OsPNH1 regulates leaf development and maintenance of the shoot apical meristem in rice

Authors


*For correspondence (fax +81 52 789 5226; e-mail makoto@nuagr1.agr.nagoya-u-ac.jp)

Summary

The Arabidopsis PINHEAD/ZWILLE (PNH/ZLL) gene is thought to play an important role in the formation of the shoot apical meristem (SAM) and in leaf adaxial cell specification. To investigate the molecular mechanisms of rice development, we have isolated a rice homologue of PNH/ZLL, called OsPNH1. Around the SAM, OsPNH1 was strongly expressed in developing leaf primordia, specifically in the presumptive vascular domains, developing vascular tissues, a few cell-layers of the adaxial region, and future bundle sheath extension cells. In the SAM, only weak expression was observed in the central region, whereas strong expression was detected in the mid-vein region of leaf founder cells in the peripheral SAM domain. We produced transgenic rice plants containing the antisense OsPNH1 strand. The antisense OsPNH1 plants developed malformed leaves with an altered vascular arrangement and abnormal internal structure. These plants also formed an aberrant SAM with reduced KNOX gene expression. We examined the subcellular localization of the OsPNH1-GFP fusion protein and found that it was localized in the cytoplasm. On the basis of these observations, we propose that OsPNH1 functions not only in SAM maintenance as previously thought, but also in leaf formation through vascular development.

Introduction

Leaves are produced repeatedly from the shoot apical meristem (SAM) at regular intervals (plastochron) and in regular positions (phyllotaxy) around the peripheral region of the SAM. In rice leaves, the primordia arise in a distichous phyllotaxy, i.e. 180° apart. It is thought that leaf organogenesis is initiated by a partitioning of the meristem into a region that contains future leaf founder cells. The expression of KNOX (KNotted1-like homeobOX) genes, which function to maintain the indeterminate state of the SAM, is a useful molecular marker of the leaf founder cell acquisition in the SAM (Jackson et al., 1994; Kerstetter et al., 1997; Lincoln et al., 1994; Long et al., 1996; Sentoku et al., 1999; Smith et al., 1992). The KNOX genes are specifically expressed in the indeterminate region of the SAM and downregulated in the determinate region, i.e. the founder cells.

Analyses of several mutants with altered leaf morphology have demonstrated that the proper balance of indeterminate/determinate cells in the SAM is critical for normal leaf development. For example, a maize narrow sheath (ns) mutant, which fails to form the leaf marginal domain, cannot recruit the margin of the leaf founder cells in the SAM (Scanlon et al., 1996). Similar phenomena, characterized by a failure in the recruitment of leaf founder cells or in the exchange of cell fate from indeterminate to determinate in the SAM, have also been observed in a maize mutant, leaf bladeless1 (lbl1), that has malformed leaves (Timmermans et al., 1998). These leaf mutants demonstrate that correct recruitment of leaf founder cells and/or exchange of cell fate from indeterminate to determinate is important for normal leaf development.

In Arabidopsis, several factors involved in SAM maintenance and leaf development have already been identified and characterized. A meristem-defective mutant, pinhead/zwille (pnh/zll), provides an interesting example. This mutant has a flat meristem and, as a result of a failure in the maintenance of the indeterminate state of the SAM, it often forms a single central organ instead of the meristem (Lynn et al., 1999; McConnell and Barton, 1995; Moussian et al., 1998). The PNH/ZLL gene belongs to a novel gene family that is widely conserved from eukaryotes to prokaryotes (Lynn et al., 1999; Moussian et al., 1998). PNH/ZLL expression is localized in the central region of the embryo, probably corresponding to the provascular region in the early stage of embryogenesis (Lynn et al., 1999; Moussian et al., 1998). In the post-embryonic stage, the expression of PNH/ZLL continues in the SAM, provascular and adaxial domains of developing leaf primordia (Lynn et al., 1999). KNOX gene expression is decreased in the SAM of the pnh/zll mutant, suggesting that one possible function of PNH/ZLL may be to maintain the undifferentiated state of the stem cells in the SAM (Moussian et al., 1998). Besides the maintenance of the stem cells in the SAM, the specific expression of PNH/ZLL in the adaxial domain of leaf primordia indicates another function for PNH/ZLL in leaf development (Lynn et al., 1999).

In order to understand the molecular mechanisms of SAM maintenance and leaf development in rice, a model plant of cereal crops, we isolated a rice PNH/ZLL homologue, OsPNH1. Analyses of OsPNH1 expression and the phenotypes of antisense OsPNH1 plants revealed that OsPNH1 plays an important role in rice leaf development and SAM maintenance.

Results

Cloning of a rice PNH/ZLL homologue

To isolate a rice PNH/ZLL homologue, we searched for its homologous sequence in rice ESTs using the tBLASTn algorithm. An EST clone C26544 partially encoding a similar sequence to the C-terminus region of the Arabidopsis PNH/ZLL gene was found and used to screen a rice embryo cDNA library. We isolated a 3.5-kb full-length cDNA clone designated OsPNH1 (Oryza sativapinhead1), which contained an open reading frame encoding 978 amino acid residues. The putative protein encoded by this clone shows 72.5% identity to the amino acid sequence of Arabidopsis PNH/ZLL over its entire length (Figure 1a). The Arabidopsis PNH/ZLL gene family includes at least eight members: PNH/ZLL, and AGO1–7 (Fagard et al., 2000). In rice, there are at least 10 clones encoding similar amino acid sequences in the ESTs. From these, we also cloned a cDNA designated OsAGO1 (Oryza sativa argonaute1) which shows a high similarity to Arabidopsis AGO1 (75.2% identity in the amino acid sequence over its entire length) (Figure 1b-d). The phylogenetic analysis demonstrates that the isolated rice clones OsPNH1 and OsAGO1 correspond to the Arabidopsis genes PNH/ZLL and AGO1, respectively (Figure 1b). Many proteins similar to PNH/ZLL and AGO1 have been found in other organisms (Catalanotto et al., 2000; Cerutti et al., 2000; Cox et al., 1998; Kataoka et al., 2001; Koesters et al., 1999; Schmidt et al., 1999; Sharma et al., 2001; Tabara et al., 1999; Zou et al., 1998). Cox et al. (1998) showed that the C-terminal region of these proteins, known as the Piwi box, is well conserved. Cerutti et al. (2000) also found two conserved amino acid sequences in this superfamily, known as the Piwi domain in the carboxyl terminus and the PAZ domain in the central region (Figure 1c,d), although the biochemical roles of these domains have not yet been clarified.

Figure 1.

Comparison of the amino acid sequence of OsPNH1 and related proteins.

(a) Comparison of the deduced amino acid sequence of OsPNH1 protein (GenBank accession number ABO81950) with that of the PNH/ZLL protein (AJ223508), OsAGO1 protein (ABO81951) and AGO1protein (U91995). Identical amino acid residues are shown in black boxes with white lettering. The green and blue underlining indicates the position of the conserved amino acid sequences known as the PAZ domain and Piwi domain, respectively. The orange highlighted region is a highly conserved piwi box and surrounding region, which was used for the alignment shown in (d).

(b) Phylogenetic analysis of OsPNH1, OsAGO1, PNH/ZLL, AGO1 and related Arabidopsis putative proteins. Arabidopsis putative proteins are AGO2 (GenBank accession number AC007654, protein number AAF24585.1), AGO3(AC007654, AAF24586.1), AGO4(AC005623, AAC77862.1), AGO5(AC006929, AAD21514.1), AGO6(AC003033, AAB91987.1) and AGO7(AC073178, AAG60096.1). The structural relationship was calculated by ClustalW and illustrated by TREEVIEW (http://taxonomy.zoology.gla.ac.uk/rod/treeview.html).

(c) Alignment of the PAZ domain between OsPNH1 and some PAZ domain-containing proteins, such as rabbit eIF2C (GenBank accession number AF005355), Drosophila PIWI (2507358 A), Drosophila Sting (AE00363-38) and Arabidopsis CAF (AF187317).

(d) Alignment of the region surrounding the piwi box between OsPNH1 and some piwi box-containing proteins, such as Drosophila dAGO1 (GenBank accession number AB035447), C. elegans RDE-1 (2523267 A) and Neurospora QDE-2 (AF217760).

We mapped OsPNH1 on the rice genome using RFLP analysis. OsPNH1 was mapped on Chr. 6 between the RFLP markers G122 and G1314A (data not shown). We looked for a morphological rice mutant mapped around this point but did not find any mutants.

Expression of the OsPNH1 transcript

To determine the expression pattern of OsPNH1 in various organs, we conducted an RNA gel blot analysis using the 3′ non-coding sequence of the OsPNH1 as a probe. The 3.5-kb OsPNH1 mRNA was strongly expressed in the vegetative shoot apex, inflorescence shoot apices and stems, and weakly expressed in young roots (data not shown).

We used in situ hybridization for a more precise analysis of OsPNH1 expression. Since a high level of expression was observed in the vegetative shoot apex, we first analysed OsPNH1 expression around the vegetative SAM. In the rice SAM, the leaf primordia develop from the mid-vein to marginal regions, enclosing the apex with a distichous phyllotaxy (Figure 2a). Figure 2(a) shows the shoot apex of a 30-day-old rice plant, which appears as a protrusion of the first leaf primordium (P1) and a hood-like second primordium (P2). Strong expression of OsPNH1 was observed in the developing vascular tissue of the young leaf primordia (until P4) around the vegetative meristem (Figure 2b-j). Weak expression was also detected in the central region of the meristem itself (Figure 2k,l). In early P1 leaf primordia without any morphological vascular differentiation, strong OsPNH1 expression was seen at the centre of the leaf primordia (arrows in Figure 2h-j), which corresponds to the future mid-vein region. In late P1 to early P2 leaf primordia, which had morphologically started to develop mid-vein tissue, the expression of OsPNH1 was localized in the mid-vein (arrowheads in P2 in Figure 2b-j) as well as the future large veins that cannot be morphologically observed at this stage (Figure 2f-j). In more developed primordia such as P3 and P4, OsPNH1 expression was observed not only in the vascular regions, but also as a striped pattern in the future bundle sheath extension cells (Figures 2b-j,n). As the bundle sheath extensions are first observed around the mid-vein region at the P5 stage (Kaufman, 1959), the expression of OsPNH1 in the bundle sheath extensions occurs about two plastochrons prior to their morphological development. A high level of OsPNH1 expression in leaf primordia continued until the P4 stage and then quickly disappeared after the P5 stage (data not shown). Therefore, OsPNH1 expression preceded the morphological development of vascular bundles and bundle sheath extensions in leaf primordia.

Figure 2.

In situ mRNA localization of OsPNH1 around the vegetative SAM.

(a) SEM image of the shoot apex of 30-day-old rice. P3 and more developed leaf primordia were removed for observation of the young leaf primordia and the SAM.

(b)- (j) Serial transverse sections through a 30-day-old vegetative meristem; (b) shows the most apical section of the P2 leaf primordium whereas (j) is the most basal section of P1 leaf. Strong expression is observed in the future vascular and developing provascular regions, and in the future bundle sheath extension cells (be) in the leaf primordia (P1-P4). Arrowheads indicate the position of the mid-vein in P2 and P3 primordia. Arrows in (h)-(j) indicate the mid-vein portions of the P1 primordium.

(k) and (l) show OsPNH1 expression in the SAM. (k) Transverse section through the centre region of the SAM. (l) Medial longitudinal section of the SAM. Weak expression is detected in the central region of the SAM in both sections.

(m) Transverse section of a 30-day-old vegetative meristem. The expression is observed in a few cell layers at the most adaxial side of the late P1 leaf primordia.

(n) Transverse section through a late P3 primordium of a 30-day-old shoot. OsPNH1 expression is specifically localized in the adaxial side of the future bundle sheath extension cells (be) but not in the abaxial side. P1, first leaf primordium; P2, second leaf primordium; P3, third leaf primordium; P4, fourth leaf primordium. Bars in (b) (l) and (n) = 100 µm. Bars in (k) and (m) = 50 µm.

In Arabidopsis, PNH/ZLL is expressed not only in the vascular region, but also in the adaxial region of developing leaf primordia. Similarly, we detected specific expression of OsPNH1 in the adaxial region of rice leaf primordia, although the expression pattern showed some differences from that observed in Arabidopsis. Expression of OsPNH1 in the adaxial region was first observed in a few cell layers including epidermis at the base of the late P1 primordia (Figure 2m). In more developed primordia, such as late P2, P3 and P4, expression in the adaxial side was observed in the regions where the bundle sheath extensions would later form (Figure 2b-j,n). Although bundle sheath extensions run vertically in a striped pattern from the vascular tissues to both the adaxial and abaxial sides, OsPNH1 expression in future bundle sheath extensions was strictly limited to the adaxial side.

As described in the introduction, the group of cells giving rise to the leaf primordia can be distinguished prior to their morphological initiation by the down-regulation of KNOX genes in the SAM. The rice KNOX gene OSH1 is a good molecular maker because it is specifically expressed in the SAM and is downregulated on the flank of the meristem where the initiation of lateral organ formation occurs (Sentoku et al., 1999). As shown in Figure 3a,c, the OSH1 down-regulated region (P0) was observed as a ring-shape in the SAM. To investigate whether the expression of OsPNH1 had already been turned on in the P0, we performed precise analyses of OsPNH1 expression in the SAMs of various sections. As shown in Figure 3b,d, the earliest OsPNH1 expression occurred at an area of the P0 that corresponded to the mid-vein region of the next leaf primordium. Since we have never detected the expression of OsPNH1 prior to P0, OsPNH1 expression must be initiated at about the same time as OSH1 suppression in the SAM. Such pre-primordium expression has also been observed in the case of Arabidopsis PNH/ZLL, but its expression is initiated in the P-1 stage, which is one plastochron before the P0 stage (Lynn et al., 1999). The earlier expression of Arabidopsis PNH/ZLL in the pre-primordium region than expression of OsPNH1 may be due to the difference in the plastochron length between Arabidopsis and rice; it is 1–2 days in Arabidopsis and 3–4 days in rice.

Figure 3.

Expression of OsPNH1 and OSH1 in P0 leaf primordium in the SAM.

Transverse sections through the vegetative meristem probed by the antisense RNA for OSH1(a)or OsPNH1(b).The initiation site of the P0 primordium cannot be morphologically observed but can be recognized by the down-regulation of OSH1 in the SAM. Note that the P0 area appears as a ring shape in the SAM, as is schematically illustrated by the yellow colour in (c).OsPNH1 expression at the future mid-vein portion of the P0 primordia is shown in green colour in (d).P0, the mid-vein region of the next leaf primordium site. P1, first leaf primordium; P2, second leaf primordium. Bar = 100 µm.

OsPNH1 expression continues after the transition from the vegetative to reproductive meristem. A sign of the transition to the reproductive stage in rice is a change in the phyllotaxy from distichous to 2/5. The inflorescence meristem first produces several bracts of the rachis-branches with 2/5 phyllotaxy. At this bract differentiation stage, OsPNH1 expression was observed at the predicted site of the next bract (arrowhead in Figure 4a) and in the primary rachis-branch primordium region (arrow Figure 4a). This finding indicates that OsPNH1 expression marks the position of lateral organ primordia in the SAM throughout rice development. After the bract differentiation stage, OsPNH1 was expressed in the vascular tissue of the developing rachis-branch (Figure 4b,c). In developing floral organs, OsPNH1 was expressed in the adaxial side of the palea, lemma and stamen (arrows in Figure 4d).

Figure 4.

In situ mRNA localization of OsPNH1 in the inflorescence SAM.

(a) Transverse section through the inflorescence meristem at the bract primordia differentiation stage. The first bract primordium (br) is developing. The rice bract has no lateral vascular tissue but only has a mid-vein. The phyllotaxy of the bract and rachis branch is spiral. OsPNH1 expression corresponds to the site of development of the future primary rachis branch (arrow) and the next bract primordium (arrowhead).

(b) and (c) Longitudinal sections through the inflorescence SAM at the primary rachis-branch primordia differentiation stage (b) and the secondary rachis-branch primordia differentiation stage (c). The expression is localized in the provascular tissues. (d)Longitudinal section through the floral meristem. Specific OsPNH1 expression in the adaxial region is observed in some floral organ primordia (arrows). br, bract primordia; fl, flag leaf; pb, primary branch; sb, secondary branch; st, stamen primordia; lm, lemma primordia; gl, rudimentary glume; pl, palea primordia. Bars = 100 µm.

Phenotypic analyses of OsPNH1 antisense plants

To analyse the biological function of OsPNH1, the antisense OsPNH1 transcript was expressed under the control of the actin1 (Act1) gene promoter. The Act1 promoter is a much stronger promoter with constitutive expression than the 35S promoter in transgenic rice plants (Zhang et al., 1991). More than 200 independent transformants were regenerated in this experiment. Almost all of the transgenic rice plants showed aberrant leaf morphologies and a dwarf phenotype (Table 1 and Figure 5b-e). On the basis of the phenotypic severity, these transformants were divided into four categories: ‘mild’, ‘intermediate’, ‘severe’, and ‘pinhead’ (Figure 5b-e). Transformants with the ‘mild’ phenotype showed a slightly aberrant leaf shape and weak dwarfism (Figure 5b). The ‘intermediate’ transformants showed dwarfism and aberrant erect leaves (Figure 5c). The ‘severe’ plants showed overall shoot abnormality with severe dwarfism (Figure 5d). These plants have often produced various malformed leaves, including narrow, fused, tortuous and filamentous leaves (Figure 5d). The ‘pinhead’ plants did not form real leaves but only formed tendril-like green protrusions and soon became necrotic on the culture medium (Figure 5e).

Table 1.  Phenotypic categories of primary transformants and the height/width of SAM
PhenotypeNumber of
transformants
Height/
Width of SAM
  • *

    Six-day-old wild type seedlings were examined as a control.

  • **

    The sample size is shown in parentheses.

Mild360.820 (10)**
Intermediate1270.647 (13)
Severe540.278 (7)
Pinhead15
Wild type*1.313 (10)
Figure 5.

Phenotype of transgenic plants transformed with Act1::antisense OsPNH1 cDNA constructs.

(a)-(e) Morphological variation in antisense OsPNH1 plants. (a) Morphology of the wild type plant.

(b)-(e) Typical morphology of antisense OsPNH1 transformants exhibiting the mild (b), intermediate (c), severe (d)and pinhead (e) phenotype.

(f)-(g) SAM structure of the antisense OsPNH1 plants. (f) Longitudinal section of 6-day-old wild type vegetative meristem.

(g)-(i) Longitudinal sections of the SAM of antisense OsPNH1 transformants exhibiting the mild (g), intermediate (h), and severe (i) phenotype.

(j)- (m) Morphological alteration around the SAM in antisense OsPNH1 plants. (j) Transverse section of 6-day-old wild type vegetative meristem.

(k)-(m) Transverse section around the SAM of antisense OsPNH1 plants exhibiting the mild (k), intermediate (l), and severe (m) phenotype. Filled triangles indicate the mid-vein region of the leaf primordia.

(n) Close-up view of a cross section through the wild-type mid-vein region of P5 leaf sheath. The xylem (xy)/phloem (ph) structure is clearly shown.

(o) Close-up view of a cross-section through the P5 mid-vein region of an antisense plant with the severe phenotype. The xylem/phloem structure is not clearly developed.

(p) Transverse section of the filamentous structure shown in pinhead plant (e). The vascular structure is not clearly developed.

(q) Cross-section of the area around the mid-vein region of the wild type P6 leaf sheath. The mid-vein vascular bundle is located at the abaxial side, and the airspace (as) developed at the adaxial side is divided between two bundle sheath extensions (be).

(r) Cross-section of the area around the P6 mid-vein region of an antisense plant with the severe phenotype. The mid-vein vascular bundle is not located at the abaxial side but remains in the central region. The development of the airspace is interrupted (arrow). P4, fourth leaf primordium; Bars in (a) and (b) = 5 cm (c) and (d) = 2 cm (e) = 1 cm (f) to (i) (n) and (r) = 100 µm (j) to (m) = 250 µm.

For precise analyses of these phenotypes, we studied the morphology of the SAMs in the transformants. Longitudinal sections showed that all of these transformants, except the ‘pinhead’ plants, had flat SAMs compared to that of the wild type (Figure 5g-i). The height-to-width ratio of these plants decreased gradually according to the phenotypic severity (Table 1). In the ‘pinhead’ plants, we did not find a SAM-like organ at the bottom of the tendril-like protrusions (data not shown). It has been noted previously that the shape of the SAM affects the leaf initiation pattern or phyllotaxy (Itoh et al., 2000; Jackson and Hake, 1999; Laufs et al., 1998; Leyser and Furner, 1992). Consistent with these reports, the antisense plants also showed aberrant phyllotaxy (arrowhead indicates the centre of each leaf in Figure 5j-m), which correlated to the overall phenotypic severity (compare Figure 5k,m).

Vascular development in the leaf was also affected. The number of vascular bundles was reduced in the antisense plants. For example, there were six large vascular bundles in the wild P4 leaf primordia, whereas only three large vascular bundles were seen in the antisense plants (compare Figure 5j,m). Moreover, the arrangement of vascular bundles was disturbed. In the wild type, the development of vascular bundles occurs in a bilateral symmetrical pattern around the axis of the mid-vein (three large and small vascular bundles are evident in both sides of the P4 primordium in Figure 5j). Such a well-ordered developmental pattern frequently failed in the antisense plants (for example, two large and one small vascular bundle can be seen in the upper side of the P4 primordium and one large vascular bundle occurs in the lower side in Figure 5m). The xylem/phloem structure did not develop normally (compare Figure 5n,o), and the ‘pinhead’ plants did not form any vascular structures in the tendril-like protrusions (Figure 5p).

In the wild type leaf primordia, the vascular bundles shift toward the abaxial side during primordia development, and airspaces develop in the adaxial side, resulting in the dorso-ventrality of the internal structure of the leaf primordium (Figure 5q). In the antisense plants, this movement of the vascular bundles did not occur clearly. The bundles remained in the central region of the mesophyll tissue and the formation of airspaces in the adaxial side was also affected (Figure 5r). These defects in OsPNH1 antisense plants suggest that OsPNH1 may play a role in the leaf developmental processes in rice (see Discussion).

To examine the correlation between the phenotypic severity and the sense OsPNH1 transcript suppression level, we conducted in situ hybridization analyses. The expression of the sense OsPNH1 was severely reduced in the ‘mild’ phenotype plants (Figure 6b) and hardly detected in the ‘intermediate’ and ‘severe’ plants (Figure 6c), but a high level of expression was seen in the wild type (Figure 6a). In these antisense plants, OsAGO1 was normally expressed (data not shown). These results demonstrate that the introduced antisense OsPNH1 strand was functional, suppressing the intrinsic expression, and that the phenotypic severity of the antisense plants may depend on the level of the remaining sense transcript.

Figure 6.

Expression of the OsPNH1 and OSH1 transcripts in the antisense OsPNH1 transformants.

(a)-(d) Transverse sections of the SAM hybridized with an OsPNH1 antimRNA probe. Sections are from the wild type (a) and the antisense plants with the mild (b), intermediate (c)and severe (d) phenotype. The mild plant shows low level expression of OsPNH1, but little or no expression was observed in the intermediate and severe plants.

(e)-(h) Longitudinal SAM sections hybridized with an OSH1 antimRNA probe. Sections are from wild type (e) and the antisense plants with the mild (f), intermediate (g) and severe (h) phenotype. The shape of the SAM in the antisense OsPNH1 plants was flattened with a narrower OSH1 expression domain. Bars = 100 µm.

As described above, the expression of OSH1 can be observed throughout the SAM except in the determinate region recruited as the leaf founder cells (Figure 6e, Sentoku et al., 1999). We performed in situ hybridization analyses using the antisense OSH1 probe to examine the state of determination of the malformed SAMs in the antisense transformants. In contrast to the wild type, the OSH1 expression area was reduced even in the slightly malformed SAM of the ‘mild’ plants (Figure 6f). The extent of the decrease in the OSH1 expression area in the SAM corresponded to the severity of the malformed SAM. These observations indicate that OsPNH1 functions to maintain the indeterminate state of the SAM similar to PNH/ZLL in Arabidopsis (see Discussion, Moussian et al., 1998).

Cellular localization of OsPNH1

To examine the subcellular localization of the OsPNH1 protein, we fused the entire OsPNH1 cDNA to the GFP coding region. The plasmid was expressed in onion epidermal cells and the localization was examined using confocal laser scanning microscopy. The OsPNH1-GFP protein was spread throughout the cytoplasm of the onion epidermal cells (Figure 7a), whereas SLR-GFP, as a marker for nuclear-localized protein, was specifically localized to the nucleus (Figure 7b).

Figure 7.

OsPNH-GFP is localized in the cytoplasm. (a)Fluorescence image of onion epidermal cell expressing OsPNH1-GFP. (b)Fluorescence image of onion epidermal cell expressing SLR-GFP. Bar = 100 µm.

Discussion

In this study, we investigated the function of the rice PINHEAD/ZWILLE (PNH/ZLL) homologous gene, OsPNH1. Although Arabidopsis PNH/ZLL is considered to mainly function in SAM maintenance, our analyses demonstrated that OsPNH1 is involved not only in SAM maintenance, but also in leaf development.

OsPNH1 involved in leaf vascular development in rice

OsPNH1 is strongly expressed around the vegetative shoot apex in the pro-vascular region of the leaf primordia (Figure 2); its expression always precedes morphological vein development by about one plastochron. Furthermore, suppression of OsPNH1 expression by the antisense transcript causes incomplete leaf vascular development (Figure 5n-p) and disordered spatial arrangement of vascular tissues in the leaf primordia (Figure 5j-m). This localized expression in presumptive vascular regions, together with the defects in vascular development in the OsPNH1 antisense plants, strongly suggest that OsPNH1 plays an important role in leaf vascular development. It is noteworthy that the defects in vascular development are preferentially observed in leaves but not in the roots or stem (data not shown). This indicates that the OsPNH1 product functions only in vascular development in the leaves. Consistent with this, the expression of OsPNH1 is very low in the developmental vascular tissues of the roots and stem (data not shown).

OsPNH1 may be involved in determination of the central domain of the leaf founder region

The first step in leaf development is thought to be the establishment of a group of cells known as the leaf founder cells, which are destined to become leaves (reviewed in Sylvester et al., 1996). In maize, the leaf founder cells can be divided into three domains, the central (mid-vein), lateral and marginal domains, which are regulated by a specific developmental signal(s) for each domain (Scanlon et al., 1996). As shown in Figure 3b, OsPNH1 expression is already initiated in the central domain of the leaf founder region in the SAM. This finding indicates that OsPNH1 may be a factor that regulates the specific developmental signalling of the central domain in leaf founder cells. We consider that the determination of the central domain in leaf founder cells is the initial important event for the development of organized phyllotaxy, because the initial protrusion of the leaf primordium occurs at the central domain in the leaf founder cells (Figure 2a). Actually, defects in the localized expression of OsPNH1 in the antisense plants induce random phyllotaxy, probably due to a failure in the determination of the central domain of leaf founder cells (Figure 5j-m).

Role of OsPNH1 in maintenance of the indeterminate state of the SAM

Arabidopsis PNH/ZLL is considered to be a factor that maintains the stem cells of the SAM (Lynn et al., 1999; McConnell and Barton, 1995; Moussian et al., 1998). In antisense OsPNH1 transformants, defects occur in the SAM. In particular, the SAM size is gradually reduced in response to the suppression of intrinsic OsPNH1 levels (Table 1 and Figure 6a-c). The lower dome structure of the SAM can be observed in plants with the mild and intermediate phenotypes (Figure 5g,h), but plants with the severe phenotype have a flat meristem the same as in Arabidopsis pnh/zll (Figure 5i and McConnell and Barton, 1995; Moussian et al., 1998). The SAM of weak OsPNH1 antisense plants contains a reduction in the number of cells in the indeterminate state, as shown by a narrower OSH1 expression region (Figure 6e,f). These findings indicate that OsPNH1 plays a role in the maintenance of the indeterminate state of the stem cells in the SAM. The weak expression of OsPNH1 in the meristem centre supports the theory that OsPNH1 functions to maintain the indeterminate nature of the stem cells (Figures 2k,l).

Does OsPNH1 function in the adaxial leaf domain?

It has been suggested that Arabidopsis PNH/ZLL may play a role in promoting adaxial cell fate, based on its expression pattern and the mutant phenotype. PNH/ZLL expression has been observed in the adaxial leaf domain, and the failure of meristem formation in the pnh/zll mutant is thought to be the result of defects in the competency of SAM formation, which is the adaxial character of leaves (Lynn et al., 1999). We detected OsPNH1 expression in the adaxial region of the leaf, but the expression pattern was different from that observed previously in Arabidopsis. Adaxial expression of OsPNH1 in rice leaf primordia is restricted to a few cell layers in the most adaxial side of P1 to P4 leaf primordia (P3 and P4 in Figure 2b-j,m), and is also seen in the adaxial region of future bundle sheath extensions in P2 to P4 leaf primordia (P3 and P4 in Figures 2b-j.n). In contrast, Arabidopsis PNH/ZLL expression is widely spread around the entire adaxial region of the leaf primordia.

The internal adaxial area of the rice basal region (leaf sheath) primarily consists of airspace and bundle sheath extensions (Figure 5q). The airspace is the tissue containing large, gas-filled intercellular spaces (lacunae) localized between two bundle sheath extensions. In OsPNH1 antisense plants, organized airspace formation in the adaxial leaf domain is disrupted and results in the failure of the development of bundle sheath extensions (Figure 5k-l,q). Consequently, the overall internal leaf structure did not show the characteristic dorsoventrality that is found in normal leaves. In contrast to the Arabidopsis pinhead mutants, the competency of axial meristem formation is maintained in the antisense plants. When the surface cells of these transgenic leaves are examined using a scanning electron microscope (SEM), it is apparent that the epidermal cell structure is defective in both the adaxial and abaxial regions, but it does not show a homeotic conversion from adaxial to abaxial characteristics (data not shown). These results indicate that OsPNH1 would contribute to the construction of the internal structure of the adaxial portion of the leaf, but not to the determination of adaxial cell fate or promotion of the adaxial competency.

Molecular function of OsPNH1

Although the molecular function of OsPNH1 could not be clarified in this study, the analyses of some homologous proteins from other organisms provide a clue as to the possible role of OsPNH1. A recent report revealed that the family to which OsPNH1 belongs has two conserved regions, the Piwi and PAZ domains (Cerutti et al., 2000). The Piwi domain is large, with more than 300 amino acids localized at the C-terminal region, and the PAZ domain consists of about 200 amino acids and is localized in the central region of the these proteins. The Piwi domain has been found in many organisms including prokaryotes, whereas the PAZ domain appears to be restricted to eukaryotes (Cerutti et al., 2000). There is evidence to suggest that some genes belonging to this family are involved in stem cell maintenance or gene silencing. For example, Drosophila PIWI, human HIWI and Arabidposis PINHEAD/ZWILL (PNH/ZLL) have been shown to be required for stem cell maintenance (Cox et al., 2000; Moussian et al., 1998; Sharma et al., 2001), while Drosophila STING, C. elegans RDE-1, N. crassa QDE-2 appear to be required for post-transcriptional gene silencing (PTGS) mechanisms (Catalanotto et al., 2000; Schmidt et al., 1999; Tabara et al., 1999). Interestingly, Arabidopsis ARGONAUTE1 (AGO1) has been shown to be required for both stem cell maintenance and post-transcriptional gene silencing (PTGS) (Bohmert et al., 1998; Fagard et al., 2000; Lynn et al., 1999). These results indicate that this gene family may either be implicated in stem cell maintenance mechanisms through the control of PTGS, or may have a dual function in both stem cell maintenance and PTGS. The finding of this study indicate that OsPNH1 is at least involved in stem cell maintenance

OsPNH1-GFP expression is localized in the cytoplasm (Figure 7a), a characteristic it shares with the human AGO1 homologue, dAGO1, and rabbit eIF2C (Kataoka et al., 2001; Zou et al., 1998), but not with the Drosophila PIWI protein, which is distributed in nucleus (Cox et al., 2000). Both of the mammalian proteins localized in the cytoplasm are considered to interact with RNA molecules (Kataoka et al., 2001; Roy et al., 1988; Zou et al., 1998). In Arabidopsis, a protein known as CAF has been recently identified (Jacobsen et al., 1999). The CAF protein contains a PAZ domain and, based on its characteristic primary structure and loss-of-function phenotypes, is thought to be involved in RNA processing around the shoot apical region. All of these observations lead us to speculate that OsPNH1 may be involved in the metabolism of certain RNA molecules. The simplest scenario, for example, is that OsPNH1 is involved in promoting the stability of OSH1 mRNA in the SAM to maintain the stem cells. Future analyses of genes associated with RNA metabolism will reveal the nature of the involvement of RNA processing and stabilization in the developmental processes of some lateral organs from the SAM.

Experimental procedures

Plant material and growth conditions

Rice plants (Oryza sativa L., cv. Taichung 65) were used for the analyses. The plants were grown in the field or in a greenhouse at 30°C (day) and 24°C (night).

Isolation of OsPNH1 from a rice cDNA library

A clone, rice expressed sequenced tag (EST) C26544, encoding a homologous sequence to the Arabidopsis PINHEAD/ZWILLE gene was obtained from the DNA bank at the Ministry of Agriculture, Forestry and Fisheries of Japan. The C26544 sequence without a poly (A) tail was used as a probe to screen a rice embryo cDNA library. Hybridization was performed in 50% Formamide, 6× SSC, 5× Denhardt's solution, 0.5% SDS, and 0.1 mg ml−1 salmon sperm DNA at 42°C for 14 h and filters were washed in 2× SSC, 0.1% SDS at room temperature and then further washed in 0.2× SSC, 0.1% SDS at 65°C. Nucleotide sequences were determined by the dideoxynucleotide chain-termination method using an automated sequencing system (ABI373A). The cDNA clone was completely sequenced on both strands.

Mapping of OsPNH1 in the rice recombinant inbred lines

To map OsPNH1 on the rice chromosome, we used 71 recombinant inbred lines from a cross between two rice cultivars, Asominori (a Japonica rice) and IR24 (an Indica rice) (kindly provided by Dr Yoshimura of Kyushu University, Japan). Restriction fragment length polymorphism (RFLP) analysis was performed using a probe specific for OsPNH1. The linkage analysis was calculated using the MAPMAKER program (Whitehead Institute for Biomedical Research/Massachusetts Institute of Technology Centre for Genome Research, Wilmington, NC, USA).

In situ hybridization

In situ hybridization with digoxigenin-labelled RNA, produced from the OsPNH1 coding region without a poly (A) tail, was conducted as described previously (Kouchi and Hata, 1993). Tissues were fixed in 4% (w/v) paraformaldehyde and 0.25% glutaraldehyde in 0.1 m sodium phosphate buffer, overnight at 4°C, dehydrated through a graded ethanol series followed by a t-butanol series (Sass, 1958), and finally embedded in Paraplast Plus (Oxford Labware, St. Louis, MO, USA). Microtome sections (7–10 µm thick) were applied to glass slides treated with Vectabond (Vector Laboratories, CA, USA). In situ hybridization with digoxigenin-labelled sense or antisense RNA was conducted according to the method of Kouchi and Hata (1993).

DNA construction for transgenic rice plants and rice transformation

For antisense expression of the OsPNH1 cDNA in transgenic plants of rice, a promoter-terminator cassette (pBIAct1nos) containing the Act1 promoter (Zhang et al., 1991) and the NOS terminator was prepared by substituting the Act1 promoter for the 35S promoter in the hygromycin-resistant binary vector, pBI35Snos (Sato et al., 1999), between the HindIII and XbaI sites. The 3.5kb OsPNH1 cDNA was cloned in the antisense orientation to the XbaI-SmaI site of pBIAct1nos. This construct was introduced into the rice cultivar Taichung 65 by Agrobacterium tumefaciens-mediated transformation, as described by Hiei et al. (1994).

Histological analysis

For plastic sections, tissues were fixed in FAA (formalin: glacial acetic acid: 70% ethanol = 1 : 1 : 18). After rinsing with 0.1 m sodium phosphate buffer (pH 7.2), the tissues were dehydrated in a graded ethanol series, embedded in Technovit 7100 resin (Kulzer & Co. GmbH, Wehrheim, FRG), and sectioned at 3–5 µm. For in situ hybridization, tissues were fixed overnight at 4°C in 4% (w/v) paraformaldehyde and 0.25% glutaraldehyde in 0.1 m sodium phosphate buffer, and then dehydrated in a graded t-butanol series. They were embedded in Paraplast Plus (Oxford Labware, St. Louis, MO, USA), and sectioned at 8 µm. Microtome sections were stained with Delafield's hematoxylin.

Transient expression assay

The entire OsPNH1 cDNA was cloned as an in-frame C-terminal fusion to the green fluorescent protein (GFP) in the CaMV35S-GFP-NOS cassette vector (Chiu et al., 1996). The SLENDER RICE (SLR)-GFP fusion construct (Itoh et al., 2001) was used as the nuclear localized control. The GFP fusion protein was expressed in onion bulb epidermal cells using a particle-mediated DNA delivery system (PDS-1000/He; Bio-Rad, CA, USA). After bombardment, the tissues were incubated in darkness. Prior to observation, the tissues were soaked in 2 µg µl−1 DAPI (Dojindo, Kumamoto, Japan) solution for visualization of the nucleus. The stained samples were observed through a confocal microscanning laser microscope (MRC-1024; Bio-Rad, CA, USA). The images were obtained using a combination of a 488-nm laser excitation and 515 nm emission filter.

Scanning electron microscopy (SEM)

Tissues were fixed in FAA (formalin: glacial acetic acid: 70% ethanol = 1 : 1 : 18). After rinsing with 0.1 m sodium phosphate buffer (pH 7.2), the tissues were dehydrated in a graded ethanol series, and then 100% ethanol was replaced with 3-methyl-butyl-acetate. Samples were critical-point-dried, sputter-coated with platinum and observed with a scanning electron microscope (model S-4000, Hitachi, Tokyo, Japan) at an accelerating voltage of 10 kV.

Acknowledgements

We are grateful to Dr Kenzo Nakamura for providing the vector for rice transformation and Dr Yasuo Niwa (Shizuoka Prefectual University) for providing the 35S-GFP-NOS cassette vector. This research was supported by a Grant-in-Aid for Scientific Research on Priority Areas (Molecular Mechanisms Controlling Multicellular Organization of Plants) from the Ministry of Education, Science and Culture (Japan) and Special Coordinating Funds for Promoting Science and Technology from the Science and Technology Agency (Japan) to M. M., and from Research Fellowships of the Japan Society for the Promotion of Science for Young Scientists to A. N.

Ancillary