GASSHO1 and GASSHO2 encoding a putative leucine-rich repeat transmembrane-type receptor kinase are essential for the normal development of the epidermal surface in Arabidopsis embryos


(fax: +81 19 621 6107; e-mail


Receptor-like kinases (RLKs) containing leucine-rich repeats (LRRs) act as both signal receptor and signal transducer in ligand-mediated communication between cells. It is believed that many LRR-RLKs are present in the Arabidopsis genome, but the functions of most are unknown. We recently identified Bnms4D-82, an expressed sequence tag (EST) in Brassica napus that encodes an LRR-RLK and is expressed at an early stage of its microspore embryogenesis. To elucidate the function of this gene we used GASSHO1 (GSO1) and GSO2, two Arabidopsis genes with a high degree of homology with Bnms4D-82. The products of transcripts of GSO1 and GSO2 accumulate in parts of the embryo and in seedlings, but not in true leaves. Plants that lacked both GSO1 and GSO2 exhibited pleiotropy, including abnormal bending of embryos, ectopic adhesion between cotyledons, a highly permeable epidermal structure, and an abnormal pattern of distribution of stomata on cotyledons in embryos and seedlings. However, plants homozygous for either gso1-1 or gso2-1 had no visible abnormality. These results suggest that GASSHO genes are essential for the formation of a normal epidermal surface during embryogenesis.


In multicellular organisms, communication between cells through cell membranes and cell walls is an important mechanism for coordinating the functioning of many cells. An important channel for the exchange of information among cells is the secretion and detection of signal molecules, called ligands, in intercellular spaces. The detection of released ligands at the cell surface triggers signals inside the cell by means of phosphorylation, resulting in regulation of gene expression (reviewed by Takayama and Sakagami, 2002). Receptor-like kinases (RLKs), a key component of communication between cells, sustain ligand reception and intercellular signal transduction. The first RLK in plants was identified in Zea mays (Walker and Zhang, 1990). Since then, RLKs and their functions have been identified and characterized in Arabidopsis thaliana, and in a few other plant species. RLKs form one of the largest protein families, with more than 600 members, and account for approximately 60% of the entire kinase superfamily in Arabidopsis (Shiu and Bleecker, 2001, 2003).

Receptor-like kinases containing leucine-rich repeats (LRR-RLKs) form a large proportion of the RLK family in the Arabidopsis genome, and consist of three domains: an extracellular LRR domain, a single transmembrane domain, and a cytoplasmic serine/threonine kinase domain (Shiu and Bleecker, 2001). This subfamily includes some well-studied members, the functions of which were deduced from phenotypic analysis using corresponding mutants and/or ectopic overexpression lines. It was shown that LRR-RLKs have an important role in diverse plant signal transduction pathways during growth and development. The ERECTA gene, isolated as a causal gene of the erecta mutant in Arabidopsis, regulates organ shape and inflorescence architecture (Torii et al., 1996; Yokoyama et al., 1998). The Arabidopsis CLAVATA1 (CLV1) gene helps to maintain a balance between proliferation and differentiation in apical meristems (Clark et al., 1997; Schoof et al., 2000). Three CLAVATA1-related receptor kinases, BARELY ANY MERISTEM 1 (BAM1), BAM2 and BAM3, act positively in meristem development, and are required for the development of the male and female gametophyte and leaves (DeYoung et al., 2006). SERK (SOMATIC EMBRYOGENESIS RECEPTOR-LIKE KINASE) was identified as a marker gene indicating embryonic competence in Daucus carota (Schmidt et al., 1997). Hecht et al. (2001) demonstrated that ectopic expression of AtSERK1, a gene homologous to SERK in Athaliana, promotes somatic embryogenesis in culture. In addition, many aspects of plant development are regulated by LRR-RLKs, such as HAESA for the promotion of floral organ abscission (Jinn et al., 2000), RECEPTOR-LIKE PROTEIN KINASE 2 (RPK2) for the development of the anther (Mizuno et al., 2007), EXTRA SPOROGENOUS CELLS (EXS)/EXCESS MICROSPOROCYTES 1 (EMS1) and HAIKU 2 (IKU2) for seed growth (Canales et al., 2002; Luo et al., 2005; Zhao et al., 2002), and BRASSINOSTEROID INTENSIVE 1 (BRI1) and BRI1-ASSOCIATED RECEPTOR KINASE (BAK1) for brassinosteroid signaling (Li and Chory, 1997; Li et al., 2002). Although LRR-RLKs have important roles in various developmental processes, there has been no report of any contribution of LRR-RLKs to the establishment of the epidermis and its surface structure, except in Nodine et al. (2007), who reported that RECEPTER-LIKE PROTEIN KINASE 1 (EPK1) and TOADSTOOL 2 (TOAD2) are redundantly required for normal protoderm development during embryogenesis.

In Arabidopsis, the biochemical and molecular mechanism(s) underlying cuticle generation and deposition were deduced from mutant lines with a defective cuticle, which exhibit inter-organ fusion, retarded cell growth and differentiation, and a high level of permeability. Many genes encoding key enzymes for the biosynthesis of wax (Aarts et al., 1995; Chen et al., 2003; Krolikowski et al., 2003; Pruitt et al., 2000; Rowland et al., 2006; Todd et al., 1999) and cutin (Schnurr et al., 2004; Xiao et al., 2004), and for the export of wax (Pighin et al., 2004), have been reported. A few indirect regulators acting in transcriptional regulation (Aharoni et al., 2004; Brown et al., 2004; Hooker et al., 2007) and signal transduction (Tanaka et al., 2001, 2007; Watanabe et al., 2004) for cuticle generation and/or establishment of epidermal cells have been identified, and essential roles of RLKs for epidermal development have been demonstrated. In Z. mays, CRINKLY4 (CR4) is essential for epidermal differentiation, and the lack of CR4 leads to aberrant cell size and morphology, and inter-organ fusion (Becraft et al., 1996; Jinn et al., 2000). In Arabidopsis, an ACR4a homologue of CR4 is expressed in the protoderm and outer layer of the shoot apical meristem, and is involved in normal epidermal development (Tanaka et al., 2002; Watanabe et al., 2004). ABNORMAL LEAF SHAPE 2 (ALE2); an RLK related to epidermal development was recently discovered, and was shown to act in the same signal pathway as ACR4 (Tanaka et al., 2007). However, details of the signal pathway for epidermal development are unclear; the interrelationships of the regulators are unknown, and there are likely to be other components for signaling that have not yet been identified. Here, we report the identification of two novel LRR-RLKs acting in epidermal development.

In an earlier study with Brassica napus, we isolated the genes expressed differentially during the early stage of microspore embryogenesis (Tsuwamoto et al., 2007). Of the large number of expressed sequence tags (ESTs) isolated, Bnms4D-82 (DDBJ accession no. BP999916), which encodes an LRR-RLK, was expressed specifically in androgenic microspores and zygotic embryogenesis. In order to clarify the function of this gene, we carried out the functional analysis of Arabidopsis genes GASSHO1 (GSO1) and GASSHO2 (GSO2), which have a high level of homology to Bnms4D-82. Plants that lacked both GSO1 and GSO2 developed various defects, including a high degree of permeability, abnormal distribution and density of stomata, inhibited cell elongation, and abnormal fusion of plant parts in embryos and seedlings, similar to those shown by some Arabidopsis mutants with a defective cuticle. These results indicate that the two GASSHO genes are essential for the establishment of a normal epidermal surface, and may be involved in cuticle generation at the embryo stage of Arabidopsis.


Identification of gso1-1 and gso2-1 mutant lines

GSO1 (At4g20140) and GSO2 (At5g44700) genes encode putative LRR-RLKs that belong to the LRR XI subfamily (Figure S1), and are very similar to each other (85% of their amino acids are identical). In order to characterize GSO1 and GSO2, we obtained two Arabidopsis T-DNA insertion lines, SALK_064029 and SALK_130637, carrying a T-DNA insertion in GSO1 and GSO2, respectively, from the SALK T-DNA insertion collection (Alonso et al., 2003), and which were called lines gso1-1 and gso2-1, respectively. As shown in Figure 1(a), the first exons of GSO1 and GSO2 are separated by the insertion of each T-DNA region in mutant alleles; therefore, null alleles of GSO1 and GSO2 were expected in each mutant. To verify this, the gso1-1 and gso2-1 homozygotes were shown by RT-PCR to lack the GSO1 and GSO2 transcripts, as expected (Figure 1b).

Figure 1.

 T-DNA insertion position of gso1-1 and gso2-1 mutant lines and expression analysis of GSO1 and GSO2.
(a) T-DNA insertion site of gso1-1 and gso2-1. Boxes and lines represent exons and introns, respectively. Boxes show the positions and relative size of each predicted domain. Arrows indicate the general locations and directions of primers improved for the detection of a T-DNA insertion and RT-PCR. Scale bar: 500 bp.
(b) RT-PCR analysis of GSO1 and GSO2 in homozygous gso1-1 and gso2-1. To confirm the elimination of intact GSO1 and GSO2 transcripts, cDNAs derived from siliques of gso1-1 and gso2-1 homozygotes were used as templates for RT-PCR.
(c) RT-PCR analysis of GSO1 and GSO2 in various organs.

Organ-specific expression of GSO1 and GSO2 in developing embryos and seedlings

RT-PCR analysis of mRNA showed that the GSO genes were expressed in siliques, seeds, flower buds and roots, but not in leaves or stems (Figure 1c). The levels of the GSO1 and GSO2 transcripts were greater in siliques and seeds than in flower buds. In roots, the transcripts of GSO1 were present in only very small quantities, and GSO2 was not expressed. Aside from this difference, the general expression patterns of GSO1 and GSO2 were very similar, suggesting the possibility of functional redundancy between the two.

The promoter GUS assay was used to further analyze the expression profiles of GSO1 and GSO2. We constructed two binary vectors, GSO1pGUS and GSO2pGUS, containing the GUS gene, following each promoter sequence of GSO1 and GSO2 (Figure 2a). Arabidopsis plants were transformed with the constructs, and at least 10 independent lines for each construct were prepared for the GUS assay. GUS activity was detected in the developing embryos, seedlings and flower buds in both GSO1pGUS and GSO2pGUS transformants. During embryogenesis, uniform expression of GSO1 extended from the globular embryo to the mature cotyledonary embryo (Figure 2b, A–D). An identical pattern of GUS expression was found in GSO2pGUS transformants, except that GSO2 was not expressed in the embryo suspensor (Figure 2b,f–i). After germination, both were expressed in whole cotyledons and in the hypocotyl (Figure 2b,e,j). As predicted by the RT-PCR analysis, additional expression of GSO1 in roots was found in the promoter-GUS assay (Figure 2b,e,k), whereas GUS activity was not detected in true leaves (Figure 2b,k,l). In flower buds, GSO1 was expressed in filaments and in the stigmas (Figure 2b,m,o), whereas GSO2 was expressed in pollen grains and the separation layer between the bud and the peduncle (Figure 2b,n,p).

Figure 2.

 Expression pattern of GSO1 and GSO2 in developing embryos embedded in ovules, seedlings and flower buds.
(a) Structure of constructs used for promoter GUS assay.
(b) GUS staining of transgenic GSO1pGUS and GSO2pGUS Arabidopsis plants. GUS expression patterns in developing embryos and seedlings of transgenic GSO1pGUS (A–E, K, M and O) and GSO2pGUS (F–J, L, N and P) plants were determined at different stages. A and F: globular stage. B and G: late-heart stage. C and H: late-torpedo stage. D and I: mature embryos at cotyledon stage. E and J: seedlings with cotyledons. K and L: seedlings with true leaves. M and N: flower buds. O and P: anther from open flower. Scale bars: 100 μm in A–D, F–I, O and P, and 1 mm in E and J–N.

Morphology during embryogenesis

To determine whether GSO1 and GSO2 contribute to the normal development of embryos and seedlings, gso1-1 and gso2-1 homozygotes were screened by genomic PCR, selfed, and then examined for morphological abnormalities. Embryos at various developmental stages, mature seeds, and seedlings of gso1-1 and gso2-1 did not differ morphologically from those of the wild type. Because GSO1 and GSO2 were expressed synchronously, and their mutants did not exhibit different phenotypes, we produced gso1-1gso2-1 double mutant lines to confirm the functional redundancy of GSO1 and GSO2.

When developing embryos were examined under a light microscope, the gso1-1 gso2-1 embryos were indistinguishable from those of the wild type until the early heart stage (Figure 3a,f). However, at the heart–torpedo transition stage, they were expanded laterally, with cotyledons that were more concave and divergent than those of the wild type (Figure 3b,g). Furthermore, at the late torpedo stage, a cotyledon of the gso1-1 gso2-1 embryos adhered to the peripheral tissue of the endosperm, whereas those of the wild type developed normally (Figure 3c,h). In the final stages, the gso1-1 gso2-1 embryos were bent in reverse, as compared with wild-type embryos, perhaps as a result of adhesion between cotyledons and the peripheral tissue of the endosperm (Figure 3d,e,i,j). Because of this, gso1-1gso2-1 seeds appeared slightly contorted. Despite their slight contortion, the germination rate of gso1-1 gso2-1 seeds was unaffected (data not shown).

Figure 3.

 Morphological analysis of developing embryos in wild-type and gso1-1 gso2-1 homozygous plants. Ovules from wild-type (a–e) and gso1-1 gso2-1 homozygous (f–j) plants containing developing embryos were examined at various stages of seed development.
a and f: heart stage.
b and g: late-heart stage.
c and h: late-torpedo stage.
d and i: curled cotyledon stage.
e and j: mature embryo. Scale bar: 100 μm.

Seedling growth

To observe the seedlings of both gso1-1 gso2-1 and the wild type, the seeds were sown on an MS plate (in vitro conditions), treated at 4°C for 4 days for stratification and then grown at 25°C. At 4 days after stratification treatment (4 DAS), seedlings of the wild type had a well-elongated hypocotyl and separated cotyledons (Figure 4a), whereas gso1-1 gso2-1 seedlings showed several abnormalities: the hypocotyl was short and the cotyledons were concave (Figure 4b), and in 84 of 127 seedlings (66%) the cotyledons adhered each other at their adaxial surfaces (Figure 4c). At 6 DAS, true leaves were visible from the opening between the edges of the adherent cotyledons in gso1-1 gso2-1 seedlings (Figure 4d); as the true leaves expanded, the adherent cotyledons were gradually pushed apart and eventually separated, and no phenotypic difference was apparent between the two groups of seedlings after the emergence of true leaves.

Figure 4.

 Abnormal phenotypes of gso1-1 gso2-1 seedlings and adult plants.
a: wild-type seedlings.
b: gso1-1 gso2-1 seedling with spread-out cotyledons.
c: gso1-1 gso2-1 seedling with adhered cotyledons at 4 days after stratification treatment (4 DAS).
d: gso1-1 gso2-1 seedling with true leaves from lateral apertures of attached cotyledons at 6 DAS. Arrows indicate true leaves.
e: overview of 40-day-old wild-type (right) and gso1-1 gso2-1 (left) plants. The growth of gso1-1 gso2-1 plants was limited markedly under open conditions. Scale bar: 1 mm in (a)–(d) and 5 cm in (e).

On the other hand, 34 of 36 gso1-1 gso2-1 seedlings grown in pots under open conditions withered (Figure 4e), and gso1-1 gso2-1 seedlings grown in pots under high humidity conditions (each pot was covered with a polyethylene bag to maintain high humidity) showed no lethality (data not shown). These results suggest that the ability to retain water had been affected in gso1-1 gso2-1 seedlings.

Permeability of the epidermis

We used the toluidine blue (TB) test (Tanaka et al., 2004), a technique for making defects in the cuticle rapidly visible, to confirm the high level of permeability of the epidermal surface in gso1-1 gso2-1 seedlings. Seedlings of plant lines, wild-type, gso1-1, gso2-1 and gso1-1 gso2-1, were grown under in vitro conditions and then examined. In gso1-1 gso2-1 seedlings at 4 DAS, TB passed through the cuticle and permeated into the inside, whereas the cuticle was intact in all other lines (Figure 5a–d). The true leaves of the wild-type and gso1-1 gso2-1 lines repelled the TB solution, indicating that the true leaves of gso1-1 gso2-1 have an intact and normal cuticle (Figure 5f,g). Therefore, it can be inferred that the function of GSO1 and GSO2 may be limited to juvenile stages, that is the embryonic cotyledons and hypocotyl. This is supported by their specific expression in juvenile organs, as already shown.

Figure 5.

 Estimation of permeability by the toluidine blue test.
a and f: seedlings of wild-type at 4 days after stratification treatment (4 DAS) and 6 DAS, respectively.
b: seedling of gso1-1 at 4 DAS.
c: seedling of gso2-1 at 4 DAS. d and g: seedlings of gso1-1 gso2-1 at 4 DAS and 6 DAS, respectively. e: seedling of gso1-1 gso2-1 harboring GSO2 transgene at 4 DAS. Scale bar: 1 mm.

Genetic complementation of gso1-1 gso2-1

For genetic complementation, gso1-1 gso2-1 plants were transformed with a genomic fragment containing full length of GSO1 or GSO2, and homozygotes for each transgene were screened from progenies of the obtained T0 plants (the homozygosity of plants was inspected by segregation of hygromycin resistance). As shown in Figure 5(e), seedlings of gso1-1 gso2-1 with the GSO2 transgene repelled the TB solution. In this assay, at least three independent complemented lines for each of GSO1 and GSO2 were used, and the seedlings of gso1-1 gso2-1 with the GSO1 transgene showed the same results as with GSO2 (data not shown). Furthermore, aberrant embryogenesis as observed in gso1-1gso2-1 (shown in Figure 3) was not observed in all of the complemented lines (data not shown). Such complete restoration by introgression of GSO1 or GSO2 indicates that various abnormalities were caused by the lack of GSO1 and GSO2.

Cellular anatomy of developing seedlings

We observed the external structure of gso1-1 gso2-1 seedlings by scanning electron microscopy (SEM), and the internal structure (through transverse sections) by light microscopy. In wild-type seedlings, the epidermal cells of the hypocotyl were elongated and their surface was smooth (Figure 6a,c). By contrast, in gso1-1 gso2-1 seedlings the hypocotyl was short, and its epidermal cells were compressed and showed a rough surface (Figure 6b,d); in addition, the endosperm tissue was partially adherent to the surface of the cotyledons (Figure 6e). The SEM revealed that the cotyledons of gso1-1 gso2-1 adhered to each other with their epidermal surfaces (Figure 6f).

Figure 6.

 Changes in the epidermal structures of the gso1-1 gso2-1 seedlings.
a–f: SEM observations of surface structures in wild-type (a, c) and gso1-1 gso2-1 (b, d, e and f) seedlings.
a and b: hypocotyl.
c and d: magnified images of a and b, respectively.
e: adhered cotyledons of a gso1-1 gso2-1 seedling. Arrows point to accretions from maternal tissues.
f: magnified image of (e), focused on the adhered region of cotyledons.
g and h: transverse sections of hypocotyls of wild-type (g) and gso1-1 gso2-1 (h) plants. Arrows indicate disorganized epidermal cells.
i: comparison of hypocotyl length.
j: comparison of epidermal cell length of hypocotyls. Scale bars: 50 μm in (a)–(d) and (f), 250 μm in (e), and 100 μm in (g) and (h).

The internal structure was not affected by the disruption of GSO1 and GSO2: the basic radial pattern was unchanged, although cells in the cortex and the epidermis were arrayed irregularly in a few places (Figure 6g,h). The length of the hypocotyl was measured in gso1-1, gso2-1, gso1-1 gso2-1 and in the wild-type at 4 DAS (n = 30), and the length of epidermal cells in the hypocotyls of gso1-1 gso2-1 and wild-type seedlings was measured (Figure 6i,j). The hypocotyl in gso1-1 gso2-1 was clearly shorter than that in other lines. However, the lengths did not differ significantly among wild type, gso1-1 and gso2-1. The epidermal cells of gso1-1gso2-1 were significantly shorter than those of the wild type. The hypocotyl length was 2.6-fold smaller than the wild type, and the epidermal cell length was 2.1-fold smaller than the wild type.

External and internal structures of cotyledons

The external and internal structures of the gso1-1 gso2-1 cotyledons were observed by the methods used with the hypocotyls. Most stomata are surrounded by three subsidiary cells (anisocytic), and are distributed equally over the surface of the cotyledons in Arabidopsis. Furthermore, stomata are never adjacent (adhering to the one-cell spacing rule; Yang and Sack, 1995). These features were apparent at 4 DAS in the abaxial epidermis of cotyledons of wild-type plants grown under in vitro conditions (Figure 7a,e). By contrast, cotyledons of the gso1-1 gso2-1 seedlings had more stomata and did not conform to the one-cell-spacing rule. As seen in Figure 7(c), the stomatal index of these cotyledons was 1.6 times that of the wild-type. Also, most of the stomata were distributed in clusters and were surrounded by many unexpanded subsidiary cells (Figure 7f). The surface covering the stomata appeared to be abraded (Figure 7d).

Figure 7.

 External and internal structures of the gso1-1 gso2-1 cotyledons. SEM observations of epidermis on the abaxial side of cotyledons at 4 DAS (a–f). Transverse sections of gso1-1 gso2-1 cotyledons at 4 DAS stained by toluidine blue (g and h).
a: wild type. b: wild type stoma. c: gso1-1 gso2-1. d: gso1-1 gso2-1 stoma. e and f: magnified images of a and c, respectively. g: internal structure of fused gso1-1 gso2-1 cotyledons. h: internal structure of wild-type cotyledons. Abbreviations: ab, abaxial side; ad, adaxial side. Scale bar: 100 μm in a, c and g, 25 μm in e and f, and 50 μm in h.

Almost all cells constituting the mesophyll of gso1-1gso2-1 were thicker than those of the wild type, resulting in cotyledons that were about twice as thick as those of the wild type (Figure 7g,h). None of these abnormalities was observed in true leaves, suggesting that the role of GSO1 and GSO2 is confined to the embryo. No defect was found in the roots or flower buds of gso1-1 gso2-1 plants, although expression of GSO1 and/or GSO2 was detected in these organs (Figure 2). This result may indicate no function of GSO1 and GSO2 in these organs, or the existence of another unidentified LRR-RLK that is interchangeable with GSO1 and/or GSO2.


Plants that lacked both GSO1 and GSO2 genes encoding LRR-RLKs showed pleiotropy only during the embryo stage and the seedling stage. The absence of abnormalities in gso1-1 gso2-1 plants transformed by either GSO1 or GSO2 confirms the cause-and-effect relationship between these LRR-RLKs and the phenotypes exhibited in gso1-1 gso2-1 plants. Accordingly, the existence of an unknown signal pathway, or pathways, that requires either GSO1 or GSO2 for the normal development of epidermal structure in the embryo, is predicted. Because of its molecular identity, some cases of a single LRR-RLK being implicated in disparate events are known. For instance, the ERECTA gene has a role in the development of the inflorescence and in the pattern of stomatal distribution (Shpak et al., 2005), control of transpiration efficiency (Masle et al., 2005) and resistance to bacterial wilt (Godiard et al., 2003). The gso1-1 gso2-1 double mutant displayed various pleiotropic phenotypes. However, we consider the defect in the epidermal surface to be fundamental, causing all the other phenotypes, because of its similarity to some other Arabidopsis mutants with a defective cuticle, and we propose that GSO1 and GSO2 may mediate unknown local signal(s) essential for the formation of a normal epidermal surface during the development of the embryo.

Some Arabidopsis plants harboring mutations in genes required for wax and/or cutin biosynthesis/deposition generate a defective cuticle that cannot prevent fusion of different plant parts (Aarts et al., 1995; Chen et al., 2003; Jenks et al., 1995, 1996; Krolikowski et al., 2003; Kurata et al., 2003; Lolle et al., 1998; Pruitt et al., 2000; Wellesen et al., 2001; Yephremov et al., 1999; Zheng et al., 2005), as was the case with the cotyledons of gso1-1 gso2-1. A deficit of the waxy layer is known to impair an ability to retain water (Riederer and Schreiber, 2001); conversely, excessive deposition of wax is reported to decrease permeability because of the altered ultrastructure of the cuticle (Aharoni et al., 2004). We found an abnormal, rough epidermal surface (Figures 6 and 7), and a drastically increased permeability in gso1-1gso2-1 seedlings (Figure 5), suggesting that the fused cotyledons and low level of water retention are secondary effects of a defective epidermis.

During embryo development, more than 70% of gso1-1gso2-1 embryos were abnormally bent (Figure 3). We consider such bending to be yet another effect of a defective epidermal surface that prevents normal separation of plant organs (as seen in Figure 8). This hypothesis is supported by the remnants of endosperm tissue on the epidermal surface of the gso1-1 gso2-1 seedlings (Figure 6e), and indicates that a normal epidermal structure in embryos is necessary to prevent abnormal fusion between the embryo and surrounding tissues, such as the endosperm. Similar results were obtained by Tanaka et al. (2001, 2002), who found embryos abnormally fused with the endosperm in the abnormal leaf shape 1 (ale1) mutant, and with the seed coat in the Arabidopsis thaliana homologue of crinkly 4 (acr4) mutant. The former gene encodes a subtilisin-like serine protease expressed in the endosperm surrounding an embryo, and the latter encodes an RLK expressed in only the protoderm during embryogenesis, and both have a role in independent signal pathways underlying the differentiation of epidermal structure, including the cuticle (Watanabe et al., 2004). Recently, Tanaka et al. (2007) reported that the ALE2 acts in the same process as ACR4, and that ACR4, ALE1 and ALE2 promote the establishment of epidermal structures through the expression of epidermis-specific genes. Although the relationships of two GSO proteins with these factors were not examined, similar functions are expected because of similar phenotypes in each mutant embryo. In contrast, despite ectopic adhesion during embryogenesis, these mutants did not show the abnormal bending and adhesion of cotyledons found in gso1-1gso2-1, which implies that another factor is involved in the abnormal bending. We regard the distinctive form of the malformed embryo to be a crucial factor: at the heart–torpedo transition stage, the gso1-1 gso2-1 embryo, just before developing abnormal bending, displayed distinctly outspread cotyledons, and may have been able to access and adhere the peripheral tissue of the endosperm more easily than acr4 and ale1 embryos. Furthermore, such abnormal bending may facilitate tighter adhesion of cotyledons in gso1-1gso2-1 by exerting pressure. On the basis of these considerations, we conclude that the aberrant bending is an incidental result of a defective epidermis.

Figure 8.

 Putative mechanism of irregular bending of the gso1-1 gso2-1 embryos.
At the early-heart stage, no difference was apparent between wild type and the gso1-1 gso2-1 mutant (a). The wild-type embryo continues to expand with straight, not spread-out, cotyledons at the heart–torpedo transition stage (b). In contrast, formation of concave and spreading cotyledons is observed in the gso1-1 gso2-1 embryo at this stage (c). In addition, because of abnormal epidermal structures such as a deficient cuticle layer, the embryo expands and an apical part of the embryo is adhered to the endosperm peripheral tissue. At the late-torpedo stage, the wild-type embryo with intact epidermal structure is bent along the inner wall (d), resulting in a normal mature embryo (f); in the gso1-1 gso2-1 embryo, the cotyledon adhering to the endosperm peripheral tissue arrests ordinary vertical development (e), and causes the embryo to bend in reverse (g).

We cannot identify the immediate cause of the malformed gso1-1 gso2-1 embryos from our results. However, it is likely that the naked embryo with a defective epidermal structure, e.g. lacking a protective coat in the form of the cuticle, is affected by the environment, because the cuticle offers essential protection against biotic and abiotic environmental factors (Barnes et al., 1996; Eigenbrode and Espelie, 1995; Jenks et al., 1994). The gso1-1 gso2-1 seedlings exposed to the environment showed serious histological defects, including short hypocotyl, repression of longitudinal expansion of epidermal cells, and thickened and partly disorganized tissues. Similar histological defects were found in the wax2 and eceriferum10 (cer10) mutants, which cannot generate an intact cuticle because of disruption of the biosynthesis of wax and cutin (Chen et al., 2003; Zheng et al., 2005). We consider that the repression of longitudinal expansion of hypocotyl cells was a major cause of the shorter hypocotyls of gso1-1 gso2-1, because the ratio of diminution in length between hypocotyl and epidermal cells was similar. Although the length of the internal cells was not measured, serious histological defects observed in gso1-1 gso2-1 hypocotyls are compatible with the repression of longitudinal expansion of internal cells. Furthermore, in adhered cotyledons, epidermal cells on the abaxial, exposed exterior were smaller than those on the adaxial, less exposed exterior (Figure 7g). These observations support our contention that the histological defects in gso1-1 gso2-1 are the result of severe stress caused by abnormal epidermal structures, such as a defective cuticle, that offer poor protection against an adverse environment.

In addition to the features described above, the distribution pattern of stomata was different in some mutants with an abnormal cuticle. Of these, cer1 and cer6 wax-accumulation mutants increase the stomatal index (Gray et al., 2000), as was seen with the gso1-1 gso2-1 mutants. On the contrary, mutations in WAX2 and overexpression of the AP2 domain transcription factor SHINE1/WIN1, which result in considerable wax deposition, decrease the stomatal index (Aharoni et al., 2004; Brown et al., 2004; Chen et al., 2003). Although the relationship between the stomatal density and a defective cuticle are not well understood, these results indicate that alteration of wax biosynthesis is related to abnormal stomatal density. Gray et al. (2000) proposed a possible model: in normal plants, the cuticle overlying the epidermal cells may act as a passage for the diffusion of molecules that are released from guard cells, and repress the ability of epidermal cells to develop into stomata. The higher stomatal index in gso1-1 gso2-1 plants may have been caused by the restricted diffusion of such a repressor resulting from the defective development of the epidermal surface.

We have described the pleiotropic phenotypes of gso1-1 gso2-1 and the expression profiles of each gene, and conclude that GASSHO is essential for the development of normal epidermal structure in embryos. Although further investigations, such as transmission electron microscopy (TEM) observations of the epidermal surface and determination of the changes in composition of the cuticle in gso1-1 gso2-1 seedlings, remain to be carried out, the phenotypic similarity between gso1-1 gso2-1 and some known cuticle mutants strongly implies the involvement of GASSHO genes in cuticle generation via an unknown pathway. The action of GSO1 and GSO2 proteins as LRR-RLKs, e.g. intracellular localization, catalytic activity, and the specific interaction with some other effectors that participate in the GASSHO signal pathway, is unknown, and additional information is required on exactly how GSO1 and GSO2 contribute to the normal development of the epidermis.

Experimental procedures

Plant lines and growth conditions

Arabidopsis thaliana ecotype Columbia (Col-0) was used as the wild type. The T-DNA insertion lines gso1-1 (SALK_064029) and gso2-1 (SALK_130637) were identified in the SIGNAL database ( and were obtained from the Arabidopsis Biological Resource Center (ABRC, To screen for T-DNA insertions in GSO1 and GSO2, PCR analysis was carried out with the gene-specific primers GS1F (5′-GGTCAACCGGGTATCATCAACA-3′) and GS1R (5′-AAGCGAATTTCCATCAAGAGAC-3′) for GSO1, and GS2F (5′-AACTTGCAAGCTCTCAATGCGT-3′) and GS2R 5′-TGGTGTCATAATTCCCGGTTAG-3′) for GSO2, in combination with T-DNA left border primer Lba1 (5′-TGGTTCACGTAGTGGGCCATCG-3′). To generate the gso1-1 gso2-1 double mutant line, gso1-1 and gso2-1 homozygotes were crossed reciprocally, and the subsequent generation was selfed. The resultant progeny was screened by PCR, as described above, for gso1-1 gso2-1 homozygotes. Seeds of each line were sown on pots filled with a mixture of vermiculite and rock wool (open conditions) or sterilized seeds were sown on MS plates of agar (0.8%)-solidified Murashige and Skoog medium (in vitro conditions). After sowing, the pots and MS plates were kept at 4°C for 4 days for stratification, and were then transferred to an environmental chamber at 25°C under a 16-h:8-h light:dark photoperiod. The age of plants is given in DAS.

RT-PCR analysis

Poly(A)+ RNA was extracted from leaves, stems, roots, flower buds, siliques and seeds of wild-type plants with a Micro Fast Track Kit (Invitrogen, First-stranded cDNA synthesis from 150 ng of mRNA was conducted with a First-Strand cDNA Synthesis Kit (Amersham, PCR analysis was performed for 1 min at 94°C followed by 28 cycles of 30 sec at 94°C, 30 sec at 57°C and 1 min at 72°C, with the gene-specific primers used for the genotyping of T-DNA insertion lines. The same procedure was used to amplify GSO1 and GSO2 transcripts from cDNAs derived from gso1-1 and gso2-1 homozygous plants in order to confirm the knock-out of GSO1 and GSO2 transcripts; the cDNA was prepared from siliques of gso1-1 and gso2-1 homozygotes. As a control in all RT-PCR reactions, the transcript of the actin gene was amplified for 25 cycles of 30 sec at 94°C, 30 sec at 57°C and 30 sec at 72°C using the primers ACTf (5′-GGTGTGTCTCACACTGTGCCAA-3′) and ACTr (5′-AATTTCCCGCTCGGCTGTT-3′).

Plasmid construction and plant transformation

Genomic fragments containing the entire GSO1 (5649 bp) and GSO2 (5710 bp) genes were amplified by PCR with the following primers: GS1-cF (5′-CTATTACTAATAAACAACTA-3′), GS1-cR (5′-GGTTCTAGAGTTAGATAGGTT-3′), GS2-cF (5′-TGAGAAACTAACACACTTCAC-3′) and GS2-cR (5′-AGAAACTATAATCGCAAACGA-3′). Each of the amplified DNA fragments was blunted and cloned directly into the HindIII/SalI-digested pIG121 Hm vector using a BKL kit (TaKaRa, These vectors were named GS1c and GS2c, respectively, and were used for the genetic complementation of the gso1-1 gso2-1 double mutant line.

To fuse the GSO1 and GSO2 promoters to the GUS gene, the putative promoters of GSO1 and GSO2, namely the 2598-bp fragment upstream of the ATG of GSO1, and the 2350-bp fragment upstream of the ATG of GSO2, were amplified by PCR with the following primers: GS1-pF (5′-CTCGAGGGTTAAGTATAGAATGCACAGAAAG-3′; XhoI site attached to the 5′ end) and GS1-pR (5′-GTCGACTTCGTCGTCTTATGGTTTGGTTTAC-3′; SalI site attached to the 5’ end) for the GSO1 promoter, and GS2-pF (5′-AAGCTTTCCATAAGATAAGACTTGTACAC-3′; HindIII site attached to the 5’ end) and GS2-pR (5′-GTCGACTTCAAGAAACAAGAGATCTGTG-3′; SalI site attached to the 5’ end) for the GSO2 promoter. Each PCR product was cloned into a pCR2.1-TOPO vector (Invitrogen). The GSO1 and GSO2 promoter fragments were digested by restriction enzymes corresponding to the restriction sites attached to the primers, and were cloned into a pBI121 vector in place of the CaMV 35S promoter, resulting in GS1pGUS and GS2pGUS vectors, respectively.

The GS1c and GS2c vectors were electroporated into Agrobacterium tumefaciens strain EHA101 and were used for transforming the gso1-1 gso2-1 homozygous plant by the floral-dip method (Bechtold and Pelletier, 1998). As to the GS1pGUS and GS2pGUS vectors, A. tumefaciens strain EHA105 was used for transforming the wild-type plants. Seeds were screened on MS medium (0.8% agar) with 50 mg l−1 kanamycin for GS1pGUS and GS2pGUS transformants, or 30 mg l−1 hygromycin B for GS1c and GS2c transformants.

GUS assay

In the GS1pGUS and GS2pGUS transformants, various organs from plants grown on soil or MS plates were examined for their GUS expression patterns. GUS staining was performed as described by Jefferson (1987). The dissected organs were transferred to a solution of 200 mm sodium phosphate buffer (pH 7.0), 12.5 mm potassium ferricyanide, 12.5 mm potassium ferrocyanide, 0.3% Triton X-100, 20% methanol and 38.3 mm 5-bromo-4-chloro-3-indolyl-β-d-glucuronide, and were kept overnight at 37°C. The stained organs were washed with 70% ethanol and cleared in lactophenol (water/glycerol/lactate/phenol, 1:1:1:2, by volume) overnight. The cleared samples were observed by light microscopy.

Toluidine blue test

A partially modified TB test (Tanaka et al., 2004) was used to visualize the defective epidermis in seedlings. At 4 and 6 DAS, seedlings grown under in vitro conditions were dipped into an aqueous solution of 0.05% (w/v) TB that had been sterilized by passage through a filter of 0.45-μm pore size. Five minutes later, the seedlings were transferred to water and were then washed gently three times to remove any residual TB.


To observe embryonic development, ovules containing developing embryos were dissected from green siliques of gso1-1 gso2-1 homozygous and wild-type plants at various stages, and were cleared overnight in Hoyer’s solution (Liu and Meinke, 1998). The cleared ovules were observed by light microscopy using a differential interference contrast unit. Morphological changes in seedlings of the gso1-1 gso2-1 homozygote were studied as follows. Seeds were sown and germinated under in vitro conditions, as described above, and seedlings were observed under a stereomicroscope for an overview and for the measurement of the hypocotyl length. For transverse sections and SEM, the seedlings were fixed overnight at 4°C in a fixative containing 0.25% glutaraldehyde and 4% paraformaldehyde in 50 mm sodium phosphate buffer (pH 7.0), and were then dehydrated through a graded ethanol series. Some of the fixed seedlings were then embedded in Technovit 7100 (Heraeus Kulzer, and sliced into 10-μm-thick transverse sections. Sections were stained in 0.05% TB and viewed under a light microscope. The remaining dehydrated seedlings were soaked in isoamyl acetate, dried to the critical point and coated with gold for SEM (S-250N FE-SEM; Hitachi, However, with this method we could not obtain a clear image of the epidermal cells of the cotyledons. To prevent deformation of the cotyledon epidermal cells during fixation, we tried the rubber imaging technique. Provil Novo (Heraeus Kulzer) was applied to the abaxial side of the cotyledon and then peeled away carefully, resulting in a mold that could be replicated. The mold was filled with epoxy resin and cured at 60°C for 1 h. The completed cast was removed, sputter-coated with gold and used for SEM observation. However, alterations in the minute surface structure could not be imaged by this method, and the common technique described above was used.

Stomatal indices for the abaxial surface of wild-type and gso1-1 gso2-1 cotyledons at 4 DAS were calculated by the following formula: [stomatal density/(stomatal density + epidermal pavement cell density)] × 100 (Woodward, 1987). Stomatal density and epidermal pavement cell density within a 200-μm × 200-μm square were measured by SEM and averaged over eight replications, with resin casts made independently each time.


This work was supported by a Grant-in-Aid for Scientific Research (KAKENHI) from the Ministry of Education, Culture, Sports, Science and Technology, Japan. RT is a recipient of the student support program of Grants-in-Aid for the 21st Century COE Program from the Ministry of Education, Culture, Sports, Science and Technology, Japan. The authors thank Ayako Ueyama for technical assistance with the SEM.