BRL1 and BRL3 are novel brassinosteroid receptors that function in vascular differentiation in Arabidopsis

Brassinosteroids (BRs) are plant steroid hormones crucial for cell growth and plant morphogenesis. They were first identified by their ability to elongate plant stems when applied exogenously(Grove et al., 1979; Mandava, 1988). Since then, Arabidopsis genetics has permitted the identification of a wealth of BR synthesis and signal transduction components. The phenotypic characterization of mutants affecting BR synthesis and signal transduction has revealed an essential role for BRs in cell elongation and differentiation. bri1 is the only BR-insensitive loss-of-function mutant identified so far by genetic screens (Clouse et al.,1996; Kauschmann et al.,1996; Li and Chory,1997). Dark-grown bri1 mutants resemble BR-deficient mutants, exhibiting severe dwarfism with dark green epinastic leaves, and reduced apical dominance and fertility(Clouse et al., 1996; Kauschmann et al., 1996; Li et al., 1996; Szekeres et al., 1996; Azpiroz et al., 1998; Choe et al., 1999; Choe et al., 2001; Noguchi et al., 1999).

BRI1 encodes a membrane-localized leucine-rich repeat receptor-like kinase (LRR-RLK). The extracellular LRR domain is connected by a single pass transmembrane domain to a typical Ser/Thr kinase(Li and Chory, 1997; Friedrichsen et al., 2000). While more than 200 LRR-RLKs have been identified in the Arabidopsisgenome (Yin et al., 2002b;Shiu and Bleecker, 2001), only a few have been assigned a function. Moreover,most of these receptors have no known ligand. Genetic and biochemical evidence has demonstrated that BRI1 is an essential component of a receptor complex that binds brassinolide (BL) with high affinity(Wang et al., 2001). Although understanding how BRs are perceived by the BRI1 receptor remains to be elucidated, other signaling components operating downstream of BRI1 have been identified. BRI1 associated receptor kinase 1 (BAK1), a much smaller RLK containing only five LRRs, was identified in an activation-tagging screen for bri1 suppressors (Li et al.,2002) and in a yeast two-hybrid screen for BRI1 interacting proteins (Nam and Li, 2002). BAK1 thus acts in close proximity with BRI1, perhaps as its co-receptor. Brassinosteroid insensitive 2 (BIN2)/UCU1/DWF12, one of the 10 Arabidopsis GSK-3/Shaggy–like kinases, negatively regulates BR signaling (Li and Nam, 2002; Pérez-Pérez et al.,2002; Choe et al.,2002), while bri1 Suppressor 1 (BSU1), a nuclear-localized Ser/Thr phosphatase, positively regulates BR signaling(Mora-García et al.,2004). When BR levels are low, proteins in the BES1/BZR1 family are hyperphosphorylated by BIN2 and targeted for degradation. Upon BR perception, dephosphorylated BES1/BZR1 proteins accumulate in the nucleus,where they presumably regulate the transcription of BR-regulated genes(Wang et al., 2002; Yin et al., 2002a). Many BR-induced genes are related to cell wall loosening enzymes, supporting the role of BRs in cell expansion (Kauschmann et al., 1996; Nicol et al.,1998; Yin et al.,2002a; Goda et al.,2004).

Early physiological studies using Zinnia explants have shown that BRs play an important role in promoting xylem differentiation (reviewed by Fukuda, 1997). The BL synthesis inhibitors, uniconazole and brassinazole (BRZ)(Iwasaki and Shibaoka, 1991; Asami et al., 2000), prevent xylem differentiation in the Zinnia cell cultures, which can be restored by exogenous application of BL. Furthermore, the levels of BR intermediates have been shown to peak at the transition from undifferentiated cells to tracheary elements (Yamamoto et al., 2001). BRs also appear to promote xylem differentiation in cress plants (Nagata et al.,2001). In Arabidopsis, two mutants defective in BR-synthesis, cpd and dwarf7, exhibit vascular differentiation defects (Szekeres et al.,1996; Choe et al.,1999); however, vascular phenotypes in BR biosynthetic and signaling mutants have not been characterized in detail. How BRs regulate vascular development in Arabidopsis and whether xylem differentiation is controlled by specific BR-receptors are still unanswered questions.

We identified three proteins from the Arabidopsis genome,designated BRL1, BRL2 and BRL3, with high sequence identity to BRI1. The BRL2 gene corresponds to the previously identified Vascular Highway 1 (VH1), which appears to be required to maintain provascular differentiation in the leaves(Clay and Nelson, 2001). Here,we show that BRL1 and BRL3, but not BRL2, can complement bri1 mutant phenotypes when expressed under the control of the BRI1 promoter. We show that BRL1 and BRL3, but not BRL2, can bind BL with high affinity. Unlike BRI1, which is expressed ubiquitously, the BRL genes are predominantly expressed in the vascular tissues. Consistent with this observation, we found a brl1 mutant had increased phloem and reduced xylem differentiation compared with the wild type. Double mutant combinations of bri1 brl1 and triple bri1 brl1 brl3 mutants displayed aberrant vascular differentiation. We propose that BRL1 and BRL3 are novel plasma-membrane-localized steroid receptors that function specifically in provascular differentiation to maintain proper xylem:phloem ratios.

Ecotypes Columbia (Col-0) and Wassilewskija (Ws-2) were used as the wild types. Seeds were germinated either on 1/2 MS medium (Caisson) plus 1% sucrose or directly in soil. Plants were grown under fluorescent light (12 hours light/12 hours dark cycles) except for the BL-binding assays, in which plants were grown under short-day conditions (9 hours light/16 hours dark). Seeds were kept at 4°C for at least 72 hours before light treatment.

Screening and isolation of a T-DNA insertion in the brl1-1 line was done as described by the Arabidopsis knock-out Facility at the University of Wisconsin(http://www.biotech.wisc.edu/Arabidopsis/). The brl1-2, brl2 and blr3 mutants were isolated from the T-DNA Express Collection at the Salk Institute(http://signal.salk.edu/cgi-bin/tdnaexpress). The gene-specific primers used for PCR genotyping of the T-DNA insertions and the location of the T-DNA insertions for every allele are provided in Table S1(in the supplementary material).

Tissue samples were fixed overnight in 4% (v/v) glutaraldehyde in 25 mM NaPO4 buffer, pH 7.4, and then dehydrated with a graded series of ethanol and infiltrated with polymethacryl resin Technovit 7100 (Heraeus Kulzer, Germany),followed by embedding and polymerization in Technovit 7100. Sections (4 μm)were done using a Leica Jung Autocut microtome (Wetzlar, Germany). The tissue sections were stained with 0.1% Toluidine Blue in 0.1 M NaPO4 buffer, pH 7.0,and observed under bright-field illumination through a Zeiss Axioskop microscope. Images were processed using Photoshop 7.0. To calculate the number of phloem cells, pictures of each 4 μm transverse section were taken at 40× magnification under a Zeiss microscope. Then each vascular bundle was printed separately. The phloem cells were delimited with a pen and hand counted for each bundle of every plant. The average of phloem cells per plant was calculated using a total of 12 plants. Histochemical staining of plants expressing the GUS reporter fusions was basically done as reported in Baima et al. (Baima et al., 1995).

Visualization of green fluorescent protein (GFP)-tagged BRLs was done on 7-day-old light-grown hypocotyls using a BioRad Radiance 2100 microscope. The images were captured using the Kalman filter (n=5), and then analyzed with Confocal Assistant software.

BRL1, BRL2 and BRL3 genes were amplified from bacterial artificial chromosome (BAC) clones F20N2, F14H20 and MRP15, respectively, and cloned into a modified pBluescript plasmid that contains a 1.7 kb BRI1-promoter sequence and the RbcS-E9 terminator. After sequencing verification, the pBRI1-BRL-E9ter fragments were cleaved from the pBluescript plasmids and cloned into the pPZP212 binary vector. Site-directed mutagenesis was carried out using Stratagene's QuickChange Site-Directed Mutagenesis kit to introduce G597E (corresponding to bri1-113 mutation) or E1056K (corresponding to bri1-1mutation), mutation into the pBRI1-BRL1-E9ter construct. Plant transformation into the weak bri1-301 mutants by floral dipping was done according to Clough and Bent (Clough and Bent, 1998). Due to unknown reasons, the pBRI1-BRL2-E9ter plasmid could not be transformed into Agrobacterium cells. Instead, we constructed a GFP-tagged version of the BRL2 transgene and were able to transform the resulting pBRI1-BRL2::GFP-E9ter construct into the bri1-301mutants.

For BRI1, a 1.7 kb fragment including the 5′ utr and promoter region was amplified from BAC F23K16 and cloned between the HindIII/BamHI sites in vector pBI101.3. Transgenic plants were selected on kanamycin. For BRL genes, a 1.72 kb SpeI/HindIII restriction fragment of the BRL1promoter and a 750 bp NsiI/AflII restriction fragment of the BRL3 promoter were individually ligated in frame to the GUS gene of the binary vector pCB308. Transgenic plants were selected for resistance to the herbicide glufosinate. Homozygous plants were used for GUS histological staining.

These experiments were performed essentially as reported in Wang et al.(Wang et al., 2001), with slight modifications. For ³[H]-BL binding assays with immunoprecipitated proteins, 4-week-old T3 plants expressing GFP-tagged BRLs were grown under short-day conditions. The membrane fractions were extracted from 15 g of rosette leaves using 2 ml/g tissue of BRI1 extraction buffer at 4°C (Wang et al., 2001). Membranes were immunoprecipitated using 5 μl of anti-GFP antibodies(Molecular Probes) and Protein A agarose beads (200 μl of 50% slurry immobilized protein A), (Pierce). After extensive washing, the immunoprecipitate was aliquoted in 50 μl/sample and was used for individual binding experiments. Specific BL binding and data analysis was done according to Wang et al. (Wang et al.,2001). Immunoprecipitated GFP fusion proteins were analyzed by Western blotting with anti-GFP antibodies (1/3000 dilution).

To identify novel receptors that may function in BR perception, a BLAST search of the Arabidopsis genome with the kinase domain of BRI1 was conducted, identifying three BRI1-like genes(Fig. 1A), designated BRL1 (At1g55610), BRL2 (At2g01950), and BRL3(At3g13380). These genes comprised large intronless open reading frames (ORF)of 3498, 3429 and 3483 base pairs encoding predicted proteins of 1166(estimated molecular mass of 127 kD), 1143 (125 kD), and 1161 amino acids (127 kD), respectively. A full-length cDNA clone for each BRL had been identified from the Arabidopsis expressed sequence tag (EST) database(http://www.tigr.org/tdb/e2k1/ath1/ath1.shtml).

The BRI1-BRL family encodes LRR-receptor kinases with several predicted domains: an N-terminal signal peptide followed by an extracellular domain consisting of a variable number of LRR-motifs in tandem, and a single transmembrane pass followed by an intracellular Ser/Thr kinase domain(Fig. 1B). The numbers of predicted LRR-motifs were 22 in BRL2 and 23 in BRL1 and BRL3, compared with 25 in BRI1. A distinctive feature of all BRI1-like proteins is a 70-amino acid island domain essential for BRI1 to bind BRs, which was located four LRRs away from the transmembrane domain in all four proteins(Fig. 1B).

The amino acid sequence alignment is shown in Fig. 1C. BRL1 and BRL3 were the closest homologs, both sharing 43% sequence identity with BRI1, followed by BRL2 with 41%. Among the four LRR-RLKs, BRL1 and BRL3 shared the highest homology, being 80% identical. It is noteworthy that BRL1 and BRL3 contained island domains that were 93% identical, while BRL2 is the most divergent member (Fig. 1C). The finding that BRI1, BRL1, BRL2 and BRL3 constituted a subfamily of the LRR-RLK proteins led us to hypothesize that the BRLs may be BR receptors with distinct functions.

To test whether the BRL genes encode functional BR receptors,three fusion constructs containing the BRI1 promoter, the complete ORF for each BRL, and the 3′ terminator of the pea RbcS-E9 gene were generated. Each pBRI1::BRL::E9 transgene was then transformed into a weak bri1 (bri1-301) mutant background (Fig. 2). Unlike many bri1 mutants that are male sterile, bri1-301 displays a weak mutant phenotype and is fertile, and therefore is directly transformable(Fig. 2C). We found that transgenic plants expressing pBRI1-BRL1(Fig. 2D) and pBRI1-BRL3 (Fig. 2F)were able to suppress the dwarf bri1 phenotype, whereas the pBRI1-BRL2 transgenic plants were indistinguishable from the bri1-301 mutants (Fig. 2E). In addition, transgenic plants expressing pBRI1-BRL1and pBRI1-BRL3 in the Col-0 background conferred a long and curled petiole phenotype, similar to the one described for the overexpression of BRI1 under its own promoter (Wang et al.,2001), while wild-type plants transformed with pBRI1-BRL2looked similar to wild-type controls (data not shown). Thus, BRL1 and BRL3 can substitute for BRI1 function when expressed in the expression domain of BRI1. By contrast, BRL2 does not appear to function in BR signaling,despite its sequence homology with BRI1.

Sequence analyses of various bri1 alleles have previously identified several regions that are essential for BRI1 function, including the 70-amino acid island domain and the cytoplasmic kinase domain(He et al., 2000; Wang et al., 2001). To test whether the island domain and a functional kinase domain were required for BRL1 to carry out BRI1's function, two mutated pBRI1-BRL1 constructs, BRI1-BRL1-GE and BRI1-BRL1-EK, containing mutations in the 70-amino acid island and the kinase domain, respectively, were generated by PCR-based site-directed mutagenesis. The mutations G597E and E1056K correspond to bri1-113 and bri1-1 alleles(Li and Chory, 1997). These constructs were transformed individually into bri1-301. As indicated in Fig. 2G,H, neither mutated construct was able to fully rescue the mutant phenotype, confirming that both the integrity of the island domain and a functional kinase are essential for BRL1 to substitute for BRI1 function in BR-regulated plant growth.

To test whether BL is the ligand for the BRL receptors, full-length BRL1, BRL2, BRL3 were expressed as fusion proteins with GFP under the control of the strong CaMV 35S promoter. These constructs were transformed into Arabidopsis wild-type Col-0. Transgenic plants expressing 35S-BRL1::GFP, 35S-BRL2::GFP and 35S-BRL3::GFP did not show any apparent phenotypic alteration. Light-grown seedlings of these transgenic plants were examined for BRL subcellular localization using a confocal scanning laser microscope (CSLM), revealing that all three proteins were localized at the plasma membrane (Fig. 3D and data not shown). Accordingly, these membrane fractions were used to test if the BRLs were able to bind specifically to BL, using a previously defined assay for BRI1 (Wang et al., 2001). As shown in Fig. 3E, only BRL1 and BRL3 were able to bind BL, whereas no binding could be detected in BRL2-overexpressing lines. BRL1::GFP and BRL3::GFP proteins were immunoprecipitated by anti-GFP antibodies(Fig. 3C) and the specific binding activity of ³[H]-BL was calculated(Fig. 3A,B). The dissociation constants of the immunoprecipitated binding activity were calculated to be 3.6±0.07 nM for BRL1 and 53.4±0.04 nM for BRL3, as shown in Fig. 3A,B. As a positive control, we obtained a Kd=55±0.08 nM for the BRI1::GFPtransgenic plants, which are known to have a high BL-binding activity(Wang et al., 2001).

Having established that BRL1 and BRL3 encode functional BR receptors, we further investigated the organ- and tissue-specific expression patterns of these genes. Transgenic wild-type Arabidopsiscontaining intergenic 5′ regions of BRI1, BRL1 and BRL3 were used in transcriptional fusions to the histochemical GUS reporter gene (Fig. 4). The expression of pBRI1::GUS was detected postembryonically, in dividing and elongating cells of roots and hypocotyls of young seedlings, and in the emerging leaves and petioles of mature plants(Fig. 4C,F,I), consistent with the previously described role for BRI1 in BR-promoted cell elongation(Friedrichsen et al., 2000). However, pBRI1::GUS expression was not particularly enriched in vascular cells (Fig. 4C). Conversely, both BRL1 and BRL3 appeared to be predominantly expressed in the vascular tissues. Seven-day-old transgenic seedlings showed activity of the BRL1 promoter in the columella cells of the root tip,in the vascular initials in the meristematic region of the roots, and in the vascular tissues (Fig. 4A). This expression pattern appeared during late embryogenesis when pBRL1::GUS was expressed in the root tip(Fig. 4J), specifically in the columella cells and in the vascular initial cells. After germination, GUS staining was detected in the stele cells, and in the early differentiation zone of the root (Fig. 4K),where the expression continues from the root to the hypocotyls and cotyledons following the mid-vein (Fig. 4D). In mature plants, pBRL1::GUS expression traced the vasculature of the leaf, predominantly in the mid-vein(Fig. 4G), and was found in the vascular bundles of inflorescence stems. Histological examination of inflorescence stems revealed that the BRL1 expression is associated with the procambial cells of the vascular bundles(Fig. 4M), located between the differentiating xylem and the phloem.

By contrast to BRL1, the pBRL3 activity became apparent only during postembryonic development. It exhibited a very discrete pattern of expression, preferentially in the two protophloem cell files at the elongation zone of the root (Fig. 4B). The expression in these two cell files attenuated as the phloem cells differentiated in the upper root. In cotyledons and leaves, pBRL3::GUS was expressed in phloem cells (shown in Fig. 4E,H and detailed in 4N,O). The GUS expression started at the cotyledons and shoot apex, moving toward the basal part of the leaves, where the expression became attenuated(Fig. 4E,H). Complementary to pBRL1::GUS, which is expressed in the main vein, the expression domain of pBRL3::GUS was more intense in the secondary and tertiary veins and in the upper part of the cotyledons and leaves(Fig. 4N). No GUS staining could be detected in the inflorescence stems (data not shown).

To summarize, the vascular tissue-preferential expression of BRL1and BRL3 was common in all organs. In addition, the spatial complementation of this expression pattern was related to the level of maturation of vascular development, with BRL1 preferentially expressed in more mature cells than BRL3. In agreement, BRL1and BRL3 appeared to be preferentially expressed in the provascular tissues, in a tissue-specific microarray analysis(Birnbaum et al., 2003)available at the AREX database(www.arex.org). This unique expression pattern suggests that BR signaling through BRL1 and BRL3 may play a role in provascular organogenesis in plants.

To investigate the role of BRL1 and BRL3 genes in vascular development, we identified T-DNA tagged lines for each gene. A total of three mutants, two alleles of brl1, brl1-1 in the Ws-2 background and brl1-2 in the Col-0 background, and a single brl3 allele in the Col-0 background, were identified. The location of the T-DNA insertions is shown in Fig. 1B, and all alleles are null, based on Northern blotting analysis (data not shown).

All the brl single mutants identified exhibited wild-type appearance at the seedling and mature plant stages(Fig. 5 A-C and data not shown). However, due to the specific expression of these promoters in vascular cells, we looked for vascular defects by histological analysis. Vasculature in the Arabidopsis inflorescence stem follows a radial pattern in which vascular bundles alternate with interfascicular parenchyma. The vascular pattern in the stem arises from the meristematic activity of the procambial cells, which give rise to the phloem (in the outer side) and the xylem (in the inner side of the bundle; stained in dark blue by Toluidine Blue in Fig. 5I), thus creating a collateral pattern (Esau, 1965; Turner and Somerville, 1997).

brl1-1 displayed a clear vascular development phenotype(Fig. 5E,F). The number of phloem cells in brl-1 was increased (93±5, n=12)compared with wild-type Ws-2 (62±5, n=12). These alterations were associated with a concomitant reduction of xylem cells in the brl1-1 mutant (Fig. 5E,F). These results indicate that BRL1 promotes xylem differentiation and represses phloem differentiation in the inflorescence stem in the Ws-2 background. In addition to the phenotypic defects associated with reduced elongation previously described for the bri1 mutants, we found that bri1 mutants were also altered in vascular development,affecting the xylem/phloem patterning in the shoot. bri1-5 plants,like brl1-1 mutants, had increased phloem differentiation and reduced xylem differentiation compared with wild-type vascular bundle(Fig. 5E-G). Consistent with this, a BR-deficient mutant, det2-1(Li et al., 1996), also exhibited reduced xylem and increased phloem differentiation, as well as enlarged phloem cap cells (Fig. 5L). Conversely, we found that plants overexpressing BRI1 exhibited a striking increase in the procambial activity that leads to increased xylem differentiation in the vascular bundles(Fig. 5M).

In addition, we found that brl1-1 enhanced the dwarf phenotype of bri1-5, a weak allele of bri1(Fig. 5D). bri1-5brl1-1 double mutants exhibited an overall abnormal vascular organization, affecting the differentiation of both the phloem and xylem(Fig. 5H). The small phloem cells appeared collapsed between the enlarged phloem cap cells and xylem. The vascular defects of the bri1-5brl1-1 double mutants in the Ws-2 ecotype resembled those found in bri1-101, a strong bri1allele in the Col-0 ecotype (Fig. 5P), indicating that BRL1 and BRI1 function redundantly in provascular cell differentiation in the Ws-2 background.

It is noteworthy that the brl1-1 vascular phenotype described above appears to be background-dependent. brl1-2 and brl3,as well as double mutants brl1-2 brl3, did not display any significant vascular defects in the root and the shoot in the Col background. Furthermore, the number of phloem cells found in brl1-2 (83±5, n=12) and brl3 (82±5, n=12), as well as double mutants brl1-2 brl3 (79±5, n=10), was similar to the Col wild type (81±5, n=12). Similarly, the bri1-101 brl1-2 and bri1-101 brl3 vasculature resembled that of the single mutant bri1-101 (Fig. 5I-Kand data not shown), altogether suggesting that BRI1, BRL1 and BRL3 function redundantly in vascular development, especially in the Col-0 background. To test this possibility, we generated bri1-101 brl1-2 brl3 triple mutants in the Col-0 background. The triple mutant was smaller than the single bri1-101 mutant (Fig. 5N,O). The mutant plants were not fertile and most died after making a single inflorescence stem with a few small flowers(Fig. 5O). We observed enhanced vasculature defects in the triple mutant compared with the bri1-101single mutant (Fig. 5P,Q). Xylem development in the basal part of the stem appeared to be delayed in the triple mutant, resulting in decreased xylem differentiation and an absence of differentiated interfascicular fibers (Fig. 5Q). By contrast, the small phloem cells differentiated to give rise to an unusual formation of phloem fibers (arrowhead), and bundles were not fully differentiated. The presence of phloem fibers suggests abnormal coordination in vascular bundle development, in which the phloem may differentiate at the expense of xylem due to lack of BR perception in these cells. These results indicate that BR signal transduction through the BRI1,BRL1 and BRL3 receptors functions redundantly to regulate provascular cell growth and differentiation.

In this study, we showed that BRs are perceived by a small family of receptors in Arabidopsis. Previous studies have shown that BRI1 is responsible for the major BR-binding activity in Arabidopsis(Wang et al., 2001). Here we demonstrate that two closely related LRR-receptor kinases, BRL1 and BRL3, were also capable of binding BL with high affinity. Moreover, we found that these genes were expressed exclusively in vascular cells, and that a synergistic interaction of BRL1 and BRL3 with BRI1 was required to achieve normal vascular differentiation in the plant shoot.

Biochemical and genetic evidence demonstrates that BRL1 and BRL3, but not BRL2, are bona fide BR receptors. First, BRL1 and BRL3 can rescue the bri1 mutant phenotype when expressed under the control of the BRI1 promoter (Fig. 2)and they conferred long petiole phenotypes when expressed in wild-type plants,both suggesting that functional specification of these two receptors may reside, at least in part, in their cis-regulatory elements. Second,immunoprecipitates of BRL1 and BRL3, but not of BRL2, can bind BL with high affinity (Fig. 3A,B),indicating that both BRL1 and BRL3, like BRI1, are involved in BR perception. However, it is plausible that additional factors may specify the action of these receptors. In fact, signaling components downstream of BRI1 belong to multigene families, whose members probably have overlapping, as well as specific, functions (Li et al.,2002; Nam and Li,2002; Yin et al.,2002a; Wang et al.,2002; Mora-García et al., 2004). Future studies should investigate the signal transduction pathway(s) associated with the BRL receptors in comparison to what is known for BRI1. Moreover, differences in the amino acid sequences of these receptors may be functionally relevant. In the intracellular part, while the core sequences of the kinase domain are highly homologous among all members, the sequences are more divergent in the regions flanking the catalytic domain of the kinase, i.e. in the juxtamembrane region and the C-terminal extension. In the extracellular domain, the functional relevance of the number of LRRs is unclear, since BRI1 orthologs from other species with seemingly equivalent functions show some degree of variability. The intervening 70-amino acid islands probably play a more significant role, given that mutations in the extracellular domain of BRI1 cluster to this region and abolish binding of BL (Wang et al.,2001). The sequences of the island domains share a number of conserved residues, those of BRL1 and BRL3 being highly similar and distinct from those of BRI1 and BRL2. However, BRL1 binds BL with a 10-fold higher affinity than that of either BRI1 or BRL3, which suggests that regions in addition to the island domain may modulate the affinity for the ligand.

The BRL2 protein, previously described as Vascular Highway 1 (VH1), has been proposed to regulate provascular cell specification(Clay and Nelson, 2001). VH1 is expressed in the vascular tissues and loss-of-function of BRL2/VH1 in the Ler ecotype results in abnormal phloem transport due to a discontinuous phloem network that results in premature leaf senescence. Despite the sequence similarity to the BR receptors, we have shown that BRL2/VH1 does not function as a BR receptor. Sterols have been implicated in the establishment of the vascular pattern, in correct auxin transport and distribution, and more recently in leaf senescence(Carland et al., 2002; Willemsen et al., 2003; Suzuki et al., 2004; Motose et al., 2004). Since sterol-mediated vascular development appears to be required to maintain the continuity of the vascular network, it is therefore conceivable that BRL2 may bind a sterol. Our studies underscore the importance of performing biochemical assays prior to assigning a function to proteins identified through database searches.

The completed sequence of the rice genome offers the possibility of establishing a comparison between the Arabidopsis and rice BRI1 families. Fig. 1A depicts the phylogenetic relationship between all described or annotated members of the BRI1 family. Several inferences can be extracted from this tree, inasmuch as it accurately represents the evolution of the BRI1 family in angiosperms. First, all major branches include sequences of both rice and Arabidopsis, implying that the diversification inside the family predates the split between monocots and dicots. In addition, the whole family seems to have undergone an early functional division between receptors that bind BL (BRI1 and BRL1/3) and those that do not (BRL2). Finally, a relatively recent duplication event in the Arabidopsis lineage accounts for the presence of the highly homologous BRL1 and BRL3 genes,encoded in duplicated chromosomal regions. Their complementary expression pattern probably reflects a process of subfunctionalization from an ancestral function throughout the vascular tract(Lynch and Force, 2000). Analysis of the expression pattern and function of the unique rice BRL1 homolog will help to address this question.

The analysis of recessive, dwarf mutants insensitive to BL has led to the identification of BRI1 orthologs in several species other than Arabidopsis, such as OsBRI1 in rice(Yamamuro et al., 2000), UZU in barley (Chono et al.,2003), CU3 in tomato(Montoya et al., 2002) and LKA in pea (Nomura et al.,2003). It is noteworthy that all of them cluster in a highly supported group (Fig. 1A),suggesting that these proteins are the major BL receptors among angiosperms. Consistently, loss-of-function of Arabidopsis BRL1 and BRL3,described here, and BRL2 (Clay and Nelson, 2001) caused relatively subtle effects. It is remarkable that the expression of members of the BRI1 family, except BRI1 itself, is restricted to vascular tissues. It remains to be established whether this vascular specificity is an ancestral feature of these receptors, and by extension, of BL signaling, or rather a derived trait related to the functional specialization among paralogous genes.

It should be pointed out that other members of the ArabidopsisRLKs have been reported to behave similarly. In the case of the ERECTA family, both ERL1 and ERL2 act synergistically with ERECTA in defining aerial organ size(Shpak et al., 2003; Shpak et al., 2004). Like BRL1 and BRL3, ERL1 and ERL2 are also located in duplicated chromosomal regions (Shpak et al., 2004), suggesting that distinct RLK families may have evolved similarly in plants.

The molecular mechanism by which procambial cells differentiate into phloem or xylem remains largely unknown. The inner location of the vascular tissues has hampered the identification of vascular development mutants and often led to the identification of cell wall synthesis genes(Turner and Somerville, 1997; Caño-Delgado et al.,2000; Caño-Delgado et al., 2003; Taylor et al.,1999). To date, only a few mutants have been identified in Arabidopsis that specifically affect vascular patterning,(Clay and Nelson, 2001; Koizumi et al., 2000; Clay and Nelson, 2002; Bonke et al.,2003; Parker et al.,2003; Motose et al.,2004) (reviewed by Fukuda,2004).

The vascular-specific expression of BRL1 and BRL3 genes(Fig. 4) implies that these BR receptors play important roles during vascular development. Furthermore,studies using Zinnia mesophyll cell cultures have shown that exogenous application of BRs induces tracheary element differentiation in the terminal stage of xylogenesis (Yamamoto et al., 1997). Our studies in intact Arabidopsis establish that BR signaling in the provascular cells through BRL1, BRL3 and BRI1 receptors contributes to the collateral organization of the vascular bundles in the plant shoot (Fig. 5). The analysis of bri1 and brl1 mutant combinations revealed an increase in phloem and a decrease in xylem differentiated cells(Fig. 5F-H,Q), indicating that BR signaling in the procambial cells induces xylem cell differentiation and,at the same time, represses phloem differentiation. It is noteworthy that the bri1 null mutant (Fig. 5P) and the bri1-101 brl1 brl3 triple mutant(Fig. 5Q), exhibited phloem cells that are either collapsed with dramatically enlarged phloem cap cells or prematurely differentiated to phloem fibers, suggesting that different BR-signaling strength has different effects on phloem cell differentiation in the stem. Altogether, our data support the hypothesis that BRs contribute to the establishment of the collateral xylem/phloem pattern rather than just inducing xylem differentiation per se. In agreement, APL, a MYB transcription factor that determines phloem cell identity in the root, functions in promoting phloem differentiation and repressing xylem differentiation(Bonke et al., 2003), also suggesting that the differentiation of these two cell types is tightly linked.

We want to point out that the phenotypes of brl mutants appear to be affected by natural variation between different Arabidopsisaccessions. In the Ws-2 background, brl1-1 null mutants have an increased number of phloem cells compared with the wild type, while no significant differences were observed between the brl1-2 allele and the Col-0 wild type. Since no brl3 alleles were available in the Ws-2 ecotype, we attribute the differences in phenotype to the observed ecotype variation, rather than to an allele-specific effect. Furthermore, natural variation in genes that affect vascular development has been described previously for REV/IFL1, a member of the homeodomain-leucine zipper transcription factor family that controls interfascicular fiber differentiation (Talbert et al.,1995; Zhong and Ye,1999; Emery et al.,2003). Interestingly, Zinnia members of this family of transcription factors were found to be positively regulated by BRs(Demura et al., 2002; Ohashi-Ito and Fukuda, 2003). In addition, while the vh1 mutants in the Ler accession produced clear leaf senescence defects(Clay and Nelson, 2001), we did not observe these phenotypes in our mutant allele brl2 in the Col-0 ecotype under the same experimental conditions (data not shown). Several differences in allele variation between Col-0 and Ws-2 related to vascular development in the inflorescence stem of Arabidopsis have also been reported (Altamura et al.,2001). For example, phloem cap cells appeared to be larger in Ws-2 than in Col-0. We found increased size in the phloem cap cells in bri1,brl1-1, bri1-5brl1-1 and bri1-101brl1-2brl3 mutants, suggesting that natural variation in these genes may account for the differences in vascular structure between the Ws-2 and Col-0 accessions.

Several plant hormones have been implicated in vascular pattern formation. Auxin, the only hormone that has been shown to exhibit polar transport, acts as a major signal in vascular differentiation(Aloni, 1987; Sachs, 1991). In the`canalization hypothesis', the diffusion of the hormone from an auxin source induces the formation of a polar auxin transport system along a procambial cell file that differentiates into a xylem strand. However, additional molecules have been proposed to be required for vascular differentiation. Our data suggest that BR may be one of these additional molecules required for the induction of vascular cell differentiation in plants. The expression pattern of BRL1 in the provascular cells correlates with an inductive role of BRs in xylem differentiation; by contrast, the discrete expression of BRL3 in the phloem cells suggests that BRL3 may repress phloem differentiation, thereby controlling the formation of the collateral pattern in the vascular bundles. This hypothesis would be possible if BRs are transported from one cell to its neighbors. However, while polar auxin transport has been widely demonstrated, the long distance transport of BRs awaits to be demonstrated. Knowledge of where BRs are synthesized and whether they are perceived as part of a paracrine or autocrine system in the plant is crucial to further understand the role of BRs in vascular ontogeny.

We propose a model for the function of BRs in vascular cell differentiation(Fig. 6). BRs function through vascular-specific receptors BRL1 and BRL3 and the ubiquitously expressed receptor BRI1 to promote xylem differentiation and to repress phloem cell differentiation (Fig. 6A). In support of this model, phenotypes of BR-biosynthesis and -perception mutants exhibit a reduced number of xylem cells at the expense of phloem cells(Fig. 6B), indicating that the differentiation of these two cell types is reciprocally linked. Thus, BR perception by the procambial cells may act as an inductive signal through BRI1/BRL receptors in the acquisition of vascular cell fate and thus in the mutual regulation of xylem/phloem differentiation in the vascular bundle. However, it is also possible that BRs promote the division of the procambial cells in the shoot apex that will later differentiate into xylem cells. This hypothesis is supported by the phenotypes observed in the stems of BRI1 overexpressing plants (Fig. 5M), where procambial activity appeared to be enhanced compared with the wild type (Fig. 5I). By contrast, we observed that the bri1 null mutant(Fig. 5H) and bri1 br1 brl3 triple mutants (Fig. 5Q) had small bundles, probably due to a reduced division of the procambial cells.

Understanding the molecular mechanism by which plant steroid signaling induces vascular cell specification and how this signal is integrated with other signaling molecules such as sterols and auxin in the meristem is an outstanding question for the following years. Future molecular and computational analyses on BR action will be crucial to elucidate the regulatory mechanism of vascular cell specification leading to xylem/phloem differentiation.

Recently, Zhou et al. (Zhou et al.,2004) reported that the overexpression of BRL1 partially complements the phenotype of a weak bri1 mutant, thus confirming our results that BRL1 acts as a functional BR receptor.

We thank Jeff Long, Niko Geldner and Jennifer Nemhauser for critical reading of the manuscript. We thank Juan Carlos Izpisua-Belmonte for use of his confocal microscope, Gabriel Sternik for confocal assistance, and the T-DNA KO facilities at the Salk Institute and University of Wisconsin for making the mutants available. This work was funded by grants from the USDA and HFSP to J.C., a grant from NIH (GM60519) to J.L., and by HFSP fellowships to A.C.D. and S.M.G. J.C. is an Investigator of the Howard Hughes Medical Institute.