Functional characterization of a gibberellin F-box protein, PslSLY1, during plum fruit development

Abstract

Fruit development is orchestrated by a complex network of interactions between hormone signaling pathways. The phytohormone gibberellin (GA) is known to regulate a diverse range of developmental processes; however, the mechanisms of GA action in perennial fruit species are yet to be elucidated. In the current study, a GA signaling gene PslSLY1, encoding a putative F-box protein that belongs to the SLY1 (SLEEPY1)/GID2 (gibberellin-insensitive dwarf2) gene family, was isolated from Japanese plum (Prunus salicina). PslSLY1 transcript abundance declined as fruit development progressed, along with potential negative feedback regulation of PslSLY1 by GA. Subcellular localization and protein–protein interaction assays suggested that PslSLY1 functions as an active GA signaling component that interacts with the ASK1 (Arabidopsis SKP1) subunit of an SCF–ubiquitin ligase complex and with PslDELLA repressors, in a GA-independent manner. By using a domain omission strategy, we illustrated that the F-box and C-terminal domains of PslSLY1 are essential for its interactions with the downstream GA signaling components. PslSLY1 overexpression in wild-type and Arabidopsissly1.10 mutant backgrounds resulted in a dramatic enhancement in overall plant growth, presumably due to triggered GA signaling. This includes germination characteristics, stem elongation, flower structure, and fertility. Overall, our findings shed new light on the GA strategy and signaling network in commercially important perennial crops.

F-box protein, fruit development, functional biology, GA signaling, protein–protein interaction, Prunus salicina, subcellular localization

Introduction

Fruiting is a distinct feature in flowering plants and represents an important research field because of its significance for human nutrition. Post-fertilization, fruits develop rapidly via a coordination of hormonal processes, leading to wide irreversible molecular, biochemical, and structural changes. The distinctive metabolic pathways that synchronize the progression of fruit development have been the subject of considerable research in a wide range of fruit species (Seymour et al., 2013; Kumar et al., 2014). The availability of naturally occurring and induced mutants in fruit crops has greatly assisted the elucidation of mechanisms underlying the progression in fruit development. Phytohormones and their dynamic interplay act as master regulators of fruit growth and development (Ozga and Reinecke, 2003; McAtee et al., 2013). The phytohormone gibberellin (GA) is a major plant hormone that is involved in the coordination of a diverse range of functions (Richards et al., 2001; Olszewski et al., 2002; Thomas and Sun, 2004; Fleet and Sun, 2005). Levels and accumulation sites of GA and GA responsiveness are critical for flower induction and floral organ determination, as well as for fruit initiation, development, and ultimate fruit quality (Blázquez et al., 1998; Goto and Pharis, 1999; Bernier and Perilleux, 2005; Mesejo et al., 2013; El-Sharkawy et al., 2014a; Acheampong et al., 2015). Fertilization mutually triggers GA, auxin, and cytokinin pathways in ovules to initiate the flower-to-fruit transition program as well as the onset of the cell division and expansion processes associated with early embryo development (Vivian-Smith and Koltunow, 1999; Ozga and Reinecke, 2003; Davies, 2004; de Jong et al., 2009; Mariotti et al., 2011). Previous studies have demonstrated that GA levels increase commensurate with the progression of fruit development and growth (Bukovac and Nakagawa, 1967; Jackson, 1968; Yamaguchi and Takahashi, 1976; Looney and Pharis, 1986; Talon et al., 1992; El-Sharkawy et al., 2014a). On the other hand, GA-deficient mutant fruits present several developmental disorders, including malformed flowers, reduced fruit size, and delayed ripening (Walser et al., 1981; Serrani et al., 2007; de Jong et al., 2009; El-Sharkawy et al., 2012). While these studies suggested the important contribution of GA in fruit growth (Jackson, 1968; Crane, 1969; Bukvoac and Yuda, 1979; Gillaspy et al., 1993), much of the fundamentals of the mechanism of GA action in these processes remain to be elucidated.

Plants maintain complex GA signaling networks to accommodate the diverse range of often simultaneous responses mediated by the hormone. The characterization of GA-defective mutants, displaying numerous traits, has helped researchers to recognize several crucial GA signaling pathways and outcomes (Magome et al., 2004; Yasumura et al., 2007; Yamaguchi, 2008; Harberd et al., 2009; Olimpieri et al., 2011). The most profound discovery has been the identification of the central GA signaling components in plants. The identification of the soluble GA receptor, GID1, the GA suppressor regulators, DELLA, and the GA activator F-box proteins, SLY1 and SNE (SNEEZY) in Arabidopsis or GID2 and SNE in rice, has improved our fundamental understanding of how GA mediates its activity (Peng et al., 1997; Silverstone et al., 1998; McGinnis et al., 2003; Sasaki et al., 2003; Strader et al., 2004; Ueguchi-Tanaka et al., 2005). At low GA levels, DELLA proteins halt the activity of the GA-responsive, DNA-binding basic helix–loop–helix (bHLH) transcription factors (de Lucas et al., 2008; Feng et al., 2008). GA perception by the GID1 receptor stimulates the interaction of GID1 with the DELLA proteins (Ueguchi-Tanaka et al., 2005; Griffiths et al., 2006), and the establishment of a stable co-receptor GA–GID1–DELLA complex results in conformational changes in DELLA proteins, facilitating SLY1/GID2 F-box and DELLA protein binding. Subsequent DELLA binding to the ubiquitin ligase complex SCF^SLY1/GID2 led to DELLA ubiquitination and degradation. The loss of DELLA promotes plant growth due to the release of GA-responsive genes from their restraint interactor (Griffiths et al., 2006; Hirano et al., 2010).

F-box proteins are subunits of the SCF E3 ubiquitin ligase complex and play diverse roles in plant development (Zheng et al., 2002; Hare et al., 2003; Zhang et al., 2019). In GA signaling, the F-box domain binds to the Skp1 homolog, which conjugates the SLY1/GID2 proteins to Cullin, the core of the SCF complex (Dill et al., 2004; Fu et al., 2004). SLY1/GID2 relies on its C-terminal regulatory (GGF and LSL) domain to bind DELLA proteins via their C-terminal GRAS (VHIID and LHRII) domain (Dill et al., 2004; Fu et al., 2004; Hirano et al., 2010; Ariizumi et al., 2011). DELLA proteins are subsequently ubiquitinated by the E2/E3 ubiquitin ligase pathway via the SCF^SLY1/GID2 complex, and the ubiquitinated DELLA is then degraded via the 26S proteasome with the concomitant discharge of the ubiquitin moieties for reuse (Ariizumi et al., 2011; Gao et al., 2011; Sun, 2011). Although studies in Arabidopsis and rice have revealed a conserved GA signaling pathway across land plant lineages, the fact that each plant carries a distinct set of GA signaling paralogs adds extra complexity to GA response mechanisms. For instance, the GA signaling machinery in Arabidopsis consists of three GID1s, five DELLAs, and two SLY-like proteins (SLY1 and SNE). In contrast, the rice genome has only one representative of each of GID1 and DELLA, and two GID2-like proteins (GID2 and SNE) (Hauvermale et al., 2012). Two GID1s, three DELLAs, and three SLY-like proteins (two SLY1 and one SNE) have been identified in grape (Acheampong et al., 2015). Despite the dissimilarities, genetic and biochemical characterization has shown that all these paralogs are active and exhibit diverse overlapping and distinct roles in regulating different aspects of GA responses.

In the current study, a novel SLY-like gene, PslSLY1, encoding a putative F-box protein of the SLY1/GID2 gene subfamily was isolated from Japanese plum. An analysis of transcript accumulation during fruit development suggested that PslSLY1 is potentially subject to feedback regulation by GA. Assessing the localization and interaction properties of PslSLY1 and truncated versions of the protein revealed that PslSLY1 could influence GA responsiveness via the formation of an SCF^PslSLY1 ubiquitin ligase complex that presumably targets different DELLA repressors for degradation in a GA-independent manner. Overexpression of PslSLY1 in Arabidopsis showed that the protein is able to re-establish GA responsiveness in the wild type (WT) and the GA-insensitive mutant, sly1.10. The presented data provide further evidence for the conserved role of GA in the organization and the progression of fruit development, and shed light on the complex regulatory processes involved in fruit growth.

Materials and methods

Plum tissues and treatments

Flowers and fruit at different developmental stages were collected from Japanese plum (Prunus salicina L.), cultivar Early Golden (EG), as described previously (El-Sharkawy et al., 2007). To explore the influence of GA on PslSLY1 mRNA accumulation, ripe fruits were picked before the onset of climacteric ethylene production, then sterilized, and subjected to treatment with gibberellin (GA₃, 100 μM) and the GA biosynthesis inhibitor paclobutrazol (PAC, 10 μM). Untreated fruits were used as controls. After the treatments, all fruit were stored at room temperature and samples were collected at 0, 3, 6, and 9 d post-treatment for analysis. A defined mix of tissues from three replicate fruits was sampled and frozen for later analysis at each time point.

Isolation and analysis of the PslSLY1 sequence

Based on common structural characteristics among previously characterized SLY1/GID2-like F-box genes, a pair of degenerate primers (#1 and 2, Supplementary Table S1 at JXB online) were designed to amplify a fragment of the plum gene ortholog. The putative PsISLY1 amplicon was cloned, sequenced, and analyzed using BLAST (Altschul et al., 1997). Elongation of the partial cDNA clone was carried out using the 3'- and 5'-RACE kit (Life Technologies, Grand Island, NY, USA). Full-length amplification of the PslSLY1 cDNA sequence was carried out using Platinum Taq DNA Polymerase High Fidelity, following the instructions provided by the manufacturer (Life Technologies). Alignment of predicted proteins was performed using ClustalX software, and the Neighbor–Joining tree was generated with MEGA5 (Tamura et al., 2011). Structural domains were annotated using the online software ‘Smart’ (Letunic et al., 2009). The full-length genomic sequence (gPslSLY1) was isolated using the Universal Genome Walker Kit (Clontech, Palo Alto, CA, USA).

DNA and RNA extraction, and qPCR assays

Genomic DNA was extracted from young EG plum leaves according to the DNeasy Plant Mini Kit (Qiagen, Germantown, MD, USA). Total RNA extraction, DNase treatment, cDNA synthesis, and quantitative PCRs (qPCRs) were performed as described previously (El-Sharkawy et al., 2012). Gene-specific primers were designed using Primer Express (v3.0, Applied Biosystems, Carlsbad, CA, USA) (primers #3–14, Supplementary Table S1). Three technical replicates from three biological replicates for each reaction were analyzed on an ABI PRISM 7900HT Sequence Detection System (Applied Biosystems, Foster City, CA, USA). Transcript abundance was quantified using standard curves for both target and reference genes [PslAct (EF585293) and AtAct (NM_121018)], as described previously (El-Sharkawy et al., 2012). The data were presented as an average of nine replicates (±SD).

Transient expression of PslSLY1–GFP fusion proteins in tobacco protoplasts

Truncated versions of PslSLY1 proteins were generated as described previously (El-Sharkawy et al., 2014b). ORFs containing independent N-terminal (PslSLY1_∆N), F-box (PslSLY1_∆F), and C-terminal (PslSLY1_∆C) deletions of PslSLY1 were generated by PCR using gene-specific primer pairs (primers #15–22, Supplementary Table S1). Full-length and truncated PslSLY1 ORFs were fused in-frame with the green fluorescent protein (GFP) into the pGreenII vector using the BamHI site and expressed under the control of the 35S promoter. Protoplasts used for transfection were obtained from suspension-cultured tobacco BY-2 cells. Protoplasts were transfected with the constructs and analyzed for GFP fluorescence by confocal microscopy as described previously (El-Sharkawy et al., 2009).

Yeast two-hybrid (Y2H) assay

Y2H assays were carried out using the Matchmaker Gold Yeast two-hybrid System (Clontech) as described previously (El-Sharkawy et al., 2014a). Full-length and truncated PslSLY1 ORFs were inserted into the BamHI–PstI site of the pGBKT7 bait vector (GAL4-binding domain; DBD). ASK1, PslGAI, PslRGL, and PslRGA cDNAs were fused into the NdeI–BamHI site of the pGADT7 prey vector (GAL4 activation domain; AD). The resuspended yeast were spread in 96-well plates containing DDO/X/A medium in the presence or absence of 100 μM GA₃.

Bimolecular fluorescence complementation (BiFC) assay

Full-length and truncated PslSLY1 ORFs were fused into the SalI–BamHI site of the pSAT1-N vector, containing N-terminal YFP (NY). ASK1 and PslRGL sequences were cloned into the SacII–BamHI site of the pSAT1-C vector, containing C-terminal YFP (CY). The different combinations of constructs encoding NY and CY at similar concentrations were mixed and then co-transfected into protoplasts obtained from suspension-cultured tobacco BY-2 cells, as previously described (El-Sharkawy et al., 2014a). Empty BiFC vectors (NY: gene/CY and NY/CY: gene) were used as negative controls. Protoplasts were analyzed for fluorescence by confocal microscopy.

Plant transformation

Full-length and truncated PslSLY1 ORFs were fused into the SpeI–BstBI site of the pCambia1305.1 binary vector and expressed under the control of the 35S promoter. The constructed vectors were transformed into Agrobacterium tumefaciens and introduced into Arabidopsis as described previously (El-Sharkawy et al., 2012). The full-length PslSLY1 ORF was introduced into the Arabidopsis WT Ler background; however, the full-length and truncated PslSLY1 ORFs were introduced into the Arabidopsis mutant sly1.10 (McGinnis et al., 2003). T₃ homozygous independent lines from each group were grown (24 plants/group/growth condition) under long-day (LD) conditions (16:8 h light/300 μmol m⁻² s⁻¹, 23 °C:18 °C, and 65% relative humidity) and treated or not treated with PAC (5 µM). Plant materials were frozen in liquid N₂ immediately after collection and stored at −80 °C until use.

Accession numbers

The sequence data used in this study can be found in the GenBank database under the following accession numbers: PslSLY1 (MT358313), gPslSLY1 (MT358314), PslGAI (KU845589), PslRGL (KU845592), and ASK1 (U70034).

Results and discussion

In silico characterization of PslSLY1

In line with our previous research into the factors regulating stone fruit growth (El-Sharkawy et al., 2012, 2014a, 2017), this study was conducted to elucidate the function of an SLY1 F-box protein in the plum cultivar, EG. The isolated gene, PslSLY1, contains an ORF of 598 bp encoding a protein of 186 amino acids with a calculated molecular mass of 20.7 kDa and an isoelectric point of 8.44. The genomic gPslSLY1 sequence is 2159 bp in length, comprising a 1245 bp promoter sequence that carries several predicted GA and auxin cis-acting regulatory elements presumed to be involved in the regulation of PslSLY1 transcription (Supplementary Table S2). As with the majority of previously identified SLY1/GID2 gene family members (McGinnis et al., 2003; Dill et al., 2004), the plum gene is encoded by an intronless ORF. One exception is the hexaploid wheat, where the TaGID2 genes display two exons interrupted by one intron (Lou et al., 2016). Most characterized plants have only one GA-specific F-box protein that belongs to the SLY1/GID2 subfamily (McGinnis et al., 2003; Sasaki et al., 2003); however, recent studies have demonstrated the presence of two biologically functional SLY1/GID2 paralogs in grape (Acheampong et al., 2015). Sequence data mining in the genomes of Prunus species, including P. persica, P. avium, P. armeniaca, P. yedoensis, and P. dulcis (https://www.rosaceae.org), indicated that all Prunus species have only one SLY1/GID2 homolog. One exception is the ornamental flowering cherry (P. yedoensis) that has two putative paralogs.

Multiple alignments of PslSLY1 with other reported SLY1/GID2 sequences from different species indicated a high sequence divergence (32–87% similarity), largely due to extensive deviation in the N- and C-terminal domains outside the typical F-box region (Fig. 1). However, several signature structural elements commonly associated with the F-box subunit protein of SCF ubiquitin ligases were identified (McGinnis et al., 2003; Sasaki et al., 2003). The predicted amino acid sequence of PslSLY1 begins with an N-terminal domain of unknown function that displays striking sequence dissimilarity, in terms of length (61–94 amino acids in monocots and 32–48 amino acids in dicots) and homology (37–70% and 26–72% similarity in monocots and dicots, respectively), to other SLY1-like proteins. This is followed by the distinctive F-box sequence (46 amino acids) that has a greater sequence similarity (63–100%) to other SLY1 homologs. Finally, similarly to the N-terminal region, the C-terminal region displays a substantial variability in its predicted protein length (73–134 amino acids) and homology (29–86% similarity) to other SLY1-like proteins. The C-terminal region, however, has two short conserved domains, GGF (33–37 amino acids; 56–91% similarity) and LSL (26 amino acids; 66–100% similarity) that suggest functionality for this domain.

Fig. 1.

Amino acid sequence alignment of PslSLY1 with closely related sequences from M. domestica, A. thaliana, S. lycopersicum, Z. mays, and O. sativa, using ClustalX. Conserved residues are shaded in black. Dark gray shading indicates similar residues in six out of seven of the sequences, and light gray shading indicates similar residues in at least four of the sequences. The three main domains, N-terminus, F-box, and C-terminus are indicated, and the conserved GGF and LSL motifs are outlined.

The phylogenetic tree separated the GA F-box proteins into two subfamilies, SLY1/GID2 and SNE. The SLY1/GID2 subfamily can be further divided into two classes based on angiosperm monocots and eudicots (Supplementary Fig. S1). PslSLY1 has strong sequence identity to SLY1/GID2 proteins from other members of the family Rosaceae. The relationship was more distant with Arabidopsis, grape, and tomato SLY1/GID2 homologs, indicating that the evolutionary relationship is consistent with taxonomic distance.

Characterization of PslSLY1 domains

All SLY1/GID2 genes are highly conserved, with predicted proteins that maintain three typical domains generally associated with the F-box subunit of the SCF E3 ubiquitin ligase complex (Fig. 1). To determine the role of each domain in the protein function, we investigated the subcellular and biochemical characteristics of full-length ORFs and ORFs displaying a deletion in each of the three PslSLY1 domains. The complete protein is referred to as PslSLY1. PslSLY1 protein lacking the N-terminal, F-box, or C-terminal domains are labeled PslSLY1_ΔN, PslSLY1_ΔF, or PslSLY1_ΔC, respectively.

Subcellular localization of PslSLY1 and its truncated derivatives

An in silico analysis of the amino acid sequences of the SLY1/GID2 family failed to identify any motifs that can signify the localization of the protein in the plant cell. The SignalP Server, for instance, predicted that the PslSLY1 is a non-transmembrane, cytoplasmic protein. Fluorescence microscopy illustrated that control tobacco BY-2 cells transformed with the GFP gene alone showed ubiquitous fluorescence, following the predicted cytosolic localization of the GFP proteins (Fig. 2A). However, a full-length PslSLY1–GFP fusion protein localized exclusively to the nucleus (Fig. 2B), suggesting that PslSLY1 is capable of redirecting the GFP signal from the cytosol to the nucleus. This is consistent with its primary function as a component of the SCF E3 ubiquitin ligase complex that targets the DELLA proteins for degradation within the plant cell nucleus (Dill et al., 2004). To investigate the structural role of PslSLY1 domains, we explored the basis of localization of PslSLY1 proteins using independent domain removal. The elimination of the N-terminal domain did not affect the nuclear location of PslSLY1_ΔN protein (Fig. 2C). In contrast, the eradication of the F-box and C-terminal domain produced a dramatic disruption of protein localization. In PslSLY1_ΔF−GFP chimeras, a small proportion of the signal was visualized in the nucleus or cytoplasm; however, the majority of the chimeric GFP protein was detected on the exterior side of the nuclear membrane (Fig. 2D). Likewise, removal of the C-terminal domain in the PslSLY1_ΔC derivative noticeably altered the protein localization pattern (Fig. 2E). Without the C-terminus, the protein was no longer confined to the nucleus, as the GFP signal consistently covered the cytoplasm and nucleus. Treatment with GA before transfection did not alter the localization of the different PslSLY1 versions. These results indicate the fundamental role played by elements included in the F-box and C-terminal domain in ensuring appropriate localization of the PslSLY1 protein.

Fig. 2.

Subcellular localization of PslSLY1 and truncated proteins. The full-length ORF is referred to as PslSLY1. Truncated derivatives, including independent deletions of the N-terminal, F-box, and C-terminal domains, are referred to as PslSLY1_ΔN, PslSLY1_ΔF, and PslSLY1_ΔC, respectively. Constructs of either (A) GFP as control, (B) PslSLY1–GFP, (C) PslSLY1_ΔN–GFP, (D) PslSLY1_ΔF–GFP, or (E) PslSLY1_ΔC–GFP were transiently transformed into Nicotiana tabacum protoplasts. GFP fluorescence was visualized using confocal laser scanning microscopy. Fluorescence (left panel) and light micrograph (center panel) images are merged (right panel) to illustrate the different locations of the two proteins. Scale bar=10 µm. (This figure is available in color at JXB online.)

PslSLY1 is a component of an SCF complex that targets DELLA proteins

F-box subunits are assembled into an SCF complex (Zhang et al., 2019). If the PslSLY1 protein is part of a functional SCF^PslSLY1 complex, it should interact with downstream proteins involved in GA signaling (Dill et al., 2004; Fu et al., 2004). We tested the ability of PsISLY1 to interact with the Arabidopsis Skp1 protein (ASK1) using the Y2H system in GA-free or GA-containing media. Consistent with SLY1/GID2 F-box proteins, PslSLY1 was effectively capable of interacting with ASK1. The binding results suggested that GA is not essential for establishing a stable SCF^PslSLY1 complex in yeast (Fig. 3A). We further assessed the interaction properties of PslSLY1 with DELLA repressors to define whether DELLA proteins are prospective substrates of the SCF^PslSLY1 E3 ubiquitin ligase. The well-characterized DELLA proteins selected for this experiment were the plum PslGAI, PslRGL, and PslRGA that represent three subclades of DELLA proteins and play overlapping and distinct roles in mediating GA responses during plum fruit development (El-Sharkawy et al., 2017). The analysis revealed that PslSLY1 had an analogous binding capacity in a GA-independent manner for each of the PslDELLAs (Fig. 3A).

Fig. 3.

(A) Differences in interactions between PslSLY1–ASK1 and –PslDELLAs, and (B) PslSLY1 truncated derivatives (PslSLY1_ΔN, PslSLY1_ΔF, and PslSLY1_ΔC) with ASK1 and PslRGL using a yeast system. Y2H assays of the interactions were performed using PslSLY1 or derivatives as bait in yeast strain Y2HGold and ASK1 or PslDELLAs as prey in the Y187 strain. The mated yeast were grown in 96-well plates containing DDO/X/A selective medium in the presence or absence of 100 μM GA₃. All experiments were repeated at least three times. (This figure is available in color at JXB online.)

To verify the role of each PslSLY1 domain in the binding selectivity of the diverse interacting proteins, the truncated PslSLY1 ORFs were tested for interaction with ASK1 and PslRGL in the presence or absence of GA using the Y2H system (Fig. 3B). Relative to the full-length control PslSLY1, the exclusion of the N-terminal domain did not strikingly disrupt the GA-independent interaction capacity of PslSLY1_ΔN with ASK1 or PslRGL. In contrast, the deletion of either the F-box or the C-terminal domains altered protein interactions. The omission of the F-box domain hindered the interaction between the PslSLY1_ΔF protein and ASK1, and diminished the capacity for interaction with PslRGL. The PslSLY1_ΔC ORF was able to interact with ASK1, although at considerably weaker levels; however, the lack of the C-terminal domain inhibited the interaction with PslRGL. To provide further evidence for yeast results, we visualized the interaction pattern of different PslSLY1 versions using the BiFC approach (Fig. 4). The YFP fluorescence signal triggered by interactions with ASK1 was only detected with PslSLY1, PslSLY1_ΔN, and PslSLY1_ΔC derivatives. Fluorescence with PslRGL was only observed with the PslSLY1, PslSLY1_ΔN, and PslSLY1_ΔF derivatives. It has been shown that the F-box domain functions as a protein–protein interaction site during the establishment of the SCF E3 ubiquitin ligase complex and the C-terminal domain is essential for binding to the DELLA repressor substrate (Dill et al., 2004; Fu et al., 2004; Gomi et al., 2004; Ariizumi et al., 2011). The data outlined above demonstrate that PslSLY1 is a functional F-box component that can form an SCF^PslSLY1 ubiquitin ligase complex and putatively target DELLA proteins for degradation. The loss of the N-terminal domain had minor consequences for ligand–complex formation and function.

Fig. 4.

BiFC visualization of PslSLY1 and truncated derivative (PslSLY1_ΔN, PslSLY1_ΔF, and PslSLY1_ΔC) interactions with ASK1 and PslRGL. PslSLY1 and its derivatives were fused with the N-terminus (NY) of YFP. ASK1 and PslRGL were fused with the C-terminus (CY) of YFP. Different combinations of NY and CY constructs were transiently co-expressed in tobacco protoplasts. NLS–mCherry was included in each transfection to highlight the location of the nucleus. The merged image is a digital merge of bright field and fluorescent images to illustrate the interaction location. Scale bars=10 μm. (This figure is available in color at JXB online.)

GAs coordinate a wide range of systems through various response regulation mechanisms, including biosynthesis, metabolism, and signaling (Yasumura et al., 2007; Yamaguchi, 2008). In planta, GID1 receptors and DELLA repressors interact via conserved N-terminal DELLA regulatory domains after GID1 binding to the active GA. The formation of the GA–GID1–DELLA co-receptor complex leads to conformational changes in the DELLA protein that facilitates recognition between the GRAS domain of the DELLA protein and the SLY1/GID2 C-terminal domain (Dill et al., 2004; Fu et al., 2004; Hirano et al., 2010; Ariizumi et al., 2011; Lou et al., 2016). The DELLA proteins are subsequently degraded via the SCF^PslSLY1 ubiquitin-dependent proteolysis pathway. Although this mechanism is widespread, studies have demonstrated an alternative mechanism via a GA–GID1–DELLA co-receptor, proteolysis-independent pathway (Ariizumi et al., 2008, 2013).

Among the three tested PslDELLA proteins, PslRGA represents a new branch of DELLA protein evolution, with no representative in Arabidopsis (El-Sharkawy et al., 2017). PslRGA displays considerable divergence in the essential N-terminal DELLA regulatory domain, including the key DELLA, LExLE, and VHYNP motifs important for GA-dependent GID1–DELLA protein interactions. The lack of a conserved DELLA regulatory domain restricted GA–GID1–DELLA interactions in the Y2H system (Griffiths et al., 2006; Hirano et al., 2010; El-Sharkawy et al., 2017). PslRGA also appears to encode a GA-insensitive DELLA protein since PslRGA−GFP fluorescence showed resistance to the destabilizing impact of GA in Arabidopsis roots (El-Sharkawy et al., 2017). Despite this, the GRAS domain containing PslRGA displayed a strong capacity for interacting with PslSLY1 in vitro. It appears that the first step of GA–GID1–DELLA co-receptor complex assembly is imperative for the transition to DELLA suppression via the proteolysis-dependent degradation pathway.

PslSLY1 expression during fruit ontogeny

SLY1/GID2 family members have been reported to be differentially regulated in different tissues (Acheampong et al., 2015; Liu et al., 2016; Lou et al., 2016). PslSLY1 mRNA accumulation varied in an organ- and developmental stage-dependent manner (Fig. 5). This suggests that PslSLY1 is transcriptionally regulated, as proposed for its ortholog in Arabidopsis (McGinnis et al., 2003). PslSLY1 transcripts were moderately abundant in flower buds (~4 d before bloom) followed by a marked decline at full bloom. After fertilization, PslSLY1 mRNA peaked at ~7 days after bloom (DAB) and then declined during early fruit formation (~10–22 DAB). GA plays a role in diverse biological and physiological processes, especially flowering and fruit initiation (Blázquez et al., 1998; Goto and Pharis, 1999; Vivian-Smith and Koltunow, 1999; Davies, 2004; Bernier and Perilleux, 2005). It is also involved in coordinating the cell division, expansion, and embryo development during the fruit set phase (Serrani et al., 2007; de Jong et al., 2009, El-Sharkawy et al., 2014a). The accumulation profile of PslSLY1 transcripts in floral buds and directly after fertilization indicates a role in floral meristem development and flower bud initiation, as well as in the GA regulation of the flower-to-fruit transition program.

Fig. 5.

Steady-state transcript levels of PslSLY1 mRNA assessed by qPCR in EG flowers and fruit throughout development. Expression was determined in the whole fruit during stages S1 and S2 of fruit development. During S3 and S4, expression was determined in pulp (black filled bars) and in seeds (gray filled bars). Transcript accumulation was determined using qPCR on three biological replicates. Error bars represent the SD of the mean. Standard curves were used to calculate the number of molecules of the target gene per sample, which were then normalized relative to PslAct expression. The y-axis refers to the relative expression. The x-axis represents the developmental stages indicated by the number of days after bloom (DAB). (This figure is available in color at JXB online.)

Stone fruit (Prunus spp.), including plums, display a characteristic double sigmoid growth pattern during fruit development, with four distinct stages, S1–S4 (El-Sharkawy et al., 2014a). The PslSLY1 transcription profile suggested its contribution to the GA signaling network that modulates GA responses, regulating fruit growth during the early S1–S2 stages of fruit development. PslSLY1 mRNA declined gradually from early S3 stage (~57 DAB) throughout the remainder of fruit development, reaching very low levels by the post-climacteric phase (~83 DAB). In plum, transcript profiles of GA-related genes are usually aligned with their GA regulation model. When GA is abundant, the GA proteins that act as activators (i.e. GA3ox, GA20ox, GID1, and SLY1/GID2) and repressors (i.e. GA2ox and DELLA) of GA action are subject to negative feedback and positive feed-forward regulation by GA, respectively (Griffiths et al., 2006; Yasumura et al., 2007; Yamaguchi, 2008; Harberd et al., 2009). The accumulation pattern of bioactive GAs during plum fruit development (i.e. in the same developmental stages adopted in the current study) proposed that both GA₁ and GA₄ are directly implicated in S3 events via elevating the rate of cell expansion and increasing fruit size (El-Sharkawy et al., 2014a). The data indicate that PslSLY1 is an element of the GA signaling network during the S3 stage of fruit development and during the ripening phase (S4) in a distinctive GA-dependent manner. This argument is supported by high levels of GA₁ detected through ripening (El-Sharkawy et al., 2014a).

GA regulation of PslSLY1

The abundance of PslSLY1 mRNA in organs with low levels of GA might be due to feedback inhibition of expression of the gene during dynamic GA signaling. To examine this, PslSLY1 mRNA accumulation profiles were evaluated in fruits supplemented with GA or the GA biosynthesis inhibitor PAC. GA treatment markedly reduced PslSLY1 mRNA levels. Compared with the controls, PslSLY1 mRNA levels declined by ~74% in GA-treated samples collected 9 d after treatment (Supplementary Fig. S2a). In contrast, treatment with PAC steadily increased PslSLY1 transcript abundance to ~5-fold over the same treatment period (Supplementary Fig. S2b). These results suggest that in plum, the negative regulation of PslSLY1 mRNA abundance is part of the GA response in plum.

In planta function of PslSLY1

The results confirmed that PslSLY1 could act as an effective F-box protein that is able to interact with the ASK1 subunit of an SCF–ubiquitin ligase complex and potentially targeting PslDELLA repressors for ubiquitination in a GA-independent manner. In Arabidopsis and other species, it has been shown that GA relieves growth repression by promoting interactions between GID1s and DELLAs, triggering the ubiquitin-dependent proteolysis of the DELLA proteins via the SCF^SLY1/GID2 and the 26S proteasome (Yasumura et al., 2007; Ariizumi et al., 2008).

To test whether PslSLY1 protein maintains similar GA functions in planta, several independent Arabidopsis transgenic lines overexpressing PslSLY1 (SLY1/w) were generated. Plants from two independent T₃ homozygous lines (L2 and L9) were chosen for further analysis (Fig. 6A). Theoretically, ectopically expressing PslSLY1 should accelerate the GA responses and trigger overall plant growth. Accordingly, SLY1/w transgenic plants were phenotypically characterized for a range of GA-related traits. As expected, overexpression of PslSLY1 in WT Arabidopsis produced dramatic phenotypic changes consistent with elevated GA responsiveness. The phenotypic variations from the WT became more distinguishable in plants treated with the GA biosynthesis inhibitor PAC. The accumulation of transcripts involved in GA metabolism were assessed, including genes encoding enzymes for GA biosynthesis (GA3ox and GA20ox) and enzymes for GA inactivation (GA2ox) (Thomas et al., 1999; Xu et al., 1999; Sponsel and Hedden, 2004; Sun and Gubler, 2004). Consistent with the GA regulation model (Yamaguchi, 2008), the accumulation of AtGA3ox1 and AtGA20ox1 was dramatically reduced by ~61% and ~51% relative to the WT, respectively, while AtGA2ox8 steadily increased by ~2-fold in SLY1/w plants (Fig. 6B). Overexpression of PslSLY1 in WT Arabidopsis led to extensive disturbances in overall growth and development consistent with altered GA signaling (Fig. 6C). Analyses of the transgene in several transgenic lines showed that the level of PslSLY1 mRNA was usually correlated with the strength of the phenotype (i.e. the higher the level of expression of the transgene the more pronounced the effect on growth). SLY1/w plants manifested significantly higher germination rates, enhanced vegetative growth associated with increases in the number and length of internodes, precocious transition to flowering, longer pistil/stamen lengths, and larger siliques with earlier silique dehiscence (Table 1). Treatment of the SLY1/w plants with PAC (Fig. 6D) demonstrated that ectopic PslSLY1 expression enhanced the resistance to the growth impact of PAC, even for traits that presented non-significant contrasts with WT plants under optimal conditions (Table 1).

Table 1.

Vegetative and reproductive characteristics of WT (Ler) Arabidopsis plants overexpressing the PslSLY1 gene as shown in Fig. 6

Character	Condition	WT	SLY1/w
Germination rate (%)	C	90.0±1.9	98.0±0.8**
Germination time (d)	C	2.7±0.4	2.4±0.2ns
Stem height (cm)a	C	32.7±1.5	44.5±2.2**
	PAC	8.8±0.5	22.4±1.9**
Internode length (cm)a,b	C	2.4±0.2	3.0±0.3*
	PAC	0.6±0.3	1.4±0.2**
No. of internodesa	C	13.4±0.2	14.0±0.3*
	PAC	14.3±0.4	16.5±0.2**
Transition to flowering (d)c	C	28.3±0.6	26.2±0.8*
	PAC	45.3±1	36.6±0.9**
Pistil length (mm)c,d	C	2.21±0.08	2.55±0.09**
	PAC	1.52±0.07	1.98±0.06**
Stamen length (mm)c,d	C	2.16±0.04	2.39±0.07**
	PAC	1.26±0.08	1.78±0.09**
Silique initiation time (d)e	C	2.4±0.2	2.1±0.2ns
	PAC	3.7±0.3	2.8±0.3*
Silique maturation time (d)a	C	13±0.6	9.2±0.4**
	PAC	21.4±0.9	12.1±0.7**
Silique length (mm)e,f	C	10.29±0.11	12.59±0.2**
	PAC	4.56±0.07	8.52±0.12**
No. of seeds/silique e,f	C	42.7±3.2	46.3±2.6ns
	PAC	13.0±1.7	29.8±3.3**

Plants were grown under an LD photoperiod and divided into two groups: control and treated with PAC (5 μM). The measurements are the means (±SD) of ~12 plants. Statistically significant differences from the WT are indicated by * and ** for the probability levels P<0.05 and P<0.01, respectively. ns, non-significant (P>0.05).

^a The measurements were taken from adult plants and have ~10% shattered siliques.

^b The first vegetative internode in the main inflorescence.

^c Flowering time was scored upon the emergence of ~10% of flowers.

^d The measurements are the means (±SD) of 12 flowers.

^e The measurements were scored upon the emergence of 10% of siliques.

^f The measurements are the means (±SD) of 12 siliques.

Fig. 6.

(A) PslSLY1 transgene accumulation and (B) accumulation levels of the GA-related mRNAs in WT and two transgenic lines (SLY1/w-L2 and SLY1/w-L9), overexpressing the PslSLY1 sequence. Transcript accumulation was determined using qPCR on three biological replicates. Error bars represent the SD of the mean. Standard curves were used to calculate the number of molecules of the target gene per sample, which were then normalized relative to AtAct expression. The y-axis refers to the relative expression. The x-axis represents the Arabidopsis plants. ND, non-detectable. Aerial portions of the WT and SLY1/w-L2 transgenic mutant grown under LD conditions/non-treated (C) or /treated (D) with PAC (5 μM). Scale bars=10 cm.

Overexpression of PslSLY1 rescues the sly1.10 mutant

To further verify that PslSLY1 can have biologically similar functions to its orthologs, we tested its capability to overcome the loss of function of the endogenous AtSLY1 in the GA-insensitive sly1.10 mutant (McGinnis et al., 2003). Truncated PslSLY1 derivatives were introduced into the sly1.10 loss-of-function mutant line to define the strength of the GA response under the independent loss of the major PslSLY1 domains. The different PslSLY1 lines were phenotypically characterized under standard growth conditions.

Several biochemical and molecular studies illustrated the crucial role played by GA in coordinating the germination rate and the time necessary to establish germination (Lee et al., 2002; Tyler et al., 2004; Ariizumi et al., 2008, 2011). The sly1.10 mutant has a prolonged seed dormancy and germinates at a much slower rate (Ariizumi and Steber, 2007). The germination of seeds resulting from sly1.10 plants overexpressing PslSLY1_ΔF (SLY1_ΔF/s) and PslSLY1_ΔC (SLY1_ΔC/s) behaved similarly to the sly1.10 mutant, via displaying poor germination rates and duration, either with or without the addition of exogenous GA (Table 2; Supplementary Fig. S3). In contrast, the introduction of PslSLY1 or PslSLY1_ΔN ORFs restored WT germination efficiencies in the transgenic plants.

Table 2.

Vegetative and reproductive characteristics of Arabidopsis WT (Ler), sly1.10 mutant, and sly1.10 plants overexpressing PslSLY1, PslSLY1_ΔN, PslSLY1_ΔF, and PslSLY1_ΔC ORFs as shown in Fig. 7, and Supplementary Figs S3 and S4.

Character	sly1.10	SLY1/s	SLY1ΔN/s	SLY1ΔF/s	SLY1ΔC/s	WT
Germination rate (%)	11.5±2.5	80.7±1.4**	83.2±1.9**	9.1±1.1 NS	7.9±1.6ns	90.0±1.9
Germination time (d)	21.9±0.3	3.0±0.2**	2.9±0.3**	23.3±0.3ns	20.0±0.2ns	2.7±0.4
Stem height (cm)a	5.7±0.3	27.8±1.3**	26.4±0.8**	6.1±0.4ns	5.5±0.5ns	32.7±1.5
Internode length (cm)a,b	0.8±0.2	2.1±0.4**	2.0±0.7**	0.8±0.3ns	0.7±0.1ns	2.4±0.2
No. of internodesa	7.5±0.4	13.3±0.3**	13.1±0.4**	7.4±0.3ns	7.8±0.6ns	13.4±0.2
Transition to flowering (d)c	47.7±1.2	29.2±0.9**	30.1±1.1**	50.0±1.7ns	49.2±0.9ns	28.3±0.9
No. of flowers/inflorescence	16.0±1.6	48.4±5.5**	49.5±5.4**	15.1±1.5ns	17.2±2.2ns	52.3±10.5
Pistil length (mm)c,d	1.00±0.05	2.22±0.13**	2.19±0.09**	0.90±0.07ns	1.11±0.06ns	2.21±0.08
Stamen length (mm)c,d	0.92±0.03	2.15±0.05**	2.11±0.08**	0.84±0.05ns	0.95±0.02ns	2.16±0.04
Silique initiation time (d)e	3.9±0.5	2.5±0.2**	2.7±0.3**	4.1±0.6ns	3.7±0.4ns	2.4±0.2
Silique maturation time (d)a	29.0±1.6	13.7±0.9**	13.3±0.7**	31.3±1.0ns	30.8±0.9ns	13±0.6
Silique length (mm)e,f	3.91±0.06	9.96±0.15**	9.67±0.17**	4.06±0.08ns	3.93±0.09ns	10.29±0.11
No. of seeds/siliquee,f	9.8±1.5	39.7±1.6**	40.1±2.7**	10.2±1.2ns	9.9±1.3ns	42.7±3.2

Plants were grown under an LD photoperiod and divided into two groups: control and treated with PAC (5 μM). The measurements are the means (±SD) of ~12 plants. Statistically significant differences from the WT are indicated by * and ** for the probability levels P<0.05 and P<0.01, respectively. ns, non-significant (P>0.05).

^a The measurements were taken from adult plantsand have ~10% shattered siliques.

^b The first vegetative internode in the main inflorescence.

^c Flowering time was scored upon the emergence of ~10% of flowers.

^d The measurements are the means (±SD) of 12 flowers.

^e The measurements were scored upon the emergence of 10% of siliques.

^f The measurements are the means (±SD) of 12 siliques.

Several independent transgenic events were produced and one representative from each transformation, conferring a maximum transgene level, was selected for further studies (Fig. 7A). To estimate whether overexpression of PslSLY1 and its derivatives can recover the disrupted GA responsiveness in the mutant, the accumulation of GA metabolism genes was assessed in controls (WT and sly1.10 mutant) and transformed plants. Relative to the sly1.10 mutant line, sly1.10 plants transformed with PslSLY1_ΔF (SLY1_ΔF/s) and PslSLY1_ΔC (SLY1_ΔC/s) derivatives showed no meaningful differences in the transcription levels of the three genes involved in GA metabolism (Fig. 7B). However, ectopic expression of the full-length PslSLY1 (SLY1/s) and the PslSLY1_ΔN (SLY1_ΔN/s) led to a dramatic restoration of the GA metabolic pathway. It remarkably reduced transcript levels of AtGA3ox1 by ~89% and ~82%, and of AtGA20ox1 by ~87% and ~82%, respectively. In contrast, AtGA2ox8 mRNA increased by ~20- and ~18-fold in SLY1/s and SLY1_ΔN/s plants, respectively.

Fig. 7.

(A) Transgene accumulation and (B) expression levels of the different GA-related mRNAs in the WT, sly1.10 mutant, and sly1.10 plants overexpressing PslSLY1 (SLY1/s) and PslSLY1_ΔN (SLY1_ΔN/s), PslSLY1_ΔF (SLY1_ΔF/s) and PslSLY1_ΔC (SLY1_ΔC/s). Transcript accumulation was determined using qPCR on three biological replicates. Other details are as described in Fig. 6. (C) Aerial portions of sly1.10 mutant and the transgenic lines were grown under LD (16 h day/8 h night) conditions. ND, non-detectable. Scale bars=10 cm.

The sly1.10 mutant plants have reduced growth compared with WT plants (Table 2). Consistent with the results obtained from subcellular localization and protein–protein interactions, PslSLY1_ΔF and PslSLY1_ΔC plants displayed phenotypes consistent with the repression of GA action, similar to the sly1.10 mutant phenotype (Table 2; Fig. 7C). These results validated that the F-box and C-terminal domains are vital for proper SLY1/GID2 protein function (Dill et al., 2004; Fu et al., 2004; Gomi et al., 2004; Ariizumi et al., 2011). Introducing PslSLY1 and PslSLY1_ΔN ORFs into sly1.10 visibly restored growth via re-establishing the typical plant height, mainly due to enhancements in internode length and number. Relative to the control sly1.10 mutant, SLY1/s and SLY1_ΔN/s plants demonstrated significant enrichment in their overall heights by ~5- and ~4.8-fold, respectively (Table 2; Fig. 7C).

GAs are committed to the developmental events leading to reproductive competence (Cheng et al., 2004; Dill et al., 2004; Achard et al., 2007; Ariizumi et al., 2011; Davière and Achard, 2013). The sly1.10 mutant showed extensive deficiencies in all aspects of reproductive growth, a pattern that has not been altered in SLY1_ΔF/s and SLY1_ΔC/s plants (Table 2). Overexpression of PslSLY1 and PslSLY1_ΔN recovered reproductive growth to levels comparable with WT plants. Relative to the sly1.10 mutant control, SLY1/s and SLY1ΔN/s plants exhibited a visible increase in the number of flowers per inflorescence, estimated to be ~3-fold (Table 2). Moreover, SLY1/s and SLY1_ΔN/s plants began flowering ~18.5 d earlier than sly1.10 plants, with no apparent difference from WT plants. Although the sly1.10 mutant displayed generally smaller flowers, it shared a crucial character with the WT by securing a coordinated flower structure, in terms of a similar pistil/filament length ratio that ensures correct fertilization and seed set. SLY1/s and SLY1_ΔN/s plants set larger flowers with accelerated stamen and pistil growth that maintained the proper floral patterning (Table 2; Supplementary Fig. S4). The improvement in various flowering characteristics in SLY1/s and SLY1_ΔN/s resulted in an enhancement of reproductive traits, relative to the sly1.10 mutant (Table 2; Supplementary Fig. S4). In SLY1/s and SLY1_ΔN/s plants, the conclusions to silique emergence as well as silique maturation were similar to those of the WT and noticeably shorter than those of the control sly1.10. Despite the appropriate flower structure of the sly1.10 mutant, it showed a notable reduction in seed number. The overall reproductive growth deficiency suggested a strong involvement of compromised GA signaling (Dill et al., 2004; Ariizumi et al., 2008). All disorders in silique length and fertility observed in sly1.10 mutant plants were considerably recovered to WT levels in SLY1/s and SLY1_ΔN/s plants. It has been reported that GA responsiveness is perturbed in sly1.10 mutant, as the GA applications do not result in changes in growth habit (McGinnis et al., 2003; Dill et al., 2004). Increasing the levels of GA receptors (GID1 overexpression) or restoring GA signaling machinery (SLY1 overexpression) in sly1.10 plants re-establishes GA responsiveness and triggers plant growth via activating the suppression of DELLA repressors (Yasumura et al., 2007; Ariizumi et al., 2008).

PslSLY1 is an active F-box protein involved in GA signaling in plum that can efficiently interact with DELLA proteins in vitro. As in the Arabidopsis GA signaling model, the formation of the co-receptor GA–PslGID1–PslDELLA complex allows SCF^PslSLY1 to bind to and ubiquitinate PslDELLA proteins (Hirano et al., 2008, Ariizumi et al., 2011). This model predicts that SLY1 should form a complex that includes not only DELLA but also GID1 proteins. The results reported here present additional confirmation of the indispensable role of GA in organizing fruit development. They provide a foundation for advancing research into the precise nature of the complex development of fruit.

Supplementary data

The following supplementary data are available at JXB online.

Table S1. The oligonucleotide primers used for the study.

Table S2. PslSLY1 promoter analysis using the PLACE website.

Fig. S1. Evolutionary relationships in the GA signal transduction factor SLY1/GID2 and SNE gene family.

Fig. S2. PslSLY1 transcript regulation in response to the application of GA or the GA biosynthesis inhibitor PAC.

Fig. S3. Effect of overexpressing the full-length PslSLY1 and truncated derivatives in the rescue of germination defects of the sly1.10 mutant.

Fig. S4. Effect of overexpressing the full-length PslSLY1 and truncated derivatives on recovering floral organ and silique growth of the sly1.10 mutant.

Abbreviations

Abbreviations
 
• BiFC

  bimolecular fluorescence complementation

 
• DAB

  days after bloom

 
• EG

  Early Golden

 
• GA

  gibberellin

 
• WT

  wild type

 
• Y2H

  yeast two-hybrid

Acknowledgments

We thank Dr Camille M. Steber for providing the sly1.10Arabidopsis mutant, Dr Stanton Gelvin for providing the EYFP vectors, and Dr Brian Jones for critical review of the manuscript.

Author contributions

IE designed the experiments, plant transformation, phenotypical, and molecular characterization of transformed Arabidopsis, data analysis, supervised the study, generated the original version of the manuscript, and produced the final version of the manuscript. AI designed the experiment, protein–protein interaction assay, generated Y2H and BiFC constructs, and reviewed and edited the original version of the manuscript. AD performed subcellular localization, Y2H, and BiFC experiments. WE performed plant transformation, and phenotypical and molecular characterization of transformed Arabidopsis, and associated data analysis. SJ designed the experiment, provided plum materials, RNA/DNA and qPCR assay, and reviewed and edited the original and final version of the manuscript. SS designed the experiment, and reviewed and edited the original version of the manuscript.

Data availability

All data supporting the findings of this study are available within the paper and within its supplementary data published online.