Plum Fruit Development Occurs via Gibberellin–Sensitive and –Insensitive DELLA Repressors

Fruit growth depends on highly coordinated hormonal activities. The phytohormone gibberellin (GA) promotes growth by triggering degradation of the growth-repressing DELLA proteins; however, the extent to which such proteins contribute to GA-mediated fruit development remains to be clarified. Three new plum genes encoding DELLA proteins, PslGAI, PslRGL and PslRGA were isolated and functionally characterized. Analysis of expression profile during fruit development suggested that PslDELLA are transcriptionally regulated during flower and fruit ontogeny with potential positive regulation by GA and ethylene, depending on organ and developmental stage. PslGAI and PslRGL deduced proteins contain all domains present in typical DELLA proteins. However, PslRGA exhibited a degenerated DELLA domain and subsequently lacks in GID1–DELLA interaction property. PslDELLA–overexpression in WT Arabidopsis caused dramatic disruption in overall growth including root length, stem elongation, plant architecture, flower structure, fertility, and considerable retardation in development due to dramatic distortion in GA-metabolic pathway. GA treatment enhanced PslGAI/PslRGL interaction with PslGID1 receptors, causing protein destabilization and relief of growth-restraining effect. By contrast, PslRGA protein was not degraded by GA due to its inability to interact with PslGID1. Relative to other PslDELLA–mutants, PslRGA–plants displayed stronger constitutive repressive growth that was irreversible by GA application. The present results describe additional complexities in GA-signalling during plum fruit development, which may be particularly important to optimize successful reproductive growth.


Introduction
Fruit development is a multiphase process that requires a tight coordination of molecular, biochemical and structural elements. The series of modifications that control the transition of fruit growth through consequent developmental stages involve many distinctive metabolic PLOS ONE | DOI: 10.1371/journal.pone.0169440 January 11, 2017 1 / 31 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 pathways. In recent years, many molecular and genetic mechanisms underlying the action of phytohormones in fruit development have been identified, uncovering the complexity of this regulatory network [1][2][3][4]. Collectively, hormone application, endogenous hormone quantification and genetic studies support the hypothesis that fruit development is largely coordinated by hormonal interplay. Gibberellin (GA) is an essential hormone involved in diverse biological processes, leading to correct plant growth and development [5][6][7][8]. In tree fruit species, the proper establishment of reproductive growth is dependent on coordinated levels of GA at the appropriate developmental stages [9][10][11]. Application of GA resulted in visible improvement of fruit quality traits in terms of size, weight and many other characteristics [12][13]. Conversely, mutant fruits with inadequate quantities of GA exhibited a series of distortions in floral development and general reproductive growth events [1-2, 9, 14]. Although the potential impact of GA in coordinating fruit development processes has already been acknowledged [15][16][17][18], the mechanism by which these effects are achieved is still largely unknown. This may be due to the diversity of cross-talk between GA and other hormones, which are often species/ organ/developmental stage-dependent [19]. Several lines of evidence point out the essential role of GA in coordinating reproductive growth. In flowering plants, GA specifies the site of floral primordium initiation, and acts with homeotic genes to ensure proper floral organogenesis and patterning [20][21][22][23]. Molecular and genetic studies highlighted the pivotal contribution of GA during fruit-set, the term given to the onset of rapid cell division necessary for early embryo development and fruiting structure enlargement [1][2]. The transition of ovary into fruit, initiated upon successful fertilization, activates GA pathway in the ovules that acts with other hormones, particularly auxin and cytokinin, in triggering fruit-set program, thereby stimulating fruit growth [2,4,[24][25][26][27]. Previous studies have shown that the endogenous GA content readily increased along with the progression in fruit maturity and ripening [13,15,28]. These findings coupled with the stimulatory effect of exogenous GA in the fruit growth of several species suggested that GA is needed in mature fruiting tissues to allow fruit enlargement with potential involvement in ripening [1,[12][13][29][30][31]. On the other hand, the scarcity of bioactive GA during plum fruit growth caused serious developmental disorders, including growth retardation, disturbed flower patterning and limited fruit characteristics [9]. Insight into mechanisms of GA-regulated plant development has been manifested from research into GA-biosynthesis, -metabolism and -signalling pathways [19,32]. The major metabolic processes regulating GA-biosynthesis and -deactivation have been identified [33]. By contrast, the discovery of GA-receptors and downstream signalling components has been recently elucidated [34][35][36]. The central of GA-signalling are the DELLA proteins that are part of the wider GRAS family of regulatory proteins [37]. According to the relief of restraint model, DELLA proteins operate as growth-repressors and GA-mediated DELLA degradation is a critical step to overcome this restraint [38]. In agreement with their function as growthrepressors, lacking one or more of DELLA proteins within the plant elicited constitutive activation of GA-signalling pathway independent to the hormone presence in which the mutant plants exhibited GA-overdosed phenotype, including slender vegetative growth and parthenocarpic fruit development [39][40][41][42][43][44]. At low GA levels, DELLA proteins impair the activity of basic helix-loop-helix transcription factors by interacting with their DNA binding domain [45][46]. The binding of GA to its receptor GIBBERELLIN-INSENSITIVE DWARF1 (GID1) results in a conformational change that promotes interaction of GID1 with the DELLA domain of DELLA proteins [47][48][49][50]. The GA-GID1-DELLA complex is subsequently recognized by the SCF SLY1/GID2 E3 ubiquitin-ligase complex, which mediates ubiquitination of DELLA proteins. This ubiquitin mark destines the DELLA proteins for degradation via the 26S proteasome, thereby allowing growth by releasing their inhibitory interaction with GAdependent gene partners [45][46][51][52][53][54][55].
In plants, it is important to maintain optimal levels of hormone signalling to ensure normal growth. Disruption of this signalling pathway can dramatically impact plant development. Nevertheless, the severity of these phenotypic changes can vary within and among species [56][57][58][59][60][61]. For instance, the altered phenotype in some GA-deficient mutants can be easily reversed to normal by applying external GA, while others show unresponsive effect to GA treatment [62].
In the present study, three novel genes encoding DELLA proteins were isolated from Japanese plum cultivar Early Golden (Prunus salicina L.). To understand the potential involvement of various PslDELLA in fruit growth, their expression profile was assessed throughout fruit development. We next investigated PslDELLA function to provide evidence that the identified proteins are responsible for regulating the GA-responsiveness during fruit growth. Sequence analysis indicated that PslGAI and PslRGL deduced proteins contain all domains present in typical DELLA proteins; however, PslRGA lack the intact DELLA domain necessary for the GA-dependent interaction with GA-receptors, GID1. Despite this fact, PslRGA primary structure showed high similarity to the C-terminal portions of DELLA proteins, and phylogenetic and modelling structure classified it as a member of DELLA group. Analysis of yeast twohybrid (Y2H) and bimolecular fluorescence complementation (BiFC) assays indicated that PslGAI and PslRGL proteins are active repressor components that effectively interact with PslGID1 receptors in a GA-dependent manner. However, PslRGA was not able to form complex with PslGID1 proteins under different circumstances of GA-types or concentrations. Phenotypical analysis of transgenic Arabidopsis plants overexpressing each of PslDELLA confirmed the function of the three proteins as growth-repressors. Although PslGAI-and PslRGL-mutant plants were able to recover the normal growth by GA application, PslRGAplants exhibited constitutive inhibition of GA-signalling, overcoming the destabilization effect of GA. Finally, we provided several lines of evidence that PslRGA encode a strong stable DELLA protein independent of GA action and this was mainly due to critical substitutions occurring within the essential DELLA domain.

Plum tissues and treatments
Flowers and fruits from sequential developmental stages were harvested from Japanese plum cultivar Early Golden (EG) as described previously [64]. Since the seed is inseparable in S1 and S2 growth phases, the whole fruit tissue was used for RNA extraction, while in S3 and S4 stages the pulp tissue was carefully separated from the seed for RNA analysis. To evaluate the potential ethylene-dependent regulation of PslDELLA during plum fruit ripening, mature EG fruit (76 DAB) were harvested before autocatalytic ethylene production had risen, surface sterilized, and subjected to various treatments. These included propylene (1000 μl l -1 ), the ethyleneinhibitor 1-MCP (1 μl l -1 ) and water-dipped fruit were used as control. Fruit were sampled at different stages of ethylene production (non-climacteric, pre-climacteric, climacteric and postclimacteric), by assessing ethylene evolution. In 1-MCP treatment samples were collected at similar age to that of control fruit. In all cases, mixed tissues of at least twelve fruit (distributed into 3 biological replicates) at the same age or displaying a similar ethylene production were used for mRNA extraction and analysis. All samples were frozen in liquid nitrogen immediately after collection and stored at −80˚C.

Isolation and in silico analysis of PslDELLA sequences
Based on the sequence similarity among various DELLA cDNAs, a pair of degenerate primers (S1 Table) was designed in the conserved regions to amplify the plum orthologs from EG cDNA under stringent primer hybridization conditions. Fragments from several independent PCR reactions were cloned, sequenced and compared with database sequences using the BLAST program [65]. Extension of the partial cDNA clones were carried out using the 5'-and 3'-RACE kit (Invitrogen, Burlington, ON, Canada). Full-length amplification of cDNA sequences designated PslGAI, PslRGL and PslRGA was carried out using Platinum Taq DNA Polymerase High Fidelity, following the instructions provided by the manufacturer (Invitrogen). The names of the individual plum DELLA introduced here are not intended to imply functional homology to specific Arabidopsis DELLA protein. Since there is two different alleles of PslRGL and PslRGA (a & b), unless mentioned otherwise PslRGL and PslRGA will be always referred to PslRGLa and PslRGAa, respectively. The group of the three genes and proteins PslGAI, PslRGLa and PslRGAa are referred to as PslDELLA and PslDELLA, respectively. Alignment of predicted proteins was performed using ClustalX and the neighbor-joining tree was generated with MEGA5 [66]. Full-length genomic sequences were isolated using the Accu-Prime Pfx (Invitrogen). To determine the function of PslRGA sequence, mutated version of PslRGL and PslRGA designated PslRGL .MU and PslRGA .MU , respectively; were generated using the QuikChange site-directed mutagenesis kit (Stratagene, San Diego, CA, USA). Changes were generated within the N-terminal DELLA and TVHYNP motifs of PslRGL sequence to mimic PslRGA and conversely in PslRGA to simulate that of PslRGL sequence.

DNA, RNA extractions and qPCR assays
Genomic DNA was extracted from young plum leaves according to the DNeasy Plant Maxi Kit (Qiagen, Mississauga, ON, Canada). Total RNA extraction, DNase treatment, cDNA synthesis, and qPCR reactions were performed as described previously [9]. Gene-specific primers were designed using Primer Express (v3.0, Applied Biosystems, Carlsbad, CA, USA) (S1 Table). Three independent biological replicates for each reaction were run on an ABI PRISM 7900HT Sequence Detection System (Applied Biosystems) and each experiment was repeated three times. Transcript abundance was quantified using standard curves for both target and reference genes [PslAct (EF585293), AtAct (NM_121018)], which were generated from serial dilutions of PCR products from corresponding cDNAs. The data were present as an average ±SD.

Observation of GFP fluorescence
Full-length coding sequences of PslGAI, PslRGL, PslRGA, and the mutated PslRGL .MU and PslRGA .MU versions were fused in frame with the GFP into the pGreenII vector using the BamHI site and expressed under the control of the 35S promoter. For protoplasts assay, the different constructs were transfected into protoplasts from suspension cultured tobacco BY-2 cells exposed to different treatments that alter GA-response, including 100 μM GA 3 and/or 10 μM of GA-biosynthesis inhibitor paclobutrazol (PAC). Non-treated protoplasts were used as a mock control. For Arabidopsis assay, transgenic seeds independently expressing the different PslDELLA−GFP chimeric proteins (excluding PslRGL .MU ) were germinated in standard MS growth medium. After 5 days, the roots were observed initially for GFP fluorescence and then exposed to respective treatments as described above for 60, 120 and 240 minutes. After each time point, root samples were mounted on microscope slides and analyzed for GFP fluorescence using confocal microscopy as described previously [67]. All assays were repeated three times. model for solvent of Amber12 software package. The binding energy of the respective PslDELLA proteins to individual GA 3 -PslGID1 complex was then calculated through obtaining the electrostatic components of the thermodynamic cycle corresponding to a protein-protein binding event [68]. The electrostatic energies were calculated using the numerical Poisson-Boltzmann solver algorithm APBS software version 1.3 [69].

Bimolecular fluorescence complementation (BiFC) assay
For constructs used in the BiFC experiment, the N-terminal (pSAT1-N) and C-terminal (pSAT1-C) EYFP vectors were used. The full-length of all PslDELLA, including the mutated PslRGL .MU and PslRGA .MU versions were fused into the SacII-BamHI site of the pSAT1-C vector. Consequently, plum GA-receptors PslGID1b and 1c were inserted into the BglII-BamHI of the pSAT1-N vector. 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 in the presence or absence of 100 μM GA 3 , as described previously [13]. All assays were repeated at least three times and visualized using confocal microscopy.
Yeast two-hybrid (Y2H) assays Y2H assays were performed with the Matchmaker Gold Yeast two-hybrid System (Clontech, Palo Alto, CA, USA). PslDELLA full-length ORFs and the mutated PslRGL .MU and PslRGA .MU versions were inserted into the NdeI-BamHI site of the pGADT7 prey vector (GAL4 activation-domain; AD). PslGID1b and 1c cDNAs were fused into the BamHI-PstI and NdeI-BamHI sites of the pGBKT7 bait vector (GAL4 binding-domain; DBD), respectively. Prey and bait vectors (100 ng) were then introduced into Y2HGold and Y187 yeast strains, respectively; using Yeastmaker yeast transformation system 2. Interactions between the proteins were assayed by the mating method, according to the manufacturer's instructions, in 96-well plates containing medium with or without 100 μM GA (GA 1 , GA 3 or GA 4 ), as described previously [13]. All assays were repeated at least three independent times.

Plasmid construction and plant transformation
Full-length PslDELLA and PslRGA .MU (excluding the stop codon) were fused with the GFP reporter gene in the binary vector pGreen0029 [70]. The resulting vectors were transformed into A. tumefaciens and employed for Arabidopsis transformation, as described previously [9]. All genes under the control of the 35S promoter were introduced into wild-type (WT) Arabidopsis background Col-0. Non-transformed WT plants as well as plants transformed with empty vectors were used as controls. Different generations of transgenic plants were selected under kanamycin resistance circumstances. T3 homozygous independent lines from each transformation were grown under standard long day conditions (16:8 h light/300 μmol m −2 s −1 ; 23:18˚C and 65% relative humidity) with or without GA 3 treatment; 24 plants/transformation/ treatment. All plant materials were frozen and stored at −80˚C until use.

Isolation and structural characterization of PslDELLA cDNAs
To investigate the molecular basis of GA action in fruit development, three novel sequences closely related to the growth-repressing DELLA proteins, a subset of the plant-specific GRAS (GAI, RGA and SCARECROW) family of transcriptional regulators were isolated from Early Golden (EG) plum cultivar. PslGAI, PslRGL, and PslRGA predicted to encode proteins of 633, 593, and 537 amino acid residues with calculated molecular weights of 69.9, 64.5, and 59 kDa, respectively. The relationships between the predicted plum and Arabidopsis amino acid sequences, as indicated by percentage similarity over the whole sequence, are presented in S2 Table. The various PslDELLA showed considerable sequence deviation (50-70% similarity), mainly due to the divergence of N-terminal portions outside the DELLA domain. Nevertheless, several signature structural elements commonly associated with the DELLA subfamily were detected (Fig 1A and 1B). The deduced amino acid sequences of PslGAI and PslRGL comprise the two domains essential for the protein function, including the typical N-terminal DELLA domain (DELLA and TVHYNP motifs) and the highly conserved C-terminal GRAS domain. Both domains are necessary for the GA-dependent interaction with GA-receptors GID1 and involved in the repression function of the protein [71][72]. Although PslRGA sequence displayed structurally high similarity with other DELLA proteins, critical divergences in the key amino acid residues important for GID1-DELLA interactions were detected [50], resulting in partially conserved DELLA domain ( Fig 1A). For instance, in the DELLA motif, comprising DeLLaFLxYxV sequence; the three Leu residues are substituted by the distinct amino acids Tyr, Phe and Ala in PslRGA. While the nonpolar residue represented by F (Val 52 and Val 53 in PslGAI and PslRGL, respectively) is replaced by Asp 37 in PslRGA. Further, in the following LExLE motif with the consensus sequence MAxVAxxLExLExF; the first amino acid Met is substituted by Leu and the essential nonpolar Ala residue is changed into Arg. Finally, in the TVHYNP motif (TVhynPxxLxxWxxxM), the amino acid residues Thr, Try, and Met are substituted by Ala, Glu, and Leu, respectively. While the amino acid His that is not important for GID1 interaction, but contributes to protein stabilization [73] is changed into Val. All these critical alterations in amino acid residues essential for the direct GID1 interaction surface suggested that PslRGA may function differentially by having a distinct interaction mode with GID1-like proteins in comparison with typical proteins holding conserved DELLA domain.
In an attempt to unravel the genomic and allelotype structure of the different PslDELLA genes, the full-length genomic sequences of the three genes were isolated and sequenced from EG gDNA. Consistent with DELLA gene subfamily, all plum genes exhibited a single open reading frame without any intron interruption. EG is homozygous for PslGAI; however, two different alleles were identified for PslRGL (a & b) and PslRGA (a & b) with non-synonymous alterations in nucleotide composition, leading to several changes in the predicted proteins ( Fig  1A). The two PslRGL and PslRGA alleles share 98% and 94% amino acid sequence identity, respectively; reflecting the presumed allopolyploid origins. One of the most significant differences between the two PslRGL alleles is the detection of a microsatellite region with imperfect nucleotide recreates due to the insertion of three amino acid residues Ser-Gly-Gly within the N-terminal region of PslRGLb at position 44. On the other side, all the critical structural changes within the N-terminal DELLA domain of PslRGAa were detectable in PslRGAb allele. Sequence data mining in Prunus species genome (e.g. P. persica and P. mume), the closest genomes to plum (P. salicina), identified the three PslDELLA as the only putative DELLA-like genes within the genome.
Phylogenetic analysis of PslDELLA with closely related genes from Arabidopsis indicated that PslGAI can be grouped into the clade of AtGAI and AtRGA, whereas PslRGLa and b are clustered with AtRGL-related proteins ( Fig 1C). Interestingly, PslRGAa and b form a unique distant clade without any representative from Arabidopsis orthologous, indicating that this clade may represent a new branch in DELLA protein evolution. Nonetheless, sequence data mining identified members closely related to PslRGA clade in many other plant species (S1 Fig).
To gain a broader insight into PslDELLA function, we investigated the localization compartment of their proteins. Fluorescence microscopy revealed that, as expected, the full-length PslDELLA−GFP fusions were localized exclusively in the nucleus (S2 Fig), which is consistent with their primary function as transcription regulators [74][75]. Together, these primary comparative analyses suggested that PslGAI and PslRGL might be involved in GA-signalling in a manner similar to that of other characteristic DELLA proteins via interaction with the GAreceptor GID1-like proteins [48]. However, the distinct PslRGA protein might exhibit particular function due to the disrupted DELLA domain. This hypothesis was tested using a number of biochemical and biological approaches, as described below.

Molecular modelling of PslDELLA proteins
To determine whether the putative plum proteins exhibited similar function to those of Arabidopsis DELLA, three-dimensional modelling of PslDELLA proteins was generated and compared with AtGAI as a template. Analysis of the predicted structures indicated that all PslDELLA proteins are highly similar to AtGAI protein (S3 Fig). The two typical DELLA and GRAS domains as well as the links between the domains constituted the differences among the four proteins. Although PslGAI and PslRGL exhibited few amino acid alterations within the N-terminal DELLA domain, these changes are not in the contact residues with GID1-like proteins, but rather in the amino acids that contribute to the conformation of the protein in this area. In the meantime, the 3-D structures did not establish a clear difference between the divergent PslRGA protein and the other DELLA proteins holding complete domain. This is probably due to the conserved long C-terminal GRAS domain within the structure. Consequently, the binding energies of the three PslDELLA to the previously characterized plum GA-receptors PslGID1b and 1c [13] in the presence of GA 3 molecule were calculated to determine the PslDELLA's interaction capacities. Data analysis revealed that the three PslDELLA proteins displayed differential binding energy features to the different PslGID1-like proteins. The binding energies of PslGAI and PslRGL to PslGID1s were found to be within the range of −3 and −17 kcal mol −1 , respectively. However, PslRGA did not show any obvious binding ability for either of PslGID1 proteins with binding energy estimated at~23 kcal mol −1 .

Properties of PslGID1-PslDELLA interaction
Considerable progress has been made in elucidating the molecular basis of GA action [8]. Perception of bioactive GA by its GID1 receptors promotes the direct interactions between GID1 and DELLA domain of GA-repressors DELLA [76][77]. To determine whether PslDELLA possess a comparable function as those of Arabidopsis, the interactions between PslGID1 and PslDELLA were assessed in yeast system in the presence or absence of GA 3 (Fig 2A). The analysis revealed a clear divergence in terms of the interaction capacity and preference. Previous studies have shown that interactions between the Arabidopsis GID1 and DELLA are enhanced in yeast cells in the presence of bioactive GA [47]. Similarly, the binding results confirmed the essential GA-induced assembly of stable GA-PslGID1-PslDELLA complex in yeast. PslGAI and PslRGL were effectively able to interact with both PslGID1s; however, they showed differential binding efficacy to a specific PslGID1 protein. Their capacity to form complex with PslGID1b was much stronger than PslGID1c. Conversely, PslRGA protein did not bind to any of PslGID1s, even in the presence of GA. It has been reported that the two PslGID1s perfectly interacted with Arabidopsis GAI and RGL1 proteins in yeast system [13]. Nevertheless, this may be different in the case of PslGID1 and PslDELLA, where sequence and conformational differences may confer some levels of specificity in PslGID1-PslDELLA pairing.
To provide additional evidence, we attempted to visualize the direct GA-triggered interactions between PslGID1 and PslDELLA using BiFC approach. Tobacco protoplasts supplemented with 0 and 100 μM GA 3 were co-transfected with the various combinations of NY-PslGID1 and CY-PslDELLA constructs (Fig 2B). The YFP signal caused by interaction between PslGID1 and PslDELLA was only detected in protoplasts pre-treated with GA. Although, the untreated cells should contain endogenous GA, no fluorescence signals were observed in cells grown in GA-free medium (data not shown). This is probably due to the scarcity of active GA content that is not sufficient to promote interaction between the fluorescence-labeled PslGID1 and PslDELLA. Consistent with yeast assays, both PslGAI and PslRGL proteins exhibited high activity to form complexes with both PslGID1s; however, PslRGA did not interact with any of the PslGID1 proteins. GA orchestrates a broad range of processes and many levels of regulation are known to be involved in determining GA-responses, including biosynthesis, metabolism and signalling [8,78]. Our data and those of others [11,79] showed that another level of regulation exists in terms of GA-GID1-DELLA binding capacity and preference.

PslGID1-PslDELLA interaction is GA-type-dependent
It was reported that the interaction preference of GID1-DELLA proteins in yeast cells is dependent on the structural features of the bioactive GA used in the reaction [79]. Bioactive GAs, GA 1 , GA 3 and GA 4 , share three common structural traits, including a hydroxyl group on C-3β, a carboxyl group on C-6, and a lactone between C-4 and C-10, in which the 3β-hydroxyl group can be exchanged for other functional groups at C-2 and/or C-3 positions [33]. Interaction capacity between PslDELLA (PslGAI, PslRGL and PslRGA) and GA-receptors (PslGID1b and PslGID1c) using Y2H and BiFC approaches. Y2H assays (A) were performed using PslDELLA as prey in Y187 yeast strain and PslGID1 as bait in Y2HGold yeast strain. The mated yeast was grown in 96-well plates containing DDO/X/A medium in the presence or absence of 100 μM GA 3 . For in planta BiFC assay (B), PslDELLA sequences were fused with the C-terminus (CY) of YFP; PslGID1 were fused with the N-terminus (NY) of YFP. Different combinations of NY and CY constructs were transiently co-expressed in GA-treated tobacco protoplasts. NLS-mCherry was included in each transfection to highlight the location of the nucleus. YFP fluorescence is yellow; the merged image is a digital merge of bright field and fluorescent images to illustrate the interaction location; bars = 10 μm. All experiments were repeated at least three times. However, GA 1 and GA 3 differ from GA 4 by the presence of hydroxyl group at C-13 that can influence the activity of the GA structure [80]. Accordingly, it is possible to speculate that the reduced interaction activity of PslGAI and PslRGL with PslGID1c, and the lack of proteinprotein interaction between PslRGA and both plum GA-receptors are due to using GA 3 as a mediator of the reaction. To test this hypothesis, we examined the interaction property of different PslGID1-PslDELLA in the presence of GA 1 and GA 4 ; the most abundant bioactive GA involved in plum fruit development [13]. Interaction assays in GA-free and GA 3 -containing mediums were included as controls (Fig 3). The effect of GA 4 on the PslGID1-PslGAI/PslRGL interactions was comparable to that of GA 3 and the reduced interaction activity of the two PslDELLA proteins with PslGID1c remained detectable. However, GA 1 showed generally lower activity in triggering the interaction between PslGAI/PslRGL and PslGID1s than GA 3 or GA 4 , with no visible interaction between PslGAI and PslGID1c. Altogether, the results of Y2H experiments suggested that there is substrate preference among the PslGID1-like proteins, which further depends on the structure of bioactive GA. PslGAI and PslRGL are generally better substrates for PslGID1b than for PslGID1c. In the same time, bioactive GA 3 and GA 4 are better mediators for the interaction than GA 1 . One of the most outstanding questions in GA biology is how the hormone controls so many different aspects of plant growth and development. On the basis of Y2H results, it is possible that different GA-PslGID1-PslDELLA complexes have diverse biochemical properties that enable specialized functions.
Despite the type of GA tested, the interactions between PslRGA and PslGID1 proteins were undetectable, confirming the loss of interaction capability of that protein under different circumstances of bioactive GA in yeast cells. Conserved DELLA domain is essential for GA-GID1-DELLA interaction, since any deletion or point substitution results in loss-of-interaction ability despite the presence of GA [47,49,58,75]. Recently, a grape DELLA protein, VvDELLA3, exhibiting high sequence similarity to PslRGA has been characterized [11]. Although, both VvDELLA3 and PslRGA share the disrupted DELLA domain, VvDELLA3 displayed selective interaction with VvGID1-like proteins. Further, the ability of VvDELLA3 protein degradation in response to GA treatment suggests its active function as a GA-sensitive repressor. Sequence comparison between PslRGA and VvDELLA3 highlighted more pivotal substitutions within the DELLA motifs that could potentially account for abolished interaction between PslRGA and PslGID1-like proteins (S4 Fig). By contrast, Fleck and Harberd [75] provide evidence that the distinct Arabidopsis DELLA proteins, GAI and RGL1, are GA-insensitive stable proteins, as they  do not disappear from the nucleus in response to GA treatment and the plants overexpressing each of these two proteins exhibited dwarf, GA non-responsive phenotype.

PslRGA substitutions abolish the GA-dependent complex formation with PslGID1s
The previous results indicated that disrupted DELLA domain of PslRGA might potentially contribute for the lack of interaction with PslGID1-like proteins. To test this hypothesis, the activeinteractor PslRGL and the non-active PslRGA proteins were subjected to a series of mutations. By comparing the amino acid residues of several DELLA sequences from different plant species, particularly between PslRGAa, PslRGAb and VvDLLA3, it appeared that Phe-35, Val-77 and Glu-86 residues are unique to PslRGAa (S4 Fig). Therefore, we mutated these three amino acids in PslRGAa into Leu, His and Try, respectively; to simulate active DELLAs. Similarly, the corresponding amino acids Leu-51, His-93 and Try-102 in PslRGL were changed into their analogs in PslRGAa. The mutated versions of both proteins were designated PslRGL .MU and PslRGA .MU , respectively (Fig 4A). Assessing the localization of the two mutated versions revealed that both proteins remained targeting the nucleus compartment, indicating that the generated mutations did not affect their potential function as transcription regulators (Fig 4B). The consequences of these mutations were evaluated by assessing the changes in the dynamic of interaction property of the original ORFs (as a control) and ORFs carrying mutations, using both Y2H and BiFC approaches (Fig 4C and 4D). Relative to control ORFs, PslRGL .MU protein lost the GA-dependent capacity to bind to any of the PslGID1 proteins. In contrast, PslRGA .MU protein accomplished successful GA-dependent interaction, but only with PslGID1b protein.
The previous results provided strong evidence that the changes within the DELLA domain of PslRGA is the cause of abolished interaction with PslGID1-like proteins.

PslDELLA expression during fruit ontogeny
Earlier studies reported that the expression of DELLA genes differ among various developmental stages. Whereas AtRGA and AtGAI are highly expressed in most tissues, AtRGL1, AtRGL2, and AtRGL3 are mainly expressed in germinating seeds, young seedlings, and flowers [81]. Hence, the expression level of PslDELLA genes was assessed during various developmental stages to provide further credence about their role in regulating fruit growth. An initial screen of the five PslDELLA transcripts (PslGAI, PslRGLa/b, PslRGAa/b) indicated that all are expressed with no significant difference between the -a and -b gene pairing (data not shown). Consequently, the qPCR assays were performed on the -a gene variant of PslRGL and PslRGA. Although transcripts of PslDELLA were ubiquitously expressed, their accumulation profile appears to be organ-and developmental stage-dependent (Fig 5). This preliminary analysis indicated that these GA-negative signalling components might be transcriptionally regulated, as suggested for their orthologs in Arabidopsis [81].
All PslDELLA transcripts were abundantly expressed in flower buds (~−4 DAB), but showed distinct accumulation pattern afterward. PslRGL transcripts gradually declined along with development, reaching low levels by the end of fruit initiation (~22 DAB). The signal of PslRGL detected in flower buds represents the highest abundance among the whole experiment. Conversely, PslGAI and PslRGA steadily increased along with flower development, peaking soon after fertilization,~10 DAB. Subsequently, both transcripts behaved similarly to that of PslRGL mRNA by decreasing to their low levels at the end of fruit-set. GA is involved in diverse biological processes, particularly flowering and fruit initiation [21][22]24]. It actively promotes flowering through regulating floral meristem identity genes [20] and floral integrator genes [23]. Similarly, GA is needed to organize the abundant cell division, expansion and embryo development during fruit-set phase [1][2]. The abundance of the three transcripts during flowering and fruit-set suggested a dominant task of PslDELLA in regulating GA-response during this stage. Recent evidence suggested that ethylene is involved in both the control of the ovule lifespan and the determination of the pistil/ fruit fate. The proposed model suggests that ethylene may modulate the onset of ovule senescence and, consequently, the window of GA fruit-set responsiveness by altering GA-perception and -signalling. Though an actual mechanism remains unidentified, it was suggested that the ethylene produced in ovules would modulate the excessive GA-response by stabilizing the DELLAs via CTR1 [82][83]. Interestingly, a remarkable increase in the transcription   of several ethylene-associated genes was eventually detected in plum during flower to fruit transition (i.e. in the same developmental stages used in the present study) [31,67]. Genetic and biochemical analyses have shown that the five AtDELLA are actively involved in GA-signalling and they exhibit both overlapping and distinct roles in regulating GA-responsive growth [81,[84][85][86]. For instance, only RGA has been shown to prominently mediate GA effects on flower development, whereas GAI, RGA, RGL2 and RGL1 play the main role in the regulation of GA-dependent fruit initiation [44,[87][88]. The accumulation profile of the different PslDELLA suggested the contribution of the three transcripts in regulating floral meristem identity and flower bud initiation. However, only PslGAI and PslRGA are apparently more involved in mediating the GA-dependent events of fruit-set.
Stone fruits (Prunus spp.), including plum, exhibits a typical double sigmoid growth pattern during fruit development with four distinct stages; S1-S4 [13]. During S1 (27-37 DAB), the three transcripts increased to form a modest peak by~32 DAB. Subsequently, PslRGL and PslRGA mRNAs gradually declined to reach relatively low levels by the end of S2 (~52 DAB); however that of PslGAI continue expressed at nearly constant moderate levels. Throughout fruit development, it is almost certain that the series of modifications that make the fruit proceed through the consequent developmental stages involve many different pathways, including the GA pathway. During S1, the GA is needed to organize the intense cell division and expansion [13]. In S2, there is hardly any increase in fruit size, as the fruit enter a period of growth dormancy. Therefore, the significant accumulation of PslGAI transcripts seemed to be associated with the lignification of the endocarp, the only developmental process occurring during this stage [9,13]. It was demonstrated that GA mediates lignin formation and deposition by polymerization of pre-formed monomers [89]. These results suggested that all PslDELLA should be active components of the GA-signal network that regulate fruit growth during immature S1 stage; however, only PslGAI is the dominant player in modulating the GAresponses during S2 phase. Comparing PslDELLA expression profile with the changes in GA contents from flowering until the end of S2-stage suggested that their accumulation is potentially associated with the growth signature events that are triggered in a GA-dependent manner, when GA-biosynthesis and -signalling actively occurred [9,13]. The abundance of PslDELLA in GA-rich tissues may be caused by rapid turnover of PslDELLA proteins or due to feedback regulation of PslDELLA transcription during active GA-signalling. Further, it was demonstrated that the GA-upregulated OsSLR1 expression site is corresponding to the site of GA action, so its expression should be affected by GA levels [90].
During S3 maturation phase (57-77 DAB); the expression profile of the three transcripts remained slightly different. In early S3 stage (57-62 DAB), PslGAI signal was greatly detected and dramatically decreased to its basal levels afterward. However, those of PslRGL and PslRGA remained at low levels. Through S4, where most ripening-related metabolic changes occurred in an ethylene-dependent manner, all PslDELLA were scarcely detectable, signifying a minor contribution during mature growth phase. Interestingly, the decline in PslDELLA transcripts is associated with accelerated cell division and expansion events, resulting in visible enlargement in fruit size [13]. GA-mediated responses are under the tight regulation of growth-repressing DELLA proteins. According to the "relief of restraint" model, any activation of GA-signalling requires degradation of DELLA proteins [38,44]. Therefore, the down-regulation of PslDELLA in mature fruiting tissues can enable fruit expansion and relief fruit growth, reaching their standard size. The effect of GA application in increasing fruit size and weight of several fruit species, including plum, strongly support this hypothesis [1,13,91].
Although, the levels of the three transcripts were hardly detected, a slight increase in their signal was observed at~82 DAB. Remarkably, these minor increases coincided with the climacteric ethylene production peak [64]. Accordingly, it is tempting to speculate that PslDELLA are potentially regulated by ethylene during fruit ripening. To confirm this hypothesis, the expression of the three transcripts was assessed in EG fruit pre-treated with the ethylene stimulator propylene and the ethylene response inhibitor 1-MCP. The results provided further credence to the potential feedforward regulation of PslDELLA by ethylene in mature fruiting tissues (S5 Fig). As expected, propylene-treated fruit exhibited rapid and brief ripening profile in association with increased ethylene levels. In contrast, all fruit treated with 1-MCP were unable to ripen autonomously and their ethylene production remained low. Propylene treatment caused dramatic increase in all PslDELLA and this correlated well with the changes of ethylene production during fruit ripening (R 2 = 0.96; P<0.01). Conversely, 1-MCP treatment abolished ethylene-induced PslDELLA expression. Previous studies suggested a cross-talk between GA and ethylene in the regulation of different aspects of plant development [82,92]. However, the nature of interaction between the two hormones (positive or negative) is dependent on the developmental and environmental circumstances. Apparently, in mature fruit, ethylene alters GA-responses by directly or indirectly enhancing DELLA transcription and/or increasing DELLA proteins stability [8,93].

Overexpression of PslDELLA in WT Arabidopsis
To examine PslDELLA-like protein function in planta, the three genes were introduced separately into WT Arabidopsis background (Col). A number of independent transgenic lines (15 to 26 lines / transformation) were obtained and confirmed by qPCR analysis. However, only two homozygous T3 representatives from each transformation were selected for further studies on the basis of differential transgene levels (Fig 6A). Transformed plants with empty vector were phenotypically indistinguishable from WT (data not shown). According to the model suggested by Achard and Genschik [78], DELLAs restrain plant growth, whereas GA promotes growth by targeting DELLAs for destruction. Therefore, increasing the amount of DELLArepressors within the plant should lead to artefacts due to over-saturation in the system that typically affect the GA-signalling machinery, causing changes in plant phenotype consistent with aberrant DELLA protein accumulation [78,94]. To better characterize the resultant phenotypes, the expression of some Arabidopsis genes involved in GA-metabolism was assessed. GA-homeostasis in a variety of plant species has been found to be tightly linked to the activities of enzymes involved in GA-biosynthesis and -catabolism [19,95]. When GA levels and/or responsiveness are high, genes encoding enzymes for GA-biosynthesis (AtGA20ox and AtGA3ox) and enzymes for GA-inactivation (AtGA2ox) are subject to negative-feedback and positive-feedforward regulation, respectively [33]. Consistent with the GA-regulation model, the accumulation of AtGA2ox8 strongly declined in PslGAI-, PslRGL-and PslRGA-plants by 64%, 60%, and 74%, respectively. While AtGA20ox1 and AtGA3ox1 steadily increased bỹ 5.1-,~4.4-, and~6.2-fold, and~3.6-,~4.2-and~5.5-fold in PslGAI-, PslRGL-and PslRGAplants, respectively (Fig 6B). The previous data suggested that the over-accumulation of PslDELLA in transgenic plants was able to alter the feedback and feedforward regulation of GA-metabolism pathway. We further characterized the molecular basis of the GA-signalling disturbance in transgenic plants by quantifying the levels of the five endogenous AtDELLA mRNAs. However, no significant differences between WT and transgenic plants in the levels of AtDELLAs were detected, confirming that the resulting phenotypes were due to the selective introduction of PslDELLA transgene (data not shown). Overexpression of PslDELLAs in WT Arabidopsis led to dramatic disturbances in general growth performance consistent with impaired GA-responses. All the plants overexpressing PslDELLA proteins exhibited a severe dwarf phenotype; however, the repressing activity of PslRGA was always much stronger than that of PslGAI and PslRGL proteins (Fig 6C).

Developmental phenotypes of PslDELLA lines
Despite the advances in our understanding of the molecular basis of GA action, it remains unclear how these key phytohormones promote growth. Overexpressing PslDELLA in Arabidopsis visibly perturb overall plants growth behavior, including rooting capacity, plants architecture, and general vegetative and reproductive growth. Bioactive GA plays crucial roles in coordinating different plant growth aspects [5]. Thus, the interruption in GA-signalling pathway due to PslDELLA-overexpression can explain the distortion in different growth incidence of transgenic plants. Consequently, application of bioactive GA in such GA-deficient mutants has convenient implications in identifying the GA-dependent growth processes. Accordingly, the different PslDELLA-events were phenotypically characterized for some of well-known GA-dependent traits under standard growth conditions and in response to GA 3 treatment.
All PslDELLA-plants exhibited compact shoot growth associated with slender root formation and proliferation of lateral roots. The root length of PslGAI-, PslRGL-and PslRGA−plants were enhanced by~0.7-, 1.1-, and 0.9-fold, respectively (Fig 7). By providing an exogenous supply of bioactive GA 3 in the culture medium, we tested the GA-response of the independent Arabidopsis lines. Excluding PslRGA−plants, GA treatment caused a rapid stem elongation concomitant with a severe reduction in the formation of adventitious roots. By contrast, PslRGA−plants were unaffected by the treatment, producing compact shoots and elongated roots despite the GA 3 incorporated in the medium.
The role of GA in plant development has been well characterized [5,96]. Nevertheless, the way by how GA regulates plant development is still poorly understood [97][98]. The compact stem growth along with the accelerated root characteristics are a common behavior in GAdeficient mutants [47,[99][100][101]. Recent studies suggested that GA inhibited root growth by suppressing lateral root formation in a DELLA-dependent pathway [97,[101][102][103]. The conflict phenomenon of GA effects in shoot and root growth has been previously reported in several plants species [102][103][104][105]. The fact that PslRGA overexpression, as other PslDELLA promote lateral root formation, but selectively overcomes the inhibitory effect of GA on root formation support the idea that PslRGA encodes a functional DELLA-repressor, but insensitive to GA presence.
All transgenic Arabidopsis plants overexpressing the different PslDELLA genes showed compact growth due to substantial decline in the length of all stem growth-related characters, in which PslRGA-plants had the strongest dwarfing effect. Relative to WT, PslGAI-, PslRGLand PslRGA-plants exhibited significant reduction in their overall heights by~83%, 86% and 92%, respectively ( Table 1, Fig 8). Moreover, PslGAI-and PslRGL-plants architecture was visibly different due to development of multiple branching architecture in association with numerous, but notably short internodes (Table 1, Fig 8B and 8C). Among the different transgenic events, PslGAI-plants displayed the highest branched structure followed by PslRGLplants; however, such structure was not evident in PslRGA-plants. The altered branching pattern PslDELLA-plants is probably due to constitutive GA-response within the axillary bud meristem. Doust and his colleagues [106] have identified quantitative trait loci in foxtail millet for branching architecture, including genes encoding GA biosynthetic enzymes. In the LAT-ERAL SUPPRESSOR (ls) mutant of tomato, in which axillary bud growth is repressed, the GA content of these buds is higher than in those of the wild type [107]. Apparently, PslDELLArepressors alter plant structure directly or indirectly by triggering LS protein [37]. However, this pattern seems to be more associated with GA-sensitive DELLA-repressor.
The transgenic plants differentially responded to the application of bioactive GA 3 (Table 1; Fig 8F). GA treatment rescued to certain extent PslGAI-and PslRGL-plants' height mainly due to extending internode length with no considerable changes in internode number. Relative to control untreated plants, no changes in the branch architecture were observed (Fig 8B, 8C and  8D). While, the compact phenotype of PslGAI-and PslRGL-mutants can be partially reversed by GA treatment, PslRGA-plants were not affected by GA presence (Table 1; Fig 8D and 8F) even after increasing the doses of GA to 500 and 1000 μM. With respect to the typical DELLA proteins with intact domains, putative DELLA-repressors holding disrupted DELLA domain are less responsive to GA-induced degradation, indicating that these proteins may function as constitutive suppressors of GA-signalling independent of GA action [49,[58][59]90]. Thus, we hypothesize that the degenerated DELLA domain in PslRGA is the cause of GA-insensitive phenotype observed in corresponding plants. Apparently, such proteins operate to maintain a basal level of growth-restraint in specific tissues or at certain points of development despite the presence or absence of bioactive GA [108][109]. If this is correct, the generated version PslRGA .MU with recovered DELLA domain should respond to GA treatment on the bases of its active interaction property detected in yeast system (Fig 4). Hence, transgenic Arabidopsis plants overexpressing PslRGA .MU sequence were generated and its phenotypical growth characteristics in the presence or absence of GA 3 were evaluated. Interestingly, PslRGA .MU -overexpression conferred a compact growth phenotype that is not obviously distinguishable from that occurred by the native PslRGA transgene (Table 1; Fig 8D and 8F). Contrary to PslRGAplants, when PslRGA .MU -plants treated with GA they developed elongated internodes, resulting in partial recovery.
Plants can adopt a wide variety of environmental forms. The plasticity plays important roles in ecosystems, agriculture and landscape aesthetics. Under stressful circumstances, plants can rapidly respond to the environmental stimuli by rebuilding their system architecture to modify whole plant strategies, avoiding the environmental impact with maintaining productivity. The fundamental importance of these processes has prompted considerable research into how plants perceived the alert signal and how they governed the subsequent changes in growth behavior. Recent studies proposed that GA-signalling permits flexible and appropriate modulation of plant growth in response to changes in natural environments [63,86,93,110]. Therefore, it is possible to speculate that GA is one of the key players that regulate the plant's decision if exposed to unfavorable environmental conditions through upregulating DELLArepressors, leading to impaired growth rate. If this is the case, it is obvious that the over-accumulation of PslDELLA in transgenic plants will turn-on the alert signal, resulting in not only reduced stem growth but also enhanced rooting system. Gibberellins are involved in the developmental events leading to reproductive competence, as well as in floral determination and commitment [85,93]. Under standard growth conditions, PslDELLA-overexpression caused substantial disorder in all phenotypical and phenological characteristics of reproductive growth. GA treatment was able to recover the different aspects of reproductive growth disruption in all mutants, excluding those of PslRGA-plants; however, the recovery remained visibly less than WT. Conversely, PslRGA-plants continued showing insensitive response to GA.
The number of flowers/inflorescence noticeably decreased in different transgenic events. In addition, the transition to flowering was considerably delayed in PslGAI-, PslRGL-, and PslRGA-plants by~17,~22, and~35 days, respectively. Further, PslDELLA-plants displayed generally much smaller flower size and their filaments were usually shorter than their pistil ( Table 1, Fig 9A). Such variation between the stamens and pistil can cause a major reduction in fertility, especially in self-pollinated species. Excluding PslRGA-flowers, GA application restored flowers number, flowering time and proper flower structure (Table 1).
Relative to WT, the time from flowering to silique maturation was significantly delayed bỹ 13, 16 and 19 days in PslGAI-, PslRGL-, and the two PslRGA-related plants, respectively (Table 1). GA treatment slightly delayed silique maturity in WT plants. This contradictory response is probably due to reach over-dose levels of the hormone, suggesting the importance of optimal GA levels to ensure proper growth and development. By contrast, the treatment considerably reduced siliques shattering duration, but substantially remained longer than WT. Furthermore, both silique length and seed number were drastically reduced in PslDELLAplants (Table 1, Fig 9B). Silique lengths of PslGAI-, PslRGL-, and the two PslRGA-related plants were reduced by~79%, 77% and 89%, respectively. Although all mutants exhibited significant reduction in seed number potentially due to compromised flower structure, the seeds were completely developed, but exhibited delayed germination estimated by 2 days later than WT in all PslDELLA-mutants. Analysis of different growth aspects in response to GA application highlighted the PslRGA .MU -plants as the strongest mutant resisting the stimulatory effect of GA in re-establishing typical growth, suggesting the existence of other obstacles within the sequence that still impairs PslRGA .MU -AtGID1 interactions.

GA-insensitivity due to PslRGA stability
Ultimate GA-response is the result of antagonistic reaction between GA activation and suppression mechanisms [111]. Therefore, any disturbance in these machineries can modify the response of plants to active GA. The GA-insensitivity observed in transgenic PslRGA-plants along with the ability of GA treatment to rescue PslRGA .MU -plants suggested that PslRGA is a highly stable DELLA protein. To confirm this hypothesis, we determined the kinetics change in detectable level of nuclear PslDELLA−GFP fluorescence in transgenic Arabidopsis roots independently expressing the different PslDELLA−GFP proteins after 60, 120 and 240 minutes of GA 3 treatment. Using this approach allowed us to monitor any alteration in the dynamic of PslDELLA−GFP proteins degradation in response to exogenously applied GA, which is informative because PslDELLA−GFP proteins are functionally active in respective transgenic plants.
Fluorescence intensity of PslGAI-and PslRGL−GFP chimeric proteins in root cell nuclei decreased substantially within 120 minutes of GA treatment and disappeared after 240 minutes, indicative of complete degradation of both proteins (Fig 10A and 10B), as demonstrated previously for several GA-sensitive DELLA proteins [63,[112][113][114]. Conversely, this dynamic GA-induced degradation of the two PslDELLA proteins was not seen in transgenic roots carrying PslRGA. PslRGA−GFP fluorescence showed full resistance to the destabilizing impact of GA and did not display any changes in fluorescence intensity 240 minutes after the onset of GA treatment (Fig 10C).
The stability of PslRGA protein was further confirmed by assessing the response of modified PslRGA .MU −GFP expressed in Arabidopsis root tips to GA treatment. Although PslRGA .
MU −GFP showed slower degradation rate in response to GA presence comparing with PslGAI and PslRGL proteins, the generated protein was readily degradable and almost disappeared 240 minutes post GA treatment (Fig 10D). One possible explanation for the relative persist signal of PslRGA .MU −GFP for longer period post GA treatment would be attributable to the existence of other substitutions within the sequence that impair PslRGA .MU -AtGID interactions.
In subsequent experiments, we transiently expressed the native ORFs PslRGL−GFB and PslRGA−GFP as well as their modified derivatives (PslRGL .MU −GFP and PslRGA .MU −GFP) in tobacco protoplasts treated with GA, paclobutrazol (PAC), and a joint treatment of PAC followed by GA to avoid the involvement of endogenous GA effect. Consistent with the previous results, the application of GA to protoplasts carrying the PslRGL−GFP and PslRGA .MU −GFP derivatives induced the disappearance of GFP florescence signal (Fig 11A and 11D). However, the intensity of nuclear signal in protoplasts transfected with PslRGA−GFP and more interestingly that of PslRGL .MU −GFP holding degenerated DELLA domain was not affected by GA (Fig 11B and 11C). Because PAC inhibits GA-biosynthesis, we sought to determine whether PAC treatment would have a different effect on PslDELLA protein degradation. PAC treatment enhanced the protein stabilization, as determined by the strong GFP signal detected with all tested chimeric proteins (Fig 11). We then examined the response of different proteins to a combined treatment of PAC+GA to confirm that the rapid loss of GFP fluorescence is GAdependent. The responses of different proteins to the combined treatment were similar to their response to protoplasts treated with GA only (Fig 11). Thus, GA activity seemed to cause the reduced level of the PslRGL and PslRGA .MU proteins. These results further support the critical contribution of conserved DELLA domain in mediating the GA-dependent DELLA degradation. The stability of DELLA proteins was demonstrated previously for plant mutants exhibiting naturally occurred and/or intentionally induced mutations (point mutation, deletion or truncation) within the critically important DELLA domain [58,90,94,103,[115][116][117][118]. The absence of the conserved DELLA domain abolished the interaction of corresponding proteins with GID1, affecting their subsequent degradation via the ubiquitin/26S proteasome pathway [53][54].
The consistency of PslDELLA responses to GA using different approaches suggested that plum fruit development is actively regulated by three types of DELLA transcription factors in which two of them encode GA-sensitive proteins (PslGAI and PslRGL); however, the third one is GA-insensitive (PslRGA). This raises the question-why plum trees comprise a GAinsensitive DELLA protein within its genome? The destabilization of PslGAI and PslRGL likely released the growth-restraining effects of the two proteins in the fruiting tissues. It could be speculated that because DELLA-less forms of proteins, as PslRGA, are more resistant to GA- Fluorescence were monitored with confocal laser scanning microscopy after treatment with water (mock) or GA 3 (100 μM) for 60, 120 and 240 minutes.
doi:10.1371/journal.pone.0169440.g010 dependent degradation, they will have strong effect in controlling GA-signalling that coordinate fruit growth. Thus, such proteins can be present in plants and function as a backup system in place, avoiding the unnecessary excessive accumulation of GA-signalling. Our observation that PslRGA−overexpression inhibits the expansion of reproductive growth, overcoming the destabilization effect of GA application supports this hypothesis.
Supporting Information S1 Fig. Evolutionary relationships of DELLA proteins. The evolutionary distances were computed using the Poisson correction method. The analysis involved 31 amino acid sequences from different plant species that belong to monocots and dicots, including P.