Arabidopsis thaliana brassinosteroid signaling kinases (BSKs) constitute a receptor-like cytoplasmic kinase sub-family (RLCK-XII) with 12 members. Previous analysis demonstrated a positive role for BSK1 and BSK3 in the initial steps of brassinosteroid (BR) signal transduction. To investigate the function of BSKs in plant growth and BR signaling, we characterized T-DNA insertion lines for eight BSK genes (BSK1–BSK8) and multiple mutant combinations. Simultaneous elimination of three BSK genes caused alterations in growth and the BR response, and the most severe phenotypes were observed in the bsk3,4,7,8 quadruple and bsk3,4,6,7,8 pentuple mutants, which displayed reduced rosette size, leaf curling and enhanced leaf inclination. In addition, upon treatment with 24-epibrassinolide, these mutants showed reduced hypocotyl elongation, enhanced root growth and alteration in the expression of BR-responsive genes. Some mutant combinations also showed antagonistic interactions. In support of a redundant function in BR signaling, multiple BSKs interacted in vivo with the BR receptor BRI1, and served as its phosphorylation substrates in vitro. The BIN2 and BIL2 GSK3-like kinases, which are negative regulators of BR signaling, interacted in vivo with BSKs and phosphorylated them in vitro, probably at different sites to BRI1. This study demonstrates redundant biological functions for BSKs, and suggests the existence of a regulatory link between BSKs and GSK3-like kinases.
Brassinosteroids (BRs) are steroidal hormones that are involved in regulation of multiple physiological and developmental processes in plants (Mandava, 1988; Clouse and Sasse, 1998; Yang et al., 2011). These include cell proliferation, senescence, male fertility, induction of flowering, fruit ripening, and stress responses (He et al., 2001; Nakaya et al., 2002; Symons et al., 2006; Bajguz and Hayat, 2009; Li et al., 2010; Ye et al., 2010). Plants defective in BR biosynthesis or signaling exhibit phenotypes such as extreme dwarfism, curled leaves, male sterility, photomorphogenesis in the dark, and altered vascular development (Chory et al., 1991; Clouse et al., 1996; Szekeres et al., 1996; Li et al., 2001).
Genetic, biochemical and molecular studies combined with proteomic and genomic approaches have enabled delineation of the BR signal transduction pathway in detail (Clouse, 2011; Kim and Wang, 2011). BRs are perceived at the cell surface by the leucine-rich repeat (LRR) receptor-like kinase BRI1 (Li and Chory, 1997; Hothorn et al., 2011; She et al., 2011). This perception initiates signaling cascades, resulting in activation of BZR transcription factors, which regulate BR-responsive genes (Yin et al., 2002; He et al., 2005). In the absence of BRs, the plasma membrane-associated protein BKI1 interacts with BRI1 and inhibits its kinase activity (Wang and Chory, 2006). When BRI1 is inactive, the downstream GSK3-like protein kinase BIN2 phosphorylates the transcription factors BZR1 and BZR2/BES1, leading to inhibition of their DNA-binding activity (Li and Nam, 2002; Vert and Chory, 2006), retention in the cytoplasm by 14-3-3 proteins (Gampala et al., 2007; Ryu et al., 2007), and degradation (He et al., 2002). Binding of BRs to the extracellular domain of BRI1 causes release of BKI1 (Wang and Chory, 2006; Jaillais et al., 2011), association of BRI1 with the LRR receptor kinase BAK1, and activation of its kinase activity (Wang et al., 2008). Substrates of BRI1 phosphorylation include the receptor-like cytoplasmic kinases (RLCKs) BSK1 and CDG1 (Tang et al., 2008; Kim et al., 2011). Once phosphorylated, BSK1 interacts with the phosphatase BSU1 and presumably activates it (Kim et al., 2009). In turn, BSU1 dephosphorylates and inactivates BIN2 (Kim et al., 2009). By an independent route, CDG1 phosphorylates BSU1 and enhances BSU1–BIN2 binding, thus contributing to effective dephosphorylation of BIN2 (Kim et al., 2011). Upon inactivation of BIN2, BZR1 and BZR2/BES1 undergo dephosphorylation by PP2A (Tang et al., 2011), accumulate in the nucleus, and regulate expression of BR-responsive genes (Sun et al., 2010; Yu et al., 2011).
BSK1 belongs to a sub-family of RLCKs (RLCK-XII) that includes 12 putative kinases in Arabidopsis (BSK1–12) (Tang et al., 2008). BSKs share homology in their sequence, and contain a putative kinase catalytic domain at the N-terminus and tetratricopeptide repeats at the C-terminus. Putative myristoylation sites may be responsible for their association with the plasma membrane. Biochemical and genetic analyses demonstrated that certain BSKs play a role in BR signaling (Tang et al., 2008; Kim et al., 2009). BSK1 and BSK3 were shown to interact with BRI1 in vivo and to be phosphorylated by BRI1 in vitro (Tang et al., 2008). BSK1 Ser230 was identified as the major site phosphorylated by BRI1 (Tang et al., 2008). Interestingly, the BRI1–BSK1 interaction was weakened by BR treatment, suggesting release of BSK1 from BRI1 after phosphorylation (Tang et al., 2008). Moreover, BRI1 phosphorylation of BSK1 at Ser230 enhanced the interaction of BSK1 with the phosphatase BSU1 (Kim et al., 2009). Based on this evidence, it was proposed that, upon BR perception, BSK1 is phosphorylated by BRI1 and released from the receptor complex to interact with and activate BSU1. However, how the BSK1–BSU1 interaction enhances the ability of BSU1 to dephosphorylate BIN2 and to regulate its activity is unknown. Gain-of-function genetic analysis supports the involvement of BSKs in BR signaling downstream of BRI1 and upstream of BIN2. In fact, over-expression of the BSK1, BSK3 or BSK5 genes partially suppressed the dwarf phenotypes of the weak bri1-5 mutant allele, while over-expression of BSK3 failed to rescue the phenotype of the bin2-1 mutant allele (Tang et al., 2008). However, the function of most BSKs has not been demonstrated by loss-of-function genetic analysis. A study of T-DNA insertion mutants for the BSK2, BSK3, BSK4, BSK5 and BSK12 genes did not reveal any evident growth or developmental phenotype (Tang et al., 2008). In addition, when bsk mutants were analyzed for their response to BRs, only the bsk3 mutant displayed hypersensitivity to brassinazole, an inhibitor of BR biosynthesis, and reduced sensitivity to brassinolide. The lack of phenotypic alteration in most of the mutants analyzed may be attributed to functional redundancy among BSKs.
We used a reverse genetic approach to study the function of BSKs in plant growth and to assess their contribution to the BR response. By using T-DNA insertions within eight of the 12 BSK genes and various mutant combinations, we demonstrated that at least five BSKs (i.e. BSK3, BSK4, BSK6, BSK7 and BSK8) play a partial overlapping role in plant growth and BR signaling. Biochemical analysis confirmed that multiple BSKs interact with BRI1 and serve as substrates of BRI1 phosphorylation. Moreover, BSKs were found to interact with BIN2 and the related kinase BIN2-like 2 (BIL2). BIN2 and BIL2 also phosphorylated BSKs in vitro, suggesting a novel regulatory link between these important components of BR signaling.
Mutations in multiple BSK genes induce morphological and developmental phenotypes
Members of the BSK family of receptor-like cytoplasmic kinases (Figure S1) play an important role in BR signaling (Tang et al., 2008). Because disruption of either BR biosynthesis or signaling results in growth defects (Clouse, 2011; Kim and Wang, 2011; Yang et al., 2011), we used a reverse genetic approach to analyze the function of BSKs in plant growth and development. Homozygous T-DNA insertion lines were isolated for eight of the 12 Arabidopsis BSK genes, and their growth was monitored under short-day conditions (see Figure S2 for sites of T-DNA insertion in BSK genes). Lack of detectable expression of BSK transcripts in the T-DNA lines was demonstrated by RT-PCR (Figure S3). As shown in Figure 1, growth of the bsk single mutants, measured as leaf area at 40 days after sowing, was indistinguishable from that of wild-type Col-0.
To examine genetic interactions between bsk mutations, higher-order mutants were generated and monitored for growth under short-day conditions (Figure 1). These included double (bsk3,4, bsk6,7 and bsk7,8), triple (bsk3,4,6, bsk3,4,7 and bsk3,4,8), quadruple (bsk3,4,6,7, bsk3,4,6,8 and bsk3,4,7,8) and pentuple (bsk1,3,4,6,7 and bsk3,4,6,7,8) mutants (see Figure S4 for the crossing strategy). The double mutants did not display any phenotype different to that of wild-type. However, when mutations in BSK6, BSK7 or BSK8 were introduced into the bsk3,4 mutant, the resulting bsk3,4,6, bsk3,4,7 and bsk3,4,8 triple mutants displayed significantly reduced growth compared to wild-type. Addition of mutations in BSK7 or BSK8 to the bsk3,4,6 mutant did not enhance the phenotype of the resulting bsk3,4,6,7 and bsk3,4,6,8 quadruple mutants. Interestingly, despite the lack of a phenotype in the bsk7,8 mutant, simultaneous mutation of bsk7 and bsk8 in the bsk3,4 or bsk3,4,6 backgrounds had a severe effect on growth of the respective quadruple (bsk3,4,7,8) and pentuple (bsk3,4,6,7,8) mutants. BSK3, BSK4, BSK7 and BSK8 belong to a common clade in the phylogenetic tree of the BSK family (Figure S1) and may exert a redundant function on growth. Finally, addition of a bsk1 mutation to the bsk3,4,6,7 mutant, resulting in the bsk1,3,4,6,7 pentuple mutant, partially suppressed the bsk3,4,6,7 phenotype. Similar differences between bsk mutant and wild-type plants were detected under long-day conditions, with the exception of the bsk3,4 mutant, whose growth was significantly reduced compared to the wild-type only under long-day conditions (Figure S5).
The bsk3,4,7,8 and bsk3,4,6,7,8 mutants, which showed the most severe reduced growth, also exhibited additional developmental phenotypes compared to wild-type plants (Figure 2). These included downward curled leaves with reduced width, length and number at all stages of growth, and with enhanced inclination (Figure 2). Because the morphological phenotypes of the bsk3,4,7,8 and bsk3,4,6,7,8 mutants were indistinguishable, the bsk6 mutation appeared to have a neutral effect in the bsk3,4,7,8 background. To exclude the possibility that the phenotypes observed in the bsk3,4,7,8 mutant were due to a background mutation, growth phenotypes were analyzed in the F2 progeny of the cross bsk3,4,7 x bsk3,4,8, which segregates for insertions in bsk7 and bsk8 and was used to isolate the quadruple mutant bsk3,4,7,8 (Figure S4). By analyzing 77 plants of this progeny representing nine genotypes, we observed that the growth phenotype co-segregated with the presence of an homozygous insertion in bsk7, or both bsk7 and bsk8, as well as an heterozygous insertion in bsk7, bsk8 or both (Figure S6).
Together, these results indicate that BSK family members are required for normal plant growth and development, and suggest a certain degree of functional overlap between BSK3, BSK4, BSK6, BSK7 and BSK8.
BSK mutants display reduced sensitivity to exogenous BR
BSK1, BSK3 and BSK5 have been genetically implicated in BR signaling (Tang et al., 2008). To explore the contribution of various BSKs to BR signaling, we examined growth of the bsk mutants described above in the presence of the BR 24-epibrassinolide (BL). Hypocotyl elongation and root growth inhibition of Arabidopsis seedlings, which are typical responses to exogenous application of BL, were quantified 7 days after sowing (Figure 3) (The absolute hypocotyl and root lengths of mutants are shown in Table S1). In agreement with previous observations (Tang et al., 2008), the bsk3 single mutant was less sensitive to BL than the wild-type Col-0, with reduced hypocotyl elongation and reduced root growth inhibition. In contrast, all the other single mutants were similar to the wild-type. The bsk6,7 and bsk7,8 double mutants were not affected in BL sensitivity, while the response of the bsk3,4 mutant was significantly different to wild-type but similar to the bsk3 mutant, indicating that the bsk4 mutation had a neutral effect on BL sensitivity in the bsk3 background. When a mutation in bsk6, bsk7 or bsk8 was introduced into the bsk3,4 mutant, the resulting bsk3,4,6, bsk3,4,7 and bsk3,4,8 triple mutants showed reduced hypocotyl elongation, but were only marginally affected in terms of root growth inhibition compared to bsk3,4. This indicates that BSK6, BSK7 and BSK8 share a certain degree of functional redundancy with either BSK3, BSK4 or both. Addition of a bsk8 mutation to the bsk3,4,7 mutant reduced both hypocotyl elongation and root growth inhibition of the resulting bsk3,4,7,8 quadruple mutant, while introduction of a bsk7 or bsk8 mutation into the bsk3,4,6 mutant reduced root growth inhibition but not hypocotyl growth of the resulting bsk3,4,6,7 and bsk3,4,6,8 quadruple mutants. However, the bsk3,4,7,8 quadruple mutant, which was most significantly affected in growth, showed the lowest sensitivity to BL. Finally, the bsk3,4,6,7,8 pentuple mutant displayed a reduced sensitivity to BL that was not significantly different from that of bsk3,4,7,8. These results provide genetic evidence that BSK6, BSK7 and BSK8, in addition to BSK1, BSK3 and BSK5, play a role in BR signaling.
Surprisingly, when a bsk1 mutation was introduced into the bsk3,4,6,7 mutant, the response to BL of the resulting bsk1,3,4,6,7 pentuple mutant was restored to the wild-type level in terms of inhibition of root growth (Figure 3b), and to the bsk3 level in terms of hypocotyl elongation (Figure 3a), suggesting an antagonistic interaction between BSK1 and other family members. To further explore the genetic interactions of BSK1 with BSK3, BSK4, BSK6, BSK7 or BSK8, a bsk1 mutation was introduced in various bsk mutant backgrounds, and sensitivity to BL was analyzed in the resulting mutants (Figure 4). In contrast to the effect of adding a bsk1 mutation into the bsk3,4,6,7 mutant, coupling a bsk1 mutation with bsk3, bsk6,7 or bsk3,4,6,7,8 (resulting in bsk1,3, bsk1,6,7 and bsk1,3,4,6,7,8 mutants, respectively) did not alter their sensitivity to BL. These results indicate complex genetic interactions between BSK genes, and suggest distinct functional characteristics for BSK1 compared to other family members.
BR-regulated expression of the SAUR-AC1 and DWF4 genes is altered in bsk mutants
To analyze the contribution of BSKs to the BR response at the molecular level, we treated bsk mutants with BL and monitored changes in expression of the BR-regulated genes SAUR-AC1 and DWF4 (Goda et al., 2002; Kim et al., 2006). SAUR-AC1 expression was induced approximately threefold in wild-type Col-0 Arabidopsis seedlings in response to BL (Figure 5a). Upon treatment of bsk mutants with BL, the bsk1, bsk4 and bsk5 single mutants displayed a significantly reduced induction of SAUR-AC1 expression compared to the wild-type. The SAUR-AC1 expression level was significantly higher in the bsk1 mutant than in the wild-type in the absence or presence of BL. SAUR-AC1 expression was only marginally induced by BL treatment in the bsk3,4 and bsk7,8 double mutants, which are mutated in pairs of phylogenetically related BSK genes, and was not induced at all in the bsk3,4,7,8 quadruple mutant and the bsk3,4,6,7,8 pentuple mutant. This correlated well with the drastic reduction in BL sensitivity observed for the bsk3,4, bsk3,4,7,8 and bsk3,4,6,7,8 mutant combinations (Figure 3). A significant reduction in SAUR-AC1 induction was also observed in the bsk3,4,6 and bsk3,4,7 triple, bsk3,4,6,7 and bsk3,4,6,8 quadruple, and bsk1,3,4,6,7 pentuple mutants. Surprisingly, a mutation of bsk6, bsk7 or bsk8 in the bsk3,4 background exerted an antagonistic effect on SAUR-AC1 expression, restoring a certain degree of BL inducibility in the resulting bsk3,4,6, bsk3,4,7 and bsk3,4,8 triple mutants.
The DWF4 gene encodes a key enzyme of BR biosynthesis, and its expression is negatively regulated by BR (Kim et al., 2006). As expected, BL treatment in wild-type Col-0 seedlings reduced the expression of DWF4 to almost 12% relative to untreated plants (Figure 5b). Under the same conditions, five of the eight bsk single mutants (bsk2, bsk3, bsk5, bsk6 and bsk7) retained significantly higher DWF4 expression compared to wild-type plants. In line with the effect observed on SAUR-AC1 expression, simultaneous mutation of the closely related genes BSK3 and BSK4 (in bsk3,4, bsk3,4,7, bsk3,4,6,7 and bsk3,4,6,8) or BSK3, BSK4, BSK7 and BSK8 (in bsk3,4,7,8 and bsk3,4,6,7,8) caused significant de-repression of DWF4 expression upon BL treatment. However, DWF4 regulation by BL was similar to wild-type in the bsk7,8 double mutant, in contrast to the altered regulation of SAUR-AC1 observed in this mutant. Several bsk mutations had an antagonistic effect on DWF4 expression. For example, when a mutation in BSK8 was introduced into the bsk3,4 background, which showed impaired DWF4 regulation by BL, repression of DWF4 in the resulting bsk3,4,8 triple mutant was similar to wild-type. In addition, as observed for the physiological response of mutant seedlings to BL treatment (Figure 3), the pentuple mutant bsk1,3,4,6,7 displayed a DWF4 expression pattern similar to that of the wild-type, in contrast to that of the bsk3,4,6,7 quadruple mutant. Finally, despite the significant reduction in DWF4 repression by BL in the bsk6 and bsk3,4 mutants, the bsk3,4,6 triple mutant displayed a DWF4 expression pattern similar to that of the wild-type.
BRI1 physically interacts with and phosphorylates several BSKs
BSK1 and BSK3 were previously shown to be phosphorylated by the BRI1 receptor kinase (Tang et al., 2008). Because BSK protein sequences are highly conserved, we hypothesized that additional BSKs function in BR signaling downstream of BRI1. To explore this possibility, we examined the ability of BRI1 to interact with and phosphorylate various BSKs. We employed bimolecular fluorescence complementation (BiFC) techniques to test in vivo interactions between BRI1 and BSK5, BSK6 or BSK11. The known interaction between BRI1 and BSK1 (Tang et al., 2008) was utilized as a positive control in these experiments. Epidermal cells of Nicotiana benthamiana leaves co-expressing BRI1 fused to the N-terminal half of YFP (nYFP) and BSK1, BSK5, BSK6 or BSK11 fused to the C-terminal half of YFP (cYFP) showed a strong fluorescence signal, indicating that multiple BSKs interact with BRI1 (Figure 6a). Cells co-expressing either BRI1–nYFP with cYFP–BIN2, or BAK1–nYFP with BSK1–cYFP, BSK5–cYFP, BSK6–cYFP or BSK11–cYFP, showed no fluorescence signals.
Next, we performed kinase assays to test the ability of BRI1 to phosphorylate BSK5, BSK6, BSK8 and BSK11. BSKs were expressed in Escherichia coli as glutathione-S-transferase (GST) fusion proteins, and their phosphorylation by the BRI1 kinase domain fused to maltose binding protein (MBP–BRI1-KD) was tested in vitro. All the tested GST–BSKs were phosphorylated by MBP–BRI1-KD, as previously observed for GST–BSK1 and GST–BSK3 (Figure 6b). BRI1 phosphorylated BSKs with similar efficiency with the exception of BSK3, which served as a weak phosphorylation substrate. These results support the hypothesis that multiple BSKs play a role in transducing the BR signal from the BRI1 receptor to downstream components of the pathway.
BSK1 Ser230, located in the activation loop of the kinase, was previously identified as the major site of BRI1 phosphorylation (Tang et al., 2008). To test whether serine residues corresponding to BSK1 Ser230 are phosphorylated by BRI1 in other BSKs (Figure 6c), we mutated BSK6 Ser210 and BSK8 Ser213 to alanine and monitored in vitro phosphorylation of the mutant proteins by BRI1. In contrast to phosphorylation of GST–BSK1S230A, phosphorylation of GST–BSK6S210A and GST–BSK8S213A by MBP–BRI1-KD was not significantly reduced compared to the wild-type proteins (Figure 6b), indicating that the site preferentially phosphorylated by BRI1 varies among BSKs and suggesting the existence of various mechanisms of BSK regulation by BRI1.
BSKs physically interact with BIN2 and other BSK family members
To identify additional proteins that physically interact with BSKs, BSK5, BSK6 and BSK11 were individually used as bait in a yeast two-hybrid screen of an Arabidopsis cDNA library. In these screens, BSK11 and BSK6 interacted with the negative regulator of BR signaling BIN2 and its homolog BIL2, respectively (Figure 7a). In addition, BSK5 interacted with the sibling protein BSK1 (Figure 7a). BiFC assays were then performed to test in vivo the interaction of BIN2 with various BSKs (i.e. BSK1, BSK6 and BSK8; BSK11 was not tested because the BSK11–nYFP fusion could not be expressed in plants). N. benthamiana epidermal cells expressing cYFP–BIN2 and BSK1–nYFP, BSK6–nYFP or BSK8–nYFP, but not BAK1–nYFP or BRI1–nYFP, showed strong fluorescence signals (Figure 7b), indicating that multiple BSKs interacted in vivo with BIN2. BiFC assays also validated the interaction between BSK1 and BSK5, and uncovered the ability of additional BSKs to interact with other family members. In fact, strong fluorescence signals were observed in cell expressing the following combinations of BSK–n/cYFP fusion proteins: BSK1–nYFP with BSK5–cYFP, BSK6–cYFP or BSK11–cYFP (Figure 7c), BSK6–nYFP with BSK1–cYFP, BSK5–cYFP or BSK11–cYFP (Figure 7d), and BSK8–nYFP with BSK1–cYFP, BSK5–cYFP, BSK6–cYFP or BSK11–cYFP (Figure 7e). No fluorescence signal was detected in cells co-expressing BSK1–nYFP, BSK6–nYFP or BSK8–nYFP with BAK1–cYFP (Figure 7c–e). Finally, strong fluorescence was observed in cells co-expressing BSK1–nYFP and BSK1–cYFP, or BSK6–nYFP and BSK6–cYFP (Figure 7f). These results suggest regulatory interactions between BSKs and BIN2, and raise the possibility that BSKs function as homo- or heterodimers.
The BIN2 and BIL2 GSK3-like kinases phosphorylate BSKs
BIN2 and its homolog BIL2 are GSK3-like kinases that have been previously shown to be catalytically active in vitro (Li and Nam, 2002; Yan et al., 2009). To explore the biological relevance of the interaction of BSKs with GSK3-like kinases, we tested whether BIN2 and BIL2 phosphorylate BSKs. Six BSK family members (i.e. BSK1, BSK3, BSK5, BSK6, BSK8 and BSK11) were expressed in E. coli as GST fusions. Phosphorylation of GST–BSKs by BIN2 fused to MBP (MBP–BIN2), and BIL2 fused to GST (GST–BIL2) was tested in vitro. All tested GST–BSK fusions were efficiently phosphorylated by MBP–BIN2 and GST–BIL2, but not by MBP–BIN2K69R and GST–BIL2K99R, which are kinase-deficient forms of the two enzymes (Figure 8). Under the same conditions, GST alone was not phosphorylated by MBP–BIN2, indicating that phosphorylation of GST–BSK fusions by MBP–BIN2 occurred in the BSK part of the protein (Figure S7). Although BSKs contain a putative kinase domain, no kinase activity was detected for the GST–BSK fusions. To further confirm the specificity of BSK phosphorylation by BIN2, we performed kinase assays in the presence of lithium, an inhibitor of GSK3-like kinase (Klein and Melton, 1996; Zhao et al., 2002). Phosphorylation of GST–BSK fusions by MBP–BIN2 and autophosphorylation of MBP–BIN2 were gradually reduced by increasing concentrations of lithium (Figure 9a). In addition, MBP–BIN2 did not phosphorylate a kinase-deficient mutant of the mitogen-activated protein kinase MPK3 (MPK3K70R) in tomato (Solanum lycopersicum) (Figure S8), and BSKs were not phosphorylated by the tomato kinase Pti1 (Figure S9).
Next, we tested whether Ser230, the major site of BSK1 phosphorylated by BRI1 (Figure 6) (Tang et al., 2008), is also a BSK1 phosphorylation site for BIN2 and BIL2. As shown in Figure 9(b), both GST–BSK1 and GST–BSK1S230A were equally phosphorylated by MBP–BIN2, as well as by GST–BIL2. This suggests that BSKs are phosphorylated on a residue other than Ser230, and are thus differentially modified by BRI1 and GSK3-like kinases.
bsk mutations affect sensitivity to brassinosteroids
A role in BR signal transduction for BSK family members was first identified by phosphoproteomic and biochemical analysis (Tang et al., 2008). Loss-of-function genetic evidence for the involvement of BSK genes in BR signaling has been provided for BSK3 only (Tang et al., 2008). In this study, examination of mutants in eight BSK genes confirmed the reduced sensitivity to BR for the bsk3 mutant, but no effect was observed in other single mutants. Introduction of additional bsk mutations into the bsk3 mutant decreased sensitivity to BR, demonstrating functional overlap between BSK3, BSK6, BSK7 and BSK8. The lowest BR sensitivity was achieved in the quadruple mutant bsk3,4,7,8, and was not enhanced by additional mutations in BSK6 (bsk3,4,6,7,8) or BSK1 (bsk1,3,4,6,7,8). Interestingly, BSK3, BSK4, BSK7 and BSK8 belong to the same clade of the BSK family phylogenetic tree and appear to play a prominent role in regulating BR signaling. While the effects of bsk mutations were generally additive, the bsk1 mutation displayed more complex interactions with other mutations. Coupling of bsk1 with bsk3,4,6,7 partially restored sensitivity to BR, implying an antagonistic interaction between these mutations. However, a similar effect was not observed when bsk1 was coupled with either bsk3, bsk6,7 or bsk3,4,6,7,8. To explain these observations, we propose a model where BSKs compete among themselves to establish productive interactions with BRI1 and differ in their signaling ability. Our genetic analysis supports the hypothesis that BSK1 is efficient in competition for BRI1, but is a weak signal transducer compared to other BSKs. The lower signaling ability of BSK1 is in agreement with the previous observation that the BSK1 gene, when over-expressed, is less efficient than BSK3 or BSK5 in rescuing the bri1-5 mutant phenotype (Tang et al., 2008). Based on these assumptions, introduction of a bsk1 mutation into the bsk3,4,6,7 background allows BRI1-mediated activation of a functional BSK (or BSKs) that is a weaker competitor for BRI1 but has a higher signaling ability than BSK1. This results in the higher BR sensitivity of bsk1,3,4,6,7 compared to bsk3,4,6,7. BSK8 is a possible candidate that comes into play when BSK1 is mutated, based on the evidence that a bsk8 mutation in the bsk1,3,4,6,7 background reduced the BR sensitivity of the resulting bsk1,3,4,6,7,8 mutants to bsk3,4,6,7 levels. According to this model, the bsk1 mutation does not display an antagonistic effect in bsk3 and bsk6,7 backgrounds because other functional BSKs in these mutants have a dominant effect on BSK1. Antagonistic effects were previously observed between members of gene families participating in hormone signaling, such as type A and type B Arabidopsis response regulators involved in cytokinin signaling (Mason et al., 2004; To et al., 2004). These effects suggest that, although members of a protein family play similar functions, they may have different biochemical characteristics. Therefore, patterns and levels of BSK expression may contribute to the nature and fine-tuning of the BR response. In this regard, analysis of gene expression using a publicly available database with the web-based application Geninvestigator (https://www.genevestigator.com/gv/plant.jsp) revealed that BSK1–BSK8 are expressed ubiquitously and at various levels in various tissues of Arabidopsis plants, while BSK9–BSK12 are expressed only in roots and reproductive tissues (Figure S10a). In addition, each of the BSKs displays a unique expression pattern at various developmental stages (Figure S10b).
The role of multiple BSKs as positive regulators of BR signaling was confirmed by determining the expression in bsk mutants of two BR-responsive genes: the BZR1 target gene DWF4 (He et al., 2005) and the BZR2/BES1 target gene SAUR-AC1 (Yin et al., 2002). Consistent with previous reports, BL treatment stimulated SAUR-AC1 expression, but inhibited the accumulation of DWF4 transcripts (Goda et al., 2002; Yan et al., 2009). As expected, the bsk3,4,7,8 and bsk3,4,6,7,8 mutants, which displayed the most pronounced reduction in BR sensitivity, showed a reduction in SAUR-AC1 induction and DWF4 repression. Interestingly, several bsk single mutants showed alterations of gene expression for one or both BR-responsive genes, suggesting that, although the corresponding genes are involved in BR signaling, the extent of their contribution to signal transduction is not sufficient to alter the BR sensitivity of the mutant. In addition, the differential effect of bsk mutations on SAUR-AC1 and DWF4 gene expression suggests that BSKs may generate different signa-ling outputs and affect the expression of different gene subsets.
bsk mutations affect plant growth and development
The most severe growth phenotypes were observed in the bsk3,4,7,8 and bsk3,4,6,7,8 mutants, which displayed reduced rosette size, leaf curling and enhanced leaf inclination. However, this morphological phenotype differed to some extent from that characteristic of mutants in central components of BR biosynthesis and signaling. Although the bsk3,4,7,8 and bsk3,4,6,7,8 mutants were indeed smaller than the wild-type control, their leaf morphology was quite different to that observed in BR biosynthesis and signaling dwarf mutants, which often have a compact rosette of short and round leaves (e.g. det2, Chory et al., 1991; cpd, Szekeres et al., 1996; dwf1, Takahashi et al., 1995; dwf4, Choe et al., 1998; bri1, Clouse et al., 1996; Li and Chory, 1997; bin2-1, Li and Nam, 2002). A possible explanation for this difference is that the mutated genes are involved only in a subset of BR responses, as suggested by the different SAUR-AC1 and DWF4 expression in the various single mutants. It is also likely that the observed phenotype reflects a functional overlap between the mutant genes and additional BSK genes that remain functional. Similarly, other signaling components acting downstream of BRI1 are encoded by small families of genes with redundant functions. For example, mutants in two of four members of the BAK1 family exhibited a weaker phenotype than bri1 alleles, while simultaneous mutations in three BAK1 family members caused a phenotype similar to a null bri1 mutant (Albrecht et al., 2008; Gou et al., 2012). Likewise, knockdown by RNAi of two BSU1-related phosphatases resulted in subtle developmental phenotypes (Mora-Garcia et al., 2004), while a loss-of-function mutant of all four family members showed a severe dwarf phenotype (Kim et al., 2009). Functional redundancy was also demonstrated for the GSK3-like kinase BIN2 and its homologs BIN2-like1 (BIL1) and BIN2-like2 (BIL2), and has been proposed for other GSK3-like kinases, based on the observation that a bin2 bil1 bil2 triple mutant still responded to BR (Yan et al., 2009). Alternatively, the differences between the phenotypes for bsk3,4,7,8 or bsk3,4,6,7,8 and known BR biosynthesis and signaling mutants may be ascribed to parallel functioning of BSKs and members of RLCK sub-family VIIc at the same node of the BR signaling pathway (Kim et al., 2011). BRI1 was shown to phosphorylate and activate the RLCK-VIIc CDG1, which in turn phosphorylates and activates BSU1 (Kim et al., 2011). Similarly, BSKs, which are phosphorylated by BRI1 and interact with BSU1, were proposed to mediate signal transduction from BRI1 to BSU1 (Kim et al., 2009). However, because over-expression of CDG1 did not complement the reduced BR sensitivity of a bsk3 mutant, their products probably play independent functions, even if a functional overlap between these genes exists (Kim et al., 2009). Finally, it is also possible that, in addition to their role in brassinosteroid signaling, certain BSK family members participate in additional physiological and developmental processes by targeting proteins other than BIN2 and related components of BR signal transduction, resulting in additional phenotypic changes. In support of this possibility, a mutation in BSK5 was recently reported to affect the response of Arabidopsis seedlings to salinity and abscisic acid (Li et al., 2012).
A striking phenotype observed in the bsk3,4,7,8 and bsk3,4,6,7,8 mutants was the inclination of the rosette leaves away from the soil surface. Although this phenotype was not previously reported for BR signaling or biosynthesis mutants in Arabidopsis, similar phenotypes have been observed in rice (Oryza sativa). For example, the rice BR signaling mutations d61-1 and d61-2, which are weak alleles of OsBRI1, a homolog of Arabidopsis BRI1, display a semi-dwarf phenotype and more erect leaves (Yamamuro et al., 2000). An erect leaf phenotype was also observed for rice brassinosteroid-deficient mutants, such as osdwarf4-1, brd1 and d2 (Hong et al., 2002, 2003; Sakamoto et al., 2006). BR was shown to influence leaf erectness in rice by enhancing cell division in the adaxial epidermis at the lamina joint (the junction between the leaf blade and the culm) (Yamamuro et al., 2000; Zhang et al., 2009). Recently, the Increased Leaf Inclination 1 (ILI1) protein, a helix-loop-helix transcription factor, was shown to be involved in BR-mediated elongation of rice lamina joint cells (Zhang et al., 2009). In addition, ILI1 and its interacting partner IBH1, an additional helix-loop-helix transcription factor, as well as their Arabidopsis homologs, were found to antagonize each other with respect to regulation of cell elongation and to be direct targets of the BR-regulated transcription factor BZR1 (Zhang et al., 2009). It is also possible that, in Arabidopsis, BSKs affect leaf inclination by participating in developmental pathways independent of the canonical BR pathway. Interestingly, BIN2 was recently found to directly regulate a mitogen-activated protein kinase cascade involved in stomatal development (Kim et al., 2012), and BAK1 family members were shown to control root growth via a BR-independent pathway (Du et al., 2012).
The role of BSKs in BR signaling
BSK1 and BSK3 were shown to interact with BRI1 and to be substrates of BRI1 phosphorylation (Tang et al., 2008). Here, we demonstrated that BSK5, BSK6, BSK8 and BSK11 are phosphorylated by BRI1, and BSK5, BSK6 and BSK11 also interact with BRI1. These data support genetic evidence indicating that multiple BSKs function as positive regulators of BR signaling. The major BSK1 site phosphorylated by BRI1 was identified as Ser230 (Tang et al., 2008), and its phosphorylation promotes interaction of BSK1 with the phosphatase BSU1 (Kim et al., 2009). In our experiments, mutations of BSK1 Ser230 reduced BRI1 phosphorylation by 87%, but mutations in the corresponding residues of BSK6 and BSK8 only reduced BRI1 phosphorylation by 12%. This indicates that the site preferentially phosphorylated by BRI1 varies among BSKs. It is possible that the Ser, Tyr and Thr residues that are adjacent in BSK6 and BSK8 to the Ser corresponding to BSK1 Ser230 are phosphorylated by BRI1, and their phosphorylation is functionally equivalent to that of BSK1 Ser230. Alternatively, phosphorylation of BSKs by BRI1 at different sites may reflect distinct mechanisms of activation, and result in different outputs or various degrees of signaling efficiency.
The BSK1–BSU1 interaction was proposed to result in activation of BSU1, which in turn deactivates the GSK3-like kinases BIN2 and BIL2 (Kim et al., 2011). However, how BSKs activate BSU1 remains unknown. Our observation that BSKs interact in vivo with BIN2 supports the possibility that BSKs function as a scaffold or docking platform facilitating the interaction between BSU1 and BIN2. A scaffold function for BSKs is in agreement with the lack of kinase activity detected for these putative kinases (this study; Bayer et al., 2009; Kim et al., 2011). Indeed, BSKs lack conserved residues that are considered essential for kinase activity (Kim et al., 2011). A scaffold activity has been proposed for kinases that are predicted to be catalytically inactive and are thus termed pseudokinases (Zeqiraj and van Aalten, 2010). For example, the mammalian kinase suppressor of Ras1 and 2 acts as scaffold, bringing together components of a mitogen-activated protein kinase cascade and regulating signaling output (McKay et al., 2009). Interestingly, scaffold proteins such as 14-3-3 family members function as homo- or heterodimers (Darling et al., 2005). We found that BSKs physically interact in yeast and in planta, suggesting that they form homo- or heterodimers. In a possible model, upon BRI1 activation, one BSK molecule of a dimer recruits BSU1 and the other BSK molecule of the dimer recruits BIN2, allowing efficient signal transfer from BSU1 to BIN2. However, it is possible that BSKs structurally compensate for the lack of conserved catalytic residues and acquire kinase activity upon phosphorylation by BRI1. Once activated, they may directly inactivate BIN2, bypassing BSU1, through a phosphorylation-mediated mechanism. Adding another level of complexity, we found that several BSKs are phosphorylated by BIN2 and BIL2. It is tempting to hypothesize that, under low BR concentrations, BIN2 family members phosphorylate and inactivate BSKs, similarly to their effect on BZR1 and BES1 transcription factors and on the mitogen-activated protein kinase kinase kinase YODA (Kim and Wang, 2011; Kim et al., 2012).
Plant material and growth conditions
The Arabidopsis thaliana T-DNA insertion lines in ecotype Col-0 were obtained from the Arabidopsis Biological Resource Center (Columbus, OH). Gene-specific primers were used with T-DNA-specific primers (Table S2) to identify and confirm T-DNA insertions by PCR (Alonso et al., 2003). Mutants of higher order were generated as detailed in Figure S4 by standard crossing procedures. Progeny homozygous for all the alleles were identified in the F2 and F3 generations by PCR. For phenotypic analysis, seeds were sown on Soilrite mix (Agrekal Habonim Industries, http://agrekal.co.il) and stratified for 2 days at 4°C in the dark. Plants were grown under either long-day (16 h light/8 h dark) or short-day (8 h light/16 h dark) conditions at 25°C in a growth chamber.
Sterilized seeds were sown in square plates on 0.5 x Murashige and Skoog (MS) medium with 1% w/v sucrose, supplemented with or without 24-epibrassinolide (BL) (1 μm) (Sigma-Aldrich, http://www.sigmaaldrich.com). Plates were cold-treated for 2 days at 4°C, and incubated in the vertical position in a growth chamber at 25°C under 16 h light/8 h dark conditions. Seven-day-old seedlings were scanned with a size reference (millimetric paper), and their hypocotyl and root lengths were measured using ImageJ (Collins, 2007). For quantitative RT-PCR analysis, 1-week-old seedlings were acclimated in liquid MS medium with constant shaking. After 4 h, BL (100 nm) was added to the medium. Samples were deep-frozen after 2 h treatment, and stored at −80°C.
Quantitative RT-PCR analysis
RNA was extracted from 1-week-old seedlings using the SV total RNA isolation system (Promega, http://www.promega.com). RNA samples were reverse-transcribed and used for quantitative RT-PCR (Melech-Bonfil and Sessa, 2010). The analysis was repeated at least three times with independent biological samples. The GAPDH gene (encoding glyceraldehyde 3-phosphate dehydrogenase) was used for normalization. Primers used for quantitative RT-PCR are listed in Table S2.
BSKs and BIL2 were fused to the C-terminus of glutathione-S-transferase (GST) in the pGEX-4T1 vector (GE Healthcare, http://www3.gehealthcare.com), whereas BIN2 and the BRI1 kinase domain were fused to the C-terminus of the maltose binding protein (MBP) in the pMAL-c2x vector (New England Biolabs, http://www.neb.com). E. coli BL21 (DE3) pLysS containing the desired vector were grown at 37°C to an OD600 of 0.6-0.8. Expression of recombinant proteins was induced using 0.5 mm isopropyl-d-thiogalactopyranoside for 6 h at 18°C. GST and MBP fusions were affinity-purified using glutathione–agarose beads (Sigma-Aldrich) and amylose resin (New England Biolabs), respectively.
Mutagenesis of BSKs, BIN2 and BIL2 fused to GST or MBP was performed in the plasmids pGEX-4T1 and pMAL-c2x using a QuikChange™ kit (Stratagene, http://www.stratagene.com) and the oligonucleotides listed in Table S2.
In vitro kinase assay
Kinase assays were performed by incubating 2 μg wild-type or mutant GST–BSK fusions at room temperature for 30 min in a mixture containing 20 mm Tris/HCl (pH 7.5), 10 mm MgCl2, 5 mm dithiothreitol, 20 μm ATP, 2 μCi [γ-32P]ATP (3000 Ci/mmol; PerkinElmer Inc., http://www.perkinelmer.com) and 0.5 μg MBP–BRI1, MBP–BIN2, MBP–BIN2K69R, GST–BIL2 or GST–BIL2K99R. Proteins were fractionated by 10% SDS–PAGE, stained using Coomassie Brilliant Blue, and exposed to autoradiography.
Yeast two-hybrid screens
BSKs were fused to the GAL4 DNA binding domain in the bait vector pAS1 or pGBT9, while BIN2, BIL2 and BSK1 were fused to the GAL4 activation domain in the prey vector pACT2 (Gietz et al.,1997). Yeast two-hybrid screens were performed using BSK5, BSK6 and BSK11 as baits, as described in Methods S1.
Bimolecular fluorescence complementation (BiFC)
BiFC expression vectors (Citovsky et al., 2006) were obtained from the Arabidopsis Biological Resource Center. BSK1, BSK5, BSK6, BSK8, BRI1 and BAK1 were cloned in-frame to the N- or C-terminal fragment of the enhanced yellow fluorescence protein (EYFP) in vectors pSAT1-nEYFP-N1 and pSAT1-cEYFP-N1 (Citovsky et al.,2006). BSK11 was cloned into pSAT1A-nEYFP-N1 and pSAT1A-cEYFP-N1, while BIN2 was cloned into pSAT1-nEYFP-C1 and pSAT1-cEYFP-C1. Expression cassettes were cloned into the binary vector pPZP-RCS2-nptII (Chung et al.,2005), and transformed into Agrobacterium tumefaciens strain GV2260. Overnight-grown Agrobacterium cultures were used for transient expression, and YFP signals were analyzed as described in Methods S1.
We thank Tamar Kuplik, Osnat Yanai and Ido Nir for technical assistance, Kevan Shokat (Department of Cellular and Molecular Pharmacology, University of California, San Francisco, CA, USA), Chao Zhang (Department of Chemistry, University of Southern California, Los Angeles, CA, USA) and Jasmina Allen (Department of Cellular and Molecular Pharmacology, University of California, San Francisco, CA, USA) for helpful discussions, and Sigal Savaldi-Goldstein (Department of Biology, Technion-Israel Institute of Technology, Haifa, Israel) and members of the Sessa laboratory for critical comments on the manuscript. This research was supported by a grant from the US–Israel Bi-national Science Foundation to G.S. and N.O. (BSF 2007091).