Introduction

Genetic studies of Arabidopsis and Anthirrinum mutants have led to the ABC model of floral organ determination, in which the combinatorial action of three regulatory functions leads to organ flower formation (Coen and Meyerowitz 1991). Based on single-mutant, double-mutant, and triple-mutant analyses, the ABC model proposes three classes of combinatorial-acting floral organ identity genes, called A, B, and C. A functions specify sepals in the first floral whorl; A plus B specify petals in the second whorl; B plus C specify stamens in the third whorl; and C specifies carpels in the fourth whorl (Coen and Meyerowitz 1991). A few years after the ABC model was proposed, studies on ovule development in petunia led to the isolation and characterization of two MADS-box genes, Floral Binding Protein 7 (FBP7) and FBP11, whose expression is turned on just before ovule primordia arise and is maintained in the seed coat after fertilization (Angenent et al. 1995; Colombo et al. 1995; Favaro et al. 2003; Ferrario et al. 2003). Moreover, ectopic expression and co-suppression of these genes in petunia and Arabidopsis thaliana indicated that their activity is necessary for determining ovule identity, an activity that has been proposed as a separate floral organ identity function, the D-function gene (Angenent et al. 1995; Colombo et al. 1995; Favaro et al. 2003). As a result of these studies, the ABC model has become more complex, and now includes a new class of genes referred to as the E-function genes, SEPALLATA1 (SEP1), SEP2, and SEP3, which are required for B and C floral organ identity functions (Egea-Cortines and Davis 2000; Pelaz et al. 2000; Honma and Goto 2001; Jack 2001).

In Arabidopsis thaliana, SEP1, SEP2, SEP3, and SEP4 are required for specifying the identity of all four whorls of the floral organ, and for floral meristem determinacy (Honma and Goto 2000; Pelaz et al. 2000, 2001; Ditta et al. 2004). SEP1, SEP2, and SEP4 are expressed throughout the floral meristem at stage 2, slightly earlier than SEP3, which is expressed in a region corresponding to the inner three whorls just before the initiation of floral organ primordia (Flanagan and Ma 1994; Savidge et al. 1995; Mandel and Yanofsky 1998; Ditta et al. 2004). Subsequently, SEP1 and SEP2 expressions persist in all floral organ primordial, and SEP3 is expressed in the inner three whorls. In contrast, SEP4 becomes more highly expressed in the central dome than in the sepals. Genetic studies of quadruple mutants indicated that the SEP genes are functionally redundant in the regulation of floral organ development and have redundant roles in promoting floral meristem determinacy (Pelaz et al. 2000; Ditta et al. 2004). In petunia, which is distantly related to Arabidopsis, SEP-like genes function similarly to those in Arabidopsis. Vandenbussche et al. (2003b) reported that FBP2 and FBP5 in petunia were required for B, C, and D floral organ identity functions, based on the analysis of fbp2fbp5 double mutants. Moreover, two MADS-box genes in silver birch (Betula pendula: Betulaceae), BpMADS1 and BpMADS6 (Lemmetyinen et al. 2004), and two genes in apple (Malus domestica: Rosaceae), MdMADS3 and MdMADS4 (Sung et al. 2000), are phylogenetic and functional homologues to SEP-like genes. These results imply that these sets of genes have similar functions among these plants and perhaps function in more distantly related groups as well.

The monocotyledonous plants are a large group, but many of the molecular studies of MADS-box genes in monocots are carried out in the grass species, tulip (Kanno et al. 2003), Commelina and Tradescantia (Ochiai et al. 2004), Phalaenopsis (Tsai et al. 2004), Agapanthus (Nakamura et al. 2005), Muscari (Nakada et al. 2006), and Dendrobium (Skipper et al. 2006). Garden asparagus (Asparagus officinalis L.) is a dioecious monocot species belonging to the genus Asparagus (Asparagaceae), which is a large genus comprising 100–300 species. Because this genus contains hermaphrodite and dioecious species (Kanno et al. 1997; Ochiai et al. 2002; Fukuda et al. 2005), the genus Asparagus is an important and interesting model for studying sex differentiation and floral development. Several MADS-box genes have been isolated and characterized from A. officinalis (Caporali et al. 2000; Park et al. 2003, 2004; Losa et al. 2004; Ito et al. 2005) and from the ornamental species, A. virgatus (Yun et al. 2004a, b). On the basis of the expression analyses of isolates, Caporali et al. (2000) and Losa et al. (2004) reported the existence of SEP-like genes in A. officinalis (AOM1 and AOM4), although an additional SEP-like gene isolated in this study was not well understood. Here, in order to clarify the function of SEP-like genes in the genus Asparagus, we isolated and characterized three SEP-like genes, AOMADS1, AOMADS2, and AOMADS3, from A. officinalis, and three genes, AVMADS1, AVMADS2, and AVMADS3 from A. virgatus. In addition, we examined the evolutionary histories of these genes in Asparagus species.

Materials and methods

Plant material

The roots, shoots, phylloclades, and flower buds of A. officinalis L. cv. Mary Washington 500 W and A. virgatus Baker were collected from the experimental field at Tohoku University. All materials were frozen in liquid nitrogen immediately after collection and stored at −80°C. For Northern blot analysis, we used 1–2-mm long flower buds, which are representative of a late developmental stage. Vouchers sampled in this study have been deposited in the Herbarium, Graduate School of Science, Tohoku University (TUS).

Isolation of cDNA clones

Partial cDNAs from A. officinalis and A. virgatus were isolated by the 3′ rapid amplification of cDNA ends (RACE) method (Frohman et al. 1988; Münster et al. 1997). Poly(A)+ RNA prepared from flower buds of A. officinalis and A. virgatus were used as templates. PCR amplification of the first cDNA strand was done using two MADS-box-specific primers (5′-GATCAAGMGSATCGAGAA-3′ and 5′-GATGAAGMGSATCGAGAA-3′) and an adapter primer. The PCR products were extracted and purified from agarose gels using a QIAquick Gel Extraction kit (Qiagen), then sub-cloned into the pGEM-T Easy Vector (Promega, USA). Upstream sequences overlapping the 3′ fragments were isolated by 5′ RACE using the 5′/3′ RACE kit (Roche). cDNA clones with complete open reading frames were isolated by PCR, using primers located in the 5′ and 3′ UTR regions, cDNA pools as template. Sequencing was performed using the ABI PRISM DNA Sequencing kit (Perkin-Elmer Applied Biosystems) according to the manufacturer’s protocol.

Phylogenetic analysis

We used 124 amino acid sequences from different MADS-domains and K-domains for the phylogenetic analysis. To construct phylogenetic trees for SEP-like genes, the amino acid sequences were aligned using CLUSTAL X (Thompson et al. 1997), based on the BLOSUM protein weight matrix. Phylogenetic relationships were analyzed by the neighbor joining (NJ) method and the maximum-likelihood (ML) method. For NJ analysis, we constructed phylogenetic tree which is drawn by Njplot (Perriére and Gouy 1996). Bootstrap values of this phylogenetic tree were derived from 1,000 replicate runs (Thompson et al. 1994; http://www.ddbj.nig.ac.jp/E-mail/homology.html). For ML analysis, we used the PROML program of the PHYLIP v. 3.6 program package (Felsenstein 2004). We used JTT with constant rates across sites for this analysis (Jones et al. 1992). All indels were counted as missing. We performed ten random sequence addition searches using the J option, and global branch swapping using the G option, in order to identify the ML tree with the best log-likelihood. We then performed a bootstrap analysis (Felsenstein 1985) with 100 replications using the SEQBOOT program of PHYLIP. The GenBank accession numbers of the amino acid sequences used are: ACamAGL2 (AY850184: Acorus americanus), AGL6 (M55554: Arabidopsis thaliana), AGLG1 (AY244507: Triticum monococcum), AktSEP3-1 (AY627628: Akebia trifoliata), AMtrAGL2 (AY850179: Amborella trichopoda), AMtrAGL9 (AY850178: Amborella trichopoda), AOM3 (AY383559: A. officinalis), AOM4 (AY383560: A. officinalis), AOMADS1 (DQ344502: A. officinalis), AOMADS2 (DQ344503: A. officinalis), AOMADS3 (DQ344504: A. officinalis), AP1 (Z16421: Arabidopsis thaliana), ApMADS3 (AB079261: Agapanthus praecox), AVMADS1 (DQ344499: Asparagus virgatus), AVMADS2 (DQ344500: Asparagus virgatus), AVMADS3 (DQ344501: Asparagus virgatus), BoAGL3-a (AJ508052: Brassica oleracea), BpMADS1 (AJ252070: Betula pendula), CAGL2 (AF135962: Cucumis sativus), CaMADS1 (AF129875: Capsicum annuum), CDM44 (AY173057: Chrysanthemum morifolium), CMB1 (Q39685: Dianthus caryophyllus), CsSEP3 (AY397763: Chloranthus spicatus), DAL1 (X80902: Picea abies), DcMADS5 (AJ271151: Daucus carota), DEFH49 (X95467: Antirrhinum majus), DEFH72 (X95468: Antirrhinum majus), DEFH200 (X95469: Antirrhinum majus), DlMADS7 (AY599760: Dendrocalamus latiflorus), DlMADS17 (AY599754: Dendrocalamus latiflorus), DOMADS3 (AF198176: Dendrobium grex), EGM3 (AF029977: Eucalyptus grandis), FaMADS (AF484683: Fragaria ananassa), FBP2 (M91666: Petunia hybrida), FBP9 (AF335236: Petunia hybrida), FBP23 (AF335241: Petunia hybrida), FDRMADS1 (AF141966: Oryza sativa), GGM9 (AJ132215: Gnetum gnemon), GGM11 (AJ132217: Gnetum gnemon), GHMADS1 (Accession No. AF538965: Gossipium hirsutum), HcSEP1 (AB089157: Houttuynia cordata), HcSEP2 (AB089158: Houttuynia cordata), HcSEP3 (AB089159: Houttuynia cordata), HvAGL6 (AY541067: Hordeum vulgare), HvAGL9 (AY541068: Hordeum vulgare), LeMADS1 (AY294329: Lycopersicon esculentum), LeMADS5 (AY294330: Lycopersicon esculentum), LpMADS4 (AY198329: Lolium perenne), LpMADS5 (AY198330: Lolium perenne), LpMADS6 (AY198331: Lolium perenne), LpMADS7 (AY198332: Lolium perenne), M79 (Y15008: Oryza sativa), MAGL4 (AF185574: Populus tremuloides), MdMADS1 (U78947: Malus domestica), MdMADS3 (U78949: Malus domestica), MdMADS8 (AJ001681: Malus domestica), MdMADS9 (AJ001682: Malus domestica), MdMADS11 (AJ000763: Malus domestica), MpMADS3 (AB050645: Magnolia praecocissima), MpMADS4 (AB050646: Magnolia praecocissima), MpMADS13 (AB050655: Magnolia praecocissima), MTF1 (AJ223318: Pisum sativum), NsMADS3 (AF068722: Nicotiana sylvestris), NtMADS4 (AF068723: Nicotiana tabacum), NUadAGL2 (AY850183: Nuphar advena), OM1 (Q38694: Aranda deborah), OsMADS1 (L34271: Oryza sativa) OsMADS6 (U78782: Oryza sativa), OsMADS7 (U78891: Oryza sativa), OsMADS8 (U78892: Oryza sativa), OsMADS45 (U31994: Oryza sativa), PaMADS1 (AF372840: Poa annua), PapnSEP3 (AY306174: Papaver nudicaule), PatSEP1 (AY306166: Pachysandra terminalis), PdMADS (AY313211: Prunus dulcis), PhSEP3 (AY306171: Petunia hybrida), PMADS12 (AY370527: Petunia hybrida), PtM6 (AY235222: Populus tremuloides), RbAGL6 (AY306184: Ranunculus bulbosus), RMADS217 (AY551922: Oryza sativa), SaMADSD (Y08626: Sinapis alba), SEP1 (B39534: Arabidopsis thaliana), SEP2 (AY727621: Arabidopsis thaliana), SEP3 (O22456: Arabidopsis thaliana), SEP4 (P29383: Arabidopsis thaliana), SlSEP1 (AB162019: Silene latifolia), SlSEP3 (AB162020: Silene latifolia), SQUA (X63701: Antirrhinum majus), SvAGL6 (AY306188: Syringa vulgaris), SvSEP3 (AY306186: Syringa vulgaris), TAGL2 (AY098738: Lycopersicon esculentum), TaMADS (AF543316: Triticum aestivum), TaMADS12 (AB007505: Triticum aestivum), TDR5 (X60758: Lycopersicon esculentum), VvMADS2 (AF373601: Vitis vinifera), VvMADS3 (AF373602: Vitis vinifera), VvMADS4 (AF373603: Vitis vinifera), ZAG3 (L46397: Zea mays), ZAG5 (L46398: Zea mays), ZMM6 (AJ430692: Zea mays), ZMM7 (Y09302: Zea mays), ZMM24 (AJ430638: Zea mays), ZMM27 (AJ430694: Zea mays), ZMM31 (AJ430640: Zea mays).

Southern blot analysis

Total DNA was extracted from phylloclades of A. officinalis as described by Honda and Hirai (1990). Purified total DNA was digested with three restriction enzymes (EcoRI, HindIII, and BamHI) and separated on a 0.6% agarose gel. Hybridization was performed at 42°C in a solution containing 50% formamide, ×5 SSC, 0.1% N-lauroylsarcosine, 0.02% SDS, and 2% blocking reagent (Roche). Gene-specific probes were labeled with the PCR DIG Probe Synthesis Kit (Roche). Washes were performed twice in ×2 SSC, 0.1% SDS at room temperature for 5 min, and twice in ×0.5 SSC, 0.1% SDS at 65°C for 10 min.

Northern blot analysis

Total RNA was isolated from roots, stems, phylloclades, and flower buds of A. officinalis by the method of Chomczynski and Sacchi (1987). Total RNA from each sample (10 μg each) was separated by electrophoresis on a 1.2% agarose gel containing ×1 MOPS solution and 5% formaldehyde, then transferred to positively charged nylon membranes (Roche). Hybridization was carried out at 65°C overnight in a solution of ×5 SSC, 50% formamide, 0.02% SDS, 0.1% N-lauroylsarcosine, and 2% (w/v) blocking reagent (Roche) using the same DIG-labeled gene-specific probes that were used for Southern hybridization.

Gene-specific RT-PCR

Expression of the AOMADS3 in each floral organ of A. officinalis was characterized by PCR analysis of cDNA isolated from dissected tissues. Gene-specific primers were designed based on known sequences from AOMADS3. The following thermocycling conditions were employed: (94°C, 2 min) × 1 cycle; (94°C, 30 s; 54°C, 30 s; 72°C, 60 s) × 35 cycles; (72°C, 10 min) × 1 cycle. Amplified products were run on a 1% agarose gel and digitally photographed.

RT-PCR RFLP analysis

RT-PCR RFLP analyses of AOMADS1 and AOMADS2 genes were conducted to analyze the relative expression levels of AOMADS1 and AOMADS2. There are nucleotide substitutions present in AOMADS1 and AOMADS2 that alter their susceptibility to certain restriction enzymes. We used PstI (CTGCAG) to discriminate between the two genes, because the AOMADS1 cDNA contains a PstI site in I-region, whereas the AOMADS2 cDNA does not (Fig. 7A).

cDNA prepared from flower buds of A. officinalis was used as a template. The following thermocycling conditions were employed: (94°C, 2 min) × 1 cycle; (94°C, 30 s; 54°C, 30 s; 72°C, 60 s) × 35 cycles; (72°C, 10 min) × 1 cycle. Amplified products were digested with PstI, separated on a 1% agarose gel, and digitally photographed.

Results

Sequencing and phylogenetic analyses of the SEP-like genes in A. officinalis and A. virgatus

We isolated several cDNA clones of MADS-box genes from A. officinalis using the RACE method. BLAST searches revealed that three of these clones had high sequence similarity to the AOM1 gene (Caporali et al. 2000). This suggested that the three cDNA clones represented SEP-like genes, and we named them AOMADS1, AOMADS2, and AOMADS3. Using 5′ RACE, the 5′ region of each of the cDNAs was obtained, and full-length cDNA clones were isolated by PCR using gene-specific primers. At least two independent clones were sequenced for each gene. Using a similar procedure, we isolated three SEP-like genes from A. virgatus, designated AVMADS1, AVMADS2, and AVMADS3.

The predicted amino acid sequence of AOMADS1 and AOMADS2 were almost identical to those of AOM1 and AOM4, respectively, so that these genes are likely to be the same gene (Caporali et al. 2000; Losa et al. 2004). Amino acid sequence alignment of several SEP-like genes revealed that the SEP-like genes from A. officinalis and A. virgatus have well-conserved MADS-domains and K-domains. Moreover, we identified three regions that followed the K-domain sequence that was also conserved among previously published sequences (Fig. 1). We termed these regions “SEP motif I,” “SEP/AGL6 motif,” and “SEP motif II.” Of these motifs, the SEP/AGL6 motif is especially well conserved, not only in SEP-like genes, but also in AGL6-like genes, which are the outgroup lineages. Vandenbussche et al. (2003a) and Zahn et al. (2005) have also suggested that this region is a typical motif in SEP-like genes. Both SEP motif I and SEP motif II were conserved in the SEP-like lineage, but not in the outgroup lineages, although there were some exceptions (Fig. 1).

Fig. 1
figure 1

Alignment of the C-terminal regions of the predicted amino acid sequences for select representatives of SEP-like genes, and their outgroup lineage, AGL6-like genes. Three highly conserved regions, SEP-motif I, SEP/AGL6-motif, and SEP-motif II, are indicated with boxes. Residues in these regions that are highly conserved with respect to the SEP1 consensus sequences are shaded

Full size image

We conducted phylogenetic analyses of SEP-like genes to determine the phylogenetic positions of the SEP-like genes in A. officinalis and A. virgatus that we isolated in this study. We used as our data set most of the published genes in the MADS-box gene family. To construct the phylogenetic tree, we used the amino acid sequences of the well conserved MADS-domains and K-domains. Figure 2 shows the result of the NJ analysis and Fig. 3 shows the result of the ML analysis. AGL6-like genes, AP1 and SQUA, were used as outgroups in both phylogenetic analyses. In both analyses, all SEP-like genes comprise a monophyletic group with high support value, and this group is the sister to the monophyletic group of AGL6-like genes (Figs. 2, 3). In the monophyletic group of SEP-like genes, each group of genes from A. officinalis and A. virgatus: AOMADS1/AOM1 and AVMADS1, AOMADS2/AOM4 and AVMADS2, and AOMADS3 and AVMADS3, respectively, is also monophyly. Among the SEP-like genes in A. officinalis and A. virgatus, AOMADS1/AOM1 and AVMADS1 are more closely related to AOMADS2/AOM4 and AVMADS2 than to AOMADS3 and AVMADS3.

Fig. 2
figure 2

Phylogenetic trees of SEP-like genes generated by the NJ method. Numbers represent bootstrap values of 50% or more support

Full size image
Fig. 3
figure 3

Phylogenetic tree for SEP-like genes obtained using the ML method. The log-likelihood of the best ML tree is -4,577.43. The numbers below the branches are the bootstrap support values ≥50%. SEP-like genes in Asparagus species are indicated in bold

Full size image

Genomic organization and expression analyses of SEP-like genes from A. officinalis

To determine the copy number of AOMADS1/AOM1, AOMADS2/AOM4, and AOMADS3 in the A. officinalis genome, we performed genomic Southern blot hybridization. Total DNA was digested with EcoRI, HindIII, or BamHI and hybridized with the C-terminal region of each cDNA as a gene-specific probe. When probed with an AOMADS1-specific probe, a single hybridization band was observed in the samples digested with EcoRI and HindIII, whereas two bands were detected in the samples digested with BamHI, indicating that the AOMADS1 gene occurs as a single copy in the A. officinalis genome (Fig. 4). Using an AOMADS2-specific probe, we observed several bands, which seemed to be the result of cross-hybridization with AOMADS1 gene. These results of AOMADS1 and AOMADS2 were consistent with previous report of AOM1 and AOM4, respectively (Caporali et al. 2000; Losa et al. 2004). Several bands were also observed for AOMADS3 gene, indicating that it is present as a low-copy gene (Fig. 4).

Fig. 4
figure 4

Southern blot analyses of AOMADS1, AOMADS2, and AOMADS3 in A. officinalis. Each lane contains 20 μg of total DNA digested with BamHI (B), EcoRI (E), or HindIII (H), as indicated above the lanes

Full size image

The expression patterns of the SEP-like genes in A. officinalis were analyzed by Northern hybridization analysis using total RNA isolated from roots, stems, phylloclades, and flower buds. We observed that AOMADS1/AOM1 and AOMADS2/AOM4 were expressed specifically in flower buds (Fig. 5). No transcripts for AOMADS1/AOM1 and AOMADS2/AOM4 were detected in vegetative tissues such as roots, stems, and phylloclades. It should be noted, however, that as the sequence homology between AOMADS1/AOM1 and AOMADS2/AOM4 is very high, it is difficult to distinguish between the expression of one or the other of the two genes. Expression of AOMADS3 could not be detected in any organs by Northern hybridization analysis, so we performed RT-PCR analysis (Fig. 6). RT-PCR analysis revealed that AOMADS3 was also expressed specifically in flower buds (Fig. 6). These results indicated that expression of AOMADS3 is lower than that of AOMADS1/AOM1 and AOMADS2/AOM4 in flower buds (Figs. 5, 6).

Fig. 5
figure 5

Northern blot analyses of AOMADS1 and AOMADS2 in A. officinalis. Each lane contains 10 μg of total RNA from roots (R), stems (S), phylloclades (P), and flower buds (F). Ethidium bromide-staining of rRNA is shown below the panel of Northern blot hybridization

Full size image
Fig. 6
figure 6

Gene-specific RT-PCR of AOMADS3 using cDNA from dissected organs of A. officinalis. AOEF1A, encoding elongation factor 1A gene, is used as a positive control. R roots; S stems; P phylloclades; F flower buds

Full size image

To analyze the relative amounts of AOMADS1/AOM1 and AOMADS2/AOM4 transcripts, RT-PCR RFLP analyses were performed using cDNA pools from male and female flower buds from A. officinalis. The results showed that there was no significant difference in the amounts of AOMADS1/AOM1 and AOMADS2/AOM4 transcripts in male and female flower buds (Fig. 7).

Fig. 7
figure 7

RT-PCR RFLP analysis of AOMADS1 and AOMADS2. a The schema of expected restriction sites for distinguishing AOMADS1 and AOMADS2 by RT-PCR RFLP. b RT-PCR RFLP profiles of AOMADS1 and AOMADS2 in A. officinalis. Lane 1 male flower buds; lane 2 female flower buds; lane 3 cDNA of AOMADS1, lane 4 cDNA mixture of AOMADS1 and AOMADS2 (1:1); lane 5 cDNA of AOMADS2

Full size image

Discussion

SEP-like genes have highly conserved C-terminal motifs

Recent observations of a wide range of MADS-box genes have the potential to illuminate many of the developmental processes that underlie plant organ development (Alvarez-Buylla et al. 2000; Theissen et al. 2000). Moreover, alignment studies of various angiosperm MADS-box genes have revealed that some lineages of the MADS-box gene families contain highly conserved motifs (Kramer et al. 1998, 2004; Vandenbussche et al. 2003a; Yun et al. 2004b). For instance, Kramer et al. (1998) demonstrated that the GLO-like and DEF-like genes (B-class genes) encode the characteristic amino acid motifs of each lineage: the paleoAP3 motif, euAP3 motif, PI motif-derived, and PI motif. Moreover, Kramer et al. (2004) defined two characteristic amino acid motifs in the C-terminal regions of AGAMOUS (AG)-like genes, AG motifs I and II. In AG-like genes, Yun et al. (2004b) reported that nine amino acid residues following AG motif II were conserved among previous published sequences of various monocots, and termed it the MD motif. Furthermore, from the alignment of the C-terminal regions of many MIKC-type MADS-box genes from a wide range of angiosperms, representing all major gene families, Vandenbussche et al. (2003a) reported that most lineages of MADS-box gene family contained several conserved motifs in their C-terminal regions. These results indicate that members of the same MADS-box gene subfamily usually contain highly conserved C-terminal motifs.

Conserved C-terminal motifs in the SEP-like genes have been reported as the “AGL2 motif,” “ZMM3 motif,” and “ZMM7 motif” (Vandenbussche et al. 2003a). Of these three motifs, ZMM3 and ZMM7 motifs are conserved in four grass species: Hordeum vulgare, Oryza sativa, Triticum aestivum, and Zea mays (Vandenbussche et al. 2003a). Recently, Zahn et al. (2005) have also reported the conserved C-terminal motifs in SEP-like genes, SEP I motif and SEP II motif. We compared the previously published sequences of SEP-like and AGL6-like genes from a large range of seed plants, including A. officinalis and A. virgatus (Fig. 1). The result is that amino acid residues in the C-terminal region are highly conserved among SEP-like genes in seed plants. It is very interesting to note that a part of the SEP I motif (Zahn et al. 2005) is well conserved among the AGL6-like genes as well (Fig. 1), so that we termed this region as the “SEP/AGL6 motif.” The eight amino acid residues before the sequence of the SEP/AGL6 motif, which is a part of SEP I motif (Zahn et al. 2005), was conserved among previously published sequences of SEP-like genes in seed plants, therefore, we termed this region as “SEP motif I.” As these motifs are conserved among more distantly related species, they may play important roles in determining partner specificity in higher-order complex formation or that they contain activation domains. However, the SEP/AGL6 motif is conserved not only in SEP-like genes, but also in AGL6-like genes, so the specialized functions of SEP-like genes are not characterized solely by the SEP/AGL6 motif. We found another conserved motif, termed as “SEP motif II,” downstream of the SEP/AGL6 motif, which was previously called SEP II motif (Zahn et al. 2005). The SEP motif I is more highly conserved than the SEP motif II, suggesting that the SEP motif II may have additional functions and that the SEP motif I involves in the original functions of the SEP-like genes.

There are a variety of roles for the predicted C-terminal sequence motifs in plants. For example, the first half of the C-terminal regions of DEFICIENS and GLOBOSA in Antirrhinum majus appears to be essential for in vitro ternary complex formation with the SQUAMOSA protein (Egea-Cortines et al. 1999). In addition, Lamb and Irish (2003) reported that the PI motif in the C-terminal of the PISTILLATA in Arabidopsis is necessary for the specification of organ identity, although the PI motif was positioned outside the functional MADS-domains and K-domains. In the HOX ortholog proteins, which are the most important transcriptional factors involved in determining the body plan in animals, the gain or loss of the QA motif in the C-terminal region of Ultrabithorax (Ubx) contributes to the evolution of hexapod body patterns in animals (Galant and Carroll 2002; Ronshaugen et al. 2002). Thus, specific protein sequences outside the MADS-domain of MADS-box genes and the homeodomain of HOX ortholog genes are linked to the generation of diversity in the plant and the insect body plans (Galant and Carroll 2002; Levine 2002; Ronshaugen et al. 2002; Lamb and Irish 2003). In the view of these previous results, the well-conserved C-terminal SEP-motif I, SEP/AGL6-motif, and SEP-motif II identified in our study are particularly intriguing, as they may play a role in generating floral diversity. Moreover, the conservation of sequence specificity in SEP motif I, SEP/AGL6 motif, and SEP motif II may reflect a similar conservation of biochemical and/or functional activity. Additional studies using sequences that lack C-terminal SEP motif I, the SEP/AGL6 motif, and SEP motif II are needed in order to examine the role of these motifs in complex formation.

Perspectives on the functional evolution of SEP-like genes in Asparagus

In plants, gene duplication and divergence of the MADS-box genes has likely played a fundamental role in the elaboration of the plant body plan (Theissen et al. 2000). Moreover, the MADS-box genes have undergone a significant amount of gene duplication, and it is the increased number of MADS-box genes, as well as their recruitment to new roles, that has likely contributed to the evolution of new plant morphologies (Theissen et al. 2000). SEP-like genes are a separate subfamily of the MADS-box gene family (Becker and Theissen 2003). Ermolaeva et al. (2003) indicated that SEP1 and SEP2 have recently undergone duplication in the evolution of flowering plants, but their relationships to SEP3 and SEP4 were not clear. Some studies have suggested that SEP1 and SEP2 were more closely related to SEP4 than to SEP3 (Yu and Goh 2000; Lemmetyinen et al. 2004; Zahn et al. 2005), whereas others suggested that SEP3 was the closest relative of SEP1 and SEP2 (Purugganan 1998; Lawton-Rauh et al. 2000; Sung et al. 2000; Becker and Theissen 2003; Parenicova et al. 2003; Vandenbussche et al. 2003b). Our phylogenetic analysis by the NJ method supports the latter hypothesis, showing that SEP3 is more closely related to SEP1 and SEP2 (Fig. 2), whereas the other analysis by the ML method showed that the phylogenetic relationship of these four genes are not clear (Fig. 3).

In A. officinalis, Caporali et al. (2000) and Losa et al. (2004) suggested that the expression profiles of AOM1, AOM3, and AOM4 were similar to those of the SEP genes in Arabidopsis thaliana. Moreover, AOM1 and AOM4 could not be distinguished based on in situ hybridization analyses (Caporali et al. 2000; Losa et al. 2004). We isolated AOMADS1 and AOMADS2, which seem to be the same as AOM1 and AOM4, respectively. The results of our expression analysis of AOMADS1/AOM1 and AOMADS2/AOM4 indicated that these genes were expressed in the flower buds (Fig. 5), and there was no significant difference in the amounts of AOMADS1/AOM1 and AOMADS2/AOM4 transcripts (Fig. 7b). As the expression patterns and the amount of the transcripts of these genes were very similar, these genes could be redundant. Our phylogenetic analysis showed that AOMADS1/AOM1 and AOMADS2/AOM4 were related to Arabidopsis SEPALLATA genes (Figs. 2, 3), suggesting that AOMADS1/AOM1/AVMADS1 and AOMADS2/AOM4/AVMADS2 may be the functional homologues to Arabidopsis SEPALLATA genes. In addition to AOMADS1/AOM1/AVMADS1 and AOMADS2/AOM4/AVMADS2, we isolated two additional SEP-like genes, AOMADS3 from A. officinalis and AVMADS3 from A. virgatus. Although AOMADS1/AOM1/AVMADS1 and AOMADS2/AOM4/AVMADS2 are closely related to SEP3 gene, which were also described in the previous studies (Zahn et al. 2005; Malcomber and Kellogg 2005), AOMADS3 and AVMADS3 were classified in different clade which is far related to SEP3 gene (Figs. 2, 3). The alignment of the amino acid sequences of SEP-like genes showed that AOMADS3 and AVMADS3 genes lacked the SEP-motif II in their C-terminal sequences (Fig. 1). It is interesting to note that in the group of SEP-like genes that included AOMADS3, AVMADS3, and sister groups, all had either lost or had a poorly conserved SEP-motif II in their C-terminus (Fig. 1). Our expression analyses of SEP-like genes in A. officinalis revealed that AOMADS1/AOM1 and AOMADS2/AOM4 were expressed at equal levels in the flower buds (Figs. 5, 7) and that the amount of AOMADS3 transcript was lower than that of AOMADS1/AOM1 and AOMADS2/AOM4 (Figs. 5, 6). Therefore, it seems that AOMADS3 may have undergone a functional shift from the original function of AOMADS1/AOM1 and AOMADS2/AOM4. AOMADS3 would have any functions for floral development in asparagus because this gene specifically expressed in flower buds, although the expression level is very weak.

Expression analyses of AOMADS1/AOM1 and AOMADS2/AOM4 revealed that there was no difference in the level of expression of each gene (Fig. 7; Caporali et al. 2000; Losa et al. 2004). As these two genes appear to be the products of a recent duplication event (Figs. 2, 3), each gene may not have accumulated enough mutational changes to result in a change in their functions. However, a recent study on Arabidopsis thaliana indicated the presence of some nucleotide sites in the MADS-domains and K-domains that had high probabilities of positive Darwinian selection. Their results indicated that these sites play important roles in the acquisition of novel functions (Martinez-Castilla and Alvarez-Buylla 2003). Therefore, it is likely that functional changes in AOMADS1/AOM1 and AOMADS2/AOM4 of A. officinalis occur as a result of changes at a few sites in each gene and may be a major source of innovation in floral development. Further comparative analyses using transgenic plants will illuminate the functional evolutionary histories of not only AOMADS1/AOM1 and AOMADS2/AOM4 but also AOMADS3 of A. officinalis.