A total of 30 tree peonies (cv. ‘Xueyingtaohua’) were planted and grown for four years in a greenhouse at Qingdao Agricultural University (Qingdao, China) (22 ± 3 °C day, 15 ± 3 °C night, natural 10 h light/14 h dark photoperiod). Five slightly opening flowers in similar bud states were selected and removed from each test plant. Samples were delivered to the laboratory within one hour of harvest and stems were pruned to 15 cm, immersed in distilled water, and placed in an incubator (20 °C, 14 h light/10 h dark) for 36 hours. Twenty cut flowers were randomly selected from each group and treated with either a solution containing 100 μM, 200 μM IAA (Indole-3-Acetic Acid, IAA) and 100 μM TIBA (2,3,5-triiodobenzoic acid, TIBA). The remaining samples were still placed in distilled water as the control group. 10 cut flowers were randomly selected from each treatment group and control group for shedding rate statistics. Following treatment, organs of interest were individually removed from the flowers. Double-sided blades were used to excise the ovary by cutting along its boundary with the receptacle. Whole stamens were removed and kept intact. The abscission zone was represented by the 2 mm segment located at the junction between the petals and the receptacle. The proximal and distal portions of this section were represented by the upper 2 mm (including petal tissue) and lower 1mm (including the ring-shaped receptacle), respectively. Three cut flowers were selected for each sampling stage for in situ hybridization, fluorescence quantification, and endogenous IAA determination experiments, respectively.

The known PIN protein sequences of Arabidopsis thaliana [TAIR - Home Page (arabidopsis.org)] and Oryza sativa [Rice Genome Annotation Project (uga.edu)]were downloaded to blast tree peony genomewide (Yuan et al., 2022). The HMMER software was utilized with default parameters to conduct a search for candidate PIN protein sequences according to transmembrane domain (PF03547.22), followed by further identification of the sequences through a homology search using the BLAST default parameters. The candidate tree peony PIN genes were validated using the SMART web tool [SMART: Main page (embl-heidelberg.de)], and manually curated to remove truncated and redundant proteins. The identified tree peony PIN proteins were designated based on their phylogenetic relationships with Arabidopsis homologs. The subcellular localization information of the PsPIN protein sequences were analyzed by the Softberry service platform ProtComp 9.0 (SoftBerry - ProtComp (Plant) HELP) (Yu et al., 2008a). The MapInspect software (http://mapinspect.software.informer.com/) (Wu et al., 2019) was used to analyze physical position and chromosomal localization of PsPIN genes. The MEME web server [Introduction - MEME Suite (meme-suite.org)] (Bailey et al., 2015) was employed to identify intricate motifs of PsPIN proteins, utilizing the default parameter settings except for a maximum motif count of 20. A total of 112 PIN amino acid sequences from 11 species (Arabidopsis thaliana, Paeonia ostii, Solanum lycopersicum, Solanum tuberosum, Gossypium raimondii, Zea may, Oryza sativa, Cystopteris fragilis, Picea abies, Populus trichocarpa, and Glycine max) were retrieved from the NCBI database. Multiple sequence alignment was performed using ClustalW program in MEGA-X software (Kumar et al., 2018), and the phylogenetic tree was constructed using maximum likelihood (ML) with 1000 bootstraps.

The PsPIN gene structure (exon-intron) was defined using the online software Gene Structure Display Server (GSDS) [Gene Structure Display Server 2.0 (gao-lab.org)] (Hu et al., 2015). DNAMAN software (https://www.lynnon.com/pc/framepc.html) and TMHMM Serverv.2.0 (http://www.cbs.dtu.dk/services/TMHMM/) were applied to perform protein family sequence alignment and transmembrane structure analysis. The cis-regulatory elements in the promoter region (2,000 bp upstream of the starting codon) of the PsPINs were searched by the online program of PlantCARE [PlantCARE, a database of plant promoters and their cis-acting regulatory elements (ugent.be)] (Lescot et al., 2002). Gene structure, cis-acting element numbers and responsive functions were visualized using TBtools (Chen et al., 2020). NetPhos 3.1 web server (NetPhos 3.1 - DTU Health Tech - Bioinformatic Services) was used to analyze the phosphorylation sites of the PsPIN proteins.

Total RNA was extracted from 100 mg of each sample using a Steady Pure Plant RNA Extraction Kit (TaKaRa, Beijing, China) according to the manufacturer’s instructions. The first strand of cDNA was synthesized from 1 µg of total RNA using a HiScript III RT SuperMix (Vazyme, Nanjing, China) for qPCR reverse transcriptase at a final volume of 50 µL. qRT-PCR was performed using a SYBR^® Premix Ex Taq™II Kit (TaKaRa, Dalian, China). All reactions were performed in triplicate. The relative expression level was calculated by the 2^-ΔΔCt method (Livak and Schmittgen, 2001). The primers used for qRT-PCR are listed Supplementary Table S8 .

IAA standards, antigens, and antibodies were obtained from Qingdao Agricultural University. All samples were homogenized and extracted in cold 80% methanol (4°C) containing 1mmol/L butylated hydroxytoluene for 4 h. Samples were then centrifuged at 4000rpm for 15 min (4°C). Extracts were passed through a C18 Sep-Pak cartridge (Waters, USA), dried using N₂, and then dissolved in a mixture comprising 2 mL PBS (pH 7.5), 0.1% Tween 20 (v/v), and 0.1% gelatine (w/v). Next, 96-well microfiltration plates were coated with an IAA antigen in NaHCO₃ buffer (50 mmol/L, pH 9.6) and incubated overnight at 4°C. Plates were then washed four times with a mixture of PBS (pH 7.5) and 0.1% Tween 20 (v/v). Subsequently, all wells were filled with 50 μL of IAA antibodies along with a 50 μL aliquot of either a sample extract or IAA standard. Plates were then incubated at 37°C for 30 min. After an additional wash with the PBS and Tween 20 mixture, wells were supplemented with IgG horseradish peroxidase, incubated for 30min at 37°C, and washed a final time. Finally, ortho-phenylenediamine was supplied to act as a substrate before a final 30 min incubation at 37°C. Absorbance was detected at 490 nm.

The AZ samples were collected at 0, 24, 48, and 72 hours after flowering for this assay. The in vitro transcription vector pGEM-T was constructed using a PsPIN4-specific fragment amplified. with the Antisense probe sequence is listed in Supplementary Table S9 . Needle labeling was performed following the protocols provided by the DIG RNA Labeling Kit (SP6/T7) (Roche, USA) and DIG Nucleic Acid Detection Kit (Roche, USA). The probe concentrations used was 211.7 ng/ul for PIN. The protocol for in situ hybridization was modified based on Abcam’s guidelines ( In situ hybridization (ISH) protocol | Abcam). Detailed steps for paraffin sectioning referred to Shi (Shi et al., 2017), finally, the sections were stained overnight with NBT/BCIP (Roche, Germany) at 37°C. Subsequently, the samples were examined under a Nikon Eclipse ci optical microscope (Nikon, Japan).

The overexpression vector 35S:PsPIN4 was generated by cloning the CDS of PsPIN4 into the vector pCAMBIA1300. The primers used for PsPIN4 CDS amplification are listed in Supplementary Table S8 . Transgenic Arabidopsis plants were obtained according to the floral dip transformation method (Clough and Bent, 1998) and were then used for phenotype observation and further assays. The pBS was quantified as the force in gram equivalents required for pulling a petal from flower (Butenko et al., 2003), as determined by using a digital force gauge (Model: HF-2, Lunjie Electromechanical Instrument Co., Ltd., Shanghai). A total of 20 petals per position were measured.

A promoter sequence of 2,000 bp upstream of the starting codon was cloned and fused into pCAMBIA1381 vector with a GUS reporter gene. The primers used for PsPIN4 promoter amplification are listed in Supplementary Table S8 . Arabidopsis homozygous T₃ generation plants were utilized for GUS staining analysis. Samples at different stages (1, 2, 3, and 4 weeks) as well as floral organs of different positions (P1-P7) were selected to observe the expression of ProPsPIN4 throughout the whole plant using GUS staining. The samples were performed in the staining solution (50 mM phosphate buffer (pH7.2), 0.1% (v/v) Triton ™ X-100, 0.5 mM K₄Fe (CN) 6H₂O, 0.5 mM K₃Fe (CN) ₆, and 0.5 mM X-Gluc) at 37°C in the dark. Then 95% ethanol was applied to decolorize the tissues and examined under the ZEISSV11 stereoscope (Nikon Japan).

For immunoblot analyses, total proteins were extracted from 0.1 g of four transgenic lines (Arabidopsis thaliana at 1,2,3, and 4 weeks of age and floral organs at 2,4, and 6 positions) and NT plants in 3 ml of extraction buffer [50 mM Tris–HCl (pH 8), 10 mM MgCl₂, 1 mM EDTA 0.5 M (pH 8), 5% glycerol, 1 mM DTT, 0.1% Triton X-100 and 1 mM (PMSF)]. The homogenate was placed on ice for 5 min, and then centrifuged at 13,000 × g for 15 min. The proteins concentrations of the resulting supernatant were determined by the Bradford method (Bradford, 1976). Total proteins samples were then electrophoretically separated on 12% (w/v) SDS–PAGE gels and blotted by the semi-dry method (iBlotTM Gel Transfer Stack) onto a nitrocellulose membrane (Invitrogen). GUS protein detection was carried by the same protocol described by Ben Saad et al. (Ben-Saad et al., 2015).

The PsPIN4 ORF without stop codon was amplified with the primers PsPIN4-GFPF and PsPIN4-GFPR ( Supplementary Table S8 ). The PsPIN4::GFP fusion expression vector was constructed and used to transform Agrobacterium (Agrobacterium tumefaciens) GV3101. Agrobacterium was cultured in LB liquid medium (with kanamycin and rifampicin) until an OD₆₀₀ of 0.6–0.8 was obtained. Further, the cells were centrifuged at 5,000 rpm for 10 minutes. The cell pellet was diluted to obtain an OD₆₀₀ of 0.6–0.8 with a solution of 10 mM MES, 200 mM acetylsyringone, and 10 mM MgCl₂ (pH 5.6), and injected into tobacco leaves. The infected plants were cultured in darkness for 24 hours, and then in normal conditions (25°C, 16 hours light/8 hours dark) for 24 hours. The nuclear labeled protein GHD7::CFP and the membrane labeled protein PIP2;1::RFP were interchanged for accurate localization, and the fluorescence was observed using a laser confocal microscope.

Statistical analysis was performed by GraphPad Prism 7 (GraphPad Software Inc. San Diego, CA. USA) and t-tests with P < 0.05 (∗) and P < 0.01 (∗∗) was used to indicate statistically significant level. All results were presented as the means ± standard deviation (SD) and at least three replicates were set up.

In this study, a total of 8 PsPIN genes were identified in tree peony and were renamed PsPIN1, PsPIN2a-b, PsPIN4, PsPIN5, PsPIN6a-b and PsPIN8 ( Table 1 ) according to Arabidopsis homologs. Information regarding these genes and their corresponding proteins including gene ID, location, number of exons, protein length (aa), molecular weight (MW), theoretical isoelectric point (pI), and subcellular location are showed in Table 1 and Supplementary Table S1 . Among the 8 PsPIN proteins, the lengths and molecular weights varied greatly, ranging from 90 aa (PsPIN6b) to 654 aa (PsPIN4) and from 10192.01 Da to 70955.96 Da, respectively. Our prediction of subcellular localization revealed that PsPIN proteins are primarily distributed on the cell membrane. In addition, the presence of shorter protein sequences may be attributed to the genome annotation process.

Through comparisons with the tree peony genome, our results indicate that the newly identified PsPIN genes are distributed across three chromosomes ( Figure 1 ), the majority of which were located on chromosomes I and V (each containing three PsPIN members). In view of this, we searched the tree peony genome for segmental and tandem duplicate gene pairs and found no repeat events among PsPIN family members.

To elucidate the evolutionary relationship within the PsPIN gene family, we used the MEGA-X software to construct a phylogenetic tree from the inferred full-length amino acid sequences of 112 PIN protein sequences collected from 11 species. Based on the classification methods of the AtPINs, the tree revealed six distinct branches (AtPIN1, AtPIN2, AtPIN5, AtPIN6, AtPIN8, and AtPIN3/4/7), The 8 newly identified PsPIN proteins were categorized into six groups: PsPIN1, PsPIN2a-b, PsPIN4, PsPIN5, PsPIN6a-b and PsPIN8. The evolutionary analysis revealed a robust correlation between peony and poplar PIN. We employed the MEME (Multiple Em for Motif Elicitation) to unveil the motif compositions of the 112 PIN proteins and identified 20 conserved motifs through structural analysis ( Figure 2 ). In the “long” PIN group, PsPIN1 lacked only the 17th motif, whereas PsPIN2a lacked motifs 9, 12, 17, 18 and 20. Potentially as a result of incomplete genome annotation, only one or two motifs were identified in PsPIN2b, PsPIN6a, and PsPIN6b. Conversely, “short” PINs such as PsPIN5 and PsPIN8 were found to contain shorter central hydrophilic rings and lack the typical 10, 15, 11, 9, 18, 17, 12 and 14 motifs.

The structural characteristics of the gene family were investigated by integrating the PIN genes from Arabidopsis, rice, and tree peony. The results illustrated great variability, with the quantity of introns among the PsPINs ranging from 3 to 6 ( Figure 3A ). Specifically, PsPIN5 and PsPIN8 each contained four introns, while PsPIN1, PsPIN2a, and PsPIN4 contained five introns, and PsPIN2b and PsPIN6a contained six introns. Due to potentially incomplete genomic annotation, only three introns were detected in PsPIN6b. It is important to highlight the absence of UTR regions associated with PsPIN2 and PsPIN6, which also displayed shorter exon segments. Additionally, PsPIN2b possessed notably longer gene spans.

The transmembrane motif of the PsPIN protein was presented for further structural analysis ( Figure 3B ). We identified a distinct central hydrophilic ring (HL), which separates the two hydrophobic domains (HD) located at the N-terminus and C-terminus. The variances in length recorded in the central hydrophilic region contribute greatly to PsPIN diversity.PsPIN2a, PsPIN4, and PsPIN5 possessed five transmembrane helices (Helix1-Helix5) at the N-terminus, whereas PsPIN1 and PsPIN8 lacked the third transmembrane helix. Specifically, only PsPIN2a contained five transmembrane helices at the C-terminus, while PsPIN1 and PsPIN4 had 4 helices, and PsPIN8 had only 3 helices. The eighth helix of PsPIN2b, PsPIN6a, and PsPIN6b differed significantly from other members. The conserved NPXXY structure, which is crucial membrane and receptor proteins interaction in endocytosis, was found near the hydrophobic and hydrophilic regions of the C-terminus. Notably, the tyrosine (Y) in PsPIN8 was replaced with histidine (H). Amino acid fragments are shown in Supplementary Table S4 .

The PlantCare online program was utilized to identify putative cis-elements in the 2,000 bp promoter regions of the PsPIN genes. A total of 41 cis-regulatory elements were identified and classified into three categories: growth and development, phytohormone, and stress response ( Figure 4 ). Growth and development responsive elements primarily included photoresponse (MRE, TCT-motif, AE-box, G-box, Box 4, I-box, GATA-motif, GT1-motif, ACE, TCCC-motif, TGACG-motif, AAAC-motif, Gap-box, CGTCA-motif, GA-motif, L-box and ATCT-motif), zein metabolism regulation (O^2–site), endosperm expression (GCN4-motif) and meristem expression (GAT-box). Phytohormone responsive elements were comprised of responses to abscisic acid (ABRE), methyl jasmonate (TGACG-motif, CGTCA-motif), gibberellin (P-box, GARE-motif, and TATC-box), auxin (CATATG-site, ARF-binding, ACTTTA-site, TGA-element, and AuxRR-core) and salicylic acid (TCA-element). Elements associated with stress response mainly included responses to wounding (WUN-motif), defense and stress response (TC-rich), stress (STRE), low temperature (LTR, W box), drought (MBS), anaerobic induction (ARE) and other complex stress elements (MYB and MYC). Interestingly, with the exception of PsPIN2b, all members of the PIN gene family, exhibited various types of auxin -responsive elements.

The online software NetPhos 3.1 Server was used to predict post-translational protein phosphorylation. Phosphorylation sites were abundant among the 8 PsPIN proteins, with serine (Ser) representing the most frequently modified site, followed by threonine (Thr) and tyrosine (Tyr) ( Figure 5 ; Table 2 ). The concentration of prediction sites in the center of the hydrophilic ring resulted in significantly fewer sites accounting for PsPIN5 and PsPIN8.

Gene ontology analysis was performed to elucidate the functional classifications of PsPIN genes and explore their involvement in various biological processes ( Figure 6A ). The studied genes were most highly represented in the “Biological Process” ontology by the following GO terms: “regulation of biological quality,” “homeostatic process,” “regulation of auxin-mediated signaling pathway,” and “obsolete auxin homeostasis”. Similarly, genes were prevalent in the “cell periphery” and “obsolete intracellular part” terms of the “Cellular Component” ontology. Notably, the only observed cellular function among the 8 genes was related to auxin transmembrane transport activity. This analysis indicates that PsPIN genes are likely involved in various regulatory pathways in plants, including cell morphogenesis, gravitropism, embryo development, and root system development. Furthermore, this analysis revealed new functional categories, such as response to hypoxia, plasmodesma regulation, monosaccharide response, and gametophyte development. Taken together, these findings provide valuable insights for future investigations into PsPIN functional divergence.

To better understand the function of PIN genes in tree peony, we next assessed the expression of PsPIN genes in nine distinct tissues (roots, stems, leaves, petals, receptacles, stamens, ovaries, bracts, and sepals) using qRT-PCR analysis ( Figure 6B ). The result revealed a minimal expression of PsPIN2 and PsPIN6 in any tissue and an exclusive expression of PsPIN8 in the ovary tissue. In contrast, the expression of PsPIN1, PsPIN4, and PsPIN5 varied across different tissues. Notably, PsPIN4 exhibited a significantly high expression in the petal tissue.

Auxin’s influence on the rate of petal shedding in cut peony flowers was examined through the use of IAA and TIBA ( Figure 7A ). The average abscission time following IAA treatment (153.6 h) was 29.30% longer when compared with the control (118.8 h). Conversely, TIBA treatment was found to significantly accelerated the abscission process, resulting in an average abscission time of only 94.8 hours ( Figure 7B ).

The quantification of internal plant IAA following exogenous IAA and TIBA treatments was used to reflect auxin’s influence. Following IAA treatment, there with a significant increase in the distribution of the hormone within the distal part, proximal part, and AZ. After 72 h, IAA levels within the distal portion and the entire AZ were elevated compared to the control, suggesting that shedding had not been initiated. TIBA treatment, however, decreased the quantities of IAA in the three tissues, especially in the distal portion and AZ. Following 48 h, auxin was only detected in very low quantities in both, reaching stabilization in the middle to late flowering period ( Figure 7D ).

AZ initiation is dependents on auxin level changes both internally and in surrounding tissues. However, the crucial functions performed by PIN transporters in this process, particularly in flower organ shedding, have seldom been studied. To screen potential PsPINs involved in petal abscission initiation, We analyzed the expression patterns of PsPIN1, PsPIN4, and PsPIN5 in various regions (distal part, proximal part, and AZ) during flowering stages ( Figure 7E ). The expression of PsPIN1 was consistently low across all tissues. In the distal part, PsPIN4 expression initially increase rapidly at 12 h, followed by a gradual decline. A similar pattern was observed in the proximal part and AZ, with an initial upward trend before 36 h followed by subsequent decline. Notable, the expression of PsPIN4 in the distal part was lower than that in the proximal part after 36 h. The expression of PsPIN5 exhibited a generally declining trend across all three tissues, with a significant increase observed at 60 h in the distal part. These findings highlight PsPIN4 as a candidate gene involved in the regulating of petal abscission.

The application of in situ hybridization enabled the visualization of PsPIN4 expression patterns during petal shedding. The results confirmed the findings of qRT-PCR analysis. Initially, the were only minimal hybridization signals were detected within the AZ, but by 24 h, signals were widespread throughout the entire region ( Figure 7C ). Nevertheless, strong signals persisted on the upper surface of proximal parts, as well as within the vascular bundles and their periphery. By 72 h of flowering, the signals marginally localized within the vascular bundles of the receptacle, reflecting a decline in auxin transport efficiency mediated by PsPIN4. To further investigate the underlying relationship between PsPIN and auxin, exogenous IAA was applied 36 hours after flowering and PsPIN4 expression was analyzed after an additional 48 h ( Figure 7F ). The localized changes in IAA concentration provided feedback on the expression of PsPIN4, actively regulating the auxin distribution among tissues. A significant increase in PsPIN4 expression was observed in the distal part, AZ, and proximal part. Due to continuous IAA supply, PsPIN4 expression was consistently elevated and did not initiate the abscission layer at 72 h. Furthermore, higher expression levels were more evident with a treatment concentration of 200 μM compared to 100 μM. Overall, our study suggests that PsPIN4-mediated auxin transport may be involved in petal abscission.

The ectopic expression of the PsPIN4 promoter was employed to further elucidate the functional role of PsPIN4. The GUS signal in transgenic Arabidopsis was recorded at 1, 2, 3, 4, and 7 weeks respectively. Western blot showed that the GUS fusion protein driven by the PsPIN4 promoter was effectively expressed in the experimental group ( Figure 8C ). The GUS signal was predominantly distributed in the roots, leaves, petals, sepals and receptacle ( Figure 8A ). In newly formed leaves, the signal was primarily localized to the main vein. As the leaves matured, apart from the main vein, the signal exhibited a punctate distribution throughout the entire area. However, this punctate distribution was less pronounced in aged leaves. Before three weeks of age, the signal spread throughout the entire root and then weakened at its base while becoming concentrated in lateral and apical regions. Our focus of observation encompassed seven positions from top to bottom of the inflorescence for analyzing the signal within flower organs ( Figure 8B ). The signal at the apex of the stigma remained relatively stable during the early and middle stages, but exhibited a significant enhancement at positions 6 and 7. The signal within the petals initially increased and subsequently decreased, with peak intensity observed at position 3 and position 4. This signal was primarily localized along the floral veins. Additionally, there was a gradual signal increase at the junction between petals and receptors, which was particularly evident during early pod development.

The subcellular localization analysis revealed that PsPIN4::GFP exhibited co-localization in both the nucleus and plasma membrane, with a slightly stronger fluorescence signal observed in the nucleus compared to the plasma membrane ( Figure 8D ).

To further investigate the functions of PsPIN4, we examined its impact on floral organ shedding in Arabidopsis. In wild-type Col Arabidopsis, the floral organs began to fully drop at position 8. Conversely, transgenic Arabidopsis lines overexpressing PsPIN4 (35S:PsPIN4-3, 35S:PsPIN4-7, and 35S:PsPIN4-8) exhibited a significant delay in petal abscission time, with floral organs fully dropping at position 12 or 13 ( Figure 9A ). BLADE-ON-PETIOLE (BOP) was used as a gene marker for AZ formation (McKim et al., 2008). The expression of AtBOP1 and AtBOP2 were significantly reduced at position 4 and 5 in 35S:PsPIN4 compared to Col ( Figure 9B ). Furthermore, the delay in floral organ abscission was quantified by measuring petal break strength (pBS) ( Figure 9C ). The transgenic lines required more force than Col to tear petals from flowers at positions 5 and 6. Additionally, flowers with “just-visible” white petals were selected and cultured on IAA plates to record the abscission rate ( Figure 9D ). Initially, the abscission process was found to be significantly delayed in Col plants grown on IAA plates compared to MS plates. However, abscission rates for transgenic lines grown on MS plates and Col grown on IAA plates were similar. Additionally, the 35S:PsPIN4 cultures on the IAA plate demonstrated prolonged organ abscission when compared to all other groups, resulting in only a 47.93% abscission rate at 120 h. While further examining the abscission process of Atpin4, we observed premature abscission of flower organs, with dropping occurring at positions 7 or 8 ( Figure 9E ). Interestingly, the introduction of PsPIN4 under the control of the CaMV35S promoter into the Atpin4 background seemed to rescue this phenotype, as floral organs completely detached at positions 11 or 12 with a dropout rate similar to that of 35S:PsPIN4 ( Figure 9A ). The pBS measurement and abscission rate of the IAA-treated plate further confirmed these results ( Figure 9F ). Moreover, the abscission rates of the Atpin4 and Col groups on the IAA plate were similar at 120 h, while the 35S:PsPIN4/Atpin4 groups displayed significantly lower abscission rates than the other groups ( Figure 9G ). The pBS at positions 5 and 6 also indicated that the repair line could effectively delay petal shedding ( Figures 9E, F ). Collectively, this data strongly supports the role of PsPIN4 in delaying floral organ abscission in Arabidopsis.