Abstract The KNOTTED1-LIKE HOMEOBOX PROTEIN1 (SlKD1) is a master abscission regulator in tomato (Solanum lycopersicum). Here, we identified an SlKD1-interacting transcription factor GATA transcription factor 6 (SlGATA6), which is required for maintaining the auxin-response gradient and preventing abscission. SlGATA6 up-regulates the expression of SlLAX2 and SlIAA3. The AUXIN RESISTANT/LIKE AUXIN RESISTANT (AUX/LAX) proteins SlLAX2-dependent asymmetric auxin distribution causes differential accumulation of Auxin/Indole-3-Acetic Acid 3 (SlIAA3) and its homolog SlIAA32 across different abscission zone cells. It is also required for SUMOylation of AUXIN RESPONSE FACTOR 2a (SlARF2a), a key suppressor of auxin signaling and abscission initiator. Moreover, SlIAA3 and SlIAA32 depress SUMOylated SlARF2a, thus suppressing SlARF2a function. The interaction between SlKD1 and SlGATA6 suppresses SlGATA6 binding to the promoters of SlLAX2 and SlIAA3, thereby disrupting the auxin-response gradient and triggering abscission. This regulatory mechanism is conserved under low light–induced abscission in diverse Solanaceae plants. Our findings reveal a critical role of SlKD1 in modulating the auxin-response gradient and abscission initiation. __________________________________________________________________ SlGATA6 fine-tunes the auxin-response gradient to prevent abscission. INTRODUCTION In plants, abscission usually takes place in a specific site called the abscission zone (AZ) ([52]1, [53]2). Auxin is an important inhibitor of abscission. Flowers and fruits are rich sources of auxin, which are transferred from these organs to other regions via AZ. This polar transport of auxin forms an auxin gradient in the AZ and thus inhibits abscission ([54]3–[55]5). Impairment of the continuous polar flow of auxin through the AZ abolishes the auxin-response gradient, resulting in ethylene sensitivity in the AZ cells and initiating abscission ([56]6–[57]9). The initiation of abscission can be triggered by developmental signals or environmental stimuli, such as darkness ([58]10), low light ([59]11, [60]12), and drought ([61]13); however, the mechanism underlying the interruption of the auxin-response gradient to initiate abscission is still obscure. Interrupting the auxin flux by using the auxin transport inhibitor N1-naphthylphthalamic acid notably accelerates abscission ([62]14). The polar auxin transporters AUXIN RESISTANT/LIKE AUXIN RESISTANT (AUX/LAX) and PIN-FORMED (PIN) are responsible for auxin influx and efflux, respectively ([63]15–[64]17). Down-regulating the expression of specific AUX/LAX or PIN genes causes premature abscission ([65]7, [66]18). The local auxin response in AZ cells is mediated by auxin signaling. The auxin-promoted degradation of Aux/IAAs (AUXIN/indole-3-acetic acids) releases the repression imposed on AUXIN-RESPONSIVE FACTOR (ARF) proteins to transcriptionally activate or repress downstream auxin-responsive genes. Different Aux/IAA-ARF modules are involved in various aspects of development ([67]19–[68]21). The core auxin regulatory genes involved in abscission have largely been identified. For example, in Arabidopsis (Arabidopsis thaliana), loss-of-function mutants of ARF2 exhibit delayed abscission, suggesting that ARF2 plays a dominant role in initiating abscission ([69]22, [70]23). Furthermore, microarray assays in tomato indicated that ARF expression levels did not change during pedicel abscission induced by manual removal of flowers, suggesting that abscission is regulated in an Aux/IAA expression-dependent mechanism ([71]3). In agreement with this idea, the expression levels of six Aux/IAA genes in red cestrum (Cestrum elegans) were negatively correlated with floret abscission ([72]24). Overexpression of axr3-1, a gain-of-function, semidominant allele of IAA17, notably delayed floral organ abscission in Arabidopsis ([73]7). Knocking down RhIAA16 transcript levels by virus-induced gene silencing (VIGS) notably accelerated petal abscission in rose (Rosa hybrida), suggesting that Aux/IAAs prevent premature abscission ([74]25). In plants, C2C2-GATA-transcription factors are evolutionarily conserved transcription factors ([75]26). The GATA genes of tomato (Solanum lycopersicum), Arabidopsis, and rice (Oryza sativa) can be divided into four subfamilies (I to IV) ([76]26, [77]27). Members from subfamily I have been reported to participate in auxin signaling. For instance, Arabidopsis GATA2 restricts cell division involved in auxin-mediated root elongation, and the overexpression of its encoding gene decreased β-glucuronidase (GUS) activity derived from a DR5:GUS transgene in roots by down-regulating the expression of PIN1 and a suite of auxin-response genes ([78]28). Posttranscriptional regulation also modulates GATA function. For example, darkness induces the proteasomal degradation of Arabidopsis GATA2 in a CONSTITUTIVE PHOTOMORPHOGENIC 1 E3 ubiquitin ligase–dependent manner to prevent photomorphogenesis ([79]29). In tomato, the basic leucine zipper transcription factor LONG HYPOCOTYL 5 interacts with SlGATA17 and suppresses SlGATA17 expression to regulate seed germination during stress ([80]30). However, whether GATAs contribute to auxin signaling in the context of plant abscission and their underlying regulatory mechanisms are still unknown. KNOX proteins comprise a small family of homeobox proteins with three–amino acid loop extensions and can be divided into three subclasses: I, II, and M ([81]31). In tomato, three KNOX genes, TOMATO KNOTTED3 (TKN3), TNK4, and KNOTTED1-LIKE HOMEOBOX PROTEIN1 (KD1), are strongly expressed in the AZ of pedicels connecting the fruits and the plant ([82]32). In our previous study, we reported that several transcription factors, WUSCHEL (WUS), KD1, and FRUTFULL 2 (FUL2), participate in low light– and auxin depletion–induced abscission ([83]12). Genetic evidence indicated that SlWUS acted upstream of SlKD1 to regulate these two types of abscissions. Several studies have demonstrated that SlKD1 modulates the auxin concentration and response gradient in the AZ and plays a critical role in initiating abscission ([84]12, [85]32, [86]33). However, the class M KNOX protein SlKD1 lacks a clear DNA-binding homeodomain, suggesting that SlKD1-mediated abscission might depend on other proteins that bind to DNA. Here, we identified SlGATA6 as interacting with SlKD1; this interaction suppresses the binding of SlGATA6 to the SlLAX2 and SlIAA3 promoters and inhibits their transcription. We provide evidence that the auxin gradient, established by SlLAX2, stabilizes the noncanonical AUX/IAA protein SlIAA32, which interacts and inhibits SlARF2a, a major player in abolishing the auxin-response gradient across the AZ, thus accelerating abscission, together with SlIAA3. Together, our results describe a molecular mechanism whereby auxin depletion– and low light–induced SlKD1 expression breaks the auxin-response gradient and initiates abscission. RESULTS SlKD1 directly interacts with SlGATA6 We previously showed that SlKD1 influences the auxin-response gradient and auxin concentration to enhance abscission inside the AZ ([87]12, [88]32). However, because SlKD1 lacks a DNA-binding homeodomain, SlKD1-mediated abscission is likely to require other proteins that bind to DNA. To identify the possible regulatory proteins interacting with SlKD1, we performed a yeast two-hybrid (Y2H) screen using a cDNA library prepared from total RNA extracted from tomato flower AZs at different times following flower removal (0, 2, 4, 8, 12, and 16 hours). Among 17 independent candidate interactors, we noticed a GATA transcription factor, SlGATA6 ([89]Fig. 1A and data S1). Phylogenetic analysis suggested that SlGATA6 belongs to subfamily I and is most closely related to Arabidopsis GATA6 (fig. S1A). A subcellular localization assay indicated that a SlGATA6–green fluorescent protein (GFP) fusion accumulates in the nucleus (fig. S1B). To determine which regions of SlKD1 and SlGATA6 interact in yeast, we divided SlKD1 into two fragments and SlGATA6 into three based on their conserved functional domains ([90]Fig. 1B). We observed an interaction between the C terminus of SlKD1 containing the KNOX2 domain and the C-terminal ZnF domain of SlGATA6 ([91]Fig. 1C). We validated this interaction in vitro by performing a pull-down assay with recombinant full-length SlKD1-His and SlGATA6 fused to glutathione S-transferase (SlGATA6-GST), as well as with truncated SlKD1[C]-His and SlGATA6[C]-GST ([92]Fig. 1, D and E). To validate their interaction in vivo, we performed coimmunoprecipitation (co-IP) assays by coexpressing SlKD1-Flag (encoding SlKD1 with a Flag tag) and SlGATA6-GFP in Nicotiana benthamiana leaves, which revealed that SlGATA6 associates with SlKD1 in plant cells ([93]Fig. 1F). A bimolecular fluorescence complementation (BiFC) assay indicated that the SlKD1-SlGATA6 interaction occurs in the nucleus. Deleting the C-terminal region of SlKD1 (SlKD1[ΔC]) or the C terminus of SlGATA6 (SlGATA6[ΔC]), which were both critical for their interaction in yeast, abolished the SlKD1-SlGATA6 interaction in plant cells ([94]Fig. 1G). These results indicate that SlGATA6 directly interacts with SlKD1. Fig. 1. SlGATA6 interacts with SlKD1 and inhibits abscission. [95]Fig. 1. [96]Open in a new tab (A) Y2H assay showing that SlKD1 interacts with SlGATA6. SlKD1 was cloned into the pGBKT7 vector [with GAL4 DNA-binding domain (BD)]; SlGATA6 was ligated into the pGADT7 vector [with GAL4 activation domain (AD)]. DDO, double dropout medium (synthetic defined medium lacking Trp and Leu). QDO, quadruple dropout medium (synthetic defined medium lacking Trp, Leu, His, and Ade). X, X-a-gal. AbA, Aureobasidin A. (B) Diagrams of full-length and various truncated forms of SlKD1 and SlGATA6 used in (C), (E), and (G). KNOX, KNOTTED1-like homeobox domain. G1 to G4, GATA low complexity domains. ZnF-GATA, zinc-finger DNA-binding domain. Numbers denote amino acids. (C) Y2H assay showing that SlKD1[C] directly interact with SlGATA6[C]. (D) Pull-down assays of the interaction between full-length SlKD1 and SlGATA6. (E) Pull-down assays for SlKD1 and SlGATA6 fragments interaction. (F) Immunoprecipitation assay showing the interaction of SlKD1 with SlGATA6 in vivo. SlKD1-Flag was coexpressed with SlGATA6-GFP or GFP in N. benthamiana leaves. Protein complexes were immunoprecipitated using GFP beads and analyzed by immunoblot. (G) BiFC assays showing the interaction between SlKD1 and SlGATA6 in planta. DAPI, 4′,6-diamidino-2-phenylindole; YFP, yellow fluorescent protein. Scale bars, 25 μm. (H) RNA in situ hybridization for SlGATA6 transcripts in flower pedicels. The sense probe was used as a negative control. AZ, abscission zone. Scale bars, 200 μm. (I) Number of hours required to achieve 50% pedicel abscission in the WT, Slgata6, and SlGATA6-OE plants. Boxplots present data of six independent tests with at least 15 pedicels in each. Significant differences were determined by one-way analysis of variance (ANOVA) with Dunnett’s test compared to the WT; ****P < 0.0001. Box plot shows maxima, first quartile, median, third quartile, and minima. h, hours. SlGATA6 plays a negative role in flower pedicel abscission A reverse-transcription quantitative PCR (RT-qPCR) assay indicated that SlGATA6 is abundantly expressed in the AZ of flower pedicels at time 0 hours before flower removal (fig. S1C); we obtained independent support of this claim by RNA in situ hybridization using a specific antisense probe for SlGATA6 transcripts ([97]Fig. 1H). SlGATA6 expression decreased quickly before remaining at a low level following flower removal, which induces pedicel abscission caused by auxin depletion (fig. S1D). To understand the function of SlGATA6 in abscission, we generated SlGATA6 loss-of-function mutants using CRISPR-Cas9 gene editing and SlGATA6 overexpression (SlGATA6-OE) lines (fig. S1, E to G). We observed accelerated abscission of the pedicel in three independent Slgata6 mutant lines and delayed abscission in SlGATA6-OE lines compared to wild-type (WT) plants after flower removal ([98]Fig. 1I). These results confirm that SlGATA6 inhibits flower pedicel abscission. RNA-seq analysis shows that SlGATA6 regulates auxin-related pathways To investigate the mechanisms underlying the accelerated abscission in the Slgata6 mutants, we explored differential gene expression in the AZ at 4 hours after flower removal between WT and Slgata6 plants by transcriptome deep sequencing [RNA sequencing (RNA-seq)]. We identified 1316 significantly differentially expressed genes [DEGs; absolute |log[2] fold-change| ≥ 1, false discovery rate (FDR) <0.01] in Slgata6 plants compared to WT plants, of which 429 were up-regulated genes and 887 were down-regulated genes (fig. S2A and data S2). A Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis to classify the top 10 significant pathways enriched in DEGs determined that the “plant hormone signal transduction pathway” is the most enriched ([99]Fig. 2A). A closer look at the enriched DEGs in the plant hormone signal transduction pathway revealed that these genes are responsible for auxin homeostasis and signaling (26 out of 33, data S3). The down-regulated genes encode 13 AUX/IAA proteins, one ARF-type transcription factor, three SMALL AUXIN-UPREGULATED RNAs (SAURs), one auxin transporter (LAX2), and five GRETCHEN HAGEN3s (GH3s); genes encoding a SAUR and two GH3s were up-regulated ([100]Fig. 2B). To validate the data from the RNA-seq analysis, we measured the relative transcript levels of six auxin-related genes using RT-qPCR (fig. S2B). The changes in gene expression seen using RNA-seq and RT-qPCR were consistent for all six genes, confirming the validity of the RNA-seq data. Fig. 2. SlGATA6 positively and directly regulates the expression of auxin-related genes. [101]Fig. 2. [102]Open in a new tab (A) KEGG pathway enrichment analysis of DEGs as determined by RNA-seq analysis of AC and Slgata6 AZs. The numbers in parentheses mean P values. (B) Heatmap representation of auxin-related DEGs in WT and Slgata6. Relative mRNA expression levels (reads per kilobase per million reads) in RNA-seq data from two biological replicates were used for the analysis using TBtools ([103]62). (C) Sequence logo showing the SlGATA6 DNA-binding motif. (D) Venn diagram showing the extent of overlap between SlGATA6-bound genes and up-regulated and down-regulated genes in Slgata6 AZs. (E) Genome browser windows of SlGATA6-bound loci SlLAX2 and SlIAA3, showing the distribution of reads obtained by DAP-seq. The DAP-seq reads are color coded: green denotes the input reads; red and blue, the two replicates for SlGATA6-bound reads; the peak position is indicated by the red shaded area. (F) EMSAs using probes from the SlLAX2 and SlIAA3 promoters demonstrating that recombinant SlGATA6-GST binds to the SlIAA3 and SlLAX2 promoters. Red letters indicate the SlGATA6 binding motif identified by DAP-seq or its mutated variant used in the mutant probes. (G) Promoter assays showing that SlGATA6 activates the transcription of SlIAA3 and SlLAX2. SlIAA3 and SlLAX2 promoters were cloned upstream of the GUS reporter gene. N. benthamiana leaves were coinfiltrated with the GUS reporter constructs containing either the intact SlLAX2 or SlIAA3 promoters (pro) or mutant variants (mpro) harboring mutations in the SlGATA6-binding site and the 35S:SlGATA6 effector construct or empty vector (control). The values are means ± SD from three biological replicates. One-way ANOVA with Tukey’s test was used to assess statistical significance relative to 35S:empty; ****P < 0.0001; ns, no significant difference. SlGATA6 regulates auxin-related genes by binding to their promoters To define the specific target genes directly regulated by SlGATA6, we used DNA affinity purification sequencing (DAP-seq), which is an in vitro DNA-binding assay used to globally capture the DNA-binding sites of a given transcription factor in the genomic context ([104]34). We obtained 34,707 highly reliable SlGATA6-binding peaks, of which ~33.2% were located upstream from transcription start sites (≤2.5 kb upstream of the start codon, fig. S2C). KEGG enrichment analysis established that these putative SlGATA6 target genes are mainly enriched in pathways related to “biosynthesis of secondary metabolites” and “plant hormone signal transduction” (fig. S2D). We identified a total of 19,096 highly reliable SlGATA6-binding sites located within the regulatory regions of 5400 genes (data S4). The consensus sequence for the SlGATA6 binding motif was GAT(G/C)CAGAT(C/G)(C/T) ([105]Fig. 2C). Notably, this motif includes the core sequence GATC, a previously identified GATA6 recognition motif in Arabidopsis ([106]35). We compared the 1316 DEGs derived from the RNA-seq analysis to the above 5400 SlGATA6-bound genes from our DAP-seq data to identify those direct SlGATA6-regulated target genes. In total, 271 genes (175 down-regulated and 96 up-regulated) were shared by the DAP-seq and RNA-seq datasets, suggesting that they are direct targets of SlGATA6 ([107]Fig. 2D and data S5); about 20.6% of the DEGs (271 of 1316) contained the SlGATA6-binding site in their regulatory regions. Among the 271 overlapping genes, we focused on 4 genes involved in the auxin pathway: 3 AUX/IAA genes (SlIAA3, SlIAA11, and SlIAA13) and 1 auxin transporter gene (SlLAX2) ([108]Fig. 2E and fig. S3, A and B), down-regulated in Slgata6 plants and whose promoters were bound by SlGATA6 ([109]Fig. 2B). To validate the binding of SlGATA6 to the promoters of these four auxin-related genes, we turned to electrophoretic mobility shift assays (EMSAs). Recombinant SlGATA6-GST showed direct binding to probes derived from all four auxin-related genes and containing the predicted SlGATA6-binding site ([110]Fig. 2F and fig. S3, C and D). To determine how SlGATA6 regulates these four genes, we used a β-GLUCURONIDASE (GUS) transactivation assay in N. benthamiana leaves. When the 35S:SlGATA6 effector construct was coinfiltrated with the intact promoter reporter constructs for SlLAX2, SlIAA3, SlIAA11, or SlIAA13, we observed an increase in relative GUS activity, but not with mutated versions (all bases mutate into A) of these promoters with a mutation in the predicted SlGATA6-binding site ([111]Fig. 2G and fig. S3E). These findings demonstrate that SlGATA6 activates the transcription from the SlLAX2, SlIAA3, SlIAA11, and SlIAA13 promoters. SlLAX2 delays abscission by affecting auxin concentration across AZ To explore their effects on abscission, we used VIGS to produce individual tomato plants knocked down for SlLAX2, SlIAA3, SlIAA11, or SlIAA13. Silencing SlLAX2 or SlIAA3 resulted in accelerated abscission, while silencing SlIAA11 or SlIAA13 had no significant effect on abscission (fig. S4). To eliminate the redundant functions of SlIAAs, we conducted concurrent silencing of SlIAA3 and SlIAA11, SlIAA3 and SlIAA13, SlIAA11 and SlIAA13, and also silenced all three SlIAAs simultaneously to investigate their functional roles. Only SlIAA3 was involved in abscission (fig. S5). We further detected the expression levels of SlLAX2 and SlIAA3 in WT and SlGATA6 mutants by RT-qPCR and found that the expression levels of SlLAX2 and SlIAA3 were significantly down-regulated in mutants compared with WT (fig. S6). Therefore, we chose SlLAX2 and SlIAA3 for further characterization. We first explored the roles of SlLAX2 in establishing the auxin concentration in the AZ, as LAXs are involved in local auxin accumulation ([112]15). RT-qPCR analysis and RNA in situ hybridization assays indicated that SlLAX2 transcripts are abundant in the AZ at time 0 hours before flower removal (fig. S7, A to C). Moreover, SlLAX2 expression sharply decreased following flower removal (fig. S7D), similar to SlGATA6 expression (fig. S1D). When we generated knockout mutant lines for SlLAX2 using CRISPR-Cas9 and SlLAX2 overexpression lines (fig. S7, E to G), we determined that Sllax2 plants exhibit significantly accelerated abscission, while SlLAX2-OE lines had significantly delayed abscission compared to the WT ([113]Fig. 3A). Knockout of SlLAX2 significantly decreased the auxin concentration in the AZ compared to that in the WT; by contrast, the auxin concentration was significantly higher in SlLAX2-OE lines than in the WT ([114]Fig. 3B). To examine auxin signaling output in the pedicel, we used DR5:GUS lines in which GUS is driven by the auxin-responsive synthetic promoter element DR5 ([115]12, [116]32, [117]36). We introduced the DR5:GUS transgene into the Sllax2 lines by crossing and stained the pedicels of Sllax2 DR5:GUS plants to reveal GUS activity. We detected little GUS staining in the region proximal to the AZ or within the AZ itself, in contrast to DR5:GUS plants in the AC background, indicating that loss of SlLAX2 function results in a lower auxin signaling output across the pedicel ([118]Fig. 3C). These results indicate that SlLAX2-dependent auxin influx is important for auxin accumulation in the AZ, which inhibits abscission. Fig. 3. SlLAX2, SlIAA3, and SlIAA32 suppress tomato pedicel abscission. [119]Fig. 3. [120]Open in a new tab (A) Number of hours required for WT, Sllax2, and SlLAX2-OE plants to achieve 50% pedicel abscission. (B) Auxin contents in the AZ of WT, Sllax2, and SlLAX2-OE plants. The values are means ± SD from three biological replicates. (C) GUS staining pattern derived from DR5:GUS reporter in the AZ of WT, Sllax2, Sliaa32, and Sliaa3 plants. Transverse sections were made from distal region, distal side of the AZ (DAZ), proximal side of the AZ (PAZ), and proximal region, as indicated. Images were digitally extracted for comparison. Three independent plant specimens were used, each with at least 10 pedicels. Scale bars, 1 mm. (D) Cell-free degradation test to examine the influence of naphthaleneacetic acid (NAA) or l-kynurenine (l-Kyn) treatment on SlIAA32 degradation. Equal amounts of recombinant SlIAA32-His were incubated with total protein extracts from WT AZs treated with water (control), NAA, or l-Kyn for 30 or 60 min. The presence of SlIAA32 was determined using an anti-His antibody. Actin as a loading control. (E) In vivo degradation assay of SlIAA32 in WT, SlLAX2-OE, and Sllax2 plants. The pedicel AZs were treated with 100 μM cycloheximide (CHX) for different times. (F to I) Number of hours required for the WT and Sliaa32 lines (F), SlLAX2-OE and Sliaa32 SlLAX2-OE (G), WT and Sliaa3 lines (H), and Sliaa3, Sliaa32, and Sliaa3 Sliaa32 (I) plants to achieve 50% pedicel abscission. Significant differences were determined by one-way ANOVA with Tukey’s test [(A), (G), (H), and (I)] or Dunnett’s test [(B) and (F)]; *P < 0.05, ***P < 0.001, ****P < 0.0001; different lowercase letters indicate significant differences (P < 0.05). Boxplots show data from six independent replicates, each consisting of at least 15 pedicels, with maxima, first quartile, median, third quartile, and minima. h, hours. SlLAX2-dependent auxin concentration stabilizes SlIAA32 to maintain auxin-response gradient Aux/IAA proteins commonly harbor four conserved amino acid sequence patterns, referred to as domains I, II, III, and IV. The F-box protein TRANSPORT INHIBITOR RESPONSE 1 (TIR1) interacts with domain II of Aux/IAAs in an auxin-dependent manner and leads to the destabilization of Aux/IAA proteins ([121]37, [122]38). Noncanonical AUX/IAAs lack the canonical domain II and cannot be degraded by the TIR1 complex. However, high concentrations of auxin can stabilize noncanonical AUX/IAAs to mediate a local auxin response ([123]20, [124]38). High concentrations of auxin activate TRANSMEMBRANE KINASE 1 (TMK1) at the cell surface and lead to the cleavage of the TMK1 C terminus in Arabidopsis ([125]20). Phosphorylation of the noncanonical Aux/IAAs IAA32 and IAA34 by the TMK1 C-terminal domain enhances IAA32 and IAA34 protein stability and plays an important role in apical hook formation ([126]20). SlIAA32 and SlIAA33 are noncanonical AUX/IAAs in tomato ([127]39), but only SlIAA32 was highly expressed in the AZ (fig. S8A). RNA in situ hybridization indicated that SlIAA32 transcripts are abundant in the AZ (fig. S8, B and C). Furthermore, cell-free degradation assays indicated that auxin treatment enhances the stability of SlIAA32, while application of the auxin biosynthesis inhibitor l-kynurenine (l-Kyn) decreased its stability ([128]Fig. 3D). Relative SlIAA32 transcript levels were similar in WT, SlLAX2-OE, and Sllax2 plants (fig. S8D); however, SlIAA32 accumulated to higher levels in SlLAX2-OE lines and to lower levels in Sllax2 lines compared to the WT, indicating that SlIAA32 may be more stable in SlLAX2-OE lines and less stable in Sllax2 lines ([129]Fig. 3E). Knockout of SlIAA32 significantly accelerated abscission compared to that in the WT (fig. S8E and [130]Fig. 3F). The auxin-response gradient was much weaker in Sliaa32 mutant lines than in the WT, as indicated by the DR5:GUS reporter ([131]Fig. 3C). In addition, while exogenous application of the synthetic auxin naphthaleneacetic acid (NAA) to treat the flower pedicel of WT and Sliaa32 mutant lines, we found that Sliaa32 mutants were less sensitive to auxin treatment, reaching a higher abscission rate than WT at the same period (fig. S8F). We crossed SlLAX2-OE to Sliaa32 and established that Sliaa32 SlLAX2-OE plants display increased abscission compared to SlLAX2-OE ([132]Fig. 3G). These results demonstrate that SlLAX2-dependent auxin concentration functions upstream of SlIAA32 to maintain the auxin-response gradient across the AZ and inhibit abscission. SlIAA3 delays abscission and functions independently of SlIAA32 We explored the roles of SlIAA3 in the AZ, as AUX/IAAs are involved in auxin signaling. RT-qPCR analysis and RNA in situ hybridization assays indicated that SlIAA3 is abundantly expressed in the AZ at time 0 hours before flower removal (fig. S9, A to C), with SlIAA3 expression sharply decreasing following flower removal (fig. S9D). We knocked out SlIAA3 by CRISPR-Cas9 and found that, compared to the WT, knockout of SlIAA3 significantly accelerates abscission ([133]Fig. 3H and fig. S9E). Sliaa3 DR5:GUS plants also showed a severely impaired auxin-response gradient with almost no GUS staining detected in the AZ or proximal regions ([134]Fig. 3C). These results indicate that SlIAA3-dependent auxin signaling is important for forming the auxin-response gradient to prevent abscission. To further explore the relationship between SlIAA32 and SlIAA3 in mediating abscission, we generated the Sliaa3 Sliaa32 double mutant by genetic crossing of two single mutants. Sliaa3 Sliaa32 plants had higher abscission levels than the respective Sliaa3 and Sliaa32 single mutants, indicating that SlIAA32 and SlIAA3 nonredundantly prevent abscission ([135]Fig. 3I). SlIAA3/SlIAA32 delay abscission by directly interacting with SUMOylated SlARF2a Aux/IAAs primarily block the function of ARFs in the auxin signaling cascade ([136]37, [137]40, [138]41). To determine which ARFs are repressed by SlIAA3 and SlIAA32 during tomato flower pedicel abscission, we assessed SlARF protein abundance in the AZ using liquid chromatography–tandem mass spectrometry (LC-MS/MS). We detected four SlARFs (SlARF2a, SlARF4, SlARF6, and SlARF14) in the AZ (data S8). To determine their potential effect on abscission, we used VIGS to knock down the transcript levels of each SlARF gene individually and monitored the abscission rate in the resulting knockdown plants. Only the SlARF2a-VIGS plants showed delayed abscission relative to the WT, whereas the other VIGS plants had no effect (fig. S10). RT-qPCR analysis and RNA in situ hybridization assays indicated that SlARF2a is abundantly expressed in the AZ at time 0 hours before flower removal (fig. S11, A to C). Furthermore, knockout of SlARF2a by CRISPR-Cas9 significantly delayed abscission, while SlARF2a overexpression lines showed accelerated abscission ([139]Fig. 4A and fig. S11, D to F), demonstrating that SlARF2a promotes abscission. SlARF2a is a core positive regulator of ethylene signaling during tomato fruit ripening ([140]42). We found that after ethylene treatment, compared with WT plants, the tomato pedicel explants of Slarf2a plants exhibited a significant delay in abscission (fig. S11G), indicating that SlARF2a is the key factor for the AZ to respond to ethylene and promote abscission. We also introduced the DR5:GUS reporter into SlARF2a-OE lines by crossing and found that SlARF2a overexpression strongly impairs the auxin-response gradient (fig. S11H), similar to the Sliaa3 and Sliaa32 lines ([141]Fig. 3C), indicating that SlARF2a is a powerful auxin-response depressor. On the basis of these results, we speculate that SlIAA3 and SlIAA32 inhibit abscission by repressing SlARF2a function; however, SlIAA3 and SlIAA32 failed to interact with SlARF2a in a Y2H assay (fig. S11I). Fig. 4. SlIAA3 and SlIAA32 interact with SlARF2a in a SUMO-dependent manner to delay abscission. [142]Fig. 4. [143]Open in a new tab (A) Number of hours required for WT, Slarf2a, and SlARF2a-OE plants to achieve 50% abscission. (B) Diagram of SlARF2a domains and three predicted SUMOylated sites. (C) Diagrams of SlIAA3 and SlIAA32 showing putative SIMs. (D) Transient expression showing that mutating all SlARF2a SUMOylatable lysine sites to arginine residues in SlARF2a-Flag (3*K/R) inhibits SUMOylation with SUMO1-HA but not for WT SlARF2a or single K104R mutant. (E) Number of hours required for WT, Slarf2a #6, Slarf2a #6 gSlARF2a (WT), and Slarf2a #6 gSlARF2a (3*R/K) plants to achieve 50% abscission. (F) Immunoprecipitation (Anti-Flag) reveals that auxin treatment induces SUMOylation of SlARF2a-Flag but not SlARF2a-Flag (3*K/R), while auxin depletion decreases SUMOylation. FR, flower removal. (G) Immunoprecipitation (Anti-Flag) showing that knockout of SlLAX2 decreases SlARF2a SUMOylation. (H and I) Transient expression of SlIAA3 (WT-SIM) or SlIAA3 (SIM mutant) (H) or SlIAA32 (WT-SIM) or SlIAA32 (SIM mutant) (I) with SlARF2a-Flag or SlARF2a-Flag (3*K/R), followed by immunoprecipitation and immunoblot analysis revealing that SlIAA3 and SlIAA32 interact with SlARF2a in SIM- and SUMO-dependent manner. (J to O) Number of hours required for WT and SlIAA3-OE (J); WT and SlIAA32-OE (K); WT, Sliaa3, and SlIAA3dIImSIMm-OE Sliaa3 (L); WT, Sliaa32, and SlIAA32SIMm-OE Sliaa32 (M); WT, Sliaa3, and Sliaa3 Slarf2a (N); and WT, Sliaa32, and Sliaa32 Slarf2a (O) plants to achieve 50% abscission. dIIm, mutated domain II; SIMm, mutated SIM domain. Actin as the loading control. Significant differences were determined by one-way ANOVA with Dunnett’s [(A), (E), (J), and (K)] and Tukey’s [(L) and (M)] test; **P < 0.01; ***P < 0.001; ****P < 0.0001; different lowercase letters indicate significant differences (P < 0.05). Boxplots show data from six independent tests, each consisting of at least 15 pedicels with maxima, first quartile, median, third quartile, and minima. h, hours. In Arabidopsis, the interaction between ARF7 and IAA3 is dependent on the small ubiquitin-like modifier molecule SUMO for the regulation of root branching toward water ([144]43). Bioinformatics analysis revealed that SlARF2a contains three putative SUMOylation sites, namely, at lysine residues K104, K686, and K632 ([145]Fig. 4B); in addition, SlIAA3 and SlIAA32 contain a SUMO-interaction motif (SIM) ([146]Fig. 4C). We therefore tested whether SlIAA3 and SlIAA32 interact with SlARF2a in a SUMO-dependent manner. We confirmed that SlARF2a is a target for SUMOylation by coexpressing SlARF2a-Flag and SUMO-HA (encoding SUMO with a HA tag) in N. benthamiana leaves; mutating the three potential SlARF2a SUMOylatable motifs from lysine to arginine (non-SUMOylatable SlARF2a^3K/R) prevented the addition of SUMO onto SlARF2a ([147]Fig. 4D). To test the effect of SlARF2a SUMOylation on the auxin-response gradient and abscission, we expressed SUMOylatable SlARF2a and non-SUMOylatable SlARF2a^3K/R in the Slarf2a background under the control of the SlARF2a promoter. SlARF2a^3K/R had a greater ability to impair the auxin-response gradient and induce abscission than intact SUMOylatable SlARF2a ([148]Fig. 4E). Using SlARF2a-Flag plants, we determined that flower removal represses SlARF2a SUMOylation, while exogenous application of NAA promoted SlARF2a SUMOylation ([149]Fig. 4F). In agreement with this result, we detected much less SUMOylated SlARF2a in Sllax2 than in the WT ([150]Fig. 4G). These results indicate that SlLAX2-dependent auxin concentration is important for SlARF2a SUMOylation, which is required for maintaining the auxin-response gradient and preventing abscission. Co-IP assays in N. benthamiana leaves indicated that SUMOylated SlARF2a interacts with SlIAA3 and SlIAA32, while non-SUMOylatable SlARF2a^3K/R failed to do so ([151]Fig. 4, H and I). Coexpressing SlIAA3 or SlIAA32 in combination with SlARF2a-Flag and SUMO-HA in N. benthamiana does not affect the SUMOylation of SlARF2a (fig. S11J). Furthermore, mutant variants of SlIAA3 and SlIAA32 with mutations in the SIM domain no longer interacted with SlARF2a ([152]Fig. 4, H and I), underscoring the importance of SlARF2a SUMOylation in the SlARF2a-SlIAA3/SlIAA32 interaction. As domain II is responsible for Aux/IAA stability, and an amino acid substitution in domain II can stabilize the protein ([153]20, [154]38), we generated a dexamethasone-inducible transgenic line for inducible expression of SlIAA3[dIIm] (with the domain II of SlIAA3 mutated) ([155]44), SlIAA3[dIImSIMm] (with the domain II and SIM domains of SlIAA3 mutated), SlIAA32, and SlIAA32[SIMm] (with the SIM domain of SlIAA32 mutated) ([156]43) (fig. S11, K to N). While mutating the SIM domain of SlIAA3 and SlIAA32 had little effect on abscission, the dexamethasone-induced expression of SlIAA3[dIIm] or SlIAA32 delayed abscission ([157]Fig. 4, J and K). Moreover, overexpression of SlIAA3[dIImSIMm]/SlIAA32[SIMm] failed to restore the mutant phenotype in the backgrounds of Sliaa3 and Sliaa32 mutants ([158]Fig. 4, L and M). All these results demonstrate the importance of the SIM domain in inhibiting abscission. We then crossed Slarf2a to Sliaa3 DR5:GUS and Sliaa32 DR5:GUS plants and stained for GUS activity in pedicels. Knockout of SlARF2a in the Sliaa3 and Sliaa32 single mutant backgrounds strongly enhanced the auxin response across the AZ (fig. S11H) and resulted in a delayed abscission phenotype ([159]Fig. 4, N and O). These results indicate that SlIAA3 and SlIAA32 negatively regulate SlARF2a function by interacting with SUMOylated SlARF2a to maintain the auxin-response gradient and prevent abscission. SlGATA6 establishes auxin-response gradient across AZ To explore the role of SlGATA6 in auxin homeostasis and signaling during abscission, we measured the auxin content in the AZ of WT, Slgata6, and SlGATA6-OE lines. Compared to the WT, knockout and overexpression of SlGATA6 decreased and increased auxin levels in the AZ, respectively (fig. S12A). Moreover, following treatment with NAA (50 ng·g^−1), which erased the difference in the auxin concentration among the different lines, the abscission rate of Slgata6 plants remained higher than that of the WT, but SlGATA6-OE plants showed a much delayed abscission rate (fig. S12B). In addition, Slgata6 DR5:GUS plants displayed a severely compromised auxin-response gradient, with almost undetectable GUS staining in the proximal region of the AZ and the AZ (fig. S12C). These results indicate that both the auxin concentration and response are regulated by SlGATA6. Auxin, SlIAA3, and SlIAA32 distribution is tightly regulated by SlGATA6 To gain insight into the role of SlGATA6 in establishing the distribution of auxin, SlIAA3, and SlIAA32 in the AZ, we investigated where auxin accumulated in the pedicel AZ by immunolocalization on longitudinal pedicel sections with a monoclonal anti-IAA antibody ([160]45). The cellular auxin concentration was high in distal regions and the AZ proper, while weaker in proximal AZ regions ([161]Fig. 5, A and C). We used Calcofluor white (CFW, cell wall dye) to distinguish individual cells and made transverse sections to further study the distribution of auxin ([162]Fig. 5, B, D, and E). Compared with the proximal side of AZ cells, there are more cells containing high auxin signal in the distal region, but not higher signal in the cell of distal region; thus, an auxin gradient with a higher concentration at the distal end and a lower concentration at the proximal end is formed in the AZ. Knockout of SlGATA6 largely abolished the auxin gradient from the distal to proximal AZ regions, with a similar phenotype observed in the knockout of SlLAX2 ([163]Fig. 5, A to E). Fig. 5. SlGATA6 regulates the distribution of auxin, SlIAA3, and SlIAA32 throughout the AZ. [164]Fig. 5. [165]Open in a new tab (A and B) Immunolocalization of IAA in AZ of WT, Slgata6, and Sllax2 plants. Longitudinal section view (A); transverse section view (B). Arrow indicates intracellular auxin. (C) Quantification of the fluorescence intensity in (A). (D) In transverse section view, we divided the area from pith to epidermis into nine layers for quantification. (E) Quantification of the fluorescence intensity in (B). (F and G) Immunolocalization of SlIAA32 and SlIAA3-mCherry in AZ of WT and Slgata6 plants. Longitudinal section view (F); transverse section view (G). Solid triangles represent SlIAA32 protein, and hollow triangles represent SlIAA3 protein. Arrows indicate the presence of both SlIAA32 and SlIAA3 proteins. (H and I) Quantification of the fluorescence intensity in [(F) and (G)]. Scale bars, 200 μm [(A), (B), (F), and (G) complete view] and 20 μm [(B) and (G) magnified view]. Quantification of signal intensity from the distal side of the AZ (DAZ) to the proximal side of the AZ (PAZ) (6 ≤ n ≤ 10) was conducted using ImageJ [(C) and (H)]. Quantification of signal intensity at the distal side of the AZ (DAZ) and the proximal side of the AZ (PAZ) (6 ≤ n ≤ 8) was conducted using ImageJ [(E) and (I)]. Shades around lines depict 95% confidence interval [(C) and (H)]. We generated proSlIAA3:SlIAA3-mCherry transgenic tomato plants and assessed SlIAA3 accumulation by using an antibody against mCherry. The mCherry signal was more abundant in the proximal side of the AZ compared to the distal part of the AZ, while knockdown of SlIAA3 by VIGS led to a marked decrease of SlIAA3 accumulation in the proximal side of the AZ (fig. S13, A and B). We also investigated SlIAA3 and SlIAA32 distribution by double immunofluorescence labeling using monoclonal antibodies specifically recognizing mCherry or SlIAA32. In proSlIAA3:SlIAA3-mCherry plants, SlIAA32 highly accumulated in the distal side of the AZ, but not in the proximal side of the AZ, similar to the auxin distribution; we observed an opposite accumulation pattern for SlIAA3, with lower SlIAA3 accumulation in the distal side of the AZ than in the proximal side of the AZ ([166]Fig. 5, F to I). These results suggest that the changes in auxin concentration and signaling in cells of different AZ regions control abscission. We crossed Slgata6 to the proSlIAA3:SlIAA3-mCherry line and measured SlIAA3 and SlIAA32 accumulation in the resulting Slgata6 proSlIAA3:SlIAA3-mCherry line by double immunofluorescence labeling. Knockout of SlGATA6 abolished SlIAA3-mCherry and SlIAA32 accumulation across the AZ ([167]Fig. 5, F to I), which was highly similar to the pattern seen following flower removal in proSlIAA3:SlIAA3-mCherry lines (4 hours after flower removal) (fig. S13, C to F). These results indicate that SlIAA3 and SlIAA32 distribution across the AZ is tightly regulated by SlGATA6. SlARF2a functions downstream of SlGATA6-SlIAA3/SlLAX2 module to accelerate abscission To further explore the relationship between SlLAX2 or SlIAA3 and SlGATA6 during abscission, we crossed Sliaa3 to Sllax2 to obtain Sliaa3 Sllax2 double mutant lines, then crossed Sliaa3, Sllax2, and Sliaa3 Sllax2 mutant lines to SlGATA6-OE plants. We observed that either single or double knockout of SlIAA3 and SlLAX2 in SlGATA6-OE plants significantly accelerated abscission compared with SlGATA6-OE plants, and there was no significant difference between the WT and Sliaa3 Sllax2 SlGATA6-OE plants (fig. S14A). We also crossed Slgata6 to the Slarf2a mutant: The resulting Slgata6 Slarf2a double mutant line displayed a delayed abscission phenotype compared to that of Slgata6 (fig. S14B). These results indicate that SlLAX2 and SlIAA3 function nonredundantly and act downstream of SlGATA6, while SlARF2a functions downstream of SlGATA6-SlIAA3/SlLAX2 module to modulate abscission. SlKD1 induced abscission by repressing the function of SlGATA6 Our previous research indicated that SlKD1 expression is up-regulated in abscission induced by flower removal (auxin gradient depletion) and under low-light conditions, promoting tomato pedicel abscission ([168]12, [169]32), and RT-qPCR assay further indicated that (fig. S15). Low light also induced SlGATA6 expression (fig. S16A). However, in low light, Slgata6 mutants showed more flower drop until fruit set, while fewer flowers dropped in SlGATA6-OE lines ([170]Fig. 6A and fig. S16B). This finding indicates that SlGATA6 inhibits abscission. Fig. 6. Auxin depletion and low light induce SlKD1 to repress the SlGATA6-SlLAX2/SlIAA3 module to promote abscission. [171]Fig. 6. [172]Open in a new tab (A) Flower drop phenotypes of WT, Slgata6, and SlGATA6-OE plants. Arrows indicate AZ. Scale bar, 1 cm. Images were digitally extracted for comparison. (B and C) Number of hours needed for WT, SlKD1-RNAi, Slgata6 SlKD1-RNAi, Sliaa3 SlKD1-RNAi, Sllax2 SlKD1-RNAi, and Sliaa3 Sllax2 SlKD1-RNAi plants to achieve 50% abscission (B) and frequency of flower abscission (C). n = 6. (D) EMSA showing that SlKD1 does not bind to the SlLAX2 and SlIAA3 promoters but inhibits the binding of SlGATA6 to their promoters. (E) Promoter activity assay indicating that SlKID1, but not SlKD1[ΔC], represses the transcriptional induction of proSlLAX2:GUS or proSlIAA3:GUS by SlGATA6. Values are means ± SD from three biological replicates. (F) Chromatin immunoprecipitation followed by quantitative PCR (ChIP-qPCR) analysis of SlGATA6-Flag enrichment at the SlLAX2 and SlIAA3 promoters in SlGATA6-OE and SlGATA6-OE SlKD1-RNAi lines following flower removal or shading treatment. Upper panel shows the localization of the PCR products tested for enrichment (top, black) and SlGATA6 binding sites (red). FR, flower removal. (G) Tomato pedicel AZ was divided into three zones using laser microdissection: distal AZ, AZ, and proximal AZ. Expression levels for each gene are represented according to a color scale. (H and I) Number of hours needed for WT, SlKD1-OX, and Slarf2a SlKD1-OX plants to achieve 50% abscission (H) and frequency of flower abscission (I). n = 6. One-way ANOVA with Tukey’s test [(B), (C), (E), (H), and (I)] was used to assess statistical significance; *P < 0.05, **P < 0.01, ****P < 0.0001; different lowercase letters indicate significant differences (P < 0.05); ns, no significant difference. Boxplots and violin plots show maxima, first quartile, median, third quartile, and minima. To study the relationship between SlKD1 and SlGATA6, we crossed SlKD1-RNAi lines with Slgata6 mutants and assessed the abscission phenotype of Slgata6 SlKD1-RNAi line. After auxin depletion imposed by flower removal, knocking out SlGATA6 in the SlKD1-RNAi background suppressed the delayed abscission phenotype of the SlKD1-RNAi lines ([173]Fig. 6B). Similarly, under low light, knockout of SlGATA6 in the SlKD1-RNAi line resulted in more flower dropping compared to SlKD1-RNAi lines ([174]Fig. 6C). We then examined plants knocked out for SlLAX2 and/or SlIAA3 in the SlKD1-RNAi background. After auxin depletion, Slliaa3 SlKD1-RNAi plants reached 50% abscission in 24.6 hours, while Sllax2 SlKD1-RNAi plants reached 50% abscission in 26.3 hours, and Slliaa3 Sllax2 SlKD1-RNAi plants reached 50% abscission in 16.2 hours ([175]Fig. 6B). Low light induced 43.7% flower drop in Sliaa3 SlKD1-RNAi, 42.7% in Sllax2 SlKD1-RNAi, and 55.3% in Sliaa3 Sllax2 SlKD1-RNAi ([176]Fig. 6C). This result indicates that SlKD1 induces abscission via SlGATA6. Because SlKD1 interacts with the C-terminal ZnF domain of SlGATA6, and the ZnF domain is the DNA-binding domain, we conducted an EMSA to assess the influence of this interaction on the transcriptional activation activity of SlGATA6. EMSAs indicated that SlGATA6 can bind to the SlIAA3 and SlLAX2 promoters, while SlKD1 alone could not. When increasing amounts of recombinant SlKD1 were added, the binding of SlGATA6 to the SlIAA3 and SlLAX2 promoters was weaker; however, SlKD1[ΔC] lacking the C terminus failed to impair the binding of SlGATA6 to the SlIAA3 and SlLAX2 promoters ([177]Fig. 6D). Next, we investigated whether the SlKD1-SlGATA interaction affected the binding of SlGATA6 to the SlIAA3 and SlLAX2 promoters in N. benthamiana leaves using a GUS transactivation assay. Coinfiltration of 35S:SlKD1 with 35S:SlGATA6 and proSlIAA3:GUS or proSlLAX2:GUS significantly inhibited the relative GUS activity induced by 35S:SlGATA6 alone, while coinfiltration of 35S:SlGATA6 with 35S:SlKD1[ΔC] and proSlIAA3:GUS or proSlLAX2:GUS did not affect GUS activity ([178]Fig. 6E). Then, we crossed SlGATA6-Flag-OE lines to SlKD1-RNAi (RNA interference) lines and performed chromatin immunoprecipitation–quantitative PCR (ChIP-qPCR) analysis to determine whether SlKD1 affects SlGATA6 binding to the SlIAA3 and SlLAX2 promoters in vivo. SlGATA6-Flag-OE and SlGATA6-Flag-OE SlKD1-RNAi plants showed no significant difference in terms of their SlGATA6 binding to the SlIAA3 or SlLAX2 promoters under normal conditions. However, SlGATA6-Flag-OE SlKD1-RNAi plants showed higher binding activities for SlGATA6 to the SlIAA3 and SlLAX2 promoters than SlGATA6-Flag-OE after flower removal and under low light ([179]Fig. 6F). In agreement with this result, SlIAA3 and SlLAX2 expression levels were higher in SlGATA6-Flag-OE SlKD1-RNAi plants than in SlGATA6-Flag-OE plants under low-light and auxin depletion conditions (fig. S16C). All these results suggest that SlKD1 represses the transcriptional function of SlGATA6 to induce SlLAX2 and SlIAA3 transcription during abscission induced by auxin depletion brought upon by flower removal and low light. As the KD1 shows less effect on GATA6 binding to the SlIAA3 or SlLAX2 promoters under normal conditions, we further explored the expression levels of SlLAX2 and SlIAA3 in SlKD1-OX plants and WT. The results indicated that SlLAX2 and SlIAA3 of SlKD1-OX plants were slightly low but not significant compared to those in WT (fig. S17). To further understand the impact of SlKD1 on the transcriptional regulation of SlGATA6, we first examined the detailed expression location of SlGATA6 and SlLAX2 in the AZ ([180]Fig. 6G). A detailed localization of SlGATA6 and SlLAX2 transcript in specific cell layers in WT plants showed that SlGATA6 level remained high in the distal side of AZ but was low on the proximal of AZ, SlLAX2 showed a similar phenotype under normal conditions, and the expression level of SlLAX2 is positively correlated with that of SlGATA6. However, under low-light conditions, the transcriptional level of SlGATA6 shows an increase in the distal side of the AZ, while that of SlLAX2 did not. The expression level of SlLAX2 is not correlated with that of SlGATA6. After knockdown of SlKD1, under low-light conditions, the expression patterns of SlLAX2 and SlGATA6 became similar and exhibited a high positive correlation. Those results indicated that SlKD1 plays a minor role in regulating SlGATA6 activities under normal conditions, probably by posttranslation modified to regulate its unstable protein. Under low-light and flower removal conditions, the stable SlKD1 exerts an inhibitory effect on the function of SlGATA6. Furthermore, we crossed SlKD1-OE lines to Slarf2a lines; after auxin depletion, knockout of SlARF2a in the SlKD1-OE background delayed the accelerated abscission of SlKD1-OE lines ([181]Fig. 6H) and repressed the enhanced flower drop of SlKD1-OE under low-light conditions ([182]Fig. 6I). These results suggest that the SlKD1-regulated SlGATA6-SlLAX2/SlIAA3 module is involved in auxin depletion– and low light–induced abscission. The KD1-GATA6 module is conserved in Solanaceae during abscission Compared to other tomato TKN homologs, the KNOX2 domain in SlKD1 appears to be unique ([183]Fig. 7A). SlGATA6 failed to interact with other members of the KONX family genes that contain the KNOX2 domain in Y2H assay (fig. S18), suggesting that SlKD1 plays a unique role in mediating tomato pedicel abscission. Moreover, we discovered that the KNOX2 domain of SlKD1 is highly conserved in the other Solanaceae species potato (Solanum tuberosum), eggplant (Solanum melongena), and pepper (Capsicum annuum) ([184]Fig. 7B). In addition, all of these species have a protein related to SlGATA6 with a similar ZnF domain ([185]Fig. 7C). Shading can also cause flower drop in potato, eggplant, and pepper plants ([186]Fig. 7, D, G, and J). Thus, we asked whether the KD1-GATA module might be conserved in the shading-induced flower abscission of Solanaceae species. To verify this assumption, we used VIGS to knock down KD1 or GATA6 in potato, eggplant, and pepper and measured the flower abscission rate under low-light conditions ([187]Fig. 7). The knockdown of each KD1 ortholog resulted in decreased flower drop under shading, while the knockdown of each GATA6 ortholog led to the opposite phenotype. Thus, the KD1-GATA6 module is conserved in these three Solanaceae crops. Together, our data indicate that flower abscission may be regulated by the KD1-GATA6 module, which is deeply conserved in Solanaceae species. Fig. 7. The KD1-GATA6 module is conserved in three Solanaceae crops. [188]Fig. 7. [189]Open in a new tab (A) Alignment of the KNOX2 domain from SlKD1 and other SlTKN proteins. (B) Alignment of the tomato KD1-KNOX2 domain of SlKID1 with that from the orthologs of other Solanaceae crops. (C) Alignment of the GATA6-ZnF GATA domain from SlGATA6 with that from the orthologs of other Solanaceae crops. (D) Potato flower abscission phenotype under low light. The orange arrows indicate the AZ. Scale bars, 1 cm. (E) Silencing efficiency of StKD1 and StGATA6 in their respective VIGS plants as determined by RT-qPCR. Values are means ± SD from four biological replicates. (F) Frequency of flower abscission in the pTRV2-empty, pTRV2:StKD1, and pTRV2:StGATA6 plants under shading. Boxplots show data from four independent tests, each consisting of 10 to 14 flowers. Violin plot shows maxima, first quartile, median, third quartile, and minima. (G) Eggplant flower abscission phenotype under low light. The orange arrow indicates the AZ. Scale bars, 1 cm. (H) Silencing efficiency of SmKD1 and SmGATA6 in their respective VIGS plants as determined by RT-qPCR. (I) Frequency of flower abscission in the pTRV2-empty, pTRV2:SmKD1, and pTRV2:SmGATA6 plants under shading. n = 30 to 35. (J) Pepper flower abscission phenotype under low light. The orange arrow indicates the AZ. Scale bars, 1 cm. (K) Silencing efficiency of CaKD1 and CaGATA6 in their respective VIGS plants as determined by RT-qPCR. (L) Frequency of flower abscission in the pTRV2-empty, pTRV2:CaKD1, and pTRV2:CaGATA6 plants under shading. n = 30 to 35. One-way ANOVA with Dunnett’s test [(E) and (F)], two-tailed Student’s t test [(H) and (K)], and Fisher’s exact test [(I) and (L)] was used to assess statistical significance; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Violin plots show maxima, first quartile, median, third quartile, and minima. DISCUSSION Local auxin-response gradient in AZ is necessary for inhibiting abscission. Destruction of auxin-response gradient induces abscission, and a complex network that involves multiple auxin-related genes spatially and temporally regulates this process. Here, we report the role and regulatory mechanism of low light and auxin depletion on interrupting the auxin-response gradient and inducing abscission. The immunofluorescence results show that there are more cells holding high auxin at the distal side of the AZ than at the proximal side ([190]Fig. 5). This difference causes a higher concentration of auxin at the distal end than the proximal end, resulting in an auxin gradient across the AZ. SlLAX2 is the key to the formation of this auxin gradient, which is important for maintaining auxin-response gradient and preventing abscission in two aspects. First, SlARF2a plays a major role in depressing the auxin-response and positively regulating ethylene response; hence, it was identified as a vital abscission inducer ([191]Fig. 4 and fig. S11G). The auxin-dependent SUMOylation site within SlARF2a is vital for inhibiting its function. Second, the high concentration of auxin that depends on SlLAX2 promotes the stability of the noncanonical Aux/IAA, SlIAA32, which maintains the auxin-response gradient and prevents abscission by interacting with SlARF2a in a SlLAX2-dependent SUMOylation manner ([192]Fig. 4 and fig. S11H). Moreover, we found that SlIAA3 was highly expressed in the AZ and preferentially accumulated in the proximal region of the AZ, where the auxin concentration is relatively low ([193]Fig. 5). We also revealed that SlIAA3 maintains the auxin-response gradient and represses abscission by interacting with and repressing the SUMOylation of SlARF2a, thereby inhibiting its function ([194]Fig. 4, H, J and L). Considering that SlIAA3 is a canonical Aux/IAA and is degraded in an auxin-dependent pathway, the stability of the noncanonical Aux/IAA SlIAA32 increased as the auxin concentration increased, and both of them contribute to repress the function of SlARF2a. We reasonably deduced that the high auxin concentration of the AZ cell promotes auxin response mainly through SlIAA32, with little contribution from SlIAA3. The middle auxin level in the AZ may exert its function through the combined action of SlIAA32 and SlIAA3, while the low auxin concentration functions through SlIAA3 alone. Although SlIAA32 and SlIAA3 exhibit opposite distribution patterns across the AZ, the combined signal of these two auxin-response enhancers is higher on the distal side than on the proximal side. We suggested that the cells in the distal side of AZ mainly rely on SlIAA32 and slightly rely on SlIAA3 to maintain a high auxin response. Meanwhile, the cells in the AZ use both SlIAA32 and SlIAA3 to maintain a moderate auxin response, and the cells in the proximal side of the AZ mainly depend on SlIAA3 and a small amount of SlIAA32 to form a relatively low auxin response ([195]Fig. 8A). Fig. 8. Schematic model of how the GATA6-LAX2/IAA3 module inhibits abscission. [196]Fig. 8. [197]Open in a new tab (A) GATA6-LAX2/IAA3 module established auxin-response gradient across AZ. (B) Model of low light–induced modulation of tomato flower pedicel abscission by KD1-GATA6. The DAP-seq indicated that SlGATA6 promotes SlLAX2 and SlIAA3 expression by directly binding to a “GATC” motif in the SlLAX2 and SlIAA3 promoters ([198]Fig. 2). Genetic evidence indicates that SlLAX2 and SlIAA3 act downstream of GATA6 and play nonredundant roles in inhibiting abscission. We found that SlGATA6 does not directly bind to the promoter of SlIAA32. Using the Plant Care ([199]http://bioinformatics.psb.ugent.be/webtools/plantcare/html/) to explore the possible upstream regulators of SlIAA32, we identified several light-responsive elements, including G-Box, Box 4, Box II, GT1-motif, and TCCC-motif, and drought-responsive elements (MBS). Combined with SlIAA3 that functions downstream of SlGATA6 and participates in abscission induced by auxin depletion and low light, in contrast, SlIAA32 is potentially engaged in responses to light and drought. It is also reasonably deduced that having both SIAA3 and SIIAA32 could be important in a different way; these proteins, which work at different levels of auxin concentration, can provide a robust repression of SIARF2a during the fluctuation of auxin flow, which may be affected by environmental conditions. Multiple studies indicate that environmental stresses or auxin depletion induce abscission, which requires an interruption of the auxin-response gradient ([200]9, [201]12, [202]46, [203]47). SlKD1 transcript levels increase under auxin depletion and low-light conditions to induce abscission ([204]12, [205]32). SlKD1 physically interacts with SlGATA6 and the interactions between them occurred between the C terminus of SlKD1 containing the KNOX2 domain and the C-terminal ZnF domain of SlGATA6. SlKD1 could not bind to the promoters of SlLAX2 or SlIAA3 on its own, but coexpression of SlKD1 and SlGATA6 suppressed the binding of SlGATA6 to the SlLAX2 and SlIAA3 promoters, as indicated by EMSAs and ChIP-PCR ([206]Fig. 6). Knockdown of SlKD1 delayed pedicel abscission induced by auxin depletion and the flower drop induced by low light, while knockout of SlGATA6 abolished the lower abscission rate and flower drop defects seen in the SlKD1-RNAi line. We observed a similar abscission phenotype in Sliaa3 SlKD1-RNAi, Sllax2 SlKD1-RNAi, and Sliaa3 Sllax2 SlKD1-RNAi lines. Moreover, knockout of SlARF2a significantly slowed down the abscission rate of SlKD1-OE ([207]Fig. 6). These results indicate that auxin depletion and low light induce pedicel abscission and flower drop by inducing SlKD1 to repress the SlGATA6-SlIAA3/SlLAX2 module. However, this inhibition is obvious under low-light and auxin depletion conditions, but not under normal conditions. The regulation mechanism of low light and flower removal on posttranslation modification of SlKD1 needs to be studied in future works. The KNOX2 domain of KD1 and the C-terminal ZnF domain of GATA6 are conserved among solanaceous species. Solanaceous family yields are typically lost in low-light conditions. Our research has demonstrated that the KD1-GATA6 module helped in inducing flower abscission in solanaceous species under low-light conditions. Our study provides KD1 as a potential target for inhibiting abscission and enhancing the yields of solanaceous crops under low-light conditions through gene editing technology. In this study, we revealed that low light–induced SlKD1 expression repressed the function of the SlGATA6-SlIAA3/SlLAX2 module to induce abscission, uncovering a regulatory mechanism by which low light mediates GATA function ([208]Fig. 8B). Under normal conditions, the SlGATA6-SlIAA3/SlLAX2 module maintains the auxin-response gradient across the AZ to prevent abscission. Auxin depletion and low light promotes SlKD1 transcription, which down-regulates SlLAX2 and SlIAA3 expression, thus initiating abscission. MATERIALS AND METHODS Plant materials and growth conditions The tomato (S. lycopersicum) cultivar “Ailsa Craig” was used as the WT in this study. Our laboratory was responsible for maintaining seeds for the SlKD1-OX, SlKD1-RNAi, and DR5:GUS transgenic tomato lines ([209]12). Cultivated potato (S. tuberosum), eggplant (S. melongena), and pepper (C. annuum) were used for VIGS. Tomato, potato, eggplant, and pepper plants were cultivated in a controlled greenhouse environment, maintaining a temperature of 25°C for 16 hours during the day, followed by a temperature of 15°C for 8 hours at night. The experimental procedure for shading treatment and inducing fruit drop was conducted in accordance with a previously outlined methodology ([210]11, [211]12). The photosynthetically active radiation levels under normal light and low light were 600 and 180 μmol m^−2 s^−1, respectively. N. benthamiana plants were used for biochemical experiments and grown under a 25°/18°C (day/night) temperature regime with a relative humidity of 60%. Plasmid construction and tomato transformation Transgenic tomato plants were generated using AC as the background. The CRISPR-Cas9 vectors were composed of the Cas9 endonuclease gene and two single guide RNAs (sgRNAs), as detailed in the study conducted by Mao et al. ([212]48). Two target sequences per gene of interest, sgRNA1 and sgRNA2, were designed based on target sites in the exons of SlGATA6, SlLAX2, SlIAA3, SlIAA32, and SlARF2a (see data S7). To conduct phenotypic tests, T2 homozygous lines lacking the Cas9 transgene were identified by sequencing PCR products generated from the appropriate target areas. The primers used in this study are listed in data S7. The pTA7002-SlIAA3dIIm, pTA7002-SlIAA3dIImSIMm, and pTA7002-SlIAA32SIMm constructs were generated by cloning the synthesized coding sequence encoding a domain II mutant of SlIAA3 or the SIM domain of SlIAA3/SlIAA32 into the pTA7002 vector, which had been digested with the restriction enzymes Xho I and Spe I. The full-length coding sequence of SlIAA32 was also cloned by PCR and introduced into the pTA7002 vector. All resulting plasmids were introduced individually into Agrobacterium (Agrobacterium tumefaciens) strain LBA4404 via electroporation. The presence of the construct in each plant was assessed by PCR, using primers specifically designed to target the hygromycin resistance gene. Gene expression levels in transgenic plants were analyzed following treatment with a 1 mg/liter dexamethasone solution. Three T2 lines with the highest expression levels were chosen for phenotypic analysis and experiments. The SlGATA6-OE, SlLAX2-OE, and SlARF2a-OE plants were generated by cloning the full-length coding sequence of SlGATA6, SlLAX2, and SlARF2a individually in pCAMBIA1300-Flag vectors. SlARF2a-OE^3* K/R plants were generated by cloning the synthesized coding sequence of SlIARF2a^3* K/R into the pCAMBIA1300-Flag vector, which had been digested with the restriction enzymes Eco RI and Bam HI. The proSlIAA3:SlIAA3-mCherry plants were generated by cloning a 2000-bp promoter fragment and full-length coding sequence of SlIAA3 in the p2301-mCherry vector. The resulting constructs were introduced into Agrobacterium (strain LBA4404). Each construct was transformed into tomato (AC cultivar) using the leaf disc cocultivation method ([213]12). Specific primers were used to detect SlGATA6, SlLAX2, and SlARF2a expression in transgenic plants, respectively. All primers used in this study are listed in data S7. Moreover, on the basis of the morphological and histological data of mutant and transgenic plants, we discovered that, in comparison with WT plants, all the mutant/transgenic plants developed normal AZs. The differences in abscission between them and the WT were exhibited in the varying lengths of time necessary for the abscission (fig. S19). Pedicel abscission assays The experiments looking at flower pedicel abscission were conducted according to the methodology outlined in previous studies ([214]12, [215]49). In short, to analyze flower pedicel abscission rate, we used at least 15 flowers for each experiment. Flowers were manually removed (to eliminate the source of auxin and then promotes abscission), and flower pedicels were placed in 1% agar media for incubation. The time of flower removal was recorded as 0 hours and the number of flower pedicel abscission at various time intervals was then calculated as a percentage of the overall number. For 50% pedicel drop, a nonlinear fit was performed using Origin software, and the time required to achieve 50% abscission was computed. The tomato was treated with low light before flowering, and the treatment was continued until the fruit was set. The percentage of flower drop for WT and mutant plants was calculated under normal and low-light conditions. We measured 20 to 30 flowers in four inflorescences in each experiment. Virus-induced gene silencing The VIGS experiments were mainly carried out as described in previous research ([216]50). The VIGS tool available on the Sol Genomics Network ([217]https://solgenomics.net/) website was used to design the specific fragments for knocking down SlLAX2, SlIAA3, SlIAA11, SlIAA13, SlARF2a, SlARF4, SlARF6, SlARF14, StKD1, StGATA6, SmKD1, SmGATA6, CaKD1, and CaGATA6. Each fragment was amplified through PCR and then inserted into the pTRV2 vector. The resulting constructs, the pTRV2 empty vector, and the pTRV1 vector were separately introduced into Agrobacterium strain GV3101. As previously described ([218]51–[219]54), Agrobacterium cultures harboring pTRV1 or pTRV2 or each pTRV2 derivative were suspended using buffer (10 mM MES, 10 mM MgCl[2], and 200 mΜ acetosyringone, pH 5.6) and mixed in a 1:1 (v/v) ratio before infiltrating the stem of tomato, potato, eggplant, or pepper plants. The silencing efficiency was assessed by RT-qPCR at the time of flower blooming. The primers are listed in data S7. RNA extraction and RT-qPCR analysis The AZs of various plants were collected and frozen in liquid nitrogen and preserved at −80°C. An RNA Pure Plant Kit (CWBIO, Cambridge, MA, USA) was used to extract total RNA from samples. One microgram of total RNA was converted into first-strand cDNA using a PrimeScript 1st strand cDNA Synthesis Kit (Takara). qPCR was performed using SYBR Premix (Takara) on a qTOWER3/G real-time instrument (Analytik Jena). SlACTIN, StEF1-α, SmACTIN, and CaACTIN genes were used as internal controls for their respective plant species. Relative gene expression levels were calculated using the 2^−ΔΔCt method ([220]55). Detailed qPCR primer sequences are listed in data S7. Sequence alignment and phylogenetic analysis The amino acid sequences of the KNOX2 domain from KNT in tomato and the KNOX2 domain and ZnF-GATA6 domains from potato, eggplant, and pepper proteins were obtained from the Sol Genomics Network ([221]https://solgenomics.net/) website by BLAST search. The multiple sequence alignment was conducted using MEGA7 software. Phylogenetic analysis using the amino acid sequences for the tomato and Arabidopsis GATA proteins was performed using MEGA7 software with the neighbor-joining method with a bootstrap test of 1000 replicates. Y2H assay A cDNA library was constructed from mRNA extracted from flower AZs harvested at 0, 2, 4, 8, 12, and 16 hours following abscission induced by auxin depletion (flower removal) using the Make Your Own Mate & Plate Library System (Clontech, [222]http://clontech.com/). The full-length coding sequence of SlKD1 was introduced into the pGBKT7 vector; the resulting construct was used as bait. The above library was screened using a Yeastmaker Yeast Transformation System 2 kit (Clontech, 630439). The sequences encoding the fragments SlKD1[N] (amino acids 1 to 79) and SlKD1[C] (amino acids 80 to 171) were individually cloned into the pGBKT7 vector. The sequences encoding the fragments KNOX2 domain of SlTKN1 to SlTKN18 were individually cloned into the pGBKT7 vector. The full-length coding sequence of SlGATA6, SlGATA6[N] (encoding amino acids 1 to 157 of SlGATA6), SlGATA6[M] (encoding amino acids 158 to 233 of SlGATA6), and SlGATA6[C] (encoding amino acids 234 to 325 of SlGATA6) sequences were cloned into the pGADT7 vector. All primers used in this study are listed in data S7. The resulting pGBKT7 and pGADT7 constructs were cotransformed as appropriate pairs into the yeast strain Y2H-Gold. The transformants were plated on synthetic defined medium lacking Leu and Trp and allowed to grow at 30°C for 3 to 5 days. Positive transformants were spotted onto synthetic defined medium lacking Leu, Trp, His, and Ade and containing Aureobasidin A (AbA; 150 ng/ml) and allowed to grow at 30°C for 3 to 5 days. The combination of BD-53 and AD-T was used as a positive control, and the combination of BD-Lam and AD-T was used as a negative control. Protein extraction and immunoblotting Total protein from tomato pedicel AZ or N. benthamiana leaves was extracted using radioimmunoprecipitation assay buffer (Solarbio, R0010). Immunoblot analysis was performed using the Magnetic IP/Co-IP Kit (Thermo Fisher Scientific, 88804). The antibodies used were as follows: anti-GST (1:8000 dilution; Solarbio, K200006M), anti-His (1:3000 dilution; Solarbio, K200060M), anti-GFP (1:3000 dilution; Solarbio, K114305P), anti-Flag (1:5000 dilution; CST, #14793), anti-Actin (1:2000 dilution; Biopm, PMK085S), anti-HA (1:3000 dilution; Solarbio, K007440P), anti-MYC (1:3000 dilution; Solarbio, K106458P), anti-SUMO1 (1:1000 dilution; Abcam, ab5316), and anti-SlIAA32 (SSEYLLNHATTLPSVYY; 1:1000 dilution; synthesized by Abmart Shanghai Co. Ltd. synthesis). The secondary antibodies used were goat anti-rabbit IgG (H + L)–horseradish peroxidase conjugate (1:30,000 dilution; EASYBIO, BE0101) and goat anti-mouse IgG (H + L)–horseradish peroxidase conjugate (1:3000 dilution; Bio-Rad, 170-6516). The membranes were incubated with ECL chemiluminescent substrate (Super sensitive) (Biosharp, BL523) before the signals were captured on an Azure Biosystems C600 (America). Pull-down assay The full-length or C-terminal coding sequences of SlKD1 and SlGATA6 were individually amplified by PCR from WT cDNA and cloned into pET30a and pGEX-6P-1, respectively. Primer sequences are listed in data S7. The plasmids were then individually transformed into Escherichia coli strain BL21. One positive colony per construct was grown at 37°C until the OD[600] (optical density at 600 nm) reached 0.6 to 0.8. Protein production was induced for 16 hours at 18°C by the addition of 0.25 mM isopropyl-β-d-thiogalactopyranoside (IPTG). Recombinant SlKD1-His or SlKD1c-His was purified using Ni-NTA agarose (Beyotime, P2241). Recombinant SlGATA6-GST, SlGATA6c-GST, or GST (negative control) was incubated with GST beads (Beyotime, P2262-1). SlKD1-His or SlKD1c-His was added to the GST beads covered with SlGATA6-GST, SlGATA6c-GST, or GST. After incubation at 4°C for 1 hour, the beads were washed three times with phosphate-buffered saline (PBS, pH 7.2 to 7.4). The bound proteins were then eluted with elution buffer (10 mM GSH, 8 mM Na[2]HPO[4], 136 mM NaCl, 2 mM KH[2]PO[4], and 2.6 mM KCl). The eluted proteins were detected by immunoblotting as above. Immunoprecipitation assay Total protein from tomato pedicel AZ from SlARF2a-Flag OE, SlARF2a^3*K/R-Flag OE, and Sllax2 SlARF2a-Flag OE plants was extracted with lysis buffer (Beyotime, P0043), to which the protease inhibitor phenylmethylsulfonyl fluoride (PMSF) was added (1 mM, Beyotime, ST506-2). The mixture was subjected to high-speed centrifugation at 14,000g for 15 min at 4°C. To the resulting supernatant, 50 μl of anti-Flag beads (Beyotime, P2115) was added, followed by incubation on ice for 2 hours. The beads were collected by magnetic separation rack and then washed three times with 1 ml of cold lysis buffer each time. After the final wash, 100 μl of preheated (95°C) 1× SDS-loading buffer was added to the beads to elute the immunocomplex. The immunoprecipitated proteins were examined by immunoblotting using anti-Flag and anti-SUMO1 antibodies. Co-IP assay Transient expression in N. benthamiana leaves was used for co-IP. The full-length coding sequences of SlKD1 and SlGATA6 were individually amplified from AC cDNA and cloned into the pCAMBIA1300-Flag vector and pCAMBIA1300-GFP vector, respectively. For the construction of SUMO-HA, SlARF2a-Flag, SlIAA3-MYC, and SlIAA32-MYC constructs, the full-length coding sequences of SUMO, SlARF2a, SlIAA3, and SlIAA32 were individually amplified from WT cDNA using specific primers and introduced into the pRI101-AN vector with the appropriate tag ([223]56). Primers used for vector construction are listed in data S7. The resulting plasmids were individually transformed into Agrobacterium strain GV3101 for infiltration of N. benthamiana leaves. After 2 days, leaves were collected and homogenized in lysis buffer (with PMSF) at 4°C for 30 min. The mixture was subjected to high-speed centrifugation at 14,000g for 15 min at 4°C. The resulting supernatant was incubated with 50 μl of anti-GFP beads (Beyotime, P2132) or anti-Flag beads (Beyotime, P2115) for 2 hours at 4°C. Subsequent steps were as described above for immunoprecipitation. Proteins were probed by immunoblotting with anti-GFP, anti-Flag, anti-MYC, and anti-HA antibodies. Subcellular localization and BiFC assays For subcellular localization, the full-length coding sequence of SlGATA6 was amplified by PCR from WT cDNA and cloned into the pCAMBIA1300-GFP vector. For BiFC, the full-length coding sequence of SlKD1 and a fragment encoding SlKD1[C] or SlKD1[ΔC] were cloned in the pCAMBIA1300-nYFP vector; the full-length coding sequence of SlGATA6 and a fragment encoding SlGATA6[C] or SlGATA6[ΔC] were cloned in the pCAMBIA1300-nYFP vector. The primers used for vector construction are described in data S7. The resulting plasmids were individually transformed into Agrobacterium strain GV3101 for infiltration of N. benthamiana leaves. Protoplasts were created 3 days after infiltration by immersing converted leaves in the Plant Protoplasts Isolation Kit (Beyotime, C0362S) and gently agitating for 1 hour in the dark. The protoplast suspensions were centrifuged at 200g for 1 min. The supernatant was discarded, and the pellet was gently resuspended in the remaining supernatant. Fluorescence in N. benthamiana leaves was captured at 3 days after infiltration using a Leica TCS SP8 81-1557 confocal laser scanning microscope. Excitation/emission wavelengths were 488 nm/506 to 538 nm for YFP and GFP, respectively. NF-YA4-mCherry and 4′,6-diamidino-2-phenylindole (DAPI, Beyotime, C1002) were used as nuclear localization markers, with excitation/emission wavelengths of 359 nm/457 nm (mCherry) and 587 nm/610 nm (DAPI). RNA in situ hybridization RNA in situ hybridization was conducted as outlined in a recent study by Wang et al. ([224]57). In brief, a specific fragment of the coding sequence of SlGATA6, SlLAX2, SlIAA3, SlIAA32, and SlARF2a was amplified and subsequently inserted into the pSPT18 and pSPT19 vectors (Roche, Basel, Switzerland). The antisense and sense RNA probes were synthesized using SP6 and T7 RNA polymerase, respectively, following the established methodology outlined in the DIG Oligonucleotide 3′-End Labeling Kit (Roche). Probe information is given in data S7. RNA-seq analysis TRIzol Reagent (Invitrogen, 15596026) was used to extract total RNA from the AZ of AC and SlGATA6 knockout transgenic plants, using two biological replicates. The concentration and purity of total RNA were assessed using a 2100 Bioanalyzer (Agilent). Two micrograms of total RNA was used for library construction using a KC Stranded mRNA Library Prep Kit for Illumina (catalog no. DR08402, Wuhan Seqhealth Co. Ltd., China). Products corresponding to 200 to 500 bp were purified, quantified, and eventually sequenced using a DNBSEQ-T7 sequencer (MGI Tech Co. Ltd., China) as 150-bp paired-end reads. The generation and sequencing of the libraries were conducted by Wuhan Kangce Technology Co. Ltd. (Wuhan, China). Using Trimmomatic (version 0.36), the raw sequencing data were filtered, and the clean reads were mapped to the tomato reference genome (version: SL4.0) using STRA software (version 2.5.3a) with default parameters. The analysis of DEGs was conducted using the edgeR package (version 3.12.1). The DEGs were filtered based on an absolute log[2] fold change of at least 1 and an FDR below 0.01. KEGG pathway enrichment analysis was implemented using KOBAS software (version 2.1.1). The validity of the RNA-seq data was verified through RT-qPCR on a randomly selected gene. The sequences of primers can be found in data S7. DAP-seq and data analysis DAP-seq was carried out according to a previously published method ([225]12). The full-length coding sequence of SlGATA6 was cloned into the pET15b-Halo vector, carrying a sequence encoding the Halo tag. A genomic DNA library was prepared from tomato AZs following the guidelines provided by the manufacturer (Illumina). Recombinant SlGATA6-Hola protein was purified and the DNA targets were enriched, followed by sequencing on an Illumina HiSeq 4000 instrument (Gene Denovo Biotechnology Co., China). The clean reads were aligned to the tomato reference genome (version SL4.0) using the Bowtie2 program (version 2.2.5). The binding peaks of SlGATA6 were identified using MACS2 software (version 2.1.2). The MEME suite ([226]http://meme-suite.org/) was used to detect motifs. Electrophoretic mobility shift assays The full-length coding sequence of SlGATA6 was cloned into the pGEX-6P-1 vector; the resulting plasmid was transformed into E. coli Rosetta cells. One positive colony was grown at 37°C until the OD[600] reached 0.6 to 0.8. Protein production was induced for 16 hours at 18°C by the addition of 0.25 mM IPTG. Recombinant SlGATA6-GST was purified using a GST-Tag Protein Purification Kit (Beyotime, P2260S). Parts of promoter sequences of SlLAX2, SlIAA3, SlIAA11, and SlIAA13, tagged with biotin, are depicted in data S7. The 5′ biotin-labeled probe was synthetized by Saibaisheng Company (China). A nonlabeled probe was used as a competitor, while the labeled mutant probe served as a negative control. EMSA was performed using a Chemiluminescent EMSA Kit (Beyotime). GUS staining and activity assay GUS staining and activity assays were carried out according to published methods ([227]58). For GUS staining, flower pedicels were incubated in GUS staining solution (Real-Times Biotechnology Co. Beijing, China) at 37°C for 24 hours in the dark. Three distinct plants were chosen for each genotype, from which at least 20 flower pedicels were collected. For GUS activity assays, the full-length coding sequence of SlGATA6 and SlKD1 and fragments of the SlKD1 coding sequence (SlKD1[ΔC]) were individually cloned into the binary pRI101 vector. A 2200-bp promoter fragment for SlLAX2, SlIAA3, SlIAA11, and SlIAA13 was cloned in the pBI101 vector upstream of the GUS reporter gene. The resulting plasmids were individually introduced into Agrobacterium strain EHA105 for infiltration of N. benthamiana leaves as described above. GUS activity was measured as previously described ([228]58) with three biological replicates. The primers can be found in data S7. Determination of auxin concentration The content of IAA was measured as previously described ([229]58). In brief, the IAA content was determined using LC-MS/MS analysis. The AZ segments of at least 40 flower pedicels were frozen and ground in liquid nitrogen, and then freeze dried in a vacuum oven set at −80°C. From each plant, 50 mg of powder was dissolved in 1 ml of a solution made of formic acid, water, and methanol (15:4:1, v/v/v). To each sample, 10 μl of an internal standard was added at a concentration of 100 ng/ml (186006963, Waters). The mixture was centrifuged at 4°C for 5 min at 8000g. After transferring to sterile plastic microtubes, the supernatant was evaporated until completely dry and then reconstituted in 100 μl of 80% (v/v) methanol and passed through a 0.22-μm membrane filter for LC-MS/MS analysis. Protein stability analysis The full-length coding sequence of SlIAA32 was amplified by PCR from WT cDNA and cloned into the pET30a vector. The primers can be found in data S7. Protein production was induced as above. Recombinant SlIAA32-His was purified using Ni-NTA agarose (Beyotime, P2241). For cell-free degradation assays, total protein extracts were prepared from AZs treated with water only (negative control), NAA (50 μg/g; Sigma-Aldrich, 317918), or 10 μM l-Kyn (Sigma-Aldrich, K8625) using protein extraction buffer (25 mM tris-HCl, pH 7.5, 10 mM NaCl, 10 mM MgCl[2], 4 mM PMSF, 5 mM dithiothreitol, and 10 mM adenosine triphosphate). Protein concentration was determined with a bicinchoninic acid (BCA) kit (Thermo Fisher Scientific, 23227). Each cell-free degradation reaction contained 500 μg of total proteins and 100 ng of recombinant SlIAA32-His. Aliquots of the mixtures were collected after incubation at 28°C for 0, 30, or 60 min. Immunoblotting with anti-His antibodies was used to determine the amount of SlIAA32-His remaining at each time point. For in vivo protein stability assays, total protein was extracted from AZ in WT, SlLAX2-OE, and Sllax2 lines, then separated by SDS-PAGE. ACTIN was used as the loading control, which was detected by immunoblotting with anti-ACTIN antibodies. The relative band intensity was determined using ImageJ software ([230]https://imagej.net/ij/). The protein content of the 0-hour sample was set to 1. Immunofluorescence analysis IAA distribution was determined as previously described ([231]45). Freshly prepared AZ samples were prefixed for 2 hours in 3% (w/v) 1-ethyl-3-carbodiimide (Sigma-Aldrich, 341006) at 28°C before being transferred to FAA (Servicebio, G1108). The samples were dehydrated through a graded ethanol series (30, 50, 70, 80, 90, and 100%, all v/v). After dehydration, the samples were immersed in xylene and then Paraplast (Thermo Fisher Scientific) for 1 hour each time before embedding in 100% (w/v) Paraplast. The embedded samples were sliced into 10-mm slices. The sections were incubated with 1:100 (v/v) dilutions of anti-IAA monoclonal antibody (Sigma-Aldrich, A0855) at 4°C (12 to 16 hours), followed by Alexa Fluor 488–labeled anti-mouse IgG antibody [1:400 (v/v), Servicebio, GB25301] for 1 hour at room temperature in the dark. For cell wall staining, sections were immersed in 0.01% CFW (18909; Sigma-Aldrich) and subjected to dark treatment for 5 to 10 min before observation. The fluorescence signals were captured with an ortho-fluorescent microscope (Nikon), with an excitation wavelength of 465 to 495 nm and an emission wavelength of 515 to 555 nm (for IAA) and an excitation wavelength of 355 nm and an emission wavelength of 440 nm (for CFW). For SlIAA3/SlIAA32 double immunofluorescence, the sections were incubated overnight with 1:300 (v/v) dilutions of anti-mCherry antibody and 1:200 (v/v) dilution of anti-SlIAA32 antibody at 4°C, followed by incubation with CY3-labeled anti-mouse IgG antibody [1:300 (v/v), Servicebio, GB21301] and Alexa Fluor 488–labeled anti-rabbit IgG antibody [1:400 (v/v), Servicebio, GB25303] for 1 hour at room temperature in the dark. For nucleus staining, sections were immersed in DAPI and subjected to dark treatment for 5 to 10 min before observation, with excitation/emission wavelengths of 587 nm/610 nm (DAPI). The fluorescence signals were captured with an ortho-fluorescent microscope (Nikon), with an excitation wavelength of 465 to 495 nm and an emission wavelength of 515 to 555 nm for Alexa Fluor 488 detection (for SlIAA32) and an excitation wavelength of 510 to 560 nm and an emission wavelength of 590 nm for mCherry (for SlIAA3). All fluorescence intensities were quantified by ImageJ. Measurements of fluorescence intensity were made for the longitudinal sections of tomato pedicel that extended from the proximal side of the AZ to the distal side over a distance of 200 μm. Within each 10-μm-length range, randomly choose 10 to 20 nuclear signals for measurement, then note the average value of each signal as the measurement result. The region between the pith and the epidermis was uniformly separated into nine layers for the cross-sectional slices, and the fluorescence intensity of each layer was assessed independently. Each experiment comprised 6 to 10 biological replicates. LC-MS/MS analysis LC-MS/MS analysis was used to identify which SlARF proteins were specifically enriched in pedicel AZs. The experiment was carried out with reference to previous studies ([232]59). Total protein was extracted from AC AZs in protein lysate buffer [100 mM NH[4]HCO[3], pH 8, 8 M urea, and 0.2% (w/v) SDS]. A total of 5 mg of total protein was digested with 1 ml of 50 mM NH[4]HCO[3] containing 50 μg of Lys-C/trypsin protease mix (Promega, Madison, WI) at 37°C for 12 to 16 hours. All peptides were purified using StageTips (Thermo Fisher Scientific, 87782) before being analyzed by LC-MS/MS. Peptide concentration was determined using a BCA assay. Shotgun proteomics analysis was conducted using an EASY-nLCTM 1200 UHPLC system (Thermo Fisher Scientific) and a Q Exactive HF-X mass spectrometer (Thermo Fisher Scientific) in data-dependent acquisition mode. The raw files were processed using Proteome Discoverer software (Thermo Fisher Scientific, version 2.4) retrieval (MS1 tolerance, 10 ppm; MS2 tolerance, 0.02 Da; missed cleavage, 2). The peptide fragments were searched against the NCBI database ([233]https://ncbi.nlm.nih.gov/). ChIP-qPCR The ChIP assays were conducted in accordance with previous methods ([234]60). A total of 1 g of AZ tissue was collected from AC, SlGATA6-Flag OE, and SlGATA6-Flag OE SlKD1-RNAi plants. The obtained tissue samples were then cross-linked under vacuum in 3% (w/v) formaldehyde. The chromatin was fragmented to an average size of roughly 500 bp using a sonicator (Sonic Ruptor 400, Omni, USA). Immunoprecipitation was performed using 20 μl of anti-Flag antibody (CST, #14793). Following the collection of immunoprecipitated proteins using protein A beads (Thermo Fisher Scientific, 80103G), proteins were removed using proteinase K digestion (Thermo Fisher Scientific, 26160), followed by reverse cross-linking. The quantification of the immunoprecipitated DNA was performed using qPCR with SYBR Green dye on a qTOWER3/G real-time system (Analytik Jena). The primers can be found in data S7. Laser microdissection and gene expression analysis The laser microdissection experiments were performed following our previous methods ([235]61). Laser microdissection was performed using a PALM MicroBeam system (Carl Zeiss, Germany). The frozen sections (10 μm) of WT and SlKD1-RNAi AZs were placed on the MembraneSlide 1.0 PEN (Zeiss, No. 415101-4401-000). Incubate the sections in ice-cold 100% ethanol for 2 to 3 min to dehydrate. Specific pedicel tissue sections (distal side of AZ, AZ, and proximal side of AZ) were removed from tissue slices (n = 5) and collected in AdhesiveCap 500 tubes (Zeiss). The tubes were frozen in liquid nitrogen and stored at −80°C. Total RNA was extracted from tissue samples using the RNeasy Micro Kit (Qiagen), following the manufacturer’s instructions. Then, RT-qPCR analysis was conducted on the isolated RNA following the previously outlined methods. Statistical analysis Experiments were conducted with three separate biological replicates unless noted otherwise. The figures or figure legends include specific statistical parameters for each experiment. Statistical analyses were performed using GraphPad Prism (v9). Accession numbers Sequence data from this article can be found in Sol Genomics Network ([236]https://solgenomics.net/) or NCBI ([237]https://ncbi.nlm.nih.gov/) under the accession numbers listed in data S7. Acknowledgments