytologist (2016) Ó 2016 The Authors www.newphytologist.com New Phytologist Ó 2016 New Phytologist Trust New Phytologist Research 3 cDNA was synthesized from 1 lg of total RNA using Chromoplast isolation and protein extraction SuperScript III transcriptase (Invitrogen). Quantitative RT- PCR was run on LightCycle 480 (Roche) according to the Crude chromoplasts from watermelon flesh were isolated and manufacturer’s instructions. All analyses were repeated three purified with a discontinuous sucrose gradient according to the times. The relative expression levels of ClPHT4;2 (gene ID: methods of Wang et al. (2013). The total intact chromoplast pro- Cla017962), ClbZIP1 (gene ID: Cla017696), ClbZIP2 (gene teins were extracted from fruit that was harvested at 10, 18, 26 ID: Cla014572), ClPSY (gene ID: Cla009122), ClPDS (gene and 34 DAP and ground in liquid nitrogen. Total proteins were ID: Cla010898), and ClCHYb (gene ID: Cla011420) were separated by sodium dodecyl sulfate polyacrylamide gel elec- normalized against that of the watermelon ACTIN gene trophoresis (SDS-PAGE). (gene ID: Cla007792) transcript and averaged using three biological replicates. The 50 and 30 ends of the transcripts Anti-ClPHT4;2 antibody production and western blot were determined using SMARTer RACE 50/30 kit (Clontech analysis Laboratories Inc., Beijing, China). The primer sequences used are listed in Supporting Information Table S1. A ClPHT4;2-specific antibody was generated in mouse by Abmart using 12 special N-terminal peptides. Using recombinant plasmid pET28a (+)-ClPHT4;2/BL21(DE3) expression systems, Vector construction and Agrobacterium-mediated transfor- the ClPHT4;2 protein was produced. Western blotting was per- mation of watermelon formed to evaluate the specificity and titer of the anti-ClPHT4;2 ClPHT4;2 knockdown in red-fleshed 97103 was completed antibody using bacterial-expressed recombinant ClPHT4;2. Four using the pYBA1311 vector for the RNA interference (RNAi) antisera were selected, and one (against the peptide analysis (Yan et al., 2012). To generate a hairpin RNAi con- PNYPSRFSTKKP) was tested using watermelon chromoplast struct, a highly specific 180 bp fragment of ClPHT4;2 CDS proteins separated in gradient SDS polyacrylamide gels (T = 4– was amplified by PCR primers with different adaptors. This 20%) (Bio-Rad). fragment was inserted into KpnI/XhoI sites and ClaI/BamHI sites of pYBA1311 and used to transform Agrobacterium Subcellular localization analysis tumefaciens strain C58/ATCC 33970. The Agrobacterium- meditated transformation protocol was modified from that To determine the subcellular localization of ClPHT4;2, described by Lin et al. (2012). A PCR amplication with speci- ClPHT4;2 was inserted into the pYBA1138-mCherry vector with fic primers (upstream primer of the 35S promoter and down- mCherry at the C terminus of ClPHT4;2, driven by the constitu- stream primer in the ClPHT4;2 gene) and AS013 PAT/bar Kit tive 35S promoter. Plasmids were purified and transformed into (Envirologix Inc., Portland, ME, USA) were used to confirm watermelon fruit protoplasts according to a slightly modified pre- the transgene insertion in transformed watermelons. The viously reported method (Yoo et al., 2007). The mCherry fluo- primer sequences used are listed in Table S1. rescence and chromoplast autofluorescence were analyzed using LSM700 microscope at wavelengths of 587 and 500 nm for exci- tation and 610 and 509 nm for emission, respectively. Functional complementation of the MB192 yeast mutant by ClPHT4;2 protein Transmission electron and immunoelectron microscopy The open reading frames of ClPHT4;2, AtPHT4;2, truncated ClPHT4;2M1-64, and truncated AtPHT4;2M1-44 were cloned Watermelon fruit flesh was fixed in 2% paraformaldehyde into the yeast expression vector pDR-GW-GFP (the green fluo- and 1% glutaraldehyde in 50 mmol l 1 Pipes buffer (pH 6.9) rescent protein (GFP) fusion proteins were expressed under the for 4 h at room temperature, followed by three rinses with PMA1 promoter; Loque et al., 2007) via the Gateway LR reac- the Pipes buffer. Specimens were then postfixed in 1% osmic tion. The constructs were transformed into the yeast Pi-uptake- acid (diluted in distilled water) for 1.5 h at room temperature. defective mutant MB192 (Bun-Ya et al., 1991). The exact subcel- After three washes with distilled water for 30 min each, the lular locations of the PHT proteins in these positive clones were fruit flesh was dehydrated in an ethanol series and embedded analyzed using the LSM700 laser-scanning confocal microscope in LR White acrylic resin (Sigma). The polymerization of LR (Zeiss). Positive clones were harvested, washed in Pi-free medium White was affected by heat-curing the resin at 46°C for 16 h. and then subjected to YNB medium containing Pi concentrations Ultrathin sections were collected on Formvar-coated gold of 20, 40 or 60 lM. In order to determine the kinetic properties grids. Immunofluorescence labeling used the method described of the ClPHT4;2 transporter, Pi uptake experiments using 32Pi by Wen & Li (2009) with modifications. The primary anti- were conducted using the transformed yeast. Cell samples of c. body was purified anti-ClPHT4;2 and the negative control 1 mg fresh yeast were used following the method described by consisted of mouse nonimmune serum. To determine the rel- Pavon et al. (2008) with modifications. The data were analyzed ative distribution of ClPHT4;2 in fruit flesh cells, the gold using the software SIGMAPLOT (v.10.0) to determine the Km value particles in the chromoplast, mitochondrion, endoplasmic for truncated ClPHT4;2 and AtPHT4;2 proteins regarding Pi reticulum, or other parts were counted. Five individual fruit uptake. flesh cells were analyzed, and the samples were observed and

Ó 2016 The Authors New Phytologist (2016) New Phytologist Ó 2016 New Phytologist Trust www.newphytologist.com New 4 Research Phytologist

photographed under a JEM-100S electron microscope (voltage effects of ethylene on flesh color development, whole fruits were 80 V). submerged for 5 min in a solution of 2 mM ethephon (ethylene- releasing reagent) or 1.5 mM AVG (an inhibitor of ethylene biosynthesis) in 0.1% Tween 20. Controls were submerged for Yeast one-hybrid assay the same time in 0.1% Tween 20. Submerged treatments were Yeast one-hybrid (Y1H) assay was completed using the Match- repeated twice over 4 d. The carotenoid contents were evaluated maker Gold Yeast One-Hybrid System (Clontech). According to by HPLC, and RNA was extracted from this treated flesh 4 d the single nucleotide polymorphism (SNP) in the ClPHT4;2 pro- later. Each treatment was repeated with at least 10 fruits. moters in 97103 and PI296341-FR, an ABRE motif and a 340 bp region were chosen to clone into the pAbAi vector using Results the primers listed in Table S1. Plasmid, SMART-amplified ds cDNA and linearized pGADT7-Rec were transformed into the Ultrastructural characterization of plastid development in Y1H Gold strain. The DNA–protein interaction was assessed red- and pale-fleshed watermelon accessions based on the growth of the cotransformants on SD/-Leu medium containing aureobasidin A (AbA), according to the manufac- The chromoplasts in watermelon fruits are derived from noncol- turer’s protocol. Positive clones were confirmed and sequenced. ored plastids (Egea et al., 2010). We comparatively examined the plastid development during fruit development in the red-fleshed cultivar 97103 and the pale-fleshed line PI296341-FR. Electrophoretic mobility shift assay (EMSA) Carotenoid biosynthesis results in the production of clearly visi- The total ClbZIP1 and ClbZIP2 proteins were fused in frame ble, intensely dark and symmetrical globules in chromoplasts with his tag and expressed in Escherichia coli. The recombinant (Bangalore et al., 2008). Very few visible carotenoid crystals were protein was purified by HIS-agarose affinity chromatography. detected in cells at the immature white stage (10 d after pollina- EMSA was performed using the Light Shift Chemiluminescent tion, DAP) in 97103 according to the transmission electron EMSA kit (Thermo Fisher Scientific, Shanghai, China) according microscopy analysis (Fig. 1a). Compared with the large number to the manufacturer’s instructions. The biotin-labeled DNA frag- of developing proplastids in 97103 flesh cells, there were few pro- ments that are listed in Table S1 were synthesized and used as plastids in PI296341-FR at the same developmental stage probes, while the same fragments lacking a biotin label were used (Fig. 1e). The cells of immature watermelon in the white-pink as the competitors in the EMSA. (18 DAP) stage showed the presence of carotenoid crystal globular structures within the chromoplasts in 97103, and some membrane envelopes of developed chromoplasts ruptured to release Transient assays in Nicotiana benthamiana leaves the pigment globules (Fig. 1b). In the red (26 DAP) stage of To generate the reporter construct pClPHT4;2:LUC, a 1.5 kb 97103, many chromoplasts became less organized asymmetrical promoter sequence of ClPHT4;2 from 97103 or PI296341-FR structures, and carotenoid globules released from the broken was fused to the luciferase reporter gene using the BamHI and envelope were clearly seen (Fig. 1c). In the 34 DAP stage of SalI sites of the pCAMBIA1381Z vector. Full-length ClbZIP1 97103, the structure of the chromoplasts disappeared, and a large and ClbZIP2 coding sequences were amplified and cloned into number of carotenoid globules were observed to scatter in the the BamHI/ClaI and BamHI/SalI sites of the pTCK303/1460 cytoplast (Fig. 1d), suggesting the fragile of the mature chromo- vector to generate effector plasmids. The Agrobacterium-mediated plasts. By contrast, in the pale-colored PI296341-FR, only a few infiltration of N. benthamiana leaves was performed as previously noncolored plastids turned into chromoplasts that accumulated described (Chen et al., 2011). Images were captured with carotenoid globules in flesh cells during ripening (Fig. 1f,g). NightSHADE LB985 (with INDIGO software). Ten independent Additionally, even in the very late stages of fruit development, leaves were analyzed. The experiments were repeated at least three the crystals and remaining stroma were surrounded by an intact times. plastid envelope in PI296341-FR cells (Fig. 1h). We also detected two or three intact chromoplasts similar to those in PI296341- FR cells in the ripening fruit flesh cells of the white-fleshed acces- ABA, fluridone, ethephon, aminoethoxyvinylglycine (AVG), sion SANBAI (C. lanatus ssp. mucosospermus). These results indi- sucrose and glucose treatments of developing fruits cate that a mass of carotenoid globules in chromoplasts produce Developing fruits from 97103 and PI296341-FR were treated at colorful flesh, while only one or two chromoplasts developed in 10 DAP. Aqueous solution (500 ll) of 100 lM ABA aqueous pale-colored flesh cell. solution, 50 lM fluridone (ABA biosynthesis inhibitor) aqueous solution, 5% glucose, and 5% sucrose or distilled water (control) Differential expression of candidate plastid transporter was injected into the fruit with a 1 ml sterile hypodermic syringe. genes Distilled water was used as the control solution. For injection, the syringe needle was pushed into the fruit carpopodium, and In previous transcriptome profiling experiments (Guo et al., the solution was slowly injected into the fruits. At 4 d after injec- 2011, 2015), several of the differentially expressed genes in tion, these treated fruits were plucked for analysis. To study the 97103 were associated with the transporting ions and small

New Phytologist (2016) Ó 2016 The Authors www.newphytologist.com New Phytologist Ó 2016 New Phytologist Trust New Phytologist Research 5

(a) (b) (c) (d)

(e) (f) (g) (h)

Fig. 1 Ultrastructures of watermelon chromoplasts in 97103 and PI296341-FR fruits. (a–d) Longitudinal section of 97103 fruits at: (a) 10 d after pollination (DAP), (b) 18 DAP, (c) 26 DAP, and (d) 34 DAP, Bars, 4 cm. Electron microscopy images of the flesh cells in the above fruits. Plastids are outlined by orange circles, Bars, 2 lm. (d) Degraded chromoplast membranes and the scattered carotenoid globules in the cytoplast. (e–h) Longitudinal section of PI296341- FR fruits at: (e) 10 DAP, (f) 18 DAP, (g) 34 DAP, and (h) 42 DAP. Electron microscopy images of the flesh cells in the same fruits. Bars, 2 lm. Chromoplasts are outlined with orange circles. molecules. Some of these transporters may be important during 2.713, normalized by reads per million sequence) at 10 DAP, but chromoplast development. As nonphotosynthetic plastids, devel- which increased dramatically, peaking at 26 DAP (485.941), and oping chromoplasts need a certain amount of membrane trans- then decreased slightly at 34 DAP (373.845) in red-fleshed porter proteins to exchange metabolites with the surrounding 97103. By contrast, the Cla017962 expression in the flesh of cytosol (Fischer, 2011). We identified 75 candidate plastid PI296341-FR was stable during fruit development and much envelope membrane transporters in the watermelon genome lower than that in 97103 (Table S2). These results indicate that (Table S2). According to our comparative transcriptome profiles Cla017962 might be important during watermelon fruit flesh (Guo et al., 2015), 10 differentially expressed genes were identi- development. fied among these candidates between 97103 and PI296341-FR The 97103 Cla017962 gene was observed to contain 3743 during fruit development (Table S2). Further analysis indicated base pairs with eight exons and seven introns (Fig. S1a). The cor- that three candidate transporter genes (Cla012668, Cla017029 responding full-length complementary DNA (cDNA) was iso- and Cla017962) were differentially expressed in the flesh of lated through 50 and 30 RACE, which contained a 31 bp 50- 97103, consistent with a series of physiological and biochemical untranslated region and a 301 bp 30-untranslated region changes in flesh color formation in 97103. Cla012668 and (GenBank no. KU963293) (Fig. S1b). According to TargetP Cla017029 were down-regulated in 97103 fruit. Cla012668 was analysis (http://www.cbs.dtu.dk/services/TargetP/), Cla017962 annotated as a plastid glucose transporter that is homologous to encoded a 521-amino-acid protein with a 64-amino-acid putative At5g16150. Cla017029, homolog to At3g01280, was annotated plastid transit peptide (Fig. S1c). The TMHMM online tool as a porin/voltage-dependent anion-selective channel protein. (http://www.cbs.dtu.dk/services/TMHMM-2.0/) predicted that Cla017962, which is a homolog to A. thaliana Pi transporter the protein consists of 11 transmembrane helices (Fig. S1d). A At2g38060 (AtPHT4;2), was the only one that was up-regulated sequence comparison revealed amino acid substitutions and dele- during 97103 flesh ripening. Thus, we focused on Cla017962, tions in the 97103 and PI296341-FR Cla017962 proteins which was expressed at very low levels (transcript abundance: (Fig. S1e). Cla017962 was the most homologous to the

Ó 2016 The Authors New Phytologist (2016) New Phytologist Ó 2016 New Phytologist Trust www.newphytologist.com New 6 Research Phytologist

characterized phosphate transporters from Arabidopsis, rice and (a) tomato (Fig. S2), showing 67.3% amino acid sequence identity to AtPHT4;2 in Arabidopsis. Thus, Cla017962 was predicted to be ClPHT4;2 in watermelon.

ClPHT4;2 transcript abundance is positively correlated with carotenoid contents in different watermelon accessions To gain insights into the temporal and spatial transcription patterns and putative functions of ClPHT4;2, real-time PCR was performed to analyze the transcription abundance in various tis- sues and organs, including roots, stems, leaves, flowers, and fruits. The results showed that ClPHT4;2 transcript abundance (b) increased 10 times from 10 to 26 DAP during 97103 fruit development, while its expression level in PI296341-FR remained consistent and low throughout fruit development (Fig. 2a). In other plant parts, ClPHT4;2 was mainly expressed in the roots and maintained a low level in stems, leaves and flowers (Fig. 2a). To clarify the relationship between the ClPHT4;2 expression and fruit flesh color development, we measured the ClPHT4;2 (c) transcript abundances, the content of four carotenoids, and the sugar content (soluble solids content, °Brix) of 96 RILs. Pearson’s correlation analysis showed that the transcript abundance of ClPHT4;2 in the RIL population was significantly positively correlated with the total carotenoid content (r = 0.5097, t = 5.7744, P < 0.0001) (Table S3). We also analyzed 102 representative watermelon accessions to characterize the relationship between ClPHT4;2 transcriptions and flesh carotenoid contents. The 102 representative natural accessions included three subspecies and major varieties in watermelon: 61 cultivars in C. lanatus ssp. vulgaris, 12 wild accessions (d) in C. lanatus ssp. lanatus and 29 accessions in C. lanatus ssp. mucosospermus (Table S4). In these accessions, ClPHT4;2 expression levels were also correlated with flesh carotenoid content (r = 0.8317, t = 14.9787, P < 0.0001). Darker watermelon flesh colors were associated with higher ClPHT4;2 transcript abundance.

Knockdown of ClPHT4;2 in the red-fleshed cultivar decreased the carotenoids accumulation in fruit flesh To clarify the effects of ClPHT4;2 during watermelon fruit flesh development, we generated a ClPHT4;2-RNAi construct to Fig. 2 ClPHT4;2 expression patterns and the fruit phenotypes of the transform the red-fleshed cultivar GS33 (Table S4). Expression ClPHT4;2-RNAi transgenic lines. (a) ClPHT4;2 (Cla017962) expression analysis of the resulting transgenic population helped to identify patterns in different parts of the red-fleshed cultivar 97103 and pale- 21 independent T plants, which were propagated for further fleshed line PI296341-FR. The ClPHT4;2 expression levels were measured 0 by real-time PCR in roots, stems, leaves, female and male flowers, and analysis. Seven transgenic lines were tested with reduced fruits at different stages (10, 18, 26, and 34 d after pollination, DAP). ClPHT4;2 mRNA levels, and three of them (clpht4R2, clpht4R3, Watermelon ACTIN was used as an internal control. Error bars indicate clpht4R11) with different ClPHT4;2 gene expression levels were SD (n = 10). A significant difference (**, P < 0.01) was observed selected (Fig. 2b,d). Line clpht4R9, which did not show any between the PI296341-FR and 97103 plants. (b) Fruit longitudinal section reduction of ClPHT4;2 transcripts, was used as a negative control of ClPHT4;2 knockdown lines and the control at 34 DAP. Bars, 1 cm. (c) Carotenoid and sugar content (°Brix) in the transgentic fruits presented in (Fig. 2b). In these transgenic lines, the mutated phenotype was (b). (d) Differential expression of ClPHT4;2 and selected carotenoid only observed in fruit flesh. Lines clpht4R2 and clpht4R3 exhib- biosynthesis genes in the ClPHT4;2 knockdown lines presented in (b). ited severe reductions of ClPHT4;2 expression, and the fruit ClPHT4;2, ClPSY, ClPDS and ClCHYb expression levels were measured by carotenoid contents decreased significantly compared with the real-time PCR. Watermelon ACTIN was used as an internal control. Error = control (Fig. 2c), while, in line clpht4R11, in which the bars indicate SD (n 3).

New Phytologist (2016) Ó 2016 The Authors www.newphytologist.com New Phytologist Ó 2016 New Phytologist Trust New Phytologist Research 7

ClPHT4;2 expression was decreased to a lesser extent, the fruit putative transit peptide in the N terminus) or empty vector color was slightly diluted. Flesh carotenoid contents in these (pDR-gw-eGFP) and grown in yeast nitrogen base medium con- transgenic lines were consistent with the ClPHT4;2 transcription taining different concentrations of Pi for the production of GFP abundance. Compared with the obvious deceases in carotenoid fusion proteins. We observed that ClPHT4;2-GFP was localized content, the changes in sugar content (°Brix) were less obvious to the yeast vacuole, whereas the other four fusion proteins had a (Fig. 2c). The transcript abundance of ClPSY, ClPDS and ubiquitous distribution (Fig. 3a). While the mutant cells carrying ClCHYb (beta-carotene hydroxylase), which are involved in the empty expression vector or ClPHT4;2 grew well only at Pi carotenoid biosynthesis, were also down-regulated in these trans- concentrations > 60 lM, the transformants expressing genic lines (Fig. 2d). Decreases of flesh carotenoid contents in the ClPHT4;2M1-64, AtPHT4;2 or AtPHT4;2M1-44 grew well at Pi ClPHT4;2 RNAi-transgenic lines indicated that the high concentrations > 40 lM. No transformants could grow in the ClPHT4;2 expression levels are necessary for flesh color develop- medium containing 20 lM Pi (Fig. 3b). The growth rate was ment in red-fleshed cultivar. positively correlated with increasing Pi concentrations, clearly demonstrating that at micromolar concentrations, Pi limits growth. A 32Pi-labeled substrate was then used to determine the ClPHT4;2 complements a yeast phosphate-uptake- kinetic properties of the ClPHT4;2 transporter. Increasing exter- defective mutant nal 32Pi concentrations induced a rapid increase in the uptake Complementation analysis involving the yeast phosphate-uptake rate, which approached saturation at higher substrate concentra- mutant MB192 was used to obtain biochemical evidence that tions (Fig. 3c). Our data revealed that the Pi uptake as mediated ClPHT4;2 functions as a Pi transporter. AtPHT4;2 was used as a by ClPHT4;2 followed Michaelis–Menten kinetics, with an positive control (Irigoyen et al., 2011). The yeast cells were trans- apparent mean Km of 0.44 mM Pi based on the average of three formed with ClPHT4;2, ClPHT4;2M1-64 (deleted the 64- independent experiments. Together, the data suggested that amino-acid putative transit peptide in the N-terminus), ClPHT4;2 functions as a plasma membrane Pi transporter in AtPHT4;2, AtPHT4;2M1-44 (deleted the 44-amino-acid yeast cells, mediating Pi uptake at micromolar concentrations.

(a) (b) a

Fig. 3 Functional expression of ClPHT4;2 in b c yeast. (a) Subcellular localization of PHT-GFP in yeast. (a) ClPHT4;2-GFP; (b) vector control; (c) ClPHT4;2(D1-64)-GFP; (d) AtPHT4;2-GFP; (e) AtPHT4;2(D1-44)-GFP. Bars 5 lm. (b) Complementation of a yeast inorganic phosphate (Pi) transport mutant de with PHT4 genes grown in 60, 40, and 20 lM Pi-limiting SD media. Yeast MB192 cells harboring either an empty vector (control) or the indicated PHT4 cDNA constructs were grown in SD medium containing 220 lMPitoOD600 = 1. Equal volumes of 10-fold serial dilutions were (c) applied to different Pi-limited media (60, 40, or 20 lM; pH 5.5) and then incubated at 30°C for 3 d. The AtPHT4;2(D1-44) cDNA and ClPHT4;2(D1-64) cDNA lack the region encoding an N-terminal 44- or 64-amino- acid putative transit peptide, and a new ATG start codon has been added. (c) Kinetics of Pi uptake. The rates of Pi uptake in yeast MB192 cells expressing AtPHT4;2 or ClPHT4;2(D1-64) or carrying the vector control were determined with increasing external Pi concentrations and ﬁtted by nonlinear regression according to the Michaelis–Menten equation. Values shown are means SE for three independent experiments.

Ó 2016 The Authors New Phytologist (2016) New Phytologist Ó 2016 New Phytologist Trust www.newphytologist.com New 8 Research Phytologist

ClPHT4;2 expression level changes (Fig. 4b), while the decrease ClPHT4;2 is a chromoplast-localized protein in ClPHT4;2 content at the ripe stage (34DAP) may have The presence of a potential plastid-targeting sequence and the occurred because of ruptured chromoplast. gene expression pattern of ClPHT4;2 prompted us to hypothesize We examined the subcellular localization of ClPHT4;2 in that the ClPHT4;2 protein is localized to chromoplasts in water- watermelon fruit protoplasts. The ClPHT4;2-mCherry fusion melon flesh. To test this hypothesis, the specific anti-ClPHT4;2 protein and free mCherry (vector control) were transiently pro- antibody was generated (see ‘Experimental procedures’ section), duced in flesh protoplasts that were isolated from 97103 fruits at and the intact chromoplasts that were isolated from different fruit 14 DAP. The fluorescence signals of all of the ClPHT4;2- developmental stages (Fig. 4a) were subjected to western blot mCherry fusion proteins coincided precisely with chromoplast analysis. A single cross-reacting band between 35 and 40 kD was autofluorescence (green), which was in contrast to the ubiquitous detected, as shown in Fig. 4(b). During 97103 fruit development, distribution of free mCherry (Fig. 4c). To determine the precise the increase in ClPHT4;2 abundance was consistent with localization of ClPHT4;2 in watermelon fruit cells, ultrathin

(a) (c)

(b)

(d) (e)

(f)

Fig. 4 ClPHT4;2 localized in watermelon fruit chromoplasts. (a) Chromoplasts were isolated by sucrose gradient centrifugation as described in the Materials and Methods section. Intact chromoplasts were collected from layers between 17–30% and 30–50%. (b) Immunoblot analysis of the expression levels of ClPHT4;2 protein in chromoplasts. Isolated chromoplasts were washed extensively to remove contaminated proteins. Anti-ClPHT4;2 was used to measure ClPHT4;2 expression (top). ACTIN was used for normalization (bottom). (c) Subcellular localization of ClPHT4;2. ClPHT4;2-mCherry and mCherry were transiently produced in watermelon fruit protoplasts, incubated in the dark for 12 h and observed with a confocal microscope. Bars, 20 lm. (d) Immunogold-based localization of ClPHT4;2 in watermelon ﬂesh cells at 10 DAP. The protein reacting with anti-ClPHT4;2 serum mainly resides in the plasma membrane of proplastids. An enlargement of the red frame area is shown. Bars, 0.5 lm. (e) Negative control of the immunogold localization. Gold particles were scattered throughout the cytoplasm or vacuoles in a control serum. An enlargement of the red frame area is shown. Bars, 0.5 lm. (f) Immunogold-based localization of ClPHT4;2 in chromoplasts. An enlargement of the red frames areas is shown. Bars, 1 lm.

New Phytologist (2016) Ó 2016 The Authors www.newphytologist.com New Phytologist Ó 2016 New Phytologist Trust New Phytologist Research 9 sections of fruits at 10 DAP were analyzed by immunogold elec- the accumulation of sugar, while it was maintained at low levels tron micoscopy (Fig. 4d). Most of the gold particles were local- in PI296341-FR flesh (Fig. 5a). Cla014572, termed ClbZIP2, ized in the proplastids. Specific gold particle localization patterns was a homolog to Arabidopsis S1 bZIP AtbZIP44 (protein iden- were not observed in control sections that were incubated with tity 49.13%), with one upstream open reading frame (uORF) the negative control serum (Fig. 4e). We also detected gold parti- present in the 50-leader sequence of Cla014572 homolog to cles in fruit chromoplasts at 18 DAP (Fig. 4f). A quantitative Arabidopsis group-S bZIP factor family, and members in this analysis of the distribution of gold particle indicated a preferen- family own a SC-uORF in its 50-leader, which is important for tial association of ClPHT4;2 with the chromoplast, accounting sucrose-induced repression of translation. (Fig. S5). ClbZIP2 for 57% of the total gold particles in fruits at 18 DAP (Table 1). expression was repressed during fruit ripening in 97103, but These results imply that ClPHT4;2 is localized in the chromo- remained at a high level in PI296341-FR (Fig. 5b). plasts. To test the binding activity to different parts of ClPHT4;2 promoter regions, Y1H experiments were carried out. ClbZIP1 and ClbZIP2 bound to the ABRE-box elements and the 340 bp Distinct ABRE and ERE cis-elements were found in the red- fragment of the 97103 ClPHT4;2 promoter region containing fleshed cultivar 97103 ClPHT4;2 promoter region, and two two ABRE motifs. However, they were unable to bind to the cor- ABRE-binding bZIP TFs were identified responding promoter region in PI296341-FR (Fig. 5c). The To identify the trans-acting factors regulating ClPHT4;2 differ- EMSA results confirmed the selective binding of ClbZIP1 and ential expression, we analyzed the 1.5 kb promoter regions of ClbZIP2 to the ABREs of the ClPHT4;2-97103 promoter ClPHT4;2 from the 97103 and PI296341-FR genomes. The (Fig. 5d). Next, using a well-established transient expression assay putative cis-elements in these two promoters were predicted using of Nicotiana benthamiana leaves, we verified the inductive effects the PlantCARE online tool (http://bioinformatics.psb.ugent.be/ of ClbZIP1 on the expression of a firefly luciferase gene (LUC) webtools/plantcare/html/) (Rombauts et al., 1999). SNPs in these reporter gene under the control of the 1.5-kb ClPHT4;2 pro- two promoter regions led to the loss of an ethylene-responsive moter from 97103 or PI296341-FR plant. Coexpression of the element ERE motif (ATTTCAAA) and two ABRE motifs (T(C/ 35Spro:ClbZIP1 construct with the 97103-ClPHT4;2pro:LUC T)ACGT(T/G)) in PI296341-FR genome (Fig. S3). These three resulted in an obvious activation of luminescence intensity motifs were absent in all pale-fleshed C. lanatus ssp. lanatus acces- (Fig. 5e), suggesting that ClbZIP1 activated reporter gene expres- sions listed in Table S4. These two cis-acting regulatory motifs sion. By contrast, in the coexpression experiment involving the are regarded as TF-binding sites (Kumar et al., 2012; Sarkar & PI296341-FR ClPHT4;2 promoter, which does not contain the Lahiri, 2013; Liu et al., 2015). To clarify the transcript regulation two ABRE motifs, the LUC intensity was unaffected. In a parallel of ClPHT4;2, we used a Y1H assay to isolate ABRE and ERE coexpression experiment, ClbZIP2 did not induce reporter gene binding factor (s). The bait containing the ERE motif without expression (Fig. 5f), possibly because of the loss of heterodimer any AD vectors induced reporter gene expression, indicating that partners, as reported in Arabidopsis (Weltmeier et al., 2009). some yeast TFs can bind to this motif. No such self-activation These results suggest that ClbZIP1 and ClbZIP2 bind to the was observed with the ABRE motif. Three ABRE copies were 97103 ClPHT4;2 promoter to regulate this gene expression. used as bait. We screened c. 3.0 9 106 clones and identified 36 positive colonies that were capable of growing in the presence of ClPHT4;2 is transcriptionally induced by ABA, ethephon, aureobasidin A. The plasmids from the positive clones were and glucose sequenced and eight annotated DNA-binding proteins were listed in Table S5. Among these candidate binding proteins, we Analyses of cis-elements and candidate TFs revealed that ABA, focused on the two proposed bZIP transcriptional factors: ethylene and sugar may regulate ClPHT4;2 gene expression in Cla017696 and Cla014572. Based on genomic database infor- 97103. ABA, fluridone, ethephon, AVG, sucrose and glucose mation and RACE analysis, we obtained the full-length mRNA were applied to developing fruits (10 DAP) to investigate their sequences of Cla014572 (GenBank no. KU963292) and effects on ClPHT4;2 expression and flesh color development. Cla017696 (GenBank no. KU963291). Cla017696, termed Exposure to ABA accelerated flesh color development in 97103, ClbZIP1, was a homolog of Arabidopsis ABI5 TF (protein iden- whereas fluridone treatment blocked this process. Exogenous tity 51.80%). The full-length ClbZIP1 mRNA sequence is pro- applications of glucose, but not sucrose, stimulated the flesh color vided in Fig. S4. During watermelon fruit development, development in 97103 (Fig. 6a). Ethephon treatment also pro- ClbZIP1 expression was highly induced in 97103 flesh following moted flesh pigmentation, while AVG had little effect on this process (Fig. 6b). No obvious changes to flesh color were Table 1 Subcellular distribution of anti-ClPHT4;2 in watermelon fruit flesh cells of watermelon (mean SD) (n = 5) observed in the treated PI296341-FR fruits (Fig. 6a,b). In 97103 fruit flesh, ABA, ethylene and glucose treatments promoted the Chromoplast (%) Mitochondrion (%) ER (%) Others (%) carotenoid content and ClPHT4;2 expression, while the fluridone treatment blocked these processes. Sucrose treatment 57 1.45 11 1.32 8 1.27 24 2.04 induced the ClPHT4;2 expression, but had no obvious effect on Values represent the percentages of the total labeling in distinct locations flesh color development in 97103 flesh (Fig. 6c,d). However, no in fruit flesh cells of watermelon. ER, endoplasmic reticulum. treatment influenced the carotenoid content and ClPHT4;2

Ó 2016 The Authors New Phytologist (2016) New Phytologist Ó 2016 New Phytologist Trust www.newphytologist.com New 10 Research Phytologist

Fig. 5 Binding of ClbZIP1 and ClbZIP2 to the 97103-ClPHT4;2 promoter region. (a, b) Sugar contents (°Brix) and expression levels of ClbZIP1 (a) and ClbZIP2 (b) at different fruit development stages in 97103 and PI296341-FR. °Brix values (blue line) were measured as described in the Materials and Methods section. Gene expression levels were measured by real-time PCR. Watermelon ACTIN was used as an internal control. Error bars indicate SD (n = 5). (c) Yeast one-hybrid assay. Yeast that was transformed with the indicated vector was diluted and inoculated onto leucine-negative SD media (SD/- Leu). Interaction was tested on leucine-negative SD media (SD/-Leu) with aureobasidin A (AbA). Bait 1, ABRE in 97013; bait 2, ABREmu in PI296341-FR; bait 3, 340 bp 97103-ClPHT4;2 promoter region containing the two ABREs; and bait 4, 340 bp PI296341-FR-ClPHT4;2 promoter region. ClbZIP1 and ClbZIP2 bound to the ABRE and a 340 bp region in the 97103-ClPHT4;2 promoter but not the corresponding region in PI296341-FR. (d) Electrophoretic mobility shift assayEMSA showed that ClbZIP1 and ClbZIP2 proteins can bind to the ABRE motif in the promoter of ClPHT4;2. A protein and DNA complex can be detected when ClbZIP1-polyhistidine and ClbZIP2-polyhistidine fusions were incubated with an ABRE motif from the ClPHT4;2 promoter of 97103. The detected DNA-binding was reduced by the addition of a 100-fold molar excess of an unlabeled oligonucleotide probe (line 3). The DNA binding activity was undetectable when incubated with the coordinate ABREmu region in PI296341-FR (line 4). (e) An LCI assay showed the interactions with the ClbZIP1/2 and ClPHT4;2 1.5-kb promoter in Nicotiana benthamiana leaves. Two effectors and two reporters are shown on the top lines. Representative images of N. benthamiana leaves 60 h after infiltration are presented (activated status (reporter + effector) and nonactivated status (reporter only)). Transient expression assays revealed that only in region 4 could effector ClbZIP1 activate 97103-ClPHT4;2pro:LUC expression. The bottom lines indicate the infiltrated constructs and treatments. (f) Quantitative analysis of luminescence intensity in (e) based on three independent measurements. Error bars represent SD (n = 3). Asterisks denote Student’s t-test significance compared with control plants: **, P < 0.01.

expression in PI296341-FR fruit flesh (Fig. 6c,d). These results their roles in chloroplasts, root heterotrophic plastids and Golgi strongly suggest that ABA, ethylene and sugar are important in bodies (Guo et al., 2008). Our results reveal that the PHT4 trans- the regulation of ClPHT4;2 expression and flesh color develop- porter family takes part in fruit flesh chromoplast development ment in red-fleshed cultivar. and carotenoid accumulation. Tomato chromoplasts were recently observed to synthesize Discussion ATP through chromorespiration activites of respiratory bioenergetic organelles during fruit ripening (Pateraki et al., 2013; We identified ClPHT4;2 as a new Pi transporter protein local- Renato et al., 2014). Wang et al. (2013) also observed high ized in watermelon flesh chromoplast membrane to transport Pi metabolic activities and an abundance of transport-related pro- between the cytosol and chromoplasts. High ClPHT4;2 expres- teins in watermelon chromoplasts. The accumulated Pi resulting sion levels were significantly correlated with increased flesh from these reactions needs to be transported during watermelon carotenoid content in various watermelon accessions. Knock- chromoplast development. ClPHT4;2 was analyzed as a potential down of ClPHT4;2 expression weakened the flesh color forma- Pi transporter in a yeast system. Down-regulated ClPHT4;2 tran- tion in RNAi transgenic watermelon lines. These results provided scription influenced chromoplast development and inhibited the strong evidence that ClPHT4;2 influences flesh color develop- carotenogenesis pathway. In the ClPHT4;2-RNAi transgenic ment. ClPHT4;2 belongs to the PHT4 transporter family, and is lines, the transcription abundances of three tested carotenoid the first member of this family confirmed to be involved in fruit biosynthesis genes (i.e., PSY, PDS and CHYb) decreased. flesh color development and carotenoid accumulation. A previous ClPHT4;2 may function as a dynamic Pi transporter affecting Pi functional analysis of the Arabidopsis PHT4 proteins uncovered homeostasis during energy metabolism in chromoplast.

New Phytologist (2016) Ó 2016 The Authors www.newphytologist.com New Phytologist Ó 2016 New Phytologist Trust New Phytologist Research 11

Fig. 6 Effects of ABA, fluridone, sucrose, glucose, ethephon and aminoethoxyvinylglycine (AVG) on flesh color development and on the relative expression of ClPHT4;2. (a) Changes in fruit flesh after treatment with ABA, fluridone, sucrose, glucose and the control to 97103 (upper row) and PI296341-FR (lower row) watermelon fruits. Bars, 3 cm. (b) Fruit flesh changes after ethephon and AVG treatments. (c) Carotenoid content of the fruits presented in (a) and (b). Asterisks denote Student’s t-test significance compared with control plants: *, P < 0.05, **, P < 0.01. (d) Effect of treatments on ClPHT4;2 transcript accumulation. Watermelon ACTIN was used as an internal control. Error bars indicate SD (n = 3). A significant difference (*, P < 0.05; **, P < 0.01) was found compared with control plants.

ClPHT4;2 may serve as two-way valves with transport direction globular structures containing pigment in cultivated watermelon dependent on the Pi electrochemical gradient and proton-motive ripe flesh were considered as a special chromoplast category forces, similar to other PHT4 family members (Guo et al., 2008; (Bangalore et al., 2008). We detected the carotenoid globule release Fischer, 2011). We hypothesized that the transport of Pi between progress that does not occur in the wild watermelon species. We subcellular compartments is central to various metabolic activities propose that the high accumulation of carotenoids during chromo- during chromoplast development, with ClPHT4;2 playing an plast development induces the chromoplast rupture, which is neces- important role. A comprehensive characterization of the physio- sary for flesh color development in watermelon cultivars. The logical relationship between Pi transport and carotenoid biosyn- chromoplast development and carotenoid accumulation pathways thesis will require additional research. The decreased abundance are arrested in wild pale-colored lines. The chromoplast structures of ClPHT4;2 in ClPHT4;2-RNAi lines did not completely pre- stay intact at the ripe stage of many carotenoid-storing raw fruits vent the development of flesh color in red-fleshed watermelon and vegetable (Jeffery et al., 2012). Intact chromoplasts in ripening lines. This may have been because of the presence of other Pi tomato can decrease the rates of carotenoid solubilization when transporters located in the plastid membrane, such as triose exposed to digestive forces (Jeffery et al.,2012).Tomatoneedsheat P/phosphate translocators or nucleoside transporters (Fischer, treatment to increase the lycopene absorption efficiency, while the 2011). lycopene from fresh watermelon juice is available directly (Naz et al., Chromoplasts are enclosed by two membranes that are impor- 2014). The lycopene in watermelon is regarded as more available to tant for keeping structural stability and connecting the chromo- human assumption than that in tomato (Edwards et al., 2003; Naz plast metabolic activities with those of the other plant cell parts et al., 2014). We believe that many released carotenoid globules in (Fischer, 2011). We observed that many carotenoid globules were cultivated watermelon flesh cells are responsible for the high released from the collapsed chromoplasts to cytoplasm and accu- lycopene absorption efficiency. mulated in colored flesh cells during cultivated watermelon fruit Fruit quality formation is a complex, genetically programmed ripening, even though these carotenoid globules should be sur- process that is regulated by complex networks involving multiple rounded by a membrane. The breakdown of chromoplast and pathways (Seymour et al., 2013). The phytohormone and sugar the lycopene bodies that are ‘free’ in the cytoplasm had not been signaling pathways are important for watermelon fruit ripening reported in watermelon before. In previous studies, those dark and flesh color formation (Grassi et al., 2013). Using the same

Ó 2016 The Authors New Phytologist (2016) New Phytologist Ó 2016 New Phytologist Trust www.newphytologist.com New 12 Research Phytologist

(a)

(b)

Fig. 7 Hypothetical models of ClPHT4;2 transcriptional regulation in cultivated and wild watermelon lines. (a) Proposed model for the induction of ClPHT4;2 following treatments with ABA, ethylene, and sugar in colored ﬂesh cells from cultivated line. ABA and sugar treatments regulate the activation of ClPHT4;2 expression by TF-ClbZIP1. Sugar content may lead to sucrose-induced repression of translation of ClbZIP2. Ethylene activates gene expression through an unknown transcription factor (TF). (b) Without the relevant cis-elements in the wild watermelon promoter region, ClPHT4;2 transcription cannot be regulated by phytohormone or sugar signaling pathways.

RIL population derived from a cross between 97103 and our collection consisting of 1197 Citrullus accessions (Zhang PI296341-FR, we previously reported a major quantitative trait et al., 2016), which covered the genetic diversity of the Citrullus locus in chromosome 2 associated with economically important species. High ClPHT4;2 expression levels were observed in all traits, including °Brix, flesh color, fruit length and fruit width tested watermelon lines with colored fruit flesh. (Ren et al., 2014; Zhang et al., 2014). Here, we clarified the In summary, high ClPHT4;2 expression levels are necessary downstream effect gene ClPHT4;2 in chromoplast development for watermelon flesh color formation. Additionally, they were that responded to biotic signals. Two bZIP transcription factors, important during the domestication of flesh color in modern ClbZIP1 and ClbZIP2, which influence ClPHT4;2 expression in watermelons. We proposed a possible regulatory relationship red-fleshed watermelons, are regulated by ABA and sugar to dif- between the carbohydrate accumulation and the flesh color devel- ferent extents. We developed a model for the sugar and phyto- opment. This relationship may partly explain the co- hormone signaling that mediates ClPHT4;2 transcription in development of flesh color and sweetness from pale-fleshed, no- cultivated watermelons which differs from that in wild accessions sweet ancestors to modern dessert cultivars. (Fig. 7). During fruit development and sugar accumulation in cultivated watermelon, ABA and sugar signaling pathways activated ClPHT4;2 expression through ABRE motifs and the Acknowledgements related ClbZIP1 TF. Another TF, ClbZIP2, is directly repressed We thank Dr Li Li of Cornell University for providing invaluable by sugar. Ethylene also activates ClPHT4;2 transcription, with a suggestions during the project. We are grateful to Dr Shuizhang conserved ERE motif and the related unknown TFs that are Fei of Iowa State University for improving the language in this probably involved in ethylene response. Although watermelon article. This research was supported financially by grants from fruits have traditionally been classified as nonclimacteric, they are the Ministry of Science and Technology of The People’s very sensitive to exogenous ethylene, which leads to accelerated Republic of China (2016YFD0100506), NSFC Research softening, water-soaking (i.e. abnormal ripening), and induced Program (31361140355, 31401893, 31272184, and 31301738), expression of specific genes in cultivated lines (Karakurt & the Beijing Scholar Program (BSP026), the Beijing Nova Pro- Huber, 2002). These regulatory activities may be absent in the gram (2016B039), the Beijing Excellent Talents Program pale-fleshed fruits of C. lanatus ssp. lanatus because of the missing (2014000021223TD03) and the Ministry of Agriculture of critical cis-elements in the promoter regions. China (CARS-26). It is believed that colored flesh and sweet cultivated watermelons originate from their pale-colored and no-sweet progenitors. During domestication, the various biochemical and physiological Author contributions changes occurring in cultivated watermelons and their progenitors J.Z. conducted most of the experiments and helped to write the required considerable coordinated changes in gene expression. article. S.G. and Y.R. analyzed the data. H.Z. and G.G. supervised Clarifying the complex and dynamic gene expression patterns and the experiments. M.Z. and G.W. provided technical assistance to regulatory pathways and their effect on plant growth and develop- J.Z., while M.Z., H.H. and F.L. contributed to the phenotype ment has recently become one of the primary objectives of plant analysis. Y.X. conceived the study and revised the article. biologists (Francesconi & Lehner, 2014). Quantitative differences in gene expression are thought to contribute to phenotypic differences among individuals (G€oring et al., 2007). In this study, the References variability in ClPHT4;2 expression was consistent with the accu- Bang H, Davis AR, Kim S, Leskovar DI, King SR. 2010. Flesh color inheritance mulation of fruit flesh pigmentation in 198 watermelon lines. Of and gene interactions among canary yellow, pale yellow, and red watermelon. these, the 102 representative natural accessions were selected from Journal of the American Society for Horticultural Science 135: 362–368.

New Phytologist (2016) Ó 2016 The Authors www.newphytologist.com New Phytologist Ó 2016 New Phytologist Trust New Phytologist Research 13

Bang H, Kim S, Leskovar D, King S. 2007. Development of a codominant Jeffrey C. 2001. Cucurbitaceae. In: Hanelt P, ed. Mansfeld’s encyclopedia of CAPS marker for allelic selection between canary yellow and red watermelon agricultural and horticultural crops 3. Berlin, Germany: Springer-Verlag, 1510– based on SNP in lycopene b-cyclase (LCYB) gene. Molecular Breeding 20: 1557. 63–72. Kang BS, Zhao W, Hou YB, Tian P. 2010. Expression of carotenogenic genes Bangalore DV, Mcglynn WG, Scott DD. 2008. Effects of fruit maturity on during the development and ripening of watermelon fruit. Scientia Horticulture watermelon ultrastructure and intracellular lycopene distribution. Journal of 124: 368–375. Food Science 73: S222–S228. Karakurt Y, Huber D. 2002. Cell wall-degrading enzymes and pectin solubility Bun-Ya M, Nishimura M, Harashima S, Oshima Y. 1991. The PHO84 gene of and depolymerization in immature and ripe watermelon (Citrullus lanatus) Saccharomyces cerevisiae encodes an inorganic phosphate transporter. Molecular fruit in response to exogenous ethylene. Physiologia Plantarum 116: 398–405. and Cellular Biology 11: 3229–3238. Koch KE. 1996. Carbohydrate-modulated gene expression in plants. Annual Chen Q, Sun J, Zhai Q, Zhou W, Qi L, Xu L, Wang B, Chen R, Jiang H, Qi J Review of Plant Physiology and Plant Molecular Biology 47: 509–540. et al. 2011. The basic Helix-Loop-Helix transcription factor MYC2 directly Kumar R, Sharma MK, Kapoor S, Tyagi AK, Sharma AK. 2012. Transcriptome represses PLETHORA expression during jasmonate-mediated modulation of analysis of rin mutant fruit and in silico analysis of promoters of differentially the root stem cell niche in Arabidopsis. Plant Cell 23: 3335–3352. regulated genes provides insight into LeMADS-RIN-regulated ethylene- Cheng WH, Endo A, Zhou L, Penney J, Chen HC, Arroyo A, Leon P, Nambara dependent as well as ethylene-independent aspects of ripening in tomato. E, Asami T, Seo M et al. 2002. A unique short-chain dehydrogenase/reductase Molecular Genetics and Genomics 287: 189–203. in Arabidopsis glucose signaling and abscisic acid biosynthesis and functions. Lee JM, Joung JG, McQuinn R, Chung MY, Fei Z, Tieman D, Klee H, Plant Cell 14: 2723–2743. Giovannoni J. 2012. Combined transcriptome, genetic diversity and Chomicki G, Renner SS. 2015. Watermelon origin solved with molecular metabolite profiling in tomato fruit reveals that the ethyleneresponse factor phylogenetics including Linnaean material: another example of museomics. SlERF6 plays an important role in ripening and carotenoid accumulation. New Phytologist 205: 526–532. Plant Journal 70: 191–204. Collins JK, Perkins-Veazie P, Roberts W. 2006. Lycopene: from plants to Li L, Yuan H. 2013. Chromoplast biogenesis and carotenoid accumulation. humans. HortScience 41: 135–1144. Archives of Biochemistry and Biophysics 539: 102–109. Edwards AJ, Vinyard BT, Wiley ER, Brown ED, Collins JK, Perkins-Veazie P, Lin CY, Ku HM, Chiang YH, Ho HY, Yu TA, Jan FJ. 2012. Development of Baker RA, Clevidence BA. 2003. Consumption of watermelon juice increases transgenic watermelon resistant to Cucumber mosaic virus and Watermelon plasma concentrations of lycopene and b-carotene in humans. Journal of mosaic virus by using a single chimeric transgene construct. Transgenic Research Nutrition 133: 1043–1050. 21: 983–993. Egea I, Barsan C, Bian W, Purgatto E, Latche A, Chervin C, Bouzayen M, Pech Liu TY, Huang TK, Yang SY, Hong YT, Huang SM, Wang FN, Chiang SF, JC. 2010. Chromoplast differentiation: current status and perspectives. Plant Tsai SY, Lu WC, Chiou TJ. 2016. Identification of plant vacuolar transporters and Cell Physiology 51: 1601–1611. mediating phosphate storage. Nature Communications 7: 11095. Fischer K. 2011. The import and export business in plastids: transport processes Liu M, Pirrello J, Chervin C, Roustan JP, Bouzayen M. 2015. Ethylene control across the inner envelope membrane. Plant Physiology 155: 1511–1519. of fruit ripening: revisiting the complex network of transcriptional regulation. Flugge€ UI, H€ausler RE, Ludewig F, Gierth M. 2011. The role of transporters in Plant Physiology 169: 2380–2390. supplying energy to plant plastids. Journal of Experimental Botany 62: 2381– Loque D, Lalonde S, Looger LL, von Wiren N, Frommer WB. 2007. A cytosolic 2392. trans-activation domain essential for ammonium uptake. Nature 446: 195– Francesconi M, Lehner B. 2014. The effects of genetic variation on gene 198. expression dynamics during development. Nature 505: 208–211. Naz A, Butt MS, Sultan MT, Qayyum MM, Niaz RS. 2014. Watermelon G€oring HH, Curran JE, Johnson MP, Dyer TD, Charlesworth J, Cole SA, lycopene and allied health claims. Excli Journal 13: 650–660. Jowett JB, Abraham LJ, Rainwater DL, Comuzzie AG et al. 2007. Discovery Nisar N, Li L, Lu S, Khin NC, Pogson BJ. 2015. Carotenoid metabolism in of expression QTLs using large-scale transcriptional profiling in human plants. Molecular Plant 8:68–82. lymphocytes. Nature Genetics 39: 1208–1216. Paris HS. 2015. Origin and emergence of the sweet dessert watermelon, Citrullus Grassi S, Piro G, Lee JM, Zheng Y, Fei Z, Dalessandro G, Giovannoni JJ, lanatus. Annals of Botany 116: 133–148. Lenucci MS. 2013. Comparative genomics reveals candidate carotenoid Pateraki I, Renato M, Azcon-Bieto J, Boronat A. 2013. An ATP pathway regulators of ripening watermelon fruit. BMC Genomics 14: 781. synthaseharboring an atypical g-subunit is involved in ATP synthesis in tomato Guo B, Jin Y, Wussler C, Blancaflor E, Motes C, Versaw WK. 2008. Functional fruit chromoplasts. Plant Journal 74:74–85. analysis of the Arabidopsis PHT4 family of intracellular phosphate Pavon LR, Lundh F, Lundin B, Mishra A, Persson BL, Spetea C. 2008. transporters. New Phytologist 177: 889–898. Arabidopsis ANTR1 is a thylakoid Na+-dependent phosphate transporter Guo S, Liu J, Zheng Y, Huang M, Zhang H, Gong G, He H, Ren Y, Zhong S, functional characterization in Escherichia coli. Journal of Biological Chemistry Fei Z et al. 2011. Characterization of transcriptome dynamics during 283: 13520–13527. watermelon fruit development: sequencing, assembly, annotation and gene Quick WP, Neuhaus HE. 1996. Evidence for two types of phosphate expression profiles. BMC Genomics 12: 454. translocators in sweet-pepper (Capsicum annuum L.) fruit chromoplasts. Guo S, Sun H, Zhang H, Liu J, Ren Y, Gong G, Jiao C, Zheng Y, Yang W, Fei Biochemical Journal 320:7–10. Z et al. 2015. Comparative transcriptiome analysis of cultivated and wild Rausch C, Bucher M. 2002. Molecular mechanisms of phosphate transport in watermelon during fruit development. PLoS One 10: e0130267. plants. Planta 216:23–37. Guo S, Zhang J, Sun H, Salse J, Lucas WJ, Zhang H, Zheng Y, Mao L, Ren Y, Ren Y, McGregor C, Zhang Y, Gong G, Zhang H, Guo S, Sun H, Cai W, Wang Z et al. 2013. The draft genome of watermelon (Citrullus lanatus) and Zhang J, Xu Y. 2014. An integrated genetic map based on four mapping resequencing of 20 diverse accessions. Nature Genetics 45:51–58. populations and quantitative trait loci associated with economically important Henderson WR, Scott GH, Wehner TC. 1998. Interaction of flesh color genes traits in watermelon (Citrullus lanatus). BMC Plant Biology 14: 33. in watermelon. Journal of Heredity 89:50–53. Renato M, Pateraki I, Boronat A, Azcon-Bieto J. 2014. Tomato fruit Irigoyen S, Karlsson PM, Kuruvilla J, Spetea C, Versaw WK. 2011. The sink- chromoplasts behave as respiratory bioenergetic organelles during ripening. specific plastidic phosphate transporter PHT4;2 influences starch accumulation Plant Physiology 166: 920–933. and leaf size in Arabidopsis. Plant Physiology 157: 1765–1777. Rolland F, Baena-Gonzalez E, Sheen J. 2006. Sugar sensing and signaling in Jeffery J, Holzenburg A, Kin S. 2012. Physical barriers to carotenoid plants: conserved and novel mechanisms. Annual Review of Plant Biology 57: bioaccessibility. Ultrastructure survey of chromoplast and cell wall morphology 675–709. in nine carotenoid-containing fruits and vegetables. Journal of the Science of Rombauts S, Dehais P, Van Montagu M, Rouze P. 1999. PlantCARE, a plant Food and Agriculture 92: 2594–2602. cis-acting regulatory element database. Nucleic Acids Research 27: 295–296.

Ó 2016 The Authors New Phytologist (2016) New Phytologist Ó 2016 New Phytologist Trust www.newphytologist.com New 14 Research Phytologist

Ruan YL. 2014. Sucrose metabolism: gateway to diverse carbon use and sugar Supporting Information signaling. Annual Review of Plant Biology 65:33–67. Sarkar AK, Lahiri A. 2013. Specificity determinants for the abscisic acid response Additional Supporting Information may be found online in the element. FEBS Open Bio 3: 101–105. Supporting Information tab for this article: Seymour GB, Østergaard L, Chapman NH, Knapp S, Martin C. 2013. Fruit development and ripening. Annual Review of Plant Biology 64: 219–241. Smeekens S, Ma J, Hanson J, Rolland F. 2010. Sugar signals and molecular Fig. S1 Gene structure and expression pattern of ClPHT4;2 networks controlling plant growth. Current Opinion in Plant Biology 13: 274– (Cla017962). 279. Su L, Diretto G, Purgatto E, Danoun S, Zouine M, Li Z, Roustan JP, Bouzayen Fig. S2 Phylogenetic tree of the ClPHT, AtPHT, OsPHT, and M, Giuliano G, Chervin C. 2015. Carotenoid accumulation during tomato tomato PHT gene families. fruit ripening is modulated by the auxin-ethylene balance. BMC Plant Biology 15: 114. Wang YQ, Yang Y, Fei Z, Yuan H, Fish T, Thannhauser TW, Mazourek M, Fig. S3 Comparison of the 900 bp ClPHT4;2 promoter Kochian LV, Wang X, Li L. 2013. Proteomic analysis of chromoplasts from six sequences from 97103 and PI296341-FR watermelon plants. crop species reveals insights into chromoplast function and development. – Journal of Experimental Botany 64: 949 961. Fig. S4 ClbZIP1 mRNA sequence and its similarity to the closely Welsch R, Maass D, Voegel T, Dellapenna D, Beyer P. 2007. Transcription factor RAP2.2 and its interacting partner SINAT2: stable related Arabidopsis thaliana ABI5 mRNAABI5 protein. elements in the carotenogenesis of Arabidopsis leaves. Plant Physiology 145: 1073–1085. Fig. S5 ClbZIP2 mRNA sequence and its similarity to the closely Welsch R, Medina J, Giuliano G, Beyer P, von Lintig J. 2003. Structural and related Arabidopsis thaliana AtbZIP44 protein. functional characterization of the phytoene synthase promoter from Arabidopsis. Planta 216: 523–534. Weltmeier F, Rahmani F, Ehlert A, Dietrich K, Schutze€ K, Wang X, Chaban C, Table S1. Primers used for genotyping, probe synthesis, cloning, Hanson J, Teige M, Harter K et al. 2009. Expression patterns within the and expression analysis. Arabidopsis C/S1 bZIP transcription factor network: availability of heterodimerization partners controls gene expression during stress response and Table S2. Candidate plastid transporter genes and their expres- – development. Plant Molecular Biology 69: 107 119. sion levels during watermelon fruit development. Wen L, Li Y. 2009. Distribution of an ankyrin-repeat protein on the endoplasmic reticulum in Arabidopsis. Journal of Integrative Plant Biology 51: 140–146. Wiese A, Elzinga N, Wobbes B, Smeekens S. 2004. A conserved upstream open Table S3. Contents of four main carotenoids, °Brix values, and reading frame mediates sucrose-induced repression of translation. Plant Cell 16: ClPHT4;2 transcript abundances in the recombinant inbred line 1717–1729. population. Yan X, Wang H, Ye Y, Zeng G, Ma R, Mi F, Yao L. 2012. pYBA100: an ease-of- use binary vector with LoxP-FRT recombinase site for plant transformation. ° Molecular Plant Breeding 10: 371–379. Table S4. Contents of four main carotenoids, Brix values, and Yoo SD, Cho YH, Sheen J. 2007. Arabidopsis mesophyll protoplasts: a versatile ClPHT4;2 transcript abundances in the wild germplasm popula- cell system for transient gene expression analysis. Nature Protocols 2: 1565– tion. 1572. Yuan H, Zhang J, Coimbatore D, Li L. 2015. Carotenoid metabolism and Table S5. Characteristics of the cDNA clones (binding protein) regulation in horticultural crops. Horticulture Research 2: 15036. Zhang H, Fan J, Guo S, Ren Y, Gong G, Zhang J, Wen Y, Davis A, Xu Y. 2016. isolated during yeast one-hybrid screening. Genetic diversity, population structure, and formation of a core collection of 1197 Citrullus accessions. HortScience 51:23–29. Please note: Wiley Blackwell are not responsible for the content Zhang J, Gong G, Guo S, Ren Y, Zhang H, Xu Y. 2014. Fine mapping or functionality of any Supporting Information supplied by the of the flesh color controlling genes in watermelon (Citrullus lanatus). In: authors. Any queries (other than missing material) should be Harbor B, ed. Curcurbitaceae 2014 Proceedings, Michigan, USA. Alexandria, VA: American Society for Horticultural Science Press, 111– directed to the New Phytologist Central Office. 116.

New Phytologist (2016) Ó 2016 The Authors www.newphytologist.com New Phytologist Ó 2016 New Phytologist Trust