Comparative analysis of the RTFL family
The RTFL family is widely conserved among land plants and shares no sequence similarities with identified proteins or well-characterized motifs (Narita et al. 2004; Wen et al. 2004). Therefore, we investigated the biological functions of RTFL orthologs to increase our understanding of their effects on the control of plant organogenesis. SALAD is a motif-based database for plant comparative proteomics. This program can be used to predict biological function based on the hypothesis that proteins with similar motifs have similar biochemical properties and thus related biological functions (Mihara et al. 2009). Consequently, we collected 188 RTFLs among 22 species with full-length amino acid sequences for comparative analysis using SALAD (Table 1, see the whole amino acid sequences of 188 RTFLs in Supplemental Table S1). A total of 8 RTFLs were excluded by SALAD due to the low similarity calculated using the MEME software (Bailey et al. 2009), and the remaining 180 RTFLs were shown in the comparative analysis (Fig. 1, see the complete tree in Fig. S1). These sequences cover a wide range of land plant lineages, including liverworts, moss, gymnosperms, and angiosperms.
A total of 73 motifs were identified using the MEME suite (Bailey et al. 2009) among 180 RTFLs (see all motif sequences in Supplemental Table S2; see the nomination rules of motifs in the “Materials and Methods”). The RTF domain, which was used for blasting RTFL members, was presented as Motif 1 among all RTFLs (Figs. 1, S1). The N-terminal region of the RTFL family is less conserved among RTFL members, and no predictable signal peptides have been identified (Narita et al. 2004; Wen et al. 2004). This was confirmed by our results since the remaining 72 variable motifs (Motifs 2–73) were mostly identified around the N-terminus without any signal motifs (Figs. 1, S1).
The 180 RTFLs were grouped into four clades (Clades 1–4; Fig. 1) based on the motif patterns. Conserved motifs could be found in an interspecific manner. Excluding Motif 1 that defined the RTFL family, Motifs 2–12 were found in various species from liverworts to angiosperms (Figs. 1, S1). In Clade 4, Motifs 9, 19, 59 and 71 were found in Marchantia polymorpha (liverwort), Physcomitrella patens (moss), and Picea sitchensis (gymnosperm) (Figs. 1d, blue, green and yellow-green arrows, S1, respectively). Motifs 2 and 3 were found among RTFLs of eudicots and monocots in Clade 2 and Clade 3 (Figs. 1b, c, S1). In addition, Motifs 2–4 and 7–9 were found among species in Clade 2, which consisted of only eudicots (Figs. 1b, S1). Clade 1 consisted of two long RTFLs with diverse motifs (Figs. 1e, S1). Conserved motifs could also be found within specific families. Two types of motif combinations (Motifs 3, 4 and 8; Motifs 4, 7, and 9) were specific in Brassicaceae, and were also observed in Arabidopsis thaliana, A. lyrata subsp. lyrata, Thellungiella parvula, and T. halophila (Figs. 1b, S1). Motifs 2–4 and 8 were conserved among RTFLs in Glycine max and M. truncatula of Leguminosae (Figs. 1b, c, S1). Motifs 2, 3, and 10 were conserved among all species of Gramineae, which we compared to O. sativa, Brachypodium distachyon, Sorghum bicolor, Zea mays, and Hordeum vulgare var. distichum (Figs. 1c, S1). According to the motif patterns, RTFL members in Arabidopsis could be divided into five subgroups (Table 2); Subgroup 1 contained Motifs 1, 3 and 4, which were shared by RTFL 2 and 3 in Clade 2 (Figs. 1b, S1); RTFL4 and 5 in Clade 2 (Figs. 1b, S1) were grouped into Subgroup 2, both of which contained Motif 1, 4, 7 and 9; Subgroup 3 included ROT4, RTFL1 and RTFL 7–11 in Clade 3 (Figs. 1c, S1), with a pattern of Motifs 1 and 2 in common; RTFL15–19 and 21 in Clade 4 (Figs. 1d, S1) were grouped into Subgroup 4, with Motifs 1 and 6 in common; the remaining RTFLs (RTFL 6, 12, 13, 14, 20, 22 and 23) in Clade 4 (Figs. 1d, S1) showed diverse motif patterns with only Motif 1/functional RTF domain in common, and thus were grouped into Subgroup 5 (RTFL13 was excluded by SALAD in Figs. 1, S1, but was included in the same clade with RTFL14 when analyzed with lower amounts of RTFLs, unpublished data).
Phylogenetic analysis of RTFL members in Arabidopsis and O. sativa
Based on the above comparative analysis, we examined RTFL diversity between Arabidopsis (an eudicot) and O. sativa (a monocot), both of which are common model plants that have been fully sequenced (The Arabidopsis Genome Initiative 2000; mads Genome Sequencing Project 2005). A total of 90 % of Arabidopsis genes are believed to have homologs in the rice genome (International Rice Genome Sequencing Project 2005), and here 20 RTFL orthologous members in O. sativa exhibited diverse motif patterns (Table 3; Figs. 1, S1). Therefore, we generated a phylogenetic tree of 43 RTFL members from Arabidopsis and O. sativa (Fig. 2a) based on the conserved RTF sequences (identified as Motif 1 in Figs. 1, S1) for two reasons: (a) RTF domain/Motif 1 of Arabidopsis is sufficient to induce the RTFL-overexpression phenotypes in leaves and fruits (Ikeuchi et al. 2010); (b) RTF domain/Motif 1 is the only sequence/motif conserved among all RTFLs in Arabidopsis and O. sativa (Figs. 1, S1). The short length of RTF domains resulted in weak bootstrap values (data not shown), but the general topological relationships were observed regardless of the analysis parameters. Os01t0972300 was phylogenetically clustered into the same clade with ROT4 (Fig. 2a; arrows) based on RTF sequences/Motif 1, although they were in different clades based on the comparative analysis of whole amino acids sequences/whole motif patterns (Fig. 1c, d, purple and orange arrows). Os01t0972300 encodes 124 amino acids and was named OsRTFL3 in this report, which follows the nomination of OsRTFL1 and OsRTFL2 in Narita et al. (2004).
OsRTFL3 has similar functions as ROT4 in the development of above-ground organs
The long evolutionary history and conserved sequence of RTFL peptides are indicative of their essential functions in land plant evolution. To investigate whether the RTFL family is functionally conserved between eudicots and monocots, OsRTFL3 was constructed under the 35S promoter of the Cauliflower mosaic virus (CaMV35S) and transformed into wild-type Arabidopsis (Col-0). We established five independent transgenic lines and selected one line for further study after confirming that all individuals showed fundamentally similar phenotypes. Transgenic plants overexpressing ROT4 were used in Narita et al. (2004), with the coding sequence of ROT4 constructed under CaMV35S. The high expression level of ROT4 and OsRTFL3 was confirmed using RT-PCR in the transgenic plants (Fig. 2b). No detectable amplification of ROT4 was observed in the wild-type Arabidopsis under our PCR conditions, which could be explained by the low expression level. These two overexpressing lines were termed ROT4 o/x and OsRTFL3 o/x in the following content.
We next compared the morphology of wild-type plants, ROT4 o/x, and OsRTFL3 o/x. Both ROT4 o/x and OsRTFL3 o/x showed a pronounced reduction in organ size (Fig. 3a, b). The reduction in blade area, petiole length, blade length, and width of OsRTFL3 o/x was more significant when compared with wild type and ROT4
o/x (Figs. 3c,4a, b; P < 0.001, paired student’s t test). OsRTFL3 o/x also showed a short-organ phenotype in inflorescences and fruits, similar to ROT4 o/x (Fig. 3d, e). In addition, fruits of OsRTFL3 o/x were wider than those of wild type and ROT4 o/x (Fig. 3e). Although filaments and stamens were much shorter in OsRTFL3 o/x, they could reach the stigma at later developmental stages. Therefore, OsRTFL3 o/x was fully fertile, as was ROT4 o/x (Narita et al. 2004). The above comparison of gross morphology demonstrated that phenotypes of OsRTFL3 o/x in shoots were similar to ROT4 o/x, but were quantitatively different.
Organ size is determined by both cell size and number (Tsukaya 2006). To examine whether the reduced leaf size of ROT4 o/x and OsRTFL3 o/x were induced by a decrease in cell number and/or size, the number and size of palisade cells in the first rosette leaves of wild-type plants, ROT4 o/x, and OsRTFL3 o/x were measured. The total number of palisade cells per leaf blade in both ROT4 o/x and OsRTFL3 o/x decreased significantly, with a more severe reduction in OsRTFL3 o/x (Fig. 4c). To confirm whether the decreased cell number was related to the effect of ROT4 o/x and OsRTFL3 o/x on leaf shape, the palisade cell numbers in both the leaf-length and leaf-width direction were counted. The results showed that the cell number of OsRTFL3 o/x and ROT4 o/x in the leaf-length direction decreased in a similar pattern as the decrease in total cell number in the subepidermal layer. However, cell numbers along the leaf-width direction in OsRTFL3 o/x significantly decreased compared with wild type and ROT4 o/x (Fig. 4d). Similarly, the size of palisade cells in both ROT4 o/x and OsRTFL3 o/x significantly decreased, with a more severe reduction in OsRTFL3 o/x (Fig. 4e). This pattern indicates that both ROT4 o/x and OsRTFL3 control polar cell proliferation as well as cell expansion in the lateral organs, suggesting that OsRTFL3 has a similar function as ROT4 in the control of organogenesis when overexpressed in Arabidopsis. In addition, OsRTFL3 showed a unique function in negatively regulating the cell number along the leaf-width axis when overexpressed, which was not observed in ROT4 o/x lines.
OsRTFL3 o/x inhibited root growth
Although at least six RTFL members in Arabidopsis have been overexpressed and resulted in the dominant “round-leaf” phenotype, no significant differences were observed in morphological features of roots between wild type and RTFL overexpressors (Narita et al. 2004; Wen et al. 2004). However, according to our observations, both ROT4 o/x and OsRTFL3 o/x generated shorter primary roots, and OsRTFL3 o/x exhibited a more severe phenotype (Fig. 5a). The rate of root elongation decreased severely in OsRTFL3 o/x based on the time-course analysis, which was also observed in ROT4 o/x, but more mildly. (Fig. 5b). In addition, the roots of OsRTFL3 o/x almost stopped elongating around the fifth day after germination, while the elongation rate began to accelerate in the wild type and ROT4 o/x. Inhibition was observed in another three independent lines of OsRTFL3 o/x, and all OsRTFL3 o/x lines had a capability of generating lateral roots (Figs. 5a, S2). The developmental defects in root growth of both ROT4 o/x and OsRTFL3 o/x were inconsistent with the previous RTFL-overexpressing phenotypes observed in Arabidopsis. The phenotypes of OsRTFL3 o/x regarding the regulation of cell numbers along the leaf-width axis and root growth suggested that OsRTFL3 o/x may have unique functions in the control of organogenesis, in addition to the common functions as ROT4.