Abstract

In many eukaryotic organisms, homeobox genes are important regulators that specify the cell fate and body plan in early embryogenesis. In this study, a gene designated OSTF1 (Oryza sativa transcription factor 1) encoding a homeodomain protein in rice was isolated and characterized. The encoded OSTF1, although sharing only approximately 51% sequence identity with other HD-GL2 members, contains four characteristic motifs (an N-terminal acidic region, a homeodomain, a truncated leucine zipper, and a START domain). OSTF1 was detected as a single copy gene in rice. The transcripts were absent in young panicle or mature spikelet before anthesis, but appeared very early in the pollinated grain with a transient profile. In vegetative tissues examined, expression was only detectable in root. In situ hybridization analysis on developing grains revealed that OSTF1 was strongly and uniformly expressed in the embryo at the globular stage and preferentially localized to the protoderm at 3–6 d after pollination. Expression was also detectable in the integument and throughout the endosperm. Although OSTF1 is not closely related to the remaining HD-GL2 members in sequences, this gene exhibits an analogous epidermis-preferential expression pattern.

Introduction

The homeobox genes are important regulatory genes in the specification of cell fate and body plan at the early stage of embryogenesis in higher organisms. These genes have been found to be clustered together in the same order along the chromosome as being expressed along the antero-posterior body axis in animals (Ingham 1988, Lawrence and Morata 1994). However, they are dispersed throughout the genome in plants although small clusters have been reported (Liu et al. 1999, Tavares et al. 2000). The homeobox genes share a consensus DNA sequence of 180 bp called the homeobox. The encoded 60-amino-acid homeodomain (HD) contains a characteristic DNA-binding structure formed by three α-helices separated by a loop and a turn. Through interactions of the HD with specific DNA sequences, these proteins act as transcription factors and turn on target gene expressions in a precise spatial and temporal pattern (Gehring et al. 1994a, Gehring et al. 1994b).

The homeobox genes in plants have been cloned from various species including gymnosperm (Ingouff et al. 2001, Sundas-Larsson et al. 1998) and ferns (Aso et al. 1999). Based on amino acid sequences within the HD and other characteristic motifs, they have been catalogued into five families: plant HD finger (PHD-finger), HD-BEL1, HD-knotted1 (HD-KN1), HD-leucine zipper (HD-ZIP), and HD-glabra2 (HD-GL2, also known as HD-ZIP IV). These families exhibit distinct expression profiles inferring their specific regulatory roles in tissue or organ differentiation (Chan et al. 1998). As an example, the transcription of the HD-KN1 genes is undertaken specifically in the meristemic regions of vegetative shoot, but not in the differentiated L1 layer or leaf primordia. This expression pattern suggests that the HD-KN1 transcription factor may promote indeterminate growth and its down-regulation is critical for initiation of the determinate development of lateral organs (Sato et al. 1996, Smith et al. 1992).

Unlike the HD-KN1 type, genes of the HD-GL2 family display a different spatial specificity in expression. In Arabidopsis, the GL2 gene is specifically expressed in the outer cell layer of shoot and root, although it has opposite functions in these two organs. This gene is considered to promote trichome differentiation in shoot (Rerie et al. 1994) but to inhibit root hair formation (Di Cristina et al. 1996). Moreover, the expression of ATML1 is restricted to the protoderm cell layer in the 16-cell embryo and remains in the L1 layer of shoot apical meristem, inflorescence meristem, and floral meristem throughout the diploid life cycle (Lu et al. 1996, Sessions et al. 1999). The ZmOCL1, ZmOCL3, ZmOCL4, and ZmOCL5 genes in maize also essentially exhibit the L1/protederm/epidermis-specific expression patterns during early embryogenesis and in meristematic regions or young organ primordia at later stages (Ingram et al. 1999, Ingram et al. 2000). These observations show that the expressions of HD-GL2 genes are restricted to the epidermal tissues or their precursor cells. The ZmOCL2 gene in maize is the only exception. Although it is meristem-specific, expression of this gene is restricted to the subepidermal cell layers (Ingram et al. 2000). The HD-GL2 may represent a unique family of homeobox genes that specialize in control of lateral patterning in embryos and in maintaining cell-layer identity in meristematic regions (Chan et al. 1998).

The early zygotic divisions of many dicot and monocot species are relatively regular (Natesh and Rau 1984). In Arabidopsis, the protoderm is the first tissue that differentiates as early as the 16-cell embryo stage. The lineages and characteristics of the epidermal cells are well elucidated, not only by histological and ultrastructural studies, but also by characterizing the expression patterns of epidermis-specific genes such as ATML1 (Lin and Schiefelbein 2001, Lu et al. 1996). In rice, the embryonic cleavage pattern at the very early stage has been recognized as regular, but is considered to be irregular after the ten-cell stage (Suzuki et al. 1992). This assumption suggests positional controls rather than the cell lineage effect in determining cell fates. However, the difficulties in resolving the complexity of cell division planes by ultrastructural studies may lead to such a conclusion. Therefore, studying the early molecular markers associated with cell lineage may bring new insights into the early embryo development of rice.

To investigate the transcriptional factors with specification on tissue or cell-layer identity, we cloned and characterized a HD-GL2 homeobox gene termed Oryza sativa transcription factor 1 (OSTF1). The encoded OSTF1 contains the characteristic motifs of HD-GL2 proteins but shares a relatively low sequence similarity with ATML1, ZmOCL5, or O39. The transient- and epidermis-preferential expression profile of OSTF1 in developing grains suggests its role in early embryo and endosperm development.

Results

Molecular cloning of OSTF1 cDNA

In order to isolate the HD-GL2 homeobox gene from rice, the EST databank (http://www.ncbi.nlm.nih.gov/) was assumed to have information on some HD-GL2 cDNA fragments. After performing the TBLASTN search (Altschul et al. 1997) using ATML1 protein sequence as a query, a rice EST clone (C74009) was found to encode a putative polypeptide with 65% sequence similarity to ATML1 between amino acid residues 576–633. The DNA fragment corresponding to C74009 was amplified and used as a probe to screen cDNA library of rice young panicle. Together with RT-PCR and 5′-RACE, the three DNA fragments obtained were assembled according to their overlapped regions to yield pOSTF1-6 (see Materials and Methods). A full-length cDNA, designated OSTF1, was revealed to encode a protein with 709 amino acids (predicted molecular mass = 78.2 kDa and pI = 7.4). The 5′- and 3′-untranslated regions were 90 and 297 bases long, respectively. A putative polyadenylation signal was observed 142 bases upstream from the poly(A) tail. Sequence comparison of the full-length cDNA with its genomic DNA showed that exons are spread over 6 kb and split by 11 introns. Except for the fact that intron I (2,100 bp) and intron II (737 bp) were relatively large, sizes of the remaining introns ranged from 83 to 299 bp (Fig. 1). All introns began with GT and end with AG, consistent with the donor/acceptor splice rule.

OSTF1 is a member of the HD-GL2 family

The predicted amino acid sequence of OSTF1 has a HD at its N-terminal region between residues 65 and 124 (Fig. 1). A BLASTP search of the GenBank database revealed that OSTF1 is similar to ATML1, ZmOCL5, and O39 in sequence. However, the latter three proteins are highly homologous to each other (about 82% similarity) whereas OSTF1 has only about 51% similarity to any of them. A phylogenetic tree constructed using the HD sequences from different families of plant HD proteins also showed that OSTF1 is clustered within the HD-GL2 family, although relatively divergent from all the other members (Fig. 2). The phylogenetic comparison based on the full-length protein sequence exhibited a similar result except that ZmOCL2 fell into the same group as ANL2, ZmOCL1, and ZmOCL3 (data not shown).

To gain additional information about phylogenetic relationships between OSTF1 and other HD-GL2 genes, the ten introns (I–X, Fig. 1) located in the coding sequence of OSTF1 were compared with that of ATML1 (GenBank accession number AL035527) (Sessions et al. 1999) and ANL2 genes (GenBank accession number AF077335) (Kubo et al. 1999). Except for the fact that intron I at the region encoding variable acidic domain (see below) cannot be compared, the positions of other introns are conserved among these three genes. Only intron IV in OSTF1 is absent in ATML1, and neither intron IV nor IX are found in ANL2. This conservation in exon/intron organization again suggests that OSTF1 is a member of the HD-GL2 family.

Predicted structural features of OSTF1

In contrast to most HD proteins with HDs at the middle or in the C-terminal region (Chan et al. 1998), this domain in OSTF1 is N-terminal located (Fig. 3A). When aligned with several highly homologous HD-GL2 sequences (Fig. 3B), OSTF1 is less conserved but contains the sequence characteristics of HD-GL2 proteins described by Lu et al. (1996). The N-terminal end of its HD features basic amino acid clusters, which may act as a nuclear localization signal (Hicks et al. 1995). Moreover, the sequence in the helix-III region (QVKFWFQNkRTQmK) is strictly conserved. This region is the recognition helix that provides sequence-specific interactions with the major groove of target DNA (Gehring et al. 1994b). A homology search of the GenBank database with this sequence consistently recovered only HD-GL2 proteins (data not shown), indicating the conservation between the sequence in helix-III and the overall architecture of protein.

Outside the HD proper, OSTF1 also shares several common elements with other HD-GL2 proteins (Fig. 3). Firstly, an N-terminal region preceding the HD is enriched with acidic amino acids (Fig. 1). Similar acidic domains, although highly variable in length and sequences, were found in other HD-GL2 proteins. Secondly, a putative truncated leucine zipper motif is located adjacent to the C-terminal of the HD (Fig. 3B). The pattern of sequence alignment suggests a conserved structure of six heptad-repeats (I–VI) with mainly leucine at the “d” positions. A loop constituted by 18 amino acids interrupts between heptad-repeat III and IV. In the Athb-10 (GL2) from Arabidopsis, this motif has been demonstrated to mediate protein dimerization by a domain exchange experiment (Di Cristina et al. 1996). It is worth to noting that a “Cys-X-X-Cys” sequence with uncharged amino acid residues in the neighboring region is consistently found in OSTF1 and other HD-GL2 members. Intra- or inter-molecular disulfide bonds may form between these two cysteine residues to stabilize the leucine zipper scaffold. Thirdly, predicted by PSI-BLAST searches (Altschul et al. 1997), a StAR-related lipid transfer (START) domain (~220 amino acid residues), sharing 55–60% sequence similarity with its corresponding domain in other HD-GL2 proteins, is located in the middle region of the protein (Fig. 3B). This domain has also been identified in steroidogenic acute regulatory protein (StAR) orthologues, phosphatidylcholine (PC)-transfer proteins, RHOA-specific GTPase-activating proteins (RHOGAP) from mammals, and in pleckstrin homology (PH)-domain proteins from nematode (Ponting and Aravind 1999). Study on the crystal structure of the START domain in the StAR protein has revealed a hydrophobic tunnel large enough to bind a single cholesterol molecule (Tsujishita and Hurley 2000).

Genomic Southern blot analysis

To examine the copy number of OSTF1 gene in rice, a blot of the rice genomic DNA digested with various restriction enzymes was used. Although the hybridization probe contained a small region of the conserved homeobox sequence, only one band was detected at each lane when washed under high stringency conditions (Fig. 4). This finding suggests that cross-hybridizations between the probe and other homeobox genes did not occur and OSTF1 is a single copy gene in rice. A databank search (http://www.rice-research.org) revealed that the rice genome contains nine putative HD-GL2 genes. Among their translated protein sequences, OSTF1 represents the most divergent one from the well-characterized ATML1/ZmOCL5/O39 subfamily (data not shown).

Expression of OSTF1 gene in different organs

The expression patterns of OSTF1 gene in different organs at various developmental stages were examined by Northern blot analysis. Using the probe for Southern hybridization analysis, a single band was detected in grains at 1–5 d after pollination (DAP), but not in young panicles, meiosis panicles, or flowering spikelets even after extended exposure. Although DNA fragments corresponding to OSTF1 gene were found in the EST (C74009) and cDNA (pOSTF1-1) clones from the young panicles, their expression was not detected by Northern blotting (Fig. 5A). The hybridized signal was estimated to be ~2.8 kb, consistent with the size of the isolated cDNA (Fig. 5B).

To characterize the temporal expression profile of OSTF1 during embryogenesis, grains at different DAP were collected for Northern blotting. Signals were first detected at 3 DAP. After peaking at 4–5 DAP, the expression gradually diminished to a barely detectable level at 8 DAP (Fig. 5B). Furthermore, 1–5 DAP grains were separated into the upper half (containing the hull, pericarp, and endosperm) and lower half (containing the hull, pericarp, endosperm, and embryo). Signals of transcript were detected in both parts with similar intensities (Fig. 5B). These findings suggest that the expression of OSTF1 is not embryo-specific. Among the vegetative tissues examined, the expression was strong in the root but barely detectable in the callus, green seedling, mature leaf, or old leaf (Fig. 5C).

In situ localization of OSTF1 mRNA

In order to further define the spatial expression patterns of OSTF1, in situ hybridization analysis was conducted on embryo and root. Consistent results were observed for serial samples on slides and between hybridizations. In the globular embryo at 2 DAP, OSTF1 transcripts accumulated rather uniformly in the embryo proper. Clear signals were also detected in both layers of the integument (Fig. 6A). At 3 DAP, the embryo consisted of ~200 cells and was flat at the ventral side but spherical at the dorsal side. The expression of OSTF1 appeared stronger at the outer cell layer of the embryo. The signals in the integument faded. Moreover, OSTF1 transcripts were detectable throughout the endosperm, in which cellularization had already been completed and starch had begun to accumulate. The signals were stronger in the outer endosperm layer (or prealeurone layer), which later develops into the aleurone layer (Fig. 6B). Such an epidermis-preferential expression pattern was also observed in embryos undergoing shoot apex differentiation at 4 DAP (Fig. 6C). By 6 DAP while embryos were forming the leaf primordia, signals were found in the integument, the scutellar epithelium, and the embryo axis (Fig. 6D). Such observations again suggest that the expression of OSTF1 is not epidermis-specific. No signal was found within the pericarp or nucellus.

At 8 DAP, embryo development was close to mature in morphology, and the endosperm was filled with starch. At this stage, transcripts in the embryo itself and in the endosperm were greatly diminished (Fig. 6E). These findings are consistent with the results of Northern blot analysis (Fig. 5B). Signals in the integument were still detectable at 8 DAP but totally disappeared at about 10 DAP (data not shown), the stage at which the seed became desiccated. No hybridization signal was detected when the sense probe was used (data not shown). These findings lead to the conclusion that the expression of OSTF1 switches from a ubiquitous pattern in the proembryogenic masses to an outer cell layer-preferential localization during early embryo development.

In the root tissues, the expression of OSTF1 was not detected in the root cap or meristematic region (data not shown). In the elongation area of root from 3-day-old seedlings, signals appeared strongly in the epidermis and root hairs. The semi-developed central cylinder also exhibited a weak expression. Interestingly, instead of being evenly distributed within cells, the signals were localized only at the root-hair forming side in the epidermal cells (Fig. 6F).

Discussion

This report describes the isolation of a homeobox gene and characterization of its expression in rice. We found that the isolated OSTF1 gene encodes a protein that possesses the characteristics of HD-GL2 transcription factors. These characteristics are a sequential combination of an acidic domain, a HD, a truncated leucine zipper motif, a START domain, and an uncharacterized C-terminal region (Fig. 3A). HD proteins with this specific organization have been found only in plants but are not restricted to the HD-GL2 family. Members of the HD-ZIP III family such as Athb-8, Athb-9, Athb-14, and REVOLUTA/IFL1 (Otsuga et al. 2001, Sessa et al. 1998, Zhong and Ye 1999) also possess the same domain organization. However, a longer uncharacterized C-terminal region, a canonical leucine zipper motif, and small length differences between helix motifs were found in the HD-ZIP III but not in the HD-GL2 proteins (Sessa et al. 1998). Moreover, the consensus sequence in the helix-III of HD-ZIP III (QIKVWFQNRRCREK) (Sessa et al. 1998) is slightly different from that of HD-GL2 (QVKFWFQNkRTQmK) (Fig. 3B). These differences may contribute to variations in their protein–DNA interaction modes and DNA-recognition specificities. Although sharing similar structures, these two families of transcription factors may activate (or repress) different sets of genes and be involved in various aspects of plant development. Indeed, unlike the HD-GL2 genes preferentially expressed in epidermis, the Athb-8 mRNA is restricted to the procambial regions of embryo and developing organs (Baima et al. 1995). Moreover, the IFL-1/REVOLUTA1 regulates fiber differentiation in the vascular regions of Arabidopsis (Zhong and Ye 1999).

Both the HD-GL2 and HD-ZIP III proteins possess a START domain (Fig. 3). In animals, the StAR protein that contains the START domain acts as a sterol transfer protein (Kallen et al. 1998) and can bind cholesterol in vitro (Tsujishita and Hurley 2000). However, in addition to proteins involved in lipid transport, the START domain exists in various categories of proteins for signal transduction or transcriptional regulation (Ponting and Aravind 1999). Hence it may act as the crossroads where signals and protein activities engage. By analogy with the nuclear steroid receptors in animals (Evans 1988), the plant HD-GL2 proteins may be able to mediate transcription in a steroid-dependent manner. Brassinosteroids (BRs), a group of plant hormones constituted by polyhydroxy steroids (Schumacher and Chory 2000), are the possible ligands for the HD-GL2 proteins. Based on the buried pocket-like binding structure of the START domain (Tsujishita and Hurley 2000), a diffusible lipid molecule is likely to be its binding target. Such a structural consideration supports the possibility of BRs-binding to the HD-GL2 proteins. However, only the protein sequence of this domain has been reported in the plant kingdom (Ponting and Aravind 1999). Further identification of the putative plant ligands binding to the START domain may bring new insights into signal transductions and transcriptional regulations.

HD-GL2 apparently has several members in the plant genome. An in silico analysis of the Arabidopsis genome revealed more than 24 candidates (data not shown), very close to the number estimated previously (Tavares et al. 2000). Since about 70% of the identified genes in Arabidopsis have at least one gene paralogue (Walbot 2000), there may be about 14 HD-GL2 genes in the small Arabidopsis genome. Similarly, nine genes were found in the rice genome database (http://www.rice-research.org). They are expected to have similar biochemical activities but not necessarily the same biological roles. Ectopic overexpression of an ATML1-like gene in Norway spruce leads to an early block in the development of somatic embryos in which the smooth surfaces are absent (Ingouff et al. 2001). Since the physiological roles of OSTF1 in rice remain unclear, further investigations are now progressing by transgenic plant analysis.

In this study, we revealed the temporal and spatial expression pattern of OSTF1 during early embryogenesis in rice. Rice embryos develop much faster than those of other cereal plants. The globular stage lasts only 3 DAP, and most of the morphogenetic events in embryogenesis are accomplished within 1 week (Sato et al. 1996). Although the embryo may not have signs of organ differentiation on day 2 after pollination, transcripts of OSTF1 are already highly accumulated (Fig. 6A). This early onset of OSTF1 expression and its putative role as a transcription factor imply that OSTF1 may function in a regulatory process involving organ or tissue determination. At 3 DAP, OSTF1 switches from a ubiquitous pattern into a much more epidermis-specific expression in the early embryo (Fig. 6B–D). These changes are very similar to what occurs in some of the angiosperm HD-GL2 genes (Ingram et al. 1999, Lu et al. 1996).

In Arabidopsis, the L1 box, an 8-bp motif (5′-TAAATG(C/T)A-3′), has been identified as the target of ATML1 protein and conserved within promoter regions of the L1 layer-specific genes (Abe et al. 2001). Similar to the L1 box, a sequence “TAAATGCC” with only one base difference was found 436 bp upstream from the cloned OSTF1 cDNA (data not shown). These findings indicate that dicot and monocot plants may have common transcriptional factors and cis-regulatory elements driving epidermis-specific expressions, although they exhibit distinct patterns in embryo development. However, unlike the spatial specificity of the remaining HD-GL2 members, OSTF1 transcripts are never totally absent from the inner cell mass of embryo or endosperm at the early developmental stages (Fig. 6). This preferential, but not specific, expression of OSTF1 in epidermis is consistent with the fact that OSTF1 does not belong to any subfamily of the HD-GL2 proteins (Fig. 2). OSTF1 represents a HD-GL2 member with least sequence similarity to the identified genes from various plants.

Materials and Methods

Growth conditions and collection procedures for plant materials

Rice plants (Oryza sativa cv. TNG67) were grown in the field or in pots containing field soil from February to June. Young panicles were collected by hand-dissection. Meiosis panicles were collected from plants (internode below flag leaf = 1–5 cm) containing spikelets undergoing meiosis. Flowering spikelets were collected while the lemma and palea remained open and pollen was releasing. Rice grains were marked and collected each day from 1 to 9 DAP. The grains collected at 1–5 DAP were mixed and separated into upper and lower parts by cutting at the middle. Roots were collected from water-cultured plants 3–5 d after germination. Samples for RNA extraction were pooled, immediately frozen in liquid nitrogen, and then stored at –80°C before analysis.

Construction of cDNA libraries

An oligo (dT)-primed cDNA library was constructed using poly(A)+ RNA isolated from young panicles (1–10 cm) by the ZAP-cDNA/Gigapack III Gold Cloning Kit (Stratagene, La Jolla, CA, U.S.A.). The lambda phage was transfected into E. coli XL1-Blue and displayed on plates.

Isolation of OSTF1 cDNA clone and in silico analysis of the cognate genomic DNA

The cDNA fragment corresponding to expressed sequence tag (EST) C74009 was amplified by reverse transcription (RT)-PCR using the forward primer 5′-GCAGAACAACGGGTACGACGC-3′ (bases 1,842–1,862 on OSTF1 cDNA, Fig. 1) and reverse primer 5′-CACAGACCGGGAATACATG-3′ (bases 2,118–2,100 on the reverse strand of OSTF1 cDNA). The poly(A)+ RNA isolated from the young panicles were reverse transcribed and used as templates for PCR (template denatured at 94°C for 2 min, then 35 cycles of 94°C for 30 s, 55°C for 1 min, and 72°C for 1 min). After subcloning a predicted fragment of 277 bp, the sequence was confirmed and used as a probe for cDNA library screening. For library screening, prehybridization was performed in 0.1% SDS, 6× SSC, 5× Denhardt’s solution containing 1 mg ml–1 herring sperm DNA at 50°C for 2 h. Hybridization was then carried out in 50 ml of the same buffer with isotope-labeled probe at 50°C overnight. Filters were washed once with 2× SSC at room temperature for 10 min, twice with 0.2× SSC and 0.1% SDS at 65°C for 20 min, and then were exposed to X-ray film for 4 d. After screening on about 6×105 plaques, a total of 18 strongly hybridized clones were isolated and sequenced. The clone with the longest insert (pOSTF1-1, 1,251 bp) contained a 5′-truncated cDNA. As the sequence is highly homologous with ATML1, the cDNA was named as OSTF1 (O. sativa transcription factor 1).

To obtain the full-length cDNA, 5′-rapid amplification of cDNA ends (RACE) was subsequently performed but failed. Considering that the HDs in HD-GL2 proteins are usually N-terminal-located, degenerate primer (5′-CA(AG)GT(AGCT)AA(AG)TT(CT)TGGTT(CT)CA(AG) AA-(CT)-3′) was designed according to the conserved helix III sequence QVKFWFQN in the HD of ATML1. RT-PCR was performed using the degenerate primers and the reverse primer for EST C74009 (see above). Using cDNAs from young grains at 1–5 DAP as templates, a ~1.7 kb fragment was successfully amplified and subcloned (pOSTF1-2). This fragment was determined to be a part of a homeobox cDNA. The 5′-end of the cDNA was further obtained by 5′-RACE using the SMART™ RACE cDNA amplification kit (Clontech, Palo Alto, CA, U.S.A.) and a gene-specific reverse primer (5′-GACTTGCGGAGCTGTCTTGTTGT-3′, bases 1,443–1,421 on the reverse strand of OSTF1 cDNA, Fig. 1). The amplified ~1.4 kb fragment was subcloned (pOSTF1-3). The three clones obtained from cDNA library screening (pOSTF1-1), RT-PCR (pOSTF1-2), and 5′-RACE (pOSTF1-3), respectively, containing the 5′-, middle- and 3′-region of OSTF1 were assembled to yield pOSTF1-5 with a 5′-truncated cDNA of OSTF1 and pOSTF1-6 containing a full-length version.

The cognate genomic sequence of OSTF1 was obtained by searching the databank (http://www.rice-research.org) using the BLASTN program (Altschul et al. 1997). One BAC clone, OSM131619, was found to contain cDNA of OSTF1 between residues 8680 and 14705. The exon/intron boundaries were determined by aligning the OSM131619 and OSTF1 cDNA sequences with manual refinement.

DNA sequencing

All plasmid DNA were purified by the Nucleospin Miniprep kit (Clontech) from LB broth cultures (5 ml) containing ampicillin (50 µg ml–1) incubated at 37°C overnight. DNA sequencing analysis was conducted using the ABI PRISM377 DNA sequencer (Applied Biosystems, Foster City, CA, U.S.A.).

Phylogenetic analysis

HDs constituted by ~60 amino acids were obtained from the GenBank databases and were aligned using the PILEUP program in the Genetics Computer Group package (GCG, v.10.2). Phylograms were generated with the PAUPsearch software in GCG using the Kimura algorithm for distance matrix and Neighbor-joining algorithm for tree reconstruction with bootstrap analysis. The phylogenetic tree was displayed using the TreeView software (Page 1996).

DNA and RNA gel blot analyses

Genomic DNA from rice was isolated by minipreparation (McCarty 1986). DNA was digested with different restriction enzymes at 37°C overnight. After resolving the restriction fragments on a 0.8% agarose gel with TBE, the gel was denatured in 0.5 M NaOH and 1.5 M NaCl for 40 min and neutralized with 1.0 M Tris-HCl (pH 7.4) and 1.5 M NaCl for 40 min. The sample was then transferred onto a nylon membrane (Micron Separations, Westborough, MA, U.S.A.) by capillary blotting in 20× SSC. Total RNA was isolated from various tissues by the guanidine isothiocyanate procedure (Trizol® reagent, Gibco-BRL, Rockville, MD, U.S.A.). Transcript size in the Northern blot analysis was estimated using an RNA marker (Novagen, Madison, WI, U.S.A.) loaded on the same gel. Gel-eluted EcoRI–BamHI DNA fragment of pOSTF1-5 (bases 415–975 on OSTF1 cDNA, Fig. 1) excluding most of the homeobox region was labeled with 32P by the random-priming method (Rediprime™ II, Amersham Pharmacia Biotech, NJ, U.S.A.). For the rice actin probe, a cDNA fragment was amplified by RT-PCR using primers 5′-GGTAGAAGATGGCTGACGCC-3′ and 5′-TTAGAAGCATTTCCTGTGC-3′ designed according to the RAc1 gene (GenBank accession no. X16280). Prehybridization of DNA or RNA blots was performed in 7% SDS, 0.25 M Na2HPO4 (pH 7.2) at 65°C for 2 h. Hybridization was carried out in the same buffer (20 ml) with the isotope-labeled probe at 65°C for 18 h. Autoradiographs were taken after washing the membranes twice in 5% SDS, 20 mM Na2HPO4 (pH 7.2) at 65°C for 40 min and twice in 1% SDS and 20 mM Na2HPO4 (pH 7.2) at 65°C for 30 min.

RNA in situ hybridization

Restriction fragments BamHI–SacI and PstI–EcoRI corresponding to bases 975–1,328 and 1,844–2,118 of OSTF1 cDNA (Fig. 1) were subcloned from pOSTF1-2 into pSPT18 (Roche, Mannheim, Germany) to generate pOSTF1-7 and pOSTF1-8, respectively. Riboprobes were in vitro transcribed by T7 (antisense) or T3 (sense) RNA polymerase using linearized pOSTF1-7 and pOSTF1-8 templates and digoxigenin-11-UTP (Roche).

Tissues for in situ hybridization were fixed with 3% paraformaldehyde and 0.25% glutarldehyde in 100 mM Na-phosphate (pH 7.0) and 130 mM NaCl buffer and then embedded in paraffin wax (Schwarzacher and Heslop-Harrison 2000). Ribbons of paraffin sections (7 µm thick) were affixed to slide coated with Vectabond (Vector Labs, Burlingame, CA, U.S.A.). Serial sections from continuous samples were analyzed with sense or antisense probes. In situ hybridization and signal detection were performed according to Coen et al. (1990). Products of hybridization were detected with a DIG Nucleic Acid Detection Kit (Roche). The hybridization signal viewed under light-field microscope is bluish purple.

Acknowledgments

This work was supported by the National Science Council, Republic of China (Grants NSC 87-2611-B-005-005-B29 and NSC 88-2311-B-005-010). We are grateful to Dr. Johnson Wang for providing the rice materials and his technical assistance, and to Dr. C. C. Hu who helped in the phylogenetic analysis. We thank the Pharmacia rice-research.org program for providing information on the rice genomic sequence.

3

Present address: U-VISION Biotechnology Incorporation, 132 Lane 235, Pao-Chiao Road, Shin-Tien City, Taipei, 231 Taiwan.

4

Corresponding author: E-mail, wmleu@nchu.edu.tw; Fax, +886-4-2285-6791.

Fig. 1 Nucleotide (upper) and deduced amino acid sequence (lower) of OSTF1 cDNA. Numbers in the right and left tabulate nucleotides and amino acids sequence, respectively. The HD, truncated leucine zipper motif, and START domain are shown on gray background and indicated by (a), (b), and (c), respectively. The acidic amino acids preceding the HD are shown in italic. The presumable polyadenylation signal is underlined. The arrowheads indicate the position of introns, with size from I–XI is 2100, 723, 83, 299, 124, 126, 102, 158, 90, 94, and 107 bp, sequentially.

Fig. 1 Nucleotide (upper) and deduced amino acid sequence (lower) of OSTF1 cDNA. Numbers in the right and left tabulate nucleotides and amino acids sequence, respectively. The HD, truncated leucine zipper motif, and START domain are shown on gray background and indicated by (a), (b), and (c), respectively. The acidic amino acids preceding the HD are shown in italic. The presumable polyadenylation signal is underlined. The arrowheads indicate the position of introns, with size from I–XI is 2100, 723, 83, 299, 124, 126, 102, 158, 90, 94, and 107 bp, sequentially.

Fig. 2 A neighbor-joining tree derived from the phylogenetic comparison of the OSTF1 homeodomain with other plant homeodomains. In constructing this tree, HDs constituted by ~60 amino acids were used. Numbers along the branches are bootstrap values (100 replicates). The scale bar indicates the genetic distance for 0.1 amino acid substitutions/site. GenBank accession numbers for the sequences are as follows: PRHP (L21975); HAT3.1 (X69512); ATH1 (X80126); BELL1 (U39944); KNAT3 (X92392); OSH45 (D49703/4); OSH1 (D16507); STM (U32344); Athb-8 (Z50851); REV/IFL (AF233592); Athb-9 (Y10922); Athb-14 (Y11122); HAT9 (U09341); HAT4 (M90394); Athb-4 (Y09582); Oshox1 (X96681); Athb-5 (X67033); Athb-7 (X67032); Athb-3 (X62644); Athb-1 (X58821); OSTF1 (AF317882); ZMOCL2 (AJ250984); GL2 (L32873); ZMOCL1 (Y17898); ZMOCL3 (AJ250985); ANL2 (AF077335); ZMOCL4 (AJ250986); ATML1 (U37589); ZMOCL5 (AJ250987); O39 (U34743).

Fig. 2 A neighbor-joining tree derived from the phylogenetic comparison of the OSTF1 homeodomain with other plant homeodomains. In constructing this tree, HDs constituted by ~60 amino acids were used. Numbers along the branches are bootstrap values (100 replicates). The scale bar indicates the genetic distance for 0.1 amino acid substitutions/site. GenBank accession numbers for the sequences are as follows: PRHP (L21975); HAT3.1 (X69512); ATH1 (X80126); BELL1 (U39944); KNAT3 (X92392); OSH45 (D49703/4); OSH1 (D16507); STM (U32344); Athb-8 (Z50851); REV/IFL (AF233592); Athb-9 (Y10922); Athb-14 (Y11122); HAT9 (U09341); HAT4 (M90394); Athb-4 (Y09582); Oshox1 (X96681); Athb-5 (X67033); Athb-7 (X67032); Athb-3 (X62644); Athb-1 (X58821); OSTF1 (AF317882); ZMOCL2 (AJ250984); GL2 (L32873); ZMOCL1 (Y17898); ZMOCL3 (AJ250985); ANL2 (AF077335); ZMOCL4 (AJ250986); ATML1 (U37589); ZMOCL5 (AJ250987); O39 (U34743).

Fig. 3 The structure of HD-GL2 protein and sequence comparison within the HD-GL2 family. (A) Schematic diagram illustrating the domains conserved among the HD-GL2 members. (B) Alignment of the sequence in HD, truncated leucine zipper motif, and START domain of OSTF1 with the corresponding sequences of Phalaenopsis O39, Arabidopsis ATML1, ANL2 and GL2. The alignment was performed by the Clustal W method (v.1.7) (Thompson et al. 1994) with PAM 250 residue weight table. All matched residues are shaded with solid black; conserve and similar residues are shaded with dark-gray and light-gray color, respectively. A consensus sequence from invariant and similar amino acids is shown in upper and lower cases, respectively. In the HD consensus sequence, plus symbols represent basic amino acids, and stars indicate invariant amino acid residues in all HDs. In the truncated leucine zipper motif, the amino acid residues within each heptad repeat are labeled a–g, in which the d-residues are usually a leucine or hydrophobic amino acids. The a-residues facing the same side on α-helices are usually hydrophobic amino acids. Residues “e” and “g”, which border the hydrophobic core, frequently are charged amino acids (Branden and Tooze 1998). Stars on the consensus sequence mark two conserved cysteine residues found in the loop region. In the START domain, a cluster of hydrophobic residues proposed to be part of the lipid-binding site is underlined (Moonen et al. 1979).

Fig. 3 The structure of HD-GL2 protein and sequence comparison within the HD-GL2 family. (A) Schematic diagram illustrating the domains conserved among the HD-GL2 members. (B) Alignment of the sequence in HD, truncated leucine zipper motif, and START domain of OSTF1 with the corresponding sequences of Phalaenopsis O39, Arabidopsis ATML1, ANL2 and GL2. The alignment was performed by the Clustal W method (v.1.7) (Thompson et al. 1994) with PAM 250 residue weight table. All matched residues are shaded with solid black; conserve and similar residues are shaded with dark-gray and light-gray color, respectively. A consensus sequence from invariant and similar amino acids is shown in upper and lower cases, respectively. In the HD consensus sequence, plus symbols represent basic amino acids, and stars indicate invariant amino acid residues in all HDs. In the truncated leucine zipper motif, the amino acid residues within each heptad repeat are labeled a–g, in which the d-residues are usually a leucine or hydrophobic amino acids. The a-residues facing the same side on α-helices are usually hydrophobic amino acids. Residues “e” and “g”, which border the hydrophobic core, frequently are charged amino acids (Branden and Tooze 1998). Stars on the consensus sequence mark two conserved cysteine residues found in the loop region. In the START domain, a cluster of hydrophobic residues proposed to be part of the lipid-binding site is underlined (Moonen et al. 1979).

Fig. 4 DNA gel blot analysis of the OSTF1 gene. Each lane was loaded with 10 µg of the genomic DNA from rice digested by the indicated restriction enzyme (B, BamHI; E, EcoRI; H, HindIII; X1, XbaI; X2, XhoI). The DNA markers are indicated in kb.

Fig. 4 DNA gel blot analysis of the OSTF1 gene. Each lane was loaded with 10 µg of the genomic DNA from rice digested by the indicated restriction enzyme (B, BamHI; E, EcoRI; H, HindIII; X1, XbaI; X2, XhoI). The DNA markers are indicated in kb.

Fig. 5 Detection of OSTF1 transcript by Northern blot hybridization. Rice samples were collected as described in Materials and Methods. Total RNA (20 µg) was denatured and electrophoresed in a 1% agarose gel in the presence of formaldehyde and then transferred onto a nylon membrane. (A) YP1, 1–2 cm young panicles; YP2, 3–10 cm young panicles; MP, meiosis panicles; FS, flowering spikelets; G, a mixture of grains at 1–5 DAP. X-ray films were exposed for 1 d (upper panel) or 4 d (lower panel). Photograph of the ethidium bromide-stained gel used in this experiment is provided. (B) Numbers at the top represent time points at which grains were harvested in DAP. U, upper half of grains at 1–5 DAP; L, lower half of grains at 1–5 DAP. The same blot was re-probed with a rice actin DNA as a control for the amount of RNA loaded in each lane. The actin probe hybridized to a ~1.8 kb band. The approximate transcript size was determined by the RNA molecular weight marker on the right. (C) FS, flowering spikelets; C, callus; R, roots; GS, green seedlings; ML, mature leaves; OL, old leaves; G, a mixture of grains at 1–5 DAP.

Fig. 5 Detection of OSTF1 transcript by Northern blot hybridization. Rice samples were collected as described in Materials and Methods. Total RNA (20 µg) was denatured and electrophoresed in a 1% agarose gel in the presence of formaldehyde and then transferred onto a nylon membrane. (A) YP1, 1–2 cm young panicles; YP2, 3–10 cm young panicles; MP, meiosis panicles; FS, flowering spikelets; G, a mixture of grains at 1–5 DAP. X-ray films were exposed for 1 d (upper panel) or 4 d (lower panel). Photograph of the ethidium bromide-stained gel used in this experiment is provided. (B) Numbers at the top represent time points at which grains were harvested in DAP. U, upper half of grains at 1–5 DAP; L, lower half of grains at 1–5 DAP. The same blot was re-probed with a rice actin DNA as a control for the amount of RNA loaded in each lane. The actin probe hybridized to a ~1.8 kb band. The approximate transcript size was determined by the RNA molecular weight marker on the right. (C) FS, flowering spikelets; C, callus; R, roots; GS, green seedlings; ML, mature leaves; OL, old leaves; G, a mixture of grains at 1–5 DAP.

Fig. 6 Spatial expression pattern of OSTF1 in embryo and root of rice. Median longitudinal sections of embryos at 2 (A), 3 (B), 4 (C), 6 (D) and 8 DAP (E) or root from 3-day-old seedlings (F) were probed with digoxigenin-labeled antisense strand of OSTF1 RNA. Bluish purple staining corresponds to hybridization signals. The integument and prealeurone layer are indicated with arrow and arrowhead, respectively. Cc, central cylinder; Co, coleoptile; Ec, epidermis and cortex; Ep, epiblast; Em, embryo; En, endosperm; Lp, leaf primordia; Nu, nucellus; Pe, pericarp; Rh, root hair; Sa, shoot apex; Sc, scutellum. Bar = 100 µm.

Fig. 6 Spatial expression pattern of OSTF1 in embryo and root of rice. Median longitudinal sections of embryos at 2 (A), 3 (B), 4 (C), 6 (D) and 8 DAP (E) or root from 3-day-old seedlings (F) were probed with digoxigenin-labeled antisense strand of OSTF1 RNA. Bluish purple staining corresponds to hybridization signals. The integument and prealeurone layer are indicated with arrow and arrowhead, respectively. Cc, central cylinder; Co, coleoptile; Ec, epidermis and cortex; Ep, epiblast; Em, embryo; En, endosperm; Lp, leaf primordia; Nu, nucellus; Pe, pericarp; Rh, root hair; Sa, shoot apex; Sc, scutellum. Bar = 100 µm.

The nucleotide sequence reported in this paper has been submitted to GenBank under accession number AF317882.

References

Abe, M., Takahashi, T. and Komeda, Y. (
2001
) Identification of a cis-regulatory element for L1 layer-specific gene expression, which is targeted by an L1-specific homeodomain protein.
Plant J.
 
26
:
487
–494.
Altschul, S.F., Madden, T.L., Schaffer, A.A., Zhang, J., Zhang, Z., Miller, W. and Lipman, D.J. (
1997
) Gapped BLAST and PSI-BLAST: a new generation of protein database search programs.
Nucl. Acids Res.
 
25
:
3389
–3402.
Aso, K., Kato, M., Banks, J.A. and Hasebe, M. (
1999
) Characterization of homeodomain-leucine zipper genes in the fern Ceratopteris richardii and the evolution of the homeodomain-leucine zipper gene family in vascular plants.
Mol. Biol. Evol.
 
16
:
544
–552.
Baima, S., Nobili, F., Sessa, G., Lucchetti, S., Ruberti, I. and Morelli, G. (
1995
) The expression of the Athb-8 homeobox gene is restricted to provascular cells in Arabidopsis thaliana.
Development
 
121
:
4171
–4182.
Branden, C. and Tooze, J. (
1998
) Specific transcription factors belong to a few families. In Introduction to Protein Structure, 2nd edition. pp.
175
–203. Garland Publishing, New York, NY.
Chan, R.L., Gago, G.M., Palena, C.M. and Gonzalez, D.H. (
1998
) Homeoboxes in plant development.
Biochim. Biophys. Acta
 
1442
:
1
–19.
Coen, E.S., Romero, J.M., Doyle, S., Elliott, R., Murphy, G. and Carpenter, R. (
1990
) Floricaula: a homeotic gene required for flower development in Antirrhinum majus.
Cell
 
63
:
1311
–1322.
Di Cristina, M., Sessa, G., Dolan, L., Linstead, P., Baima, S., Ruberti, I. and Morelli, G. (
1996
) The Arabidopsis Athb-10 (GLABRA2) is an HD-Zip protein required for regulation of root hair development.
Plant J.
 
10
:
393
–402.
Evans, R.M. (
1988
) The steroid and thyroid hormone receptor superfamily.
Science
 
240
:
889
–895.
Gehring, W.J., Affolter, M. and Burglin, T. (
1994
) Homeodomain proteins.
Annu. Rev. Biochem.
 
63
:
487
–526.
Gehring, W.J., Qian, Y.Q., Billeter, M., Furukubo-Tokunaga, K., Schier, A.F., Resendez-Perez, D., Affolter, M., Otting, G. and Wuthrich, K. (
1994
) Homeodomain-DNA recognition.
Cell
 
78
:
211
–223.
Genetics Computer Group (GCG), Wisconsin Package Version 10.2, Madison, WI.
Hicks, G.R., Smith, H.M., Shieh, M. and Raikhel, N.V. (
1995
) Three classes of nuclear import signals bind to plant nuclei.
Plant Physiol.
 
107
:
1055
–1058.
Ingham, P.W. (
1988
) The molecular genetics of embryonic pattern formation in Drosophila.
Nature
 
335
:
25
–34.
Ingouff, M., Farbos, I., Lagercrantz, U. and von Arnold, S. (
2001
) PAHB1 is an evolutionary conserved HD-GL2 homeobox gene expressed in the protoderm during Norway spruce embryo development.
Genesis
 
30
:
220
–230.
Ingram, G.C., Boisnard-Lorig, C., Dumas, C. and Rogowsky, P.M. (
2000
) Expression patterns of genes encoding HD-ZipIV homeodomain proteins define specific domains in maize embryos and meristems.
Plant J.
 
22
:
401
–414.
Ingram, G.C., Magnard, J.L., Vergne, P., Dumas, C. and Rogowsky, P.M. (
1999
) ZmOCL1, an HDGL2 family homeobox gene, is expressed in the outer cell layer throughout maize development.
Plant Mol. Biol.
 
40
:
343
–354.
Kallen, C.B., Billheimer, J.T., Summers, S.A., Stayrook, S.E., Lewis, M. and Strauss, J.F., 3rd (
1998
) Steroidogenic acute regulatory protein (StAR) is a sterol transfer protein.
J. Biol. Chem.
 
273
:
26285
–26288.
Kubo, H., Peeters, A.J., Aarts, M.G., Pereira, A. and Koornneef, M. (
1999
) ANTHOCYANINLESS2, a homeobox gene affecting anthocyanin distribution and root development in Arabidopsis.
Plant Cell
 
11
:
1217
–1226.
Lawrence, P.A. and Morata, G. (
1994
) Homeobox genes: their function in Drosophila segmentation and pattern formation.
Cell
 
78
:
181
–189.
Lin, Y. and Schiefelbein, J. (
2001
) Embryonic control of epidermal cell patterning in the root and hypocotyl of Arabidopsis.
Development
 
128
:
3697
–3705.
Liu, L., White, M.J. and MacRae, T.H. (
1999
) Transcription factors and their genes in higher plants functional domains, evolution and regulation.
Eur. J. Biochem.
 
262
:
247
–257.
Lu, P., Porat, R., Nadeau, J.A. and O’Neill, S.D. (
1996
) Identification of a meristem L1 layer-specific gene in Arabidopsis that is expressed during embryonic pattern formation and defines a new class of homeobox genes.
Plant Cell
 
8
:
2155
–2168.
McCarty, E.R. (
1986
) A simple method for extraction of DNA from maize tissue.
Maize Genet. Coop. Newsl.
 
60
:
61
.
Moonen, P., Haagsman, H.P., Van Deenen, L.L. and Wirtz, K.W. (
1979
) Determination of the hydrophobic binding site of phosphatidylcholine exchange protein with photosensitive phosphatidylcholine.
Eur. J. Biochem.
 
99
:
439
–445.
Natesh, S. and Rau, M.A. (
1984
) Embryology of angiosperms. In The Embryo. Edited by Johri, B.M. pp.
377
–443. Springer-Verlag, Berlin.
Otsuga, D., DeGuzman, B., Prigge, M.J., Drews, G.N. and Clark, S.E. (
2001
) REVOLUTA regulates meristem initiation at lateral positions.
Plant J.
 
25
:
223
–236.
Page, R.D.M. (
1996
) TREEVIEW: An application to display phylogenetic trees on personal computers.
Comput. Appl. Biosci.
 
12
:
357
–358.
Ponting, C.P. and Aravind, L. (
1999
) START: a lipid-binding domain in StAR, HD-ZIP and signalling proteins.
Trends Biochem. Sci.
 
24
:
130
–132.
Rerie, W.G., Feldmann, K.A. and Marks, M.D. (
1994
) The GLABRA2 gene encodes a homeodomain protein required for normal trichome development in Arabidopsis.
Genes Dev.
 
8
:
1388
–1399.
Sato, Y., Hong, S.K., Tagiri, A., Kitano, H., Yamamoto, N., Nagato, Y. and Matsuoka, M. (
1996
) A rice homeobox gene, OSH1, is expressed before organ differentiation in a specific region during early embryogenesis.
Proc. Natl. Acad. Sci. USA
 
93
:
8117
–8122.
Schumacher, K. and Chory, J. (
2000
) Brassinosteroid signal transduction: still casting the actors.
Curr. Opin. Plant Biol.
 
3
:
79
–84.
Schwarzacher, T. and Heslop-Harrison, P. (
2000
) Preparation of sectioned, whole mount or centrifuged material. In Practical in situ Hybridization, pp.
22
–79. BIOS Scientific Publishers Limited, Oxford.
Sessa, G., Steindler, C., Morelli, G. and Ruberti, I. (
1998
) The Arabidopsis Athb-8, -9 and -14 genes are members of a small gene family coding for highly related HD-ZIP proteins.
Plant Mol. Biol.
 
38
:
609
–622.
Sessions, A., Weigel, D. and Yanofsky, M.F. (
1999
) The Arabidopsis thaliana MERISTEM LAYER 1 promoter specifies epidermal expression in meristems and young primordia.
Plant J.
 
20
:
259
–263.
Smith, L.G., Greene, B., Veit, B. and Hake, S. (
1992
) A dominant mutation in the maize homeobox gene, Knotted-1, causes its ectopic expression in leaf cells with altered fates.
Development
 
116
:
21
–30.
Sundas-Larsson, A., Svenson, M., Liao, H. and Engstrom, P. (
1998
) A homeobox gene with potential developmental control function in the meristem of the conifer Picea abies.
Proc. Natl. Acad. Sci. USA
 
95
:
15118
–15122.
Suzuki, K., Taniguchi, T. and Maeda, E. (
1992
) Ultrastructure and cleavage pattern of rice proembryos.
Jpn. J. Crop Sci.
 
61
:
292
–303.
Tavares, R., Aubourg, S., Lecharny, A. and Kreis, M. (
2000
) Organization and structural evolution of four multigene families in Arabidopsis thaliana: AtLCAD, AtLGT, AtMYST and AtHD-GL2.
Plant Mol. Biol.
 
42
:
703
–717.
Thompson, J.D., Higgins, D.G. and Gibson, T.J. (
1994
) CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice.
Nucl. Acids Res.
 
22
:
4673
–4680.
Tsujishita, Y. and Hurley, J.H. (
2000
) Structure and lipid transport mechanism of a StAR-related domain.
Nat. Struct. Biol.
 
7
:
408
–414.
Walbot, V. (
2000
) Arabidopsis thaliana genome. A green chapter in the book of life.
Nature
 
408
:
794
–795.
Zhong, R. and Ye, Z.H. (
1999
) IFL1, a gene regulating interfascicular fiber differentiation in Arabidopsis, encodes a homeodomain-leucine zipper protein.
Plant Cell
 
11
:
2139
–2152.