Genome-wide characterization of the tomato GASA family identifies SlGASA1 as a repressor of fruit ripening

Abstract Gibberellins (GAs) play crucial roles in a wide range of developmental processes and stress responses in plants. However, the roles of GA-responsive genes in tomato (Solanum lycopersicum) fruit development remain largely unknown. Here, we identify 17 GASA (Gibberellic Acid-Stimulated Arabidopsis) family genes in tomato. These genes encode proteins with a cleavable signal peptide at their N terminus and a conserved GASA domain at their C terminus. The expression levels of all tomato GASA family genes were responsive to exogenous GA treatment, but adding ethylene eliminated this effect. Comprehensive expression profiling of SlGASA family genes showed that SlGASA1 follows a ripening-associated expression pattern, with low expression levels during fruit ripening, suggesting it plays a negative role in regulating ripening. Overexpressing SlGASA1 using a ripening-specific promoter delayed the onset of fruit ripening, whereas SlGASA1-knockdown fruits displayed accelerated ripening. Consistent with their delayed ripening, SlGASA1-overexpressing fruits showed significantly reduced ethylene production and carotenoid contents compared to the wild type. Moreover, ripening-related genes were downregulated in SlGASA1-overexpressing fruits but upregulated in SlGASA1-knockdown fruits compared to the wild type. Yeast two-hybrid, co-immunoprecipitation, transactivation, and DNA pull-down assays indicated that SlGASA1 interacts with the key ripening regulator FRUITFULL1 and represses its activation of the ethylene biosynthesis genes ACS2 and ACO1. Our findings shed new light on the role and mode of action of a GA-responsive gene in tomato fruit ripening.


Introduction
Gibberellins (GAs) are a class of tetracyclic diterpenoid phytohormones that regulate diverse aspects of plant development and stress responses, including stem elongation, cell division, seed germination, trichome formation, lateral root formation, f lower and fruit development, and resistance to both biotic and abiotic stress [1][2][3][4][5][6][7][8]. GA signaling is based on the E3 ubiquitin ligase-mediated polyubiquitination and subsequent proteolytic degradation of inhibitory DELLA proteins [2]. When bioactive GA binds to the GA receptor GID1 (GIBBERELLIN-INSENSITIVE DWARF1), the activated receptor binds to DELLA proteins and enables their polyubiquitylation by the ubiquitin ligase E3 SKP1-CULLIN-F-box (SCF) complex and their subsequent degradation via 26S proteasomemediated proteolysis. The proteolysis of DELLAs releases their transcription factor partners from inhibition and triggers the transcriptional reprogramming of GA-responsive genes [2].
Tomato is not only an important cash crop but also a model plant for studying fruit development and climacteric ripening [32]. Using tomato as a model, ethylene was shown to play an essential role in controlling climacteric fruit ripening [33,34]. In addition to ethylene, several transcription factors such as RIPENING INHIBITOR (RIN), NON-RIPENING (NOR), FRUITFULL1/2 (FUL1/2), APETALA2a (AP2a), and TOMATO AGAMOUS-LIKE 1 (TAGL1) act as key regulators of ripening [33,[35][36][37][38][39]. Interestingly, GA was recently shown to play a negative role in tomato fruit ripening in an ethylene-dependent manner [40]. Nevertheless, the roles of GA-responsive genes in tomato fruit development and ripening remain elusive.
Here, by identifying and analysing members of the tomato GASA gene family, a family known to be responsive to GA, we revealed that SlGASA1, a member of the SlGASA family, plays a negative role in fruit ripening by repressing the effect of FUL1 on regulating genes involved in ethylene biosynthesis. Our study provides new insights into the role of this GA-responsive gene in the regulatory network controlling fruit ripening.

Identification of 17 GASA family genes in the tomato genome
We identified 17 GASA genes in the whole tomato genome (SL4.0) using Hmmsearch and multi-sequence alignment based on the protein sequences of GASA family members in Arabidopsis. All GASAs contain a signal peptide of 18-29 amino acids at their N terminus, a variable region with polar amino acid residues in the middle of the protein sequence, and a GASA domain (PF02704) at the C terminus (Fig. S1, see online supplementary material). In addition to SlGASA1 (also known as GAST1, as reported in 1992 [9]), the 16 other SlGASA family genes were named based on their locations on the tomato chromosomes. Detailed information about these genes/proteins, including their chromosomal coordinates, isoelectric points, molecular weights, protein lengths, and open reading frame (ORF) lengths, is provided in Table S1 (see online  supplementary material).

Analysis of the conserved domains and promoter sequences of tomato GASA/GASA family members
To further investigate the diversity of SlGASA protein structure and to predict their putative functions, we identified their conserved domains and motifs based on searches of the Pfam and MEME databases. As shown in Fig. S2 (see online supplementary material), all SlGASA family members harbor the GASA domain and three conserved motifs (i.e. motif 1, motif 2, and motif 3). Motif 3 is linked to the conserved signal peptide of the GASA family, while motifs 1 and 2 form the GASA domain.
To predict the potential roles of SlGASA family genes in different developmental steps, we predicted the presence of cisacting elements in their promoter regions using the PlantCARE database. We identified cis-acting elements which are related to plant development, abiotic and biotic stresses, and plant hormone responsiveness (Data S1, see online supplementary material). Interestingly, we only detected the typical GA-responsive element (TATC-box) in the promoter of SlGASA1 but not in that of other members of the SlGASA family, pointing to a potential role for SlGASA1 in GA-mediated processes (Data S1, see online supplementary material).

Responsiveness of SlGASA family gene expression to exogenous GA treatment
Since SlGASA1 was first identified as a GA-inducible gene, we assessed the responsiveness of SlGASA family genes to treatment with GA 3 and the GA biosynthesis inhibitor paclobutrazol (PAC) via reverse transcription quantitative PCR (RT-qPCR). The transcript levels of most SlGASA genes (SlGASA1, SlGASA6, SlGASA10, SlGASA14, SlGASA4, SlGASA11, SlGASA7, SlGASA9, SlGASA16, SlGASA12, SlGASA2, SlGASA8, SlGASA5, and SlGASA15) were upregulated by GA 3 treatment ( Fig. 2A; Fig. S3, see online supplementary material). Among these genes, only SlGASA1 expression was repressed by PAC treatment ( Fig. 2A; Fig. S3, see online supplementary material). However, SlGASA3 and SlGASA13 expression was repressed by GA 3 and activated by PAC treatment ( Fig. 2A; Fig. S3, see online supplementary material). Interestingly, the application of exogenous ethylene to GA 3 -treated fruits eliminated the effect of GA 3 on SlGASA transcript levels ( Fig. 2A; Fig. S3, see online supplementary material). These results suggest that all SlGASA family members are sensitive to exogenous GA treatment and that ethylene can repress this GA responsiveness in tomato.

Expression profiles of SlGASAs in various tissues and developmental stages
To investigate the potential functions of SlGASA family genes in tomato, we performed RT-qPCR to examine the accumulation of SlGASA transcripts in various tissues and different developmental stages (including root, stem, leaf, bud, f lower tissue, and 11 fruit tissues at different developmental and ripening stages). As shown in Fig. 2B, the 17 SlGASA family genes can be divided into four subgroups based on their expression patterns during plant growth and fruit development. Subgroup I consists of two SlGASA genes (SlGASA13 and SlGASA17) whose transcript accumulation peaks at the breaker (Br) stage. Subgroup II contains eight genes (SlGASA10, 15, 9, 11, 2, 7, 6, and 16) with preferential expression in flower buds ( Interestingly, among all SlGASA family members, SlGASA1, SlGASA3, and SlGASA5 transcripts declined in abundance after the breaker stage ( Fig. 2B; Fig. S4, see online supplementary material), suggesting that these genes play negative roles in fruit ripening. We mined public transcriptome deep sequencing (RNAseq) data [41] to validate our results. Based on this analysis, SlGASA13 and SlGASA17 transcripts accumulated during fruit ripening, while most other SlGASA genes, especially SlGASA1, SlGASA3, and SlGASA5, exhibited decreased expression during fruit ripening (Fig. S5, see online supplementary material). These expression profiles support the notion that SlGASA family genes are involved in fruit ripening.

Subcellular localization of SlGASA1 and its responsiveness to ethylene
GAs were recently shown to play a negative role in fruit ripening [40]. To investigate the functional significance of GA-responsive genes in the ripening process, we selected SlGASA1 for further study as it was the only SlGASA family member induced by GA but repressed by PAC and was downregulated during ripening. Subcellular localization performed in Nicotiana benthamiana leaf protoplasts showed that SlGASA1 localizes to both the cytoplasm and the nucleus (Fig. 3A). To explore whether SlGASA1 is regulated by ethylene, we examined the effect of treating mature green (MG) tomato fruit with ethylene or 1-MCP (1-methylcyclopropene, an ethylene perception inhibitor) on the transcript levels of this gene by RT-qPCR; transcripts of the ethylene-induced gene E4 were used as a control. The transcript accumulation of SlGASA1 was repressed by ethylene but induced by 1-MCP, suggesting a negative role of SlGASA1 in fruit ripening (Fig. 3B).

Overexpression of SlGASA1 in tomato leads to a delay in fruit ripening
Because SlGASA1 expression was significantly downregulated at the ripening initiation stage (i.e. breaker stage), we generated SlGASA1 overexpression (SlGASA1-OE) lines using a fruit ripeningspecific promoter (E8) to investigate the role of SlGASA1 in fruit ripening. We obtained more than 10 independent transgenic lines and performed RT-qPCR to assess the relative accumulation of SlGASA1 transcripts in breaker stage fruits. Three independent T 2 lines (L1, L2, L3) that showed significantly higher SlGASA1 transcript levels (in the range 15-60-fold) compared to the wild type (WT) were selected for physiological and molecular analyses (Fig. 4A).
The SlGASA1-OE plants showed a significant delay in the onset of ripening compared to the WT (Fig. 4B). Under our cultivation conditions, the average time from anthesis to the breaker stage in WT fruits was approximately 38 days (Fig. 4C), while for SlGASA1-OE fruits, the average time was 43 (L1), 43 (L2), and 42 (L3) days (Fig. 4C). The color change measurements (hue angle values) confirmed the delayed ripening of SlGASA1-OE fruits compared to the WT after the breaker stages (Fig. 4D). We also monitored ethylene emission during ripening, finding that ethylene production is significantly reduced at the breaker (Br) and breaker+3 stages (Br + 3) in fruits of the three SlGASA1-OE lines (Fig. 4E). These results indicate that in addition to a delay in ripening initiation, fruit ripening after the breaker stage is also delayed in the SlGASA1-OE lines.  Carotenoid accumulation is an important parameter indicating the progress of ripening. To further investigate the inf luence of SlGASA1 overexpression on fruit ripening, we measured the carotenoid contents in the WT and SlGASA1-OE lines. The contents of lycopene (red) and β-carotene (orangeyellow), two critical carotenoid components responsible for tomato color, were significantly reduced in SlGASA1-OE fruits after the breaker stage ( Fig. 4F and G). These results further support the finding that ripening is delayed in SlGASA1-OE fruits, and they point to a repressive role for SlGASA1 in tomato fruit ripening.

Expression of ripening-related genes is suppressed in SlGASA1-OE fruits
The ripening-delayed phenotype prompted us to examine the expression levels of ripening-associated genes in SlGASA1-OE lines. In line with the delayed ripening and decreased ethylene production of these lines, the transcript levels of ACS2, ACS4, and ACO1, three key ethylene biosynthesis genes, were significantly reduced in SlGASA1-OE fruits compared to the WT at the breaker stage (Fig. 5). Moreover, the transcript levels of PSY1, PDS, andβ-CYC, encoding critical regulators of f lux through the carotenoid pathway, were reduced in SlGASA1-OE fruits (Fig. 5), which is consistent with their decreased carotenoid accumulation. Furthermore, the transcript levels of genes associated with cell wall modification and fruit softening, such as PG2a (POLYGALACTUR-ONASE 2a) and PL (PECTATE LYASE), were also significantly lower in SlGASA1-OE fruits than in WT at the breaker stage (Fig. 5). In addition, the expression levels of RIN, NOR, and FUL1, three key ripening regulators [42], were significantly downregulated in the SlGASA1-OE lines (Fig. 5). Interestingly, the relative transcript levels of DEMETER-LIKE PROTEIN 2 (DML2), encoding a key DNA demethylase that controls fruit ripening in tomato, were also lower in L2 and L3 compared to the WT (Fig. 5). These results further support the repressive role of SlGASA1 in regulating fruit ripening in tomato.

Silencing of SlGASA1 leads to accelerated fruit ripening
To gain more insight into the role of SlGASA1 in fruit ripening, we used virus-induced gene silencing (VIGS) to knockdown SlGASA1 transcript levels in tomato fruit. Accordingly, we infiltrated WT fruits at 30 DPA with Agrobacterium (Agrobacterium tumefaciens) harboring either pTRV2-SlGASA1 or pTRV2 (as the control). The SlGASA1-silenced fruits exhibited earlier ripening than the control (Fig. 6A). In control fruits, the average time from anthesis to the breaker stage was approximately 37 days, whereas SlGASA1-silenced fruits reached the breaker stage at 33 DPA (Fig. 6B). RT-qPCR analysis showed that SlGASA1 transcript levels are significantly lower in SlGASA1-silenced lines compared to the control (Fig. 6C). By contrast, the expression levels of SlGASA8, a family member closely related to SlGASA1, displayed no significant change in SlGASA1-silenced lines (Fig. 6D), suggesting that the VIGS construct is specific for SlGASA1. In line with their accelerated fruit ripening, the transcript levels of ripening-associated genes including ACS2, ACO1, PSY1, PL, E8, RIN, and FUL1 were significantly higher in SlGASA1-silenced fruits than those of the control at 37 DPA (Fig. 6E). These results further support the notion that SlGASA1 represses fruit ripening.

SlGASA1 interacts with the ripening regulator FUL1 and represses its upregulation of ACS2 and ACO1
To explore the possible mechanism underlying the role of SlGASA1 in regulating fruit ripening, we examined the interaction between SlGASA1 and a set of ripening regulators: RIN, NOR, AP2a, TAG1, FUL1, and EIL1-4 (ETHYLENE-INSENSITIVE3-LIKE 1-4) by performing yeast two-hybrid assays. SlGASA1 interacted with FUL1 (Fig. 7A), a key MADS-box transcription factor in the fruit ripening regulatory network, but not with any of the other transcription factors tested. To validate the interaction between SlGASA1 and FUL1 in vivo, we conducted co-immunoprecipitation (Co-IP) assays in N. benthamiana leaves. SlGASA1-HA co-precipitated with FUL1-FLAG, but not with GFP-FLAG, following immunoprecipitation with an anti-FLAG antibody (Fig. 7B). These results indicate that SlGASA1 and FUL1 interact both in vitro and in vivo.  A Screening for ripening regulators that interact with SlGASA1 by yeast two-hybrid. SlGASA1 protein was fused to the GAL4 DNA-binding domain as bait, and RIN, NOR, AP2a, TAG1, FUL1, and EIL1-4 were fused with the GAL4 activation domain as preys. Interactions between P53 and T7 and between SlGASA1-BD and empty AD were used as positive and negative controls, respectively. B Co-immunoprecipitation assay validating the interaction of SlGASA1 and FUL1. SlGAST1-HA and FUL1-FLAG, FUL1-FLAG and GFP-HA were transiently co-infiltrated in N. benthamiana leaves as described in the 'Materials and methods' section.
Because both ACS2 and ACO1, which are key ethylene biosynthesis genes controlling climacteric fruit ripening, are direct targets of FUL1 and were significantly downregulated in SlGASA1-OE fruits but upregulated in SlGASA1-silenced fruits, we reasoned that the interaction between SlGASA1 and FUL1 might affect the transcriptional regulation of ACS2 and ACO1 by FUL1. To test our hypothesis, we performed dual-luciferase reporter assays in N. benthamiana leaves (Fig. 8A). Transient expression of FUL1 in N. benthamiana leaves enhanced the promoter activity of ACS2 and ACO1, but co-expression of FUL1 and SlGASA1 significantly reduced the transcriptional activation of these two genes (Fig. 8B). To examine whether SlGASA1 recruits FUL1 to the ACS2 and ACO1  s t-test). C Binding of SlGASA1 and FUL1 to the ACS2 and ACO1 promoters. A DNA pull-down assay showed that SlGASA1 and FUL1 bind to the ACrG box region in the ACO1 or ACS2 promoter. GST-FUL1, SlGASA1-HIS, and GST proteins were produced and purified in Escherichia coli. SlGASA1-HIS and GST-FUL1, SlGASA1-HIS and GST were incubated with ACO1 or ACS2 promoter probes, and the interactions were detected with anti-GST antibody.
promoters, we performed DNA pull-down assays with biotinlabeled ACS2 and ACO1 promoters. The ACS2 or ACO1 promoter pulled down SlGASA1 only in the presence of FUL1 (Fig. 8C), suggesting that SlGASA1 and FUL1 form a complex to repress the transcription of ACS2 and ACO1. Taken together, these results suggest that SlGASA1 regulates fruit ripening by repressing the transcriptional regulation of key ripening-associated genes by FUL1.

Discussion
GA plays a negative role in fruit ripening in tomato [40,43], but the roles of GA-responsive genes in fruit ripening remain largely unknown. Here, by identifying GASA family genes in tomato, we revealed that SlGASA1, a GA-responsive gene with low expression after the breaker stage, encodes a repressor of fruit ripening that represses the transcriptional activation of the ethylene biosynthesis genes ACS2 and ACO1 by FUL1. Our findings of the role and mode of action of SlGASA1 in fruit ripening provide new insights into the regulatory mechanism of climacteric fruit ripening, and they extend our understanding of the roles of GASA family members in various developmental processes.
GASA family genes have been identified in a number of plant species, and different family genes have been found to be involved a wide range of developmental processes, including cell elongation [16,17,21], the f loral transition [17,28], fruit and seed development [13,21,28], defense against pathogens [5][6][7][8], and resistance to abiotic stress [6]. Among the 17 SlGASA family genes, genes in subgroup II were specifically expressed in f lower buds ( Fig. 2B; Fig. S4, see online supplementary material), suggesting that they might function in the f loral transition. Subgroup IV genes were mainly expressed in roots and fruits during early development ( Fig. 2B; Fig. S4, see online supplementary material), supporting the involvement of these subgroup members in root and fruit development. Interestingly, in contrast to the high expression levels of the two SlGASA genes in subgroup I, the four SlGASA genes in subgroup III were expressed at low levels after the onset of ripening ( Fig. 2B; Fig. S4, see online supplementary material), suggesting that they play a negative role in fruit ripening. Notably, the four genes in subgroup III, which were expressed at low levels during ripening, were highly expressed in vegetative and reproductive tissues, suggesting that they play important roles in vegetative and reproductive development. All these expression patterns indicate that GASA family genes play conserved roles in different plants. However, since most GASA family members have not been functionally characterized, future work should focus on revealing the roles and underlying regulatory mechanisms of GASA family genes during different developmental processes and stress responses.
Although GASA family genes were shown to be involved in a wide range of developmental process, the functions of GASA genes in fruit ripening remain unknown. In addition to tomato, GASA family members have been identified in strawberry and apple, which also have f leshy fruits. In strawberry, two FaGASA family genes, FaGAST1 and FaGAST2, were associated with fruit ripening based on their expression during fruit ripening [20,21]. However, their functions were reported to be related to fruit size determination [20,21], and whether they are involved in fruit ripening remains elusive. Several MdGASA family members in apple, such as MdGASA3, MdGASA13, and MdGASA26, are also highly expressed in fruits [12], but their functional significance in fruit ripening remains unclear. Thus, our demonstration of the role and mode of action of SlGASA1 in fruit ripening in tomato extends our knowledge of the functions of GASA family members in f leshy fruits. Nonetheless, in addition to SlGASA1, the roles and underlying regulatory mechanisms of other SlGASA family members in fruit ripening requires further investigation, as several SlGASA genes displayed ripening-associated expression patterns, including SlGASA13 and SlGASA17, whose transcripts accumulated to high levels during ripening.
Identifying ripening regulatory genes and revealing their underlying mechanisms represent important steps toward dissecting the ripening regulatory networks of fruits. Our study identified a novel ripening gene and revealed the underlying mechanism of the encoded protein in regulating fruit ripening. Like most reported regulators that mediate ripening processes in an ethylene-dependent manner [33,34,42,44], SlGASA1 also regulates the ripening of tomato fruit by affecting ethylene biosynthesis. Interestingly, our results showed that SlGASA1 is unable to directly bind to the promoters of the ethylene biosynthesis genes ACS2 and ACO1. Instead, it recruits the key ripening transcription factor FUL1 and represses the activation of ACS2 and ACO1 transcription by FUL1. Our findings reveal a novel regulatory mechanism controlling fruit ripening. However, the molecular mechanisms underlying how the interaction between SlGASA1 and FUL1 represses the transcription of ethylene biosynthesis genes require further investigation.

Data collection and tomato GASA family identification
Protein sequences of tomato GAST1 (Solyc02g089350) and GASAs in Arabidopsis were obtained from the database of NCBI (http:// www.ncbi.nlm.nih.gov/). The tomato (SL4.0) reference genome sequence and annotations were obtained from the SGN database (https://solgenomics.net/).
Based on studies of GASA proteins in Arabidopsis, the conserved domain of GASA proteins was identified as the GASA domain (PF02704). We preliminarily identified 20 putative GASA proteins encoded in the tomato genome using a Hidden Markov model (HMM). We manually determined whether these putative GASA proteins contained a cleavable signal peptide in their N termini, ultimately characterizing 17 GASA proteins in tomato. In addition to Solyc02g089350 (SlGASA1), we named the 16 remaining SlGASA genes in tomato based on their chromosomal positions. The basic information including molecular weight (Mw) and isoelectric point (pI) of all SlGASA proteins were obtained by the online program Expasy.

Protein sequence alignment and phylogenetic tree construction
The alignment of the protein sequences of AtGASAs in Arabidopsis, OsGASAs in rice and SlGASAs in tomato was performed using MEGA X. The bad alignment regions of all protein sequences participating in topology structure construction were removed by trimAl tool [45]. A phylogenetic tree was then generated by the method of maximum-likelihood (ML) with poisson correction and 1000 bootstrap replicates in IQ-TREE [46]. The visualization of the phylogenetic tree was performed using Evolview (www.evolgenius.info/). We also built a phylogenetic tree using the sequences of tomato SlGASA proteins alone, and the results were consistent with the results obtained using the previous approach. The protein sequences of SlGASA family are provided in Data S2 (see online supplementary material).

Analyses of conserved motifs, conserved domains, and cis-acting elements in SlGASA promoters
The analysis of the conserved motifs of SlGASA proteins were performed by the program of MEME (5.05) (http://meme.nbcr.net/ meme/), and the conserved domains of tomato GASA proteins were predicted using Pfam (http://pfam.xfam.org/). The upstream 2.0 kb genomic DNA sequences from the translation initiation codon (ATG) were taken as the promoter regions of SlGASAs.
The cis-elements of the SlGASA promoter regions were detected by PlantCare with default parameters. Phytohormone-related ciselements are summarized in Data S1 (see online supplementary material). The visualization of the relevant data was conducted using TBTools [4].

Plasmid construction and tomato genetic transformation
To generate the E8:SlGASA1-pBin19 construct, the coding sequence of SlGASA1 was amplified and cloned into the vector of pDONR207, followed by mobilization into the Gateway vector pBin19. The primers used for construction of E8:SlGASA1-pBin19 are listed in Data S3 (see online supplementary material). The tomato genetic transformation mediated by Agrobacterium (A. tumefaciens) was performed as described previously [47]. Half-strength Murashige and Skoog medium with kanamycin (100 mg L −1 ) was used to select the transgenic tomato lines.

Subcellular localization
To generate the 35S:SlGASA1:GFP construct, we amplified the SlGASA1 coding sequence excluded the stop codon and inserted into a vector harboring 35S:GFP. The 35S:SlGASA1:GFP construct and the empty 35S:GFP vector were then transferred into Nicotiana benthamiana leaf protoplasts as previously described [48]. Following culture in the dark for 22 hours, the subcellular location was observed under a confocal laser scanning microscope.

Tomato growing conditions
The wild type and transgenic tomato plants (S. lycopersicum L. cv Micro-Tom) used in this study were grown in a greenhouse with the conditions descried in Deng et al. [49].

VIGS experiments
Regarding the VIGS experiments, WT tomato fruits were collected at 30 DPA and injected with 100 μL Agrobacterium culture containing the pTRV2-SlGASA1 construct or pTRV2 (control) into fruit tissues through the calyx scar. Seven days after injection, fruit pericarp samples were frozen by liquid nitrogen and stored at −80 • C.

RNA isolation and RT-qPCR
Tomato pericarp samples for RT-qPCR analysis were collected at different growth and developmental stages. A plant RNA extraction kit (BIOFIT, Chengdu, China) was used to isolate the total RNA from different tomato tissues. The procedure including removal of genomic DNA, cDNA generation, and quantitative PCR (qPCR) analysis were performed as described in Deng et al. [49]. The sequences of primers used for RT-qPCR are provided in Data S3 (see online supplementary material).

Transcriptome profiling
The RNA-seq data used for transcriptomic analysis were obtained from our previous publication [41] which is available in the Big Data Center's Genome Sequence Archive (http://bigd.big.ac.cn/) under accession numbers CRA001723 and CRA001712. Transcriptome profiling was conducted as described in our previous article [41], clean reads were mapped to the tomato reference genome (version 4.0) using HISAT2 [50] and then normalized to reads TPM (transcript per million).

Phytohormone treatments
To treat fruits with GA 3 or the GA biosynthetic inhibitor paclobutrazol (PAC), WT tomato fruits were collected at 32 DPA and injected through the calyx scar with 100 ppm GA 3 , 100 ppm PAC or distilled water (control) using a syringe. After 2 days, a group of GA 3 -treated fruits was treated with 100 ppm ethylene for 24 hours in a sealed jar; 20 fruits were used per treatment. The treated fruit samples were then stored at −80 • C until use.
For the treatments of ethylene and 1-MCP, WT fruits at the MG stage were treated with ethylene (100 ppm) for 12 hours or 1-MCP (40 ppm) for 16 hours in sealed jars. Treated fruit samples were stored at −80 • C.

Measurement of fruit color
WT and transgenic tomato fruits were collected at different ripening stages. Their color measurement was performed as described by Deng et al. [51].

Measurement of ethylene content
WT and transgenic tomato fruits at different ripening stages were picked and incubated at room temperature for 2 hours in open 150-mL jars. Each jar with a single fruit were sealed and then kept for 2 hours at room temperature, and 1 mL gas samples from headspace were collected with a syringe and analysed using an Agilent 7890B gas chromatograph. Ethylene content was quantified as describe in Deng et al. [49].

Carotenoid and chlorophyll measurements
WT and transgenic tomato fruits were collected at different ripening stages, and the extraction and measurement of carotenoid were conducted as described in Deng et al. [51].

Yeast two-hybrid assay
The full-length coding sequences of RIN, NOR, EIL1-4, TAG1, and FUL1 were cloned into pGADT7 for generation of the prey constructs. The coding sequence of SlGASA1 was inserted into pGBKT7 as the bait construct. Yeast AH109 cells were cotransformed with different combination of the bait and prey constructs and grown on SD (synthetic defined) medium without Leu and Trp (SD-Leu-Trp) for 2-4 days. Interactions were tested based on the growth of yeast cultures on selection medium (SD-Leu-Trp-His-Ade).

Dual-luciferase reporter assay
The vectors used for this transient transactivation assay were generated according to the GoldenBraid2.0 cloning strategy. Constructs of 35S:SlGASA1 and 35S:FUL1 were used as effectors, while promoter sequences of ACO1 and ACS2 were individually cloned upstream of the Luciferase (LUC) coding sequence to serve as the reporters. Renilla luciferase gene was used as an internal control. The Dual-Luciferase reporter assay system (Promega, Madison, WI, USA) was used to test the ratio of LUC to REN.

Co-IP assay
The CDS of SlGASA1 or FUL1 was cloned into the pBTEX-FLAG or pBTEX-HA vector to generate the FUL1-FLAG or SlGASA1-HA overexpression construct, respectively. SlGASA1-HA and FUL1-FLAG were co-expressed in N. benthamiana leaves through infiltration mediated by Agrobacterium, and FUL1-FLAG and FUL1-FLAG were co-infiltrated in N. benthamiana leaves to serve as a control. Total proteins were obtained from the N. benthamiana leaves at 48-72 hours after Agrobacterium infiltration and incubated with anti-FLAG M2 magnetic beads (Sigma, M8823) for 12-16 h at 4 • C. The isolated proteins were examined by immunoblot analysis using anti-HA (CST, #2367) and anti-FLAG (CST #14793) antibodies.