Initiation of shoot apical meristem in rice: characterization of four SHOOTLESS genes

In flowering plants, the fundamental body plan is constructed during embryogenesis. The fertilized egg first sets up apical-basal, radial and dorsal-ventral axes, which are later partitioned to differentiate region-specific organs. The shoot and root apical meristems arise as terminal elements of apical-basal axes. The establishment of shoot apical meristem (SAM) during embryogenesis is a key event in plant development, because the above-ground body plan depends on the activity of SAM through the production of leaves, axial buds and stem. A genetic approach utilizing developmental mutants is an efficient way to address how SAM is established in the plant embryo. Mutants affecting SAM differentiation have been identified in several plant species. In Arabidopsis, shoot meristemless (stm) mutations include strong, intermediate and weak alleles. The strong alleles cause deletion of the SAM, thus indicating that STM is required for the initiation and maintenance of the SAM throughout the life cycle of the plant (Barton and Poethig, 1993; Clark et al., 1996; Endrizzi et al.,1996). Double mutant analyses reveal that STM acts competitively with CLAVATA1 to regulate meristem size in the upstream of WUSCHEL and ZWILLE (Clark et al., 1993, 1995, 1996; Jurgens et al., 1994; Laux et al., 1996; Leyser and Furner, 1992). In addition, STM encodes a member of the KNOTTED class of homeodomain proteins (Long et al., 1996). A petunia NO APICAL MERISTEM (NAM) gene is considered to be associated with the determination of SAM position, as deduced from the anatomical observation, sequence data, and in situ hybridization analysis (Souer et al., 1996). These genes play important roles in the establishment of the SAM, but the genetic program leading to SAM formation must involve a number of genes. In maize, at least three mutations cause a shootless phenotype without any other disruption (Sheridan and Clark, 1987, 1993; Sheridan, 1988; Clark and Sheridan, 1991), although they have not yet been well characterized. Also, in rice, at least four loci cause a shootless phenotype by their recessive mutations (Hong et al., 1995). Therefore, more mutations must be analyzed for elucidating the genetic cascade toward SAM establishment.

To date, a few genes giving the loss-of-SAM phenotype by the recessive mutations have been cloned. Arabidopsis homeobox gene STM is the only cloned gene known to function in the initiation of SAM. Although the maize KN1 gene is homologous to STM and has been suggested to be involved in SAM differentiation, the loss-of-function mutation of KN1 produces almost no phenotypical change in the embryo (Kerstetter et al., 1997). However, considering the expression site of KN1 (Smith et al., 1995) and the polyploid origin of maize (Gaut and Doebly, 1997), KN1 has been suggested to be associated with the differentiation and maintenance of SAM sharing redundancy with other genes (Kerstetter et al., 1997). Rice homeobox gene, OSH1, a presumed ortholog of KN1 (Matsuoka et al., 1993), is expressed in a specific region of globular embryos where SAM is differentiated later (Sato et al., 1996). Although there is no direct evidence that OSH1 is involved in SAM initiation, OSH1 will be a useful molecular marker of embryo differentiation. Apart from the SAM, grasses have highly developed and complicated embryos, which raises controversial problems concerning the identities and functions of embryo-specific organs. For example, the cotyledon in grass embryos has been variously interpreted, based on anatomical and histological studies (Brown, 1960). Because genetic aspects have not been explored, analyses of embryo mutants will give new insights on these problems.

We have identified nine shootless mutants of rice derived from four loci (Hong et al., 1995). The characterization of shootless mutants in rice will shed light on the genetic regulation of SAM initiation. Furthermore, the analyses of shootless mutants will provide new information on the relationships among embryonic organs and on the nature of leaf primordial development.

We used nine recessive shootless mutants of rice (Oryza sativa L.) derived from four independent loci; two alleles (shl1-1 and shl1-2) at SHL1 (SHOOTLESS 1), five alleles (shl2-1, shl2-2, shl2-3, shl2-4 and shl2-5) at SHL2, shl3 at SHL3, and shl4 at SHL4 (Hong et al., 1995). All the mutants have a genetic background of cv. Taichung 65 except shl2-5, which was derived from cv. Kinmaze, and were obtained from chemical mutagenesis using N-methyl-N-nitrosourea. These mutations were maintained in a heterozygous state. We used wild-type siblings of each mutant for the comparison with the mutant phenotypes.

Mature shootless seeds were identified under a dissecting microscope by their external appearance, with the absence of the coleoptile, and were confirmed by the absence of shoot germination within 2 days after imbibition. The seeds were sterilized with 2% sodium hypochlorite for 30 minutes, and washed with sterilized water. For the induction of calli, they were inoculated onto N6 medium (Chu et al., 1975) supplemented with 30 g/l sucrose, 2 mg/l 2,4-dichlorophenoxyacetic acid and 2 g/l Gelrite. The pH was adjusted to 5.8. The cultures were incubated at 28°C. When calli proliferated to approximately 5 mm in diameter, they were transferred onto regeneration medium containing MS salt and vitamins (Murashige and Skoog, 1962), 30 g/l sorbitol, 30 g/l sucrose, 2 g/l casamino acid, 2 mg/l benzylaminopurine, 1 mg/l naphthalene acetic acid and 4 g/l Gelrite. After 3 weeks of culture, calli, which did not regenerate shoots, were transferred onto new regeneration medium. Calli at various stages after the transfer onto regeneration medium were fixed for plastic sectioning, scanning electron microscopy and in situ hybridization.

Developing seeds at various stages and mature seeds were fixed in FAA (formalin:glacial acetic acid:70% ethanol, 1:1:18), and dehydrated in a graded ethanol series. They were embedded in a resin, Technovit 7100 (Kurzer, Germany) after polymerization at 45°C and 3-5 μm thick sections were cut. To observe the regeneration process, calli were sectioned in the same way. Sections were stained with Toluidine Blue and observed with a light microscope.

Calli were fixed with 2.5% glutaraldehyde in 0.1 M sodium phosphate buffer. After rinsing with the same buffer, they were post-fixed in 1% osmium tetroxide for 3 hours, then rinsed again. Then they were dehydrated in a graded ethanol series, transferred into 3-methyl-buthyl-acetate, and critical-point-dried. They were sputter-coated with platinum, and observed with a scanning electron microscope (Hitachi S-4000, Tokyo) at an accelerating voltage of 15 kV.

Embryos and calli were fixed in 4% paraformaldehyde and 0.25% glutaraldehyde in 0.1 M sodium phosphate buffer. Then they were dehydrated in a graded ethanol series, replaced with xylene and embedded in Paraplast Plus (Oxford Labware, St Louis, MO). Microtome sections at 8 μm thick were applied onto slide glasses coated with Vectabond (Vector Lab. Burlingame, CA). Digoxygenin-labeled antisense probes were prepared from the coding regions without poly(A) region of RAmy1A and OSH1. RAmy1A gene was a kind gift from Dr J. Yamaguchi, Nagoya University. In situ hybridization and detection of the signals were carried out by the methods of Kouchi and Hata (1993).

Mature embryos of shl mutants were characterized by the lack of a shoot (shoot apical meristem and three foliar leaves) (Fig. 1).

In addition, two embryo-specific organs flanking the shoot apex, coleoptile and epiblast, were also lost in these mutants, indicating that the differentiation of coleoptile and epiblast was tightly linked with SAM. Radicles were apparently normal in these mutants except for shl3, in which the radicle was initiated exogenously, germinated precociously, and became necrotic in the mature dry seeds (Fig. 1I). The embryos of two alleles of SHL1, five alleles of SHL2 and shl4 could not be morphologically distinguished from one another. The radicles of the shootless mutants (except for shl3) germinated normally when dry mature seeds were inoculated on moistened filter paper. It was not apparent whether the apical part of shl embryos was differentiated into the scutellum or not. Anatomical observation revealed that in all the mutants except shl3, the epidermal cells in the apical dorsal region exhibited a palisade-shape, which was characteristic of scutellar epithelium (Fig. 2). This suggests that the scutellum develops normally in shl1, shl2 and shl4. To confirm this, we performed in situ hybridization on mature seeds imbibed for 3 days using RAmy1A, a major α-amylase gene of rice, as a probe. In the wild type, RAmy1A is strongly expressed in the scutellar epithelium and aleurone layer of germinating seeds (Sugimoto et al., 1998). In shl1, shl2 and shl4 embryos, hybridization signals were detected in the epidermal cells facing the endosperm (Fig. 3). Therefore, we conclude that the scutellum is differentiated normally and is functional in these shl embryos. Anatomical and in situ hybridization studies on mature embryos indicate that the differentiation of scutellum is regulated independently of SAM, but that of coleoptile and epiblast depends on the presence of SAM.

In the wild-type embryo, the globular stage lasts for 3 days after pollination (Fig. 4A). At this stage, dorsiventrality of embryo is evident from the gradient of cell size, larger cells in the dorsal region and smaller cells in the ventral region. At 4 days after pollination (4 DAP), shoot and radicle apices are first observed microscopically immediately after the protrusion of coleoptile (Fig. 4B). Most of the embryonic organs are apparent at 5 DAP (Fig. 4C). In embryos of the shl mutants, the protrusion of coleoptile did not occur at 4 DAP (Fig. 4D). SAM and epiblast were also not differentiated, while the radicle was initiated at 5 DAP (Fig. 4E) and normally developed at 7 DAP (Fig. 4F), as in the wild type. In embryos of shl1, shl2 and shl4, normal body axes were apparent at 5 DAP. No sign of SAM differentiation could be recognized in any of the shl mutants.

The failure of shoot formation during embryogenesis does not necessarily prevent adventitious shoot formation. If a mutation resulting in the loss of SAM is associated with an embryo-specific pattern preceding SAM initiation, and adventitious shoot initiation does not require that embryo-specific pattern, adventitious shoots may be produced from the shl calli. We therefore examined shoot regeneration from scutellum-derived calli of the shl mutants, except for shl3, whose embryos were not viable in mature seeds and the calli could not be induced. In contrast to the regeneration of numerous adventitious shoots from the wild-type calli (Fig. 5A), no adventitious shoots were produced from calli of any shl mutant (Fig. 5B). Therefore, it is concluded that SHL1, SHL2 and SHL4 are generally indispensable for shoot formation. Although shoots were never produced, small leaf-like organs were observed frequently in shl mutant calli soon after transplantation onto the regeneration-medium (Fig. 6A). These organs were also recognized in the wild-type calli. They had trichomes, stomatas and vascular bundles, indicating that they were adventitious leaves (Fig. 6B). These adventitious leaves soon turned brown and showed no further growth. A SAM-like structure at the base of these adventitious leaves could not be detected by histological and anatomical observation (Fig. 6C,D). The absence of SAM in the vicinity of adventitious leaves was confirmed by in situ hybridization using OSH1 that was expressed in the SAM as probe. At the early stage of adventitious shoot development from the wild-type calli, leaves were produced from the SAM at points where OSH1 RNA was detected (Fig. 7A). However, SAM and OSH1 expression was not recognized at the base of incipient adventitious leaves (Fig. 7B). Accordingly, leaves can be differentiated independently of SAM, and the above SHL genes are associated with the SAM differentiation but not with the initiation of leaves.

Because most shootless mutants were not morphologically distinguishable from one another, we could not specify the functional differences between the above SHL genes. We therefore examined the expression pattern of a rice homeobox gene, OSH1. In the wild-type embryo, OSH1 is first expressed in a distinct region of the globular embryo where the shoot is to be formed later, and at 5 DAP, it is transcribed in SAM, epiblast and the adjacent tissues (Sato et al., 1996). Because OSH1 expression is maintained in the SAM throughout the life cycle, it may be required for the determination and maintenance of SAM (Matsuoka et al., 1993). In situ hybridization was done on embryos at 4-6 DAP when shootless embryos were easily identified by the lack of coleoptile. In two shl1 alleles, shl1-1 and shl1-2, OSH1 was expressed in a quite restricted region, much narrower than that of the wild type (Fig. 8B,C). Similarly, hybridization signals of OSH1 were observed in a narrow region in each of the five shl2 alleles (Fig. 8D-H). In contrast, shl3 and shl4 embryos showed a normal expression pattern of OSH1 (Fig. 8I,J). These results indicate that shl1 and shl2 largely reduce a region that is specified by the OSH1, and their wild-type genes may be required for maintaining OSH1 expression. On the other hand, SHL3 and SHL4 will function downstream or independently of OSH1 because the expression pattern of OSH1 is not modified in shl3 and shl4.

Relatively few genes that show a SAM-less phenotype as a result of the loss of function mutations, have been cloned to date. In Arabidopsis, it has been suggested that STM is required for the specification of SAM cells in the embryo and for maintaining undifferentiated cells in the center of the SAM (Endrizzi et al., 1996). NAM of petunia seems to determine the positions of SAM and leaf primordia rather than to regulate directly the differentiation of SAM (Souer et al., 1996). In tomato, defective embryo and meristems has been cloned, but its function is still unknown (Keddie et al., 1998). Presumably many genes are involved in the initiation of SAM, so more mutations must be identified in order to elucidate the cascade of genes leading to SAM initiation.

The present study reveals that at least four loci are involved in the SAM initiation in rice. In shl1, shl2 and shl4, radicle and scutellum are produced in the basal region and in the apical dorsal region, respectively, and in all the mutants, OSH1 is expressed in the ventral region of the embryos. Accordingly, apical-basal and dorsal-ventral axes are not disturbed by the shl mutations. A shoot regeneration experiment shows that SHL genes are indispensable for the adventitious shoot formation. These results indicate that SHL genes are not involved in embryonic axialization, and seem to function in establishing the SAM-associated region. The results of an in situ hybridization experiment using OSH1 as a probe indicate that shl1 and shl2 affect the expression domain of OSH1, which is first expressed in a region of globular embryo where the SAM develops later (Sato et al., 1996). OSH1 is considered to be an ortholog of KN1 (Matsuoka et al., 1993) and to play an important role in establishing the shoot-specific region (Sato et al., 1996). KN1 is thought to be required for maintaining an indeterminate state of SAM cells (Hake et al., 1995; Sinha et al., 1993; Smith et al., 1995). In barley, the Hooded mutation of the KN1-type homeobox gene causes an ectopic floral meristem (Miller et al., 1995). These homeobox genes are thus involved in the initiation and maintenance of SAM. Furthermore, shl1 and shl2 are considered to have defects in the shoot meristem-specific regionalization. Both SHL1 and SHL2 may be involved in the establishment of cell identity required for OSH1 expression and for SAM initiation. On the other hand, shl3 and shl4 do not modify OSH1 expression. In rice, an orl1 mutant is known in which both SAM and radicle are deleted (Hong et al., 1995), but OSH1 is expressed normally (Sato et al., 1996). Taking all these results together, our knowledge of genetic cascades leading to the initiation of apical meristems in rice embryo can be summarized as follows. SHL1 and SHL2 function upstream of OSH1, and regulate the expression of OSH1, whereas SHL3 and SHL4 function downstream or independently of OSH1. In contrast to SHL genes, which are involved specifically in shoot differentiation, ORL1 is required for both shoot and radicle initiation downstream of OSH1.

To date, very few genes are known that regulate the expression of homeobox genes. Recently, in Arabidopsis embryogenesis, CUC genes have been shown to participate in the SAM initiation upstream of STM, a KN1-type homeobox gene (Aida et al., 1999). Maize rough sheath 2 represses the expression of knox homeobox genes in leaves (Schneeberger et al., 1998; Timmermans et al., 1999; Tsiantis et al., 1999). Although homeobox genes will play central roles, SAM initiation requires many other genes upstream and downstream of the homeobox genes. We have identified many other mutations that cause the loss of shoot and/or radicle in rice (Hong et al., 1995), which will greatly help the understanding of SAM initiation mechanism.

Because the shl mutations affect the differentiation of other embryonic organs as well as SAM, we can discuss the interrelationships among embryonic organs based on the characteristics of shl mutants. As is well known, grasses have complex embryos comprising shoot, radicle, and several embryo-specific organs such as scutellum, coleoptile and epiblast, the functions of which remain controversial, as reviewed by Brown (1960). The scutellum is considered to be homologous to either the entire cotyledon or a part of the cotyledon. The coleoptile has also been suggested to be equivalent to the cotyledon, a part of the cotyledon such as the sheath or ligule, an axial bud, a leaf, or two cotyledons grown together (Sargant and Arber, 1915; Roth, 1955). The epiblast may also be homologous to a variety of organs; a cotyledon, a part of the cotyledon such as an auricle or ligule, outgrowth of cotyledon or axis, or a leaf (Boyd, 1931; Sargant and Arber, 1915; Roth, 1955).

Among various interpretations proposed, the most popular one is that the counterpart of the cotyledon in the grass embryo is the scutellum and coleoptile. This premise is mainly based on the functional aspects of the cotyledon, which is a leaf-like storage organ for absorbing food from endosperm and protects the plumule during germination (Eames, 1961). However, this interpretation is not consistent with our observation that the scutellum and coleoptile are genetically independent, as indicated in shl mutants.

In addition, the developmental ambiguity of the SAM-cotyledon relationship makes this problem more difficult. The cotyledon may be developmentally defined as the first lateral organ of SAM, but the cotyledons initiate without SAM in the Arabidopsis stm mutant (Barton and Poethig, 1993), suggesting that the initiation of SAM and the differentiation of cotyledon are genetically separable. However, in stm mutants, SAM has recently been suggested to degrade at a very early stage of its establishment (Kaplan and Cooke, 1997; Kerstetter and Hake, 1997), and the cotyledon may be developmentally related to SAM. In this case, the cotyledon is the first lateral organ of SAM. We found that the differentiation of coleoptile and epiblast depends on SAM but the scutellum differentiates independently of SAM. If the cotyledon is the first lateral organ of SAM, it would be equivalent to the coleoptile and/or epiblast in grass embryos. However, if the cotyledon is independent of SAM, it would be homologous to the scutellum. In any case, the scutellum is developmentally discriminated from the coleoptile and epiblast, and it is impossible that both scutellum and coleoptile constitute the cotyledon.

Most of the leaf primordia initiate from SAM as a consequence of the change in cell division and growth patterns of a particular group of cells in the SAM. It is not confirmed, however, whether tissues other than SAM are able to produce leaves or not. Several researchers have proposed the concept of leaf initiation in the absence of an identifiable shoot apex in the fern Pteridium (Whittier, 1962), dicotyledonous Begonia hispida (Sattler and Maier, 1977) and so on. In a maize mutant Lax midrib1-O, a leaf initiates from a leaf (Schichnes et al., 1997). From the analysis of this mutant, it is concluded that the signal for leaf initiation need not come from the shoot apex. These examples show that leaves may initiate from non-SAM tissues. However, this concept remains controversial because the above examples may be only rare cases and the evidence on the lack of SAM is not persuasive.

In the present study, adventitious leaves initiated from calli of wild type and shl that did not differentiate a SAM. We conclude that the initiation of leaves does not require SHL genes and the SAM activity. Thus, a cascade of genes involved in leaf initiation can begin to work without signals from the SAM. However, the absence of apparent dorsiventrality and sufficient laminal growth in the adventitious leaves suggests that signals from the SAM are needed for the full development of leaves.

We would like to express our thanks to Dr Junji Yamaguchi (Nagoya University) for generously providing cDNA of RAmy1A. This work was supported in part by a Grant-in-Aid for Scientific Research on Priority Areas from the Ministry of Education, Science and Culture of Japan.