INTRODUCTION
Steroid hormones are employed both by animals and plants. Brassinosteroids, a group of plant steroid hormones, are essential for growth and development of plants (
Clouse and Sasse, 1998). While steroid hormones are perceived by intracellular receptors in animals, brassinosteroid-insensitive 1 (BRI1), a membrane protein, has been established as the major receptor for brassinosteroids in plants (
Li and Chory, 1997;
He et al., 2000;
Wang et al., 2001;
Kinoshita et al., 2005). BRI1 belongs to a family of leucine-rich repeat receptor like kinases (LRR-RLKs) with more than 200 members in
Arabidopsis (
Shiu and Bleecker, 2001). A typical structure of proteins in this family contains an extracelluar LRR domain, a single-pass transmembrane region and an intracelluar kinase domain. While many LRR-RLKs have unknown functions, BRI1 and its signaling pathway are well characterized (
Vert et al., 2005;
Kim and Wang, 2010;
Wang et al., 2012). Upon brassinosteroid binding, BRI1 kinase is activated to a basal level, which results in phosphorylation and release of BRI1 kinase inhibitor 1 (BKI1) from the membrane (
Wang and Chory, 2006;
Jaillais et al., 2011b). BRI1-associated kinase 1 (BAK1), another LRR-RLK with only five LRRs, thus interacts with and fully activates BRI1 (
Li et al., 2002;
Nam and Li, 2002;
Wang et al., 2008). Then BRI1 triggers a signaling pathway and regulates the expression of downstream genes. Nevertheless, it is still not fully clear how BRI1 is activated upon ligand binding.
Brassinosteroids are critical for vascular development (
Fukuda, 2004). Consistently, the
bri1 mutants in
Arabidopsis and rice both have abnormal vascular phenotype (
Cano-Delgado et al., 2004;
Nakamura et al., 2006). All the three homologs of BRI1 in
Arabidopsis (BRL1, BRL2 and BRL3) are specifically expressed in vascular cells (
Clay and Nelson, 2002;
Cano-Delgado et al., 2004), albeit with some different patterns. BRL1 and BRL3 have been suggested to promote xylem differentiation and repress phloem formation (
Cano-Delgado et al., 2004), while loss of VH1/BRL2 causes defective vein pattern formation and abnormal vascular transport in leaves (
Clay and Nelson, 2002;
Ceserani et al., 2009). BRL1 and BRL3, but not BRL2, rescue the phenotype of
bri1 mutants when overexpressed (
Cano-Delgado et al., 2004;
Zhou et al., 2004). Supporting the functional studies, biochemical assays showed that BRL1, BRL3 and BRI1 strongly bind to brassinolide with dissociation constants of 3.6 ± 0.07 nmol/L, 53.4 ± 0.04 nmol/L and 55 ± 0.08 nmol/L (
Cano-Delgado et al., 2004), respectively. In contrast, BRL2 showed no specific brassinolide binding activity (
Cano-Delgado et al., 2004).
Structural studies of free BRI1(LRR) and its complex with brassinolide provide important insight into steroid hormone recognition by plants (
Hothorn et al., 2011;
She et al., 2011). BRI1 harbors an LRR domain interrupted by an island domain. Brassinolide induces and binds to a hydrophobic pocket between the non-LRR island domain and the inner surface of several neighboring C-terminal LRRs of BRI1. However, the mechanism underlying the differential binding of brassinolide to its receptors remains less well understood. Here we report the crystal structure of BRL1(LRR) in complex with brassinolide and a modeled structure of the brassinolide binding domain of BRL2. The structure reveals the molecular basis for the different binding affinities of the BRI1 family of receptors with brassinolide. Based on the structural studies and others’ data, we also suggest possible mechanisms for the activation of BRI1 family receptors.
RESULTS
Overall structure of BRL1(LRR)
The ectodmain of the BRL1(LRR) (residues 25–758) was expressed and purified as previously described (
She et al., 2011). The structure of BRL1(LRR) (residues 31–60, 64–757) in complex with brassinolide was solved to 2.5 Ǻ using molecular replacement (Table 1). Structure superposition showed that the structure of BRL1(LRR) is similar to that of BRI1(LRR) (
Hothorn et al., 2011;
She et al., 2011) (Fig. 1). Like BRI1(LRR), BRL1(LRR) was monomeric in solution (data not shown) and structure. BRL1(LRR) contains 24 LRRs and forms a right-handed superhelix with a similar size to BRI1(LRR) (Fig. 1B). While the convex side of BRL1 consists of helices or loops, the concave side is composed of 24 parallel β strands of regular LRRs and one antiparallel β strand from N-terminal cap (Fig. 1A). As observed in the BRI1(LRR) structure (
Hothorn et al., 2011;
She et al., 2011) and another plant LRR protein PGIP2 (
Di Matteo et al., 2003), the β strands on the convex side also exist in BRL1(LRR) (Fig. 1A). This is likely due to the plant-specific consensus sequence L/fXGxI/vP (X and x stand for polar and any amino acids respectively) (Fig. 2) found in many LRR-RLKs. The N- and C-terminal caps, like those of BRI1(LRR), shield the two hydrophobic ends of the LRR solenoid from solvent (Fig. 1A). Seven potential glycosylation sites (Asn97, Asn157, Asn227, Asn257, Asn362, Asn532 and Asn558) are defined by sufficient electron density (Figs. 1B and 2). Four of them are conserved in BRl1(LRR) at the structural level (Fig. 1B). Interestingly, the carbohydrate moieties of both Asn532 and Asn558 in BRL1(LRR) but not Asn532 itself as observed in BRI1(LRR) pack against the island domain (Fig. 1B) and appear to have a role in stabilizing the conformation of the brassinolide binding site. Seven of the eight disulfide bonds are conserved between BRL1(LRR) and BRI1(LRR), except Cys471–Cys498 present in BRL1(LRR) (
She et al., 2011) (Figs. 1A and 2). Like that of BRI1, the island domain of BRL1 locates in concave side and comprises three antiparallel β strands and a 3
10-helix (Fig. 3A).
Brassinolide recognition by BRL1
Brassinolide well defined by electron density (Fig. 3B and 3D) binds to a hydrophobic pocket formed between the island domain and the inner surface of the solenoid of BRL1(LRR), which highly resembles that of BRI1(LRR) with an RMSD of 0.76 Ǻ over aligned 205 CA around the binding pocket (Fig. 3A). Brassinolide, in particular the distal side chain, adopts a similar extended conformation when binding to BRL1(LRR) and BRI1(LRR) (Fig. 3C). Most of the interactions, including hydrophobic interactions and hydrogen bonding, of brassinolide with BRI1(LRR) and BRL1(LRR) are highly conserved (
Hothorn et al., 2011;
She et al., 2011) (Figs. 3C and 4). As in the BRI1(LRR), the fused ring moiety of brassinolide (ring A–D) occupies most of the binding surface in the structure of BRL1(LRR) (Fig. 3C). Phe586 and Tyr627 of BRL1(LRR) interact with the island domain side of the fused ring moiety (Fig. 3C), while Gln666 in BRL1(LRR), substituting Phe681 in BRI1, establishes Van Der Waals contacts with the LRRs side of brassinolide (Fig. 3C). Nearly perpendicular to the fused ring moiety, the distal side chain (C24–28) of brassinolide anchors to a hydrophobic cavity formed by Ile527 and Trp551 from LRRs side and Phe584, Leu601 and Thr631 from island domain side (Fig. 5A). As in BRI1(LRR), the C23 hydroxyl group forms hydrogen bonds, but with the backbone nitrogen and oxygen of Met632 in BRL1(LRR) (Fig. 3C).
Notable structural differences, however, occur to the end of the fused ring portion (Fig. 3C). Compared to that in BRI1(LRR), this region of brassinolide binds deeper into the conserved pocket of BRL1(LRR), leading brassinolide to bury a larger surface area in BRL1(LRR) (388 Ǻ
2) than in BRI1(LRR) (360 Ǻ
2). These observations suggest a higher binding affinity of brassinolide with BRL1 than with BRI1, supporting previous biochemical data (
Cano-Delgado et al., 2004). While most of the brassinolide-interacting residues from the island domain are conserved between BRL1 and BRI1, the nonconserved Arg588 of BRL1 makes tight packing against the carbonyl oxygen in ring B of brassinolide via cation-pi interaction (Fig. 3B and 3C), further blocking brassinolide from solvent region. In contrast, Lys601 in BRI1, the equivalent of Arg588 in BRL1, is not directly involved in interaction with brassinolide. The varied residues of the brassinolide-binding pocket from the LRR domain appear to be critical for a deeper binding of brassinolide to BRL1(LRR). In the structure of brassinolide-bound BRI1(LRR), Asn705 supports brassinolide via contact of its hydrophobic portion with ring A (Fig. 3C), thus blocking brassinolide from binding further to the bottom of the binding pocket. In BRL1, Gly690 with no side chain substitutes Asn705 of BRI1, allowing ring A of brassinolide to fall down and form Van Der Waals interactions with Val667, Gly690 and Val691 from underneath (Fig. 3C). Together, our structural analyses offer an explanation for the higher affinity of brassinolide with BRL1 than with BRI1.
Structural model of BRL2
BRL2, unlike BRL1 or BRL3, is unable to recover the wild type phenotype of a
bri1 mutant and has no specific binding to brassinolide despite their high sequence homology (
Cano-Delgado et al., 2004). These results appear a little surprising, because the residues around the brassinolide binding site are largely conserved among the three proteins (Fig. 4). In fact, BRL2 (residues 497–729) has a higher sequence homology with BRL1 than with BRI1. Additionally, the two residues, Arg588 and Gly690 of BRL1 that contribute to a higher binding affinity to brassinolide are also conserved in BRL2 (Fig. 4). These data suggest that there exist other non-conserved residues in BRL2 responsible for its inability to recognize brassinolide. The high sequence homology with BRL1 allowed us to build a structure model of BRL2 with high confidence. We therefore modeled a structure for BRL2 (residues 497–729) using BRL1 as the template with the program MODELLER (
Eswar et al., 2008). The modeled partial BRL2 structure shows that the bulky and negatively-charged residue Glu614 substitutes Ile642 of BRL1, which is located at the inner end of brassinolide binding pocket (Fig. 5). We explored all possible rotamers of the side chain of Glu614 in COOT and found that one preferred rotamer without steric clashes with other parts of the binding pocket is close to the hydrophobic cavity anchoring the side chain of brassinolide (Fig. 5B). Thus, this mutation in BRL2 may act to disrupt the brassinolide binding by changing the hydrophobicity of this area. Supporting the significance of this hydrophobic cavity in brassinolide recognition by BRI1, introduction of polar groups at the end of side chain significantly attenuates brassinolide bioactivity (
Back and Pharis, 2003)and the hydroxylation of C26 by an enzyme BAS1 is shown to be one way to inactivate brassinosteroids in plants (
Neff et al., 1999).
DISCUSSION
We have solved the crystal structure of the BRI1 homolog BRL1 in complex with brassinolide in the current study. While BRI1(LRR) and BRL1(LRR) share a highly conserved structure (Fig. 3A and 4), the subtle differences around their brassinolide binding sites generate striking effects on their ability to recognize brassinolide (
Cano-Delgado et al., 2004) (Fig. 3C). The interaction with Arg588 of BRL1 contributes to a higher binding affinity of brassinolide with BRL1 than with BRI1 (Fig. 3C). The non-conserved Gly690 of BRL1 appears to further strengthen its recognition of brassinolide (Figs. 3C and 4). The fact that many BRI1 homologs harbor a small side chain residue (Ala or Gly) at the corresponding position of Gly690 of BRL1 suggests a conserved role played by this residue in recognizing brassinolide (Fig. 4). Our modeling study suggests that a change in one residue around the side chain binding cavity can have a deleterious effect on the recognition of brassinolide by BRI1 family of receptors (Fig. 5). Interestingly, this residue, Glu614 of
Arabidopsis BRL2, is conserved in rice BRL2 (OsBRL2) (
Nakamura et al., 2006), but not in other BRI1 homologs (Fig. 4), suggesting that it may be specific for BRL2 function. However, it is unclear why BRL3 exhibited a low affinity with brassinolide than BRL1, because all amino acids involved in brassinolide recognition are conserved between BRL1 and BRL3 (Fig. 4). The lack of commercially available radio-labeled brassinolide as well as the extremely low solubility of brassinolide rendered it difficult for us to test the non-conserved residues critical for selection of BRI1 family of receptors by brassinolide. Nonetheless, our data provide some structural insights for differential binding of brassinolide to the BRI1 family of receptors and allow a more efficient design of plant growth regulators for agricultural practice.
The mechanism of receptor activation of BRI1 family receptors is not fully clear. The crystal packing and the gel filtration (
Hothorn et al., 2011;
She et al., 2011) assay didn’t support the model in which ligand induces dimerization of the extracelluar domains of BRI1. The most marked conformational change upon hormone binding occurs to the interdomain loops that connect the island domain and LRRs in the structure of BRI1(LRR) (
She et al., 2011). In BRL1(LRR), the interdomain loops also adopt similar conformations with BRI1(LRR) (Fig. 3A). This conformation change may thus create a binding surface to interact with other protein. Two BRI1 mutants Gly644Asp and Thr750Ile, located around the ligand binding pocket (Fig. 3A), show loss-of-function phenotypes (
Noguchi et al., 1999;
Friedrichsen et al., 2000) but still bind brassinolide well (
Wang et al., 2001;
Kinoshita et al., 2005). These two residues are conserved among the BRI1 family receptors (Fig. 4), which may be key residues in extracelluar domain responsible for receptor activation. When animals’ steroid hormones are recognized by nuclear receptors, they are almost fully buried into the largely helical structures of ligand binding domains of their receptors and their binding can induce a dramatic shift of the C-termial activation helix to initiate downstream signaling (
Renaud and Moras, 2000;
Huang et al., 2010). In contrast, one side of the fused ring of brassinosteroid is nearly solvent exposed in the structures of BRI1(LRR) and BRL1(LRR) (Fig. 3A). This unshielded region may provide additional sites for protein-protein interaction. The extracelluar domain of the BRI1 coreceptor, BAK1, has been shown to provide a platform for receptor/coreceptor complex formation (
Jaillais et al., 2011a). Therefore, it is likely that the extracelluar domains of BAK1 and BRI1 form heterodimers to trigger the downstream signaling pathway. However, it still cannot exclude the possibility that some unknown players may bridge the interaction of extracelluar domains of BAK1 and BRI1 family proteins, as suggested by others (
Nam and Li, 2002). Further structural and functional studies are required to elucidate the mechanism for the activation of BRI1 family receptors.
MATERIALS AND METHODS
Protein expression and purification
The LRR domain of BRL1 (residues 25–758) from Arabidopsis with an engineered C-terminal 6× His tag was generated by standard PCR-based cloning strategy and its identity was confirmed by sequencing. The protein was expressed in sf21 cells using the vector pFastBac 1 (Invitrogen) with a modified N-terminal Hemolin peptide. One litre of cells (2.0 × 106 cells/mL) was infected with 20 mL baculovirus using a multiplicity of infection of 4 at 28°C, and protein was harvested from the media after 48 h. The protein was purified using Ni-NTA (Novagen) and size-exclusion chromatography (Superdex 200, Pharmacia) in buffer (10 mmol/L Tris, pH 8.0, 100 mmol/L NaCl) at 4°C. For crystallization of BRL1(LRR), the purified protein was concentrated to about 3.0 mg/mL.
Crystallization, data collection, structure determination and refinement
Crystals of BRL1(LRR) were generated by mixing the protein with an equal amount of well solution (1 mL) by the hanging-drop vapor-diffusion method. A mixture of brassinolide (in DMSO) and BRL1(LRR) with a molar ratio of 10:1 was centrifuged at 14,000 g for 2 min to remove insoluble brassinolide and then used for generating crystals of their complex. The buffer producing crystals of brassinolide-bound BRL1(LRR) contained 0.2 mol/L ammonium citrate dibasic and 20% (
w/v) polyethylene glycol (PEG) 3350. Crystals grew to their maximum size (0.1 × 0.1 × 0.05 mm
3) within 7 days at 18°C. The diffraction data set was collected at the Shanghai Synchrotron Radiation Facility (SSRF) at beam line BL17U1 using a CCD detector. Crystals of brassinolide-bound BRL1(LRR) belong to space group C2. For data collection, the crystals were equilibrated in a cryoprotectant buffer containing reservoir buffer plus 20.0% (
v/v) glycerol. The data were processed using HKL2000 (
Otwinowski and Minor, 1997). Molecular replacement (MR) with the program PHASER (
McCoy et al., 2007) was used to solve the crystal structure of brassinolide-bound BRL1(LRR). The atomic coordinates of BRI1 (PDB ID: 3RGZ) were used as the initial searching model. The model from MR was built with the program COOT (
Emsley and Cowtan, 2004) and subsequently subjected to refinement by the program PHENIX (
Adams et al., 2002).The structure figures were prepared using PyMOL (http://www.pymol.org).
Homology modeling of BRL2
Homology model of BRL2 (residues 497–729) was built with MODELLER (
Eswar et al., 2008) using BRL1 as the template. BRL2 (residues 497–729) shares ~53% sequence identity with BRL1.
Higher Education Press and Springer-Verlag Berlin Heidelberg 2013