To identify where gibberellin (GA) biosynthesis and signaling occur, we analyzed the expression of four genes involved in GA biosynthesis, GA 20-oxidase1 and GA 20-oxidase2 (OsGA20ox1 and OsGA20ox2), and GA 3-oxidase1 and GA 3-oxidase2 (OsGA3ox1 and OsGA3ox2), and two genes involved in GA signaling, namely, the gene encoding the α-subunit of the heterotrimeric GTP-binding protein (Gα), and SLENDER RICE1 (SLR1), which encodes a repressor of GA signaling. At the vegetative stage, the expression of OsGA20ox2, OsGA3ox2, Gα, and SLR1 was observed in rapidly elongating or dividing organs and tissues, whereas the expression of OsGA20ox1 or OsGA3ox1 could not be detected. At the inflorescence or floral stage, the expression of OsGA20ox2, OsGA3ox2, Gα, and SLR1 was also observed in the shoot meristems and stamen primordia. The overlapping expression of genes for GA biosynthesis and signaling indicates that in these tissues and organs, active GA biosynthesis occurs at the same site as does GA signaling. In contrast, no GA-biosynthesis genes were expressed in the aleurone cells of the endosperm; however, the two GA-signaling genes were actively expressed, indicating that the aleurone does not produce bioactive GAs, but can perceive GAs. The expression of OsGA20ox1 and OsGA3ox1 was observed only in the epithelium of the embryo and the tapetum of the anther. Based on the specific expression pattern of OsGA20ox1 and OsGA3ox1 in these tissues, we discuss the unique nature of the epithelium and the tapetum in terms of GA biosynthesis. The epithelium and the tapetum are considered to be an important source of bioactive GAs for aleurone and other organs of the flower, respectively.
Gibberellins (GAs) are a large family of tetracyclic, diterpenoid compounds, some of which function as endogenous plant growth regulators. Through phenotypic analyses of mutants with reduced GA production, it has been revealed that bioactive GAs play an essential role in many aspects of plant growth and development, such as stem elongation, flower and fruit development, and seed germination (reviewed in Ross et al., 1997).
Despite considerable effort, it has not yet been determined precisely where the bioactive GAs are synthesized in plants, or which cells/tissues are targetted by the bioactive GAs to initiate GA-mediated biologic actions. Quantitative analyses using combined gas chromatography–mass spectrometry (GC–MS) and bioassays with dwarf plants have revealed that GAs are mainly present in actively growing and elongating tissues, such as shoot apices, young leaves, and flowers (Jones and Phillips, 1966; Kobayashi et al., 1988; Potts et al., 1982). This suggests that GAs are primarily synthesized at the site of their action. In contrast, there is some evidence for the presence of GAs in xylem and phloem exudates, indicating a long-distance transport of GAs through these tissues (Hoad, 1995; Lang, 1970). Grafting experiments support the idea of active transport of intermediates of active GAs and also bioactive GA itself (Katsumi et al., 1983; Proebsting et al., 1992; Reid et al., 1983). The contradictory results obtained from these conventional experiments cannot pin-point where bioactive GAs are synthesized and perceived. Indeed, the precise amount of each GA that is present in a specific tissue is difficult to determine, even using highly sensitive GC–MS techniques, because most GAs occur in very small amounts and are not bioactive. Thus, we have to take a new approach to clarify the location of synthesis and signaling of bioactive GAs.
Several genes for GA biosynthesis have been cloned from various plant species, including Arabidopsis, tomato, pea, potato, and rice (reviewed in Hedden and Phillips, 2000). Using promoter:β-glucuronidase (GUS) reporter expression or in situ RNA hybridization, the expression pattern of these GA-biosynthetic genes during plant development has been studied to provide more precise information on where bioactive GAs are produced. These approaches have revealed that the expression of the genes for GA biosynthesis is strictly regulated during plant growth and development (Cowling et al., 1998; Itoh et al., 1999; Kang et al., 1999; Phillips et al., 1995; Rebers et al., 1999; Silverstone et al., 1997). These studies have provided some information on the expression pattern of the GA-biosynthesis genes. However, on the basis of these studies, we cannot simply conclude where active GAs are produced or which cells respond to the GA signals, because intermediates of the GA biosynthesis pathway or the bioactive GAs may be transported. Actually, Yamaguchi et al. (2001) found distinct cell-specific expression patterns of early and late GA-biosynthesis genes during Arabidopsis seed germination. Based on this observation, Yamaguchi et al. (2001) proposed that GA biosynthesis takes place in two separate locations during Arabidopsis seed germination, and intercellular transport of an intermediate of the GA biosynthesis pathway is required to produce bioactive GAs. This study reinforces the need to examine the expression of the enzyme that catalyzes the final step of GA biosynthesis, GA 3-oxidase, in order to determine the exact site of bioactive GA biosynthesis.
There is another important point to consider when predicting the site of GA biosynthesis on the basis of the expression pattern of GA-biosynthesis genes, i.e. the copy number of these genes. For example, there are four genes encoding GA 3-oxidase or a putative GA 3-oxidase in the Arabidopsis genome that have a different expression pattern in the Arabidopsis plants, and at least two of these catalyze 3β-hydroxylation, which is the final step in the GA-biosynthesis pathway (Yamaguchi et al., 1998). This indicates that we have to examine the expression pattern of all the GA 3-oxidase genes in order to identify all the sites of bioactive GA biosynthesis.
Paying careful attention to these points, we have analyzed where bioactive GAs are synthesized and which cells or tissues respond to GA signals in rice plants. In this study, we examined the expression pattern of six kinds of genes involved in GA biosynthesis or GA signaling. Our results indicate that the sites where bioactive GAs are actively synthesized almost overlap with the sites that are actively involved in GA signaling, with the exception of the aleurone cells of the endosperm, in which the cells actively respond to the GA signal but do not produce GAs.
Construction of chimeric genes
In this study, we used four kinds of genes involved in GA biosynthesis and two involved in GA signaling. Figure 1(a) summarizes the functional relationship between these genes.
In previous studies, we identified two copies of the gene encoding GA 3-oxidase (GA 3-oxidase1 (OsGA3ox1) and GA 3-oxidase2 (OsGA3ox2)) in rice (Itoh et al., 2001). We accessed the rice genome sequence database to determine whether other genes encoding GA 3-oxidase are present in the rice genome. We searched the database using several regions of GA 3-oxidases from various plants, including rice, Arabidopsis, tobacco, and pea, but could not identify any new genes. We also performed a genomic DNA gel blot analysis using the coding regions of OsGA3ox1 and OsGA3ox2 as probes under low-stringency conditions, but we detected bands corresponding to the genomic sequences of only these two genes (data not shown). Thus, we concluded that the rice genome has only two GA 3-oxidase genes. To construct OsGA3ox1:GUS, we used a sequence containing the 5′-flanking region – first and second exons, first and second introns, and a partial region of the third exon (Figure 1b) – because the sequence of the first intron often affects the expression of OsGA3ox1 (Itoh et al., 1999; Silverstone et al., 1997). Similarly, we used a sequence containing the 5′-flanking region – first exon, first intron, and a partial region of the second exon – as a promoter of OsGA3ox2 (D18; Figure 1b).
We also examined the expression pattern of another GA-biosynthesis enzyme, GA 20-oxidase, which catalyzes successive oxidations at the C-20 site of GAs. The steps catalyzed by GA 20-oxidase produce substrates for GA 3-oxidase (Figure 1a). The expression of GA 20-oxidase is regulated by various factors, including the level of active GAs and physical conditions (Martin et al., 1996; Phillips et al., 1995; Toyomasu et al., 1997; Wu et al., 1996), and therefore, it is considered that the steps catalyzed by this enzyme are critical for regulating the level of active GAs. We also examined whether GA 20-oxidase is expressed at the same site as GA 3-oxidase. At the onset of this study, two kinds of GA 20-oxidase genes, GA 20-oxidase1 (OsGA20ox1) and GA 20-oxidase2 (OsGA20ox2; SD1), had been identified in rice (Ashikari et al., 2002; Sasaki et al., 2002; Toyomasu et al., 1997). To construct OsGA20ox1:GUS, we used a sequence containing the 5′-flanking region and a partial region of the first exon (Figure 1b), because the OsGA20ox1 gene has no introns. OsGA20ox2:GUS was constructed using a sequence containing the 5′-flanking region – first exon, first intron, and a partial region of the second exon as a promoter (Figure 1b).
Two genes were used as molecular markers of GA signaling: one encodes the α-subunit of the heterotrimeric G protein (Gα; D1), and the other encodes the SLENDER RICE1 protein (SLR1). The knock-out mutant of Gα (d1) has a dwarf phenotype and reduced sensitivity to exogenously applied GA (Ueguchi-Tanaka et al., 2000), suggesting that Gα functions as a positive regulator of GA signaling. Therefore, we considered that the expression of this protein should reflect GA signaling. To construct Gα:GUS, we used a sequence containing the 5′-flanking region – first and second exons, first and second introns, and a partial region of the third exon (Figure 1b). The gene encoding the SLR1 protein was used as a marker to detect cells actively responding to GA (see Discussion). SLR1:GUS was constructed using a sequence containing the 5′-flanking region and a partial region of the first exon (Figure 1b), because the SLR1 gene has no introns.
At least several T1 plants of transgenic rice were used to analyze GUS expression for each construct. We always confirmed that the GUS expression pattern observed in the T1 plants was inherited by the next generation (data not shown).
Overview of gene expression by RNA gel blot analysis or RT-PCR
Before a detailed study of the expression pattern of each gene was undertaken, we obtained an overview of the expression pattern of these genes using RNA gel blot or RT-PCR analysis (Figure 2). Because the level of expression of the GA-biosynthesis genes was low and the DNA sequences in their coding regions were quite similar, we could not detect the positive bands by RNA gel blot analysis, using specific probe for each gene. Therefore, we performed a semiquantitative RT–PCR to assess the level of expression of these genes in various organs. OsGA3ox1 expression was observed only in the flower, whereas OsGA3ox2 was expressed in almost all the organs tested, but occurred at a higher level in the stem and the flower. The PCR product from the root RNA was not detected in the experimental conditions shown in Figure 2 (25 cycles), but it was observed when the number of cycles was increased to 35 (data not shown), indicating that OsGA3ox2 is also expressed in the root at a low level (see below). OsGA20ox1 was expressed in the flower with low expression in other organs, and OsGA20ox2 was expressed in all the organs, but at a higher level in the leaf and the flower. These results indicate that both OsGA3ox2 and OsGA20ox2 function in vegetative and reproductive organs, whereas OsGA3ox1 and OsGA20ox2 preferentially function in the reproductive organs.
We also examined the pattern of expression of the Gα and the SLR1 genes by RNA gel blot analysis (Figure 2). Both genes were expressed in almost all the organs tested, and the level of expression was high in the young leaf, stem, and flower, but low in the root. The organs showing higher expression of these genes correlated with those that were actively developing and/or elongating. Because the promotion of cell division and elongation is regarded as an important function of GA, the expression pattern of these genes is consistent with sites that will be expected to actively perceive the GA signal.
Expression of the marker genes in germinating seeds
First, we examined the expression of the marker genes in the embryo and the endosperm of germinating seeds. The aleurone cells perceive the GA signal and trigger the expression of α-amylase via intracellular GA signaling (Gubler et al., 1995). The expression of all the GA-biosynthetic genes tested, namely, OsGA3ox1 and OsGA3ox2, and OsGA20ox1 and OsGA20ox2, was localized in the embryo; however, two different expression patterns were evident. OsGA3ox1 and OsGA20ox1 were expressed only in the epithelium, whereas OsGA3ox2 and OsGA20ox2 were expressed both in the epithelium and in the developing shoot region (Figure 3a–d). The GUS expression pattern of OsGA3ox1 and OsGA3ox2 was same as their in situ hybridization patterns, which we previously reported (Kaneko et al., 2002). The expression of OsGA3ox2 and OsGA20ox2 around the shoot apex was observed not only at the embryonic stage but also at the vegetative stage (see later). In the embryo, the expression pattern of the Gα and the SLR1 genes was essentially the same as that of OsGA3ox2 and OsGA20ox2, and both the genes were expressed in the epithelium and the developing shoot region (Figure 3e,f). The expression of these genes was not limited to the embryo, but also occurred in the aleurone layer of the endosperm. These results indicate that GA biosynthesis is localized in the embryo, especially in the epithelium and the shoot apex, and that GA signaling occurs in the same regions of the embryo and also extends to the aleurone layer of the endosperm.
Expression of the marker genes in developing seedlings and roots
Next, we studied the expression of the marker genes in young seedlings (Figure 4). GUS activity was observed in young leaves surrounding the shoot apical meristem (SAM) of the transgenic plants carrying OsGA3ox2:GUS, OsGA20ox2:GUS, Gα:GUS, and SLR1:GUS, whereas no or very low activity was found in the developed leaves and coleoptile. The promoter of OsGA20ox1 produced low GUS activity around the SAM, but no GUS activity was induced by the OsGA3ox1 promoter, indicating that OsGA3ox1 does not function and OsGA20ox1 weakly functions in the young seedlings.
For a more precise analysis of the site of GA biosynthesis and signaling in the SAM, we performed an in situ hybridization study of OsGA3ox2, OsGA20ox2, Gα, and SLR1 expression around the SAM. Signals for all the genes tested were observed in young leaf primordia in the same pattern as the GUS staining of each chimeric construct (Figure 5). Signals were detected at the basal peripheral regions of the SAM (arrows in Figure 5), which correspond to young leaf primordia, P1 or P0, and only low expression of the genes occurred in the corpus region of the SAM. The preferential expression of the GA-related genes in the young leaf primordia, but not in the corpus region of the SAM, may reflect an involvement of these genes in the determination of cell fate in the SAM (see Discussion).
We also examined the expression of the marker genes in the root. Localized GUS staining at the root tip was observed in the transformants carrying the OsGA3ox2:GUS, OsGA20ox2:GUS, Gα:GUS, and SLR1:GUS constructs (data not shown). The expression of these genes at the root tip is consistent with the previous results that GA biosynthesis and signaling is limited to the root tip (Barlow, 1992; Yaxley et al., 2001). No GUS activity was driven by the OsGA3ox1 or OsGA20ox1 promoters (data not shown), indicating that these genes do not function in the root.
Expression of the marker genes in elongating stem
Stem elongation is one of the most well-known GA-dependent biologic events. To investigate the pattern of expression of the GA-related genes in the elongating stem, we analyzed GUS expression in the transgenic plants (Figure 6). At the stage we examined, stem elongation occurred only at the lowest internode, and the SAM was localized just above the upper node of the elongating fifth internode, as indicated by the arrows in Figure 6. As in the shoot apex at the vegetative stage, strong GUS staining was observed in young leaves surrounding the SAM in the transformants carrying OsGA3ox2:GUS, OsGA20ox2:GUS, Gα:GUS, and SLR1:GUS (Figure 6b–e), whereas the transformants with OsGA3ox1:GUS and OsGA20ox1:GUS did not show GUS activity (data not shown).
Strong GUS activity was also observed in the basal region of the elongating internodes of the transformants (Figure 6b–e). The rice internode can be divided into three parts: divisional zone, elongating zone, and elongated zone, from the bottom to the top (Figure 6a). The region with strong GUS activity corresponded to the divisional zone immediately above the node and also overlapped with the elongation zone. The intensity of GUS activity in the region overlapping the elongation zone was reduced gradually from the bottom to the top, perhaps corresponding to a decline in the activity of cell elongation. In fact, the boundary between the elongation and the elongated zones is not clear-cut, but is continuous and is characterized by decreasing elongation activity. Thus, these results demonstrate that the occurrence of cell division and elongation events correspond to the regions of active GA biosynthesis and signaling.
Expression of the marker genes around the inflorescence and floral meristems
Expression of the marker genes was also examined around the inflorescence and floral meristems. Because GUS activity under the control of the promoters for OsGA3ox2, OsGA20ox2, Gα, and SLR1 was highly expressed in inflorescence, and floral tissues and organs, we could not localize the GUS expression to specific tissues or organs, and GUS activity could not be detected using the promoter for OsGA3ox1 and OsGA20ox1 (data not shown). Thus, we performed in situ hybridization of OsGA3ox2, OsGA20ox2, Gα, and SLR1. In longitudinal sections of the inflorescence shoots at the primary rachis primordium differentiation stage, a high-level expression of these genes was observed in the inflorescence meristem itself, and also at a moderate level in the leaf primordia (Figure 7a,f,k,p). Following the transition from the primary to the secondary rachis differentiation stage, the localized expression of the genes continued in the inflorescence meristem, but was reduced in the leaf primordia, relative to that detected in the primary rachis meristem (Figure 7b,g,l,q). Following floral induction, their expression continued in the whole floral meristem (Figure 7c,h,m,r).
When the development of the stamen could be observed morphologically, high-level expression of the GA-biosynthesis genes (OsGA3ox2 and OsGA20ox2) was localized to the primordia of the stamen, whereas only low expression of these genes occurred in other floral organs, such as the glume, lemma, palea, and pistil (Figure 7d,i). The expression of the GA-signaling genes (Gα and SLR1) also preferentially occurred in the stamen primordia, but was less specific than that of the GA-biosynthesis genes (Figure 7n,s). These observations demonstrate that active GA biosynthesis and signaling occur in the inflorescence and floral meristems and also in the stamen primordia, following the floral organ developmental stage. The results also suggest that bioactive GA synthesis occurs in stamen primordia more specifically than GA signaling (see Discussion). As a control, we also performed in situ hybridization of sections at the floral organ differentiation stage using the sense probe for each gene (Figure 7e,j,o,t). No signals were detected in these sections, indicating that the staining observed in the sections, hybridized with the antisense probes, reflects the localization of the mRNA of each gene.
We observed the expression of the marker genes in developing anthers by in situ hybridization (Figure 8). Interestingly, all the genes we tested were expressed at a high level in the tapetum. Simultaneous expression of all the genes tested is unusual, and it was observed only in the tapetum and the epithelium at various stages of development of rice plants studied here. It is possible that the tapetum has a unique role in GA biosynthesis in a similar way that the epithelium has for supplying bioactive GAs to the aleurone layer (see Discussion).
Molecular markers for GA biosynthesis and signaling
We used four GA-synthesis genes and two GA-signaling genes as molecular markers for the sites of GA biosynthesis and signaling in rice plants. To identify the location of bioactive GA synthesis based on the expression pattern of the GA biosynthesis enzymes, the gene encoding GA 3-oxidase should provide the best indicator as this enzyme catalyzes the final step in GA biosynthesis to produce the bioactive GAs, GA1 and GA4 (Figure 1a). Thus, we used this gene as a molecular marker for GA biosynthesis. We carefully searched for genes encoding GA 3-oxidase in the rice genome because the enzyme is encoded by a small multigene family in several plant species (Hedden and Phillips, 2000). Based on a blast search of all the available rice genome databases and a genomic DNA gel blot analysis, we concluded that rice has only two genes encoding GA 3-oxidase.
We also studied the expression pattern of two GA 20-oxidase genes in rice. GA 20-oxidase catalyzes the production of the substrates for GA 3-oxidase. Therefore, if the site of expression of a GA 20-oxidase gene overlaps with that of a GA 3-oxidase gene, this would strongly indicate that the bioactive GAs are produced at the site where both genes are expressed. Our results show that the expression pattern of the GA 3-oxidase and GA 20-oxidase genes overlap in rice plants, strongly suggesting that the tissues and organs where the GA 3-oxidase gene is expressed also actively synthesize the bioactive GAs.
In contrast to GA biosynthesis, relatively little is known about the mechanisms of GA-signal perception. We used two other genes as molecular markers of GA-signal transduction, namely the Gα and the SLR1 genes. A phenotypic analysis of the loss-of-function mutant of Gα (d1), and epistatic analysis of d1 with a GA-signaling mutant, slr1, indicates that Gα is involved in GA signaling (Ueguchi-Tanaka et al., 2000). On the other hand, recent studies suggest that the Gα protein is also involved in various other signal-transduction pathways, including auxin (Jones, 2002) and pathogen-resistance reactions (Suharsono et al., 2002). With this in mind, we took care to account for the possibility that in some places, Gα may not function in GA signaling. Thus, we used another molecular marker for detecting GA-signal transduction, SLR1. The SLR1 protein is considered to be a repressor protein of GA signaling in rice (Ikeda et al., 2001). Recently, we determined that SLR1 encodes a protein homologous to GAI (Peng et al., 1997) or RGA (Silverstone et al., 1998), which functions as a negative regulator of GA signaling in Arabidopsis (Ikeda et al., 2001). Further analyses have revealed that the protein is localized in nuclei to repress the GA-signaling pathway (Itoh et al., 2002). The nuclear-localized SLR1 protein is rapidly degraded by GA treatment to induce GA actions such as shoot elongation and downregulation of GA 20-oxidase expression. Interestingly, GA treatment simultaneously causes the upregulation of SLR1 transcription (Itoh et al., 2002; Ogawa et al., 2000). Thus, GA treatment results in two opposite effects on SLR1, i.e. the rapid degradation of the SLR1 protein and the induction of the SLR1 transcription.
As the expression of mRNA for the GA-related genes does not directly mean the presence of their protein products, we have to carefully discuss the sites of GA biosynthesis and signaling, based on the results of promoter-GUS and in situ hybridization. However, the expression sites of the marker genes we tested in this study were well overlapped with regions actively occurring GA-action. Thus, the tissues or organs actively expressing the marker genes should correspond to the sites of GA biosynthesis and signaling in rice plants.
GA biosynthesis and signaling in germinating seeds
In germinating seeds of rice, the GA-biosynthesis genes are expressed in the embryo in two patterns, i.e. OsGA3ox1 and OsGA20ox1 are expressed only in the epithelium, and OsGA3ox2 and OsGA20ox2 are expressed in both the epithelium and the developing shoot region (Figure 3a–d). The localization of GA biosynthesis in the epithelium and the shoot region of the embryo has been reported in a previous study using in situ hybridization (Kaneko et al., 2002). Kaneko et al. (2002) used embryonic organ-deficient rice mutants to investigate why GA is synthesized in two different sites of the embryo, and proposed that the bioactive GAs synthesized at the epithelium are essential to induce α-amylase expression in the endosperm, whereas bioactive GAs in the shoot apical region are important for the shoot development. Two sets of GA-biosynthesis enzymes are expressed in the epithelium as opposed to one set in the shoot region, possibly because highly active GA biosynthesis is needed in the epithelium to induce the expression of α-amylase and other hydrolyzing enzymes in the endosperm.
We found that the expression of the marker genes for GA signaling in the embryo overlaps with that of the GA-biosynthesis genes (Figure 3e,f), indicating that GA biosynthesis and signaling occur in the same place in the embryo. The expression of the markers Gα and SLR1 also occurs in the aleurone layer of the endosperm; however, the GA-biosynthesis genes are not expressed at this site. This finding is consistent with previous physiologic and biochemical studies, which have shown that bioactive GAs produced in the embryo are transported to the aleurone layer (Fincher, 1989), and they trigger the expression of α-amylase at the transcriptional level (Gubler et al., 1995). Our results support this model and clearly demonstrate that the aleurone layer has a capacity to perceive the GA signals, but not to synthesize the GAs, an unusual relationship between GA biosynthesis and signaling.
GA biosynthesis and signaling in the shoot apical region
We have previously discussed the relationship between GA biosynthesis and cell-fate determination around the tobacco shoot apical region (Sakamoto et al., 2001). According to our model, a KNOX homeodomain protein, NTH15, which is expressed in the corpus region of the SAM, negatively regulates the expression of GA 20-oxidase via a direct interaction with the cis-acting element of the GA 20-oxidase gene in order to maintain the indeterminate state of the corpus cells. When NTH15 is downregulated at the flanks of the SAM, the suppression of GA 20-oxidase expression is released and GA biosynthesis starts, inducing organized cell division and subsequent determination of the cell fate. The localized expression of the genes for GA biosynthesis and signaling in the rice SAM is consistent with this model. Actually, the expression of the GA-related genes is apparently restricted to the basal and the peripheral regions of the SAM, in which cell fate has been determined, rather than in the corpus region consisting of cells in an indeterminate state (Figure 5). On the other hand, expression of a rice KNOX-type homeobox gene, OSH1, is localized to the corpus region of the SAM (Sentoku et al., 1999). Thus, the model for GA biosynthesis and determination of cell fate may be adaptable to the rice vegetative SAM.
GA biosynthesis and signaling in the flower
The expression of the GA-biosynthesis genes (OsGA3ox2 and OsGA20ox2) is restricted to the stamen primordia (Figure 7d,i), whereas the GA-signaling genes (Gα and SLR1) are preferentially expressed in the stamen primordia, but are also expressed at a lower level elsewhere (Figure 7n,s). It has been reported that GAMYB in Lolium temulentum is also preferentially expressed in the stamen primordia (Gocal et al., 1999). Once the anthers are morphologically developed, all the GA-related genes that we tested are expressed at high levels in the tapetum cells (Figure 8). Bioactive GAs produced in the anthers are considered to have an important role at various steps in the later stages of flower development, such as growth and opening of petals, and anthocyanin pigmentation (Bala et al., 1985; Jacobsen and Olszewski, 1991; Plack, 1958; Weiss and Halevy, 1989). Actually, the bioactive GAs have been detected at a high level in the anthers of various kinds of plants including rice (Hasegawa et al., 1995; Pharis and King, 1985). Furthermore, the removal of the anthers prevents the normal development of the corollas of petunia flowers, but the exogenous application of bioactive GA can substitute for the anthers. Based on these observations, Weiss et al. (1995) proposed that the anthers are a source of bioactive GA for other floral organs. Our findings that the expression of the GA-biosynthesis genes is restricted to the tapetum cells of the anthers, whereas the GA-signaling genes are expressed more widely in other floral organs support this model. It is interesting that high-level expression of OsGA3ox1 and OsGA20ox1 occurs only in the tapetum and the epithelium layers, both of which are considered to be an important source of bioactive GAs for other tissues and organs. This suggests to us that, in rice plants, OsGA3ox2 and OsGA20ox2 are normally used to produce bioactive GAs, but that another set of GA 3-oxidase and GA 20-oxidase enzymes is also mobilized in specific tissues, namely the tapetum and the epithelium, to produce a high level of bioactive GAs for other tissues and organs.
The presence of two sets of GA 3-oxidase and GA 20-oxidase enzymes with different expression patterns may also explain the difference in the dominant bioactive GAs found in vegetative and reproductive organs, i.e. GA1 is a dominant bioactive GA in vegetative organs in rice, whereas GA4 is dominant in reproductive organs (Kobayashi et al., 1988). Kobayashi et al. (1989) also noted that GA4 occurs at a high level in reproductive organs even in the loss-of-function mutant of OsGA3ox2, d18dy.Hasegawa et al. (1995) studied the localization of GA4 in the anther using immunohistochemistry and found that GA4 accumulates in the tapetum. Taken together, the expression patterns of the GA 3-oxidase and GA 20-oxidase enzymes and the previous biochemical analyses of GA levels suggest that the set of OsGA3ox1 and OsGA20ox1 genes preferentially produce GA4, whereas the other set of OsGA3ox2 and OsGA20ox2 genes function, as a default, to produce GA1 during the vegetative and the reproductive stages. Further biochemical studies on these enzymes is necessary to confirm this consideration.
The rice cultivar Oryza sativa L. cv. Nipponbare (wild type) and five rice mutants, d35 (mutant of ent-kaurene oxidase), d18h (mutant of OsGA3ox2), sd1 (mutant of OsGA20ox2), d1 (mutant of Gα), and slr1 (mutant of SLR1), were used in this study. Rice seeds were immersed in water for 2 days, grown for 1 month in a greenhouse, and then transplanted to the field.
Semiquantitative RT-PCR assay and RNA gel blot analyses
mRNA levels for OsGA3ox1, OsGA3ox2, OsGA20ox1, and OsGA20ox2, in various organs, were analyzed by semiquantitative RT-PCR assay. Total RNAs isolated from various organs were treated with RNase-free DNaseI, and cDNA was produced from 5 µg of RNA with the Ready-To-Go T-primed First-strand Kit (Pharmacia Biotech, Piscataway, NJ, USA). One-tenth of the cDNA obtained by reverse transcription was PCR-amplified using suitable primers, which were localized at the different exons of each gene in order to confirm that the PCR products were derived from the cDNA rather than from the genomic sequence (forward or reverse primer for actin, TCCATCTTGGCATCTCTCAG or GTACCCGCATCAGGCATCTG; forward or reverse primer for OsGA3ox1, ATGGAGGAGTACGACTCGTCGTCGATGAGAG or CTCTGCAGGATGAAGGTGAAGAAGCCTG; forward or reverse primer for OsGA3ox2, TCTCCAAGCTCATGTGGTCCGAGGGCTA or TGGAGCACGAAGGTGAAGAAGCCCGAGT; forward or reverse primer for OsGA20ox1, TACGGGCCGACATGCGCACG or GCATGCATGTAGGAGTAGCTAGG; forward or reverse primer for OsGA20ox2, GCGCCATGGTCATCAACATCGG or AGCGCATGAGGTCGGCCCAGGT). As OsGA20ox1 does not contain any introns, we confirmed that the PCR product was derived from its cDNA by including or omitting the reverse-transcription step prior to PCR amplification. The PCR product was not produced when the reverse transcription step was omitted, confirming that the PCR product was derived from the cDNA and not from the genome sequence. Following agarose gel electrophoresis, the band intensity was examined as a relative content of the transcript.
In RNA gel blot analysis, total RNA in various organs was prepared as described by Sambrook et al. (1989). RNA (10 µg per sample) was electrophoresed and transferred to Hybond N+ nylon membrane. Hybridization was performed in 5× SSC, 10% (w/v) dextran sulfate, 0.5% (w/v) SDS, and 0.1 mg ml−1 denatured salmon sperm DNA at 65°C. Filters were washed twice in 2× SSC, 0.1% SDS at 65°C for 15 min, and once in 0.2× SSC, 0.1% SDS at 65°C for 15 min.
Construction of chimeric genes and rice transformation
The promoter sequence for each GA-related gene was digested using the appropriate restriction enzymes or amplified by PCR using the genomic clone as a template. Where PCR was used for cloning the promoter sequence, the resulting product was cloned into pCRII (Invitrogen, Carlsbad, CA, USA) and was sequenced to confirm that no base substitution had occurred during PCR. The promoter sequences were introduced at the front of the GUS reporter gene of pBI-Hm (kindly provided by Dr Kenzo Nakamura at Nagoya University) to produce a fusion with the GUS reporter gene. The chimeric constructs were introduced into the Agrobacterium tumefaciens strain, EHA101, and used to infect rice callus, according to Hiei et al. (1994). Transformed cells and plants were screened by hygromycin selection and maintained in sterile culture; regenerated plants were then grown to maturity in pots in a greenhouse. The primary transformants were self-pollinated and the resulting seeds (T1) were collected.
Histochemical analysis of GUS activity
Histochemical analysis of GUS activity was performed as previously described by Matsuoka and Sanada (1991). The histochemical GUS staining was conducted for one or more hours for strong GUS activity, or overnight for faint GUS activity.
In situ hybridization
Plant materials were fixed in 4% (w/v) paraformaldehyde and 0.25% (v/v) glutaraldehyde in 0.1 m sodium phosphate buffer (pH 7.4) overnight at 4°C, dehydrated through a graded ethanol series followed by a t-butanol series, and finally embedded in Paraplast Plus (Sherwood Medical, St Louis, MO, USA). Microtome sections (8–10 µm thick) were mounted on glass slides treated with silane. Digoxygenin-labeled RNA probes were prepared from the 3′-terminal half of each cDNA. Hybridization and immunologic detection of the hybridized probes were performed according to the method described by Kouchi and Hata (1993).
We are grateful to Ms Masako Hattori (Nagoya University, Nagoya, Japan) for her technical assistance. This work was supported, in part, by a Grant-in-Aid for Center of Excellence to Y.I. and M.M. and a Grant-in-Aid from the Program for the Promotion of Basic Research Activities for Innovation Biosciences to M.U.-T. and M.M.