Molecular Mechanism for Plant Steroid Receptor Activation by Somatic Embryogenesis Co-Receptor Kinases

See allHide authors and affiliations

Science  23 Aug 2013:
Vol. 341, Issue 6148, pp. 889-892
DOI: 10.1126/science.1242468

Steroid Receptor Signaling

Plant brassinosteroids signal to diverse pathways in plant physiology. These steroid hormones are perceived at the cell surface, where they bind to the receptor BRASSINOSTEROID INSENSITIVE 1 (BRI1). Santiago et al. (p. 889, published online 8 August) now show that somatic embryogenesis receptor kinase 1 (SERK1) complexes with BRI1. Together, these receptor kinases form the steroid binding site, with the hormone acting as a “molecular glue” that stabilizes the interaction. Hormone-induced heteromerization of BRI1 with SERK1 leads to the activation of the cytoplasmic signaling cascade, triggering plant growth and differentiation.


Brassinosteroids, which control plant growth and development, are sensed by the leucine-rich repeat (LRR) domain of the membrane receptor kinase BRASSINOSTEROID INSENSITIVE 1 (BRI1), but it is unknown how steroid binding at the cell surface activates the cytoplasmic kinase domain of the receptor. A family of somatic embryogenesis receptor kinases (SERKs) has been genetically implicated in mediating early brassinosteroid signaling events. We found a direct and steroid-dependent interaction between the BRI1 and SERK1 LRR domains by analysis of their complex crystal structure at 3.3 angstrom resolution. We show that the SERK1 LRR domain is involved in steroid sensing and, through receptor–co-receptor heteromerization, in the activation of the BRI1 signaling pathway. Our work reveals how known missense mutations in BRI1 and in SERKs modulate brassinosteroid signaling and the targeting mechanism of BRI1 receptor antagonists.

Plants have membrane-integral receptor kinases that sense diverse extracellular signals including steroid hormones (1), peptides (2), and proteins (3). The steroid receptor kinase BRI1, which controls plant growth and development (4), binds the hormone brassinolide (BL) with its LRR ectodomain (58). BL binding to BRI1 causes ordering of a ~70-residue island domain that, together with the LRR core, may form a docking platform for a co-receptor protein (7, 8). The size of the envisioned interaction surface and its proximity to the membrane suggest that the SERK family of plant receptor kinases are likely co-receptor candidates. SERK1 and SERK3 are genetic components of the brassinosteroid signaling pathway (912). BRI1 and SERK3 associate in a ligand-dependent manner (13), and this interaction can be disrupted by the BRI1 inhibitor protein BKI1 (14). The BRI1 and SERK3 kinase domains transphosphorylate each other upon steroid sensing (13).

To study the interaction between BRI1 and SERK family proteins, we first characterized the bri1sud1 gain-of-function allele (15) in molecular detail. A 2.5 Å crystal structure of bri1sud1 (table S1) shows that the Gly643 → Glu point mutation is involved in the stabilization of the island domain, a structural motif that we previously suggested to be crucial for the interaction with the co-receptor (8) (Fig. 1A). Glu643 in bri1sud1 contributes to a hydrogen-bonding network in the island domain β sheet, which is part of the hormone binding pocket (green in Fig. 1A). The bri1sud1-BL complex indicates that bri1sud1 cannot signal independent of ligand, but rather stabilizes the hormone-bound state of the receptor. Consistent with this observation, we found that bri1sud1 plants are only partially resistant to the brassinosteroid biosynthesis inhibitor brassinazole (16) (Fig. 1B).

Fig. 1 The BRI1 and SERK1 ectodomains interact upon steroid sensing.

(A) The hormone binding pocket in bri1sud1 with the LRR core (blue), the island domain (green), BL (yellow), and island domain Tyr597, Tyr599, and Tyr642 contributing to the BL binding pocket in green (sticks). Polar interactions between island domain Arg596 and the main-chain oxygen and nitrogen from Ile621 are in gray. Additional contacts by Glu643 in bri1sud1 to Ser623 and Glu624 (red) are mediated by a water molecule (red sphere). (B) bri1sud1 plants show reduced hypocotyl growth in the presence of the BL biosynthesis inhibitor brassinazole (1 μM). Hypocotyl length is in mm ± SD (n = 30); NT, nontreated. Controls (Col-0, bes1-1D, and a bri1-null allele) are shown alongside. (C) Ribbon diagram of the SERK1 ectodomain with N- and C-terminal caps (orange), the LRR core (gold), disulfide bridges (green), and N-glycosylation sites in yellow (sticks). (D) Analytical size exclusion chromatography traces: BL-bound bri1sud1 elutes as a monomer (black dotted line), as does the isolated SERK1 LRR domain (blue dotted line). The bri1sud1-BL-SERK1 complex elutes as an apparent heterodimer (red line); a mixture of bri1sud1 and SERK1 in the absence of BL yields two isolated peaks that correspond to monomeric bri1sud1 and SERK1, respectively (black line). Void (V0) and total volume (Vt) are shown, together with elution volumes for molecular mass standards (A, thyroglobulin, 669 kD; B, ferritin, 440 kD; C, aldolase, 158 kD; D, conalbumin, 75 kD; E, ovalbumin, 44 kD; F, carbonic anhydrase, 29 kD). The calculated molecular masses for the bri1sud1 and SERK1 peaks are ~125 and ~35 kD, respectively; purified bri1sud1 and SERK1 are ~110 and ~30 kD. A280, absorbance at 280 nm. (E) SDS–polyacrylamide gel electrophoresis (PAGE) analysis of peak elution fractions from the experiments shown in (D). Single-letter abbreviations for amino acid residues: E, Glu; G, Gly; I, Ile; N, Asn; R, Arg; S, Ser; Y, Tyr.

We next produced the isolated SERK1 ectodomain (residues 24 to 213) by secreted expression in insect cells and obtained crystals after mutating two of the five N-glycosylation sites (Asn115 → Asp, Asn163 → Gln). A 1.5 Å crystal structure of SERK1 (table S1) revealed the presence of five LRRs and the canonical N- and C-terminal capping motifs found in plant LRR proteins (7, 8, 17) (Fig. 1C). The size of the SERK1 LRR domain is in good agreement with the size of the envisioned docking platform in BRI1.

We took advantage of the bri1sud1 mutation to form biochemically stable ternary complexes containing bri1sud1, SERK1, and the steroid hormone, purified by size exclusion chromatography. The isolated LRR domains of both bri1sud1 and SERK1 eluted as apparent monomers from the size exclusion column, whereas the retention time of a bri1sud1-BL-SERK1 complex corresponded to a heterodimer at physiological cell wall pH (pH 4.5 to 6; Fig. 1, D and E). Complex formation was BL-dependent, which suggests that the direct physical interaction of the BRI1 and SERK1 ectodomains is mediated by the hormone itself (Fig. 1, D and E). Similar complexes could be formed using the wild-type BRI1 ectodomain, but we were unable to crystallize them (fig. S1).

Crystals of a bri1sud1-BL-SERK1 complex diffracted to 3.3 Å resolution. The structure was solved by molecular replacement, using the isolated bri1sud1 and SERK1 LRR domains as search models (table S1). The asymmetric unit contained two highly similar bri1sud1-BL-SERK1 complexes [fig. S2; root mean square deviation (RMSD) = 0.7 Å comparing 893 corresponding Cα atoms]. SERK1 bound to the previously suggested docking platform in BRI1 (8), using the inner surface of its LRR domain (Fig. 2). Superposition with the isolated bri1sud1 and SERK1 ectodomains revealed no structural rearrangements upon complex formation (RMSD = 1 Å comparing 743 corresponding bri1sud1 Cα atoms, RMSD = 0.6 Å comparing 185 corresponding SERK1 Cα atoms, respectively). The small complex interface (buried surface area ~850 Å2) involves residues originating from the BRI1 island domain, from BRI1 LRR 25, and from SERK1 LRRs 1 to 4 (Fig. 2 and Fig. 3).

Fig. 2 Architecture of the ternary bri1sud1-BL-SERK1 complex.

(A) Side and 90°-rotated front views of the complex, with the bri1sud1 (blue) and SERK1 (orange) LRR domains in surface representation. (B) Corresponding ribbon diagrams of the bri1sud1 LRR domain (blue), the BRI1 island domain (green), and the SERK1 LRR domain (orange); BL is in yellow (surface representation). The co-receptor ectodomain binds to the lower end of the BRI1 superhelix, bringing the C termini of both ectodomains, which connect to the membrane helices and to the cytoplasmic kinase domains, into proximity.

Fig. 3 Ligand chemistry and genetic missense alleles validate the BRI1-SERK1 complex interface.

(A) The N-terminal capping domain of SERK1 directly contacts the BRI1 steroid binding pocket and the 2α,3α-diol moiety of BL. Shown is a molecular surface of BRI1, with the LRR core depicted in blue and the island domain in green. Parts of the SERK1 N-cap and LRR 1 are shown as ribbon diagram (orange). Main-chain hydrogen-bond interactions of SERK1 N-cap residues (Cys58, Cys65, and Asn66) with island domain BRI1 residue Arg640 are depicted as dotted lines (black). SERK1 residues Phe61 and His62 are shown alongside with interactions to the 2α and 3α OH groups in BL shown in red. Another hydrogen-bond interaction is between SERK1 residue Thr64 and Thr729 in BRI1 LRR 25. (B) Known genetic BRI1 and SERK1 alleles map to the complex interface. Thr750 (cyan) in the BRI1 C-terminal cap contacts the conserved SERK1 residue Arg73 and is mutated to Ile in the BRI1 loss-of-function mutant bri1-102. SERK1 residue Asp123 is centrally located in the complex interface (cyan) and establishes contacts with Glu749 in the BRI1 C-terminal cap. The serk3elg gain-of-function allele, which causes mutation of the corresponding Asp122 to Asn, may further stabilize the complex interface. (C) Analytical size exclusion chromatography traces of BL-bound bri1102, which elutes as a monomer (black dotted line), as does the isolated SERK1 LRR domain (blue dotted line). A mixture of bri1102, BL, and SERK1 elutes as two isolated peaks, which suggests that the Thr750 → Ile mutation destabilizes the complex (red line; compare Fig. 1D). (D) SDS-PAGE analysis of peak elution fractions from the experiments shown in (C). Single-letter abbreviations for amino acid residues: C, Cys; D, Asp; E, Glu; F, Phe; H, His; N, Asn; R, Arg; T, Thr; V, Val; Y, Tyr.

In contrast to previously described animal innate immunity receptor complexes (18, 19), LRR capping domains are involved in the formation of the BRI1-SERK1 complex: The N-terminal cap of SERK1 folds on top of the BRI1 steroid binding pocket, where it establishes contacts with the BRI1 island domain (green in Fig. 3A), with BRI1 LRR 25 (BRI1 residues 726 to 729; Fig. 3A and fig. S5, in blue) and with the hormone itself. Here, SERK1 residue Phe61 (which appears disordered in the isolated SERK1 structure) makes a stacking interaction with the C ring of the hormone, while the neighboring histidine establishes hydrogen bonds with the 2α,3α-diol moiety of brassinolide (Fig. 3A and figs. S3A and S4). Our complex structure thus provides an explanation for the BL-dependent interaction of BRI1 and SERK1 (Fig. 1, D and E, and fig. S1) and for the critical role of the 2α and 3α hydroxyls in brassinosteroid bioactivity (20, 21). These results are consistent with observations that a 2,3-acetonide form of BL acts as a brassinosteroid pathway antagonist (22). We speculate that this compound still binds to BRI1 (8) but disrupts the interaction with the co-receptor ectodomain, thereby blocking BRI1 activation (13, 22) (fig. S3).

The C-terminal cap of BRI1 also contributes to complex formation by establishing contacts with residues originating from SERK1 LRRs 1 to 4 (Fig. 3B and figs. S4 and S5). Major nonpolar interactions are provided by SERK1 residues Tyr97, Tyr101, Tyr125, and Phe145 as well as by residue Met745 in the BRI1 C-terminal cap (fig. S5). An extensive hydrogen-bonding network is centered around BRI1 C-terminal cap residue Glu749, which is contacted by SERK1 residues Glu99, Asp123, and Arg147 from SERK1 LRRs 2, 3, and 4, respectively (Fig. 3B and fig. S5). Acidic residues in the complex interface titrate in the pH range relevant in the plant cell wall (Fig. 1, D and E, and fig. S1). All SERK1 interface residues are highly conserved among the known BRI1-interacting SERK family members from Arabidopsis and rice and in other small LRR receptor kinases (fig. S4).

The presence of two known genetic alleles in the outlined complex interface allowed us to validate our structural model. The strong loss-of-function allele bri1-102 (Thr750 → Ile; cyan in Fig. 3B;, see also fig. S5) (23), which does not affect steroid binding (24), mapped to the center of the BRI1-SERK1 interface, forming a hydrogen bond with SERK1 residue Arg73. Mutation of the respective threonine to isoleucine destabilized the complex (Fig. 3, C and D), which explains the strong phenotype (23). A gain-of-function mutation (serk3elg, Asp122 → Asn) (25) mapped to Asp123 in SERK1 (cyan in Fig. 3B; see also fig. S5). SERK1 Asp123 is part of the hydrogen-bonding network described above, and its mutation to Asn may allow for the formation of additional interactions with the BRI1 C-terminal cap. A more stable association of serk3elg with the receptor BRI1 would explain the behavior of this mutant, which displays a gain-of-function growth phenotype and more strongly associates with BRI1 upon steroid binding in vivo, even when the co-receptor kinase domain is catalytically inactive (25).

Taken together, our findings define the activation mechanism for a plant LRR receptor kinase, in which a shape-complementary co-receptor protein forms an integral part of the receptor’s ligand-binding pocket. Our pharmacologically (21, 22, 24) and genetically (11, 12, 23, 25) validated BRI1-BL-SERK1 complex structure suggests that BL acts as a “molecular glue” promoting association of the receptor with its co-receptor, which may trigger interaction of their kinase domains and subsequent activation of the signaling pathway (13). A similar activation mechanism has been suggested for soluble plant hormone receptors (26, 27). However, we emphasize that in the context of the full-length proteins, the transmembrane segments as well as the cytoplasmic kinase domains might play additional roles in receptor–co-receptor heteromerization and activation (13, 14). Also, our experiments cannot address the oligomeric state of BRI1-SERK signaling complexes at the plasma membrane, but rather define a minimal signaling unit required for receptor activation.

The interaction surface with BRI1 occupies only ~10% of SERK1’s total accessible surface area (blue in fig. S6). We used these sequence fingerprints to identify other putative BRI1 co-receptors (fig. S4). When mapping their overall sequence conservation to the molecular surface of SERK1 (orange in fig. S6), we found conserved surface patches that are much larger than required for the interaction with BRI1. SERK family co-receptors mediate the activation not only of BRI1 but of other plant receptor kinases that sense very different ligands (2830). We hypothesize that SERKs harbor several distinct, partially overlapping, but conserved surface patches, which allow them to establish specific interactions with different receptor kinases and, potentially, their ligands. In line with this idea, the serk3elg mutant acts as a gain-of-function allele in brassinosteroid signaling (25), whereas it behaves as a loss-of-function mutant in the FLAGELLIN-SENSING 2 pathway (25). Plant invertase/pectin methylesterase inhibitor proteins provide another example where distinct surface areas on a common structural scaffold allow for interaction with different protein targets (31). The widespread use of multifunctional SERKs in plant membrane signaling could enable signaling cross-talk directly at the level of ligand sensing at the cell surface and may orchestrate the plant’s response to external stimuli (15).

Supplementary Materials

Materials and Methods

Figs. S1 to S6

Table S1

References (3245)

References and Notes

  1. Acknowledgments: We thank Y. Jaillais and M. Jinek for comments on the manuscript, J. Martinez for help with data collection, A. Pauluhn for providing beam time and staff members of beamline PXII (Swiss Light Source, Villigen, Switzerland), Y. Jaillais for bri1sud1 and GABI_13E10 transgenic lines, and M. Blazquez for bes1-1D. Atomic coordinates and structure factors have been deposited in the Protein Data Bank with accession codes 4LSA (bri1sud1), 4LSC (SERK1), and 4LSX (bri1sud1-BL-SERK1). Supported by the Max Planck Society, a Federation of European Biochemical Societies long-term fellowship (J.S.), and an International Human Frontier Science Program Organization Career Development Award (M.H.). M.H. and Joanne Chory (Salk Institute) are inventors on a patent application filed by the Salk Institute for Biological Studies entitled “Structure of the Brassinosteroid Receptor BRI1, and Modulation of BRI1 Signaling” (patent 20130007910).
View Abstract

Navigate This Article