GeBP, the first member of a new gene family in Arabidopsis, encodes a nuclear protein with DNA-binding activity and is regulated by KNAT1

Authors


*For correspondence (fax +33 47651 43 36; e-mail Gilles.Vachon@ujf-grenoble.fr).

Summary

Trichomes of Arabidopsis are single-celled epidermal hair that are a useful model for studying plant cell fate determination. Trichome initiation requires the activity of the GLABROUS1 (GL1) gene whose expression in epidermal and trichome cells is dependent on the presence of a 3′-cis-regulatory element. Using a one-hybrid screen, we have isolated a cDNA, which encodes for a protein, GL1 enhancer binding protein (GeBP), that binds this regulatory element in yeast and in vitro. GeBP and its three homologues in Arabidopsis share two regions: a central region with no known motifs and a C-terminal region with a putative leucine-zipper motif. We show that both regions are necessary for trans-activation in yeast. A translational fusion with the Yellow Fluorescent Protein (YFP) indicates that GeBP is a nuclear protein whose localization is restricted to, on average, 3–5 subnuclear foci that might correspond to nucleoli. Transcriptional fusion with the GUS reporter indicates that GeBP is mainly expressed in vegetative meristematic tissues and in very young leaf primordia. We looked at GeBP expression in plants mutated in or misexpressing KNAT1, a KNOX gene, expressed in the shoot apical meristem and downregulated in leaf founder cells, and found that GeBP transcript level is regulated by KNAT1 suggesting that KNAT1 is a transcriptional activator of GeBP. This regulation suggests that GeBP is acting as a repressor of leaf cell fate.

Introduction

Trichomes are hair-like structures that are present in the epidermis of the aerial parts of most terrestrial plants. In Arabidopsis thaliana, trichomes are single cells derived from protodermal cells on young leaf primordia. Trichome cells have been used to dissect pattern formation and cell morphogenesis in plants (Hülskamp et al., 1994,1999; Szymanski et al., 2000). Trichomes provide a convenient visible marker for several aspects of leaf development or leaf identity. Trichome density, for instance, correlates with leaf polarity; leaves are polarized along their adaxial (upper)–abaxial (lower) axis with the adaxial side bearing more trichomes than the abaxial side (Bowman et al., 2002; Kerstetter et al., 2001). Trichome development has also been used as a criterion for discriminating between the juvenile and adult leaves; the first two rosette leaves (juvenile leaves) of Arabidopsis do not possess trichomes on their abaxial surface (Chien and Sussex, 1996; Hamada et al., 2000; Telfer et al., 1997). The formation of trichomes in the epidermis is also a marker for organ identity. While wild-type cotyledons do not have trichomes, mutations in various genes such as LEAFY COTYLEDON1 (LEC1) or LEC2 induce trichome development on cotyledons (Meinke et al., 1994).

To ensure the proper development of leaves or other lateral organs, many genes need to be co-ordinately regulated during their development. These processes have been the focus of a number of recent studies, which indicate that lateral organ specification is concomitant with the downregulation of genes involved in meristem establishment. The meristem establishment genes include the Knotted-like (KNOX) class I genes, such as SHOOTMERISTEMLESS (STM) (Long et al., 1996), KNAT1 (Chuck et al., 1996) and KNAT2 (Pautot et al., 2001) in Arabidopsis. KNOX genes are expressed in the shoot apical meristem (SAM), but are downregulated in leaf founder cells (Jackson et al., 1994; Lincoln et al., 1994). Immature trichomes are readily visible when leaf primordia reach 100 µm and the acquisition of competence occurs early during leaf primordia initiation. From the downregulation of KNOX genes in leaf founder cells to the specific determination of trichome cell identity on leaf primordia, many processes, such as positional cues, hormonal regulation, or control of cell division, are likely to exist to influence cells to adopt a trichome fate. We have shown that hormonal control is involved in the determination of trichome fate. Gibberellin hormones (GAs) upregulate the transcription of GLABROUS1 (GL1), an MYB gene whose function is required for trichome initiation (Perazza et al., 1998). Null mutants of GL1 do not have trichomes on any plant organs. In the GA-deficient mutant ga1-3, the GL1 transcript level is dramatically reduced and rosette leaves are essentially glabrous (Perazza et al., 1998). Application of exogenous GAs restores both GL1 expression and trichome development on newly formed leaves. The molecular mechanisms of GA action on GL1 expression are not yet known.

Whereas GL1 specifies the production of trichomes, WEREWOLF (WER), an MYB gene paralogous to GL1, establishes the development of hairless cells in root epidermis and non-stomatal cells in the hypocotyl epidermis. Recently, Lee and Schiefelbein (2001) have shown that GL1 and WER specificities originate from divergence of their regulatory sequences rather than from their coding sequences (Lee and Schiefelbein, 2001). A 497-bp 3′-cis-regulatory element of the GL1 gene whose presence is required for GL1 expression has previously been characterized (Larkin et al., 1993). This element allows GUS reporter gene expression in epidermis and trichome cells to occur, while the GL1 promoter alone does not (Oppenheimer et al., 1991).This element is referred to as the GL1 enhancer, since its action is independent of its orientation and location. A central 152-bp region of this enhancerhas been inferred as being essential for its activity (Larkin et al., 1993). This enhancer must, at some point, be the target of some of the processes occurring during leaf development to allow trichome determination.

In order to better understand trichome fate determination, we have taken a molecular approach to identify regulatory proteins that could contribute to GL1 regulation through its enhancer. This study describes the isolation of one such potential regulatory protein, GL1 enhancer binding protein (GeBP), a DNA-binding protein which is the first member of a new family in Arabidopsis. We characterized its broad functional regions in yeast and its cellular localization in plant cells. The role of GeBPin planta was addressed by generating GUS reporter lines, misexpressing GeBP in transgenic plants and isolating a transposon-tagged mutant. Finally, we show that GeBP is regulated by KNAT1, suggesting a role for GeBP in the determination of leaf cell fate.

Results

Isolation of a GL1 enhancer binding protein (GeBP)

In their paper on the characterization of the enhancer, Larkin et al. (1993) have shown that at least a portion of the essential 3′ sequences must lie between positions +2589 and +2741 of the GL1 locus. This 152-bp central part of the GL1 enhancer was cloned upstream of a HIS3 reporter gene cassette and the resulting construct was inserted into the yeast genome (Figure 1a). This yeast strain was then transformed with an Arabidopsis rosette cDNA library fused to the Gal4 activation domain (Gal4AD). Out of 7 × 106 transformed clones, five colonies showed growth on selective medium (−histidine), and were confirmed as true positive clones after a second round of yeast transformation with the five isolated plasmids. All five plasmids contained the same cDNA. The sequence of this cDNA was identical to accession # AF361590 except that 15 nucleotides were missing in the 5′ untranslated region. The longest open reading frame detected is 906-bp long and encodes a protein of 302 residues, referred to as GeBP. As shown in Figure 1(b), the Gal4AD::GeBP fusion protein could also trans-activate the lacZ reporter gene in yeast and induce a 15-fold increase in β-galactosidase activity compared to the control vector. The activation by Gal4AD::GeBP is dependent on the presence of the GL1 enhancer, since no or only a slight increase only in β-galactosidase activity was observed when the enhancer was omitted (data not shown). Finally, GeBP alone with no fused Gal4AD failed to activate lacZ (Figure 1b). This result indicates that in yeast GeBP does not act as a transcriptional activator by itself. To determine whether GeBP could bind specifically to the GL1 enhancer, a gel-shift assay was performed with purified recombinant GeBP protein. A GeBP-enhancer complex was observed in our experimental conditions (Figure 1c). The retarded signal was competed away with an excess of unlabeled enhancer but not with an excess of unrelated fragment. Taken together, these data demonstrate that GeBP is a DNA-binding protein, which specifically recognizes the GL1 enhancer in yeast and in vitro.

Figure 1.

DNA-binding activity of GeBP in yeast and in vitro.

(a) Schematic representation of the yeast genomic insertion used for the one-hybrid screen. The 152-bp GL1 enhancer was inserted upstream of the HIS3 gene and inserted into the HIS3 locus of the YM4271 yeast strain (Clontech) to give YM4271(p205) strain.

(b)β-galactosidase activity in yeast cells transformed with various vectors. Yeast cells YM4271(p226), carrying a genomic insertion with the GL1 enhancer upstream from the lacZ gene, were transformed with the vectors indicated, grown in liquid culture, permeabilized and incubated with Galacton-Star chemiluminescent substrate (Tropix). Units are expressed as Relative Light Units as given by the luminometer (Turner Designs TD-20/20). Similar trans-activation levels were seen with either HIS3 or lacZ when the whole GL1 enhancer (497 bp) was used instead of its central part (data not shown).

(c) Gel retardation with purified GeBP protein. The radiolabeled GL1 enhancer was used as a probe and incubated with no protein (lane 1), purified GeBP (lane 2), purified GeBP with increasing amounts of unlabeled enhancer (lane 3 and 4), purified GeBP with increasing amounts of unlabeled minimum lacZ promoter PCYC1 (lane 5 and 6). The open arrow indicates the specific GeBP-DNA complex. FP, free probe.

GeBP belongs to a new protein family

A search for protein motifs in the GeBP primary sequence revealed the presence of two bi-partite NLS suggesting a nuclear localization of GeBP (Figure 2a). Two partial motifs, one for a basic domain (75% similarity) and one for a leucine zipper (75% similarity) were also noticed. The arrangement and proximity of these two putative motifs suggested that GeBP could belong to the bZIP transcription factor family. However, the distance between these two motifs was longer than nine residues and therefore did not fulfill the criteria defining bZIP proteins (Jakobi et al., 2002). Sequence comparisons with databases revealed that three proteins in the Arabidopsis genome share homologies with GeBP (Figure 2b). Two of these three proteins have significant C-terminal extensions compared to GeBP. Two regions of the GeBP protein are particularly conserved among these three homologues: a central (97 residues) and a C-terminal (79 residues) region. The central region of GeBP is also found in 15 other proteins of Arabidopsis, all with unknown function (Figure 2c). Two proteins in rice and one protein in Solanum tuberosum also have this region. The S. tuberosum protein, STORE KEEPER (STK) (accession # AJ314612.1), is a putative transcriptional regulator (see Discussion). The C-terminal region of GeBP, which harbours the putative leucine zipper, is also found in three other proteins of Arabidopsis, all with unknown functions (Figure 2d). A phylogenetic tree of the 19 proteins described in databases with homologies to the central region of GeBP shows that GeBP and its three similar proteins form a distinct group (Figure 2e). No homologous proteins were identified outside the plant kingdom.

Figure 2.

Figure 2.

Amino acid sequence of GeBP and proteins similar to GeBP in plants.

(a) Amino acid sequence of GeBP with putative signatures and domains. The two NLS are underlined. The central region is indicated with boxed letters and the C-terminal region is indicated with italic letters. The putative basic domain residues are indicated with bold letters (Prosite accession number for basic domain signature: PS00036) and the putative leucine-zipper residues with stars (Prosite accession number for leucine-zipper signature: PS00029). Two F residues further downstream were also considered as part of the putative leucine zipper like in the TRAB1 protein (Hobo et al., 1999). The central and C-terminal regions partially overlap domains PD210513 and PD114488 respectively in the protein domain databank Prodom (http://prodes.toulouse.inra.fr/prodom/doc/prodom.html).

(b) Schematic representation of the GeBP protein and its similar proteins in Arabidopsis. Portion of the various proteins similar to the GeBP protein regions are indicated with corresponding identities. At2g25650 is highly similar to GeBP while At2g36340 and At5g14280 have C-terminal extensions.

(c) Alignment of sequences similar to the central region of GeBP.

(d) Alignment of protein sequences similar to the C-terminal region of GeBP. Putative leucine-zipper residues are indicated with stars. Dark shaded letters indicate identical residues. Grey shaded letters indicate similar residues. Rice accessions are underlined. Alignment was done with ClustalW (http://dot.imgen.bcm.tmc.edu:9331/multi-align/multi-align.html) and the Boxshade software (http://www.isrec.isb-sib.ch:8080/software/BOX_form.html) with a fraction of sequences that must agree for shading of 0.5.

(e) Phylogeny tree of proteins sharing homology with the central region of GeBP. Bootstrap values represent the number of occurrences of a knot among 100 alignments. The PHYLIP 2.0 construction programs (http://sdmc.krdl.org.sg:8080/˜lxzhang/phylip/) were used to generate the tree using the parsimony method.

Figure 2.

Figure 2.

Amino acid sequence of GeBP and proteins similar to GeBP in plants.

(a) Amino acid sequence of GeBP with putative signatures and domains. The two NLS are underlined. The central region is indicated with boxed letters and the C-terminal region is indicated with italic letters. The putative basic domain residues are indicated with bold letters (Prosite accession number for basic domain signature: PS00036) and the putative leucine-zipper residues with stars (Prosite accession number for leucine-zipper signature: PS00029). Two F residues further downstream were also considered as part of the putative leucine zipper like in the TRAB1 protein (Hobo et al., 1999). The central and C-terminal regions partially overlap domains PD210513 and PD114488 respectively in the protein domain databank Prodom (http://prodes.toulouse.inra.fr/prodom/doc/prodom.html).

(b) Schematic representation of the GeBP protein and its similar proteins in Arabidopsis. Portion of the various proteins similar to the GeBP protein regions are indicated with corresponding identities. At2g25650 is highly similar to GeBP while At2g36340 and At5g14280 have C-terminal extensions.

(c) Alignment of sequences similar to the central region of GeBP.

(d) Alignment of protein sequences similar to the C-terminal region of GeBP. Putative leucine-zipper residues are indicated with stars. Dark shaded letters indicate identical residues. Grey shaded letters indicate similar residues. Rice accessions are underlined. Alignment was done with ClustalW (http://dot.imgen.bcm.tmc.edu:9331/multi-align/multi-align.html) and the Boxshade software (http://www.isrec.isb-sib.ch:8080/software/BOX_form.html) with a fraction of sequences that must agree for shading of 0.5.

(e) Phylogeny tree of proteins sharing homology with the central region of GeBP. Bootstrap values represent the number of occurrences of a knot among 100 alignments. The PHYLIP 2.0 construction programs (http://sdmc.krdl.org.sg:8080/˜lxzhang/phylip/) were used to generate the tree using the parsimony method.

Figure 2.

Figure 2.

Amino acid sequence of GeBP and proteins similar to GeBP in plants.

(a) Amino acid sequence of GeBP with putative signatures and domains. The two NLS are underlined. The central region is indicated with boxed letters and the C-terminal region is indicated with italic letters. The putative basic domain residues are indicated with bold letters (Prosite accession number for basic domain signature: PS00036) and the putative leucine-zipper residues with stars (Prosite accession number for leucine-zipper signature: PS00029). Two F residues further downstream were also considered as part of the putative leucine zipper like in the TRAB1 protein (Hobo et al., 1999). The central and C-terminal regions partially overlap domains PD210513 and PD114488 respectively in the protein domain databank Prodom (http://prodes.toulouse.inra.fr/prodom/doc/prodom.html).

(b) Schematic representation of the GeBP protein and its similar proteins in Arabidopsis. Portion of the various proteins similar to the GeBP protein regions are indicated with corresponding identities. At2g25650 is highly similar to GeBP while At2g36340 and At5g14280 have C-terminal extensions.

(c) Alignment of sequences similar to the central region of GeBP.

(d) Alignment of protein sequences similar to the C-terminal region of GeBP. Putative leucine-zipper residues are indicated with stars. Dark shaded letters indicate identical residues. Grey shaded letters indicate similar residues. Rice accessions are underlined. Alignment was done with ClustalW (http://dot.imgen.bcm.tmc.edu:9331/multi-align/multi-align.html) and the Boxshade software (http://www.isrec.isb-sib.ch:8080/software/BOX_form.html) with a fraction of sequences that must agree for shading of 0.5.

(e) Phylogeny tree of proteins sharing homology with the central region of GeBP. Bootstrap values represent the number of occurrences of a knot among 100 alignments. The PHYLIP 2.0 construction programs (http://sdmc.krdl.org.sg:8080/˜lxzhang/phylip/) were used to generate the tree using the parsimony method.

The central and C-terminal regions are required for GeBP activity in yeast

We made a series of deletions in the GeBP protein to determine which regions are essential for trans-activation in yeast. The central region, the putative basic domain of the bZip motif and the C-terminal region were each deleted. The various truncated GeBP cDNAs were fused to Gal4AD coding sequence, introduced back into the His3 reporter yeast strain and plated on ’−histidine‘ selective medium to look at yeast growth (Figure 3). Deletion of the basic domain had no consequence on growth. This result suggests that GeBP does not function as a bZip transcription factor, since the basic domain is essential for DNA-binding in bZip proteins (Guiltinan and Miller, 1994). On the contrary, deletion of either the central or C-terminal regions totally abolished yeast growth. Hence, these two regions are necessary for GeBP activity in yeast.

Figure 3.

Functional regions of GeBP necessary for trans-activation in yeast.

Each half-plate was streaked with yeast strain YM4271(p205) transformed with either the Gal4AD::GeBP fusion vector (top halves), or the indicated Gal4AD::GeBP deletions (bottom halves). Yeast were grown for 3 days at 28°C on SD–Histine–Leucine medium (left) for selective growth and SD–Uracyl–Leucine medium (right) to show that GeBP deletions were not lethal.

GeBP is localized in the nucleus of plant cells

In order to determine the intracellular localization of GeBP, an in-frame translational fusion with the YFP protein was made. The results of transient expression experiments by Agrobacterium-mediated infiltration of Arabidopsis leaves are shown in Figure 4. Similar to the plants transformed with a nuclear localization control YFP::lexA::NLS, the cells transformed with the YFP::GeBP hybrid protein showed fluorescence only in the nucleus. This result suggests that the two NLSs present in the GeBP sequence are functional. However, unlike the YFP::lexA::NLS control, the YFP::GeBP protein was not evenly distributed inside the nucleus but restricted to subnuclear foci. Usually, two large spots, and two or three smaller spots were seen. Transient expression experiments were also done in tobacco which gives a much higher efficiency of transformation than Arabidopsis. The same fluorescence pattern was observed in tobacco cells with both the YFP::lexA::NLS and YFP::GeBP fusion constructs (Figure 4). We stained the nuclear DNA of tobacco cells with 4′–6-diamidino-2-phenylindole (DAPI) (Figure 4). In some cases, this staining showed that the YFP::GeBP fluorescence co-localized with nuclear areas that are not or weakly stained with DAPI. These areas might correspond to nucleoli. Hence, consistent with its function as a putative transcription factor, GeBP is a nuclear protein which may be localized in nucleoli.

Figure 4.

Cellular localization of GeBP.

Arabidopsis and tobacco leaves were infected with Agrobacterium previously transformed with vectors carrying either the YFP::lex A::NLS (nuclear localization control) or the YFP::GeBP fusion constructs. Fluorescence in plant cells was visualized by exposure to UV light under a microscope. About 50–100 fluorescent cells were observed in Arabidopsis and several hundreds of cells in tobacco. Magnification bars represent 5 μm.

GeBP is expressed in the apical meristem and young leaf primordia

A 1.3-kb GeBP promoter fragment including the 5′ UTR region with the putative ATG initiation codon was cloned upstream of the GUSgene. Transgenic lines were generated and offsprings were stained for β-glucuronidase activity to visualize GeBP expression profile. After germination, GUS staining was absent until the first true leaf primordia appeared (Figure 5a). GUS staining was then observed exclusively at the centre of rosettes in a region corresponding to the apical meristem (Figure 5b). The intensity of the staining was correlated with the number of rosette leaves that had emerged from the apex (Figure 5c). No staining was visible in emerging or mature leaves. Weak staining was occasionally detected in the emerging elongating primary stem and axillary buds (data not shown). No staining was detected in embryos, root, nor in inflorescence meristems and flowers (Figure 5d). Sections were made to determine whether GUS staining was restricted to the apical meristem of seedlings (Figure 5e,f). The strongest staining was observed within the apical meristem. GUS staining was also present in very young leaf primordia and disappeared rapidly as the primordia grew. This expression pattern is compatible with trichome initiation and GL1 expression pattern. Vascular bundles of young leaf were also stained although we could not determine whether this staining was due to passive diffusion from the meristem region.

Figure 5.

Expression pattern of GUS under control of the GeBP promoter.

Plants were grown on soil and were incubated for 8–10 h with GUS substrate, destained as described (Gallagher, 1992) and fixed when needed.

(a) A 4-day-old seedling with no visible leaf primordia.

(b) An 8-day-old seedling with two formed leaves and two emerging leaf primordia.

(c) A 14-day-old seedling with six mature leaves.

(d) A primary inflorescence shoot with no visible GUS staining. Magnification bars represent 1 cm.

(e) Cross-section of a 3-week-old seedling embedded in paraffin.

(f) Longitudinal section of a 3-week-old seedling embedded in paraffin. Magnification bars represent 500 μm. AM: apical meristem; LP: leaf primordia.

Thus, the expression of GeBP seems to be specific to the vegetative meristem and young leaf primordia and is correlated with the development of leaf primordia.

Loss of function and constitutive expression of GeBP induce no visible phenotypes

The GeBP cDNA was cloned downstream of the constitutive 35S promoter and transgenic lines were generated. Overexpressing lines were identified by real-time RT-PCR. Three of these lines, which overexpress GeBP 10–22-fold (illustrated in Figure 6 for line B), were chosen for further studies. The phenotype of these lines was indistinguishable from the wild type with respect to trichome development and the overall plant development.

Figure 6.

GeBP and GL1 transcript levels in 35S::GeBP and GT6658 mutant plants.

Real-time RT-PCR was performed on total RNA isolated from 3-week-old rosettes of wild-type, 35S::GeBP (Col-0) and GT6658 (Ler) plants.

Two lines of Arabidopsis with putative transposon insertions in the GeBP gene were identified in the Cold Spring Harbor collection (Sundaresan et al., 1995) (http://genetrap.cshl.org/): line ET1969 and line GT6658. However, after molecular analysis by PCR, the ET element was not found in the GeBP gene. We conclude that the ET1969 line does not contain a transposon in the GeBP locus. The precise location of the GT line insertion was determined by PCR and sequencing. The GT element is inserted at the beginning of the first exon of GeBP, interrupting the Serine27 codon (TCT), and oriented backwards relative to the GeBP gene. Real time RT-PCR from total RNA isolated from rosettes showed that GeBP transcript is indeed not detected in GT6658 (Figure 6). Thus, the GT6658 line is a likely loss-of-function mutant. The GT6658 homozygous line is like the wild type with respect to both plant and trichome development.

As GeBP is a potential regulator of GL1, the GL1 transcript level was measured in rosettes of 35S::GeBP and GT6658 lines compared to the wild type (Figure 6). Although the GL1 transcript level in the 35S::GeBP and GT6658 plants was on average slightly above and below the wild-type level, respectively (Figure 6), the variations in GL1 expression were not significant in our experimental conditions. The implication of additional factors or functional redundancy of the other Arabidopsis GeBP-like proteins is a possible hypothesis to explain this result (see Discussion).

KNAT1 upregulates GeBPtranscript level

Like GeBP, the ArabidopsisKNAT1 gene is expressed in meristematic tissues but not in mature leaves. Loss of KNAT1 meristem expression coincides, like GeBP, with floral induction (Lincoln et al., 1994). Furthermore, a KNAT1 homologue in tobacco, the NTH15 gene, directly regulates the expression of a gibberellin biosynthetic gene (Sakamoto et al., 2001), a hormone that also regulates GL1 expression (Perazza et al., 1998). These characteristics common to KNAT1 and GeBP prompted us to look at a possible regulation of GeBP by KNAT1 by real-time RT-PCR. In plants that constitutively express KNAT1 (35S::KNAT1), the GeBP transcript level was increased 18 times relative to the wild-type level (Figure 7). Conversely, in a KNAT1 knock-out line, the brevipedicellus mutant (bp) (Venglat et al., 2002), the GeBP transcript was nearly 100-fold less abundant relative to the wild-type level (Figure 7). This modification of GeBP expression in 35S::KNAT1 and bp plants had no consequences on GL1 expression (Figure 7) as observed in 35S::GeBP and GeBP mutant plants. GeBP expression was also monitored in plants overexpressing KNAT2, a close KNAT1 homologue with similar functions and expression patterns (Pautot et al., 2001). Interestingly, GeBP expression in these plants was identical to the wild type (data not shown).

Figure 7.

Regulation of GeBP transcript level in 35S::KNAT1 and bp mutant plants.

Real-time RT-PCR was performed on total RNA isolated from 3-week-old rosettes of wild-type, 35S::KNAT1 (No) and bp mutant (Col-0) plants.

Thus, up- and downregulation of KNAT1 expression led to up- and downregulation of GeBP transcript level in Arabidopsis. Taken together, these data demonstrate that GeBP transcript level is positively regulated by KNAT1 and suggest that GeBP is involved in leaf initiation.

Discussion

The results presented here describe the characterization of GeBP, the first member of a new protein family, which was isolated on the base of its affinity for the 3′-cis-regulatory element of the GL1 gene. No homologous GeBP proteins were found outside the plant kingdom. While three GeBP-like proteins are present in Arabidopsis, 15 additional Arabidopsis proteins have a region similar to the central region of GeBP. The function of any of these proteins is unknown. Therefore, the study of GeBP will help to determine the role of a large number of proteins in Arabidopsis. The fact that deletion of either the central or C-terminal region prevents trans-activation of the reporter gene in yeast implies that these two regions are involved directly or indirectly in DNA-binding activity. The main feature of the C-terminal region is the presence of a putative leucine zipper suggesting that GeBP can form homodimers. It is therefore possible that the GeBP central region represents a new DNA-binding domain in plants. The association of a DNA-binding domain and a leucine-zipper domain is found in several protein families such as the homeodomain-leucine zipper family (Sessa et al., 1997). In S. tuberosum, the STK protein sequence shares a 32% identity (55% similarity) with the GeBP central region. According to the description in this accession, STK would act as a transcriptional regulator of the PATATIN gene which reinforces the hypothesis that GeBP is a member of a new transcription factor family in Arabidopsis. PATATIN, a protein that belongs to the family of vacuolar glycoproteins, is the major protein constituent of potato tubers and displays broad esterase activity (Hirschberg et al., 2001). A dominant gain-of-function mutant of the patatin-like homologue, STURDY, in Arabidopsis shows plants with a wide range of phenotypes including stiff stem, thicker leaves and delayed growth. Histological studies have revealed an increase in cell number in the inflorescence stem of the mutant (Huang et al., 2001).

An intriguing feature of GeBP is its subnuclear localization in tobacco and Arabidopsis epidermal cells. These subnuclear foci are likely to correspond to nucleoli. Like many nuclear proteins targeted to the nucleolus (Bickmore and Sutherland, 2002), GeBP has no clear potential nucleolar targeting signal (NTS). It is likely that one or both NLS are efficient NTSs as previously shown for several proteins in animals (Jarrous et al., 1999; Scott et al., 2001). A number of recent studies have revealed that the nucleolus is not only a structure dedicated to ribosome biogenesis, but also plays a key role in developmental or regulatory processes. In animals, the oncogene protein ARF stabilizes the p53 protein inside the nucleolus by sequestering the Mdm2 protein thus preventing p53 degradation (Weber et al., 1999). In yeast, Nopp140 is a nucleolar protein that functions as a transcriptional activator of the alpha-1 acid glycoprotein gene (Miau et al., 1997). It remains to be seen if this is also true for GeBP.

Gibberellin hormones transcriptionally activate GL1. Because GeBP is a potential regulator of GL1, the possibility exists that GeBP is also regulated by GAs. Gibberellins can regulate the activity of genes either transcriptionally as in the case of GL1, or post-transcriptionally through protein degradation (Dill et al., 2001) or regulation of nuclear import (Amador et al., 2001). In our hands, the GeBP transcript level, measured by real-time RT-PCR, was not significantly affected in ga1-3 plants by the presence or absence of exogenous GAs. Also, the subnuclear localization of the YFP::GeBP fusion protein in ga1-3 cells is similar to the wild type in transient expression experiments. Altogether, these results suggest that GeBP activity is independent of GAs.

The affinity of GeBP for the GL1 enhancer in yeast and in vitro suggests that GeBP regulates, positively or negatively, the expression of GL1 in planta. However, no modification of GL1 transcript level was detected in 35S::GeBP plants. Two main hypotheses can account for this result. First, the GeBP-GL1 enhancer interaction could be an artifact in yeast and in vitro and have no functional significance in Arabidopsis. Although we cannot exclude this hypothesis,we believe that this is unlikely because no other cDNAs, with the exception of the GeBP cDNA that was isolated independently several times, were isolated during the screening indicating that conditions were sufficiently selective. A second hypothesis is that GeBP alone would not be sufficient to regulate GL1 expression. In this hypothesis, additional cell- or tissue-specific factors would be required for GeBP to efficiently trigger GL1 regulation. These additional factors could be dependent on the GA pathway for instance. We have noticed previously that variation of GL1 expression can be monitored only in plants with severe GA-deficiency such as in the ga1-3 null mutant background. For instance, changes in GL1 expression are not detectable when exogenous GAs are applied to wild-type plants (Ler) or in the spy5 GA-response mutant (J.C., G.V., unpublished results). Similarly, in the semi-dwarf GA-deficient mutants, ga4 and ga5, trichomes are present on leaves indicating that GL1 is not turned off. Therefore, we believe that GL1 regulation by GeBP can not usefully be monitored in wild-type plants, presumably because the GA pathway is a sensitive pathway.

Three genes similar to GeBP are present in the Arabidopsis genome. The absence of phenotype in the GT line could be explained by functional redundancy. Isolation of mutants corresponding to GeBP-like genes and obtaining double and triple mutants will provide more information about the role of this family in Arabidopsis.

The upregulation of GeBP transcript level by KNAT1 indicates that GeBP is a target, direct or indirect, of KNAT1. Interestingly, KNAT2, a close KNAT1 homologue, does not regulate GeBP transcript level. Recently, KNAT1 has been shown to be partially redundant with STM in regulating stem cell function which is not the case with KNAT2 (Byrne et al., 2002). Hence, GeBP could be specific of the KNAT1/STM pathway. To our knowledge, GeBP is the first downstream gene to be identified in the KNAT1 pathway. Although the KNAT1 recognition site is not known, a binding site for the tobacco KNOX gene NTH15 (TAAGTGAC) (Sakamoto et al., 2001) is present at location −167 to −163 relative to the putative transcription start site in the promoter of GeBP. A binding site (TGAGAG (G/C)T) close to the KN1 binding sequence (TGACAG (G/C)T) (Smith et al., 2002) was also noticed about 3-kb upstream. Although detailed promoter analyses are required, these results support the possibility that KNAT1 directly regulates GeBP transcription. KNAT1 is downregulated in leaf founder cells and its ectopic expression results in dramatic tissue transformations, including ectopic meristematic activity. Although a trichome development phenotype in bp or 35S::KNAT1 plants has not been reported, misexpression of the class I KNOX gene, LeT6, in the Lycopersicon esculentum mutant clausa induces the formation of fruit carrying a high number of ectopic trichomes (Avivi et al., 2000). The activation of GeBP by KNAT1 suggests that GeBP also plays a role in the repression of leaf cell fate. Its potential role in GL1 regulation suggests that GeBP could prevent specification of leaf epidermis cell fate. Gibberellin hormones could act antagonistically to GeBP on GL1 expression as they act antagonistically to NTH15 in tobacco (Sakamoto et al., 2001). In Arabidopsis, twogenes, SERRATE and PICKLE, seem to regulate cellular competence to respond to KNOX transcription factors, possibly by limiting accessibility to genes downstream in the KNOX pathway. Both SE and PKL encode chromatin remodeling factors. Interestingly, mutants of SE produce trichomes on the abaxial side of first rosette leaves in a GA-dose dependent manner, a situation that is not encountered in the wild type (Clarke et al., 1999; Prigge and Wagner, 2001). PKL, a gene involved in GA-dependent cellular differentiation (Ogas et al., 1997,1999) has also been shown to be involved in polarity establishment in Arabidopsis (Eshed et al., 1999) and regulates LEC1, a gene whose activity is required to repress trichome development in cotyledons (Ogas et al., 1999).

GeBP represents the first member of a new protein family. Our experimental data indicate that this family could constitute a new DNA-binding or transcription factor family with some members involved in leaf primordia development. Therefore, the study of the GeBP family will provide additional clues in understanding leaf development in Arabidopsis.

Experimental procedures

Plant material

Agrobacterium C58 (GV2260) (Deblaere et al., 1985) was used for stable transformation of Arabidopsis (Columbia ecotype) using the floral dip technique (Clough and Bent, 1998) or tobacco leaf cell transformation (Kapila et al., 1997). 35S::GeBP lines are in the Columbia ecotype, GT6658 mutant line is in the Landsberg erecta ecotype and 35S::KNAT1 line is in the Nossen ecotype. Seeds of wild type and mutants of Arabidopsis thaliana were planted on soil or surface sterilized and grown in Petri dishes on MSAR medium (Koncz et al., 1990). Plants were grown at 22°C under a photoperiod of 16 h of light/ 8 h of dark.

GeBP purification and gel retardation

The GeBP cDNA was isolated from pGAD10-GeBP as an EcoRI fragment and cloned into the pYESNT C vector (Invitrogen). Induction and purification of GeBP was done according to manufacturer‘s instructions on a nickel column under denaturing conditions. β-galactosidase expression was induced from pYES2/NT/lacZ vector (Invitrogen), purified in the same conditions as GeBP and used as negative control in gel-retardation assays to detect possible unspecific DNA-binding activities that could have co-purified. Gel retardation was performed as described previously (Riechmann et al., 1996).

Constructs

Deletions of the central region, the C-terminal region and the putative basic domain were done by ligation of two PCR-amplified portions of the GeBP cDNA, one upstream and one downstream of the region to be deleted. Details of constructs are available upon request.

Deletion of Gal4AD: the pGAD10-GeBP vector was digested with Asp718 (Klenow-filled) and XbaI to remove Gal4AD together with a large portion of the GeBP cDNA. The GeBP cDNA was then rebuilt by insertion of the XhoI (Klenow-filled)–XbaI fragment (also isolated from pGAD10-GeBP) to give rise to vector p248. The resulting construct keeps the initial Gal4AD SV40T-antigen NLS upstream from the GeBP coding sequence.

The35S::GeBP construct was made as follows: the GeBP cDNA was isolated from the yeast vector pGAD10-GeBP by partial BglII digest and inserted into the BamHI site of pCGN18 (Jack et al., 1992) downstream of the large 1.6-kb 35S promoter.

The PGeBP promoter was isolated by PCR (forward oligonucleotide: CCTAGGTCCTTCTTGAGGTGGTGAAGA; reverse oligonucleotide: CCATGGCAATTGCTTGCTTTAGACTT) and cloned into the TOPO-XL vector (Invitrogen). The XhoI–BamHI fragment was cloned into the SalI–BamHI sites of pBI101 (Clontech).

One-hybrid screen

The entire enhancer (497 bp) was estimated to be too large to perform the one-hybrid screen. Therefore, we have dissected the whole enhancer into three regions (left, centre and right regions) with the central region corresponding to the region defined as being essential in planta (Larkin et al., 1993). In yeast, this central region turned out to give extremely low background in a test for yeast growth on SD–His medium, which was not the case for the left and right fragments that allowed yeast growth even at high 3-AT concentrations. The central region of the GL1 enhancer was PCR-amplified using the forward primer 5′-GCGGCCGCAGATCTGTTGACACGTCTATTTATATTAAC-3′ and the reverse primer 5′-TCTAGAGGATCCGCGCAAAGATTACTACTGTTTTG-3′. The PCR fragment was cloned into the pGEM-T vector (Promega), sequenced and sub-cloned as a BglII (Klenow filled)–XbaI fragment between the SmaI and XbaI sites of the pHISi vector to give rise to vector p205. The GL1 enhancer was then isolated as an EcoRI–BamHI (Klenow filled) fragment from p205 and cloned into the placZi vector between EcoRI and SmaI sites (Clontech) to give rise to p226. Genomic insertions of p205 and p226 into yeast strain YM4271 was done according to the Matchmaker protocol and gave rise to strain YM4271(p205) and YM4271(p226), respectively. For the one-hybrid system (Clontech), screening was performed on SD–histidine medium supplemented with 1–3 mM 3-AT using the Arabidopsis thalianaMATCHMAKER cDNA library (Clontech) from 3-week-old rosettes.

Nuclear localization

A BglII fragment was isolated from vector pGAD-GeBP and cloned into the BglII site downstream of the YFP gene (Kato et al., 2002). The resulting fusion does not include the first 25 amino acids of GeBP. The GeBP::YFP fusion gene was cloned as an XbaI fragment into the XbaI site of the plant vector pEL103 dowstream of the 35S promoter (Kato et al., 2002). Tobacco leaves were transformed as described therein (Kato et al., 2002).

Real-time RT-PCR

Total RNA was isolated from 3-week-old rosettes grown in Petri dishes using the Rneasy plant kit (Qiagen). Reverse transcription was performed using random hexamers and reverse transcriptase (Applied Biosystem) and PCR reactions were performed according to the manufacturer on the ABI 5700-SDS (Applied Biosystem). Relative transcript levels of GL1 and GeBP were calculated with the ΔΔCt method using the adenine phosphoribosyltransferase (APT) gene as the reference gene (Cowling et al., 1998). Oligonucleotides in the APT gene (forward: 5′-TGCAATCCGACTACTTGAACGA-3′; reverse: 5′-CAAGCACATTCAACAATCTTCACTC-3′), GL1 gene (forward: 5′-AGCTCCTCGGCAATAGATGGT-3′; reverse: 5′-TGTGGCGGCAGTGATGAA-3′) and the GeBP gene (forward: 5′-CATCGAAGATCTCCACGGAATAT-3′; reverse: 5′-TGAATCCTGTCTTAGCTCTGTAATCAAC-3′) were designed with the OligoExpress 1.5 software (Perkin Elmer). GL1 transcript levels were similar in all ecotypes used (No, Col-0 and Ler). GeBP transcript levels were similar in all ecotypes used except No, which showed a 3–5-fold reduction in GeBP expression relative to Ler or Col-0.

Acknowledgements

We thank Dr Jean-Gabriel Valay (University J. Fourier, Grenoble) for providing the amplified rosette cDNA library and his advices on screening yeast. We are grateful to Dr Dominique Pontier (University J. Fourier, Grenoble) for the gift of pYFP::lexA::NLS and pEL103 vectors. We thank Leïla Zekraoui (University J. Fourier, CNRS, Grenoble) for GeBP purification. We are grateful to the Cold Spring Harbor Arabidopsis Genetrap program for providing GT and ET transposon lines. We are grateful to Dr Datla (Plant Biotechnology Institute, Saskatoon, Canada) for the gift of bp mutant seeds, Dr Sarah Hake (Plant Gene Expression Centre, Albany, CA) for providing 35S::KNAT1 seeds and Véronique Pautot (Institut National de la Recherche Agronomique, Versailles) for 35S::KNAT2::GR seeds. We would like to thank Dr Jean-Marc Bonneville and Dr Daniel Perazza for useful discussions and critical reading of the manuscript and Dr Pierre Carol for critical reading of the manuscript. We thank Nicole Potier, Jean-Pierre Alcaraz and Eliane Charpentier for their technical assistance. We thank Jean-Claude Caissard (University of Saint-Etienne, France) for mentioning the work on Clausa.

Ancillary