The mulberry (Morus alba L., ‘Hongguo 2 hao ‘) and tobacco seeds (Nicotiana benthamiana) were taken from the State Key Laboratory of Silkworm Genome Biology, Southwest University, Chongqing, China. Tobacco and mulberry seeds were grown in a mixture of vermiculite and nutritive soil (1:1) at 25°C, 40% to 60% relative humidity, under a 16h light/8h dark photoperiod. Tobacco plants were used for transient expression assays.

In June 2021, samples for transcriptome sequencing were gathered from the experimental site of the State Key Laboratory of Silkworm Genome Biology at Southwest University. Only mature and healthy mulberry fruit stalks were chosen, and a total of 12 fruit stalks were collected as a sample. To ensure accuracy, three biological replicates were used. The detached area tissues of fruit stalks that are easy to drop fruits were taken as the experimental group, marked as MD_ 1, MD_ 2, and MD_ 3. The tissues of fruit stalks that were not easy to drop fruits were taken as the control group, marked as MN_ 1, MN_ 2, and MN_ 3. After quick freezing in liquid nitrogen, samples were stored in an ultra-low temperature freezer (- 80°C).

The MaJOINTLESS promoter sequence was predicted by local blast from the genome file of mulberry, using the first base of the MaJOINTLESS promoter sequence as +1 and the first base upstream of ATG as -1. Further, 1500 bp upstream promoter sequence was cloned and named pMaJOINTLESS. The extraction of genomic DNA was performed using Plant Genomic DNA Extraction Kit (DP3111, BioTeke, Beijing, China) according to the instructions. The promoter fragments of MaJOINTLESS (1500 bp) were subcloned into the PVCT024 vector upstream of the GUS (β-glucuronidase)-coding region ( Figure 1A ). The construct was transformed into tobacco leaves. The leaves were incubated in GUS staining buffer (0.2 mL 0.5 M EDTA, 0.3 mL 40 mM X-Gluc, 0.1 mL 10% TritonX-100, 1 mL 1 M Sodium hydrogen phosphate buffer) for 3 h at 37°C in the dark. The samples were then decolorized in 100% ethanol and visualized using a Zeiss SV11 stereoscope (Carl Zeiss Jena, Oberkochen, German).

Protein sequences of all candidate genes were retrieved from the Arabidopsis genome database (https://www.arabidopsis.org/), and the tobacco genome database (https://solgenomics.net/organism/Nicotiana_benthamiana/genome). Multiple alignments were performed using the DNAMAN. Total RNA was extracted with TRIzol^® reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer’s protocol. The optical density (OD) value of RNA was measured by spectrophotometer, and the total RNA with the purity (260/280) of 1.9 to 2.0 was used for further assay. Total RNA template was reverse transcribed into cDNA with a PrimeScript RT reagent Kit with gDNA Eraser (TaKaRa, Shiga, Japan). The synthesized cDNA was diluted to 200 ng/µL as the template for qPCR. qPCR experiments were carried out on ABI7500 Real-Time PCR machine (Applied Biosystems, Foster City, CA, USA). Then qPCR was performed to quantify the RNA levels with NovoStart^® SYBR qPCR SuperMix Plus (Novoprotein, Shanghai, China). All procedures with instruments and kits were performed according to the manufacturer’s instructions and protocols. The PCR mixture (10.0 µL 2×SYBR Green Realtime Master Mix, 0.8 µL qPCR forward primer, 0.8 µL qPCR reverse primer, 6.4 µL ultrapure H2O, and 2.0 µL Template) was added to the qPCR reaction plate. The cycling parameters were as follows: 95 °C for 30 s; followed by 30 cycles of 95 °C for 3 s and 60 °C for 30 s. The qPCR experiment was repeated more than three times. After the reaction, we exported the data from the ABI7500 Real-Time PCR machine and analyzed the data using the 2^-ΔΔCT method.

N. benthamiana leaf epidermal cells were used for subcellular localization of MaABI5 and MaABF1. The open reading frames (ORFs) of MaABI5 and MaABF1 were amplified using polymerase chain reaction (PCR). Subsequently, the PCR products were introduced into the YFP-2300 vector at the XbaI and BamHI restriction sites. This process resulted in the creation of two constructs: 35Spro : MaABI5-YFP and 35Spro : MaABF1-YFP, respectively. Recombinant vectors were introduced into Agrobacterium (Agrobacterium tumefaciens) strain GV3101. The Agrobacterium culture was incubated at 28°C overnight, and the cells were collected by centrifugation at room temperature for 8 minutes at 2000g. The pellet was then resuspended in a buffer containing 10 mM MgCl2, 10 mM MES, 200μM acetosyringone, at pH 6.3 to achieve a final OD600 of 0.8. Agrobacterium cell suspensions were infiltrated into tobacco leaves. After 48h, laser confocal fluorescence microscopy (Olympus, Tokyo, Japan) was used to observe fluorescence signal in tobacco epidermal cells.

PCR amplification was performed to obtain fragments specific to NbPDS, NbABF1, and NbABI5, with lengths of 500 bp, 404 bp, and 329 bp, respectively. These fragments were individually inserted into the pTRV2 vector. Subsequently, the modified pTRV2 vectors were introduced into the Agrobacterium strain GV3101. The primers used in this study are listed in Supplementary Table 1 . The relevant Agrobacterium colonies were then cultured overnight in LB medium containing 50 mg/mL kanamycin, and 50 mg/mL rifampicin, and 50 mg/mL streptomycin Sulfate. Once the concentration of Agrobacterium reached OD of 0.6 at 600 nm, we harvested the cells through centrifugation at 2000 g for 8 min. The resulting pellet was then re-suspended in a buffer containing 10 mM MgCl₂, 10 mM MES, 200 μM acetosyringone, at a pH of 6.3. pTRV1 and pTRV2 cultures were mixed in a 1:1 (v/v) ratio and then mixture was placed in the dark for 3h at room temperature.

The functional structural domain sequences of MaABF1 and MaABI5 were amplified by PCR and named as BRLZF1 and BRLZI5, respectively. BRLZF1 and BRLZI5 were amplified into the pET28a vector and expressed in E. coli BL21 (DE3) cells. Recombinant proteins were purified and used for EMSAs along with biotin-labeled fragments (~50 bp) of the MaJOINTLESS promoters. The same fragment, but unlabeled, was used as a competitor, while a probe within the mutant ABRE-binding cis-elements (interchanging G with C or C with G) was used as a mutant competitor in the assay. The binding of protein and nucleic acid was verified with chemiluminescence kit (Beyotime, Shanghai, China).

Diagrams of the effector and reporter vectors used for dual-luciferase reporter assays are shown in Supplementary Figure 4 . The effector and reporter plasmid constructs were co-transformed into tobacco leaves using Agrobacterium infiltration. After 2–3 d, activities of the LUC and REN luciferases were measured using the Dual Luciferase Reporter Gene Assay Kit (YEASEN, Shanghai, China). The results were calculated as the ratio of fluorescence of LUC to that of REN. Three leaves from each tobacco plant were injected. For each technical replicate, three tobacco plants were used and three technical replicates were performed.

The samples were sequenced and analyzed by Beijing Genomics Institution (BGI, Shenzhen, China), using the BGISEQ-500 platform. The raw data obtained from sequencing were filtered to remove low quality, junctional contamination, and reads with too high an unknown base N content. The clean reads were then compared to the reference genome for subsequent analysis.

The DEseq2 method was used to detect DEGs between samples (Love et al., 2014). The DEGs were functionally classified according to Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) annotation results and official classification, while enrichment analysis was performed using the phyper function in R software. The P-value was then FDR-corrected, and functions with Q-value ≤ 0.05 were considered significantly enriched.

To determine the metabolic pathways and regulatory processes associated with the white mulberry gene, MapMan tool was downloaded from https://mapman.gabipd.org/. Since it did not provide the map files of mulberry, the proteome files were uploaded to the MapMan genome annotation tool, Mercator v3.6 https://www.plabipd.de/portal/mercator-sequence-annotation.This step allowed us to obtain the mapping result file, which was then uploaded to the MapMan local software. The mapping process utilized the log₂FC values of DEGs (Thimm et al., 2004).

The raw sequencing data were submitted to the NCBI sequence read archive (SRA) database with the accession number: PRJNA818862.

Data analysis and mapping were performed with the GraphPad Prism 9.5.1. All data are expressed as the mean ± standard deviation. An unpaired two-tailed Student’s t-test was used to determine statistical significance. p < 0.05 was considered to indicate a significant difference (*p < 0.05, **p < 0.01, and ***p < 0.001).

To examine the effects of different plant hormones and abiotic stresses on pMaJOINTLESS activity, tobacco plants were treated with light, darkness, water, ABA, GA3, ethephon, and 2,4-D. With GUS chemical staining leaves were not stained blue after injection of clear water, indicating that clear water had no GUS activity ( Figure 1B ). However, after injection of the positive control agrobacterium bacterium solution, they were stained blue under all seven treatments, indicating that the 35S promoter had GUS activity ( Figure 1C ). The leaves of the experimental group were stained with different shades of blue after injection of the agrobacterium bacterium solution, indicating that the pMaJOINTLESS promoter also had GUS activity and the promoter activity intensity varied for different stress treatments ( Figure 1D ). Compared with the light blue color presented by the water treatment, the four plant hormones showed a darker blue color after treatment, with the leaves stained the most deeply under ABA treatment, followed by ethephon and GA3, and 2,4-D with a lighter color.

The promoter analysis of the MaJOINTLESS gene, suggested presence of many transcription factors including ABF1 and ABI5. ABF1 and ABI5 are downstream transcription factors of the ABA signaling pathway (Fujita et al., 2011). Since MaJOINTLESS promoter activity is sensitive to ABA. Due to this ABF1 and ABI5 were selected for detailed study. The multiple sequence alignment of ABF1 and ABI5 protein sequences of Arabidopsis and tobacco showed that both have bZIP structural domains, thus indicating that these belong to the bZIP family of proteins ( Figure 2A ).

According to qPCR analysis, MaABF1 and MaABI5 genes were expressed in all tissues, MaABF1 was higher in the AZ and leaves, and MaABI5 was higher in the AZ and winter buds ( Figure 2B ). Temporal expression analysis showed that the expression of MaABF1 in the AZ exhibited a decline starting from the S1 stage. The expression reached its lowest level during the S4 stage, followed by an increase in expression, While the expression of MaABI5 in the fruit stalk displayed a gradual change from the S1 to S3 stage, and subsequently exhibited a significant increase after the S3 stage, continuing until full maturity ( Figure 2C ). These results suggest that both MaABF1 and MaABI5 genes play important roles in mulberry fruit abscission as they are highly expressed in the abscission zone. The opposite expression trends of these two genes during mulberry fruit ripening tentatively suggest that they may have opposite roles in the process. The subcellular localization of MaABF1 and MaABI5 proteins in tobacco tissues was examined by constructing plant subcellular localization vectors ( Figure 2D ). The results showed that green fluorescence was present throughout the tobacco cells in the control group, whereas in the experimental group, the fluorescent signal was only observed in the nucleus ( Figure 2E ). These findings suggest that MaABF1 and MaABI5 proteins are localized in the nucleus.

To verify whether MaABF1 and MaABI5 can bind to the MaJOINTLESS promoter, structural domain proteins BRLZF1 and BRLZI5 were expressed in a prokaryotic expression system ( Supplementary Figure 1 ). The proteins were incubated with probes for EMSA experiments, and binding bands were observed in both experimental groups compared to the negative control (no target protein), indicating that the structural domain proteins BRLZF1 and BRLZI5 can bind to the MaJOINTLESS promoter ( Figure 3A ). Since BRLZF1 and BRLZI5 are tagged with Flag, we used the Flag antibody to bind competitively to the proteins, forming a supershift band. As the antibody concentration increased, the antibody-protein binding band strengthened while the probe-protein binding band weakened ( Figure 3B ). Next, we synthesized unlabeled biotin probes as cold probes, diluted them to 5×, 10×, and 25×, and incubated them with the proteins. We found that the binding of the biotin-labeled probe to the protein weakened as the cold probe concentration increased ( Figure 3C ). Finally, mutant probes were synthesized with biotin markers and incubated them with the proteins. No band was detected in the mutant group, indicating that MaABF1 and MaABI5 are specific to the predicted binding site of the MaJOINTLESS promoter ( Figure 3D ).

To verify the regulatory role of MaABF1 and MaABI5 on the MaJOINTLESS promoter, we performed a dual-luciferase assay. MaABF1 and MaABI5 were separately constructed onto the pGreenII62-SK vector, while the MaJOINTLESS promoter was ligated to the pGreenII0800-LUC vector, and then transformed into Agrobacterium. After expression in injected tobacco leaves, the enzyme activity ratios of firefly luciferase and sea kidney luciferase were measured. The results showed that MaJOINTLESS activity was significantly enhanced when MaABF1 was expressed, indicating that MaABF1 could transcriptionally activate MaJOINTLESS ( Figure 4E ). On the other hand, when MaABI5 was expressed, MaJOINTLESS activity was significantly reduced, indicating that MaABI5 could transcriptionally repress MaJOINTLESS ( Figure 4F ). Interestingly, when both MaABI5 and MaABF1 were present, MaJOINTLESS activity was significantly reduced, suggesting a transcriptional repression effect ( Figure 4G ). We speculate that MaABI5 and MaABF1 may form a transcriptional complex, preventing MaABF1 from binding to the promoter region and leading to loss of its activation effect.

Further, MaABF1 and MaABI5 were silenced in tobacco using VIGS. Following 15 days silencing of NbPDS gene in tobacco, the newly grown leaves of all three positive plants showed varying degrees of albinism, indicating that the positive plants were already prone to gene silencing ( Figure 3A ). The new leaves of each silenced strain were collected, and wild-type tobacco was used as a control. It was found that all silenced strains expressed the CP protein of the viral capsid, while the wild type did not show the CP protein. ACTIN (GenBank: AY594294.1) was used as an internal reference gene. This indicates that the virus was successfully expressed in tobacco leaves ( Figure 3B ). To assess the effect of gene silencing, qPCR was used to determine the relative expression of NbPDS, NbABF1, and NbABI5. The results are shown in Figure 3C . NbPDS was significantly down-regulated in all positive silencing strains; NbABF1 was significantly down-regulated in TRV2-NbABF1-1 and TRV2-NbABF1-2, while no significant difference was found in TRV2-NbABF1-3; TRV2-NbABI5-2 and TRV2-NbABI5-3 had stronger silencing effects. qRT-PCR results were consistent with the semi-quantitative-PCR results.

To confirm the effect of NbABF1 on fruit abscission, the expressions of NbABI5, NbCel1, NbCel2, NbJOINTLESS, NbLs, NbWUS, and NbBOP1 were examined in the TRV2-NbABF1-1 strain. The results showed that the downregulation of NbABF1 led to a decrease in the expression of NbABI5, NbJOINTLESS, NbCel1, NbCel2, NbLs, NbWUS, and NbBOP1 ( Figure 3D ). This indicates that NbABF1 promotes the expression of fruit abscission-related genes. To verify the effect of NbABI5 on fruit abscission, we measured the expression of abscission-related genes in wild-type and silenced strains. The results are shown in Figure 3E . The downregulation of the NbABI5 gene led to an up-regulation of NbABF1, NbJOINTLESS, NbCel1, NbCel2, NbLs, NbWUS, and NbBOP1 expression suggestings that NbABI5 inhibits the expression of fruit abscission-related genes.

Our results demonstrate that MaABF1 and MaABI5 regulate the expression of MaJOINTLESS, positively and negatively, respectively ( Figure 5 ).

It has been shown that J mutants inhibit the development of AZ at the stem of tomato, exhibiting significant phenotypic differences from wild-type AZ (Roldan et al., 2017). Our results found that the longitudinal section of normal fruit stalks stained with phloroglucinol showed non-lignified structures compared to deciduous fruit stalks, and this structure is the AZ organization ( Figure 6A ). Transcriptome sequencing of mulberry AZ was performed to obtain genome-wide gene expression profiles of mulberry AZ tissues at the mature stages.

PCA was conducted to confirm the differences between the MN and MD samples, and the variance explained by the first two principal components was determined (97.54% and 2.46%, respectively). The results showed that there were differences in the MN and MD sample groups, and each sample formed a separate cluster ( Figure 6B ).

A total of 25,227 genes were identified in the stalk tissue of young mulberry fruits. To compare gene expression differences between fallen and normal fruit, DEGs were screened between fallen and normal fruit stalks at maturity using |log2foldchange| ≥ 1 and P-value ≤ 0.01 as the threshold. Among them, 556 DEGs were identified after fruit drop at maturity, of which 382 genes were up-regulated and 174 genes were down-regulated. Of all expressed genes, only 2.2% were differentially expressed ( Figure 6C ). Analysis of TFs showed that 26 transcription factors are involved in the process of mulberry ripening fruit abscission, with MYB, bHLH, AP2-EREBP, and MADS accounting for the largest proportion ( Figure 6D ). We found that the transcription factors involved in mature fruit abscission in mulberry were like those of other model plants.

To verify the authenticity of RNA-seq gene expression data obtained qRT-PCR analysis of six randomly selected DEGs was performed. PRR37 (XM_010114016.2) and alpha-galactosidase 1 (XM_024170219.1) were down-regulated, while glucan endo-1,3-beta-glucosidase (XM_010102943.2), EP3 (XM_010095799.2), germin-like protein subfamily 1 member 17 (XM_024174055.1), and HSPRO2 (XM_010096776.2) showed up regulation. These results were consistent with the RNA-Seq data ( Figure 6E ).

The GO classification of DEGs in mulberry fruit abscission can be categorized into three categories: biological processes, cellular components, and molecular functions (Gaudet et al., 2021). Significantly enriched genes include those involved in cellular and metabolic processes, cellular anatomical entities and intracellular genes, catalytic activity, and binding genes. The most abundant genes were those involved in catalytic activity, suggesting that many important enzymes may be involved in the fruiting process of mulberry, resulting in a series of catalytic reactions ( Figure 7A ).

The KEGG pathway enrichment analysis showed that most of the DEGs were concentrated in the MAPK signaling pathway, flavonoid biosynthesis, RNA degradation, circadian rhythm-plant, glycine, serine and threonine metabolism, isoquinoline alkaloid biosynthesis, and other pathways. Additionally, the amino sugar and nucleotide sugar metabolism were also involved in the abscission of mulberry fruit ( Figure 7B ). Importantly, all genes involved in this pathway were upregulated ( Supplementary Table 2 ; Table 3 ). Therefore, we speculated that the MAPK signaling pathway and flavonoid biosynthesis pathway may affect the shedding of mature mulberry fruits.

The MapMan pathway annotator was used to outline the metabolism of DEGs ( Figure 7C ). DEGs were mainly expressed in secondary metabolism, minor CHO, lipid, fermentation, TCA, and light response pathways. Interestingly, most DEGs were up-regulated during metabolism, while a few genes were down-regulated in expression during MFA of mulberry. All DEGs in the transcriptome were transformed into MapMan-recognized IDs, and the expression of each DEG in response to biotic stress was expressed by heatmap representation. It can be found that most of the up-regulated DEGs are concentrated in auxins, ABA, cell wall, beta glucanase, glutathione-S-transferase, redox state, and signaling genes, while a few down-regulated genes encode secondary metabolites and respiratory burst gene ( Figure 7D ).

These results predict that DEGs have critical roles in fruit abscission mechanism. Seven DEGs having high expression in fruit abscission (germin-like proteins (GLPs), pathogen infection-related NHL (NDR1/HIN1-like) family genes, the amphiphysin-like protein 1 (ALP1), glucosidase genes, copine family gene BON1, and acidic endogenous chitinase-like genes) were identfied. These results provide a reference for subsequent in-depth studies on the molecular mechanism of fruit abscission in mulberry.