Diversification of FT-like genes in the PEBP family contributes to the variation of flowering traits in Sapindaceae species

Abstract

Many species of Sapindaceae, such as lychee, longan, and rambutan, provide nutritious and delicious fruit. Understanding the molecular genetic mechanisms that underlie the regulation of flowering is essential for securing flower and fruit productivity. Most endogenous and exogenous flowering cues are integrated into the florigen encoded by FLOWERING LOCUS T. However, the regulatory mechanisms of flowering remain poorly understood in Sapindaceae. Here, we identified 60 phosphatidylethanolamine-binding protein-coding genes from six Sapindaceae plants. Gene duplication events led to the emergence of two or more paralogs of the FT gene that have evolved antagonistic functions in Sapindaceae. Among them, the FT1-like genes are functionally conserved and promote flowering, while the FT2-like genes likely serve as repressors that delay flowering. Importantly, we show here that the natural variation at nucleotide position − 1437 of the lychee FT1 promoter determined the binding affinity of the SVP protein (LcSVP9), which was a negative regulator of flowering, resulting in the differential expression of LcFT1, which in turn affected flowering time in lychee. This finding provides a potential molecular marker for breeding lychee. Taken together, our results reveal some crucial aspects of FT gene family genetics that underlie the regulation of flowering in Sapindaceae.

Core

The natural variation at nucleotide position − 1437 of the lychee FT1 promoter determined the binding affinity of the SVP protein (LcSVP9), a negative regulator of flowering, resulting in the differential expression of LcFT1, which in turn affected flowering time in lychee. This finding provides a potential molecular marker for lychee breeding.

Gene and accession numbers

Sequence data of Sapindaceae plants from this article can be found in the SapBase database (http://www.sapindaceae.com/Download.html)(Li et al. 2024). The remaining sequences were obtained from the Phytozome, GenBank, and Arabidopsis Information Resource (TAIR) databases. The accession numbers can be found in Table S2 and Table S6.

Introduction

Lychee (Litchi chinensis), longan (Dimocarpus longan), and rambutan (Nephelium lappaceum) are commercially important fruit trees in the Sapindaceae family, which are extensively cultivated in tropical and subtropical areas worldwide (Menzel et al. 1995; Shahrajabian et al. 2020). These species are closely related and possess valuable fruits with an edible aril (Zee et al. 1998). Although they are closely related, their flowering traits, which are directly linked to production, are not entirely identical. Lychee and longan are day-neutral plants that require prolonged exposure to low temperatures to consistently flower annually (Subhadrabandhu et al. 2000; Chen et al. 2003). In contrast, rambutan flowering is triggered by water scarcity, which occurs biannually during March to May and July to August in reaction to two short periods of arid conditions followed by intermittent rainfall (Shaari et al. 1983; Tindall, 1994 ). Longan exhibits a unique floral induction mechanism in response to potassium chlorate (KClO₃), enabling off-season and year-round fruit production, which is unparalleled in other fruit crops (Matsumoto et al. 2007). Understanding the molecular genetic processes that control flowering would help the Sapindaceae fruit industry.

Plants undergo significant physiological changes during the transition from vegetative phase to reproductive stage, which is triggered by various internal and external signals that ultimately lead to flowering. The regulatory mechanisms of flowering in model plants have been elucidated by identifying at least five genetically defined pathways (Srikanth et al. 2011). The components of these pathways can vary across species, but most internal and external signals are integrated into a few central hubs that are conserved, the most well-known of which is the florigen encoded by Flower Locus T (FT), which belongs to the phosphatidylethanolamine-binding protein (PEBP) family (Kardailsky et al. 1999; Liu et al. 2016). In plants, FT serves as a small, mobile signaling molecule that is synthesized in the leaves and moves to the shoot apex, enabling the transition to reproductive development and flowering (Corbesier et al. 2007; Mathieu et al. 2007). Despite the considerable conservation observed in flowering regulators among plant species, recent research has revealed cases where FT homologs have evolved to inhibit flowering (Blackman et al. 2010; Pin et al. 2010; Harig et al. 2012; Lee et al. 2013; Nan et al. 2014; Zhai et al. 2014). The presence of four FT homologs has been identified in the Sunflower (Helianthus annuus), with one of them exhibiting a novel repressor function attributed to a mutation (Blackman et al. 2010). Similarly, sugar beet contains the BvFT1 and BvFT2 homologs of FT with antagonistic functions (Pin et al. 2010). External loop formation of BvFT caused by the divergence of the amino acids within segment B may elucidate the opposite functions (Pin et al. 2010). The Y134 and W138 residues within segment B act to change the FT function (Pin et al. 2010). The MADS-box genes, SHORT VEGETATIVE PHASE (SVP) and SUPPRESSOR OF OVEREXPRESSION OF CONSTANS 1 (SOC1), encode essential transcription factors that play a crucial role in regulating the integration of flowering signals (Ng and Yanofsky 2011; Moon et al. 2003; Kou et al. 2022). SVP directly binds to the CArG motifs of FT and its homolog, repressing their expression in leaves (Jang et al. 2009). In contrast, SOC1 activates FT transcription in the leaf vasculature of soybean by directly binding to the CArG motif within its promoter; thus, ensuring the induction of flowering (Kou et al. 2022). SOC1 and SVP bind to CArG boxes, but their binding preferences vary. SOC1 predominantly binds to the SRF-type CArG box consensus sequence, while SVP binds to the intermediate type (Tao et al. 2012).

In contrast to mutations that impair gene function in protein-coding regions, variations occurring in either cis-regulatory elements (CREs) or regulatory regions result in alterations of transcriptional level that impact the phenotype. Variations in the MdIPT5b gene promoter region of apples results in a marked increase in cellular cytokinin levels (Feng et al. 2021). This can be attributed to the deletion of a 42-bp sequence within the promoter region, which impairs the cis-element ProRE (an ACTCAT motif), leading to stable gene expression under salt stress conditions. A variation (single nucleotide polymorphism (SNP)13 T/C) in the regulatory region of PbCPK28, a SNP-13, leads to variations in the fructose content of pear (Li et al. 2023). Similar regulatory mechanisms have been confirmed in the regulation of cold tolerance in tomatoes (Zhu et al. 2023), grain width and weight in rice (Ruan et al. 2020), and low-temperature tolerance in maize (Jiang et al. 2022). The variations in the promoter region have the potential to drive evolution, generating genetic variability that can be harnessed for domestication and breeding advances (Swinnen et al. 2016).

Despite the mechanisms of floral regulation have been thoroughly investigated in model species, such as arabidopsis (Arabidopsis thaliana) (Jang et al. 2009), tomato (Solanum lycopersicum) (Huang et al. 2021; Chong et al. 2022), and rice (Oryza sativa) (Pan et al. 2022; Tang et al. 2023), they remain poorly understood in the Sapindaceae family. Therefore, it is essential to comprehend the regulatory mechanisms that govern flowering in Sapindaceae fruit trees. Understanding the regulation of flowering time in Sapindaceae is crucial for efficient breeding and improving commercial production. Thus, in our study, we used genomic data from six Sapindaceae species and massive RNA-seq data to carry out a genome-wide characterization of the FT family in Sapindaceae and explore the variations in the coding and promoter regions of the FT genes, which contribute to the variation of flowering traits in Sapindaceae species.

Results

Gene member variations of the PEBP gene family in Sapindaceae

To investigate the composition and evolutionary relationships in the PEBP proteins from Sapindaceae, we screened the genomes of six Sapindaceae species with available genomes, including lychee (Litchi chinensis), longan (Dimocarpus longan), rambutan (Nephelium lappaceum), soapberry (Sapindus mulorossi), yellowhorn (Xanthoceras sorbifolium), and balloon vine (Cardiospermum halicacabum). Sixty PEBP proteins with complete open reading frames were identified in Sapindaceae (Fig. 1a, b, Fig. S1 and Table S1). We performed a phylogenetic analysis with 96 functionally reported PEBP proteins from 46 other major angiosperm lineages (Wickland et al. 2015; Table S2), which indicated that the PEBP genes were separated into three primary clusters of MFT-like, TFL1-like, and FT-like clades (Fig. 1a). The number of PEBP genes in Sapindaceae varied, ranging from nine to twelve copies (Fig. 1b). Lychee and longan possessed four FT genes, but there were seven FT-like genes in rambutan, which is a close relative of lychee and longan (Fig. 1b), indicating that FT gene expansion occurs in rambutan. Fewer FT-like genes were detected in balloon vine than in the other five woody Sapindaceae species, however, the MFT-like genes were expanded (Fig. 1b).

Fig. 1

The variation in gene numbers within the PEBP gene family across six Sapindaceae plants. a Phylogenetic analysis of the PEBP protein family in 52 flowering plant species. The construction of the tree was performed utilizing the Neighbor Joining in MEGA X (Kumar et al. 2018). b The PEBP gene copy numbers in each species were mapped in this table. “Clad” and “spec” are short for “caldes” and “species” respectively. c Macrosynteny across lychee, longan, rambutan, soapberry, yellowhorn and balloon vine for FT-like genes. Syntenic blocks are connected by lines. The gene pairs of FT1-like are linked with red curves, while those of FT2-like/FT3-like are connected by purple curves

To investigate the evolutionary relationships among the FT-like genes, a syntenic analysis was performed on the Sapindaceae family. The FT1 genes maintained a good syntenic relationship across Sapindaceae, indicating that FT1 is functionally conserved. The FT2 genes lost their syntenic relationship in the basal Sapindaceae plant, yellowhorn, but they were conserved in the other Sapindaceae plants. The FT1, FT2, and FT3 loci maintained good collinearity in lychee, longan, rambutan, and soapberry (Fig. 1c). In particular, the FT1 gene was tandemly duplicated into four copies in rambutan, which led to the expansion of the FT genes (Fig. 1c). A similar tandem replicate event occurred in FT1 loci of yellowhorn (Fig. 1c), and MFT1 loci of balloon vine (Fig. S2). These results suggest that tandem duplication events contribute to PEBP gene expansion in Sapindaceae.

Functional antagonism of FT homologs in floral induction

Florigen FT induces flowering in angiosperms, but some FT homologs repress flowering (Wickland et al. 2015). A previous study demonstrated that the antagonistic function of FT homologs is caused by changes in the external loop of the FT protein in sugar beets, which is encoded by segment B of the fourth exon (Pin et al. 2010). To identify the potential flowering inducer and repressor of FT homologs in Sapindaceae, the FT amino acid sequences of the segment B fragments from the six Sapindaceae species and other functionally characterized species were aligned (Fig. 2a). Tryptophan at position 138 (W138) was conserved across Sapindaceae species. This observation suggests that the amino acid substitution in position 134, either a Y (tyrosine) or a non-Y residue, may be more important for the functional divergence of FT-like homologs in Sapindaceae species (Fig. 2a). Based on the amino acid variation in position 134, we found that there are two types of FT gene in Sapindoideae, Y in FT1/FT3/FT4-like and N in FT2-like (Fig. 2a). Subsequently, we performed a folding prediction and our analysis revealed that the external loops in segment B of longan FT1 (the main floral inducer) and FT2 were conformationally different (Fig. 2b). This finding suggests that differences, particularly at residue Y134/N134, may confer functional antagonism between these two genes. We then detected the changes in expression of DlFT1 and DlFT2 in the shoot apices of longan before the initiation of floral primordia. The results revealed that the expression levels of DlFT1 and DlFT2 were nearly absent (Fig. S3a), suggesting that FT in longan is barely expressed in the shoot apices. We further explored the expression profiles of DlFT1 and DlFT2 in leaves during natural flowering, and found that a dramatic increase in DlFT1 from − 28 to − 7 days before flowering, while DlFT2 exhibited a slight decrease (Fig. 2c). Additionally, the expression of LcFT1 in lychee was enhanced during reactive oxygen species (ROS) and low-temperature treatments to induce flowering, while LcFT2 decreased expression in leaves (Lu et al. 2020a; Zhang et al. 2017) (Fig. S3c, d), suggesting their possible functional antagonism. Taken together, we infer that the FT1-like genes may be conserved flowering inducers in Sapindaceae, while the FT2-like genes may function as flowering repressors due to the key substitution at FT residue 134.

Fig. 2

Investigating the crucial regions contributing to the antagonistic functions of FTs. a The alignment of the amino acid sequence in the fourth exon for the FT inducer and inhibitor from onion (Lee et al. 2013), sugar beet (Pin et al. 2010), tobacco (Harig et al. 2012), sunflower (Blackman et al. 2010), soybean (Nan et al. 2014; Zhai et al. 2014; Kong et al. 2010), arabidopsis (Kardailsky et al. 1999; Kobayashi et al. 1999), lychee (Ding et al. 2015), longan (Winterhagen et al. 2013), rambutan, soapberry, yellowhorn and balloon vine. Red dots indicate inducers that have been functionally validated, while blue triangles indicate repressors that have undergone functional validation. Pink dots represent speculated activators, while light blue triangles represent speculated inhibitors. b The prediction of protein folding for DlFT1 and DlFT2. The molecular structures were visualized using solid three-dimensional traces in a diverse color scheme for alignment. c The expression pattern of FTs during flower induction in longan leaves was analyzed using quantitative real-time polymerase chain reaction (qRT-PCR). The − delta CT (− ΔCT) calculation was employed to normalize expression levels. The data presented represent the mean ± standard deviation (SD) from three independent biological replicates, with Actin-100 (Dil.11g013910.1) used as the internal control gene. d The phenotype of Arabidopsis with the overexpression of longan FT1 and FT2 genes. All plants were grown under long-day conditions with 16 h of light and 8 h of darkness at 23℃. The left image depicts the phenotype after 26 days of sowing, while the right image showcases the phenotype after 30 days of sowing. Col (WT) represents the wild-type Arabidopsis with a Col-0 background. OE represents the transgenic lines with DlFT overexpression. T₂ generation lines with DlFT1 and DlFT2 genes are represented by numbers 1 to 3 and 16 to 18, respectively, with a scale of 1 cm. e The statistics of the number of rosette leaves during flowering for both wild-type and transgenic Arabidopsis plants, with six plants used for each line. The statistics for wild-type Arabidopsis began approximately 30 days after sowing, while the OE-DlFT1 lines started around 26 days after sowing. The OE-DlFT2 lines commences around 36 days after sowing. The data is presented as mean ± SD. One-way analysis of variance (ANOVA) was performed. P value: *P < 0.05, **P < 0.01, ***P < 0.001

To validate whether FT1 and FT2 promote or inhibit floral transition among Sapindaceae species, DlFT1 and DlFT2 were constitutively overexpressed and driven by the CaMV 35S promoter in Arabidopsis. Under long-day conditions, DlFT1 overexpression plants started to flower when there were 7.5 rosette leaves, much earlier than WT control which flowered with 14.3 leaves (Fig. 2d, e, Fig. S3b). Conversely, the floral initiation of DlFT2 transgenic plants was dramatically delayed until bearing 31.0 leaves (Fig. 2d and e, Fig. S3b). Thus, ectopic expression of DlFT1 dramatically accelerated floral initiation, while ectopic expression of DlFT2 substantially postponed floral initiation, confirming their functional antagonism in flowering induction.

Specific insertion in the DlFT1 promoter may be involved in flowering diversification in longan

As FT1-like genes are the flowering-promoting genes in Sapindaceae, we next studied the regulatory differences in the FT1-like genes among the Sapindaceae species. Nucleotide sequences of the FT1-like genes (including 3.0 kilobases (kb) upstream and 1.0 kb downstream regions) were extracted from three longan varieties (‘SX’, ‘HML’, and ‘JDB’), three lychee varieties (‘FZX’, ‘GW’, and ‘HML’), as well as rambutan, soapberry, yellowhorn, and balloon vine for comparative analysis (Adrian et al. 2010). Interestingly, the alignment revealed the specific 321-bp insertion located between 619 and 940 nucleotides upstream from the DlFT1 start codon in the three longan cultivars (Fig. 3a, highlighted in red box). To investigate whether this 321-bp insertion harbored longan-specific CREs, we compared the CREs among the complete promoter regions of LcFT1, NlFT1, the 321-bp insertion, and the DlFT1 promoter region and excluding the 321-bp insertion (Fig. 3b). Consequently, two specific CREs within the 321-bp insertion of the DlFT1 promoter were identified (Fig. 3b, Table S3, Xu et al. 2011; Yin et al. 2016). One was an SRF-type CArG-box (CC[A/T]6GG) (Fig. 3c), which is a binding motif for the MADS domain proteins associated with flower formation (Tao et al. 2012). For example, the MADS-box protein SOC1, which activates FT transcription by directly binding to the CArG motif within its promoter (Kou et al. 2022), has the potential to bind to the SRF-type CArG-box, suggesting that a specific regulation pathway may be involved in flowering of longan (Fig. 3d).

Fig. 3

Conservation analysis of FT1-like promoter sequences in Sapindaceae species. a Pairwise alignment of FT1 nucleotide sequences linked to upstream of the ATG, spanning 3.0 kilobases, and downstream of stop codons, spanning 1.0 kilobase from six Sapindaceae species by using mVISTA (Brudno et al. 2003). The graphical output displays the base pair identity within a sliding window of 75-bp, ranging from 50 to 100%. The upper shows the FT1-like gene structure and the bottom with red box shows longan-specific conserved segments in the multiple sequence alignment of different Sapindaceae species by ClustalW2 (Larkin et al. 2007). The approximately 2.5-kb upstream sequence of the ATG in lychee, longan, and rambutan showed a high degree of conservation, marked by blue dotted box (b) Venn diagram assessing the count of CREs disparately among DlFT1_pro^Del−321 (2.5-kb upstream from the ATG of DlFT1 out of longan-specific 321-bp insertion), DlFT1_pro^In−321(longan-specific 321-bp insertion located between 619 and 940 nucleotides upstream from the ATG of DlFT1), LcFT1pro (2.5-kb upstream from the ATG of LcFT1), NlFT1_pro (2.5-kb upstream from the ATG of NlFT1-a). The red triangle highlights the unique CREs of longan-specific conserved segments in DlFT1_pro^Del−321_, DlFT1_pro^In−321, LcFT1_pro, and NlFT1_pro. c Cis-regulatory sequences of longan-specific 321-bp insertion are highlighted in b with red triangle. d Cis-element prediction on the longan-specific 321-bp insertion and red rectangle highlight the TF binding motif (TFBM) of SOC1 using MAST (Timothy et al. 1998), by which sequences with an E-value less than 10 are included in the output. The motif logo derived from JASPAR TF binding profile associated with SOC1(MA0054.1) is provided in bottom black box

Promoter diversity of tandemly duplicated NlFT1s likely contributes to their subfunctionalization in rambutan

We characterized four tandemly replicated FT1-like paralogs in rambutan (Fig. 1c). To explore whether these NlFT1s are functionally different, we analyzed their coding sequences and determined that NlFT1-a shared 95.98%, 94.25%, and 94.25% amino acid identity with NlFT1-b, NlFT1-c, and NlFT1-d, respectively (Fig. S4a), suggesting that their functions are likely similar. Additionally, we investigated the expression patterns across various tissues and noted that NlFT1-a showed a distinct expression pattern while the remaining three NlFT1 shared similar expression profiles, with their main expression occurring in flowers (Fig. 4a, Fig. S4c). We inferred that the differential gene expression pattern between NlFT1-a and NlFT1-b/c/d was due to differences in the promoter region. Therefore, we employed the phylogenetic shadowing method to investigate the similarity in the NlFT1-a and NlFT1-b/c/d promoter regions. As results, the NlFT1-b/c/d promoter region showed similar changes compared with NlFT1-a (Fig. S4b). The evolutionary phylogeny analysis revealed that the NlFT1-a promoter sequence was different from the other NlFT1s, and the NlFT1-b, NlFT1-c, and NlFT1-d promoter regions shared higher sequence similarity (Fig. 4b). Thus, NlFT1 genes may be subfunctionalized into two subgroups.

Fig. 4

Scanning the promoter sequence of FT1s in rambutan. a The gene expression patterns of FT-like genes across various tissues in rambutan. b A Neighbor Joining phylogenetic analysis of promoter sequence associated with 4 NlFT1s in rambutan. c Venn diagram compares the number of CREs differentially among 4 NlFT1s linked to promoter sequence. The red triangles indicate the count of CREs in the promoter regions of NlFT1-b, NlFT1-c, and NlFT1-d that are different from those in NlFT1-a. d The promoter regions of NlFT1-b, NlFT1-c, and NlFT1-d contain identical CREs that are distinct from those present in the NlFT1-a gene. e Identification of CREs in the promoter regions of NlFT1s. CREs are depicted as triangles with different colors. The box displays the sequence of each CRE position in the four NlFT1 genes. In the box there displayed the forward (WUN/MYB/ARE) or reverse complementary motif (One of the two MYB on the right) sequences within NlFT1 gene promoter regions

To investigate the variations between the NlFT1-a and NlFT1-b/c/d promoters, the cis-elements in the 2.5-kb NlFT1-a/b/c/d promoter sequences were analyzed. In total, 29, 35, 33, and 31 types of CREs were predicted from NlFT1-a, NlFT1-b, NlFT1-c, and NlFT1-d, respectively (Fig. 4c). The NlFT1-b/c/d promoter sequences shared four identical CREs (targeted by WUN, two MYBs, and ARE) that were absent in the NlFT1-a promoter region. All of these shared CREs in NlFT1-b/c/d were stress-response relative regulatory elements (Fig. 4d, e, Table S4, Xu et al. 2011; Yin et al. 2016), suggesting that NlFT1-b/c/d may adopt additional regulation in response to abiotic stress.

SNP in the LcFT1 promoter affects flowering time by influencing the binding affinity of SVP proteins

After comparing the promoter region of FT1-like genes among the Sapindaceae species and different FT1 loci in rambutan, we determined that some FT1-like genes may have been subfunctionalized by the evolution of their promoter sequences. Then, we examined whether the change in a single locus of the promoter region could contribute to the regulatory diversity of the FT1 gene within a single species. Based on previously published resequencing data from 47 lychee germplasms (Hu et al. 2022), 67 SNPs were detected in the LcFT1 promoter sequences (Fig. 5a). We predicted two SVP binding motifs (intermediate type CArG box) in the highly conserved 2.5-kb sequence of the LcFT1 promoter (Fig. 5a). Notably, we discovered a SNP situated within one of these binding motifs, which was widely shared across 47 distinct lychee varieties (Fig. 5b, c, Table S5). The consensus sequence of the binding motif (CTATACAAAAAGGGA(G/A)ATAA) was located from 1,452 to 1,433 nucleotides upstream from the LcFT1 start codon, with a G/A SNP positioned at 1,437-bp (Fig. 5b). Interestingly, almost all of the extremely early-maturing lychee varieties (see purple shadow in Fig. 5c) exhibited the G allele, whereas the early-ripening lychee varieties (see blue shadow in Fig. 5c) possessed both genotypes (R represents a heterozygous G/A genotype). In contrast, all late-ripening lychee varieties (see yellow shadow in Fig. 5c) exclusively exhibited the A genotype (Fig. 5c). According to the single binding site prediction using position weight matrix (PWM) scanning, the A-type motif had a higher binding affinity to SVP than the G-type motif (Fig. 5b).

Fig. 5

SNP-1,437 affected the transcriptional regulation of LcSVP9 on LcFT1. a Distribution of SNP in LcFT1 promoter sequence of 47 lychee germplasms. The dark purple square exhibits the binding motif of SVP protein predicted by MAST. b The specific binding sites of SVP in two genotypes were displayed, and the p-value indicated the comparison of their binding ability. Motif logo derived from JASPAR TF binding profile associated with SVP (MA0555.1) is provided above and below of the gray background, respectively. c Homozygous and heterozygous SNPs from each accession in SVP binding site. The red background G illustrates the G homozygous genotype at the SVP-bound SNP site, while the green background A represents the A homozygous genotype. The gray background R shows the G/A heterozygous genotype. EEM is short for "extremely early maturing."; EM is short for "early maturing."; LM is short for "late maturing." d The phenotype of Arabidopsis with the overexpression of lychee LcSVP9 genes. All plants were grown under long-day conditions with 16 h of light and 8 h of darkness at 23℃. The image depicts the phenotype after 24 days of sowing. Col (WT) represents the wild-type Arabidopsis with a Col-0 background. OE represents the transgenic lines with LcSVP9 overexpression. T₂ generation lines with LcSVP9 gene are represented by numbers 1, 4 and 7, with a scale of 1 cm. e The statistics of the number of rosette leaves during flowering for svp-31 mutants, wild-type and LcSVP9 transgenic Arabidopsis plants, with five plants used for each line. The statistical analysis for svp-31 mutants commenced around 24 days post-sowing, while wild-type Arabidopsis started approximately 27 days after sowing, and the OE-LcSVP9 lines began between 28 and 42 days after sowing. The data is presented as mean ± SD. One-way analysis of variance (ANOVA) was performed. P value: ***P < 0.001, ****P < 0.0001. f Expression levels of LcSVP9 and LcFT1 in response to ROS treatment (Lu et al. 2020a) and low temperature (Zhang et al. 2017) during floral induction according to the RNA-seq data in leaves of lychee. TPM (transcripts per million) was used to indicate gene expression levels or transcript accumulated levels. The log₂ ratio of fold change (log2FC) of the gene expression value between control and treatment was calculated according to the treatment timepoint. g EMSA was performed to compare the binding affinity of LcSVP9 in the promoter region of LcFT1^{A type} and LcFT1.^{G type} containing the SNP-1,437. h Transient dual-luciferase expression assay. The control group included an empty vector, and the data are presented as the mean ± SD derived from three biologically independent samples (black dots). One-way ANOVA p value: ns = 0.097, ***P < 0.001b

To confirm whether the G/A SNP affects the binding affinity of the SVP protein, we assessed the binding ability of these two types of motifs compared to lychee SVP. In total, there were ten SVP homologs in the lychee genome. Among them, LcSVP9 belongs to the SVP1 clade, which is associated with delayed flowering (Fig. S6, Liu et al. 2018) and exhibits preferential expression in lychee leaves (Fig. S5) similar to the Arabidopsis SVP (Hartmann et al. 2000). Thus, we presume that LcSVP9 may function similarly to AtSVP in delaying the flowering time of plants. Also, LcSVP9 exhibited the highest expression level in lychee leaves among these ten SVP genes (Fig. S5a). We hypothesize that LcSVP9 could be a key regulator within the SVP gene family. To confirm the function of LcSVP9 in regulating flowering, we introduced the LcSVP9 gene into svp-31 mutant Arabidopsis under the control of the CaMV 35S promoter. Under long-day conditions, plants overexpressing LcSVP9 initiated flowering with 11.8 to 28.6 rosette leaves, showing a significantly delayed flowering time compared with the svp-31 mutant control (7.4 rosette leaves) and a timing similar to the WT control (14.3 rosette leaves) (Fig. 5d, e, Fig. S5c). This suggests that the expression of LcSVP9 impacted flowering time, functioning as a floral repressor like AtSVP. We attempted to determine the expression levels of known flowering time genes in LcSVP9 overexpression plants. We found that the expression level of AtFT were significantly reduced, indicating that LcSVP9 regulates flowering via repressing AtFT (Fig. S5d). This finding is consistent with the observation that levels of AtFT expression were elevated in svp-31 mutants (Fig. S5d).

Furthermore, we observed a reduction in LcSVP9 levels in leaves subjected to reactive oxygen species (ROS) and low-temperature treatments, while LcFT1 exhibited contrasting expression patterns (Fig. 5f). Thus, we propose that LcSVP9 may repress the expression of LcFT1. Initially, we performed an electrophoretic mobility shift assay (EMSA) to investigate the potential impact of the − 1,437 SNP on the binding affinity between LcSVP9 and the LcFT1 promoter. The results showed that the LcSVP9 recombinant protein bound to the A-type motif but did not bind to the G-type motif (Fig. 5g), confirming that the natural variation of nucleotide position − 1,437 in the LcFT1 promoter are correlated with differential binding of LcSVP9. Subsequently, we conducted a dual-luciferase reporter assay to investigate the transcriptional activity of LcSVP9 on LcFT1^A−type and LcFT1^G−type promoters. The results showed LcSVP9 significantly reduced the expression of the LUC reporter by interacting with LcFT1^A−type promoter (Fig. 5h). In contrast, LcSVP9 lost its transcriptional repression on LUC driven by the LcFT1^G−type promoter (Fig. 5h). Therefore, the presence of the SVP-binding motif of LcFT1^A−type promoters is crucial for the inhibition of LcFT1 expression by LcSVP9. We then analyzed data from ten lychee leaf samples collected in the first half of December 2014, sourced from public transcriptome data (Lu et al. 2022). The results indicated that the expression of LcFT1 was significantly higher in the early-maturing variety (EM) compared to the late-maturing variety (LM) during the flowering induction period in most lychee (Fig. 5Sb). These findings are consistent with our initial expectations. The collective data suggest that the natural variation at nucleotide position − 1,437 of the lychee FT1 promoter influences the binding affinity of LcSVP9, a negative regulator, thereby leading to differential expression of LcFT1, which is likely implicated in regulating lychee flowering time.

Discussion

The number of PEBP genes varied greatly across the Sapindaceae species, ranging from nine to twelve copies (Fig. 1b). Most eudicot species typically have fewer than ten PEBP genes, with a few exceptions, including turnips (Brassica rapa), soybean (Glycine max), tomato (Solanum lycopersicum), and lotus (Nelumbo nucifera), where the number exceeds ten due to additional whole genome duplication (WGD) events (Liu et al. 2016. No further recent WGD events have been observed in Sapindaceae, with the exception of the two WGD events following the γ event that are shared among all dicot plants (Zheng et al. 2022; Xue et al. 2022; Hu et al. 2022; Wang et al. 2022). Therefore, local duplications are more noticeable in the expansion of the PEBP-family genes among the Sapindaceae genomes. Phylogenetic analysis indicated that FT1-like, FT2-like, and FT3-like were present in lychee, longan, rambutan, and soapberry, suggesting that duplication occurred before the species diverged (Fig. 1a). Despite that the genes located upstream and downstream of FT2-like were highly collinear across all six Sapindaceae species, a homologous FT2 gene was not identified in yellowhorn. This observation indicates that the FT2-like clade may be traced back to a common ancestor in the subfamily Sapindoideae, which encompasses the other five Sapindaceae species, excluding yellowhorn, which is in the subfamily Dodonaeoideae. No FT2-like flowering inhibitory factor has been identified in yellowhorn (Fig. 2a), which may be linked to the absence of juvenile phase characteristics in this species (Yao et al. 2013). The MFT-like gene was amplified in balloon vine compared to the other five Sapindaceae species (Fig. 1b). Gene collinearity analysis showed that balloon vine MFT1-like was arranged in tandem with six copies (Fig. S2a). The expression analysis of the six tandemly duplicated CaMF1 genes demonstrated a predominant expression pattern in the seeds (Fig. S2b). Previous studies have reported that the CaMF1 homolog MFT is involved in promoting seed germination (Danilevskaya et al. 2008). Thus, we hypothesized that the function of CaMF1 in balloon vine may also be associated with seed germination, specifically the ability of balloon vine seeds to maintain their vitality and germination capacity, even after prolonged storage, in contrast to the limited preservation capacity of lychee and longan seeds (Zhu et al. 2019; Johnston et al. 1979). This distinct characteristic is likely attributed to the amplification of CaMFT1.

A comparative analysis revealed that the 321-bp insertion in the longan promoter region, which was absent in the closely related species lychee and rambutan, contained an SRF-type CArG-box domain which would be bound by SOC1 (Fig. 3d). Furthermore, previous research has suggested that SOC1 functions as a transcriptional activator of FT in soybean leaves by directly binding to the CArG motif within the FT promoter (Kou et al. 2022). We speculate that SOC1 regulates DlFT1 expression in longan leaves by binding to the CArG box on the DlFT1 promoter within the 321-bp insertion specific to longan, thereby controlling flowering time. Longan exhibits unique floral induction in response to KClO₃ application, a capability unparalleled in other Sapindaceae species (Matsumoto et al. 2007). Therefore, we speculate that SOC1 may respond to KClO₃ treatment and bind to a species-specific motif in the DlFT1 promoter, accelerating the expression of DlFT1 to promote the transition to reproductive development and flowering. Further analysis should be performed to validate this possibility.

The NlFT1 gene undergoes tandem duplication in rambutan, resulting in four copies (Fig. 1c). Their regulatory sequences and expression patterns in different biological processes were in two distinct branches and one NlFT1-a clade, and the other three genes (NlFT1-b, NlFT1-c, and NlFT1-d) clustered in a single clade (Fig. 4a, b, Fig. S4c). This result suggests that the expression of the NlFT1-b, NlFT1-c, and NlFT1-d genes may be regulated by common CREs. We found that NlFT1-b, NlFT1-c, and NlFT1-d contained three types of identical CREs that are distinct from those found in the NlFT1-a gene. These CREs were associated with stress responses, including the WUN motif for wound responsiveness, the MYB-target motifs, and the anaerobic response (Fig. 4c, d). A previous study showed that the WUN-motif elements are bound by NAC transcription factors (Huang et al. 2017). In response to drought, the NAC-type transcription factor VASCULAR PLANT ONE-ZINC FINGER 1 (SlVOZ1) directly binds to the promoter of the major flowering-integrator gene SINGLE FLOWER TRUSS (SFT), an FT orthologue, thereby promoting the transition to flowering in tomato (Chong et al. 2022). Thus, we speculate that the three rambutan NlFT1 genes (NlFT1-b, NlFT1-c, and NlFT1-d), resulting from gene replication, regulate flowering through stress response pathways.

A comparative analysis of the regulatory regions was conducted to investigate inter-specific variations in different lychee germplasms. Variations in the CREs were identified within the regulatory regions of the FT1-like gene in lychee. Our experiments showed that natural variation occurring at position -1,437 within the LcFT1 promoter correlates with differential binding of LcSVP9 transcription factors, potentially influencing lychee flowering time. This variation in the CREs offers a potential molecular marker for lychee breeding. Modifications in CREs are linked to alterations in diverse agronomic traits (Rodríguez-Leal et al. 2017; Hendelman et al. 2021). For example, a recent study utilized cis-regulatory editing to alter the transcription of WOX9 in tomatoes, which subsequently impacted floral development and mitigated undesired outcomes (Hendelman et al. 2021). The recent application of CRISPR/Cas9-mediated genome editing technology to improve lychee varieties (Wang et al. 2023a) offers promising possibilities for biotechnological engineering of Sapindaceae species. For example, by modifying the regulatory cis-element that controls FT-like gene expression, the flowering period and the period of fresh fruit supply in commercially important fruit trees in Sapindaceae can be extended.

Materials and methods

Plant materials

This study utilized eight-year-old Dimocarpus longan cv. Shixia trees. The experimental trees were cultivated at South China Agricultural University in Guangzhou, China (lat. 23.1568° N, lon. 113.3537° E). The terminal shoots of three trees were sampled to obtain adult leaves, with three biological replicates collected for each sample.

Identification of PEBP family members from six Sapindaceae species

The conserved domain (PF01161) of the PEBP was acquired fromtp://pfam.xfam.org/ (Mistry et al. 2021; Wang et al. 2023b). TBtools (Chen et al. 2020) was utilized to perform HMMER analysis, allowing for the retrieval of protein data and the identification of potential members of the PEBP family. The identified genes were screened and validated using two online tools: Ptps://www.ebi.ac.uk/Tools/pfa/pfamscan/) (Madeira et al. 2019) and NCBI Conserved tps://www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi) (Lu et al. 2020b)_. Genes without complete PEBP domains were eliminated for further analysis.

Phylogenetic analysis and multiple sequence alignment

The PEBP protein sequences in other 46 seed plants species characterized by transgenic approaches, were retrieved from the Phytozome, GenBank and Arabidopsis Information Resource (TAIR) database (Table S2). A Neighbor Joining phylogenetic analysis was conducted on a total of 156 PEBP homologs across 52 flowering plants in MEGA X (V.10.2.6) (Kumar et al. 2018). The clades were assigned based on the observed clustering pattern among genes in Arabidopsis. The Taxonomy tool (Schoch et al. 2020) avaitps://www.ncbi.nlm.nih.gov/ was utilized to conduct a species phylogenetic tree. Pairwise alignments of FT1-like promoter sequences from the six Sapindaceae species were created using mVISTA Shuffl://genome.lbl.gov/vista) (Brudno et al. 2003). The conserved regions were aligned using ClustalW2 (V.2.1) (Larkin et al. 2007).

Synteny analysis among six Sapindaceae plants

Syntenic gene pairs were identified among six plants from the Sapindaceae fami-ly using JCVI://github.com/tanghaibao/jcvi/wiki/MCscan-(Python-version))(Tang et al. 2008). The identification of syntenic blocks for each pair of species was performed usi-ng the 'jcvi.compara.catalog ortholog –cscore = 0.7' parameter.

Analysis of CREs in the lychee, longan, and rambutan

The promoter sequence of the FT1-like gene was obtained by extracting the 2500-bp sequence upstream of the translation initiation site. To identify potential cis-regul-atory elements, promoter sequences were predicted using Plformatics.ps-b.ugent.be/webtools/plantcare/html/) (Lescot et al. 2002) aps://meme-suite.org/meme/tool-s/mast) (Timothy et al. 1998), by which sequences with an E-value less than 10 were included in the output. The motif profile of SVP protein w-as derived from://jaspar.genereg.net//) (Wasserman et al. 2004).

RNA-Seq data analysis

All raw pair-ends reads were trimmed utilizing trimmomatic (V.0.36) (Bolger et al. 2014) to remove the adapters and low-quality bases. Subsequently, cleaned reads were aligned to the reference genomes using the STAR alignment tool (V.2.7.10b) (Dobin et al. 2013). The read counts and Transcripts Per Million Reads (TPM) for the genes were computed using the StringTie (V.2.2.1) (Kovaka et al. 2019)_.

Folding prediction for the DlFT1 and DlFT2 proteins

The DeepView/Swiss-pdb viewer (V.4.0.1) (Schwede et al. 2003) was used to predict the structure of DlFT1 and DlFT2 proteins, based on the Arabidopsis FT structure (PDB ID: 1WKP). The three-dimensional protein structures of DIFT1 and DIFT2 were obtained by utilizing the SWISwissmodel.expasy.org) (Waterhouse et al. 2018).

Constructs for transgenic plants

The coding sequences of DlFT1, DlFT2 and LcSVP9 were cloned and inserted into the pEarleyGate 201 vector, which harbored the CaMV 35S promoter. The recombinant plasmids pEarleyGate 201-DlFT1, pEarleyGate 201-DlFT2 and pEarleyGate 201-LcSVP9 were introduced into Agrobacterium tumefaciens strain GV3101. Plants were transformed through the floral dip method and subsequently screened for resistance to BASTA. The T-DNA insertion line svp-31 (SALK_026551) was acquired from the Arabidopsis Biological Resource Center (ABRC) (Alonso et al. 2003).

Quantitative gene expression analysis

TRIzol, a product from Thermo Fisher Scientific, was utilized for the isolation of total RNA from longan leaves. TransScript® One-Step gDNA Removal and cDNA Synthesis SuperMix (TransGen, AT311-02) was utilized to synthesize cDNA from RNA samples with an amount of 0.5 μg. The qRT-PCR primers were synthesized by Sangon Co. Ltd. (Shanghai, China), using Primer 5.0 (Singh et al. 1998) for their design. The 2 × GoTaq® qPCR Master Mix from Promega was utilized for conducting the qRT-PCR experiment. The − delt CT and 2ˆ (− delta delta CT) calculation was employed to normalize expression levels. The primer sequences utilized can be found in Table S8.

Dual-luciferase reporter assay

The coding sequence of LcSVP9 was cloned and inserted into the pGreenII-62-SK vector, serving as the effector. The pGreen II-0800-Luc vector was utilized to inserted the promoter fragments of LcFT1^A−type and LcFT1^G−type (Fig. 5e), which were subsequently employed as reporters. The empty vector served as the negative control. The effector and reporter plasmids were separately transformed into Agrobacterium strains GV3101 and subsequently co-transfected into N. benthamiana leaves. The OD600 of the bacterial cultures was adjusted to 1.0, with an effector to reporter ratio of 9:1. Dual luciferase reporter gene assay kit (YEASEN) was employed to measure the relative levels of luciferase activity, in accordance with the instructions provided by the manufacturer.

EMSA

The coding sequence of LcSVP9 was cloned and inserted into the pGEX4T-1 vector. The recombinant protein GST-LcSVP9 was purified in Escherichia coli with Glutathione and Nibeads. The oligonucleotide probes were biotin-labeled at the 3’ end. The competition analysis reactions were conducted with the unlabeled probes. EMSA experiments were conducted utilizing the LightShift Chemiluminescent EMSA Kit (Thermo Fisher, 20148X).