For genetic analysis of mechanisms of leaf morphogenesis, we chose Arabidopsis thaliana (L.) Heynh. as a model for leaf development in dicotyledonous plants. Leaves of the angustifolia mutant were the same length as but narrower and thicker than wild-type leaves. The total number of cells in leaf blades of angustifolia plants was the same as in the wild type. At the cellular level in the angustifolia mutant it was found that the cells were smaller in the leaf-width direction and larger in the leaf-thickness direction than in wild type, revealing the function of the ANGUSTIFOLIA gene, which is to control leaf morphology by regulating polarity-specific cell elongation. The existence of similar genes that regulate leaf development in the length direction was, therefore, predicted. Three loci and several alleles associated with short-leaved mutants were newly isolated as rotundifolia mutants. The rotundifolia3 mutant had the same number of cells as the wild type, with reduced cell elongation in the leaf-length direction. The features of the angustifolia rotundifolia3 double mutant indicated that ANGUSTIFOLIA and ROTUNDIFOLIA3 genes act independently. We propose that leaf expansion in Arabidopsis involves at least two independent developmental processes: width development and length development, with the ANGUSTIFOLIA and ROTUNDIFOLIA3 genes playing different polarity-specific roles in cell elongation.

The ontogenic processes in plants are different from those in animals. The plant system arising from the shoot apex can be divided into fundamental units, known as phytomers (Evans and Grover, 1940). The shoot system is a stack of these units. The most specialized organ in spermatophytes, the flower, can also be understood in this context. The phytomer of the vegetative shoot system can be further divided into three parts: the leaf, the internode, and the lateral bud.

Leaf development involves complex developmental processes (Avery, 1933). The temporal, spatial and pattern differences in cell division and elongation in the lamina contribute to the final shapes of leaves (Maksymowych, 1963; Sunderland, 1960). The pattern of cell growth and division within the lamina is surprisingly complex (Poethig and Sussex, 1985) but the genetic mechanisms regulating each process, such as leaf determination, marginal meristem formation, and polarity recognition, remain unknown.

Our interest is focused on the processes involved in the expansion of the leaf blade. Only a few studies of leaf morphology using mutants have been reported to date (Dale, 1988; Dolan and Poethig, 1991; Harte and Hansen, 1967; McHale, 1993; Röbbelen, 1957). Arabidopsis is a useful model system for studies of plant development because of its small genome, short generation time, self compatibility, amenability to stable transformation, and the availability of numerous mutants. In particular, analysis of flower development has been quite successful in Arabidopsis (Bowman et al., 1991). Analyses have also been initiated of root development (Hauser et al., 1995), embryogenic development (Goldberg et al., 1994), and root morphogenesis under certain environmental conditions (Okada and Shimura, 1990) using this model system.

The leaves of Arabidopsis are small but characteristically shaped at each position, providing suitable materials for anatomical studies (Pyke et al., 1991). We have focused on the processes involved in leaf expansion and have characterized wild-type (wt) plants and mutants of Arabidopsis at the cellular level. The narrow-leaf mutant, angustifolia (an), was originally isolated from irradiated seeds (Rédei, 1962) and the mutation has been used as a visible marker for genetic mapping (chr.1-10.9 cM) (Hauge et al., 1993). Compared to the wt, the an mutant has narrow cotyledons, narrow rosette leaves, slightly twisted seed pods (siliques) (Rédei, 1962), and less-branched trichomes (Hülskamp et al., 1994). The AN gene was predicted to play a role in leaf blade development, but has not previously been characterized in detail. The existence of rotundifolia (rot) mutants was predicted from our characterization of the an mutant, and such mutants were isolated from irradiated seeds. Analysis of these mutants should help us dissect the regulation of leaf development.

Plant culture

Seeds were sown on rockwool and/or vermiculite moistened with MGRL medium as described elsewhere (Tsukaya et al., 1991). Plants were grown at 23°C under continuous white fluorescent light (67.4±14.0 μmol/second/m2).

For cultivation of plants on plates, seeds were sterilized in a solution of NaClO (Tsukaya et al., 1991) and sown on the medium used by Okada and Shimura (1992). The plates were incubated vertically, for observation of the growth of roots and hypocotyls, under the conditions described above.

The an single mutant line was isolated from back-crossed populations of a marker line, GPR1 (an, gl1, th1, tt4), derived from ecotype Columbia. Back-crossing was carried out with ecotype Columbia.

Mutant production and gene mapping

The rot mutants were isolated from an M2 population of fast neutronirradiated seeds of ecotype Columbia with the gl1 mutation (Lehle Seeds, Tucson, AZ, USA). rot mutants were back-crossed with ecotype Columbia to obtain single-mutant lines. Allelism tests were carried out. Mapping of mutations was performed by using mapping strain W100 (an, ap1, bp, cer2, er, gl1, hy2, py, tt3) and CS8 (ap2, bp1, cer2) both of which were derived from ecotype Landsberg erecta. The rot3 mutation was mapped by PCR-based mapping using CAPS (cleaved amplified polymorphic sequences) markers (Research Genetics, Huntsville, AL, USA), along with the mapping strain W100. The an rot3 double mutant was obtained by crossing the an mutant with the rot3 mutant.

Our genetic nomenclature, is based on the proceedings of the Third International Arabidopsis Meeting (East Lansing, MI, USA, 1987): wild-type alleles are given in capitals and italicized; mutant alleles are given in lower-case, italicized letters.

Measurement of leaf size

Growth conditions were carefully fixed and ecotypic backgrounds were established to provide reproducible characters in each strain. The leaves were numbered from the first rosette leaf that emerged after the cotyledons to the last rosette leaf. Cauline leaves were numbered independently of rosette leaves.

Leaf expansion measurements

We use these terms to describe directions of, on, and within the leaf blade: leaf-length direction, the main direction of growth of the leaf primordium, along the midrib of the leaf blade (thus, a longitudinal section of the leaf means a section made in a plane in this direction perpendicular to the adaxial/abaxial surface of the leaf); leaf-width direction, the direction at right angles to the leaf-length direction, on the adaxial/abaxial surface of the leaf (thus, a transverse section of the leaf means a section made in a plane in this direction perpendicular to the adaxial/abaxial surface of the leaf); and leaf-thickness direction, the direction perpendicular to the adaxial/abaxial surface of the leaf. These three directions form the three-dimensional axes of the leaf, being perpendicular to each other.

Histology and anatomy

Histological analyses were performed with samples in Technovit 7100 resin (Kulzer & Co. GmbH, Wehrheim, FRG) as described previously (Tsukaya et al., 1993). For transverse sections, tissue samples were cut at the exact center of the fifth rosette leaf blade. For longitudinal sections, tissue samples were cut along the midrib of the leaf before embedding.

Scanning electron microscopy (SEM)

Observations of plant materials by SEM were made as described previously (Tsukaya et al., 1993), using two systems (S-700 from Hitachi, Ibaragi, Japan; and JMS-820S from JEOL Ltd., Tokyo, Japan), and samples were photographed on TMAX 100 film (Kodak, Rochester, NY, USA).

Observations of epidermal cells

Surface shapes of epidermal cells of fully expanded fifth true leaves were examined on photographs obtained by SEM as follows. Some samples were fixed in FAA solution as described previously (Tsukaya et al., 1993) and rendered transparent by incubation overnight in a chloral hydrate solution (chloral hydrate, 200 g; glycerol 20 g; H2O, 50 ml). Other samples were centrifuged in distilled water in a 1.5-ml tube at 1×104g for 5 minutes to obtain air-free samples with chroloplasts gathered in one part of each cell. Samples from both methods were photographed under a microscope to obtain paradermal images of layers of cells. For these observations, samples were taken from the central region of the lamina, avoiding the midrib and the veins (see Fig. 7C).

Shapes of epidermal cells were evaluated in terms of numbers of protrusions of cells in the leaf-width and leaf-length directions, by a slightly modified version of the method of Tsukaya et al. (1994). A grid was placed on the central region of the lamina on a photograph of the epidermis, with one axis of the grid parallel to the leaf-length direction. The number of cell-boundary lines and cells that crossed one unit length of the grid in the leaf-length direction was then counted. The average value from seventeen units on the grid was taken as the width factor of the leaf (Fig. 7C). A length factor was calculated in a similar manner perpendicular to the direction used to calculate the width factor. These two factors indicate the tendency of cells to protrude in the directions of cell expansion. Then the number of cell-boundary lines per cell that crossed a grid was similarly calculated to define the complexity of the cell, in terms of protrusions in each direction (Fig. 7C).

Development of wild-type Arabidopsis leaves

To characterize the fundamental features of leaf development, statistical analysis was first performed with Columbia wt plants, which produced two cotyledons, nine rosette leaves, and three cauline leaves under our culture conditions (Fig. 1 and Table 1).

Table 1.
Numbers of leaves in the wild type and the mutants
graphic
graphic
Fig. 1.

(A) The morphology of leaves of the wt, the an mutant, and the rot3 mutant. The leaves in each row, from the left, are the two cotyledons, eight rosette leaves and three cauline leaves. The leaves were collected when fully expanded. (B) The morphology of wt and rot mutant plants.

Fig. 1.

(A) The morphology of leaves of the wt, the an mutant, and the rot3 mutant. The leaves in each row, from the left, are the two cotyledons, eight rosette leaves and three cauline leaves. The leaves were collected when fully expanded. (B) The morphology of wt and rot mutant plants.

Heteroblasty of wt leaves

The wt exhibited heteroblasty, as previously reported: leaves produced in the shoot system had a specific morphology at each position (Marinez-Zapter et al., 1995; Röbbelen, 1957). Cotyledons were round and small, while cauline leaves had very short petioles. The cotyledons, rosette leaves and cauline leaves were distinguishable from each other, but the morphology of rosette leaves changed with the leaf position (Fig. 1A).

For example, the ratio of leaf width to length (Fig. 2D) for the first few rosette leaves resembled that for cotyledons, whereas ratios for rosette leaves produced much later resembled the ratios for cauline leaves (Fig. 2D).

Fig. 2.

Characterization of the leaves of the wild type, the an mutant, and the rot3 mutant according to leaf position. Length of the petiole (A), length of the leaf blade (B), width of the leaf blade (C), and the ratio of the width to the length of the leaf blade (D) are shown for each strain. cot., cau.1, cau.2, and cau.3 refer to the cotyledon, first cauline leaf, second cauline leaf, and third cauline leaf, respectively. The numbers indicate positions of rosette leaves from the juvenile stage. The error bars represent standard deviations for results from more than 25 samples.

Fig. 2.

Characterization of the leaves of the wild type, the an mutant, and the rot3 mutant according to leaf position. Length of the petiole (A), length of the leaf blade (B), width of the leaf blade (C), and the ratio of the width to the length of the leaf blade (D) are shown for each strain. cot., cau.1, cau.2, and cau.3 refer to the cotyledon, first cauline leaf, second cauline leaf, and third cauline leaf, respectively. The numbers indicate positions of rosette leaves from the juvenile stage. The error bars represent standard deviations for results from more than 25 samples.

The fifth rosette leaf had the most reproducible distinguishing features of the rosette leaves. Therefore, we chose the fifth rosette leaf as representative of the specific phenotype of each strain (Figs 1A, 2). The cotyledons, which are formed during embryogenesis, and the cauline leaves, which lack petioles, were judged to be inappropriate for our analysis.

Developmental processes during leaf expansion

During the expansion of the leaf blade, the increase in the wt leaf length was directly proportional to that of the width (Fig. 3).

Fig. 3.

Growth of the fifth rosette leaves of the wild type, the an mutant, and the rot3 mutant from initiation of the primordium to full expansion. The symbols represent average results from more than five samples. Arrows show the estimated stages for the mutants in relation to the growth of the wt (see text).

Fig. 3.

Growth of the fifth rosette leaves of the wild type, the an mutant, and the rot3 mutant from initiation of the primordium to full expansion. The symbols represent average results from more than five samples. Arrows show the estimated stages for the mutants in relation to the growth of the wt (see text).

Transverse and longitudinal sections were prepared from fifth rosette leaves at the following stages: stage I, leaf length = 1.0 mm; stage II, 5.0 mm; stage III, 10.0 mm; and stage IV, 15.0 mm (Fig. 3). Throughout leaf development, the first layer of palisade cells consisted of cells neatly aligned in the paradermal plane (Fig. 4A,B), and these cells were used for enumeration of cells in the leaf-width and length directions on sections. The epidermal cells were irregularly shaped, and the xylem and phloem cells also had complex structures. Therefore, these cells were excluded from analyses.

Fig. 4.

Sections of wild-type and mutant leaves. (A) Cell development in the leaf-width direction of the fifth leaves of the wt, the an mutant and the rot3 mutant. The transverse sections reveal a region between the midvein and the leaf margin. (B) Cell development in the leaflength direction of the fifth leaves of the wt, the an mutant and the rot3 mutant. The longitudinal sections reveal a region in the center of the lamina. The sections in horizontal rows are from leaves at the same stage of growth, namely, stages I, II, and IV (see text), respectively, from the top to the bottom. The adaxial surfaces of the leaves are uppermost. (C) Transverse and longitudinal sections of leaves of the an rot3 double mutant harvested at the fully expanded stage. Bars, 100 μm.

Fig. 4.

Sections of wild-type and mutant leaves. (A) Cell development in the leaf-width direction of the fifth leaves of the wt, the an mutant and the rot3 mutant. The transverse sections reveal a region between the midvein and the leaf margin. (B) Cell development in the leaflength direction of the fifth leaves of the wt, the an mutant and the rot3 mutant. The longitudinal sections reveal a region in the center of the lamina. The sections in horizontal rows are from leaves at the same stage of growth, namely, stages I, II, and IV (see text), respectively, from the top to the bottom. The adaxial surfaces of the leaves are uppermost. (C) Transverse and longitudinal sections of leaves of the an rot3 double mutant harvested at the fully expanded stage. Bars, 100 μm.

Cell division seemed to occur throughout growth but tended to occur more frequently at earlier stages (stages I and II) in the basal part of the leaf (data not shown). This phenomenon reflects the results of Pyke et al. (1991) for the first rosette leaf of ecotype Landsberg erecta. Numbers of palisade cells in the leaf-length and width directions increased from stage II to stage IV (Table 2). Therefore, the total number of cells in the section also increased (Table 2). The number of cells in the leaf-thickness direction decreased (from 5.6 to 4.4) during this period, probably because of formation of intracellular spaces (Fig. 4A,B).

Table 2.
Anatomical analysis of the fifth rosette leaf of the wild type and the mutants
graphic
graphic

Growth of palisade cells in the first cell layer was examined to determine the effects of the organization and expansion of individual cells on the morphology of the leaf.

Transverse sections of leaves at early stages (Fig. 4A; stage I) mainly revealed unvacuolated, newly divided, rod-shaped palisade cells. As growth proceeded, transverse sections revealed expansion of cells, with subsequent elongation in the leaf-width direction (Fig. 4A; stages II and IV). In longitudinal sections at stage I, a rather higher rate of cell division was noted in the basal region of the leaf blade than in the tip region, where cell expansion was more probably occurring (data not shown). This was the only stage at which such variations were observed. Leaf sections prior to stage I had newly divided small cells in all regions, while older leaves had expanding cells in all regions. Hence, stage I seemed to be the stage at which cell elongation started in the leaf-length direction. As growth proceeded, the palisade cells in this central region of longitudinal sections expanded and then elongated in the leaflength direction (Fig. 4B; stages II and IV). From these anatomical studies, the elongation processes in two directions were clearly determined as discussed below.

The an mutant

The an mutation affects the morphology of the leaf and modified leaf organs

In order to determine how the AN gene acts, we measured leaves, hypocotyls, roots, and floral organs of wt and an mutant plants.

The leaves of the an mutant are shown in Fig. 1. Resembling the wt, the an mutant produces an average of two cotyledons, nine rosette leaves and three cauline leaves. For statistical analysis, the morphology of leaves at each position was evaluated in terms of the parameters shown in Fig. 2. The shoot system of the an mutant exhibited heteroblasty, as did that of the wt. The an mutant is known as a narrow-leaved mutant, but we found that the petiole and the leaf-blade length of the an mutant were the same as those of the wt (Fig. 2A,B). The leaf width of the an mutant was significantly reduced (Fig. 2C), and the ratio of an width to wt width, determined for each leaf position, ranged from 0.61 to 0.77.

The average length of the hypocotyl of wt plants grown on rockwool, on the seventh day after sowing, was 1.4±0.30 mm, while that of the an mutant was 1.4±0.25 mm. The average diameter of hypocotyls of the wt and the an mutant was 0.24 ±0.033 and 0.25±0.017, respectively, again showing no significant difference. an plants sown on vertically oriented agar plates were also not significantly different from the wt in terms of growth rates of hypocotyls and primary roots (Fig. 5A,B). Thus, no defects were found in the hypocotyl and primary root of the an mutant.

Fig. 5.

Growth of the hypocotyl and the root of the wild type, the an mutant and the rot3 mutant on vertically oriented culture plates. (A) Lengths of hypocotyls grown on agar medium. (B) Lengths of primary roots grown on agar medium. All measurements were made with more than 10 samples of each strain. Note that there are no significant differences between the mutants and the wt.

Fig. 5.

Growth of the hypocotyl and the root of the wild type, the an mutant and the rot3 mutant on vertically oriented culture plates. (A) Lengths of hypocotyls grown on agar medium. (B) Lengths of primary roots grown on agar medium. All measurements were made with more than 10 samples of each strain. Note that there are no significant differences between the mutants and the wt.

Floral organs can be considered to be modified leaves, and petals, sepals and siliques showed the general character of the mutation observed in the leaves (data not shown). Floral organs of the an mutant were slightly longer and narrower than those of the wt. The an seed pods (siliques) were twisted as described by Rédei (1962), perhaps because of tension created by carpels that were too narrow to contain normal-sized seeds. Thus, the an mutant had morphological defects specific to the leaves and modified leaves.

The AN gene affects leaf length from a certain stage

To determine when the AN gene exerts its effect on leaf blade morphogenesis, the growth of fifth rosette leaves was examined. SEM revealed that the formation of the leaf primordium in the an mutant did not differ morphologically from that in the wt (Fig. 6A,B). However, the growth in the leaf-width direction of the fifth-rosette leaf blade was affected by the an mutation when the leaf blade was at stage I (Fig. 3). Since the an mutant and the wt produced leaves of the same final length and the increase in length was directly proportional to that in width (Fig. 3), the growth stage of an mutant leaves was evaluated in terms of the length of the leaf blade.

Fig. 6.

(A-C) SEM images of the fifth rosette leaf primordia of the wild type (A), the an mutant (B), and the rot3 mutant (C) prior to stage I. Bar, 100 μm. (D-F) SEM images of trichomes and epidermal cells of the wild type (D), the an mutant (E) and the rot3 mutant (F). The central region of leaf blades of plants at stage IV were examined. The arrows show the leaf-length direction. Bar, 100 μm.

Fig. 6.

(A-C) SEM images of the fifth rosette leaf primordia of the wild type (A), the an mutant (B), and the rot3 mutant (C) prior to stage I. Bar, 100 μm. (D-F) SEM images of trichomes and epidermal cells of the wild type (D), the an mutant (E) and the rot3 mutant (F). The central region of leaf blades of plants at stage IV were examined. The arrows show the leaf-length direction. Bar, 100 μm.

Changes in cell morphology alter leaf morphology

At least two factors, the size of cells and the number of cells, were predicted to affect the phenotype of an leaves. To identify which factor(s) contributes to the phenotype of the an leaf blade, we performed an anatomical analysis with comparison to the wt (Fig. 4A,B). The number of vascular bundles in the an mutant was the same as in the wt (data not shown).

The wt and the an mutant had the same number of palisade cells in the leaf-length direction (Table 2). The an mutant had fewer (76%) palisade cells in the leaf-width direction and more cells in the leaf-thickness direction (145%) in the expanded leaf (stage IV). Thus, the total number of cells in the an leaf blade appeared to be the same as in the wt.

Throughout leaf expansion, the total number of palisade and spongy mesophyll cells in the an mutant resembled that in the wt in transverse sections and was larger in longitudinal sections (Table 2), reflecting the alignment of cells in three dimensions, as described above (Table 2).

The cells of the an mutant had a unique morphology that was especially evident in the elongation of palisade cells. Palisade cells exhibited restricted elongation in the leaf-width direction and enhanced elongation in the leaf-thickness direction (Fig. 4A). Transverse sections at stage I showed that the division and shapes of cells were similar to that in the wt (Fig. 4A; stage I). As stage I proceeded, elongation of cells became restricted in the leaf-width direction and enhanced in the leaf-thickness direction (Fig. 4A; stage II and IV). In longitudinal sections, the only differences observed were enhanced elongation in the leaf-thickness direction, with cells tilted towards the leaf tip (Fig. 4B; stages I and IV), possibly as a result of the initial expansion of the epidermis.

The palisade cells in the first layer were measured in the leafwidth, leaf-length, and leaf-thickness directions (Table 3). The an mutant had tilted palisade cells in longitudinal sections so measurements of these cells were difficult. Therefore, measurements were made on both sections and paradermal images (Table 3). Analysis of sections revealed that, in the an mutant, palisade cells were smaller in the leaf-width direction than in the wt (87%) while they were somewhat larger in the leaflength direction. In the leaf-thickness direction the cells were significantly larger (168%), suggesting that palisade cells expanded preferentially in the leaf-thickness direction. Analysis of paradermal images supported this result, revealing a 26% reduction in the leaf-width direction (Table 3). The cell measurements in the leaf-length direction were also reduced in this analysis, as a consequence of the tilting and overlapping of cells in this direction. Reduction in the width direction was also noted in other cells, such as spongy mesophyll cells (data not shown).

Table 3.
Morphology of palisade cells (first layer) of the wild type, the an mutant, and the rot3 mutant
graphic
graphic

Epidermal cells of the an mutant are irregular in shape with fewer protrusions in the leaf-width direction

Since epidermal cells are generally known to initiate the expansion of the leaf (Dale, 1988), we predicted that they would have unique features in the an mutant. The epidermal cells on the abaxial and adaxial sides had a complex jigsawpuzzle-like shape with many protrusions in various directions (Fig. 6D,E,F). In SEM images, the epidermal cells on the adaxial side of the fifth rosette leaf of the an mutant were rather rectangular with fewer protrusions in the leaf-width direction, as were the epidermal cells in the cotyledons of the an mutant (Tsukaya et al., 1994). For a statistical evaluation, a grid was placed on images of the epidermis of the adaxial surface, and cell-boundary lines crossing a unit length were plotted on a graph of length factors against width factors (see Materials and methods; Fig. 7A,C). These factors represent the polarity of cell elongation. The larger the value, the more polarity-specific elongation occurred in the given direction. The an mutant had smaller width factors than the wt (Fig. 7A), having reduced protrusions in the leaf-width direction. The number of cell-boundary lines per cell that crossed a grid was also calculated for the leaf-width and leaf-length direction to represent the complexity in each direction, in terms of formation of protrusions (Fig. 7B,C). The fewer protrusions in a given direction, the smaller was the factor. The an mutant gave smaller values for both factors, than the wt. Thus, the cells of the an mutant were less complex.

Fig. 7.

Characterization of epidermal cells of the wild type, the an mutant, and the rot3 mutant. (A) Evaluation of the polarity of the epidermal cells. Numbers of cell-bordering lines that crossed the grid per unit length were determined (see text) for samples from 10 fully expanded leaves (stage IV). (B) Evaluation of the complexity of epidermal cells. Numbers of cell-bordering lines crossing the grid per cell (see text) were determined for samples from more than 10 fully expanded leaves. (C) Diagrams showing the method. The left diagram shows the method for (A) and the right for (B).

Fig. 7.

Characterization of epidermal cells of the wild type, the an mutant, and the rot3 mutant. (A) Evaluation of the polarity of the epidermal cells. Numbers of cell-bordering lines that crossed the grid per unit length were determined (see text) for samples from 10 fully expanded leaves (stage IV). (B) Evaluation of the complexity of epidermal cells. Numbers of cell-bordering lines crossing the grid per cell (see text) were determined for samples from more than 10 fully expanded leaves. (C) Diagrams showing the method. The left diagram shows the method for (A) and the right for (B).

Trichomes are single-celled organs on the epidermis. Trichomes of the an mutant differed from those of the wt in terms of the number of branches (an, 1 to 2 branches; wt, 3 to 4 branches; Hülskamp, 1994) and the branching direction (an, parallel to the leaf-length axis; wt, various directions; Table 4 and Fig. 6D,E).

Table 4.
Number of branches of trichomes on the fifth rosette leaf of the wild type and the mutants
graphic
graphic

The rot mutant

Our analyses of the an mutant indicated that it may be possible to isolate mutants with a defect in the leaf-length direction. Therefore, plants with short leaves of normal width were generated and analyzed.

rot mutants with leaf-length-specific defects

Four short-leaf mutants were isolated and named rotundifolia (round leaf). Three rot loci that caused short, round leaves with short petioles (Fig. 8A,B) were identified, with rot1 having at least two alleles (Fig. 1B). Mapping revealed that the rot1 locus was located at 42.8±1.7 map units on chromosome 3. The rot3 locus showed weak linkage to the tt3 locus (ca. 61.6±22.5 cM in distance), when the F2 generation of the cross, rot3 × W100 strain, having multiple visible markers, were examined. Twenty-four rot3 mutant plants from the F2 generation of the cross, rot3 (Columbia background) × CS8 mapping strain (Landsberg erecta background), were examined for the ASA1 CAPS marker. 33 chromosomes showed the Columbia background pattern in the RFLP. Thus the rot3 locus was mapped to the upper arm of chromosome 5, near the ASA1 locus at 18.5 cM. Plants with mutations at the 3 loci had short leaf blades of normal width. The rot3 mutant was used for further analysis, having the most typical, reproducible phenotype and a leaf width closest to that of the wt (Figs 1A, 8C).

Fig. 8.

Characterization of leaves of rot mutants with mutations at different loci. More than 13 fully expanded leaves were sampled in each case (stage IV).

Fig. 8.

Characterization of leaves of rot mutants with mutations at different loci. More than 13 fully expanded leaves were sampled in each case (stage IV).

The rot3 mutant had morphological defects in the leaves (Figs 1, 2) similar to those of the an mutant. Petals, sepals, stamens and pistils were all of reduced length (data not shown). However, no other parts of the plant showed any defects in growth rate or length (Fig. 5A,B). Hence, the rot3 mutation was revealed to have morphological defects specific to leaves and modified leaf organs.

The number of rosette leaves differed from that of the wt. There were two cotyledons, twelve rosette leaves, and three cauline leaves (Table 1). Leaf blades and petioles were of significantly reduced length at all positions (Fig. 2A,B). The length of the fifth rosette leaf was significantly different (level of significance, 5%) from that of the wt while the width was not (Fig. 2B,C).

The emerging rot3 leaf primordia seemed to be identical to those of the wt and the an mutant (Fig. 6A,C). The phenotypic difference became apparent at stage I (Fig. 3) and consistently affected leaf morphology as the leaf developed. The growth of each leaf was directly proportional to its width, as shown in Fig. 3. Therefore, the growth stage of rot3 leaves was evaluated in terms of the width of leaf blades.

The rot3 mutation affects cell elongation

Anatomical analyses were carried out on rot3 leaves (Fig. 4A,B) in the same way as for the an mutant. The number of vascular bundles in the rot3 mutant was the same as in the wt (data not shown). The wt and the rot3 mutant exhibited no significant differences in the number of palisade cells in both the leaf-width and leaf-length directions in the paradermal plane at stage IV (Table 2). The rot3 mutant had slightly more cells in the leaf-thickness direction (5.6 versus 4.4, at stage IV). However, this difference did not directly indicate that the number of cell layers had increased. The slightly larger number of cells in the leaf-thickness direction of the rot3 mutant led to a slight increase in the total number of palisade and spongy mesophyll cells in transverse and longitudinal sections (Table 2).

The cells of the rot3 mutant exhibited a polarity-specific defect in elongation, as observed in the an mutation but in a perpendicular direction. When compared to the wt, the palisade cells of the rot3 mutant on paradermal images were the same size in the leaf-width direction but 9% shorter in the leaf-length direction and slightly longer in the leaf-thickness direction (Table 3). Measurements on sections revealed an 11% reduction in the leaf-length direction of the cell. In the rot3 mutant the shorter cells in the leaf-length direction was not directly correlated to that in the leaf-length direction. However, this observation can be explained by the fact that the rot3 mutant had fewer intracellular spaces than the wt in the leaf-length direction (Fig. 4). Thus, the rot3 mutation affected the elongation of cells in the leaf-length direction.

The epidermal cells of the rot3 mutant were not like those of either the wt or an mutant (Fig. 7A). The length factor was smaller than that of the wt: the cells formed fewer protrusions in the leaflength direction than the wt. Fig. 7B shows that the complexity of the epidermal cells was intermediate between those of the wt and the an mutant. There were no defects in trichome branching in the rot3 mutant (Table 4 and Fig. 6).

The an rot3 double mutant

Cross pollination yielded an an rot3 double recessive mutant (Fig. 9E) which produced two cotyledons, twelve rosette leaves, and three cauline leaves (Table 1). The double mutant exhibited reduced growth in the leaf-length and leaf-width directions, with an increase in thickness, as compared to the wt (Fig. 9A-C). The ratio of width to length of rosette leaves was similar to the wt, but leaves were of reduced size (Fig. 9D). The phenotype of the double mutant was the sum of the constituent phenotypes. The double mutant had slightly fewer cells in the leaf-width direction, with an increase in the leafthickness direction, characteristics of the an mutation, but it was normal in the leaf-length direction (Table 2). Transverse sections showed that cells were short in the leaf-width direction but elongated in the thickness direction, characteristics of the an mutant. Longitudinal sections showed that the cells were short in the leaf-length direction, a characteristic of the rot3 mutation (Fig. 4C). Measurements on paradermal images also showed that cells of the double mutant were smaller in both the leaf-width and length directions (Table 3).

Fig. 9.

(A-D) Characterization of the an mutant, the rot3 mutant, and the an rot3 double mutant, according to leaf positions. See Fig. 2 for abbreviations. (E) Leaves of the an mutant, the rot3 mutant, and the an rot3 double mutant.

Fig. 9.

(A-D) Characterization of the an mutant, the rot3 mutant, and the an rot3 double mutant, according to leaf positions. See Fig. 2 for abbreviations. (E) Leaves of the an mutant, the rot3 mutant, and the an rot3 double mutant.

The an and rot3 mutations acted independently in floral organs also. The double mutant had short narrow petals, and short stamens and pistils, as compared to the wt (data not shown).

Arabidopsis: a model for studies of leaf morphology

An attempt was made to determine the genetic basis for leaf morphogenesis, and novel parameters were used to characterize the morphology of leaves of wild-type Arabidopsis (Fig.2). The leaves exhibited heteroblasty, as shown in Fig. 1A. Analysis of fifth rosette leaves revealed that growth in the leaflength direction was directly proportional to that in the leafwidth direction (Fig. 3). Anatomical analyses revealed specific stages for such regulation at the cellular level. Transverse and longitudinal sections of plants at stage I revealed that palisade cells divide periclinally to form rod-shaped cells that are elongated in the leaf-thickness direction. From stage I, the cells begin to elongate in both the leaf-length and leaf-width direction (Fig. 4A,B). Some factor(s) might regulate cell volume, through regulation of turgor pressure or vacuolation, during the leaf expansion process. Leaves at the same position showed a variety of final sizes. This variety is plotted graphically in Fig. 3 and suggests that some unknown factor might control the termination of cell division and/or elongation. Cell division is observed throughout leaf expansion. It is not temporally fixed but is a major feature of early stages. However, polarity-specific elongation is observed mainly at later stages of growth, as discussed below.

The AN gene affects the distribution of cells and leaf expansion

The phenotype associated with the an mutation was exhibited only in leaves (Figs 1, 2) and modified leaf organs. The AN gene seems to be specific to the morphology of these organs. All of the phenotypic characteristics of the an mutant can be explained by a defect in polarity-specific expansion in the leafwidth direction.

The total number of cells forming the leaf blade in the an mutant appeared to be the same as in the wt. However, the an mutant had fewer cells in the leaf-width direction and more cells in the leaf-thickness direction. In the wt, the number of cells in the leaf-thickness direction decreases during leaf expansion, as noted above (Table 2). During leaf expansion, in general, spongy mesophyll cells cease to divide at an earlier stage and intracellular spaces form as epidermal cells expand (Dale, 1988). This process causes a decrease in the number of cells in the leaf-thickness direction. In the an mutant, the failure of cell elongation and distribution in the leaf-width direction reduces the formation of intracellular spaces. Thus, cell numbers are normal and elongation is enhanced in the leafthickness direction (Table 2).

The AN gene controls leaf morphology through polarity-specific elongation of cells

The expansion of cells in Arabidopsis involves tri-directional growth during leaf development and the an mutant has a defect in cell elongation in the leaf-width direction (Fig. 4), which results in thicker cells (Table 3). Thus, the AN gene appears to control leaf morphology via regulation of polarity-specific elongation in the leaf cells.

The AN gene does not affect morphology during the initiation and early development of leaves, having its first visible effect on leaf development when the fifth true leaf is at stage I. Other phenotypes associated with the an mutation can also be explained by the defect in cell elongation in the leafwidth direction. The an mutation affects not only the polarity of the epidermal cells but also formation of protrusions (Fig. 7B). Polarity in trichomes seems also to be regulated by the AN gene, with the an mutation reducing branch formation (cell expansion) in the leaf-width direction. Trichome morphology similar to that in the an mutant was reported in zwichel mutants (Hülskamp et al., 1994), with mutations in a gene for kinesin, a microtubule motor protein that might regulate orientation of cellulose microfibrils (Oppenheimer et al., 1995). Since the an mutation affects polarity-specific elongation of cells, the AN gene might perhaps regulate the orientation of cellulose microfibrils.

The length of leaves is controlled by the ROT3 gene through polarity-specific elongation of cells

The effect of the rot3 mutation was also only visible in leaves (Figs 1, 2) and modified leaf organs. Therefore, the ROT3 gene appears to regulate the morphology of the leaf and modified leaf organs specifically. This specificity is a unique feature of this mutant. Previously isolated elongation mutants, dwarf and diminuto, have defects in all organs (Feldmann et al., 1989; Takahashi et al., 1995).

The rot3 mutation affected the localization and shape of cells. However, the ROT3 gene seems to function more in the regulation of leaf cell morphology than in the control of leaf cell number or the distribution of cells. The total number of cells in the leaf blade of the rot3 mutant was slightly larger as that of the wt (Table 2). All the phenotypic characteristics of the rot3 mutant can be attributed to a defect in cell elongation. The ROT3 gene seems to affect leaf morphology soon after the initiation and early development of leaves (stage I), when cells in the central part of the lamina start to grow in the leaf-length direction (Fig. 4).

The morphology of the palisade cells indicated that the rot3 mutation restricts cell elongation in the leaf-length direction specifically (Table 3). The greater cell length in the leafthickness direction was attributed to a concomitant effect of this restriction. Epidermal cells of the rot3 mutant also exhibited restricted elongation in the leaf-length direction (Fig. 7A). The rot3 mutation did not have any significant effect on the trichomes, unlike the an mutation (Table 4 and Fig. 6D,F). If our hypothesis about the an mutant is correct, the trichomes of the rot3 mutant should not show changes in branching if epidermal cells retain their protrusions. As seen in Figs 6D,F, 7B, the rot3 mutant had protrusions on its epidermal cells similar to the wt. Thus, the ROT3 gene appears to control leaf morphology via regulation of polarity-specific elongation (leaf-length direction) of leaf cells.

Control of leaf expansion through polarity-specific regulation of cell elongation by the AN gene and ROT3 genes

The phenotype of the an rot3 double recessive mutant was the sum of the two mutant phenotypes. Thus, the AN and ROT3 genes play independent roles in leaf development. A model explaining the roles of the AN gene and the ROT3 gene in leaf cell expansion is shown in Fig. 10. Each plant cell is constantly subjected to turgor pressure which leads to cell expansion. To some extent, this pressure is suppressed or inhibited by polarity-specific elements, such as the microfibrils of the cell wall, whose polarity is regulated by cytoskeletal elements, for example, cortical microtubules. Since the an and rot3 mutations are recessive, we propose that the products of the AN and ROT3 genes suppress the actions of hypothetical factor(s), represented by X and Y in Fig. 10, which inhibit cell expansion in a polarity-specific manner. A suppressor of elongation in the leaf-width direction is designated X (Fig. 10). The product of the AN gene suppresses X and allows the cell to elongate in the leaf-width direction (wt; stage II). The hypothetical factor Y suppresses cell elongation in the leaf-length direction. The product of the ROT3 gene will suppress this suppressor to initiate elongation in the leaf-length direction in wt plants. In both mutations, the increase in length in the leafthickness direction is a concomitant effect of restricted elongation of cells in the leaf-width and length directions.

Fig. 10.

Model explaining the role of the products of the AN gene and the ROT3 gene in leaf development. X and Y stand for putative suppressors of cell elongation in the leaf-width direction and the leaf-length direction, respectively. See text for details.

Fig. 10.

Model explaining the role of the products of the AN gene and the ROT3 gene in leaf development. X and Y stand for putative suppressors of cell elongation in the leaf-width direction and the leaf-length direction, respectively. See text for details.

The working hypothesis represented by this model is being tested in our laboratory, with emphasis on cytoskeletal elements in these mutants, which might provide clues to the functions of the AN and ROT3 genes.

A new avenue to the studies of leaf morphogenesis

Leaf morphogenesis involves various regulatory processes that form a complex system. This is the first report of dissection of the complicated processes of leaf development, and it demonstrates the role of two independent regulators that govern polarity-specific elongation at the cell level. These processes are key processes in the two-dimensional expansion in leaf morphogenesis. Identification of genes responsible for each process in leaf development should open new avenues for research into the development of plant systems.

The authors thank Dr J. L. Bowman of Monash University for a critical reading of an early version of the manuscript and his kind encouragement, and Dr E. Aspuria and Dr M. Umeda of the University of Tokyo for comments on an early version of the manuscript. The authors also thank Prof. Y. Komeda of Hokkaido University for kindly supplying the mutant lines of Arabidopsis. The authors thank Ms K. Shinozaki, an operator of the SEM (Hitachi) at the University Museum of the University of Tokyo for her skilled assistance, as well as the curator, Prof. K. Iwatsuki, of the Botanical Gardens, Faculty of Science, University of Tokyo, for access to another SEM (JEOL). H. T. was supported by a Grant-in-Aid for Scientific Research on Priority Areas from the Ministry of Education, Science and Culture, Japan.

Avery
,
G. S.
, Jr
. (
1933
).
Structure and development of the tobacco leaf
.
Amer. J. Bot
.
20
,
565
592
.
Bowman
,
J. L.
,
Smyth
,
D. R.
and
Meyerowitz
,
E. M.
(
1991
).
Genetic interactions among floral homeotic genes of Arabidopsis
.
Development
112
,
1
20
.
Dale
,
J. E.
(
1988
).
The control of leaf expansion
.
Ann. Rev. Plant Physiol. Plant Mol. Biol
.
39
,
267
295
.
Dolan
,
L.
and
Poethig
,
S.
(
1991
).
Genetic analysis of leaf development in cotton
.
Development Suppllement
1
,
39
46
.
Evans
,
M. W.
and
Grover
,
F. O.
(
1940
).
Developmental morphology of the growing point of the shoot and the inflorescence in grasses
.
J. Agricul. Res
.
61
,
481
520
.
Feldmann
,
K. A.
,
Marks
,
M. D.
,
Christianson
,
M. L.
and
Quatrano
,
R. S.
(
1989
).
A dwarf mutant of Arabidopsis generated by T-DNA insertion mutagenesis
.
Science
243
,
1351
1354
.
Goldberg
,
R. B.
,
Paiva
,
G.
and
Yadegari
,
R.
(
1994
).
Plant embryogenesis: zygote to seed
.
Science
266
,
605
614
.
Harte
,
C.
and
Hansen
,
H.
(
1967
).
Untersuchungen über die entwicklung der blattform bei Antirrhinum majus L. und einigen mutanten
.
Planta
73
,
250
280
.
Hauge
,
B. M.
,
Hanley
,
S. M.
,
Cartinhour
,
S.
,
Cherry
,
J. M.
,
Goodmann
,
H. M.
,
Koornneef
.
M.
,
Stam
,
P.
,
Chang
,
C.
,
Kempin
,
S.
,
Medrano
,
L.
and
Meyerowitz
,
E. M.
(
1993
).
An integrated genetic/RFLP map of the Arabidopsis thaliana genome
.
Plant J
.
3
,
745
754
.
Hauser
,
M.-T.
,
Morikami
,
A.
and
Benfey
,
P. N.
(
1995
).
Conditional root expansion mutants of Arabidopsis
.
Development
121
,
1237
1252
.
Hülskamp
,
M.
,
Miséra
,
S.
and
Jürgens
,
G.
(
1994
).
Genetic dissection of trichome cell development in Arabidopsis
.
Cell
76
,
555
566
.
Martínez-Zapater
,
J.M.
,
Jarillo
,
J.A.
,
Cruz-Alvarez
,
M.
,
Rolandán
,
M.
, and
Salinas
,
J.
(
1995
).
Arabidopsis late-flowering fve mutants are affected in both vegetative and reproductive development
.
Plant J
.
7
,
543
551
.
Maksymowych
,
R.
(
1963
).
Cell division and cell elongation in leaf development of Xanthium pensylvanicum
.
Amer. J. Bot
.
50
,
891
901
.
McHale
,
N. A.
(
1993
).
LAM-1 and FAT genes control development of the leaf blade in Nicotiana sylvestris
.
Plant Cell
5
,
1029
1038
.
Okada
,
K.
and
Shimura
,
Y.
(
1990
).
Reversible root tip rotation in Arabidopsis seedlings induced by obstacle-touching stimulus
.
Science
250
,
274
276
.
Okada
,
K.
and
Shimura
,
Y.
(
1992
).
Mutational analysis of root gravitropism and phototropism of Arabidopsis thaliana seedlings
.
Aust. J. Plant Physiol
.
19
,
439
448
.
Oppenheimer
,
D.
,
Pollock
,
M.
, and
Marks
,
D.
(
1995
).
Cloning of the ZWICHEL locus of Arabidopsis and is role in trichome morphogenesis
.
Sixth International Meeting on Arabidopsis Research, Madison WI. Abstract 119
.
Poethig
,
R. S.
and
Sussex
,
I. M.
(
1985
).
The developmental morphology and growth dynamics of the tobacco leaf
.
Planta
165
,
158
169
.
Pyke
,
K. A.
,
Marrison
,
J. L.
and
Leech
,
R. M.
(
1991
).
Temporal and spatial development of the cells of the expanding first leaf of Arabidopsis thaliana (L.) Heynh
.
J. Exp. Bot
.
42
,
1407
1416
.
Rédei
,
G. P.
(
1962
).
Single locus heterosis
.
Z. Vererbungs
.
93
,
164
170
.
Röbbelen
,
G.
(
1957
).
Uber Heterophyllie bei Arabidopsis thaliana (L.) Heynh
.
Ber. dt. bot. Gres
.
70
,
39
44
.
Sunderland
,
N.
(
1960
).
Cell division and expansion in the growth of the leaf
.
J. Exp. Bot
.
11
,
68
80
.
Takahashi
,
T.
,
Gasch
,
A.
,
Nishizawa
,
N.
and
Chua
,
N.-M.
(
1995
).
The DIMINUTO gene of Arabidopsis is involved in regulating cell elongation
.
Genes Dev
.
9
,
97
107
.
Tsukaya
,
H.
,
Naito
,
S.
,
Rédei
,
G. P.
and
Komeda
,
Y.
(
1993
).
A new class of mutation in Arabidopsis thaliana, acaulis1, affecting the development of both inflorescences and leaves
.
Development
118
,
751
764
.
Tsukaya
,
H.
,
Ohshima
,
T.
,
Naito
,
S.
,
Chino
,
M.
and
Komeda
,
Y.
(
1991
).
Sugar-dependent expression of the CHS-A gene for chalcone synthase from petunia in transgenic Arabidopsis
.
Plant Physiol
.
97
,
1414
1421
.
Tsukaya
,
H.
,
Tsuge
,
T.
and
Uchimiya
,
H.
(
1994
).
The cotyledon: a superior system for studies of leaf development
.
Planta
195
,
309
312
.