Large-scale histological analysis of leaf mutants using two simple leaf observation methods: identification of novel genetic pathways governing the size and shape of leaves

Authors

  • Gorou Horiguchi,

    Corresponding author
    1. National Institute for Basic Biology, Okazaki Institute for Integrated Bioscience, Myodaiji-cho Nisigo Naka 38, Okazaki, Aichi 444-8585, Japan,
    2. School of Life Sciences, Graduate University for Advanced Studies, Hayama, Kanagawa 240-0193, Japan, and
    Search for more papers by this author
  • Ushio Fujikura,

    1. School of Life Sciences, Graduate University for Advanced Studies, Hayama, Kanagawa 240-0193, Japan, and
    Search for more papers by this author
  • Ali Ferjani,

    1. National Institute for Basic Biology, Okazaki Institute for Integrated Bioscience, Myodaiji-cho Nisigo Naka 38, Okazaki, Aichi 444-8585, Japan,
    Search for more papers by this author
    • Present address: Graduate School of Science, University of Tokyo, Tokyo 113-0033, Japan.

  • Naoko Ishikawa,

    1. National Institute for Basic Biology, Okazaki Institute for Integrated Bioscience, Myodaiji-cho Nisigo Naka 38, Okazaki, Aichi 444-8585, Japan,
    Search for more papers by this author
  • Hirokazu Tsukaya

    1. National Institute for Basic Biology, Okazaki Institute for Integrated Bioscience, Myodaiji-cho Nisigo Naka 38, Okazaki, Aichi 444-8585, Japan,
    2. Graduate School of Science, University of Tokyo, Science Building #2, 7-3-1 Hongo, Tokyo 113-0033, Japan
    Search for more papers by this author

*(fax +81 564 55 7513; e-mail ghori@nibb.ac.jp).

Summary

Observations of cellular organization are essential in understanding the mechanisms underlying leaf morphogenesis. These observations require several preparative steps, such as fixation and clearing of organs, and such procedures are time-consuming and labor-intensive for large-scale analyses. Thus, we have developed simple methods for the observation of leaf epidermal and mesophyll cells. To visualize the epidermis, a gel cast was made of the leaf surface, which was then observed under a light microscope. To visualize the leaf mesophyll cells, leaves were immersed in a solution containing Triton X-100, briefly centrifuged, and then viewed under a light microscope. These methods allowed us to conduct a histological phenome analysis for a large number of known and newly isolated leaf-shape/size mutants of Arabidopsis thaliana by measuring various parameters, including cell number, size, and distribution of cells within a leaf blade. Mutants showed changes in leaf size caused by specific increases or decreases in the number and/or size of cells. In addition, altered cell distributions in the leaf blade were observed, resulting from increases or decreases in the number of cells along the proximo-distal or medio-lateral axis, or recruitment of cells along a particular axis at the expense of other leaf parts. These results provide a phenomic view of the cellular behavior involved in organ size control and leaf-shape patterning.

Introduction

To understand plant morphogenesis, detailed observations of the cellular organization of a given organ are essential. Physical or optical sectioning of tissues and scanning electron microscopic observation are most often employed for this purpose.

Several mutants of Arabidopsis thaliana have been defined by various anatomical characteristics. In the case of leaves of A. thaliana, cross-sectioning has been used to analyze cellular phenotypes in multiple cell layers, consisting of the adaxial epidermal layer, palisade and spongy mesophyll layers, and the abaxial epidermal layer (Tsuge et al., 1996), and the organization of vascular tissues (McConnell and Barton, 1998). Scanning electron microscopy and nail-polish imprints are often used to observe structural features of the epidermis, such as the morphology and distribution of trichomes, guard cells and pavement cells (Berger and Altmann, 2000; Telfer et al., 1997; Yu et al., 2005). To vizualize vascular networks and identify mesophyll cells within the dermis, whole leaves must be fixed and cleared (Hamada et al., 2000; Tsuge et al., 1996).

Mutants displaying dramatic morphological alterations in leaf polarity, such as the disruption of bilateral symmetry and the failure to establish adaxial or abaxial identity, have been extensively analyzed histologically and have provided a wealth of knowledge regarding leaf morphogenesis (for reviews, see Tsukaya, 2002, 2005). In contrast, numerous reports have described mutant phenotypes, including alterations in leaf shape and size (e.g. Bernáet al., 1999; Pérez-Pérez et al., 2002), but detailed cellular observations of leaf phenotypes have often not been performed. Nonetheless, we believe that these histological mutant phenotypes are valuable sources of information for understanding the cellular behaviors that affect final leaf size and shape (Tsukaya, 2006). Thus, the development of easy methods for histological analysis is necessary for a comparative analysis of various leaf-shape/size mutants.

We developed two very simple techniques for histological observation of leaf tissues. Using these methods, we have been able to analyze a large number of leaf mutants isolated in our laboratory (Horiguchi et al., 2006) and by others, including hormonal mutants. To obtain a histological ’phenome’ data set, we focused on the total number and final size of leaf cells because they are the most important parameters characterizing the quantitative aspect of the leaf as a determinate organ. In addition, parameters that reflect the arrangement of cells within a leaf blade are informative of changes in leaf shape. These data allowed us to postulate diverse genetic pathways involved in leaf size and shape control, and also provided an updated view of hormonal action in leaf development. We thus established a comprehensive view of cellular behaviors affecting leaf size and shape.

Results and discussion

Observation of the leaf epidermis

Because the epidermis is a transparent tissue, it is difficult to trace accurately the jigsaw-shaped individual epidermal cells using optical microscopes. In addition, the epidermal surface is not completely flat at a microscopic level, making it difficult to focus over a wide area of tissue, unless scanning electron microscopy is used. To overcome these problems, we developed a simple observation method, the dried-gel method. A drop of 2% low-melt agarose containing 0.01% bromophenol blue pre-warmed at 50°C was placed on a glass slide, and the leaf sample was immediately gently placed on it. Once the gel solidified, the leaf material was carefully peeled off, and the remaining gel cast was left to dry for about 10 min. The gel cast was then observed without a cover glass under a differential phase contrast microscope (Figure 1). Although this method cannot be applied to serial analysis of epidermis development, it allows quicker acquisition of epidermal image than the nail polish imprinting method, which requires two preparative steps, namely dental resin imprinting and subsequent copying of the resin surface by nail polish (Berger and Altmann, 2000). The images obtained using the dried-gel method are of high quality, and the contours of individual cells, including guard cells (Figure 1a), can be easily distinguished, thereby allowing rapid measurements of cell size, shape (length, width and perimeter) and density, and stomatal density. This method can also be used to observe the epidermis of other organs, such as petioles (Figure 1b) and petals (Figure 1c). Because petal cells are conical in shape, they are quite difficult to observe under a microscope without using this method. As an example, we compared the adaxial epidermis of petals of the wild-type (Figure 1c, upper panel) and the angustifolia3 (an3) mutant, also known as grf interacting factor1 (gif1; Kim and Kende, 2004) (Figure 1c, lower panel). We previously showed that the an3 mutant has fewer, but larger, leaf cells than the wild-type (Horiguchi et al., 2005). The petal epidermal cells of the an3 mutant were also found to be larger than those of the wild-type, demonstrating the efficacy of this method.

Figure 1.

 Epidermal tissues observed using the dried-gel method.
(a) Adaxial epidermis (upper panel) and abaxial epidermis (lower panel) of leaf blades.
(b) Adaxial epidermis of leaf petiole.
(c) Adaxial epidermis of wild-type (upper panel) and an3-4 mutant (lower panel) petals.
Bars = 100 μm.

Observation of palisade cells

Hoyer's solution, which contains chloral hydrate, and its modified versions are often used in histological observations because of their tissue-clearing ability (Anderson, 1954). The optimal concentration of chloral hydrate may differ depending on the tissues examined. Therefore, we previously determined the optimal concentration of chloral hydrate solution for leaves (Tsuge et al., 1996). Leaves were fixed in formalin-acetic acid-alcohol (FAA) under a vacuum and then cleared in chloral hydrate solution (200 g chloral hydrate, 20 g glycerol, 50 ml H2O). This method reproducibly provides good images of cells (Figure 2a; note that the size of mesophyll cells in the an3 mutant [right] is larger than in the wild-type [left]).

Figure 2.

 Observation of palisade cells in fully expanded first leaves.
(a) Palisade cells in leaves from wild-type (left panel) and the an3-4 mutant (right panel) cleared using chloral hydrate solution.
(b) Palisade cells in leaves from wild-type (left panel) and the an3-4 mutant (right panel) prepared using the centrifugation method.
Bars = 100 μm.

This method, however, requires several intricate steps, such as the removal of intercellular air bubbles, fixation and clearing. Thus, we developed a very simple alternative method. Leaves were placed into a microcentrifuge tube and immersed in 0.1% Triton X-100, followed by centrifugation at 10 000 g for 1 min at room temperature. This centrifugation method not only removed air bubbles from intercellular spaces, but also sedimented the chloroplasts, greatly facilitating counting the number of cells (Figure 2b). These images are of sufficient quality for measuring the size of individual palisade cells [compare palisade cells in the wild-type and the an3 mutant (Figure 2b)]. Thus, this method could serve as an alternative for performing quick observations of leaf mesophyll cells.

Verification of leaf-shape/size mutants by the number and size of leaf cells

We applied the centrifugation method to previously isolated mutants (Horiguchi et al., 2006) and new mutant lines (the number of mutant lines has expanded from 147 to 205). We also examined the leaf phenotypes of several hormonal mutants because most of these mutants are reported to have distinct leaf phenotypes, although their cellular organization has not been examined in a strict sense. For each mutant, we measured the area, length and width of the leaf blade, determined the number and area of palisade cells in the sub-epidermal layer, and counted the number of palisade cells aligned along the proximo–distal (P–D) and medio–lateral (M–L) axes. We then calculated the leaf index (leaf blade length/width ratio) and cell number index (ratio of the number of cells in the P–D/M–L axes) (see Table S1).

The quantitative phenotypes of these plants were verified using the leaf blade area and number and size of palisade cells in the sub-epidermal layer. Most mutants examined were small-leaf mutants; large-leaf mutants were relatively rare. Changes in leaf area were associated with specific changes in the number and/or size of cells (Figure 3). Theoretically, there are nine possible classes of mutants, according to changes in cell number and size (Horiguchi et al., 2006). Mutants with normal cell number and size were associated with changes in leaf shape, and this particular class is discussed later. We previously found mutants of all possible combinations except for an increase in both cell number and size (Horiguchi et al., 2006). Here, we found that ethylene insensitive3 (ein3), ethylene response1 (etr1) and auxin response factor2 (arf2) mutants potentially belong to this previously unidentified class (Table 1). Thus, by characterizing these known hormonal mutants we are able to find mutants for all possible combinations of changes in leaf cell number and size.

Figure 3.

 Characterization of palisade cells in 205 mutant lines.
The number and size of palisade cells in each mutant were determined using the centrifugation method. Data represent the average cell number and cell size of each mutant line normalized to those of the wild-type (n = 8). Open triangles, squares and circles indicate that either cell size or cell number, or both, are statistically different from corresponding wild-type values, respectively, while closed circles indicate that neither cell size nor cell number are significantly different from the wild-type values (P < 0.05, Student's t-test).

Table 1.   Characterization of leaf organization in hormonal mutants
GenotypeLeaf area (mm2)Mesophyll cellsPavement cellsStomatal index
Area (μm2)NumberArea (μm2)Number
  1. Leaves from 25-day-old plants were used. Data are means ± SD (n = 8 for each line); relative values compared with wild-type are shown in parentheses. For each leaf, the average cell area was determined by measuring 20 palisade cells or all epidermal cells within a 0.4 μm2 area. The total number of palisade cells was determined by dividing the leaf area by the palisade cell density for each leaf. The total number of epidermal cells was estimated by dividing the mean leaf area by the mean epidermal cell size (pavement plus guard cells). Stomatal index (SI) is determined by the following formula: SI = [S/(E + S)] × 100 where S is the number of stomata per unit area and E is the number of epidermal cells per unit area (Mishra, 1997). Asterisks indicate significant differences from the wild-type (Student's t-test, P < 0.05).

Wild-type48.0 ± 3.6 (100)4186 ± 444 (100)11510 ± 1194 (100)3603 ± 208 (100)22762 (100)17.7 ± 0.7
aba2-120.6 ± 2.7* (43)3093 ± 300* (74)6524 ± 919* (57)2285 ± 170* (63)16603 (73)19.4 ± 0.4*
aba3-133.4 ± 3.8* (70)3388 ± 393* (81)9450 ± 1149* (82)3096 ± 309* (86)19737 (87)19.1 ± 0.4*
axr1-328.8 ± 2.4*(60)3116 ± 454* (75)9636 ± 884* (84)2666 ± 376* (74)18368 (81)17.7 ± 0.6
axr1-1222.5 ± 2.7* (47)2540 ± 369* (61)9343 ± 1229* (81)2033 ± 402* (56)17965 (79)16.9 ± 0.4*
arf2-169.7 ± 5.0* (145)4595 ± 666 (110)14672 ± 1715* (128)4541 ± 247* (126)28026 (123)18.9 ± 0.4*
arf7-140.8 ± 1.8* (85)3390 ± 389* (81)12522 ± 1145 (109)3444 ± 292 (96)19996 (88)17.5 ± 0.6
arf7-1 arf19-121.4 ± 3.4* (45)1824 ± 100* (44)12054 ± 2298 (105)1752 ± 162.3* (49)19955 (88)17.5 ± 0.6
ctr1-125.5 ± 0.9* (12)2085 ± 238* (50)2975 ± 286* (26)1631 ± 173* (45)6037 (27)18.9 ± 1.1*
ein3-157.5 ± 5.5* (120)4432 ± 421 (106)12993 ± 1471* (113)3847 ± 230 (107)25051 (110)17.3 ± 0.5
eto1-125.0 ± 4.0* (52)2856 ± 465* (68)8494 ± 1410* (74)2686 ± 323* (75)12760 (56)17.4 ± 0.9
etr1-165.1 ± 6.7* (136)5067 ± 612* (121)12828 ± 1304* (112)4885 ± 722* (134)21597 (94.9)16.5 ± 0.7*
jar1-136.6 ± 5.6* (76)3545 ± 310* (85)10111 ± 1088* (88)2865 ± 312* (80)22422 (99)18.3 ± 0.7

Genetic pathways controlling leaf proportions via cell proliferation

Measurements of the number of palisade cells along two leaf axes allowed us to classify mutants according to axis-dependent changes in cell number. We then examined whether such changes were correlated with changes in leaf shape. The leaf index roughly represents the overall proportions of the leaf blade. The first leaves of the wild-type are almost circular (i.e. a leaf index of approximately 1.0), thus most leaf-shape mutants have narrower/longer or shorter/wider leaf blades than the wild-type (Figure 4a,c). We compared the leaf index and cell number index within individual mutant lines. Changes in the leaf index were correlated with changes in the cell number index (Figure 4a), suggesting that cell proliferation rather than polar cell expansion is affected in these mutants.

Figure 4.

 Geometric distribution of palisade cells in leaf blades.
(a) Comparison of the leaf index (leaf blade length/width ratio) and cell number index (ration of number of cells in the proximo–distal [P–D] and medio–lateral [M–L] axes) within 205 mutant lines (n = 8 for each mutant line).
(b) Comparison of palisade cell numbers along the P–D and M–L axes. Leaf-shape mutants with differences of >10% (squares, 51 mutant lines) and <10% (circles, 29 mutant lines) in cell number index compared to the wild-type are shown (n = 8 for each mutant line). Closed squares and circles indicate the mutant lines shown in (c). The position of the wild-type is indicated by a diamond.
(c) Appearance of leaf-shape mutants. Fully expanded first leaves taken from 25-day-old plants are shown. From left to right, line 1027 (narrow leaf), line 2063 (long leaf), wild-type, line 2023 (wide leaf) and rot4-1D (short leaf).
Bar = 5 mm.

Leaf-shape mutants can be further sub-classified according to changes in the number of palisade cells along leaf axes. Mutants in which the cell number index differed by more than ±10% from that of the wild-type are shown in Figure 4(b). In many cases, the changes were caused by a greater decrease in the number of palisade cells along one axis than the other. For example, an3/gif1 produces narrow leaves because of a severe decrease in cell number along the M–L axis compared to the P–D axis (Horiguchi et al., 2005; Kim and Kende, 2004). Mutants that are associated with specific changes in cell number along one particular leaf axis have narrow, long, wide or short leaves (Figure 4c). Palisade cell numbers along the P–D or M–L axis in these mutants ranged from 75 to 125% of cell numbers in the wild-type (Figure 4b), suggesting the occurrence of positive and negative genetic pathways for leaf-shape control involving cell proliferation. The only characterized mutant in this sub-class is rot4-1D, which is associated with a specific decrease in cell number along the P–D axis (Narita et al., 2004) (Figure 4c). An alternative interpretation of axis-specific phenotypes is that these mutants are impaired in both appropriate patterning and cell proliferation; i.e. the altered cell number along a specific leaf axis is just a coincidence because of the weak phenotypes. These possibilities should be examined in future investigations of mutant alleles and their corresponding gene function.

Two other sub-classes had less obvious phenotypes in relation to the total number of palisade cells. Interestingly, these sub-classes were associated with an increase in the number of palisade cells along one axis at the expense of cells along the other axis. Thus, these mutations affected the direction of growth in proliferative tissues, rather than proliferating activity itself. Although a long-range signal that orients growth has been proposed in petals of Antirrhinum majus (Rolland-Lagan et al., 2003), how the direction of growth is guided in leaf primordia is not yet understood. Thus, these mutants may be a desirable starting material for the investigation of this issue.

Re-evaluation of hormonal mutants

The development of these simple, easy observation methods prompted us to re-characterize known hormonal mutants because they are known to affect growth and development, but lack quantitative descriptions of leaf morphology. We measured the leaf area, and the area and number of sub-epidermal palisade and adaxial epidermal pavement cells, and determined the stomatal index (Table 1).

We found several results that have not been previously reported. An early report of auxin resistant 1 (axr1) alleles suggested that axr1 mutations mainly affect cell number in relation to leaf development (Lincoln et al., 1990). However, we found that both cell number and size were reduced in axr1-3 and axr1-12 leaves, irrespective of cell type (Table 1). In contrast, when we examined the arf7 arf19 double mutant (Okushima et al., 2005b), we found a specific effect of these mutations on cell size. In the double mutant, cell proliferation was not affected at a statistically significant level, but cell size and leaf blade area were reduced to a similar extent to each other in comparison to the wild-type, as reported by Wilmoth et al. (2005). In contrast, the large-leaf phenotype of the arf2 mutant (Okushima et al., 2005a) was associated with an increase in the number and size of leaf cells (Table 1). These results indicate specific roles of particular members of the ARF gene family in promoting leaf cell expansion. The effects of mutations in other members of the ARF gene family on leaf cell proliferation and expansion should be examined.

We also found an overlooked phenotype in ethylene-related mutants. Although the constitutive triple response1 (ctr1) mutant is defective in cell expansion (Kieber et al., 1993), we found that cell proliferation was inhibited to an even greater extent than cell expansion in this mutant. The ctr1-12 mutant contained only 26% of the number of leaf cells found in the wild-type, which is one of the lowest cell numbers among all the mutants we examined, whereas its cell size was reduced to about 50% of that of the wild-type (Table 1). Likewise, the ethylene overproducer  1 (eto1) mutations (Guzmán and Ecker, 1990) also inhibited both cell expansion and proliferation (Table 1). Although the relationship between ethylene and cell division has been characterized by only a few studies (Kazama et al., 2004), the strong defect in cell proliferation in ctr1 indicates the importance of normal ethylene signaling during the proliferative phase of leaf development.

ABA and ethylene have opposing effects on leaf blade expansion (LeNoble et al., 2004; León and Sheen, 2003). Whereas etr1 and ein3 mutants produced larger leaves,aba deficient2 (aba2) and aba3 mutants produced smaller leaves than the wild-type. These differences were correlated with increases in both cell number and size for the etr1 and ein3 mutants, and decreases in both cell number and size for the aba2 and aba3 mutants (Table 1). We also examined the effect of the jasmonate resistant1 (jar1) mutation (Staswick et al., 2002) on leaf development; this mutation caused marginal decreases in cell number and size (Table 1). Thus, both cell proliferation and expansion were affected in most hormonal mutants examined, suggesting that these hormones play a role in the growth of whole organs. However, as demonstrated by the specific defect of arf7 arf19 mutants in leaf cell expansion, it is tempting to speculate that genes downstream of these hormone-related genes play a specific role in either cell proliferation or expansion. How hormonal perception and subsequent signal transduction are linked to the processes of cell proliferation and expansion remains to be determined.

Conclusions

We developed two simple methods for the histological observation of leaves that greatly facilitated the characterization of leaf mutants. Cellular-phenotype-based classification of several leaf size/shape mutants largely elucidated the genetic architecture of leaf expansion.

We showed a good correlation between leaf index and cell number index among our mutant lines. This suggests that although leaf shape in wild-type is very simple (almost circular in the case of the first leaves), the final leaf shape could be considered as the integrated output of several developmental regulatory pathways influencing morphogenesis. Two types of regulation would account for altered leaf shapes in these mutants. First, in every round of cell division, the orientation of the cell division plane relative to the leaf axes must be decided. If such a decision is affected by a mutation, leaf shape may be altered accordingly. The second possibility is that there are unknown sub-domains in leaf primordia where cell proliferation is locally promoted or inhibited. In this case, locally altered cell number is merely a consequence of mis-specification of a particular pattern rather than the cause of altered leaf shape. These possibilities are not mutually exclusive, and whether such regulations are operative in leaf morphogenesis or not is unclear. This important issue will be addressed in future work using our mutant lines.

For organ-size control, we have been able to classify specific mutants positively or negatively affecting either cell proliferation or expansion. These mutants will serve as valuable tools to examine the details of developmental processes in comparison with hormonal mutations, which have pleiotropic effects on cell proliferation and expansion in most cases. In terms of the cellular patterning involved in two-dimensional leaf expansion, we identified new polarity-dependent cell-proliferation mutants. Our mutant analysis classified mutants by similar cellular phenotypes. Mutations from the same or opposite classes may be involved in the same developmental pathway. Furthermore, mutants in the same class may be sub-classified by comparing their leaf development kinetics. Together with our earlier work on polarity-dependent cell expansion, revealed by the characterization of the an and rot3 mutants (Kim et al., 1998, 1999, 2002; Tsuge et al., 1996), the in-depth analysis of each mutant, as well as the examination of genetic interactions between different mutant classes, will reveal the organization of the gene network for leaf blade expansion.

Finally, we are planning to donate mutants isolated by us and described in this study to public seed banks after the accomplishment of more detailed characterization of several mutants. We hope that our mutant collection serves the Arabidopsis research community as a useful tool for developmental biology.

Experimental procedures

Plant materials

Wild-type plants and all mutants were of the Col-0 background. Mutants with a line number below 2000 were isolated from X-ray or gamma-ray-irradiated M2 populations, and those with a number 2000 and over were from T-DNA-tagged T2 populations. Seeds were sown on rock wool (Nittobo, Tokyo, Japan), and seedlings were grown at 22°C under a 16 h light/8 h dark cycle at a light intensity of 40 μmol m−2 sec−1, and were watered daily with 0.5 g l−1 Hyponex solution (Hyponex Japan, Osaka, Japan). The seeds of aba2-1, aba3-1, axr1-3, axr1-12, arf2-1, arf7-1,arf7-1arf19-1, ctr1-12, ein3-1, eto1-1, etr1-1 and jar1-1 mutants were obtained from the Arabidopsis Biological Resource Center (ABRC, Columbus, OH, USA).

Histological analysis

Whole leaves and leaf cells were observed under a stereoscopic microscope (MZ16a) and a Nomarski differential interference contrast microscope (DMRX E) (both from Leica Microsystems, Tokyo, Japan), respectively. Palisade cells in the sub-epidermal layer and adaxial epidermal cells in the center of the leaf blade between the mid-vein and the leaf margin were examined. The density of palisade cells per unit area in this region (0.4 μm2) was determined, and the area of the leaf blade was divided by this value to calculate the total number of palisade cells in the sub-epidermal layer. For estimation of the total number of epidermal cells, the mean leaf blade area was divided by the mean epidermal cell size per unit area (0.4 μm2). To determine the cell area, 20 palisade cells from each leaf were measured. To determine the cell number index, palisade cells were counted along the P–D axis about five-cell distance away from the mid-vein, and along the M–L axis at the widest region of the leaf blade. The stomatal index (SI) was determined according to the formula, SI = [S/(E + S)] × 100, where S is the number of stomata per unit area and E is the number of epidermal cells per unit leaf area (Mishra, 1997). All measurements were performed using Image J software (http://rsb.info.nih.gov/ij/).

Acknowledgements

We thank Dr H. Fukaki (Nara Institute of Science and Technology, Nara, Japan), Dr T. Shikanai (Kyushu University, Fukuoka, Japan) and Dr K. Torii (University of Washington, WA, USA), for seeds of the sgr1-1, paa1-4 and er-102 mutants, respectively. We thank Ms C. Yamaguchi and Ms M. Nagura for daily care of the plants. This work was supported by a Grant-in-Aid for Scientific Research on Priority Areas and for Young Scientists (B) from the Ministry of Education, Culture, Sports, Science and Technology of Japan, by a Grant-in-Aid for Creative Scientific Research from the Japan Society for the Promotion of Science, and by grants from the Bio-Design Program of the Ministry of Agriculture, Forestry and Fishes of Japan, the Toray Science Foundation, and the Sumitomo Foundation.

Ancillary