Introduction

MADS-box genes encode a large family of transcriptional regulators with diverse developmental functions. However, the floral organ identity genes are, by far, the most intensively researched. These encode a family of transcription factors that have key roles in specifying the identity of each whorl of angiosperm flowers. Recent progress in understanding the floral developmental program has predominantly come from the study of two model systems, Arabidopsis thaliana and Antirrhinum majus. Genetic studies with these two species have shown that floral organ identities are determined by a combination of five classes of genes (A, B, C, D and E). This work has produced an expanded ABCDE model of floral development and also the “floral quartet” model that predicts how the different floral homeotic genes interact at the molecular level to specify the various floral organ identities (Coen and Meyerowitz 1991; Theissen and Saedler 2001). In the ABCDE model, A- and E-class genes specify sepal structures, A + B + E classes determine petal formation, B + C + E direct the formation of stamens, and C + E class factors interact to specify the identity of carpels; D- and E- class genes are involved in ovule development. In the model plant Arabidopsis, class A genes are represented by APETALA1 (AP1) and APETALA2 (AP2; Mandel et al. 1992; Jofuku et al. 1994), class B genes by APETALA3 (AP3) and PISTILLATA (PI; Goto and Meyerowitz 1994; Yang et al. 2003), class C genes by AGAMOUS (AG; Yanofsky et al. 1990), class D genes by SEEDSTICK (STK; formerly AGL11) (Pinyopich et al. 2003; Kaufmann et al. 2005) and class E genes are represented by SEPALLATA1, 2, 3 and 4 (SEP1, 2, 3 and 4; previously known as AGL2, 4, 9 and 3, respectively; Pelaz et al. 2000; Honma and Goto 2001; Ditta et al. 2004). With the exception of AP2, each of these ABCDE floral homeotic genes represents a MIKCC-type MADS-box gene, this term referring to their conserved structure comprising an M (MADS), I (intervening), K (keratin-like) and C (C-terminal) domain (Sommer et al. 1990; Bowman et al. 1991, 1993; Becker and Theissen 2003).

Platanus acerifolia is a monoecious tree species that is frequently used in city landscaping schemes due to its desirable vegetative growth characteristics. However, undesirable traits of P. acerifolia include the abundant release of pollen and seed hairs that pollute the environment and are known to be problematic as allergens. Thus, floral organ identity MADS-box genes are of interest to us as potential targets to engineer male- and/or female-sterile lines (Li et al. 2008).

By studying the MADS-box genes in P. acerifolia, we hope to further clarify the phylogenetic position of the Platanaceae family. The isolation of these full-length MADS-box gene clones offers the potential for genetic transformation techniques (Li et al. 2007) to be applied to develop male and/or female sterile lines in P. acerifolia by gene silencing (Wesley et al. 2001; Helliwell et al. 2002).

Materials and methods

Biological material, RNA isolation and sequencing

All plant material used in this study was taken from trees of Platanus acerifolia Willd. growing within the campus grounds of Huazhong Agricultural University, Wuhan, China. The trees were of mixed ages, with 0.5-year-old seedlings or 4-year-old seedlings having been generated by the germination of seeds taken from 40-year-old adult trees. During the period April–December 2007, various tissue samples were collected from the juvenile and adult plants. Juvenile samples comprised: leaves from 0.5-year-old seedlings (JL); stems from 0.5-year-old seedlings (JS); roots of 0.5-year-old seedlings (JR); subpetiolar buds of 4-year-old saplings (JSB). Samples from the 40-year-old adult trees included: mature embryos (ME); subpetiolar buds (ASB, Fig.S1-b, c); leaves (AL); stem tissue (AS). Inflorescences were also collected from the adult trees: male inflorescences designated 7MF, 8MF, 10MF, 12MF, 3MF and 4MF were collected during the months July (stamen and pistil differentiation phase, Fig.S1-h; Fig.S2-b), August (completion of stamen and pistil differentiation, Fig.S1-i; Fig.S2-c), September (there is no obvious difference in differentiation on spherical head but size, Fig.S1-i; Fig.S2-d), December (stop of floral growth and development, there is no obvious difference comparing with September but size), March (microspore mother cell in meiosis) and April (pollen mature, stigma to be red), respectively; female inflorescences designated 12FF and 4FF were collected during December and April. Male inflorescences at specific ontogenetic stages were collected on July 15, August 15, October 16, December 15, March 15 and April 16 of 2007. Female inflorescences at various ontogenetic stages were collected on April 16 and December 15, 2007. All samples were immediately frozen in liquid nitrogen but, in the case of the flower samples, bud scales were removed before freezing. Total RNA was prepared from the individual tissues using CTAB (Li et al. 2008).The construction of double-stranded cDNA, to be used for rapid amplification of cDNA end (RACE), was performed using the SMART™ cDNA Library Construction Kit (Li et al. 2008). For amplification of the P. acerifolia MADS-box sequences, 3′-RACE experiments were conducted using a 3′ PCR primer and a degenerate primer (PMADSF, 5′-GTKCTHTGYGAYGCYGARRTTGC-3′) (additional data are given in Online Resource Supplementary Table S1) that corresponded to the conserved MADS-box amino acid sequence, VLCDAEV (Van der Linden et al. 2002). The PCR reagents comprised 5 μL 10× DNA polymerase buffer, 3 μL MgCl2 (25 mM each), 1 μL 10 mM dNTP (2.5 mM each), 1 μL of each GSP (gene specific primer) (additional data are given in Online Resource Supplementary Table S1) (10 mM each), 1 μL template, 0.5 μL DNA polymerase (MBI Fermentas) and adjusted with water to a final volume of 50 μL. Touchdown-PCR reactions were initially heated to 94°C for 2 min, followed by 5 cycles of 94°C for 20 s, 64°C for 50 s, 72°C for 90 s; the annealing temperature was reduced in each subsequent set of 5-cycles by 3°C until it reached 58°C, after which there followed 25 cycles of: 94°C for 20 s, 55°C for 50 s, 72°C for 90 s; a final extension step was performed at 72°C for 10 min. 5′-RACE was performed with a 5′ PCR primer and gene-specific primer (GSP) (additional data are given in Online Resource Supplementary Table S1). The reaction was initially heated at 94°C for 2 min, followed by 35 cycles of 94°C for 30 s, 60°C for 45 s and 72°C for 90 s, and followed by a final extension step at 72°C for 10 min. The PCR products were purified with a gel PCR purification kit (Omega, USA) and cloned into the pMD-18T Vector system (Takara, Japan). Ligation products were transformed into Escherichia coli DH5a-competent cells (Takara, Japan), following the manufacturer’s instructions. Sequencing of cDNA from three independent clones was performed using bigdye terminator V3.1 cycle sequencing kit with an ABI 3700 sequencer.

Sequence alignment and phylogenetic analysis

The sequences of selected species were downloaded from the NCBI GenBank (http://www.ncbi.nlm.nih.gov). The various taxa included in the analysis (additional data are given in Online Resource Supplementary Table S2–5) were chosen to represent angiosperm diversity at the order level. All homologues or paralogues from each selected taxon were included in the analysis.

The DNA and amino acid sequences of A-, B-, C- and E-class MADS-box genes were aligned by ClustalX 1.83 respectively (Thompson et al. 1997). The conserved MIK or MIKC domains were included for phylogenetic analysis. We first used Model Test 3.7 to select the best-fit distance models and parameters for the DNA datasets with MIKC domains (Posada and Crandall 1998), and Tree-Puzzle 5.2 for the amino-acid datasets with MIK domains (Schmidt et al. 2002). The individual maximum-likelihood (ML) trees for each of the A-, B-, C- or E-class MADS-box genes were constructed by Phyml 2.4.4 with GTR + I + Γ model. The combined ML tree for all four classes of MADS-box genes was built with JTT + I + Γ model (Guindon and Gascuel 2003). The bootstrap support values (500 replications) were also computed. Neighbor-joining (NJ) bootstrap analysis (2,000 replications) with the maximum composite likelihood model for the DNAs and the Poisson correction for the amino-acids were performed by MEGA 4 (Tamura et al. 2007). The 50% majority rule consensus trees were inferred using MrBayes v3.12 with the GTR + I + Γ model for DNAs and WAG + I + Γ model for amino-acids (Ronquist and Huelsenbeck 2003). Metropolis-coupled MCMC (MCMCMC) method from a random starting tree was initiated in the Bayesian inference and run for 2,000,000 generations with trees sampled every 100th generation. The first 5,000 trees were discarded as ‘‘burn-in’’ and the subsequent 15,000 were used to calculate the consensus tree. During the analyses, AGL6, GGM2, GGM3, AcMADS600 and AGL15 were used as out-groups for the individual A-, B-, C-, E-classes and the combined four-class MADS-box genes, respectively.

Expression analysis by RT-PCR

Samples of total RNA were treated with RQ1 RNase-Free DNase (Promega, USA) in order to remove DNA contamination. Single-stranded cDNA was synthesized with M-MLV Reverse Transcriptase and Oligo (dT)15 primer (Promega, USA). In order to amplify cDNA from the different tissues we used a reverse primer located in the PlacAG 3′ UTR in conjunction with a specific forward primer, or specific forward and reverse primers in the 5′ and 3′ UTR of PlacFUL, PlacAP3, PlacSEP1 and PlacSEP3 genes (additional data are given in Online Resource Supplementary Table S6). The Histone gene of P. acerifolia was identified and primers designed to act as an internal control for the RT-PCR (Li et al. 2008).

Results

Sequence identification of MADS-box genes and conserved C-terminal motifs analysis in P. acerifolia

A total of 39 cDNA clones were isolated from P. acerifolia flowers by 3′ RACE, and 22 of these were identified as homologues of floral organ identity genes. These comprised FUL, AP3, AG, SEP1 and SEP3 homologues, each homologue type being represented by 2, 1, 1, 7 and 11 of the clones, respectively. GenBank blast and phylogenetic analysis showed these to be homologous to floral A-, B-, C- and E-class genes (Fig. S4).

Full-length P. acerifolia sequences for four of the genes (namely, PlacFUL, PlacAP3, PlacSEP1 and PlacSEP3) were obtained by RACE and a partial clone of PlacAG was also obtained. PlacFUL is 996 bp in length and has an ORF of 735 bp which encodes a predicted polypeptide of 245 aa with 5′/3′-UTR of 128 and 103 bp. PlacAP3 is 930 bp in length and has an ORF of 678 bp which encodes a predicted polypeptide of 226 aa with 5′/3′-UTR of 83 and 139 bp. PlacSEP1 is 1,120 bp in length and has an ORF of 735 bp which encodes a predicted polypeptide of 245 aa with 5′/3′-UTR of 123 and 251 bp. PlacSEP3 is 993 bp in length and has an ORF of 720 bp which encodes a predicted polypeptide of 226 aa with 5′/3′-UTR of 64 and 179 bp. Clone PlacAG contains 838 bp from the gene’s 3′-end and this contains a partial ORF of 570 bp, which encodes a predicted polypeptide of 190 aa and a 3′-UTR of 243 bp.

The predicted amino acid sequences of the genes included in this analysis have the typical “MIKC” structure of plant type II MADS domain-containing proteins (Alvarez-Buylla et al. 2000). A-, B-, C- and E-class genes each typically have a characteristic C-terminal motif. Thus, most known AP1/FUL sequences encode C-terminal regions that are rich in glutamine, although regions rich in proline, serine or glycine are also common. The C-terminal ends of all known FUL-like and euFUL codons contain a highly conserved hydrophobic six-amino-acid sequence (known as the FUL-like motif: L/MPPWML) and this is generally followed by either two basic residues or one polar and one basic residue (Fig. 1). The euAP1 sequences, instead, encode a distinct C terminus with two short conserved motifs namely, the euAP1 motif (RRNaLaLT/NLa, where “a” is an acidic residue) and the farnesylation motif (CFAT/A) which terminates the protein (Fig. 1). The PI, euAP3, TM6 and paleoAP3 sequences are each characterised by a highly conserved amino acid sequence (i.e. PI motif: F??RVQPMQPNLQE; euAP3 motif: D??TF?LLE; TM6/paleoAP3 motif: F/YG??DRLR) (Fig. 1). Most published AG or SEP sequences display two C-terminal motifs. Thus, C-class genes contain the AG I and AG II motifs, and E-class genes contain the SEP I and SEP II motif (including WML or WL) (Fig. 1). The P. acerifolia genes isolated in this study contained the respective C-terminal motifs, according to the A-, B-, C- or E-class of each gene, which the FUL-like motif of PlacFUL was LPPWML, the paleoAP3 motif of PlacAP3: YGFRDRLR, the AG I and AG II motif of PlacAG: FDSRNFLQVNQME and YSRQESIALQLG, the SEP I and SEP II motif of PlacSEP1: FFQALECNSTLQIGY and AQNVNGFIPGWML, the SEP I and SEP II motif of PlacSEP3: FFHPLECEPTLQIGY and PCVNNYMPVWLA (Fig. 1).

Fig. 1
figure 1

Representative predicted amino acid sequences alignment of ABCE genes from P. acerifolia and model plant (Arabidopsis and poplar). Conserved motifs are boxed, as defined by previous studies for the FUL/SEP motif (Litt and Irish 2003), the PI and TM6/paleo-AP3 motifs (Kramer et al. 1998), and the AG motif (Kramer et al. 2004). Gly-110 is indicated by the downward arrowhead and solid box outline

Full size image

Phylogenetic analysis

Sequence information for 599 MADS-box genes was collated by using published MADS-box gene sequences, database mining and sequences isolated by degenerate PCR and RACE analyses. These 599 sequences originated from phylogenetically diverse species and represented each of the MADS-box lineages, with 162 belonging to class A, 186 to class B, 111 to class C/D and 140 to class E (additional data are given in Online Resource Supplementary Table S2–5). The sequences of the five P. acerifolia floral organ identity gene homologues were aligned with the respective A-, B-, C- or E-class genes of the multiple angiosperm taxa for phylogenetic analysis.

The PlacFUL sequence was found to form a clade with none-core eudicot FUL-like homologues from other none-core basal eudicots, i.e. TraFUL1 and TraFUL2 of Trochodendron aralioides (Trochodendrales) (Wu et al. 2007), NEnuFL1 of Nelumbo nucifera (Nelumbonaceae, Proteales), PapsFL1 of Papaver somniferum, RbFL3 of Ranunculus bulbosus, PapsFL1 of Papaver somniferum (Ranunculales), BUseFL1, BUseFL2 and BUseFL3 of Buxus sempervirens and PatFL1 or PAteFL1 of Pachysandra terminalis (Buxales) (Fig. 2).

Fig. 2
figure 2

Phylogenetic tree of A-class genes based on maximum-likelihood (ML) analysis. AGL6 is shown as an out-group. Numbers on the branches are bootstrap values obtained from ML analysis followed by posterior probabilities from Bayesian inference (BI) and bootstrap from neighbor-joining (NJ). Only values over 50% in the bootstrap analyses and Bayesian posterior probability are indicated. The two stars indicate the AP1/FUL and euFUL/core eudicot FUL-like gene duplication, respectively, prior to the diversification of core eudicots, which might have happened very closely in time. In monocots, the two major duplication events are indicated with black circles, also the other mall-scale

Full size image

The AP3 homologue of P. acerifolia, PlacAP3, formed a sister group to PlaocAP3-1 and PlaocAP3-2 genes of P. occidentalis and formed a clade with other paleoAP3 homologues from basal eudicots, namely, AktAP3 of Akebia trifoliata and DeAP3 of Dicentra eximia (Ranunculales), MdAP3 of Meliosma dilleniifolia (Sabiaceae), NnAP3 of Nelumbo nucifera (Nelumbonaceae, Proteales), TraAP3 of Trochodendron aralioides (Trochodendrales), PpAP3-1 of Pachysandra procumbens, PtAP3-1 of Pachysandra terminalis, MdAP3-1 of Meliosma dilleniifolia (Buxales) (Fig. 3).

Fig. 3
figure 3

Phylogenetic tree of B-class genes based on maximum-likelihood (ML) analysis. GGM15 is used as outgroup. Numbers on the branches are bootstrap values obtained from ML analysis followed by posterior probabilities from Bayesian inference (BI) and bootstrap from neighbor-joining (NJ). Only values over 50% in the bootstrap analyses and Bayesian posterior probability are indicated. The two stars indicate the paleo-AP3/PI, and euAP3/TM6 gene duplication, respectively; the red triangle, a gene duplication in the paleo-AP3; black circles, the other mall-scale gene duplications

Full size image

The AG homologue from P. acerifolia, PlacAG, also aligned near the base of the core eudicots, i.e. before the separation of the two major euAG clades, SHP/PLE and AG/FAR. PlacAG formed a clade with other none-core eudicot AG homologues of the basal eudicots, namely, MdAG1 of Meliosma dilleniifolia (Sabiaceae, Buxales), TraAG1 and TraAG2 of Trochodendron aralioides (Trochodendrales), EScaAG1 of Eschscholzia californica, BgAG of Berberis gilgiana, AkqAG1 of Akebia trifoliate, RfAG2 of Ranunculus ficaria (Ranunculales) and CsAG1 of Chloranthus spicatus (Chloranthales) (Fig. 4).

Fig. 4
figure 4

Phylogenetic tree of C-class genes based on maximum-likelihood (ML) analysis. GGM3 is used as outgroup. Numbers on the branches are bootstrap values obtained from ML analysis followed by posterior probabilities from Bayesian inference (BI) and bootstrap from neighbor-joining (NJ). Only values over 50% in the bootstrap analyses and Bayesian posterior probability are indicated. The red triangle and diamond indicate the C/D and euAG/PLE gene duplication, respectively; the red triangle, a gene duplication in the SHP1/SHP2 lineage; the black circles, gene duplications in the grass C and D lineages; the black triangle, a gene duplication in the Ranunculales C lineage

Full size image

The two SEP homologues from P. acerifolia separated into two basal eudicot groups, namely, PlacSEP1 in the SEP1 clade and PlacSEP3 in the SEP3 clade. PlacSEP1 was placed close to EScaAGL2 of Eschscholzia californica (Ranunculales). PlacSEP3 formed a sister group to the cluster containing PEamAGL9.1 of Persea americana (Laurales), Ma.gr.AGL9 of Magnolia grandiflora (Magnoliales), EScaAGL9 of Eschscholzia californica (Ranunculales) and AktSEp3-1 of Akebia trifoliata (Ranunculales) (Fig. 5).

Fig. 5
figure 5

Phylogenetic tree of E-class genes based on maximum-likelihood (ML) analysis. AcMADS600 is used as outgroup. Numbers on the branches are bootstrap values obtained from ML analysis followed by posterior probabilities from Bayesian inference (BI) and bootstrap from neighbor-joining (NJ). Only values over 50% in the bootstrap analyses and Bayesian posterior probability are indicated. The red and blue star indicate AGL2/3/4-like/AGL9 and AGL2/4/AGL3 gene duplication, respectively; the triangles, a gene duplication in the gymnosperm AGL6 and the monocots AG9 lineage; black circles, the other mall-scale gene duplications

Full size image

Phylogenetic evolution analysis of sequence databases of A-, B-, C- and E-class MADS-box genes with the PlacFUL, PlacAP3, PlacAG, PlacSEP1 and PlacSEP3 sequences indicates that P. acerifolia is one of the basal eudicots and is close to Trochodendrales, Buxales, Ranunculales and Chloranthales (Fig. 6).

Fig. 6
figure 6

Outline of angiosperm phylogeny. Species used in this study are indicated according to taxonomic group

Full size image

Expression analysis

RT-PCR was used to detect the expression of the P. acerifolia floral identity gene homologues within various tissues, including the vegetative organs of juvenile and adult plants and male and female inflorescences taken at a range of developmental stages. The results obtained following 35 cycles of PCR are shown in Fig. 7. Transcripts of the genes (excluding PlacAG) were detected, albeit weakly, within some vegetative organs. Thus, PlacSEP3 transcripts were detected in embryos (ME) and seedlings roots (JR), but not in leaves (JL, AL) or stems (JS, AS). Transcripts of PlacSEP3, PlacSEP1 and PlacFUL (but not PlacAG or PlacAP3) were detected in the subpetiolar buds of adult trees (ASB). PlacSEP1 appeared to be expressed in AS but not in AL, whereas, the other four genes were not detected to any significant level in either AL or AS. All five Platanus MADS-box genes were expressed to significant levels within the reproductive structures, although the abundance of the respective transcripts varied according to the inflorescence type. PlacFUL was found to be expressed in the male inflorescence at stages 10MF and 12MF (and weakly in 3MF) and female inflorescences at stage12FF (and weakly in 4FF). PlacAP3 was expressed in all eight samples of the male and female inflorescences, with particularly strong signals detected in 12MF and 3MF and the weakest levels found in 4MF. Expression of PlacAG was detected in 8MF, 10MF and 12MF, and also 12FF and 4FF samples. PlacSEP1 and PlacSEP3 transcripts were both found throughout the male and female inflorescence samples, although levels varied with sampling stage.

Fig. 7
figure 7

RT-PCR analysis of expression of Platanus acerifolia A-, B-, C- and E-class MADS-box gene homologues (PlacFUL, PlacAP3, PlacAG, PlacSEP1 and PlacSEP3) in vegetative organs (ME mature embryo, JL juvenile leaf, JS juvenile stem, JSB juvenile subpetiolar bud, ASB adult subpetiolar bud, AL adult leaf, AS adult stem) and inflorescences (7MF, 8MF, 10MF, 12MF, 3MF, 4MF = various stages of male flower—see “Materials and methods” for details; 12 FF, 4FF = various stages of female flower). Results following 35-cycle reactions are shown. The PaHiston gene, from P. acerifolia, was included as an internal control for the reactions

Full size image

Discussion

We obtained five MADS-box homologues from Platanus acerifolia, namely, PlacFUL, PlacAP3, PlacAG, PlacSEP1 and PlacSEP3. Phylogenetic analyses grouped these Platanus genes according to the respective A-, B-, C- or E-class of MADS-box floral organ identity genes, typical of angiosperm species. In addition, the C-terminal regions of the respective P. acerifolia genes contained conserved signature motifs, as typical of A-, B-, C- or E-class genes (Vandenbussche et al. 2003) (Fig. 1). These conserved polypeptide motifs support the possibility that the P. acerifolia genes fulfill similar functions to their homologues in other flowering plants. Recent studies have indicated that a major duplication event may have occurred for floral ABC-class genes, at the base of the core eudicot lineages (Kramer et al. 1998, 2004; Litt and Irish 2003). Studies of AP3 gene phylogeny have provided evidence that such a duplication event occurred after the branching of P. acerifolia (Kramer et al. 2006). Immediately following gene duplication, a frame shift mutation occurred in one of the AP3 copies giving rise to the euAP3 lineage (Kramer et al. 2006). The results of our study with Platanus acerifolia indicate that A- and C-class gene phylogenies are also consistent with this scenario. We found that each class of the P. acerifolia homologues includes a preduplication form of the floral identity genes, as characterized by typical paleotypes of gene product motifs (Irish 2003; Vandenbussche et al. 2003) (Fig. 1). By contrast, the three PlacFUL homologues found in P. acerifolia form sister groups in A-class phylogenies (Fig. 2) and this could reflect an ancient duplication of a partial or whole chromosome within the Platanus lineage. The ABCDE model of floral development was largely derived from studies of two key model systems, Arabidopsis and Antirrhinum majus (Coen and Meyerowitz 1991; Theissen and Saedler 2001). Evidence of true A-class function has been limited to the Arabidopsis (AP1 and AP2) species (Keck et al. 2003; Ferrario et al. 2004), with AP1/FUL homologues in other model plants, such as rice and petunia, failing to demonstrate A-class function (Ferrario et al. 2004). This restricts the inference of A-class function for homologues in other plant species. Thus, although we successfully identified the AP1 homologue PlacFUL, we do not necessarily infer class-A function in P. acerifolia. Another A-function candidate, AP2, was not included in this study but is worthy of further investigation. By contrast, B and C functions appear highly conserved amongst the various B- and C-class gene homologues found in different angiosperm species (Kramer and Irish 2000; Ferrario et al. 2004). Nonetheless, direct evidence is required for functionality of the ABCE genes in Platanus and, since genetic experiments are very difficult to conduct in woody perennials, this is likely to come indirectly from complementation or over-expression experiments in Arabidopsis mutant backgrounds. In a similar evolutionary path to the B-class genes (Kramer et al. 2006), it has been proposed that the AP1 lineage resulted from a major duplication event that occurred near the base of the core eudicots, giving rise to the euAP1 and euFUL lineages of the core eudicots (Litt and Irish 2003; Shan et al. 2007). Although euAP1 is consistently expressed in sepals and petals (Irish, 2003), expression of euFUL homologues is usually restricted to the carpels and bracts (Mandel and Yanofsky 1995; Gu et al. 1998; Müller et al. 2001). Previous studies have indicated that the B-class genes of core eudicots are stably expressed in petals and stamens. However, this is not always true of the B-class genes of basal eudicots and basal angiosperms (Zahn et al. 2005b). For instance, AP3 and PI homologues from Ranunculaceae are expressed throughout the petaloid perianth and the stamens, but some homologues are also expressed in the carpels (Kramer et al. 2003). In Eupomatia, a member of the Eupomatiaceae (Magnoliales), the AP3 homologues are also weakly expressed in the calyptras and leaves, while in Magnolia, the AP3 homologue is expressed in spathaceous bracts (Kim et al. 2005). The AG-like homologues are usually expressed exclusively in stamens and carpels and play specific roles relating to C function (De Bodt et al. 2003; Irish 2003). Based on the reconstructed phylogeny of E-class genes, conducted in this study, PlacSEP3 is assigned to belong to the SEP3 clade (Malcomber and Kellogg 2005) or the AGL9 clade (Zahn et al. 2005a). PlacSEP1 appears to belong to the LOFSEP clade (Malcomber and Kellogg 2005) or, perhaps, the AGL2/3/4 clade (Zahn et al. 2005a). Expression of PlacSEP1 and PlacSEP3 was found to be strongest in female inflorescences, compared to the male flowers, and this is consistent with the expression patterns of homologous genes in other species. Our results in Platanus are consistent with the general finding that E-class homologues are usually present as multiple copies in flowering plants and may have redundant or very diverse functions (this has been discussed more fully in reviews; Malcomber and Kellogg 2005; Zahn et al. 2005a). In Arabidopsis, LFY and FUL floral meristem identity genes have been shown to be the dominant B- and C-class genes, respectively, and show the appropriate distinct expression patterns (Busch et al. 1999; Lamb et al. 2002). In P. acerifolia, however, RT-PCR and qRT-PCR results indicated that PaLFY is expressed in a similar manner to PlacFUL (data not shown here). Since the four whorls of the P. acerifolia flower are not readily separated for RT-PCR analysis, in situ hybridization studies will need to be performed to determine the detailed expression patterns.

Phylogenetic evolution analysis of A-, B-, C- and E-class MADS-box genes, PlacFUL, PlacAP3, PlacAG, PlacSEP1 and PlacSEP3, shows that P. acerifolia is one of the basal eudicots and is close to Trochodendrales, Buxales, Ranunculales and Chloranthales. Thus, together with Proteaceae and Nelumbonaceae, Platanaceae belongs to the well-established order Proteales. These three families are morphologically very disparate and the fact that they assemble together exemplifies the diversity of floral structure and organization that is present among basal eudicot lineages. At this level of angiosperm evolution, floral structure is characterized by a wide range of variations relating to the number and arrangement of the reproductive organs and there is often weak differentiation, only, of the tepals and a variable phyllotaxis (Endress 1987a, b, 1994; Drinnan et al. 1994; von Balthazar and Endress 2002; Ronse DeCraene et al. 2003; Soltis et al. 2003, 2005).

The A-, B-, C-, E-class genes from the basalmost angiosperms (Amborellaceae, Nymphaeaceae, and Austrobaileyaceae, Schisandraceae and Illiciaceae), always hold the basalmost position, followed by genes from magnoliids (Magnoliales, Laurales, Piperales, Canellales), monocots, Chloranthaceae, early-diverging eudicots, and core eudicots (Fig. 6). In early-diverging eudicots, genes from each of the Buxaceae, Trochodendrales, Proteales, Sabiaceae and Ranunculales form separate well-supported clades, although the relationships among these clades are still uncertain. Within core eudicots, three distinct gene lineages are evident, each of which contains genes from rosids, asterids, Caryophyllales, and Saxifragales, suggesting that the two gene duplication events giving rise to these three clades predated the diversification of core eudicots.

Thus, in summary, we obtained five MADS-box homologues from P. acerifolia, identified as PlacFUL, PlacAP3, PlacAG, PlacSEP1 and PlacSEP3, which group with none-core eudicot FUL-like, Paleo-AP3, none-core eudicot AG, none-core eudicot AGL2 and AGL9. Phylogenetic analyses grouped these P. acerifolia genes according to the respective A-, B-, C- or E-class of MADS-box floral organ identity genes, typical of angiosperm species. The C-terminal regions of the P. acerifolia genes possess the conserved signature motifs of the respective A-, B-, C- or E-class genes (Vandenbussche et al. 2003) (Fig. 1), and therefore it seems possible that these P. acerifolia genes may fulfill similar functions to their homologues in other flowering plants. RT-PCR analysis detected expression of the genes within the male and female inflorescences. Unfortunately, however, the anatomy of the P. acerifolia flowers prevented us from separating the four whorls for RT-PCR analysis of the individual floral structures. Therefore, more detailed, whorl-specific expression analysis requires future investigation using in situ hybridization techniques.

In conclusion, the various floral homeotic gene phylogenies consistently placed P. acerifolia very close to the basal eudicots. This putative position of P. acerifolia as a basal eudicot makes it a particularly interesting candidate species for further studies regarding the timing and role of MADS-box gene duplications.