Introduction

The reproductive transition in perennial woody plants does not occur until several years of repeated seasonal changes and alternative growth. Although many genes related to flowering regulation have been identified and a fine gene network has been described in Arabidopsis, the model herbaceous plant (Blázquez 2000, 2005), how this gene network works in woody plants remains a challenging subject in the field of plant molecular development. Recently, differences in rates of molecular evolution have been noted between woody perennials and annual herbaceous species (Dornelas et al. 2007; Wilkie et al. 2008; Yang et al. 2008). These differences have been presumed to reflect differences in generation times (Martín-Trillo and Martínez-Zapater 2002; Meilan 1997).

Extensive research on Arabidopsis has revealed that flowering can be triggered by light, temperature, gibberellins, and autonomous regulation pathways that initiate expression in several integrator genes that respond to both internal and environmental cues. These integrator genes include FLOWERING LOCUS T (FT), SUPPRESSOR OF OVEREXPRESSION OF CONSTANS1 (SOC1), and LEAFY (Blázquez 2000; Fujiwara et al. 2005). The main theory is that the vegetative growth in Arabidopsis is maintained by repressing the function of flowering genes (Koornneef et al. 1998). SHORT VEGETATIVE PHASE (SVP), a MADS-box gene, is an important flowering repressor in Arabidopsis (Hartmann et al. 2000). Recent molecular cloning and functional analyses of SVP indicated that SVP may be another central regulator of the flowering regulatory network because the gene is controlled by the autonomous, thermosensory, and gibberellin pathways and directly represses SOC1 transcription in the shoot apex and leaf (Li et al. 2008). Moreover, a moveable florigen FT, expressed in the leaf, is also modulated by SVP (Lee et al. 2007b). The SVP protein associates with the promoter regions of SOC1 and FT, where the potent repressor FLOWERING LOCUS C (FLC) binds, and SVP interacts with FLC in vivo during vegetative growth (Li et al. 2008). Some SVP/StMADS-11-like genes from Chinese cabbage (Lee et al. 2007a) and barley (Trevaskis et al. 2007) also act as flowering repressors. Another SVP homolog, PkMADS1, from the woody plant Paulownia kawakamii promoted vegetative growth (Prakash and Kumar 2002), whereas the SVP/StMADS-11-like gene from Eucalyptus (EgSVP) seems to have a slightly different function. The ectopic expression of EgSVP in Arabidopsis caused a slight delay in flowering time and produced additional inflorescences (Brill and Watson 2004). Thus, in woody plants SVP might have evolved functional diversification not seen in annual herbaceous plants.

Precocious trifoliate orange was discovered as a spontaneous mutant from wild-type trifoliate orange (Poncirus trifoliata L. Raf.; Liang et al. 1999). A major characteristic of precocious trifoliate orange is that its juvenile phase is shortened to 1 to 2 years, whereas the wild-type plant has a juvenile period of 6 to 8 years. This precocious mutant is valuable for investigating the flowering mechanism in woody plants because of its short juvenile phase and special flowering behaviors.

In a previous study, we established a two-dimensional suppressive subtractive hybridization library to identify differential gene expression in shoot meristems between the precocious mutant and wild-type trifoliate orange. A total of 274 expressed sequence tags were discovered, including FT, TFL1, BAM, and other flowering-related genes (Zhang et al. 2009). In this study, PtSVP, a differentially expressed gene, was characterized. The temporal and spatial expression patterns of PtSVP were investigated using in situ hybridization and real-time PCR. Its function was further studied using ectopic transgenic analysis.

Materials and methods

Plant materials and sample collections

Plants of precocious and wild-type trifoliate orange were grown under natural environmental conditions in experimental fields of the National Citrus Breeding Center of Huazhong Agricultural University, Wuhan, China (30°35′ N, 114°17′ E). Plants in the first year after sowing were considered to be juvenile, and adult trees of the mutants were 3–5 years old and had flowered several times. All plant tissues were sampled according to the demands of each experiment, immediately frozen in liquid nitrogen, and stored at −80°C until used.

For morphological observation and floral development, 600 buds that displayed a similar growing condition were selected and tagged when they were sprouting. Ten shoots derived from these buds were sampled every 3 days in the first month and every 7 days thereafter. The shoots were separated into segments with one node, fixed, stored in FAA, and grouped according to their bud location. Paraffin section analysis followed the method of Yao et al. (2007).

To analyze the fluctuation of PtSVP expression over the annual cycle and its relationship to flower development, total RNA for RT-RCR and real-time PCR was isolated from bud samples from adult trees at the following developmental stages: dormant buds, sprouting buds, buds on the shoots in quick growth, and mature buds. Tissue samples from three randomly selected trees were collected at approximately 2-month intervals throughout the 2006 and 2007 growth cycles.

The expression patterns and relationship of PtSVP to CiFT and CsWUS were investigated during shoot development. The spring and summer shoots at three distinct phases (before, during, and after self-pruning) were collected from adult trees of both precocious mutant and wild-type trifoliate orange. Considering that CiFT might be regulated by light, all shoots were sampled at 10:00 am to minimize the light effect. Shoot samples were collected from three groups of trees (each group containing three trees) used as biological repeats.

RNA extraction, first-strand cDNA synthesis, and PtSVP isolation

Total RNA was extracted according to the protocol of Zhang et al. (2008). Total RNA (3 mg) of each sample was treated with 3 U DNase (Promega, USA) and then further purified to remove the DNase. The total RNA yield and quality were determined spectrophotometrically at wavelengths of 230, 260, and 280 nm. Total RNA was adjusted to a final concentration of 1 μg/μl, and the first-strand cDNA synthesis was performed according to the manufacturer’s instructions.

Full-length PtSVP was obtained by rapid amplification of cDNA ends, and the PtSVP DNA sequence was identified by a BD Universal Genome Walker kit (Clontech, USA) according to the manufacturer’s instructions. The putative protein sequence was predicted using Blastx and MIT (http://genes.mit.edu/GENSCAN.html). The phylogenetic tree was constructed by the PHYLIP package using maximum likelihood.

Construction of PtSVP-GFP fusion protein and PtSVP localization

The open reading frame (ORF) without the terminator codon of PtSVP was constructed to the pCAMBIA1302 fusing to green fluorescent protein (GFP) using the restriction enzyme NcoI. The PtSVP-GFP expression vector was transformed into onion epidermis cells by particle bombardment, as described previously (Kinkema et al. 2000). After bombardment, the Petri dishes were sealed with Parafilm and placed in a 26°C incubator for about 24 h. GFP fluorescence was then detected using an 80i Nikon fluorescent microscope under a stimulating wavelength of 490 nm.

Analyzing PtSVP expression using real-time PCR

PtSVP expression level was measured by real-time PCR using SYBR green I chemistry (Qiagen, Germany). Primers for PtSVP were designed with the Primer Express software (PE Applied Biosystems, Foster City, CA, USA) and tested to ensure amplification of single discrete band with no primer-dimers. Product size was 137 bp. An amount of cDNA corresponding to 25 ng of input cDNA was used in each reaction. The real-time PCR was performed with 1 μl template of the real-time reaction mixture, 10 μl 2× SYBR Green Master Mix, 0.5 μl forward and reverse primers (10 μmol/μl), and water to a final volume of 20 μl. Gene expression levels were analyzed with ABI 7500 Sequence Detection System Software (PE Applied Biosystems) and normalized with the results of β-actin. Three biological repeats and four mechanical repeats were used for each reaction.

RNA in situ hybridization and detection

Materials for in situ hybridization were sampled and immediately fixed in RNAase-free FAA (4% formaldehyde, 10% acetic acid, 50% ethanol). Paraffin sections were made according to the method of Yao et al. (2007). A special 246-bp sequence in the middle of the PtSVP cDNA sequence was cloned into pGEM-T vector for probe synthesis (Supporting Information Table S1). Digoxigenin-labeled RNA probes were prepared using a DIG Northern Starter Kit (Roche, Germany). T7 and SP6 RNA polymerase were used to generate the sense and antisense RNA probes by in vitro transcription according to the manufacturer’s instructions. The in situ hybridization experiment was performed as described in the Cold Spring Harbor Arabidopsis Molecular Genetics Course (www.Arabidopsis.org/cshl-course/5-in_situ.html).

Arabidopsis transformation

The svp-41 mutant and wild-type (Col-0) Arabidopsis plants were used for transformation to confirm the function of PtSVP. Seeds of svp-41 were provided by P. Huijser (Max-Planck-Institut für Züchtungsforschung, Molekulare Pflanzengenetik, Cologne, Germany). The floral dipping transformation method (Clough and Bent 1998) was used in this experiment. Seeds carrying PtSVP fused to the 35S promoter were selected on medium containing 75 mg/l kanamycin and grown under long-day conditions (16 h light/8 h dark) at 25°C. The transgenic plants T1 and T2 were also confirmed by PCR amplification. To investigate flowering time, days to flowering and the number of rosette leaves were counted when plants bore a 1-cm-long inflorescence.

Overexpression of PtSVP in tobacco and phenotype analysis of the transformants

PtSVP cDNA containing full-length ORF was introduced into the binary plant transformation vector pBI121, in which the transcription of PtSVP was driven by the CaMV 35S promoter. PtSVP transgenic tobacco plants were produced according to the method of Salehi et al. (2005). A total of 219 transgenic tobacco plants were screened from the kanamycin-selected culture medium. These 219 plants together with 54 nontransgenic controls were planted in an isolated greenhouse under the same conditions with uniform management. In the 219 screened plants, 37 (T0) plants were confirmed to be true transformants by PCR detection using DNA templates with PtSVP RT–PCR primers (Supporting Information Table S1). Fifteen nontransgenic plants and 19 transformants were selected randomly for analysis of flowering time, flowering duration, number of flowers on an inflorescence, and average flower number per coflorescence. Flowering time was defined as days from transplanting in the greenhouse to blooming of the top flower. Flowering duration was the period from blooming of the top flower to formation of the last floret. Number of flowers on an inflorescence refers to the average number of flowers per inflorescence in the investigated group. Number of flowers per coflorescence was calculated as the mean of the averaged flower number on each coflorescence among the tested inflorescences. The data were analyzed using Excel (Microsoft, Redmond, WA, USA).

Yeast two-hybrid assays and library screens

Lateral buds, terminal buds, and shoot meristem at different developmental stages were sampled from precocious trifoliate orange. To prepare a representative sample of total RNA from the adult and juvenile tissues for cDNA library construction, different developmental stages of plant organs from roughly equal numbers of adult and juvenile plants were pooled. A cDNA library was constructed using the BD Matchmaker library construction and screening kits (Clontech) according to the manufacturer’s instructions. Full-length PtSVP cDNA was cloned into pGBKT7 (Clontech). Yeast cells were transformed by the LiAc/DNA/PEG method according to the Yeast Protocols Handbook from Clontech (http://www.clontech.com). Constructs were tested for autoactivation as described by the manufacturer. Colonies growing on the SD/-Leu-Trp-His-Ade media with 10 mM of the competitive inhibitor 3-amino-triazole were transferred to selective medium containing X-Gal (80 mg/ml), and the blue colonies were characterized. Screening of interaction clones was carried out according to the manufacturer’s instructions. Plasmids were isolated from colonies showing a positive (blue-colored) reaction and introduced into AH109 for confirmation of protein interaction. Inserts of the plasmids were sequenced by the Beijing Genomics Institute (Wuhan, China).

Results

Annual growth cycle and floral development course in trifoliate orange

Most citrus species have two or three periods of bud sprouting in a year to form different shoots (Davenport 1990). In wild-type trifoliate orange, only the shoots that sprout in late winter or early spring (March to April) are capable of forming a floral bud, whereas the summer (May to July) and autumn shoots (August to October) are vegetative and unable to form a floral bud (Fig. 1). In contrast, both the spring and summer shoots of the precocious mutant can generate flowers (Fig. S1). As a result, the precocious mutant has a lower ratio of leaf bud/flower bud than the wild type.

Fig. 1
figure 1

The schematic developmental cycle of wild-type trifoliate orange. Spring shoot sprouted in early March and began vegetative growth which would last for about one month. In mid April, vegetative growth of the spring shoots was terminated by self-pruning and the lateral buds on spring shoots had to select their differential determination. At this time, the meristem was un-determinated (UDM). For the flowering-competent shoots (the spring shoots), the meristem at top several bud location had the advantage to begin floral differentiation (RM, reproductive meristem) and the meristem at bottom leaves’ axilla was vegetative meristem (VM, vegetative meristem) which initiated leaf bud (vegetative) differentiation. For the flowering-incompetent shoot (the summer shoot and autumn shoot; green line) which sprouted in summer or autumn from the spring shoots or old shoots formed before the current year, all the lateral buds could only initiate leaf bud differentiation after the self-pruning

Full size image

Self-pruning is a necessary but not sufficient condition for floral bud initiation. It is a typical in all kinds of shoots of citrus plants to cease vegetative growth temporarily by automatically removing the shoot tip (about 0.5–1 cm; Fig. 1). Cytological observation revealed that the floral buds in the wild type and the precocious mutant initiated their differentiation immediately after self-pruning on spring shoots. The shoot apical meristem was in an undetermined state (Fig. S2a) and floral primordia were not been observed until the late stage of self-pruning (mid-April in Wuhan, China). After self-pruning, differentiation occurred rapidly and produced the primordia of sepal, petal, stamen, and pistil sequentially (Fig. S2b, c, d, f, g, h). The whole integrated flower bud was formed in 1 month, then fell into dormancy until late February of the following year. In wild-type trifoliate orange, the summer shoots, which do not form floral buds, begin to produce vegetative buds after self-pruning. Thus, the self-pruning appears to be a demarcation point for the meristem to initiate leaf bud or floral bud development.

Cloning and annotation of the PtSVP

The PtSVP expressed sequence tag was identified from a previously constructed suppressive subtractive hybridization library; the 3’ and 5’ rapid amplification of cDNA ends was used to obtain the full-length cDNA, which was 1087 bp long with an ORF of 654 nucleotides (GenBank accession no. FJ373210). A 3834-bp sequence of PtSVP genomic DNA (GenBank accession no. FJ373211) was obtained by DNA genome walking. Structure analysis between the full-length cDNA and genomic DNA sequence revealed that PtSVP contained eight exons and seven introns. The DNA sequence between the first exon and the eighth exon was 3129 bp, the start codon ATG was in the second exon, and the stop codon was in the eighth exon (Fig. 2a). PtSVP cDNA encoded a MADS-box protein of 218 amino acids, containing a MEF2-like motif (MADS-box) in the N-terminus and a K-box motif in the middle.

Fig. 2
figure 2

Gene structures of PtSVP and phylogenetic relationship of the predicted amino acid sequences of PtSVP with SVP-like proteins from other plant species and other representative MADS-box proteins of Arabidopsis. a Structures of the genomic sequence of PtSVP. The closed boxes denote exons, and the lines between the closed boxes denote introns. b Maximum likelihood Bayesian phylogeny of PtSVP-related MADS-box protein in Arabidopsis and the SVP from other plant species. The phylogram tree shows that the PtSVP is a homolog to AtSVP. The scale bar indicates a divergence of 0.1 amino acid substitutions per site. The unrooted tree was generated using Phylip package sofyware by neighbor-joining method. Bootstrap values from 1,000 replicates are indicated at each node. The accession numbers of MADS proteins used in this analysis are: EgSVP (AAP40641), AtSVP (NP_179840), BrSVP (AAQ55452), MEF2-LIKE (ABD66219), PkMADS1 (AAF22455), VvUnnamed (CAO48343), OsMADS22 (NP_001048193), BM1 (CAB97350), BM10 (ABM21529), INCO (CAG27846), LpMADS10 (AAZ17549), IbMADS3 (AAK27150), PnMADS1 (BAF46766), PsSVP (AAX47170), AtAGL24 (NP_194185), AtFLC (NP_001078563), AtAGL6 (NP_182089), AtAGL19 (NP_194026), AtMAF1 (NP_850979), AtSOC1 (NP_182090), AtAGL15 (NP_196883), AtAGL18 (NP_191298), AtAGL42 (NP_568952), AtAG (NP_567569), AtAGL12 (NP_565022), AtSEP1 (NP_001119230), AtAGL17 (NP_179848), AtCAL (NP_564243), AtAP1 (NP_177074), AtAGL21 (NP_195507), AtAGL16 (NP_191282), AtSEP3 (NP_564214), and AtPI (NP_197524)

Full size image

Phylogenetic analysis using amino acid sequences showed that PtSVP belongs to the SVP group, which includes the SVP homologs (StMADS11-like) from Arabidopsis and other plant species. This group was distinct from other MADS-box proteins in Arabidopsis, such as AtAGL24 and AtSOC1 (Fig. 2b). To better understand their evolutionary relationships, the SVP/StMADS-11-like genes in other plant species, including BrSVP in Chinese cabbage (Brassica; Lee et al. 2007a), EgSVP in Eucalyptus (Brill and Watson 2004), INCO in Antirrhinum (Masiero et al. 2004), and SVP/StMADS-11-like genes in rice (Sentoku et al. 2005), were included in the phylogenetic analysis. They all were clustered in the SVP group. PtSVP was closest to MEF2-LIKE, a functionally unidentified MADS-box gene from apple. In Arabidopsis, AtAGL24 is a MADS-box gene most similar to SVP (Gregis et al. 2006), but compared with PtSVP, it was even farther from SVP according to phylogenetic analysis. Thus, PtSVP might be an SVP/StMADS-11-like transcription factor.

The localization of PtSVP protein

To examine whether PtSVP was localized in the nucleus like other transcription factors, the PtSVP-GFP fusion protein was constructed to localize the PtSVP protein. The PtSVP ORF without the stop codon was fused to the 5’ end of the GFP ORF in pCAMBIA1302. The pCAMBIA1302 with CaMV 35S promoter driving GFP alone was used as a negative control. The PtSVP-GFP signal was localized in the nucleus (Fig. 3d), whereas the fluorescence of the control was localized in both the nucleus and cytoplasm (Fig. 3c). The nuclear localization of PtSVP-GFP also suggests that PtSVP might be a transcription factor.

Fig. 3
figure 3

Subcellular localization of PtSVP in cells. a, c Transient transformation with 35S::GFP. c GFP alone was widespread in the whole cell. b, d Transient transformation with 35S::PtSVP-GFP fusion construction. d PtSVP-GFP fusion protein was located in cell nucleus. Bar = 50 μm

Full size image

The annual fluctuation of PtSVP expression

To understand the relationship between floral bud initiation and the expression of PtSVP in trifoliate orange, the expression levels of PtSVP were measured in different stages of shoot, including newly formed young bud (July), mature bud (September), dormant bud (November to January), and sprouting bud (March to April) by real-time PCR (Fig. 4a).

Fig. 4
figure 4

The expression of PtSVP was investigated by real-time PCR and in situ hybridization. a Relative quantities of PtSVP in lateral vegetative buds on adult wild-type trifoliate orange. Data points represent the mean ± SE of at least four replications for the relative expression, which was normalized by the amount of the Actin control. The primers used for the analyses are shown in Supplement Table S1. b, c, d are corresponding to Nov, Mar, and Apr in panel, respectively. b The high expression of PtSVP in vegetative bud sampled in November. c The weak expression of PtSVP in the vegetative bud which was sprouting in March; d the weaker expression of PtSVP in the came-up spring shoot (in April, before self-pruning) which is flowering-competent. Contrary, PtSVP expressed strongly in the flowering-incompetent shoots, spring shoot of juvenile tree (e) and the summer shoot of adult tree (f) of the wild-type trifoliate orange. g The sense probe detection in the summer shoot tip is as a control. Bar = 50μm for panels b, d–g; 300 μm for panel c

Full size image

The expression level of PtSVP fluctuated with season, shoot growth, and self-pruning, as revealed by quantitative PCR (Fig. 4a). PtSVP was expressed abundantly in the dormant vegetative bud (November). When the vegetative bud was sprouting in March, PtSVP expression level was down-regulated, and the expression level was even lower when the vegetative bud grew into a shoot in April (near the self-pruning stage). Similar results were obtained by in situ hybridization: PtSVP was expressed strongly in vegetative buds in November (Fig. 4b) and was down-regulated in sprouting shoots (Fig. 4c). In addition, the signal intensity of in situ hybridization indicated that PtSVP was expressed at high levels in spring shoots of juvenile wild-type (Fig. 4e) and summer shoots of adult wild-type trees (Fig. 4f), which are unable to form flowers. Based on these results, we conclude that a high expression level of PtSVP might help to maintain the tissue or meristem in a dormant state, whereas the down-regulation of PtSVP expression after winter might be closely related to the shift of cell activity or the change in flowering competence of buds on adult wild-type trifoliate orange.

Expression pattern of PtSVP during self-pruning

Self-pruning may be a key stage for meristem determination. To understand the role of PtSVP in the floral transition, the dynamic expression of PtSVP during self-pruning was monitored by real-time PCR. As shown in Fig. 5, PtSVP expression was markedly down-regulated during the floral transition (self-pruning) and then increased to higher levels after the lateral buds were differentiated (20 days after self-pruning) in the flowering-competent shoots (spring shoots of both trifoliate oranges and summer shoots of the precocious mutant). However, in the flowering-incompetent shoot (summer shoots of wild type), there was an increased trend of PtSVP expression during self-pruning compared with the prior stage. The dynamic expression of PtSVP in flowering-competent and -incompetent shoots suggests that PtSVP is involved in the floral transition.

Fig. 5
figure 5

The expression pattern of PtSVP, CiFT and CsWUS during self pruning. The relative quantities of PtSVP, CiFT, and CsWUS during floral transition in spring and summer shoots of both adult precocious and wild-type trifoliate orange were investigated by real-time PCR. ACTIN was used as a housekeeping control. Real-time PCR experiments were conducted using the primers displayed in Supplement Table S1. Lane 1, 15 days before self-pruning; lane 2, during self-pruning; lane 3, 20 days after self-pruning

Full size image

Parallel analysis of the expression of two functionally confirmed genes, CiFT and CsWUS (Endo et al. 2005; Tan and Swain 2007), was carried out by real-time PCR to further investigate the role of PtSVP in floral bud initiation (Fig. 5). The expression of CsWUS was high before and after self-pruning, but it was down-regulated during the floral transition (self-pruning) in flowering-competent shoots. The expression of CiFT was low before floral transition and then up-regulated during the floral transition. Taken together, the expression patterns of these three genes suggest that the down-regulation of PtSVP is related to the floral transition. In other words, the high expression of PtSVP may maintain either vegetative or reproductive development but not directly cause development transition. It is the decrease of the expression level of PtSVP that leads to the development transition.

To precisely localize PtSVP expression in lateral buds during seasonal period of floral transition and flower differentiation, in situ hybridization was performed on sections of wild-type floral buds containing florets at various stages of development. Consistent with the results of real-time PCR, the signals of PtSVP transcripts were clearly detected in apical meristems before the floral transition (before self-pruning; Fig. 6a). But during the floral transition (at the late stage of self-pruning), the expression of PtSVP in apical meristems was down-regulated (Fig. 6b). In the course of flower differentiation, PtSVP expression was higher in differentiated cells (flower primordia) than in undifferentiated cells (apical meristems; Fig. 6c). When the flower bud fully developed and the flower apical meristem disappeared, PtSVP showed high levels in the whole flower bud except the bract (Fig. 6d). Therefore, the expression patterns of PtSVP in the lateral buds detected by in situ hybridization were consistent with the PtSVP expression pattern in whole shoots detected by real-time PCR. These results indicate that PtSVP is involved in maintaining cells in a state of ongoing activity, and its expression change is associated with floral bud initiation in trifoliate orange.

Fig. 6
figure 6

The expressions of PtSVP in lateral buds at different developmental stages of trifoliate orange were shown by in situ hybridization; a is the lateral bud before self-pruning period; b floral transition phase; c flower bud differentiation phase; d whole mature flower bud; e hybridized with a sense PtSVP probe. Bar = 50 μm. AM apical meristem, Br bract, FAM flower apical meristem, Pi pistil, SP stamen primordium

Full size image

Functional analysis of PtSVP in transgenic Arabidopsis

To evaluate the function of PtSVP in the regulation of flowering, PtSVP was ectopically expressed in svp-41 mutant and wild-type Arabidopsis driven by the 35S promoter. Arabidopsis svp-41 mutant plants exhibit a strong early-flowering phenotype under long days. If PtSVP and SVP are functionally conserved, PtSVP should rescue the SVP defect in the mutants of Arabidopsis and result in normal flowering. Fourteen independent kanamycin-resistant plants were obtained in the T1 generation. For expression analysis of PtSVP function, we randomly selected 13 T2 transgenic plants by PCR detection using DNA templates. Under long-day conditions, the flowering time of most of the transgenic lines was delayed, suggesting that PtSVP may function as a floral repressor and could partially recover the svp mutation (Fig. 7a; Table 1). Furthermore, ectopic expression of PtSVP in wild-type Arabidopsis also delayed the flowering time (Fig. 7b). These transgenic plants were grouped into three classes based on the total number of rosette leaves at flowering (Table 1). Class I plants flowered significantly later than the wild-type plants in terms of both days to flowering and the number of leaves, and the Class I plants also showed floral defects (Fig. 7d), including alterations in floral organ number, the appearance of excrescent trichomes on sepal, pale green sepals, and vestigial petals. These results suggest that the strong expression of PtSVP could disturb normal flower development. Although the class II and III plants showed fewer flower defects, they showed late flowering and had excrescent trichomes on sepal. More trichomes are usually considered to be a juvenile phenotype of Arabidopsis (Telfer et al. 1997). The overexpression of PtSVP delayed flowering time in transgenic lines and induced juvenile characteristics during the adult stage, demonstrating that the expression of SVP disturbs flower development, acts as a floral repressor similar to SVP, and is involved in organ determination. Concerning the length of the inflorescence and total number of flowers, however, the transgenic plants of 35S::PtSVP displayed obvious prolonged inflorescence and many additional flowers (Fig. 7e).

Fig. 7
figure 7

Phenotypes of PtSVP transgenic Arabidopsis. a The expression of PtSVP rescued the early-flowering phenotype of svp mutant Arabidopsis. b Ectopic expression of PtSVP in wild-type Arabidopsis delayed the flowering time. c The normal flower of wild-type Arabidopsis. d In transgenic 35S::PtSVP/WT, a flower-like structure instead of carpel and excrescent trichomes on flower. e Compared to wild-type Arabidopsis (1), 35S::PtSVP/WT plants (2, 3) grew higher and produced longer inflorescences. The longer inflorescence of 35S::PtSVP/WT was caused by the unabated growth of inflorescence, with continuously emerging new flowers. 35S::PtSVP/svp: PtSVP was expressed in the svp mutant driven by 35S promoter; WT: wild-type Arabidopsis; 35S::PtSVP/WT: PtSVP was expressed in the wild-type driven by the 35S promoter

Full size image
Table 1 Flowering time of transgenic Arabidopsis
Full size table

Ectopic expression of PtSVP in tobacco

To further examine the function of PtSVP, we constitutively expressed PtSVP in tobacco under the control of the 35S promoter. The vegetative growth of 37 transformants (T0; Supporting Information Fig. S3) was similar to that of the wild type, whereas the process of reproductive growth differed. Although the flowering time of transformants and nontransgenic tobacco plants were not significantly different with regard to anthesis of their top flower (Table 2), the coflorescence development of 35S::PtSVP transgenic tobacco was delayed at the early stage (Fig. 8b), which was evidenced by the fact that the transformants had fewer flowers in the early stage (Supporting Information Fig. S4). The duration of coflorescence development in the 35S::PtSVP transgenic tobacco plants was 1.4 times longer than in wild-type plants, and transgenic plants eventually generated significantly more florets than the wild type (Fig. 8d, e; Table 2). Although the number of coflorescences in transformants and nontransgenic plants showed no difference, the total numbers of flowers in each transgenic plant was greatly increased. The average flower number per coflorescence on 35S::PtSVP transgenic tobacco plants was about 17, significantly more than that on the nontransgenic plants (about 11, Table 2). These results suggest that PtSVP not only stunted the coflorescence transition during the early stage but also maintained the reproductive growth after the floral transition. In addition, in transgenic plants PtSVP overexpression resulted in more abnormal flowers with late development of organs from the second whorl (petal) and finally sterility (Supporting Information Fig. S5).

Table 2 Phenotype analysis of non-transgenic and 35S::PtSVP tabacco
Full size table
Fig. 8
figure 8

Phenotype of the 35S::PtSVP transgenic tobacco and the interaction experiment of PtSVP. a, c Inflorescences of the wild-type tobacco. b, d Inflorescences of 35S::PtSVP transgenic tobacco. The coflorescense development of transgenic plant was postponed at the early state (b) but was prolonged at the later stage and resulted in more florets (d). e The comparison of coflorescences between the wild-type (above) and transgenic plants (below). f Yeast two hybrid system was used to detect the interaction between PtSVP and PtAP1. Mark (1) shows blue yeast colony indicating that PtSVP interacted with PtAP1; Mark (2) is a positive control (blue yeast colony); Mark (3) is a negative control (no colony). Co, coflorescense; TF, top flower

Full size image

Interacting partners of PtSVP in citrus

To find candidate citrus proteins interacting with PtSVP, a yeast two-hybrid system was used to screen citrus cDNA expression library, using PtSVP as a bait. Approximately 1×106 yeast colonies were screened, and 17 positive clones were identified from the cDNA library. DNA sequencing showed that the 17 clones represented three genes (encoding an unknown protein, ubiquitin thiolesterase, and APETALA1-like protein). Of these selected clones, genes encoding APETALA1-like protein (PtAP1) were the most frequently identified proteins (8 clones). PtAP1 shared a very high similarity (99% at the amino acid and nucleotide sequence level in the ORF; Supporting Information Fig. S7) with CsAP1, which was demonstrated to be a functionally active homolog of AP1 (Cervera et al. 2009; Pillitteri et al. 2004). Thus, we conclude that PtSVP has the same mechanism of interacting with the AP1 homolog as reported by Gregis et al. (2006). However, a positive interaction between PtSVP and the other two proteins needs to be investigated further.

Discussion

Gene structure and evolution of PtSVP

Analysis of the full-length cDNA and the genomic DNA sequence of PtSVP indicated that there were eight exons in PtSVP (Fig. 2a), which differs from the nine exons in SVP and BrSVP (Lee et al. 2007a). In addition, the initiation codon of PtSVP is located in the second exon, whereas those of SVP and BrSVP are in the first exon. Several predicted cis-acting regulatory elements, including the MYB protein (Schmitz et al. 2002) binding site, exist in the region between the first exon and the initiation codon. This region might be an important transcriptional regulatory position, because a similar regulation pattern has been reported in other genes (Jeon et al. 2000; Jeong et al. 2006; Kikuchi et al. 2000). Nevertheless, the predicted PtSVP protein has similar functional domains as those of SVP, with a MADS-box at the N-terminus and a K-box motif in the middle, and was most similar to MEF2-LIKE, a putative SVP/StMADS-11-like protein from apple. Phylogenetic analysis of PtSVP and other SVP/StMADS-11-like proteins also indicated that PtSVP was evolutionarily close to EgSVP from Eucalyptus (Brill and Watson 2004; Prakash and Kumar 2002), SVP from Arabidopsis (Hartmann et al. 2000), and BrSVP from Chinese cabbage (Lee et al. 2007a). All the genes clustered in the SVP group (Fig. 2b) have similar functions of suppressing flowering or promoting vegetative growth. Our findings suggest that this MADS-box protein might be a candidate SVP gene in citrus.

The expression of PtSVP was closely related to floral bud initiation in trifoliate orange. SVP is a transcription factor in Arabidopsis and suppresses the floral transition by directly binding to the promoter of FT and SOC1 in the nucleus (Lee et al. 2007b; Li et al. 2008). PtSVP was localized in the nucleus (Fig. 3d), which is a feature of transcription factors. Similar to SVP, PtSVP was expressed intensively in vegetative meristems (Supporting Information Fig. S6). However, the expression pattern of PtSVP showed an annual fluctuation coincident with the flowering competence of shoots. The expression of PtSVP was down-regulated in the lateral vegetative bud after winter (Fig. 4a) and remained at a low level in the spring shoot (Fig. 4d). In the annual cycle of PtSVP expression, each down-regulation of the gene was followed by initiation of floral bud differentiation. In wild-type trifoliate orange, the spring shoot of the juvenile tree (Fig. 4e) and the summer shoot of the adult tree showed a strong expression level of PtSVP (Fig. 4f). These two kinds of shoots are incapable of floral bud formation. A similar expression pattern was reported in an SVP homolog from another perennial plant, raspberry (Rubus idaeus L.; Mazzitelli et al. 2007). This pattern was consistent with the yearly phase transition in woody plants, which does not occur in annual herbaceous plants.

Self-pruning is a self-regulation mechanism controlling the transition from vegetative to reproductive growth by removing the shoot apical meristem and top growth advantage. The earliest observable floral initiation occurs just after self-pruning. The results from real-time PCR (Fig. 5) and in situ hybridization (Fig. 6) consistently showed that the expression of PtSVP sharply declined during self-pruning, the pivotal stage when the shoots shift from vegetative to reproductive growth. Taken together, these results suggest that PtSVP negatively regulates the floral transition in trifoliate orange. The phenotype of 35S::PtSVP transgenic tobacco plants provided further evidence that the overexpression of PtSVP decelerated the floral transition of coflorescences (Fig. 8b). In Arabidopsis, SVP is associated with the FT promoter regions to regulate the expression of FT in the leaf (Lee et al. 2007b) and might be a flowering repressor. In our experiments, however, the expression patterns of CiFT and PtSVP did not shown a completely opposite trend. Therefore, whether PtSVP directly regulates CiFT in trifoliate orange must be further investigated. Our findings also suggest that the functions of SVP may be different between herbaceous and woody plants.

High expression of PtSVP prolongs reproductive growth

SVP is a transcription factor in Arabidopsis and suppresses the floral transition by directly binding to the promoter of FT and SOC1 in the nucleus (Lee et al. 2007b; Li et al. 2008). Like transcription factors, PtSVP was localized in the nucleus (Fig. 3d). In Arabidopsis, SVP is expressed in functionally determined meristems (vegetative shoot apical meristem, flower primordium, and coflorescence primordium), but not in undetermined meristems (Hartmann et al. 2000). Likewise, extensive expression of PtSVP in trifoliate orange was detected in vegetative meristems (Supporting Information Fig. S6). Thus, it appears that the expression patterns of SVP are relatively similar between herbaceous and woody plants. In addition, our results indicated that PtSVP was down-regulated in apical meristems (Fig. 6b, c), which were functionally undetermined tissues. In contrast, PtSVP was expressed strongly in the functionally determined tissues, such as vegetative shoot tips (Fig. 4e, f) and flower organ primordia (Fig. 6b, c). The expression pattern of PtSVP was somewhat similar to that of CsWUS (Fig. 5). CsWUS is functionally similar to WUS in Arabidopsis, which maintain the identities of functionally determined meristems (Laux 1996; Tan and Swain 2007). However, real-time PCR also showed that PtSVP was highly expressed in the functionally determined shoots before and after the floral transition (Fig. 5). Based on the expression pattern of PtSVP, we propose that PtSVP actively maintains the duration of both vegetative and reproductive growth.

SVP is considered to be a flowering repressor in Arabidopsis, because it functionally delays flowering time. In this study, the flowering time of PtSVP transgenic Arabidopsis plants was delayed and the 35S::PtSVP/WT showed a typical juvenile character (more trichomes). This result suggests that PtSVP regulates flowering time by maintaining the vegetative growth, which is functionally conserved to SVP in Arabidopsis. However, recent studies proved that AGL24 and SVP are also floral meristem identity genes. Ectopic AGL24 and SVP expression induces floral meristem indeterminacy by promoting the development of new ectopic floral meristems rather than causing floral reversions (Gregis et al. 2008). This finding suggested that expressions of PtSVP may control floral organ development. In this study, PtSVP induced a flower-like structure in place of the carpel (Fig. 7d). Moreover, ectopic overexpression of PtSVP in Arabidopsis and tobacco showed similar increases of flower numbers by prolonging the inflorescence length and flower formation time (Figs. 7e, 8d, e). This result coincides with the reports in transgenic of EgSVP to Arabidopsis (Brill and Watson 2004). An SVP homolog from Antirrhinum, INCO, functions not only as a negative regulator of phase transition, but also as a positive regulator of floral development (Masiero et al. 2004). Thus, we conclude that PtSVP might also play some role in maintaining reproductive growth.

In Arabidopsis, AP1 is an identity gene to promote and maintain floral meristems. AP1 binds to the promoter region of SVP to suppress its expression and then promote flower organ development after the floral transition, when the first-whorl organs are generated (Liu et al. 2007; Yu et al. 2004). The down-regulation of SVP at the proper time ensures normal flower development. This may explain why the constitutive expression of PtSVP (driven by the 35S promoter) disrupted the correlative gene expression order and thus caused abnormal flower development in Arabidopsis and tobacco. An alternative explanation of the floral defects is that the malapropos and excessive PtSVP protein may interact with AP1 or its homolog in the transgenic plants and disrupt flower development. During the floral transition in Arabidopsis, SVP interacts with other MADS-box proteins that determine floral organ identities during flower formation (DeFolter et al. 2005; Pelaz et al. 2001). This hypothesis is supported by findings that the SVP-AP1 heterodimer is involved in the recruitment of the corepressor complex for the regulation of AGAMOUS (AG) expression (Gregis et al. 2006). Thus, in our experiments the excess PtSVP may also function as SVP to effect the flower organ development of transgenic plants after the floral transition.

Regardless of the underlying mechanism, PtSVP effected different phenotypes in different genotypes (Arabidopsis and tobacco). Ectopic expression of PtSVP affected meristem determination during the reproductive transition, and thus resulted in disturbed flower development. The SVP/StMADS11 group appears to have expanded in perennials in their functions as well as phylogenetically (Jimenez et al. 2009). Perennials might use SVP/StMADS11 genes not only for functions that are required in annual plant models but also for the regulation of seasonal growth, and PtSVP might regulate the annual floral transition in trifoliate orange. These traits may include the formation of floral and vegetative bud structures, regulation of endodormancy cycling, and/or regulation of the juvenile–mature transition (Jimenez et al. 2009).

Conclusion

Our findings suggest that PtSVP plays an important role in floral bud initiation in trifoliate orange. We demonstrated that PtSVP acts as a transcription factor correlated with the floral transition and may be involved in meristem maintenance, retaining the states of both vegetative and reproductive growth. The down-regulation of PtSVP in annual spring shoots may be the cue triggering annual flowering in perennial trees. These findings provide a new understanding of the functions of the SVP clade. Further studies are required to understand how PtSVP is regulated by or regulates other genes to maintain the meristems and whether it directly regulates the flowering promoter CiFT to hamper the floral transition. Answers to these questions will greatly improve our understanding of the annual flowering mechanisms of citrus and other woody plants.