A PRIMER SET FOR SPECIFIC AMPLIFICATION OF TWO CYCLOIDEA-LIKE GENES IN THE GENISTOID CLADE OF LEGUMINOSAE SUBFAM.

Locus-specific primers were designed to amplify the homologues LEGCYC1A and LEGCYC1B of the snapdragon ( Antirrhinum majus ) floral symmetry gene CYCLOIDEA in the genistoid legumes Lupinus nanus and Cadia purpurea . These primers were tested successfully in a range of genistoid taxa. Sequence comparison between L. nanus and L. angustifolius revealed that these CYCLOIDEA-like genes can provide up to four times as many variable characters as the ribosomal internal transcribed spacers (ITS) at this taxonomic level. These genes are therefore potentially useful for studying evolutionary relationships between closely related species within the genistoid clade of legumes, which contains many large and diverse genera. Comparisons between L. nanus , Cadia purpurea and Calpurnia aurea LEGCYC sequences show that these genes may also be useful for phylogenetic studies between genera in this large clade of 1800 species.


INTRODUCTION
The most commonly used nuclear regions for phylogenetic analysis at low taxonomic levels are the internal transcribed spacers (ITS) of the nuclear 18S-5.8S-26S ribosomal DNA gene family (Alvarez & Wendel, 2003). There is a need, however, for rapidly evolving low-copy nuclear genes in systematic studies (e.g. Mort & Crawford, 2004). Multiple sources of informative molecular data are required for testing the congruence of topologies of different gene trees in order to have more reliable estimates of taxic relationships, or to investigate hybridization events (Doyle, 1992). In addition, the nature of the ITS region has certain limitations, being part of a multigene family that is homogenized through concerted evolution. Reports of incomplete concerted evolution or pseudogene evolution in this gene family suggest that sequencing of ITS may be subject to complicating factors (Alvarez & Wendel, 2003;Bailey et al., 2003). ITS divergence between closely related taxa may also be too small to resolve relationships, in part due to the short length of the ITS region (c.450 bp), and to the homogenizing effect of concerted evolution (reviewed in Hershkovitz et al., 1999). For instance, species relationships within the genistoid genus Lupinus L. could not be well resolved using ITS (Ainouche & Bayer, 1999). Molecular data from single-copy nuclear genes providing more variable characters are therefore needed to resolve rapid radiations at the species level.
CYCLOIDEA (CYC) is a rapidly evolving transcription factor that controls floral symmetry in Antirrhinum majus L. (Veronicaceae, Lamiales;Luo et al., 1996;Gübitz et al., 2003). Homologues of CYC have been isolated in the Leguminosae (Citerne et al., 2003), a family that has evolved highly specialized zygomorphic (bilaterally symmetric) flowers independently of the Lamiales. Phylogenetic analyses suggest that legume CYC (LEGCYC) genes have duplicated at least three times in the evolution of subfamily Papilionoideae (Citerne et al., 2003). These genes were found to be evolving rapidly by substitution, insertions and deletions so that unambiguous alignment of sequences in the variable regions outside the conserved TCP and R domains was possible only between closely related taxa (Citerne et al., 2003).
As part of a study to investigate whether CYC-like genes are involved in the control of floral symmetry in legumes, locus-specific primers were designed to amplify two paralogous CYC-like genes, LEGCYC1A and LEGCYC1B (Citerne et al., 2003), in closely related species belonging to the core genistoid clade (sensu Wojciechowski, 2003) with contrasting floral morphology: Lupinus nanus Doug. ex Benth., which has zygomorphic flowers characteristic of the Papilionoideae, and Cadia purpurea (Picc.) Aiton, a species with atypical near-actinomorphic flowers.
This paper describes these primers, which amplify CYC-like genes, and their possible use in phylogenetic studies within the core genistoid clade. This large clade, defined from analyses of molecular data as comprising seven different tribes, some of which were previously thought to be unrelated, is reviewed in Wojciechowski (2003). Despite considerable work on members of this group, relationships between certain genera or between species which have undergone rapid diversification are still unclear (Wojciechowski, 2003). These primers should be useful for molecular systematic studies in this clade, which contains many large genera such as Crotalaria L. (c.600 species), Aspalathus L. (c.250 species), Genista L. (c.90 species) and economically important genera such as Lupinus (c.250 species), Sophora L. and Ulex L.

MATERIALS A N D M E T H O D S
Genomic DNA of Cadia purpurea and Lupinus nanus was extracted as described in Citerne et al. (2003). The open reading frames (ORF) of LEGCYC1A and LEGCYC1B in C. purpurea and L. nanus were isolated by genome walking following a modified protocol of Siebert et al. (1995) [GenBank accession numbers: L. nanus LEGCYC1A: AY382156; L. nanus LEGCYC1B: AY382155; C. purpurea LEGCYC1A: AY225826; C. purpurea LEGCYC1B: AY225825]. An array of primers, described here, was then designed based on these sequences specifically to amplify part of each locus (Table 1, Fig. 1). PCR amplifications were carried out on L. nanus and C. purpurea DNA using Bioline Taq and reagents (Bioline, London, TABLE 1. Primers for amplification of LEGCYC1A and LEGCYC1B in genistoid legumes, with predicted annealing temperature in °C (T an ).
Location of priming sites is given in Fig. 1 Primer name UK). The reaction mix contained 2.5µl of 50mM MgCl 2 , 5µl of a 2mM dNTP mix, 10µM made up of 2.5µl of each primer (MWG Biotech, Ebersberg, Germany), 1 unit of BIOTAQ, 10-20ng of genomic DNA, and dH 2 O to make up 50µl. Cycling conditions consisted of an initial denaturation step at 94°C (3 min), followed by 30 cycles of denaturation at 94°C (1 min), annealing at 50-55°C (30 s), and extension at 72°C (1 min), followed by a final extension step at 72°C (5 min), using a DNA Engine thermal cycler (MJ Research, Inc., Waltham, MA, USA). PCR products were purified using the QIAquick PCR Purification Kit (Qiagen Ltd, Dorking, Surrey, UK) and sequenced directly using the primers described here. Dye-terminator cycle sequencing was carried out using Thermosequenase II (Amersham Pharmacia, Buckinghamshire, UK). Samples were analysed on an ABI 377 Prism Automatic DNA Sequencer (Perkin Elmer, Applied Biosystems Division, Foster City, CA, USA).
In addition, the primers described here were tested in a range of taxa, including a putative sister taxon to Cadia, Calpurnia aurea (Aiton) Benth., which has typical zygomorphic papilionoid flowers, and in taxa belonging to the sister lineage of the F IG . 1. Schematic representation of LEGCYC1 (A and B) showing the approximate location of primers described in Table 1. Black areas represent conserved domains (TCP and R) characteristic of CYC-like genes, and the hypervariable regions are paler. Forward and reverse locus-specific primers binding to a region between the TCP and R domains are complementary. core genistoid clade: Ormosia amazonica Ducke, Acosmium subelegans (Mohl.) Yakovlev and Bowdichia virgiloides Kunth (Pennington et al., 2001;Wojciechowski, 2003). Calpurnia aurea LEGCYC1A and LEGCYC1B were sequenced directly in two parts using primers LEGCYC1-F2 and LEGCYC1A-R1/LEGCYC1B-R1, and LEGCYC1A-F1/LEGCYC1B-F1 and LEGCYC1-R1 [GenBank accession numbers: DQ333954, LEGCYC1A 5p; DQ333955, LEGCYC1A 3p; DQ333956, LEGCYC1B 5p; DQ333957, LEGCYC1B 3p]. Genomic DNA of C. aurea was provided by Matt Lavin (Montana State University) and of O. amazonica, A. subelegans and B. virgiloides by Toby Pennington (Royal Botanic Garden Edinburgh).

Potential of LEGCYC genes for phylogenetics
Pairwise sequence comparison of the coding regions of LEGCYC homologues in L. nanus and L. angustifolius revealed that the two copies are diverging rapidly: 8.49% nucleotide sequence divergence between LEGCYC1A orthologues and 5.34% nucleotide sequence divergence between LEGCYC1B orthologues. Sequence comparisons including total number of substitutions are summarized in Table 2. Comparison of pairwise divergence from the nuclear ribosomal DNA ITS (3.70% sequence divergence between L. nanus and L. angustifolius ITS1+ITS2; Table 2) suggests that LEGCYC1A and LEGCYC1B sequences can provide up to four times as many variable characters as ITS at this taxonomic level. Although the level of nucleotide substitutions suggests that LEGCYC1A is evolving more rapidly than LEGCYC1B, sequence alignment between L. nanus and L. angustifolius homologues required the insertion of more gaps between LEGCYC1B than LEGCYC1A orthologues (7 and 4 gaps, respectively; Table 2). Most of these are trinucleotide repeats between LEGCYC1B orthologues, whereas there is a large indel (30 bp) between LEGCYC1A orthologues. The short intron region at the 3p end of the gene was alignable between the two Lupinus species, providing further characters potentially useful for phylogenetic analyses. The intron region of LEGCYC1A and LEGCYC1B had a similar number of substitutions (4 substitutions over 79 bp versus 3 substitutions over 92 bp, respectively), but only LEGCYC1A had indels (3 indels of 1-31 bp).
Sequence comparison between closely related genera of the genistoid clade such as Cadia and Lupinus suggested that the level of variation between LEGCYC orthologues is also phylogenetically informative at this taxonomic level. Nucleotide pairwise distances in C. purpurea and L. nanus were greater between LEGCYC1A than LEGCYC1B orthologues (17.53% versus 13.29% nucleotide sequence divergence, respectively; see Table 2). Proportionally four times more gaps (17 for LEGCYC1A and 30 for LEGCYC1B; Table 2) were required for alignment of C. purpurea and L. nanus sequence pairs than in the comparison of Lupinus species.
Despite these numerous insertion-deletion events, sequence alignment was for the most part unambiguous (details are available from the author). That these results are unrelated to the difference in floral symmetry between L. nanus (zygomorphic) and C. purpurea (actinomorphic) is borne out by comparison between L. nanus LEGCYC sequences and their orthologues in the zygomorphic putative sister species of Cadia purpurea, Calpurnia aurea, providing similar values to those described above (15.12% and 12.02% nucleotide sequence divergence between LEGCYC1A and LEGCYC1B orthologues, respectively; Table 2).
To evaluate the taxonomic range of the primers described here, some primer combinations were tested on genomic DNA from taxa in the sister group of the core clade: Ormosia amazonica, Acosmium subelegans and Bowdichia virgiloides. A single band of the expected size was amplified, with some exceptions reflecting the rapid rate of evolution of these genes, suggesting that these primers may be particularly useful within the core genistoids and their sister group, but may be less effective in a wider range of papilionoid legumes. In particular, the general primers LEGCYC1-F2, LEGCYC1-R1 were found to be most reliable in Ormosia, Acosmium and Bowdichia. The high level of variation within LEGCYC copies does however imply that, although some of these primers were found to work in taxa sister to the core genistoid clade, they may not work in certain lineages even within this group. Nevertheless, these primers are likely to be useful for phylogenetic analysis of many large genera, particularly those in the range of tribes containing Lupinus (Genisteae s.s.) and Cadia/Calpurnia (Podalyriae), as defined by current phylogenies (Wojciechowski, 2003). The phylogenetic potential of LEGCYC1A and LEGCYC1B has been confirmed since the submission of this paper by Ree et al. (2004) who successfully used the primers described here to generate a phylogeny of 15 Lupinus species. The phylogenetic utility of CYC genes in the genistoid clade mirrors their use as a variable phylogenetic marker in other families such as Gesneriaceae (Smith et al., 2004;Wang et al., 2004).

Other applications
Regulatory genes, such as transcription factors, are believed to play an important role in the control of phenotypic diversity that distinguishes species (reviewed in Cronk, 2001). The study of natural variation of regulatory genes within populations or between closely related species is fundamental for understanding the processes of selection acting on these genes (Purugganan, 2000). CYC-like genes are suspected of being involved in the control of floral symmetry in this clade of the Leguminosae (Citerne, unpublished). Although most members of the genistoid clade have typical zygomorphic papilionoid flowers, some genera such as Cadia and Acosmium have more open, actinomorphic (radially symmetric) flowers. This provides a good framework for studying developmental gene evolution in relation to changes in a related morphological trait, in this case floral symmetry.
In addition, these genes may be useful molecular markers at the population level. Single nucleotide polymorphisms (SNPs) in a CYC-like gene were found to be informative for studying population structure in Ramonda myconi (L.) Reichb. (Gesneriaceae) (Pico et al., 2002). The high rate of evolution described here between Lupinus species suggests that SNPs or trinucleotide repeats between CYC alleles could be used as population markers in legumes.

A C K N O W L E D G E M E N T S
I would like to thank Toby Pennington and Quentin Cronk for comments on the manuscript. Laboratory assistance was provided by Michelle Hollingsworth and Alex Clark. This work was part of a project funded by the Carnegie Trust for the Universities of Scotland.