Present address: Institute of Cell, Animal and Population Biology, Rutherford Building, The King's Buildings, Edinburgh EH9 3JT, UK.
Leaf shape is determined by polar cell expansion and polar cell proliferation along the leaf axes. However, the genes controlling polar cell proliferation during leaf morphogenesis are largely unknown. We identified a dominant mutant of Arabidopsis thaliana, rotundifolia4-1D (rot4-1D), which possessed short leaves and floral organs. We showed that the altered leaf shape is caused by reduced cell proliferation, specifically in the longitudinal (proximal–distal) axis of the leaf, suggesting that the ROT4 gene controls polar cell proliferation in lateral organs. The ROT4 open-reading frame (ORF) encodes a novel small peptide that had not been identified in the Arabidopsis genome annotation. Overexpression of a ROT4–green fluorescence protein (GFP) fusion protein in transgenic plants recapitulated the rot4 phenotype, suggesting that ROT4 acts to restrict cell proliferation. The ROT4–GFP fusion protein localized to the plasma membrane when expressed in transgenic Arabidopsis plants. Phylogenetic analysis indicates that ROT4 defines a novel seed plant-specific family of small peptides with 22 members in Arabidopsis, ROT FOUR LIKE1–22 (RTFL1–22). All RTFL members share a conserved 29-amino acid domain, the RTF domain, and overexpression of the ROT4 RTF domain alone is sufficient to confer a rot4-1D phenotype. Loss-of-function mutations in several RTFL genes were aphenotypic, suggesting that there may be some functional redundancy between family members. Analyses by reverse transcription-polymerase chain reaction (RT-PCR) and in situ hybridization revealed that ROT4 is expressed in the shoot apex and young leaves of wild-type plants, consistent with a role for ROT4 in controlling polarity-dependent cell proliferation during wild-type leaf morphogenesis.
Plant leaves exhibit great diversity in shape and size between species. To understand the molecular basis of such morphological diversity, it is essential to identify the genes controlling the basic processes of leaf development. The analysis of leaf-shape mutants in Arabidopsis thaliana (L.) Heynh. has helped reveal how common developmental properties of leaf organs, such as dorsoventrality (Eshed et al., 2001; Kerstetter et al., 2001; McConnell et al., 2001; Sawa et al., 1999; Siegfried et al., 1999) and symmetry (Byrne et al., 2002; Iwakawa et al., 2002; Semiarti et al., 2001; Sun et al., 2002) are established (reviewed by Tsukaya, 2002). The control of leaf shape is less well characterized, although plants have several advantages as a system for analyzing organ shape. Because plant cells are surrounded by a cell wall, and the walls of neighboring cells are cemented to one another by an extracellular matrix of pectins, there is almost no cell migration during development. In addition, programmed cell death is usually not important for determining plant organ shape. As a result, cell division and expansion are the two major factors that influence organ shapes. Many mutants defective in cell expansion processes, including cell wall biosynthesis and modification (Cho and Cosgrove, 2000; Nicol et al., 1998), turgor generation (Dennison et al., 2001; Elumalai et al., 2002), endoreduplication (Sugimoto-Shirasu et al., 2002), as well as hormone biogenesis and perception, have been characterized (for review, see Tsukaya, 2003). In addition, transgenic plants, in which expression of core cell-cycle genes, such as CDKA (cyclin-dependent kinase A);1 (Hemerly et al., 1995), CYCD (cyclin-D)3;1 (Dewitte et al., 2003), CDK inhibitors (De Veylder et al., 2001; Wang et al., 2000), E2F (E2 promoter-binding factor), and DP (E2F dimerization partner) (De Veylder et al., 2002), was manipulated, have been described. Many of these plants produced small leaves because of the inhibition of cell expansion, cell proliferation, or cell differentiation. However, the relationship between the shape of leaves and cell behavior is largely unknown.
To reveal the mechanisms of leaf-shape control, we have screened for leaf-shape mutants with a specific alteration in either the length or the width of the leaf blade, and have characterized two classes of mutants, namely angustifolia (an) and rotundifolia (rot). In an mutants, the leaves are narrow, and this is caused by a reduction in cell size specifically in the leaf-width direction (Tsuge et al., 1996). The AN gene encodes a homolog of the transcription co-repressor CtBP (C-terminal binding protein) (Kim et al., 2002). By contrast, in rot3 mutants, the leaves are rounded in shape, and this is caused by a reduction in cell size specifically along the leaf-length axis. The ROT3 gene encodes a member of the cytochrome P450 family CYP90C1, potentially involved in brassinosteroid biosynthesis (Kim et al., 1998b, 1999, 2002). These results demonstrated that leaf cell expansion along the two main axes is under independent genetic control. In contrast, control of cell division orientation during leaf development is poorly characterized. Mutants with reduced leaf cell proliferation, such as aintegumenta (ant), struwwelpeter (swp), and pointed first leaf2 (pfl2), exhibit narrow leaf phenotypes (Autran et al., 2002; Ito et al., 2000; Mizukami and Fischer, 2000). Moreover, histological characterization of the brassinosteroid-related mutants, de-etiolated2 (det2) and dwarf1 (dwf1), showed that these two mutants are defective not only in leaf cell expansion but also in leaf cell proliferation, preferentially along the leaf-length direction (Nakaya et al., 2002). Consistently, expression of CYCD3;1 is induced by brassinosteroid in det2 suspension culture (Hu et al., 2000). These observations suggest that there is genetic control of the direction of cell division within the leaf blade.
In this study, we carried out further mutant screening using an activation T-DNA tagging method to gain more insights into polarity-dependent leaf development. A newly isolated dominant mutant, named rot4-1D, develops short and rounded leaves, which are caused by a decrease in cell proliferation specifically along the leaf-length direction. ROT4 encodes a plasma membrane-localized small polypeptide of 53-amino acid residues, and defines a novel plant-specific family of small peptides. Our study will provide information for better understanding of the mechanism that links cell proliferation and leaf morphogenesis.
Isolation and phenotypic characterization of the rot4-1D mutant
An insertional mutagenesis was performed by transforming plants with the T-DNA-containing plasmid pSKI015. Insertion of this T-DNA in intergenic regions frequently results in gain-of-function mutations that increase expression of flanking genes (Weigel et al., 2000). Seeds were harvested from 2000 individual primary transformants (T1 generation), and the resulting T2 families were screened for mutations affecting leaf morphology. One family segregated for a novel short-leaf phenotype in a ratio (61 mutants:18 wild type) consistent with the presence of a single dominant mutation. Because the mutagenesis was performed in a Landsberg erecta (er) background that also carried the weak curly leaf (clf) allele, clf-9 (Goodrich et al., 1997), the mutation was introduced into an ER+CLF+ background by crossing three times to plants of Columbia ecotype. A similar dominant mutant phenotype was observed to segregate, indicating that the phenotype was independent of the ER or CLF genes. The mutation was designated rot4-1D, and all subsequent morphological analyses involved comparison of rot4-1D and wild-type sibs in the introgressed Columbia background.
The leaf blade of rot4-1D was short and rounded (Figure 1a). The reductions in leaf blade and petiole length in rot4-1D were more severe than those in leaf blade width (Figures 1b and 2a). In addition, the reduction in leaf blade length was more pronounced in leaves formed at later developmental stages (Figures 1b and 2a). In addition to the effects on leaf blade and petiole, rot4-1D floral organs and inflorescence stems were also shorter than those of wild type (Figures 1c,d and 2a). By contrast, there were no significant differences in morphological features between wild-type and rot4-1D roots (data not shown).
To test whether the reduced leaf length in rot4-1D was because of a decrease in cell number or size, or both, the numbers and sizes of the palisade cells in the first, third, and fifth rosette leaves were measured. The leaf blades of rot4-1D had fewer cells than that of the wild type, and, as in the case of the leaf blade length, this was more evident in the leaves formed at later developmental stages (Figure 2b). To relate the decrease in the cell number to the effect of rot4-1D on leaf shape (Figure 2a), the palisade cell numbers in both the leaf-length and leaf-width directions were counted. Again, the number of palisade cells in the leaf-length direction in rot4-1D decreased in a similar fashion to the decrease in total cell number in the subepidermal layer (Figure 2c). However, the cell numbers along the leaf-width direction in rot4-1D were not significantly different at any of the developmental stages examined (Figure 2c). In addition, the palisade cell sizes in bottom, middle, and top portions of first and third leaves in rot4-1D were comparable to the corresponding wild-type leaves and portions (Figures 1e and 2d). Although the cell size in the bottom portion of the fifth leaf in rot4-1D was smaller than that of wild type (Figure 2d), this might be because of a retardation in fifth leaf expansion as rot4-1D plants grow slightly slower than wild type. Cells in older leaves of rot4-1D expanded to a similar extent to wild-type cells.
Taken together, these results suggest that the main effect of the rot4-1D mutation is a decrease in cell proliferation along the leaf-length direction. The measurements of cell size and cell number do not account for the decrease in leaf width (Figure 2a). This discrepancy could be attributable to the combined effect of subtle decreases in the cell number in the leaf-width direction and cell size in rot4-1D. In fifth leaves, wild type and rot4-1D were indistinguishable with respect to the sizes and the shapes of palisade, sponge, and epidermal cells, when transverse sections were examined (data not shown). We also characterized cells in the leaf margins, as epidermal cells in the leaf margins make only longitudinal divisions in tobacco at least (Poethig and Sussex, 1985). As elsewhere, the length of the cells did not differ significantly between wild type and rot4-1D (123.1 ± 40.0 µm for wild type versus 120.3 ± 27.0 µm for rot4-1D) in the first set of foliage leaves (n = 16 for wild type and 17 for rot4-1D). As the length of the leaf margin is significantly decreased in the rot4-1D mutant (Figures 1a and 2), these data indicate that longitudinal cell division also in the margin of the leaves is severely decreased in the mutant. Taken together, cell proliferation in all regions of the leaf blade is decreased by the overexpression of ROT4.
Identification of ROT4
Southern blot analysis of the original T2 population that segregated rot4-1D indicated that a single T-DNA insertion was present (data not shown). To test whether this insertion might cause rot4-1D mutation, we analyzed the genetic linkage of the mutant phenotype with a herbicide (bialaphos)-resistance marker carried on the T-DNA. All 68 mutants in the original T2 family were resistant, whereas the 18 wild-type plants were herbicide-sensitive, consistent with the insertion being responsible for the rot4-1D mutation. To identify the affected gene, the DNA flanking the T-DNA insertion was isolated by plasmid rescue (see Experimental procedures). Sequence analysis indicated that the insertion was in an intergenic region on chromosome II between the genes At2g36980 and At2g36990 (Figure 3a). However, when the mRNA levels of these genes were determined by reverse transcription-polymerase chain reaction (RT-PCR), expression levels were similar in rot4-1D and wild-type plants (data not shown), suggesting that neither gene was likely to be responsible for the rot4-1D phenotype. Because the interval between these two genes was relatively long (about 4.5 kbp), we searched the region for potential open-reading frames (ORFs). A tblastx search was performed using the genomic DNA sequence between the stop codon of At2g36980 and the stop codon of At2g36990 as a query, and several cDNA sequences from A. thaliana and other plant species were retrieved. Although none of them perfectly matched the query genomic sequence, on the basis of the alignments and conceptual translations, a small potential ORF, which encodes 53-amino acid residues, was identified (Figure 3a,b). The ORF was located within 1 kbp from the right border of the T-DNA insert. Because pSKI015 T-DNA insertions generally affect expression of genes adjacent to the T-DNA right border, which carries cauliflower mosaic virus (CaMV) 35S enhancers (Weigel et al., 2000), it seemed likely that increased expression of this ORF was the basis of the dominant rot4-1D phenotype.
To test whether this small ORF was expressed in wild type, RT-PCR was carried out. As shown in Figure 3(c), a cDNA band of the expected size was amplified. In addition, this mRNA accumulated at a higher level in rot4-1D than in wild type (Figure 3c). Because these results strongly suggested that ROT4 encodes this small ORF, its full-length sequence was determined by 5′ rapid amplification of cDNA ends (RACE) PCR (GenBank Accession number AB107209). To confirm that this gene corresponded to ROT4, we tested whether the rot4-1D phenotype was reconstituted when wild-type plants were transformed with a construct that placed the candidate gene fused with green fluorescence protein (GFP) under control of the CaMV 35S promoter (35S::ROT4-GFP). Foliage leaves of homozygous T3 plants from three independent transformants were shorter than those of wild type, as was seen in rot4-1D (Figure 3d), confirming that the candidate gene corresponded to ROT4.
Expression pattern of ROT4
The ROT4 expression pattern was examined in various organs of wild-type plants by RT-PCR analysis using ROT4-specific primer sets. Although ROT4 was strongly expressed in flowers and flower buds, the accumulation of ROT4 mRNA was not detected in mature rosette leaves (Figure 4a). In young seedlings, however, the expression of ROT4 was clearly detected in shoots (Figure 4b). We next characterized the ROT4 expression level in shoots of seedling at different developmental stages. After stratification, plants were incubated under dark condition for 2 days, and were then grown in light for several days. The ROT4 expression level in shoots of seedlings was highest at 6 days after the shift to light (Figure 4c). Using plants at this stage, the organ-specificity of ROT4 expression was analyzed. The shoots of seedlings were dissected into the fully expanded cotyledons, the primordia of the first pair of foliage leaves, and shoot apices, which included the shoot apical meristem and younger leaf primordia. The ROT4 expression was highest in the shoot apices and somewhat reduced in the leaf primordia (Figure 4d). Expression in the mature cotyledons was extremely low. These results confirmed that ROT4 is expressed in wild-type leaves during early developmental stages. To localize the ROT4 transcript more precisely, digoxigenin-labeled probes were hybridized in situ to sections of wild-type and rot4-1D plants. The probes were synthesized from a short 143 bp region of ROT4 that lacked similarity with any other sequences in the Arabidopsis genome. However, we were unable to detect any signal when wild-type materials were probed with the ROT4 antisense probe, other than a very low background signal that was also seen when a ROT4 sense probe was used as a control (data not shown). As a further control, we also hybridized probes for the AGAMOUS (AG) gene to wild-type inflorescence sections. The characteristic AG expression pattern was seen (Figure 4e), confirming that the experimental sensitivity was adequate for detecting tissue-specific accumulation of transcripts. The sensitivity may have been lower for ROT4, because the probe necessarily derived from a much shorter region. In the case of AG, the probe was synthesized from a 1-kbp region of the AG transcript, then hydrolyzed into fragments with an average length of 150 bp. Consequently, there are about seven times as many probe-binding sites in the AG transcript as there is in the much shorter ROT4 transcript.
When the ROT4 antisense probe was hybridized to rot4-1D materials, unlike wild type, a strong signal was consistently observed (Figure 4f–l). This agreed with the RT-PCR results, which also indicated higher expression in the rot4-1D mutant than in wild type. In seedlings, there was little expression in the shoot apical meristem (Figure 4f), whereas young leaf primordia showed expression throughout the lamina of young leaf primordia (Figure 4f,g). In cotyledons and older leaves, expression was restricted to vasculature (Figure 4f,g). The expression pattern in inflorescences was complex. At stages 1 and 2 of early flower development, ROT4 was present in the center of the floral meristem, in the L3, and underlying cells. A weak signal was detected in the inflorescence meristem in the region where a floral meristem was about to initiate (Figure 4h). In stage 5 floral buds, faint expression occurred in the center of the floral meristem and also in the sepals (Figure 4i). In stage 7 flowers, strong expression was observed in the sepals and in the center of the developing stamens (Figure 4j). In stage 9 flowers, ROT4 was expressed throughout the petals, and in all other floral organs, most strongly over the vasculature (Figure 4k). At later stages of flower development, a strong signal was restricted to the vasculature of the floral organs (Figure 4l); expression was also detected in the vasculature of the stem (not shown).
The expression of ROT4 was generally consistent with extensive expression during early lateral organ formation, resolving to vascular-specific expression later in development. The analysis of other mutants generated by activation tagging has shown that the genes disrupted usually show greatly elevated levels of expression, but maintain their normal pattern of expression (Weigel et al., 2000). Because the rot4-1D mutant shows localized expression of ROT4, broadly consistent with the pattern in wild-type seedlings as crudely discerned in RT-PCR experiments, it is likely that the rot4-1D mutant also confers elevated expression of ROT4 in its normal pattern rather than ectopic expression.
Plasma membrane localization of ROT4
To determine the intracellular localization of the ROT4 protein, we examined young leaves of 35S::ROT4–GFP transgenic Arabidopsis plants using confocal microscopy. As shown in Figure 5(a,b), GFP signal was localized at the periphery of both palisade and epidermal cells in the transgenic Arabidopsis plants. The same localization was observed also in onion epidermal cells that were bombarded with the 35S::ROT4–GFP transgene (Figure 5d). To determine whether the ROT4–GFP fusion was localized in the plasma membrane or in the cell walls, plasmolysis was induced with 0.8 m mannitol. ROT4–GFP localized with the shrunken plasma membrane, indicating that it was present in the plasma membrane and not in the cell wall (Figure 5c,d). Thus, unlike the 35S::GFP control plants, in which the GFP fluorescence was mainly cytosolic (Figure 5e), the ROT4–GFP fusion protein was localized to the plasma membrane.
Microarray analysis in rot4-1D
Overexpression of ANT extends the period of CycD3 expression in leaves and eventually leads to the formation of large leaves (Krizek, 1999; Mizukami and Fischer, 2000). Therefore, we expected that expression of some cell-cycle-related genes might be altered in rot4-1D. To identify genes whose expression level was altered in rot4-1D mutants, a microarray analysis was carried out. Although cell proliferation activity is reduced in rot4-1D, expression levels of genes that control the cell proliferation, such as cell-cycle-related genes, were not detectably altered. Several genes were dramatically upregulated in rot4-1D (Table 1), but the relationship between these genes and cell proliferation is unclear. There were no genes that were downregulated below 10% levels of wild-type gene expression (data not shown).
Table 1. Genes upregulated in rot4-1D identified by gene chip analysis a
Genes for which levels of transcripts were at least three times higher in rot4-1D than in wild type (wt) are shown.
Ratio of transcript levels in the rot4-1D mutant to those in wt.
Lipid transfer protein, putative
Caffeoyl-CoA 3-O-methyltransferas, putative
Anther development protein
Lipid transfer protein-like protein
Xyloglucan endotransglycosylase, putative
Putative cysteine proteinase inhibitor B
Family II extracellular lipase
Ribosomal protein L16
Ribosomal protein L14
ROT4 is a member of a novel seed plant-specific family of small peptides
Database searches revealed 22 other putative Arabidopsis proteins with similarity to ROT4. These proteins were named ROT FOUR LIKE1–22 (RTFL1–22). Alignment revealed a 29-amino acid region that was conserved between the RTFL members (Figure 6a). The region was rich in basic amino acids, and had not been previously identified in any other proteins of known function. We named the novel region RTF domain. When a phylogenetic tree was produced from the RTF domain, no subclasses were found between the RTFL members (Figure 6b). In other plant species, such as Oryza sativa (O.s.), Glycine max (G.m.), Populus balsamifera (P.b.), etc., there are ESTs that encode RTFL members (O.s., AU182671; G.m., AW458841; P.b., BU880463). These hypothetical proteins were not found in non-seed plants and animals, suggesting that RTFL is a seed plant-specific gene family.
To further characterize RTFL function, in particular, the loss of function phenotype, we screened existing collections of insertion mutants in Arabidopis and rice for insertions that disrupted any RTFL member. Such mutations were very rare, presumably because the RTFL genes were extremely small, and we could not identify any insertion at ROT4. However, a mutant that carried a T-DNA insertion within the ORF of RTFL4 was identified within the SIGNAL collection of Arabidopsis T-DNA insertion lines (Figure 7a,c). The mutation is likely to be null because it disrupts the conserved RTFL domain, and was designated as rtfl4-1. Furthermore, we identified two mutants in Oryza (rice) that carried insertions of the Tos17 (transposon of Oryza sativa) transposon within Oryza homologs. These Oryza RTFL members were named OsRTFL1 and OsRTFL2, and encoded proteins sharing 45 and 38% amino acid identity with ROT4 within the RTF domain, respectively (Figure 7b). The two rice mutants, also predicted to be severe loss-of-function alleles, were named osrtfl1-1 and osrtfl2-1, respectively. We identified plants that were homozygous for the various mutant alleles. However, none of the mutants showed any phenotype that we could discern, in terms of either organ shape or any other gross morphological character (Figure 7c,d). These results suggest that there might be functional redundancy between several or all RTFL members, as might be expected as the family comprises more than 20 members with similar protein sequences.
The RTF domain is sufficient for ROT4 activity
Although the RTF domain is conserved between distantly related plant species, the regions outside the RTF domain at the N- and C-terminal ends of the protein are diverged. To investigate the function of the different domains of ROT4, we generated the constructs ROT4ΔN or ROT4ΔC, which expressed modified proteins that lacked either the N-terminal or the C-terminal regions of ROT4, respectively (Figure 6c). The constructs were expressed under control of the CaMV 35S promoter in transgenic Arabidopsis plants. In both cases, the resulting transgenic plants had short rosette leaves similar to those of rot4-1D mutants (Figure 6d). This result suggests that only the conserved RTF domain is required for ROT4 activity.
As in rot4-1D plants, which had short siliques, the transgenic plants also had short siliques. This phenotype was less severe for plants expressing ROT4ΔN than for those expressing ROT4ΔC. Interestingly, the siliques of both ROT4ΔN- and ROT4ΔC-overexpressing plants showed changed shape so that they resembled an arrowhead (Figure 6e). A similar phenotype was also reported for plants overexpressing the CYP78A9 gene, which encodes a cytochrome P450 enzyme. In this case, the effects on silique shape were caused by both increased cell proliferation and increased cell enlargement in the carpel valves (Ito and Meyerowitz, 2000).
ROT4 is a novel small peptide with a conserved RTF domain
We identified a novel gene that is overexpressed in an Arabidopsis mutant with short leaves (Figure 1). ROT4 encodes a small peptide and its predicted molecular weight is only 6.2 kDa. Genome-wide database searches revealed that ROT4 is a member of a novel gene family, which shares a 29-amino acid domain RTF. All these members encode small peptides. Small peptides are often involved in developmental or defense signaling, for example, CLAVATA3 (CLV3), phytosulfokine, S-locus cystein-rich protein/S-locus protein11, ENOD40 (early nodulin 40), systemin, Brick1 (Brk1), rapid alkalinization factor, and POLARIS (PLS) have been described by Casson et al. (2002), Fletcher et al. (1999), Frank and Smith (2002), Matsubayashi and Sakagami (1996), Pearce et al., (1991, 2001), Schopfer et al. (1999), Takayama et al. (2000), and Yang et al. (1993). With the exception of Brk1 and PLS, these peptides are shown to act extracellularly as peptide hormones (Ryan et al., 2002). Among the known small peptides Brk1 is unusual because it has no predictable targeting signal to membranes. This suggets that Brk1 would function in the cytoplasm. ROT4 represents another unusual example as the ROT4–GFP fusion protein is localized to the plasma membrane in leaf cells (Figure 5). Given that ROT4 has neither a predictable signal peptide nor a transmembrane domain, it is likely that ROT4 is a peripheral membrane protein or a part of a plasma membrane-localized protein complex. The RTF domain within the ROT4 peptide is likely sufficient for ROT4 activity as transgenic plants expressing ROT4ΔN or ROT4ΔC truncations have short rounded leaves similar to those of the rot4-1D mutant (Figure 6d). This result indicates that other RTFL members may have similar activities on account of their shared RTF domain, and suggests that there could be substantial functional redundancy between RTFL members.
Limitations of conventional genetic screens for identifying small genes
Because ROT4 and RTFL genes have not been annotated in the Arabidopsis genome database, it is likely that many other novel small genes may be overlooked by gene prediction programs. In addition, the small size of their coding regions means that loss-of-function mutations are likely to be rare, so that the genes may not be disrupted in conventional mutagenesis programs. For example, we were able to identify only one insertion mutant amongst the 22 RTFL members in Arabidopsis, despite the extensive collections that exist for reverse genetic analysis. Similarly, loss-of-function mutations affecting the CLV3/embryo-surrounding region (ESR)-related (CLE) gene family, which contains 28 members encoding small peptides (Cock and McCormick, 2001), are also extremely rare (Hobe et al., 2003). In addition to the relative scarcity of mutations within small genes, redundancy between family members may also limit their identification in forward genetic screens. Consistent with this, it was striking that the rtfl4-1, osrtfl1-1, and osrtfl2-1 mutants were all without gross morphological phenotypes (Figure 7b). For identification of small genes and/or genes with functional redundancy, activation tagging has several advantages. First, overexpression can often reveal a gain-of-function phenotype when loss-of-function mutations are without obvious phenotypes (Weigel et al., 2000; Wilson et al., 1996). Second, because insertions that are very distant from the promoter (at least 3.6 kbp upstream; Weigel et al., 2000) can give gain-of-function phenotypes, genes with small coding regions are nonetheless mutagenized as effectively as those with much larger coding regions. Thus, activation tagging has recently proved as an efficient means for identifying another class of small genes, those encoding microRNA (Palatnik et al., 2003).
The role of ROT4 in cell proliferation and leaf morphogenesis
When ROT4 was overexpressed, cell numbers decreased specifically in the leaf-length direction (Figures 2 and 3d). In addition, other organs, such as petals and sepals, also showed a shortened, rounded shape, suggesting that ROT4 overexpression may generally decrease polar cell proliferation in lateral organs, specifically in the leaf-length direction (Figure 1c,d). As cell proliferation in the leaf-length direction in a leaf primordium continues as long as the leaf primordium grows (Donnelly et al., 1999), the effect of ROT4 appears to be on leaf growth, but not on the establishment of the leaf primordia. The mRNA of ROT4 accumulated in wild-type shoot apices and leaf primordia, but was not detected in mature rosette leaves (Figure 4b,d). Strong expression of ROT4 was also detected in roots, flowers, and flower buds, which, like apices and young leaf primordia, are regions of active cell proliferation. Because ROT4 acts to restrict cell proliferation, it is perhaps surprising that it is mainly expressed in regions of active cell proliferation. However, a precedent for this is provided by the CINCINNATA (CIN) gene of Antirrhinum, which is also involved in the control of leaf shape. The CIN gene is proposed to restrict cell proliferation by promoting cell cycle arrest, and it is also expressed in young leaf primordia in actively dividing regions of the leaf lamina (Nath et al., 2003). The expression of CIN moves in a wave that passes from tip to base of the leaf, and that is correlated with a wave of cell cycle arrest and cell expansion that moves from tip to base. Because we were unable to precisely localize ROT4 transcripts in wild-type plants by in situ hybridization, it remains unclear whether ROT4 is expressed in a similar fashion. The creation of reporter gene constructs for ROT4 expression may help resolve this issue. In addition, it will be interesting to observe the behavior of the arrest front in rot4-1D mutants.
Two other factors that inhibit cell proliferation, ICK1 (cyclin-dependent kinase inhibitor 1) and KRP2 (kip-related protein 2), have been analyzed in detail (De Veylder et al., 2001; Wang et al., 2000). These proteins bind to CDKA;1 and D-type cyclins, and inhibit the activity of cyclin-dependent kinase (Wang et al., 1998). The characteristic phenotypes of the respective overexpressers are that they show an extreme decrease in the number of cells per leaf blade (10% of wild-type leaf cell number), and have serrated leaves. ROT4 also exerts an inhibitory effect on cell proliferation. However, a critical difference compared with the overexpression of ICK1 or KRP2 is that the reduction in cell number caused by overexpression of ROT4 is modest (Figure 2b; in the severest case, cell number of rot4-1D was 67.5% of that of wild type). Such differential inhibitory effects on cell proliferation may reflect different roles in the control of cell proliferation. Whereas ICK1 and KRP2 directly control key cell-cycle components, it is likely that ROT4 acts more indirectly. First, ROT4 is plasma membrane-localized and therefore is likely to interact indirectly through a transduction pathway with the cell-cycle machinery. In addition, ROT4 and RTFL are a seed plant-specific gene family, suggesting that they have evolved to control cell proliferation by a specialized manner.
Another important difference between plants overexpressing ICK1 or KRP2, relative to rot4-1D, is in the effect on individual cell sizes. In the previous reports, it was shown by the observations of transgenic plants expressing ICK1, KRP2, or dominant negative CDKA;1, and also ant, swp, pfl2, G-protein a subunit 1 (gpa1), and gibberellic-acid insensiteive (gai; Autran et al., 2002; De Veylder et al., 2001; Hemerly et al., 1995; Ito et al., 2000; Mizukami and Fischer, 2000; Tsukaya et al., 2002; Ullah et al., 2001; Wang et al., 2000) that when the cell number of leaves decreases, compensatory increase in cell size occur, and the reduction of leaf size is consequently alleviated (reviewed by Tsukaya, 2002, 2003). This frequently observed phenomenon raised a possibility of the existence of a compensatory system between cell division and expansion (Tsukaya, 2002, 2003). However, this does not always apply, for example, in leaves of clf mutants, both cell number and size decrease (Kim et al., 1998a). To the best of our knowledge, rot4-1D is one of the very few instances of a mutant in which a decrease in cell number does not affect cell size. The only other example that we are aware of is the ask1 mutant (Zhao et al., 1999). It should also be noted that some natural variations in leaf size and shape are solely attributable to changes in cell number (Tsukaya, 2002). Thus, it is also possible that ROT4 is involved in the control of organ shape and size. Further analysis of the function of ROT4 and RTFL peptides should help us to understand how small peptides control development of plant organs. Another important question to address is whether individual leaf cells possess proliferation potential in leaf-length and leaf-width directions independently of one another. The rot4-1D phenotype is consistent with independent control of proliferation along the two axes. Observation of the direction of cell division during leaf blade morphogenesis will provide more information to help answer this.
Plant materials and growth conditions
The rot4-1D mutant was isolated in a Landsberg er background that also carried the weak clf mutant allele, clf-9 (Goodrich et al., 1997). The rot4-1D mutant was crossed to Columbia wild type three times to remove the clf-9 and er alleles, and an identical phenotype was observed, indicating that it was independent of the clf or er background. Because the rot4-1D inflorescence phenotype resembled that of er, the absence of the er mutation was confirmed by a sequence analysis of the ER locus. For analyses of plants, seeds were sown on rock wool. After 24 h at 4°C in darkness, plants were grown under light at 40.4 ± 6.7 µmol m−2 sec−1 at 22°C. Ages of plants were given in terms of days after transfer to 22°C.
To observe palisade cells, leaf tissues were cleared with chloral hydrate solution (4 g ml−1 chloral hydrate and 0.4 g ml−1 glycerol) as described elsewhere (Tsuge et al., 1996). Individual cell sizes were measured using the imagej 1.29x program (National Institutes of Health; http://rsb.info.nih.gov/ij/).
Determination of the T-DNA insertion site in rot4-1D
To clone genomic DNA flanking the T-DNA insertion, plasmid rescue (Weigel et al., 2000) was performed. Genomic DNA was extracted from rot4-1D mutant plants, digested with EcoRI, and ligated overnight under conditions favoring intramolecular ligation. The resulting plasmid, pJG28, contained both pSKI015 and plant genomic DNA sequences with a single EcoRI site. Sequence analysis of pJG28 using a primer designed from pSKI015 sequence indicated the approximate site of the T-DNA insertion. To confirm the precise T-DNA insertion site, the following oligonucleotides were used for PCR amplification using rot4-1D genomic DNA as a template: ROT4RBsite, 5′-AAGCCGTTAGGTTTAGAGGG-3′; T7, 5′-GTAATACGACTCACTATAGGGC-3′. Amplified DNA fragments were sequenced with the ROT4RBsite primer.
Determination of the full-length ROT4 cDNA sequence
The full-length cDNA of ROT4 was established by 5′ RACE and 3′ RACE using 5′ RACE System Version 2.0 (Invitrogen Corp., Carlsbad, CA, USA) and 3′ RACE System (Invitrogen Corp.). Total RNAs were prepared from whole mature plants using the RNeasy Plant Mini Kit (Qiagen Inc., Valencia, CA, USA). The following oligonucleotides were used as gene-specific primers 1 (GSP1) of PCR for 5′ RACE: ROT4-R, 5′-TAATCAAGAGTCTTTGCGGTCG-3′; and as GSP2; 5′ RACE-GSP2, 5′-CGTGCCAGCAAACTAACATG-3′. For 3′ RACE, the following oligonucleotides were used as GSP1 and GSP2: ROT4-F, 5′-ACAAATCAATGGCACCGGAG-3′; 3′ RACE-GSP2, 5′-TGAGCCGTGCAAGACTTTTG-3′.
Generation of transgenic lines overexpressing ROT4 variants
Total RNAs were prepared from whole mature plants as above. The SuperScript one-step RT-PCR Kit (Invitrogen Corp.) was used for RT-PCR according to the manufacturer's protocol. The condition for amplification by RT-PCR was one cycle at 50°C for 15 min and 94°C for 2 min, then 35 cycles at 94°C for 15 sec, 57°C for 30 sec, and 72°C for 60 sec. ROT4-F and ROT4-R were used as primers for RT-PCR. The RT-PCR products were subcloned into pCR2.1 (Invitrogen Corp.) and the resultant vector was named ROT4/pCR2.1. To generate 35S::ROT4–GFP transgenic plants, the binary vector construct ROT4–GFP/pSMAB704 was first made as follows. ROT4 DNA was PCR-amplified from ROT4/pCR2.1 template DNA using the oligonucleotides ROT4-Gly-F (5′-CAATGGCACCGGAGGAGAATG-3′) and ROT4-Gly-R (5′-TTTCCATGGCACCTCCACCTCCACCTCCAGAGTCTTTGCGGTCGTGG-3′) in order to remove the ROT4 stop codon, and to introduce an NcoI site and a glycine linker. The product was subcloned into pCR2.1 to produce ROT4-Gly/pCR2.1. ROT4-Gly/pCR2.1 was digested by EcoRI, blunted using a DNA Blunting Kit (TaKaRa Bio Inc., Otsu, Japan), and then digested with NcoI. The vector pTH2 (Niwa et al., 1999), which carries the GFP gene under control of the CaMV 35S promoter, was digested with SalI, blunted as described above, and then digested with NcoI. These two fragments were ligated to produce ROT-GFP/pTH2. pSMAB704 was digested with HindIII and EcoRI to remove the 35S::β-glucuronidase gene and then ligated to the 35S::ROT4–GFP gene, which was excised from ROT4–GFP/pTH by HindIII and EcoRI digestion. Similarly, plasmid vector constructs ROT4ΔN/pCR2.1 and ROT4ΔC/pCR2.1 were made as follows. ROT4ΔN DNA was PCR-amplified from ROT4/pCR2.1 template DNA using the oligonucleotides ROT4dN-F (5′-ATGACTTTTGGGCAAAAGTGC-3′) and ROT4-R in order to remove the N-terminus region of ROT4 and to add the first methionine codon. ROT4ΔC DNA was PCR-amplified using the oligonucleotides ROT4-F and ROT4dC-R (5′-TCAGTCGTGCCAGCAAACTA-3′) in order to remove the C-terminus region of ROT4 and to add the stop codon. Each DNA fragment was inserted in pCR2.1. The DNA fragment insertion to the binary vector pSMAB704 was performed as described for the case of ROT4–GFP. These constructs were introduced into wild-type plants by Agrobacterium-mediated transformation using the simplified floral dip method (Clough and Bent, 1998). Transgenic plants were selected on Murashige and Skoog medium containing 2 mg ml−1 Gellan Gum (Wako, Osaka, Japan), 10 µg ml−1 bialaphos (Shinyo-Sangyo, Tokyo, Japan), and 500 µg ml−1 claforan (Aventis Pharma Ltd, Tokyo, Japan).
Reverse transcriptions were carried out using the Super Script II RT and the adapter primer of the 3′ RACE System. For amplification of ROT4 cDNA, ROT4-F and ROT4-R were used. As a control, the following oligonucleotides were used to detect the constitutively expressed ACTIN2 (ACT2) gene (An et al., 1996): ACT2-F, 5′-GAAATCACAGCACTTGCACC-3′; ACT2-R, 5′-AAGCCTTTGATCTTGAGAGC-3′. The condition for amplification by RT-PCR was one cycle at 50°C for 15 min and 94°C for 2 min, then 27 cycles for ROT4 cDNA amplification or 33 cycles for ACT2 cDNA amplification at 94°C for 15 sec, 57°C for 30 sec, and 72°C for 60 sec.
In situ hybridization
A 143-bp fragment from the 5′ end of the ROT4 gene that had no similarity with any other sequence in the Arabidopsis genome was amplified from genomic DNA by PCR using the primers ROT4-F (5′-CTTCCATTTCTTAAGCCTTTTAACTAG-3′) and ROT4-R (5′-CAAAAGTCTTGCACGGCTCACAC-3′). The product was ligated to the vector pGEM-Teasy (Sigma–Aldrich Company Ltd., Dorset, UK) to produce the clones pJG38 and pJG39, which contained the ROT4 insert in opposite orientations relative to the T7 promoter. Sense and antisense probes were synthesized using T7 RNA polymerase, and pJG39 or pJG38 templates were linearized with SalI, respectively. AG antisense probes were prepared from a pCIT565 template as described previously by Drews et al. (1991). Digoxigenin-labeled probes were prepared as described by Coen et al. (1990). The methods for tissue fixation, probe hybridization, and detection were as described by Long et al. (1996). For seedlings, plants were grown under short days and harvested after 12 days, at which stage the seedlings had two visible leaves. Inflorescences were harvested from plants grown under long days when the first flowers had opened. Tissue sections were counterstained for 5 min in 0.1% Tinopal UNPA-GX fluorescent brightener (Sigma–Aldrich Company Ltd.) dissolved in dilute NaOH, rinsed briefly in 10 mm Tris (pH 8.0), and mounted in Entellan (BDH Merck Chemicals Ltd., Dorset, UK). The sections were observed under combined light field and UV epifluorescent illumination.
Observation of the intracellular localization of ROT4
The ROT–GFP/pTH2 construct was introduced into Arabidopsis plants by floral dip transformation as above. The GFP fluorescence was observed under a confocal laser scanning microscope, TCS SP2 (Leica, Wetzlar, Germany). To carry out a plasmolysis experiment, epidermal peel was immersed in 0.8 m mannitol solution for 2 h.
Total RNAs were prepared from 25-day-old plants using RNeasy Plant Mini Kit. Gene chip ATH1 (Affymetrix, Santa Clara, CA, USA) was used as the probe array DNA for microarray analysis according to the manufacturer's protocol.
DNA sequences, which are similar to ROT4, were obtained by tblastn searches at National Center for Biotechnology Information (NCBI) (http://www.ncbi.nlm.nih.gov/BLAST/). The deduced amino acid sequences of the conserved region of ROT4 and the retrieved DNA sequences were aligned using the clustalw program (Thompson et al., 1994), and alignments were refined manually. For construction of a maximum-likelihood (ML) tree, a neighbor-joining (NJ) tree as the start tree was used for a local re-arrangement search. Then, the njdist and protml programs were used in MOLPHY Version 2.3b3 package (http://www.ism.ac.jp/software/ismlib/softother.html; Adachi and Hasegawa, 1996). The NJ tree was obtained with njdist and the ML tree was obtained with protml. The local bootstrap probability of each branch was estimated using the protml program (Himi et al., 2001; Sakakibara et al., 2001).
Isolations of rtfl4-1, osrtfl1-1, and osrtfl2-1
To find knockout mutants of Arabidopsis and rice for the cording region of RTFL members, BLASTN searches were performed at TAIR BLAST (http://www.arabidopsis.org/Blast/) or Mutant panel (http://tos.nias.affrc.go.jp/miyao/pub/tos17/). In Arabidopsis, only one mutant that had an insertion of T-DNA in the coding region of a RTFL family member was found in the SALK T-DNA insertion collection, namely accession SALK_089234, which carries an insertion in RTFL4. In O. sativa L., two mutants with a Tos17 transposon insertion in the cording region of RTFL4 were found from the rice insertion mutant database (Miyao et al., 2003), and these RTFL members were named OsRTFL1 and OsRTFL2, respectively. These candidate knockout mutant populations were kindly supplied by the Ohio Arabidopsis Stock Center (USA) and the Rice Genome Resource Center (NIAS, Japan). Each homozygous mutant was identified by PCR analysis of genomic DNA. To identify homozygotes for the T-DNA insertion in RTFL4, the following oligonucleotides were used: RTFL4-P, 5′-GTTCGTTCACCCAATGGCTC-3′; RTFL4-N, 5′-TGTTTGTGCCAGCAAACGAG-3′; and LBb1, 5′-GCGTGGACCGCTTGCTGCAACT-3′. The primer pair of RTFL4-P and LBb1 specifically amplifies DNA from the insertion allele, whereas the primer pair of RTFL4-P and RTFL4-N was used for identification of the non-insertional allele. Similarly, to identify homozygotes for Tos17 transposon insertions in the OsRTFL1 and OsRTFL2 genes, the following oligonucleotides were used: OSRTFL1-F, 5′-GCCTATGGTTTGGCATTG-3′; OSRTFL1-R, 5′-GCGCCTCGTCGTCGACGT-3′; OSRTFL2-F, 5′-ATGGCATATCCCCTCCTGT-3′; and OSRTFL2-R, 5′-TCGCTCCAGCGGAGCAACA-3′.
The authors thank Drs H. Ichikawa (NIAS, Tsukuba, Japan) and Y. Niwa (Shizuoka Prefectural University, Shizuoka, Japan) for the kind gifts of pSMAB701 and pTH2, respectively. The authors also thank Dr T. Murata (NIBB, Okazaki, Japan) for his help to observe cells using a confocal laser scanning microscope, Drs M. Hasebe (NIBB, Okazaki, Japan) and T. Nishiyama (NIBB, Okazaki, Japan) for their help in the construction of a phylogenetic tree, Dr N. Ishikawa (NIBB, Okazaki, Japan) for isolations of osrtfl1-1 and osrtfl2-1, Ms E. Takabe (NIBB, Okazaki, Japan) for experimental assistance, and Ms K. Coenen (ICMB, Edinburgh) for help with in situ hybridization. The T-DNA insertion mutants at the Arabidopsis RTFL4 locus and at the rice OsRTFL1 and OsRTFL2 loci were obtained from the collection of SALK T-DNA insertion lines (Ohio Arabidopsis Stock Center, USA) and Rice Genome Resource Center (NIAS, Japan), respectively. This study was supported by a Grant-in-Aid for Scientific Research on Priority Areas from the Ministry of Education, Science, and Culture of Japan, and by funds from the Bio-Design Program, Ministry of Agriculture, Forestry, and Fisheries, Japan. J.G. was funded by a university research fellowship from the Royal Society.
Note added in proof
Wen et al. (2004) have also recently described the effects of mis-expression of an RTH family member.