Introduction

The determination and differentiation of floral organs are regulated mainly by members of the MADS-box gene family, which encode transcription factors and are characterized by the well-conserved MADS domain (Shore and Sharrocks 1995). MADS-box genes are phylogenetically divided into two types, I and II, which probably diverged before divergence of plants and metazoans (Alvarez-Buylla et al. 2000). Characteristic of type II MADS-box genes of land plants is the domain structure composed of four regions: MADS (M), the internal region between MADS and K (I), K (K), and the C-terminal region (C) (Ma et al. 1991). The K domain, the next best-conserved domain after MADS, has been reported only from land plant MADS-box genes (Shore and Sharrocks 1995). Comparison of exon-intron structures has revealed two different groups of type II MADS-box genes: MIKCc and MIKC* (Henschel et al. 2002). All floral homeotic genes are MIKCc. MIKCc genes exist in a wide range of land plants including flowering plants and mosses. MIKC* genes are found in clubmosses and mosses (Svensson et al. 2000; Henschel et al. 2002), but the function of MIKC* genes has not been determined fully. MIKCc genes play important roles not only in floral organ development but also in other developmental processes in sporophytes, such as the switch to flowering (Hartmann et al. 2000; Samach et al. 2000), lateral root differentiation (Zhang and Forde 1998), dehiscence of fruits, siliques (Gu et al. 1998), and others (Theißen et al. 2000).

Hypotheses on the origin and evolution of flowering plant MADS-box genes with such critical functions are based on studies of MADS-box genes in gymnosperms (Shindo et al. 1999; Winter et al. 1999) and ferns (Münster et al. 1997; Hasebe et al. 1998). All previously examined MADS-box genes of the fern Ceratopteris richardii are widely expressed in both vegetative and reproductive organs, and do not diverge as in flowering plants. This suggests that recruitment of some MADS-box genes expressed and functioning in specific organs occurred in the flowering plant lineage, which was probably important for the evolution of elaborate plant body plans (Hasebe 1999; Theißen et al. 2000). Characterization of MADS-box genes in other lower land plants is important for testing this hypothesis. Cladistic analyses of morphological data and molecular phylogenetic analyses of vascular plants indicate that lycophytes are the most basal lineage in extant vascular plants (Raubeson and Jansen 1992; Hiesel et al. 1994; Kranz et al. 1995; Kenrick and Crane 1997; Daff and Nickrent 1999; Pryer et al. 2001). In this present study, a MADS-box gene of the lycophyte spikemoss Selaginella remotifolia was cloned and its expression pattern was analyzed in order to determine the evolution of MADS-box genes and plant body plans.

Materials and methods

Selaginella remotifolia Spring was collected at Showanomori Park in Chiba Prefecture, Japan. The samples were divided into various tissues and preserved in a deep freezer. Total RNA extraction and 3' and 5' RACE were performed as described by Shindo et al. (1999). The nested MADS domain-specific primers, duMADS2–2 (5′-{CAU}4AARAARGCITAYGARCTIAGYGT-3′) and AllMADS2 (5′-{CAU}4GARYTIWSIGTIYTITGYGAYGC-3′), were used to amplify the MADS-box clone.

To perform RT-PCR expression analysis, complementary DNAs were synthesized from total RNAs extracted from apices in the vegetative stage, strobili, microphylls, stems, and a mixture of rhizophores and roots. The PCR conditions were 1 cycle at 94°C for 1 min, followed by 30 cycles of 94°C for 1 min, 52°C for 1 min, and 72°C for 1.5 min, and a final step at 72°C for 5 min. A PCR amplification test was performed with the SrMADS1-specific internal primers SrMF1 (5′-TCTAAACAAGCAAAAGGAACCGCC-3′) and SrMR1 (5′-CGAAGCATCTCATTGTCCTTGTG-3′). The Selaginella remotifolia ortholog of the 6-phosphogluconate dehydrogenase gene Sr6PGD (AB086022), which was constitutively expressed in all tissues examined, was used as a positive control. Sr6PGD was amplified by PCR with the Sr6PGD-specific primers (5′-TTTCTGAGCGGACTCAAGGAGG-3′) and (5′-TAAGTGTGGGCACCAAAATAGTC-3′).

To construct a phylogenetic tree for MIKC-type MADS-box genes, amino acid sequences were obtained from EMBL/DDBJ/GenBank DNA databases and aligned using Clustal W, version 1.6 (Thompson et al. 1994). The maximum likelihood (ML) distances were calculated with ProtML under the conditions of the JTT model (Jones et al. 1992) and a Neighbor-Joining (NJ) tree was obtained with Njdist (Adachi and Hasegawa 1992–1996). The tree was further analyzed by local rearrangement search using ProtML to obtain the ML tree.

Results and discussion

A cDNA of SrMADS1 (AB086021), a MADS-box gene, was isolated from Selaginella remotifolia by 3′ and 5′ RACE using MADS domain-specific degenerate primers. Other MADS-box genes were not found, despite the use of three other primers and various PCR conditions. Start codons exist in most MIKC genes adjacent to the N-terminal of MADS domains, but putative start codons were not found between 126 bp of the 5′ region to the first nucleotide of the MADS-box of SrMADS1 (Fig. 1a). Comparison of deduced amino acid sequences of SrMADS1 with other MIKCc and MIKC* genes shows SrMADS1 to be a typical MIKCc-type MADS-box gene, lacking an extended I region (Fig. 1b).

Fig. 1a, b.
figure 1

Cloning of the SrMADS1 gene. a The structure of SrMADS1 cDNA. The symbol S and polyA indicate the stop codon and poly(A)+tail, respectively. The closed rectangle on the architecture represents MADS-box and the open rectangle indicates a K-box. b Alignment of deduced amino acid sequences of SrMADS1 and representative MIKC genes of land plants: LAMB1, 2, 4, and 6 (Lycopodium), PPM1, 3, and 4, PpMADS2 and 3 (Physcomitrella), CMADS1 (Ceratopteris), and AGAMOUS and APETALA1 (Arabidopsis). Amino acid positions indicated by bars were used for the phylogenetic analysis

To examine the phylogenetic relationship of SrMADS1 to other MADS-box genes, an ML tree was constructed using representative MADS-box sequences from dicots, monocots, gymnosperms, leptosporangiate ferns, lycophytes, and mosses. LAMB26, which are MIKCc genes of Lycopodium (a lycophyte clubmoss), form a clade as the LAMB2 group and LAMB3 and 5 closely related to LAMB4 and 6, respectively (Svensson and Engström 2002). LAMB3 and 5 were excluded from our analysis because they lack the K-box (Svensson and Engström 2002). As shown by the ML tree, SrMADS1 forms a clade with LAMB2, 4, and 6 with a high bootstrap value, and SrMADS1 is not closely related to gene(s) of the LAMB2 group (Fig. 2). Moreover, this tree indicates that SrMADS1 may be the sister of the LAMB2 group; the ML tree supports this topology with a bootstrap value of 80%, but the NJ tree does not support this topology (Fig. 2). Unlike SrMADS1, genes of the LAMB2 group do not possess additional amino acids in the N-terminal of the MADS domain . Some genes of the AG group and all genes of the CRM6 group have similar additional amino acids in the N-terminal of the MADS domains, but SrMADS1 did not cluster with AG and CRM6 clades in the phylogenetic tree (Fig. 2). Thus SrMADS1 appears to have obtained the N-terminal region originally in the lineage of SrMADS1. The expression pattern of SrMADS1 was assessed by RT-PCR using total RNA extracted from vegetative shoot tips, strobili, microphylls, stems without leaves, and rhizophores with roots. No amplification was detected in the roots or rhizophores, though we were unable to separate these (Fig. 3). Accumulation of SrMADS1 mRNA in gametophytic tissues is unknown, but SrMADS1 transcripts preferentially accumulate in whole sporophytic tissues, except in roots and rhizophores. LAMB2, 4, 5, and 6 are also expressed in a broad range of sporophytic organs of Lycopodium, though the expression pattern of LAMB3 has not been reported (Svensson and Engström 2002). The expression patterns of SrMADS1 and LAMB2, 4, 5, and 6 are similar, except that LAMB2, 4, 5, and 6 are expressed in the roots but SrMADS1 is not. The difference between these expression patterns and the results of our phylogenetic analysis suggest that the root of Selaginella was not derived from the same common ancestral organ as Lycopodium, although further analyses of root development processes in lycophytes are necessary to further test this hypothesis.

Fig. 2.
figure 2

A gene tree of MIKC-type MADS-box genes based on the maximum likelihood method. The closed circles indicate genes from lycophytes (Selaginella remotifolia and Lycopodium annotinum). Open square, square filled with an angle bar, shaded square, shaded circle, and open triangle represent dicots (Arabidopsis thaliana and Lycopersicon esculentum), monocots (Oryza sativa), gymnosperms (Gnetum gnemon, G. parvifolium, Picea abies, and Pinus radiata), ferns (Ceratopteris richardii), and mosses (Physcomitrella patens), respectively. The aligned 130 amino acid residues correspond to positions 2–58 (MADS domain), 63–70 (I region), 93–157 (K domain) from the initial methionine of APETALA1. Bootstrap values calculated with RELL method are represented above the nodes and those calculated with the NJ method are shown below. Only values over 85% with both ML and NJ are indicated

Fig. 3.
figure 3

RT-PCR analysis of SrMADS1 gene expression in sporophytic tissues. Complementary DNAs were synthesized from RNAs extracted from apices in the vegetative stage (lane 1), strobili (lane 2), microphylls (lane 3), stems (lane 4) and a mixture of roots and rhizophores (lane 5). The Sr6PGD gene was used as a positive control

Previous authors have suggested that the generally expressed fern MIKCc genes are more primitive than the MIKCc genes specifically expressed in reproductive organs (Münster et al. 1997; Hasebe et al. 1998). Our results as well as those with the MIKCc genes of Lycopodium (Svensson and Engström 2002) support this hypothesis. Lycophytes appeared during the early Devonian, approximately 400 million years ago, together with the extinct rhyniophytes and zosterophylls, which had dichotomously branching stems (Kenrick and Crane 1997) and form a sister to the clade including seed plants and ferns. Both Selaginella and Lycopodium are placed in the lycophytes, but these two genera are phylogenetically distant from each other (Kenrick and Crane 1997). Our results and those of Svensson and Engström (2002) suggest that the ancestral MIKCc gene of the primitive dichotomous plants in the early Devonian was involved in the development of basic sporophytic tissues such as shoot, stem, and sporangium.