A microRNA–transcription factor module regulates lateral organ size and patterning in Arabidopsis

Authors

  • Clayton T. Larue,

    Search for more papers by this author
    • These authors contributed equally to this work.

    • Present address: United States Department of Agriculture, Agriculture Research Service, Photosynthesis Research Unit and Department of Plant Biology, University of Illinois, Urbana, IL 61801, USA.

  • Jiangqi Wen,

    Search for more papers by this author
    • These authors contributed equally to this work.

    • §

      Plant Biology Division, The Samuel Roberts Noble Foundation, Ardmore, OK 73401, USA.

  • John C. Walker

    Corresponding authorSearch for more papers by this author

*For correspondence (fax +1 573 884 9676; e-mail walkerj@missouri.edu).

Summary

Precise regulatory mechanisms are necessary to properly control the enlargement and patterning of plant lateral organs. However, our understanding of the regulatory modules that govern both of these processes is limited. An emerging theme in plant development is microRNA (miRNA)-mediated gene regulation of transcription factors, including several NAC domain family members such as CUP-SHAPED COTYLEDON2 (CUC2). We uncovered a novel allele of CUC2, cuc2-1D, that revealed important functions of miRNAs and CUC2 in a regulatory module governing lateral organ enlargement and patterning. cuc2-1D carried a single point mutation in the CUC2 miRNA target site, disrupting miRNA targeting. Disruption of the tight balance between CUC2 and its targeting miRNA, miRNA164, led to over-accumulation of CUC2 mRNA and expansion of the CUC2 expression domain. cuc2-1D plants had enlarged vegetative and reproductive lateral organs relative to wild-type plants. Mechanistically, these enlarged organs resulted from an increase in cell proliferation that occurred over a longer developmental time frame relative to wild-type. This organ enlargement was dependent on the receptor-like kinase, ERECTA (ER). This and lateral organ patterning phenotypes in cuc2-1D suggest that miRNA164 and CUC2 are critical regulators of both processes. Therefore, we propose that miRNA164 and CUC2 form a central regulatory module that acts as a governor of lateral organ patterning and expansion.

Introduction

Plants have multiple functionally specialized lateral organs. To optimize fitness, lateral organs must be properly patterned and sized during their development to allow the plant organs to function together. As directing the patterning and enlargement of lateral organs is critical for survival, multiple robust regulatory pathways have evolved to guide their development.

In plants, our understanding of modulators of organ patterning and size is limited, although gene mutations have begun to reveal players in these processes. Hormone signaling-related genes [such as AUXIN RESPONSE FACTOR 2, ARGOS (auxin-regulated gene involved in organ size), ARGOS-LIKE and BRASSINOSTEROID INSENSITIVE1 (Clouse et al., 1996; Kauschmann et al., 1996; Hu et al., 2003, 2006; Schruff et al., 2006)] and transcription factors [such as AINTEGUMENTA and REVOLUTA/INTERFASCICULAR FIBERLESS1 (Talbert et al., 1995; Elliott et al., 1996; Krizek, 1999; Mizukami and Fischer, 2000)] have all been shown to modify lateral organ size. Flowers of plants with mutations in the protein kinase-encoding gene TOUSLED, and genes involved in auxin signaling, such as ETTIN and PINOID, among others, all show disrupted gynoecium development (Roe et al., 1993, 1997; Bennett et al., 1995; Sessions and Zambryski, 1995; Nemhauser et al., 2000). A transcription factor gene, CRABS CLAW (CRC), is important for proper gynoecium development in Arabidopsis thaliana (Alvarez and Smyth, 1999; Bowman and Smyth, 1999). The diversity of these and other genes suggests that regulation of lateral organ development in plants involves multiple players, although much remains to be understood about the underlying molecular mechanisms.

One of the mechanisms used to control gene activity in a wide range of developmental processes in both plants and animals is miRNA targeting of transcripts (Reinhart et al., 2002; Kidner and Martienssen, 2003; Bartel, 2004; Grant-Downton and Dickinson, 2005; Jones-Rhoades et al., 2006). miRNAs are short 20–24 nucleotide fragments generated from larger precursor non-coding RNAs. Mature miRNAs are incorporated into complexes where they direct targeting of mRNAs. Many of the miRNA-targeted genes in plants are transcription factors that function in developmental processes.

CUC2 is one of several Arabidopsis NAC domain transcription factors that has been shown to be miRNA-targeted (Kasschau et al., 2003; Laufs et al., 2004; Mallory et al., 2004a). Loss-of-function mutants of CUC2 in combinations with loss-of-function mutants of other family members, CUC1 or CUC3, have indicated important and partly redundant roles for these genes in maintenance of the shoot apical meristem (SAM) and development of boundary domains in emerging lateral organs (Aida et al., 1997; Vroemen et al., 2003; Hibara et al., 2006). The double mutant, cuc1 cuc2, arrests shortly after germination (Aida et al., 1997). The two embryonic cotyledons fuse into a cup-shaped structure and the SAM is missing. Loss-of-function mutants in the three-member miRNA164 family targeting CUC2 have also revealed important roles for CUC1 and CUC2. Loss-of-function of miRNA164c results in additional petals in early flowers (Baker et al., 2005), and loss-of-function of miRNA164a disrupts leaf margin patterning (Nikovics et al., 2006). Triple loss-of-function mutants of all three family members alter silique arrangement on the flowering stalk and completely inhibit valve fusion in developing siliques (Sieber et al., 2007). This demonstrates that CUC2 and miRNA164 are important regulators of many aspects of Arabidopsis development.

In a screen for modifiers of Arabidopsis fruit development, we identified a mutant, cuc2-1D, with disrupted fruit patterning and reduced fertility. The mutation was mapped to the miRNA target site of the Arabidopsis CUC2 locus. This mutation at the target site disrupted the regulatory balance between miRNA164 and CUC2. In addition to roles in lateral organ patterning, this unique miRNA target site mutant allowed us to uncover a novel role for miRNA164 and CUC2 in the regulation of lateral organ enlargement, and obtain a mechanistic understanding thereof. The data presented here suggest that miRNA164CUC2 forms a central regulatory module through which both lateral organ patterning and enlargement are governed.

Results

A mutation in the transcription factor gene CUC2 results in disrupted fruit development

During screening of an EMS-mutagenized population of Arabidopsis for genes that regulate fruit (silique) development, a mutant displaying disrupted silique patterning was isolated. The siliques were shortened and curled at the distal end, and thin tooth-like projections were found along the valve margins (Figure 1a). The mutant also showed reduced seed set (Table S1), suggesting an important function in reproduction.

Figure 1.

 Phenotypes of cuc2-1D plants and diagram of the CUC2 gene.
(a) Siliques of cuc2-1D and recapitulated cuc2-1D (Col-0 carrying a cuc2-1D transgene) plants showed abnormal silique development, with misshaped siliques and thin tooth-like projections along the valve margins, indicated by arrows.
(b) Map-based cloning uncovered a single mutation in the miRNA target site of CUC2. A schematic diagram of CUC2 is shown with exons (shaded boxes), the translational start site (arrow) and the miRNA target site (black box, not to scale). The target site is boxed in the enlarged sequence fragment. The mutated sites in cuc2-1D and the cuc2-2D allele are shown in bold and underlined. Scale bar = 1 cm.

The mutant was backcrossed to Columbia wild-type (Col-0) four times to remove extraneous EMS-induced mutations. During backcrossing, the mutant behaved in a semi-dominant manner; the fruit phenotype occurred in homozygous mutant plants, and the leaf lobation phenotype (Figure 2) was apparent in both heterozygous and homozygous plants. Additionally, patterning of the inflorescence stalk was disrupted; irregular internode elongation occurred between flowers, with clustered flowers separated by long internodes (Figure S1).

Figure 2.

 The cuc2-1D plants had an enlarged and lobed-leaf phenotype.
(a–c) Rosette leaves from Col-0 (a), cuc2-1D (b) and cuc2-1D er (c) are arranged in order (left to right) from oldest to youngest. The far right leaf is the first cauline leaf. Scale bar = 1 cm.
(d) Quantification of leaf areas confirmed an enlargement of the rosette leaves in the presence of both cuc2-1D and cuc2-2D transgenes. Addition of a mutated miRNA164a that corresponds to the cuc2-1D mutation or of a loss-of-function er allele returned the leaf area to a nearly wild-type size. The sample size was 20 for all except the three transgenic lines (Col-0 cuc2-1D, Col-0 cuc2-2D, and cuc2-1D miRNA164a C/T G/A), for which = 10. Error bars indicate SEM.

Map-based cloning isolated the mutation responsible for these phenotypes. Mutant plants were out-crossed with Landsberg erecta (Ler), and a F2 population was generated. The mutation was mapped to a roughly 60 kb region between the markers cer434898 and cer438127 on chromosome 5 containing about 20 annotated genes (http://www.arabidopsis.org). By direct sequencing, we uncovered a single mutation (C→T) in the miRNA164-targeted NAC domain transcription factor gene, CUC2 (Figure 1b) (Aida et al., 1997; Laufs et al., 2004). As this mutation was semi-dominant, we named it cuc2-1D. No additional mutations were detected at this locus.

We next confirmed that this mutation was responsible for the fruit phenotype. As this mutation was semi-dominant, the cuc2-1D phenotype was recapitulated by generating CUC2 transgenes that were stably transformed into Col-0. The transgenes contained a wild-type or cuc2-1D version of the genomic CUC2 locus. Only plants carrying the cuc2-1D transgene showed cuc2-1D phenotypes (Figures 1a and S2). These results confirm that the single mutation in CUC2 is responsible for the fruit developmental patterning defects.

Lateral organ enlargement in cuc2-1D

Careful examination of homozygous cuc2-1D plants showed that CUC2 acts as a positive regulator of lateral organ size. Rosette leaves exhibited increased medial–lateral expansion and had a greater leaf surface area (Figure 2 and Table S1). To determine whether the increased expansion was due solely to an outgrowth of leaf lobes, we also measured leaf width at the sinus (between the lobes). The cuc2-1D leaves were still significantly wider than Col-0 controls, suggesting a general enhancement of leaf expansion in cuc2-1D. This lateral organ enlargement was also observed in the recapitulated cuc2-1D plants (Figure 2d and S2). This confirms that the cuc2-1D mutation resulted in increased expansion of a vegetative lateral organ.

In addition to the role of CUC2 as a regulator of vegetative lateral organ growth, cuc2-1D reproductive lateral organs were also larger relative to Col-0. Flower pedicels (stalks supporting siliques), petals and inflorescence stems were all more elongated in cuc2-1D relative to Col-0 (Figures 3, 4 and S2, and Table S1). The primary inflorescence stalks of cuc2-1D were nearly 7 cm taller than those of Col-0 at maturity. Siliques were significantly shorter in cuc2-1D than in Col-0 (Table S1), and an increase in cuc2-1D gynoecium diameter was observed in scanning electron micrographs (Figure 5d,e). However, mis-coordination of growth at the carpel ends could mask changes in gynoecium extension. These data suggest that CUC2 acts as a global regulator of vegetative and reproductive lateral organ enlargement.

Figure 3.

 Pedicel phenotype.
(a) Pedicels were elongated (highlighted by white bars) in cuc2-1D plants.
(b) Quantification of pedicel size confirmed the longer pedicels in plants carrying both cuc2-1D and cuc2-2D transgenes, while addition of a mutated miRNA164a that corresponds to the cuc2-1D mutation or of a loss-of-function er allele resulted in reduced pedicel extension.
(c, d) Longitudinal sections of pedicel cortex cells showed similar-sized cells in both Col-0 and cuc2-1D pedicels and larger cells in cuc2-1D er pedicels (c), and this was confirmed by quantification of the cross-sectional cell areas (d).
The sample size was 40 in (b) and 100 in (d). Error bars indicate SEM. Scale bars = 0.5 cm (a) and 25 μm (c).

Figure 4.

 Petals were enlarged in cuc2-1D.
(a, b) The cuc2-1D plants had larger petals in comparison to Col-0.
(c) The cells of the cuc2-1D petals were smaller than those of Col-0 as shown by a greater number of cells along randomly drawn lines on the petal surface.
(d, e) The cuc2-1D petal buds were not larger in comparison to Col-0 before initiation of rapid growth in stage 9 (stage 8 shown).
(f) The cuc2-1D petals had an extended period of cell proliferation as shown using a cell-cycle marker line. In Col-0 petals, a burst of cell proliferation occurs at stage 9 (when rapid growth begins), which quickly subsides. In cuc2-1D petals, the burst of cell proliferation extended into stage 10 before subsiding.
The sample size was 37 in (b), 5 in (c) and ≥10 in (f). Error bars indicate SEM. Scale bars = 100 mm (a) and 25 μm (d, e).

Figure 5.

 Developmental time series of Col-0 and cuc2-1D gynoecium development.
(a) A lack of complete carpel fusion was seen shortly after carpels began to elongate, as seen in late stage 7/early stage 8 cuc2-1D flowers.
(b) This lack of fusion on the distal end of the gynoecium persisted through stage 9.
(c) The unfused zone remained in stage 11 flowers; however, it was filled with growing stigmatic papillae at this time. At stage 11, small protrusions of tissue (first appearing in late stage 9/early stage 10) were clearly visible along the carpel margins. In addition, a larger gynoecium diameter was apparent.
(d) By stage 12, developing ovules could be seen through the unfused region of the gynoecium. Note the inward curling of the stigmatic surface and enlargement of the tissue projections along the carpel margin, sometimes developing a stigmatic surface.
(e) These features became more predominant during stages 13/14.
(f) The siliques retained this incomplete valve fusion at the distal end, and projections along the valve margin.
Arrows indicate the fusion plane between the two carpels/valves. Scale bars = 50 μm (a), 100 μm (b), 150 μm (c), 250 μm (d), 500 μm (e) and 750 μm (f).

cuc2-1D shows enhanced cell proliferation

Increasing cell number, cell size or a combination of both can result in larger organs. Therefore, we determined the means by which cuc2-1D lateral organs were enlarged. The distal surface of Arabidopsis petals is composed of symmetrical cobblestone-like cells (Bowman, 1994). cuc2-1D petals contain smaller cells than Col-0 petals (Figure 4c). This suggests that CUC2 acts to promote cell division rather than cell expansion. We further confirmed this increased cell proliferation by examining longitudinal sections of the pedicels (Figure 3c). Measurements of cortex cell areas in these longitudinal sections also demonstrated the cells are not larger, further supporting the hypothesis that cuc2-1D acts to increase cell proliferation rather than cell expansion.

To further investigate the molecular mechanism underlying this lateral organ enlargement, we performed an EMS suppressor screen in the cuc2-1D background. A partial cuc2-1D suppressor was isolated, in which rosette leaf enlargement was restored to a nearly Col-0-like phenotype, with reduced rosette leaf lobation (Figure 2). However, the gynoecium-patterning defects were retained (Figure S3). This suppresser had an early stop codon mutation in the 7th exon of the Arabidopsis receptor-like kinase, ERECTA. er loss-of-function mutants show reduced cell proliferation, resulting in dwarf plants with smaller lateral organs (Torii et al., 1996). To compensate for this reduced cell proliferation, the pedicel cortex cells were larger than in ER controls (Shpak et al., 2004, 2003b). Therefore, we examined cortex cells in a cuc2-1D er double mutant. Longitudinal pedicel sections showed enlarged cell areas relative to controls (Figure 3c). As the pedicels were shorter in the cuc2-1D er double mutant relative to cuc2-1D (Figure 3b), this confirms the hypothesis that cell proliferation was restricted, and thus lateral organs are not able to enlarge as in the cuc2-1D single mutant. These data suggest that the increased lateral organ size in cuc2-1D is dependent on increased cell proliferation, mediated through an ER-dependent pathway.

While these data suggest that increased cell proliferation resulted in enlarged organs, an alternative hypothesis could be that emerging lateral organs recruit additional cells to the primordia. Thus the increased cell number in these organs is merely due to a greater number of founder cells in the lateral organ primordia. To address this possibility, we examined scanning electron micrographs of petal buds in developing flowers before the petals began rapid growth (Figure 4d,e). These petal buds are similarly sized in Col-0 and cuc2-1D backgrounds, suggesting that the petal buds are initiated similarly and that the increased organ enlargement does not occur until a later developmental stage.

cuc2-1D shows an extended period of cell proliferation

We reasoned that cell proliferation could occur at a more rapid rate or proliferation could extend over a longer developmental time frame. To address these alternative explanations and further characterize the mechanism by which the lateral organs are enlarged, we tracked mitotic activity through a cyclin:GUS reporter line. This reporter line contains an additional element resulting in rapid turnover of GUS protein, and thus only cells actively undergoing cell cycling are marked (Colon-Carmona et al., 1999). GUS activity was tracked in both Col-0 and cuc2-1D by dissecting flower petals at various developmental stages and quantifying the number of stained cells. Arabidopsis petals begin a period of rapid expansion after flower stage 8 (Smyth et al., 1990). Therefore, we tracked petals from stages 9–12. Both Col-0 and cuc2-1D flower petals showed active cell division at stage 9 (Figure 4f). Cell division activity in Col-0 petals quickly diminished, and was nearly complete by flower stage 11. However, in cuc2-1D petals, active cell division continued over a longer developmental time frame, and was not complete until flower stage 12. This suggests that the increased cell proliferation observed in cuc2-1D is a result of a developmentally extended period of cell proliferation, rather than a more rapid rate of cell proliferation.

Lack of proper miRNA targeting results in the developmental phenotypes observed in cuc2-1D

The phenotypes observed in cuc2-1D were completely different from those of previously reported cuc mutants (Aida et al., 1997; Takada et al., 2001). Unlike the cuc mutants that affect NAC domain function, the C→T mutation in cuc2-1D was located in the CUC2 miRNA target site (Figure 1b). As the mutation occurs in this site, we hypothesize that miRNA targeting may be disrupted, allowing CUC2 mRNA to over-accumulate. The mutation creates a G–U wobble basepair (bp) near the 5′ end of the miRNA, a region that requires strong base pairing for proper miRNA function (Mallory et al., 2004b). Therefore, steady-state CUC2 mRNA levels were examined. CUC2 mRNA levels were determined by quantitative RT-PCR in unopened flower buds. An approximately eightfold over-accumulation of the CUC2 mRNA in cuc2-1D relative to Col-0 was observed (Figure 6a). In addition, the recapitulation line in which a cuc2-1D transgene was transformed into a Col-0 background also showed over-accumulation of CUC2 mRNA (Figure 6a). This suggests that disruption of the miRNA target site in cuc2-1D results in CUC2 mRNA over-accumulation.

Figure 6.

CUC2 mRNA expression.
(a) Quantitative RT-PCR showed a nearly eightfold over-accumulation of CUC2 mRNA in flowers from cuc2-1D. CUC2 expression in the transgenic lines and their respective controls, grouped by vertical lines, were also tested. Error bars indicate SEM. Three biological replicates were used.
(b) A mutated version of miRNA164 was used to complement the cuc2-1D plants. The mature miRNA is boxed and the mutated sites in the mature miRNA164 and its complementary region of the hairpin loop are shown in bold and underlined.
(c) This mutated miRNA restored base pairing with the mutant cuc2-1D target site and complemented the cuc2-1D phenotypes (silique phenotype shown here, see also Figures 2 and 3, and S2). Scale bar = 1 cm.

To further investigate the hypothesis that a loss of miRNA targeting results in the cuc2-1D phenotype, a mutated miRNA164a transgene was constructed. This transgene contained two mutations in miRNA164a. One mutation (C→T) restores miRNA–target gene base pairing at the cuc2-1D target site mutation. The other mutation (G→A) restores base pairing within the hairpin loop structure of the pre-miRNA (Figure 6b). Col-0 and cuc2-1D were stably transformed with this construct. No phenotypic changes were observed in the transgenic Col-0 plants, but transgenic cuc2-1D plants reverted to a Col-0-like phenotype, with restored patterning and lateral organ enlargement (Figures 2d, 3b and 6c, and S2). Due to restored miRNA targeting, these rescued cuc2-1D plants showed no significant over-accumulation of CUC2 mRNA in flower buds relative to Col-0 (Figure 6a). An additional control comprising an miRNA164 wild-type version driven by a CaMV 35S promoter was unable to rescue the cuc2-1D phenotype. These data suggest that lack of sufficient binding by miRNA164 to the CUC2 target site disrupts this regulatory balance, and thus the CUC2 mRNA over-accumulates resulting in the mutant phenotype.

The cuc2-1D mutation modifies an amino acid (AA) in its encoded protein (serine → phenylalanine). To investigate this AA change, a third recapitulation construct was produced, generating an additional miRNA-disrupted CUC2 allele, cuc2-2D. Instead of the cuc2-1D mutation, a mutation was generated in a wobble base, 1 bp downstream from the cuc2-1D mutation site, to disrupt miRNA targeting but not the encoded AA sequence (Figure 1b). In Col-0, this transgene resulted in over-accumulation of the CUC2 transcript and a cuc2-1D-like phenotype with enlarged lateral organs and disrupted patterning (Figures 2d, 3b, 6a and S2). The results obtained using this additional allele suggest that the cuc2-1D phenotype is due to disrupted miRNA targeting rather than the AA change.

miRNA164 attenuates CUC2 mRNA and fine-tunes its expression domain

To define cellular interactions between CUC2 and miRNA164, RNA in situ hybridization experiments were preformed on floral meristems and unopened flower buds. Over-accumulation of CUC2 mRNA in cuc2-1D could be due to mis-expression outside the native expression zone, increased expression within the native expression zone or a combination of these two scenarios. The CUC2 hybridization signal in Col-0 meristem tissue was similar to that in previous reports (Ishida et al., 2000; Baker et al., 2005), with strongest expression just acropetal to emerging lateral organs (Figure 7a). In the gynoecium, expression was strongest in developing ovules and the septum (Figure 7c). cuc2-1D tissue had an enhanced signal in native expression domains (Figure 7b,d). In addition, there was expansion of the expression domain, but little mis-expression in new domains (Figure 7b,d).

Figure 7.

CUC2 RNA in situ hybridization analysis.
CUC2 RNA in situ hybridization was completed using flower tissue and apical meristem tissue shortly after transition to the floral state. In meristem tissue, CUC2 was expressed in the apical dome of cells, and the strongest staining was observed near the nodes of emerging primordia [representative images are shown in (a) and (b)]. cuc2-1D showed stronger staining in the native expression domain, with a small expansion of the expression domain. In the developing gynoecium, CUC2 was expressed in the developing ovules and septum tissue (c, d). The staining in cuc2-1D was more intense in the septum tissue, and an expansion of the expression domain was observed near the base of some developing tissue projections along the carpel margins (arrows) and at an earlier stage when the projections began to swell (insert). Note the disruption in septum development towards the distal end of the cuc2-1D gynoecium [inserts in (c) and (d)].

miRNAs can act by establishing expression domains spatially or temporally, or by attenuating the level of targeted mRNA or protein present in already established domains. The staining patterns in cuc2-1D suggest that the role of miRNA164 is to dampen the accumulation of CUC2 mRNA, but not to define CUC2 expression domains on a global scale. This dampening results in the refinement of expression domains, but does not establish domains, similar to the results seen using miRNA164 mutants (Sieber et al., 2007). Interestingly, increased cellular proliferation, at least in petals, occurred away from the locations ofCUC2 expression maxima. Indeed, the CUC2 expression maxima for developing lateral organs often occurred just acropetal to the organ node rather than within the organ proper (Figure 7). This suggests that the miRNA164–CUC2 module may influence developing lateral organs by pre-determining the extent of cellular proliferation during the primordial stages or through an intermediate signal.

CUC2 and miRNA164 function in patterning of developing fruit

Multiple aspects of gynoecium patterning are perturbed in cuc2-1D. Thin tissue projections along valve margins and curled silique ends were accompanied by incomplete valve fusion at the distal silique ends (Figure 5f). The style region was disrupted and partly engulfed by an overgrowth of valve tissue that curled inward at the silique ends.

To characterize the time course over which these phenotypes developed, flowers at various developmental stages were examined. The gynoecium begins to grow as a hollow tube from the floral meristem during flower stage 7 (Smyth et al., 1990). This tube houses developing ovules that later mature as seeds following fertilization. The gynoecium is derived from two modified leaf-like structures called carpels (Bowman, 1994). Carpels mature as valves, the walls of the silique, in maturing seed pods. In Col-0 flowers, these two carpels fuse as they arise from a central dome of cells on the meristem. However, the distal end of the cuc2-1D gynoecium was unfused throughout flower development (Figure 5a–e). In stage 11–12 flowers, the stigmatic papillae wrapped around the interior edge of the unfused cuc2-1D carpels in a crescent-like pattern instead of the uniform hat-like structure seen in Col-0 (Figure 5c,d). Developing ovules were exposed in the unfused region of the cuc2-1D gynoecium (Figure 5d,e). Therefore, we conclude that cuc2-1D patterning defects begin early in gynoecium development and progress throughout flower development.

Carpel fusion results in a region called the replum. The replum from both fusion zones is connected by a thin tissue bridge called the septum, transversing the hollow gynoecium. The carpel boundaries formed a smooth seam in Col-0, but tissue proliferation resulted in a ridge along the carpel boundary/replum in cuc2-1D stage 9 flowers, becoming finger-like projections as flower development continued (Figures 5c–e and S2), often developing stigmatic ends with papillae (Figure 5e). Additionally, septum development was incomplete at the distal end of the gynoecium (Figure S4). Together, these patterning defects result in a semi-sterile phenotype. The seed number per silique in cuc2-1D was significantly reduced, while the number of siliques per plant remained unchanged relative to Col-0 (Table S1). These results suggest that miRNA164 and CUC2 are involved in patterning of the developing gynoecium, ultimately affecting plant fertility.

cuc2-1D genetically interacts with CRC

Plants with loss-of-function of a YABBY domain transcription factor, CRC, show incomplete carpel fusion at the distal silique ends, similar to cuc2-1D (Alvarez and Smyth, 1999; Bowman and Smyth, 1999). To investigate genetic interactions between CRC and CUC2, the double mutant cuc2-1D crc was created. Double mutant plants had a synergistic phenotype in which carpel fusion was completely abolished (Figure 8), resulting in nearly sterile plants. However, this sterile phenotype was not due to defective ovule formation. Ovules were observed along the unfused carpel margins. These siliques also did not appear to display polarity defects. Close inspection of these unfused carpels showed ovules clearly initiated from the inner surface of the carpel margins (Figure S5). Additionally, occasional small siliques in which carpels were sufficiently fused produced a few viable seeds, demonstrating that seed development can proceed in this double mutant.

Figure 8.

cuc2-1D and CRC showed a synergistic interaction.
(a, d) Col-0 flower (a) and silique (d).
(b, e) crc-1 lacks carpel fusion at the distal end of the gynoecium (b) and siliques (e).
(c, f) In the double homozygous mutant cuc2-1D crc-1, carpels were completely unfused in the gynoecium (c) and siliques (f).
Scale bars = 400 μm (a–c) and 750 μm (d–f).

As both CUC2 and CRC are transcription factors, we checked the CUC1 and CUC2 expression levels in a crc background and the CRC expression level in a cuc2-1D background. However, no significant changes in expression were noted in flower clusters, suggesting an indirect interaction (Figure S6). To check CRC expression patterns in cuc2-1D, we examined tissue expression patterns using RNA in situ hybridization. CRC is expressed towards the lateral region of a young developing flower gynoecium (Bowman and Smyth, 1999). In slightly older flowers, two zones of expression can be seen in the carpels. CRC expression patterns were found to be unchanged in cuc2-1D flowers (Figure S7). This further suggests that CUC2 may interact in a parallel pathway, perhaps with a downstream factor common to CRC rather than by direct interaction with CRC. As crc is a loss-of-function allele and cuc2-1D is a gain-of-function allele, this suggests that miRNA164–CUC2 and CRC act to regulate fruit patterning in opposing fashions.

Discussion

In a screen for fruit development mutants, we uncovered a novel semi-dominant allele of the NAC domain transcription factor, CUC2, which we named cuc2-1D. The point mutation in cuc2-1D disrupted targeting by miRNA164, resulting in an over-accumulation of CUC2 mRNA. A mutated miRNA164 transgene designed to restore CUC2 target site base pairing reverted the cuc2-1D phenotype to a Col-0-like phenotype. An miRNA transgene containing only the mutation to restore miRNA target site base pairing, but no mutation to restore the pre-miRNA hairpin loop structure, or a transgene containing wild-type miRNA164 could not revert cuc2-1D to a Col-0-like phenotype (data not shown). These data confirm that the cuc2-1D phenotype is the result of disrupted miRNA targeting.

CUC2 and miRNA164 regulate lateral organ size in Arabidopsis

A surprising and interesting phenotype in the miRNA164-resistantcuc2-1D mutant was an enlargement of lateral organs. CUC2 has been proposed to function by limiting cell division and growth (Nikovics et al., 2006; Peaucelle et al., 2007; Sieber et al., 2007). The lobed leaf phenotype of miRNA164a null mutants was suggested to be due to restricted cell proliferation in the leaf sinus (Nikovics et al., 2006). Plants carrying CaMV 35S promoter-driven miRNA-resistant CUC2 transgenes had smaller organs (Sieber et al., 2007). However, analysis of the cuc2-1D mutant, which has a mutation in the endogenous CUC2 gene, suggests that CUC2 acts to promote cellular proliferation in post-embryonic lateral organs. Rosette leaves, pedicels, flower petals and inflorescence stems were enlarged relative to Col-0. The replum tissue was enlarged, with a ridge or small finger-like projections protruding from the zone of valve fusion. These phenotypes could be recapitulated using transgenes carrying the cuc2-1D mutation or an additional allele, cuc2-2D. The Col-0 phenotype could be recapitulated using mutant miR164a transgenes in a cuc2-1D background. Therefore, disrupted miRNA164 targeting and over-accumulation of CUC2 transcript result in enhanced enlargement of multiple lateral organs. These data suggest that a fine balance between CUC2 and miRNA164 activity must exist to properly guide plant development.

The cuc2-1D phenotypes probably differ from those described in an earlier report that analyzed the role of miRNA164 targeting of CUC2 (Sieber et al., 2007) due to differences in expression. In cuc2-1D, the native spatial-temporal regulatory elements are present. In previous work using 35S promoter-driven miRNA-resistant CUC2 transgenes, the smaller organs may be due to CUC2 expression outside its native expression domain. Although miRNA-resistant CUC2 transgenes have not been investigated, the native CUC2 expression domains overlapped with zones of reduced cellular proliferation relative to neighboring cells (Breuil-Broyer et al., 2004). However, the increased cellular proliferation shown in cuc2-1D is outside this zone of reduced cellular proliferation. In contrast to gain-of-function cuc2-1D plants, loss-of-function of cuc1 and cuc2 in a double mutant results in reduced cellular proliferation in the SAM, causing the SAM to be lost, and seedling arrest (Aida et al., 1997). These double mutant plants can be regenerated through callus culture, and flowers from these plants have absent or very small petals (Aida et al., 1997). Although fused cotyledons of the double cuc1 cuc2 mutant seedlings have been suggested to be due to over-proliferation of boundary cells, boundary domains in plants carrying a CUC2 transgene with a disrupted miRNA binding site are enlarged, and this has been proposed to be due to increased cellular proliferation (Laufs et al., 2004; Baker et al., 2005).

RNA in situ hybridization experiments showed CUC2 transcript over-accumulation and a small enlargement of the expression domain in the SAM of cuc2-1D. If CUC2 acts to inhibit cellular proliferation, CUC2 over-accumulation is expected to restrict the ability of the SAM to generate lateral organs, reminiscent of wuschel mutants (Laux et al., 1996). However, the silique number in cuc2-1D was unchanged, suggesting that SAM activity was not inhibited. Thus, we propose that the miRNA164–CUC2 regulatory module governs cellular proliferation in multiple tissues. In miRNA-resistant lines, CUC2 over-stimulates cellular proliferation, resulting in enlarged boundary domains and enlarged lateral organs. Therefore, CUC2 is a positive regulator of plant growth.

The cuc2-1D lateral organ enlargement was due to increased cellular proliferation and was dependent on a functional ER signaling pathway. ER also plays a role in cellular proliferation (Torii et al., 1996; Shpak et al., 2003), suggesting that cuc2-1D phenotypes may be mediated through an ER-dependent pathway. However, ER is probably not a downstream target of miRNA164–CUC2. Quantitative RT-PCR of floral and vegetative tissues found no changes in ER expression in cuc2-1D relative to Col-0, suggesting an indirect connection. Interestingly, perturbations in cuc2-1D gynoecium patterning were independent of the presence of a functional ER pathway. Thus the miRNA164–CUC2 regulatory module governs several aspects of development via multiple downstream pathways.

Mechanistically, this enhanced cellular proliferation was due to an extended period of proliferation rather than an increased proliferation rate. This suggests that miRNA164–CUC2 plays a specific role in sensing/determining when a lateral organ has reached the proper number of cells. miRNA164–CUC2 does not appear to regulate the rate of cellular proliferation or enlargement. As organ growth continues through cellular enlargement after cell cycling has been completed, this suggests that miRNA164–CUC2 specifically acts earlier in lateral organ development to help regulate the ultimate organ size. Additional regulatory modules must act at later stages. Mutants with reduced cellular proliferation are able to compensate, at least in part, through increased cellular enlargement (Wang et al., 2000; Shpak et al., 2003, 2004). Several other players also influence the cellular proliferation checkpoint, including DA1, KLUH/CYP78A5, BIG BROTHER and ANT (Mizukami and Fischer, 2000; Disch et al., 2006; Anastasiou et al., 2007; Li et al., 2008), demonstrating the importance of this checkpoint in regulating lateral organ enlargement.

CUC2–miRNA164 functions in lateral organ patterning

Use of cuc2-1D, a native CUC2 miRNA target site mutant, enabled us to gain important new insights and obtain a detailed characterization of the role of the miRNA164–CUC2 module in plant developmental patterning. Rosette leaves develop a lobed margin and patterning on flowering stalks is disrupted, similar to previous reports (Nikovics et al., 2006; Peaucelle et al., 2007; Sieber et al., 2007) using loss-of-function mutants in the miRNA164 family or CUC2 transgenes containing miRNA target site mutations. cuc2-1D silique patterning of the style, replum, valve ends and septum was also disrupted, suggesting that miRNA164–CUC2 functions in governing their growth. Alternatively, this disruption may be secondary to a mis-coordination of valve tissue growth. Patterning defects were observed in multiple lateral organs, both vegetative and reproductive, in cuc2-1D, suggesting that miRNA164–CUC2 is a critical regulator in patterning of plant lateral organs.

cuc2-1D lacked carpel fusion only at the distal end of the gynoecium. A triple loss-of-function mutant of the miRNA164 family and lines expressing a CUC2 transgene with four point mutations in the miRNA target site resulted in complete loss of carpel fusion (Nikovics et al., 2006; Sieber et al., 2007). These situations differ from that in cuc2-1D as we never observed any cuc2-1D plants with a completely unfused gynoecium. Loss-of-function of all miRNA164 family members leads to complete shutdown of miRNA164 regulation, resulting in over-accumulation of other miRNA164-targeted genes in addition to CUC2 (Sieber et al., 2007). Thus, a synergistic interaction between several miRNA164-targeted genes could result in completely unfused carpels. The modified CUC2 transgenes used by Nikovics et al. (2006) may have resulted in slightly different or stronger disruptions of flower development than in cuc2-1D, which contains a mutation in the native CUC2 locus. Taken together, we conclude that disrupted miRNA164 targeting of CUC2 inhibits carpel fusion in the developing gynoecium, together with additional players that are not yet fully known.

Mutants of the YABBY domain family member, CRC, also lack carpel fusion at the distal gynoecium end (Alvarez and Smyth, 1999; Bowman and Smyth, 1999). Therefore, we investigated the genetic relationship between CRC and miRNA164–CUC2 by generating the double mutant cuc2-1D crc. In the double mutant, we observed a nearly complete loss of carpel fusion. This synergistic interaction suggests that miRNA164–CUC2 and CRC probably act in parallel pathways. However, the CRC mutant is a loss-of-function mutant, while cuc2-1D is a gain-of-function mutant, suggesting that they act in opposing fashions: CUC2 limits carpel fusion while CRC promotes carpel fusion. A proper balance between these two pathways is required to regulate carpel fusion correctly.

Experimental procedures

Generation of transgenic plants

Plants were grown in standard growth chambers or greenhouses in peat-based media with 16 h supplemental lighting. Homozygous cuc2-1D plants were used for all experiments, although heterozygous plants did show partial organ patterning and enlargement phenotypes. Transgenic lines were generated by the floral-dip transformation method (Clough and Bent, 1998) using pCAMBIA binary vectors (Cambia, http://www.cambia.org). cuc2-1D recapitulation constructs used the genomic CUC2 locus, including 1 kb of the region 5′ of the start codon and 1 kb of the region downstream of the stop codon, without additional vector promoter elements. For each construct, 80 T1 transgenic lines were screened. cuc2-1D phenotypes were observed for both the cuc2-1D and silent cuc2-1D recapitulation constructs in 62–70% of the transgenic plants. Complementation utilized a mutated miRNA164 containing 1 kb of the genomic miRNA164a locus driven by a CaMV 35S promoter in a pCAMBIA binary vector. For each construct, 100 T1 transgenic lines were screened. In the miRNA164 construct containing both point mutations, 70 lines had Col-0-like phenotypes. Representative lines of all transgenic plants were chosen for the more detailed characterization as shown in the figures. Unless noted, all constructs were cloned using SmaI restriction sites. Point mutations were created using site-directed PCR mutagenesis in a pBluescript SK vector. Primers are listed in Table S2.

Quantitative RT-PCR

Quantitative RT-PCR was completed using three independent biological replicates. Extracts were produced using TRIzol reagent (Invitrogen Life Technologies; http://www.invitrogen.com/) according to the manufacturer’s instructions. Contaminating DNA was removed by DNase treatment, and cDNA was produced using 1 μg of total RNA. Quantitative RT-PCR was performed using ABsolute QPCR SYBR Green Premix (ABgene, http://www.abgene.com) according to the manufacturer’s instructions on an Optican 2 real-time PCR machine (MJ Research, http://www.bio-rad.com). Sample Ct values were log-transformed and standardized to an EF1α control reaction. Primers are listed in Table S2. For CUC2, primer set A does not span the miRNA target site. As a check, primer set B was generated to span this site. A similar trend was observed with both primer sets, and only the data obtained using primer set A are presented.

Mutant generation

Mutant populations were produced by imbibing 0.2 g of seed in 0.2% ethyl methanesulfonate (EMS, Sigma; http://www.sigmaaldrich.com/) for 15 h. M2 populations were screened, and mapping populations were generated by outcrossing with Ler. Mapping was completed by scoring the fruit phenotype in 500 plants; this phenotype was not dependent on a functional ER. The mutation was mapped to a region of approximately 60 kb, and sequencing was used to identify the mutant site. In the EMS screen for suppressors of the cuc2-1D phenotype, a mutant population was generated as before and M2 plants were screened. Quantitative RT-PCR was used to confirm that CUC2 mRNA over-accumulated similarly in cuc2-1D er and cuc2-1D.

RNA in situ hybridization

RNA in situ hybridization was completed essentially as described previously (Lincoln et al., 1994; Vielle-Calzada et al., 1999) and summarized here. The CUC2 probe was produced using approximately 0.5 kb of the 3′CUC2 coding sequence as described previously (Aida et al., 1999). The CRC probe was produced using the entire CRC coding region (approximately 0.5 kb). In vitro transcription used DIG RNA labeling premix and T3/T7 RNA polymerase according to the manufacturer’s instructions (Roche; http://www.roche.com) and hydrolyzed to approximately 300 bp fragments. Tissue was fixed in 4% paraformaldehyde and embedded in Paraplast (McCormick Scientific, http://www.mccormickscientific.com). Sections were cut to approximately 10 μm thickness, and attached to Colorfrost Plus slides (Fisher Scientific, http://www.fishersci.com). Slides were dewaxed and pre-treated for hybridization. After overnight hybridization, the slides were washed in 0.2× SSC and the RNase digestion step was eliminated. Detection was performed using anti-digoxigenin-AP (Roche) and Western Blue substrate (Promega; http://www.promega.com/). A CUC2 sense negative control probe did not show a specific localization pattern.

Scanning electron microscopy

Samples were fixed in 2% glutaraldehyde, 2% paraformaldehyde in 0.1 m cacodylate buffer and dehydrated in ethanol. Samples were critical point-dried using an Autosamdi 815 automatic critical point drier (Tousimis; http://tousimis.com/), and then mounted on Aluminium stubs for sputter coating with Platinum. Samples were viewed using an S-4700 scanning electron microscope (Hitachi, http://www.hitachi.com). Images for analysis of fresh tissue petal surfaces were obtained using an FEI Quanta 600F scanning electron microscope (http://www.fei.com/) in environmental mode.

Light microscopy

Pedicels were fixed as above and embedded in Epon–Spurr’s resin and sectioned at 2.5 μm on a Leica Ultracut UCT ultramicrotome (http://www.leica.com). Sections were stained with 0.05% toluidine blue and viewed using an Olympus IX70 microscope (http://www.olympus-global.com/) equipped with a Hamamatsu ORCA-ER C4742-80 digital camera (http://www.hamamatsu.com).

Cell proliferation assays

A labile cyclin:GUS reporter line (Colon-Carmona et al., 1999) was crossed into the cuc2-1D background. Flower buds were stained with GUS buffer (10 mm EDTA, 0.1% Triton X-100, 2 mm potassium ferricyanide, 2 mm potassium ferrocyanide, 100 μg ml−1 chloramphenicol, 1 mg ml−1 X-Gluc in 50 mm sodium phosphate buffer pH 7.0) for 8 h at 37°C. Petals were dissected, and the number of stained cells on each petal was recorded. At least ten petals were observed for each stage.

Image analysis

Morphological measurements were performed using FoveaPro version 4.0 (Reindeer Graphics Inc., http://www.reindeergraphics.com) plug-ins in Photoshop (Adobe, http://www.adobe.com) or MetaMorph version 6.3 (Molecular Devices, http://www.moleculardevices.com) image analysis software. Analysis of petal cells was completed by counting the number of cells in scanning electron micrographs of the distal adaxial end of the petal along five randomly generated lines of known length. The same lines were used for five petals of both Col-0 and cuc2-1D. Two-dimensional cell areas were calculated in pedicels by measuring 100 cortex cells just below the epidermal layer in longitudinal sections. Measurements of lateral organs used scanned images of the organs obtained using an Epson Perfection 3170 photo scanner (http://www.epson.com). Leaf measurements were performed on the youngest fully expanded leaf as plants began flowering. Pedicel, silique and whole-plant measurements were performed on mature plants.

Acknowledgements

The authors would like to acknowledge Huachun Wang for her assistance and input in experimentation and pre-reviewing the manuscript. The authors thank Shulan Zhang and Alex J. Gilbert for help with greenhouse work and screening mutant populations. The staff of the Columbia Electron Microscopy Core at the University of Missouri were very helpful in assisting with the scanning electron microscopy. We are grateful to members of the Walker Laboratory for helpful discussions. C.T.L. was supported by a University of Missouri-Monsanto Graduate Research Fellowship. This research was supported by grants from the Department of Energy (DE-FG02-05ER15652) and the University of Missouri Food for the 21st Century Program.

Ancillary