Lateral organ growth in seed plants is controlled in part by members of the YABBY (YAB) and class III homeodomain/leucine zipper (HD-ZIPIII) families of transcription factors. HD-ZIPIII genes appear to play a conserved role in such organs, but YAB genes have diversified, with some members of the family having specialized functions in leaves, carpels or ovule integuments. The ancestral expression patterns and timing of divergence of the various classes of YAB genes remain to be established. We isolated and evaluated the expression of one HD-ZIPIII and five YAB genes representing the five major YAB gene classes from Cabomba caroliniana, a member of the earliest-diverging angiosperms. Consistent with observations in eudicots, the FILAMENTOUS FLOWER (FIL) and YABBY5 (YAB5) genes of C. caroliniana were expressed in the abaxial regions of the leaf where new laminar segments arise, and the patterns of expression were mutually exclusive to those of HD-ZIPIII, indicating that these expression patterns are ancestral. Expression of CRABS CLAW (CRC) in the abaxial carpel wall, and of INNER NO OUTER (INO) in the abaxial outer integument of ovules was also conserved between eudicots and C. caroliniana, indicating that these patterns are primitive. However, the CRC gene was also expressed in other floral organs in C. caroliniana, and expression in stamens was also observed in another early-diverging species, Amborella trichopoda, indicating that carpel-specific expression was acquired after divergence of the Nymphaeales. The expression data and phylogeny for YAB genes suggest that the ancestral YAB gene was expressed in proliferating tissues of lateral organs.
Whereas most eudicot YAB genes are expressed in the abaxial tissues of lateral organs of the shoot, some YAB genes of the Poaceae (grass family) are expressed adaxially (Juarez et al., 2004a,b) or throughout lateral organs (Zhao et al., 2006; Dai et al., 2007). These contrasting expression patterns may be attributed to difference in regulatory pathways: HD-ZIPIII and trans-acting small interfering RNA (ta-siRNA) positively regulate expression of poaceous YAB genes (Juarez et al., 2004a,b), but repress expression of these genes in Arabidopsis (Li et al., 2005; Garcia et al., 2006; Kidner and Timmermans, 2007). These data imply that mechanisms regulating laminar expansion have diverged during angiosperm diversification.
Studies on the eudicots and grasses increasingly suggest that YAB genes are involved in the origin of the leaf, as well as in subsequent generation of the carpel and outer integument by modification of the sporophyll. These possibilities should be further tested by examination of early-diverging angiosperms and gymnosperms. Such data are also critical to trace the evolutionary course of the YAB genes, but comprehensive data are not available for the early-diverging angiosperms and gymnosperms. The present incomplete data on CRC and INO orthologs in the early-diverging angiosperms do not appear to support primordial roles in the origin of the carpel and outer integument. A CRC ortholog of Amborella trichopoda (Amborellaceae) is expressed in the stamen as well as the abaxial tissue of the carpel (Fourquin et al., 2005). Similarly, expression of an INO ortholog in the abaxial tissue of the outer integument was found in the early-diverging Nymphaea alba (Nymphaeaceae), but expression was also observed in the inner integument and nucellar tip (Yamada et al., 2003). However, these more extensive expression patterns would be primitive only if these patterns are not autapomorphies of this single early-diverging species.
In this study, we isolated and performed expression analyses on one HD-ZIPIII and five YAB genes of Cabomba caroliniana A. Gray (Figure 1a) to elucidate the evolutionary history of YAB genes in angiosperms. The genus Cabomba belongs to the Cabombaceae family of the order Nymphaeales, which, together with Amborellales and Austrobaileyales, constitute the first diverging angiosperm lineages, referred to as the ANITA grade (Saarela et al., 2007; Angiosperm Phylogeny Group, 2009). The genus Cabomba is suitable for analyses on YAB genes not only because of its basal phylogenetic position in angiosperms, but also because of its unique leaf (Figure 1b–h), which is described in more detail below.
Phylogenetic relationships among five sub-families of angiosperm YAB genes
We aligned proteins predicted from YAB genes isolated from C. caroliniana with known YAB proteins and used the alignment for phylogenetic analysis (Figure S1). As YAB genes are characteristic of seed plants (Nishiyama et al., 2003; Floyd and Bowman, 2007), gymnosperm gene(s) could be the first diverging YAB genes. However, the gymnosperm genes partitioned into two clades, thus the resulting tree could not be rooted using the gymnosperm YAB genes (Figure 2).
Each C. caroliniana YAB protein was positioned at or near the base of a separate sub-family cluster with more than 70% posterior probability. Thus we identified CcCRC, CcFIL, CcINO, CcYAB5 and CcYAB2 as C. caroliniana orthologs of the Arabidopsis CRC, FIL, INO, YAB5 and YAB2 genes, respectively (Figure 2). This result confirms that the five sub-families were already present in the common ancestor of extant angiosperms (Yamada et al., 2004; Lee et al., 2005b).
The tree suggested close relationships between YAB2 and YAB5 sub-families, with moderate support (Figure 2), as shown in a previously published maximum-likelihood tree using amino acids (Yamada et al., 2004) and a Bayesian tree using nucleotide data (Lee et al., 2005b). Close relationships between INO and CRC sub-families were also suggested with high posterior probability (96%), but this relationship is weakly supported in a maximum-likelihood tree using the Le-Gascuel model of amino acid substitutions (Guindon et al., 2010; Figure S2).
Structures of C. caroliniana and other angiosperm YAB genes
We compared the structures of YAB genes in C. caroliniana, rice (Oryza sativa) and Arabidopsis. All genes possess one intron in a common location in the zinc finger-like domain coding region, three introns in conserved locations in the YAB domain coding region, and one intron in the non-conserved region between the two conserved regions (Figure 3 and Figure S3).
A sixth intron is found in the non-conserved region downstream of the YAB domain in some genes, but this intron is absent from other genes, including C. caroliniana CcCRC (Figure 3 and Figure S3). The sixth intron is consistently absent from all YAB2 sub-family genes whose structure is available (YAB2 of Arabidopsis, and OsYAB1, OsYAB2 and OsYAB6 of rice) (Figure 3a), implying that the absence of this intron is primitive in the YAB2 sub-family. We confirmed that there is no intron in this location in CcYAB2 by amplifying this region from genomic and cDNA and showing that these two sources gave rise to PCR products of identical size (Figure S4).
Cabomba leaf development
Cabomba species exhibit heterophylly: producing both finely dissected submerged leaves and sagittate or peltate floating leaves as adaptations to their aquatic lifestyle (Figure 1a). Submerged leaf primordia emerge sub-oppositely on the flank of the SAM (Figure 1b,c). The primordia are entire at initiation (Figure 1b,d), and actively dividing cells are distributed evenly between the abaxial and marginal areas at this stage (Figure 1d). A lobe primordium with a distinct initial (Figure 1g) forms on the abaxial side near the left or right leaf margin, and proliferation of a group of cells between the main and lateral lobes becomes less active (Figure 1e). By repeating this process (Figure 1f,h), the submerged leaf becomes finely dissected (Turlier, 1984). Therefore, laminar forming cells are easily distinguished from other cells in Cabomba leaves.
A C. caroliniana co-ortholog of PHABULOSA and PHAVOLUTA is expressed in adaxial tissues of lateral organs
Previous studies have indicated that adaxial expression is conserved among HD-ZIPIII genes of seed plants (Floyd and Bowman, 2006; Floyd et al., 2006). We wished to test whether this was also true for leaves of C. caroliniana in order to determine whether adaxial–abaxial polarity is similarly established in lateral organs of this species. We isolated a HD-ZIPIII gene from C. caroliniana (CabC3HDZ1) that is a co-ortholog of Arabidopsis PHABULOSA and PHAVOLUTA (Figure S5 and Figure S6).
CabC3HDZ1 was expressed in the adaxial layers of leaf primordia and in the SAM (Figure 4a,b), but not in marginal and abaxial tissues from which lobe primordia will emerge (Figure 1g,h). Adaxial expression was also found in floral organs (Figure 4c,d). Expression was also detected in procambial strands of stems, the flower axis and leaves. In leaf procambia, the signal was somewhat stronger on the adaxial sides (Figure 4a). Therefore, at least with respect to HD-ZIPIII genes, adaxial–abaxial polarity is established in C. caroliniana as in other characterized seed plants.
CcFIL, CcYAB2 and CcYAB5 are expressed in outgrowing lamina
Expression of CcYAB5 was detected in the abaxial tissues of the distal leaf primordium in longitudinal sections (Figure 5a). The expression was extended to marginal tissues near the apices of lobe primordia (Figure 5b). Strong signal was detected in the lobe primordia emerging from the abaxial and marginal sides of the leaf primordium, but no signal was recognizable in the web-like tissue separating the lobe primordia (Figure 5b). In short, CcYAB5-expressing tissues correspond to actively proliferating cells in the leaf primordium, and the expression of CcYAB5 and CabC3HDZ1 was mutually exclusive (Figures 4a and 5a,b). CcYAB5 was expressed in procambial strands supplied to the leaf primordium, and the signal was centered on their abaxial side (Figure 5a,b). No signal was observed in procambial strands in the stem or flower axis.
In the floral buds, CcYAB5 expression was observed at the tip of sepaloid tepals, petaloid tepals and stamen primordia (Figure 5c). Strong signal was detected in the intruding tissue of the carpel primordium that develops into the placenta (Figure 5d). This signal became weaker as the carpel grew, but was still visible when the ovule primordia initiated (Figure 5e). CcYAB5 was also transcribed in the abaxial tissue of the floating leaf primordium (Figure 5d), whose expansion contributes to formation of its peltate or sagittate shape (Turlier, 1984).
In vegetative shoots, the expression pattern of CcFIL was similar to that of CcYAB5, in that it was expressed in marginal and abaxial tissues of lobe primordia and procambial strands (Figure 5f,g). However, no CcFIL expression was observed in the placental tissue. Instead, CcFIL was expressed in the procambial strands of floral buds, as well as in epidermal and mesophyll tissues of sepaloid and tepaloid tepals (Figure 5h).
CcYAB2 is expressed in the vegetative shoot in similar manner to CcFIL and CcYAB5 (Figure 5i), but no strong signal was obtained in floral buds (Figure 5j). We confirmed this result by semi-quantitative RT-PCR analysis, in which CcYAB2 amplification in floral buds was much weaker than in SAM or in leaves (Figure 6).
CcINO expression is outer integument-specific, but CcCRC is expressed in the carpel and other floral organs
Cabomba ovules have two integuments comprising two cell layers, with the inner integument initiating prior to the outer integument (Yamada et al., 2001). CcINO expression was observed in the outer epidermis of the outer integument from the time of initiation of the outer integument through to the stage when the inner integument has covered the nucellus (Figure 7a,b). No signal above background was observed in other parts of ovules or in other organs at any stage of development (Figure S7). We performed semi-quantitative RT-PCR analysis to confirm this result, but no amplification was observed, even in the floral buds, probably because CcINO-expressing tissues are scarce relative to non-expressing tissues (Figure 6). However, strong amplification was observed in the floral buds when these PCR products were further amplified using nested primers, with much weaker amplification in the SAM (Figure 6).
CcCRC expression was first detected in the petaloid tepal, stamen primordia and floral apex (Figure 7c). Weak expression was also observed in the tip of sepaloid tepal primordia (Figure 7c). CcCRC was expressed in the entire emerging carpel primordium (Figure 7d), but, after formation of the locule was initiated, its expression became limited to the abaxial tissue of the carpel wall (Figure 7e). Expression was decreased when ovule formation was initiated, and was not detected in vegetative organs (Figure 6 and Figure S7).
Expression in outgrowing lamina is an ancestral trait in FIL, YAB2 and YAB5 sub-families
Previous studies provided evidence that FIL and YAB5 genes are involved in leaf laminar outgrowth in eudicots (Sawa et al., 1999; Siegfried et al., 1999; Watanabe and Okada, 2003; Golz et al., 2004; Gleissberg et al., 2005; Sarojam et al., 2010; Yan et al., 2010) and poaceous monocots (Juarez et al., 2004a,b; Zhao et al., 2006; Dai et al., 2007). In Arabidopsis, FIL, YAB2, YAB3 and YAB5 have redundant functions, and fil yab3 yab5 or fil yab2 yab3 yab5 mutants fail to activate programs promoting laminar growth, and some SAM programs are activated in the leaf (Sarojam et al., 2010). The genes GRAMINIFOLIA (GRAM) and PROLONGATA (PROL), which are FIL and YAB5 orthologs, respectively, of Antirrhinum majus (Malpighiaceae, eurosids), have been shown to play a role in promoting laminar outgrowth because their mutants have smaller leaves than those of wild-type due to reduced frequency of cell divisions (Golz et al., 2004). A similar function has been suggested for maize (Poaceae, monocots), because ZmYAB9 and ZmYAB14 are expressed in the ectopically formed blade of the Rolled leaf1-Original mutant (Juarez et al., 2004b). We have shown that the expression domains of CcFIL, CcYAB2 and CcYAB5 correlate with actively proliferating cells of leaf primordia. Therefore, the expression patterns of the FIL, YAB2 and YAB5 sub-families in outgrowing lamina extend back to the common ancestor of extant angiosperms.
Expression in placenta is acquired in the YAB2 and YAB5 common ancestor
Placental expression of a YAB5 gene has not been previously reported, but such an expression pattern is known for a few genes of the YAB2 sub-family (Figure 2). For example, AmbF1 of Amborella trichopoda (Amborellaceae, ANITA grade) is expressed in the placental tissue (Yamada et al., 2004), and FASCICATED, a YAB2 ortholog of tomato (Solanum lycopersicum; Solanaceae, euasterids), is expressed in the placenta, regulating its size (Cong et al., 2008). These reports, as well as the CcYAB5 expression pattern, suggest that the common ancestor of YAB2 and YAB5 genes showed placental expression. Such expression was lost for CcYAB2 and the YAB2 genes of rice (Toriba et al., 2007), Vitis vinifera (Vitaceae, eudicots) (Fernandez et al., 2007) and Arabidopsis (Siegfried et al., 1999).
Based on the hypothesis that a carpel is a modified sporophyll, the placental tissue could be comparable to marginal leaf tissue (e.g. Stebbins, 1974; Endress, 1983, 2001, 2005). The expression of YAB2/YAB5 in placenta is consistent with their participation in outgrowth of modified lamina. However, expression of CcYAB5 in placenta is also contradictory to that in leaf, because CcYAB5 and CabC3HDZ1 expression overlaps in the placenta. This flexibility of CcYAB5 expression provides additional support for its lack of participation in establishment of adaxial–abaxial polarity. CcYAB5 appears to have acquired this novel role during modification of the placenta.
Expression in abaxial carpel tissues is conserved in the CRC sub-family
The proteins encoded by eudicot CRC genes terminate the floral meristem, promote gynoecium growth, and elaborate the abaxial carpel wall structure, and their expression patterns are consistent with these functions (Bowman and Smyth, 1999; Alvarez and Smyth, 2002; Lee et al., 2005a,b; Orashakova et al., 2009). Although no obvious polarity defects are observed in Arabidopsis crc mutants (Bowman and Smyth, 1999; Alvarez and Smyth, 2002) or in Eschscholzia californica (Papaveraceae, basal eudicots) plants in which expression of CRC has been down-regulated (Orashakova et al., 2009), latent function related to establishment of adaxial–abaxial polarity has been suggested for CRC because polarity defects appear when crc mutations are added to other mutant backgrounds such as gymnos or kanadi1 (Eshed et al., 1999). In contrast, the rice CRC ortholog DROOPING LEAF (DL) is expressed in the leaf mid-rib in addition to whole carpel tissue, and is necessary for termination of the floral meristem, carpel identity and mid-rib formation (Yamaguchi et al., 2004). These functions appear to be conserved in the Poaceae, as consistent expression patterns are observed among members of the family (Ishikawa et al., 2009).
Expression in whole carpel wall and leaf is derivative in the CRC sub-family because expression of AmbCRC in Amborella trichopoda is limited to abaxial carpel tissues (Fourquin et al., 2005; Orashakova et al., 2009). The specific expression of CcCRC in the abaxial carpel tissues reinforces this evolutionary scenario. Expression data for a basal monocot Asparagas asparagoides (Asparagaceae) suggest that leaf expression is acquired in the base of monocots, and polar expression in the carpel is lost elsewhere between the basal monocots and Poaceae (Nakayama et al., 2010).
CcCRC expression is detected in the floral apex before the carpel primordia bulge. Thus, ancestral CRC genes could be involved in the floral meristem termination. The expression in other floral organs observed for CcCRC differs from that of monocot and eudicot CRC genes, while expression in the stamens is shared by CcCRC and AmbCRC (Fourquin et al., 2005). This implies that carpel-specific expression was acquired after divergence of the Nymphaeales. Determination of expression patterns in younger stages of the A. trichopoda carpel would indicate whether floral meristem termination and expression in tepals are primitive in the extant angiosperms.
Abaxial expression in the outer integument is a primitive trait in the INO sub-family
NaINO, an INO ortholog of Nymphaea alba (Nymphaeaceae, ANITA grade), is expressed in the abaxial epidermis of the outer integument as seen in Arabidopsis, but is also weakly expressed in the inner integument and nucellus (Yamada et al., 2003). However, the expression pattern of CcINO exactly parallels that observed in eudicots. This indicates that exclusive expression of INO in the abaxial epidermis of the outer integument is primitive, and that expression of NaINO in the nucellus and inner integument is an autapomorphy in Nymphaea.
What are the ancestral features of angiosperm YAB genes?
As discussed above, ancestral FIL, YAB2 and YAB5 genes are expressed in outgrowing lamina in leaves, while CRC and INO genes are expressed in tissue which promotes sporophyll expansion. Thus, the ancestral YAB genes of angiosperms are involved in laminar outgrowth, supporting the hypothesis that YAB genes played a critical role in generating flat leaf morphology upon megaphyll evolution in seed plants (Floyd and Bowman, 2010; Sarojam et al., 2010).
C. caroliniana plants were collected from Oike pond in the twon of Tamaki, Mie Prefecture, Japan, or purchased in aquarium shops. Collected plants were dissected into vegetative shoots and inflorescences before freezing in liquid nitrogen. Genomic DNA was extracted from the frozen vegetative shoots using a DNeasy plant mini kit (Qiagen, http://www.qiagen.com). cDNA was synthesized using the oligo(dT)-containing Adapter primer and other components in a ‘3’ RACE system for rapid amplification of cDNA ends' (Invitrogen, http://www.invitrogen.com/) using total RNA extracted from the inflorescences.
cDNA isolation and sequence determination were performed as described by Yamada et al. (2003), except for synthesis of the cDNA used in initial isolation of CcCRC and CcYAB2 fragments. For isolation of these two genes, we first preferentially reduced the level of mRNA for CcFIL and CcYAB5 by adding primers specific for these sequences, performing an initial reverse transcription, and digesting the resulting RNA:DNA hybrids using RNase H and DNase I (Invitrogen). The remaining RNA was then used for cDNA synthesis for CcCRC and CcYAB2 isolation. These steps reduce the number of CcFIL and CcYAB5 mRNA molecules that compete with CcCRC and CcYAB2 during their amplification. PCR primer sequences are given in Table S1. cDNA sequences were deposited in the DDBJ/NCBI/GenBank database with the accession numbers shown in parentheses: CcCRC (AB553318), CcFIL (AB553316), CcINO (AB553317) and CcYAB2 (AB553319). CcYAB5 (accession number AB126655) was referred to as CcYAB2 in a previous study (Yamada et al., 2004), but was renamed here based on its orthologous relationship with Arabidopsis YAB5.
The genomic sequences of C. caroliniana YAB genes were amplified using gene-specific primers (Table S1). These sequences were deposited in DDBJ/NCBI/GenBank under accession numbers AB553320–AB553323.
A HD-ZIP III mRNA was amplified by degenerate primers (Table S1), and the isolated sequence (CabC3HDZ1) was submitted to the DDBJ/NCBI/GenBank under accession number AB553324.
Sequence and phylogenetic analyses
Gene and mRNA sequences of C. caroliniana YAB genes were compared manually to determine the intron positions. Structural information on rice and Arabidopsis YAB genes was obtained from DDBJ/NCBI/GenBank.
mRNA sequences of YAB genes were obtained from DDBJ/NCBI/GenBank, and predicted protein sequences were aligned with those of C. caroliniana using Clustal X version 1.83.1 (Thompson et al., 1997). The obtained alignment was adjusted manually using MacClade version 4.06 (Sinauer Associates Inc., http://www.sinauer.com). Bayesian phylogenetic analysis was performed using BEAST version 1.46 (Drummond and Rambaut, 2007), based on the amino acids shown in Figure S1. The Jones-Taylor-Thornton+Gamma model of amino acid substitution was used for the analysis as recommended in ProtTest version 1.2.6 (Abascal et al., 2005). The trees were sampled every 1000 steps for 5 000 000 steps, and the first 25% of trees were discarded as burn-in. Convergence of the chain was verified using Tracer version 1.4 (http://tree.bio.ed.ac.uk/software/tracer/).
The same strategy was used for alignment of HD-ZIPIII genes, and the neighbor-joining phylogeny was inferred using Clustal X version 1.83.1 (Thompson et al., 1997) using the amino acids shown in Figure S5.
In situ hybridization and morphological observations
Inflorescences or vegetative shoot apices were fixed in FAA (50% ethanol/formaldehyde/acetic acid at a ratio of 18:1:1 by volume) at 4°C for 24 h. Sample embedding and RNA probe synthesis were performed as described by Yamada et al. (2004). Antisense and sense probes were designed to hybridize with full-length open reading frames of each YAB gene, except for CcFIL and CcYAB2. The probe for CcFIL was designed to recognize the downstream sequence of the YAB domain-encoding region, while the CcYAB2 probe corresponded to nucleotides downstream of the zinc finger-like domain-encoding region (Table S1). Only a partial sequence of CabC3HDZ1 was cloned (Table S1), and sense and antisense probes were synthesized from this fragment.
Embedded samples were sectioned at 10 μm thickness. Sections were serially mounted on four or five separate glass slides, and a set of slides was used for hybridization with three or four different antisense probes and a negative control (sense probes to CcINO, CcYAB2-2 or CabC3HDZ1; Table S2). Samples were digested using 1.0 μg ml−1 Protease K (Roche, http://www.roche.com) at 37°C for 30 min prior to hybridization with RNA probes at 51°C for 16 h. Non-specifically hybridized probes were washed away by rinsing twice in 0.5× SSC at 51°C for 20 min. Blocking, antibody detection and coloration were performed as described previously (Nakayama et al., 2010). Hybridization results using a negative control are shown in Figures S7 and S8.
FAA-fixed vegetative shoots were used for histological and SEM observations. Plastic sections and SEM samples were prepared as described by Yamada et al. (2001).
Semi-quantitative RT-PCR analysis
Total RNAs were extracted from shoot apices, young leaves (0.5–3 mm long), inflorescence apices with flowers <1 mm long, floral buds (1–3 mm long) and mature flowers (5–7 mm long). In C. caroliniana, the vegetative shoot meristem is terminated as an inflorescence meristem. To avoid inclusion of inflorescence meristem, laterally formed buds <1 mm long were collected as the vegetative shoot apex. However, the inflorescence apices could include very small leaves shorter than 0.5 mm which form during vegetative phase. Contaminating genomic DNA was eliminated using a TURBO DNA-free kit (Applied Biosystems, http://www.appliedbiosystems.com) before cDNA synthesis. The concentrations of cDNAs were normalized by amplifying a glyceraldehyde 3-phosphate dehydrogenase (GAPDH) homolog (accession number AB618609). Primers used in this analysis are listed in Table S1.
We thank Drs T. Nishiuchi and R. Kofuji for helpful comments on this study. This study was supported by a grant-in-aid from the Japan Society for the Promotion of Science to T.Y.