BRASSINOSTEROID-SIGNALING KINASE 3, a plasma membrane-associated scaffold protein involved in early brassinosteroid signaling

Brassinosteroids (BRs) are steroid hormones essential for plant growth and development. The BR signaling pathway has been studied in some detail, however, the functions of the BRASSINOSTEROID-SIGNALING KINASE (BSK) family proteins in the pathway have remained elusive. Through forward genetics, we identified five semi-dominant mutations in the BSK3 gene causing BSK3 loss-of-function and decreased BR responses. We therefore investigated the function of BSK3, a receptor-like cytoplasmic kinase, in BR signaling and plant growth and development. We find that BSK3 is anchored to the plasma membrane via N-myristoylation, which is required for its function in BR signaling. The N-terminal kinase domain is crucial for BSK3 function, and the C-terminal three tandem TPR motifs contribute to BSK3/BSK3 homodimer and BSK3/BSK1 heterodimer formation. Interestingly, the effects of BSK3 on BR responses are dose-dependent, depending on its protein levels. Our genetic studies indicate that kinase dead BSK3K86R protein partially rescues the bsk3-1 mutant phenotypes. BSK3 directly interacts with the BSK family proteins (BSK3 and BSK1), BRI1 receptor kinase, BSU1 phosphatase, and BIN2 kinase. BIN2 phosphorylation of BSK3 enhances BSK3/BSK3 homodimer and BSK3/BSK1 heterodimer formation, BSK3/BRI1 interaction, and BSK3/BSU1 interaction. Furthermore, we find that BSK3 upregulates BSU1 transcript and protein levels to activate BR signaling. BSK3 is broadly expressed and plays an important role in BR-mediated root growth, shoot growth, and organ separation. Together, our findings suggest that BSK3 may function as a scaffold protein to regulate BR signaling. The results of our studies provide new insights into early BR signaling mechanisms.

The BR signaling pathway has been studied in some detail, however, the functions of the BSK family proteins in the pathway have remained elusive.The founding members of this family are BSK1 and BSK2, which were identified as BR-responsive proteins by two-dimensional difference gel electrophoresis and mass spectrometry [15].The BSK protein family is composed of twelve members, which contain an N-terminal kinase domain and C-terminal tetratricopeptide repeat (TPR) motifs [15].Although called BRASSINOSTEROID-SIGNALING KINASES, the available results indicate that not all members are involved in BR signaling.SHORT SUSPENSOR (SSP/BSK12) activates the YODA mitogen-activated protein kinase pathway to regulate suspensor development during embryogenesis [29], which has not been shown to be regulated by BRs. Biochemical and genetic studies have implicated numerous BSK members (BSK1, 2, 3, 4, 5, 6, 8, and 11) as positive regulators in BR signaling [15,30].BRI1 phosphorylation of BSK1 at Ser230 promotes BSK1 interaction with BSU1 phosphatase [15,16], however, the mechanism underlying BSK1-mediated BSU1 activation is not known [17].bsk3/4/6/7/8 pentuple mutants exhibit multiple growth defects, including reduced rosette size, leaf curling, and enhanced leaf inclination [30].However, these mutants are phenotypically very different from BR deficient and response mutants, which exhibit severe growth defects, including dark green and rounded leaves, reduced male fertility, and reduced silique growth [3,[31][32][33].Together, the functional importance of the BSK family members in BR signaling and BR-mediated plant growth and development remains unclear.
To identify novel regulators in BR signaling and responses, we took a forward genetic approach to screen for mutations that cause decreased BR responses in Arabidopsis.Our genetic screen identified five semi-dominant mutations causing decreased BR responses.Through positional cloning, we determined that the mutations occurred in the BSK3 gene (At4g00710), which encodes a receptor-like cytoplasmic kinase.Our genetic, molecular, and biochemical studies demonstrate that BSK3 plays an important role in BR signaling and BRmediated plant growth and development.

A root-based genetic screen identifies five semi-dominant mutations in BSK3 causing decreased BR responses
Previous genetic screens for mutations that cause decreased BR responses in Arabidopsis only successfully identified two BR signaling regulators, BRI1 and BIN2 [3,18,31,34,35].The identified bri1 and bin2 mutants are dwarf or semi-dwarf plants that exhibit pleiotropic growth defects, including reduced rosette size, dark green and rounded leaves, and reduced male fertility.We reasoned that previous screens missed mutants that do not exhibit bri1/bin2-like shoot phenotypes.To identify mutants affecting novel loci involved in BR signaling, we carried out a root-based genetic screen for mutations that cause decreased BR responses in Arabidopsis.Wild-type Columbia (Col) seedlings exhibited reduced root growth when grown on 0.1 μM brassinolide (BL, the most active BR) (S1A Fig) .We screened ethyl methanesulfonate (EMS)mutagenized (in Col-0) and activation-tagged (in Col-2) populations.Seedlings on 0.1 μM BL exhibiting longer roots were selected.We named these plants brassinosteroid resistant (brr) mutants.Here, we report five brr mutants (brr1, brr2, and brr3 in Col-0; brr4 and brr5 in Col-2) (Fig 1A).To determine the recessive or dominant nature of these mutations, we crossed all brr mutants into the wild-type Col-0 or Col-2 plants.Surprisingly, both heterozygous and homozygous mutants were resistant to 0.01 and 0.1 μM BL.In addition, homozygous mutants had stronger BL resistance , indicating that all five brr mutations are semi-dominant.Although brr4 and brr5 were identified from an activation-tagged population, co-segregation analyses of BL and Basta resistance suggested that the mutations were not caused by T-DNA inserts.
We used positional cloning to identify the affected gene of the brr1 mutant, which exhibited slightly reduced root growth in the absence of BL (Fig 1F).This mutant is caused by a mutation in the BSK3 gene (At4g00710), which encodes a receptor-like cytoplasmic kinase [15].The second glycine of the protein was mutated to arginine (Fig 1D).To confirm that BSK3 is responsible for the mutant phenotypes, we expressed an HA-tagged BSK3 genomic DNA under the Since all five identified bsk3 mutations are semi-dominant mutations causing BSK3 loss-offunction, we were curious whether the known bsk3-1 T-DNA insertion mutant (SALK_ 096500) [15,30] is also semi-dominant.The T-DNA insertion of bsk3-1 is located in the first intron in the 5' UTR of BSK3 (Fig 1D).RT-PCR did not detect any BSK3 transcripts in bsk3-1 (Fig 1G ), indicating that bsk3-1 is probably a null mutant.We crossed the bsk3-1 mutant into the wild-type Col-0 plants.Both heterozygous and homozygous mutant seedlings exhibited resistance to 0.01 and 0.1 μM BL.In addition, homozygous mutant seedlings exhibited stronger BL resistance (Fig 1H and 1I).These results suggest that the bsk3-1 mutant is also semidominant, indicating a dose-dependent effect of BSK3 on BR responses.

BSK3 is an N-myristoylated protein that localizes to the plasma membrane
N-myristoylation is a co-translational lipid modification.The mechanisms underlying N-myristoylation are conserved among eukaryotes.Myristic acid is linked to a protein's N-terminal glycine residue via an amide bond, which is catalyzed by an N-myristoyltransferase [36][37][38].Analysis of BSK3 protein sequence using NMT-The MYR Predictor (http://mendel.imp.ac.at/ myristate/SUPLpredictor.htm) indicated that BSK3 has a putative myristoylation sequence at its N-terminus (Fig 2A), suggesting that it may be an N-myristoylated protein.To determine whether BSK3 is an N-myristoylated protein, we performed in vitro myristoylation assays.The TNT SP6 high-yield wheat germ protein expression system was used to synthesize HA-tagged BSK3 protein in the presence of [ 3 H]myristic acid, and reaction products were analyzed by SDS-PAGE, western blot, and fluorography.Wheat germ extracts contain the N-myristoyltransferase, which can modify proteins.Consistent with the prediction by NMT-The MYR Predictor, wild-type BSK3-HA was labeled (Fig 2B).The second glycine of BSK3, a putative myristoylation site, was mutated to arginine in the bsk3-2 mutant (Fig 2A).To determine whether this G2R mutation affects BSK3 myristoylation, we performed PCR-based sitedirected mutagenesis to create BSK3 G2R mutation and examined the myristoylation of this mutant protein.It was previously reported that the G2A mutation blocks the myristoylation of N-myristoylated proteins [39].As a control, we included BSK3 G2A -HA protein in the assays.Unlike the wild-type protein, BSK3 G2A -HA and BSK3 G2R -HA mutant proteins were not modified by [ 3 H]Myristic acid (Fig 2B), indicating that these two mutations block BSK3 myristoylation.
N-myristoylation confers a tendency to associate with membranes [37], indicating that BSK3 may be a membrane-associated protein.To examine BSK3 subcellular localization, we generated transgenic Arabidopsis plants expressing BSK3-mCitrine driven by the native BSK3 promoter.BSK3-mCitrine protein localized exclusively to the plasma membrane (Fig 2C).To determine whether N-myristoylation is required for BSK3 plasma membrane localization, we examined the subcellular localization of BSK3 G2A -mCitrine and BSK3 G2R -mCitrine proteins.Unlike the wild-type protein, these two mutant proteins were mislocalized in the nucleus and cytosol (Fig 2C).To further show that BSK3-mCitrine is a plasma membrane-localized protein while BSK3 G2A -mCitrine and BSK3 G2R -mCitrine mutant proteins are not, we performed subcellular protein fractionation.Consistent with our confocal observations, BSK3-mCitrine protein was only detected in the membrane fraction (Fig 2D).By contrast, BSK3 G2A -mCitrine and BSK3 G2R -mCitrine mutant proteins were only detected in the soluble fraction (Fig 2D).These results indicate that N-myristoylation is required for BSK3 plasma membrane localization.
To determine whether BRs regulate BSK3 subcellular localization, we treated light-grown BSK3pro:BSK3-GFP seedlings with 2 μM brassinazole (BRZ, a BR biosynthesis inhibitor) for four days and 1 μM BL for two hours, respectively.Seedlings treated with BRZ and BL showed BSK3-GFP localization exclusively on the plasma membrane in the hypocotyls and roots (Fig 2E ), indicating that BRs do not regulate BSK3 plasma membrane localization.Since the effects of BSK3 on BR responses are dose-dependent, depending on BSK3 protein levels (Fig 1H and  1I), we wondered whether BSK3 protein levels are regulated by BRs.Light-grown BSK3pro:

Kinase dead BSK3 K86R protein partially rescues the bsk3-1 mutant phenotypes
All BSK family proteins contain an N-terminal kinase domain, however, contradictory results have been reported regarding whether these proteins have kinase activity in vitro.Autophosphorylation of BSK1, 3, 5, 6, 8, and 11 were not detected in kinase assay reactions containing Mg 2+ or Mn 2+ [15,30,40].However, BSK1 was shown to exhibit kinase activity, requiring Mn 2+ as a divalent cation cofactor [41,42].A luciferase-based kinase assay using affinity-purified BSK8-GFP protein from Arabidopsis seedlings showed that BSK8 has kinase activity [43].Caution should be taken, however, regarding the detected BSK8 kinase activity, as kinases copurifying with BSK8-GFP may contribute to the detected kinase activity.OsBSK3, a BSK3 ortholog in rice, was shown to exhibit weak autophosphorylation activity in kinase assay reactions containing Mg 2+ or Mn 2+ .However, a kinase dead OsBSK3 protein was not included as a negative control.Furthermore, the authors could not confidently claim that OsBSK3 is an active kinase since the autophosphorylation signals were too weak [44].Together, all these results suggest that further research is needed to conclude whether the BSK family proteins are real kinases.
To determine whether BSK3 has kinase activity, we performed kinase assays to examine BSK3 autophosphorylation.Kinase active GST-BRI1-KD and kinase dead GST-BRI1-KD K911E proteins were included as controls.Like BRI1 receptor kinase, all BSK family proteins contain the highly conserved lysine (K) residue required for binding ATP and the catalytic activity (S2 A previous study reported that kinase dead SSP/BSK12 K78R protein could fully complement the short suspensor phenotype of the ssp mutants, suggesting that SSP kinase activity may not be required for SSP function in suspensor development during embryogenesis [29].We used the same strategy to determine whether BSK3 kinase activity, if it has kinase activity in planta, is required for BSK3 function in BR signaling.We selected two independent bsk3-1 transgenic lines expressing BSK3-HA or BSK3 K86R -HA driven by the native BSK3 promoter, which exhibited similar protein levels (Fig 3C).Like BSK3-HA, kinase dead BSK3 K86R -HA protein fully complemented the short root phenotype of the bsk3-1 mutant (Fig 3D).Unexpectedly, bsk3-1 BSK3-HA seedlings exhibited hypersensitivity to 0.1 μM BL in the root elongation assays, while bsk3-1 BSK3 K86R -HA seedlings still exhibited resistance to 0.1 μM BL, although considerably weaker than bsk3-1 seedlings (Fig 3D).These results indicate that kinase dead BSK3 K86R -HA protein only partially rescues the bsk3-1 mutant phenotypes, suggesting that BSK3 kinase activity, or at least BSK3 binding to ATP, may be required for full BSK3 function in BR signaling.As described later, BSK3 may function as a scaffold protein in BR signaling, we could not rule out that the BSK3 K86R mutation may affect the scaffold function and thereby impairs BSK3 function in BR signaling.Our genetic screen identified four mutations (R156K, N182AQAL insertion, G226E, and G238S) in the kinase domain of BSK3 that confer decreased BR responses, suggesting that the kinase domain is crucial for BSK3 function in BR signaling (Figs 1D and 3A).To assess how three missense mutations (R156K, G226E, and G238S) affect BSK3 function, we used the native BSK3 promoter to express BSK3 R156K -HA, BSK3 G226E -HA, and BSK3 G238S -HA proteins in the bsk3-1 mutant, and analyzed the ability of these mutant proteins to complement the bsk3-1 mutant.As described previously, expression of wild-type BSK3-HA protein under the control of the native BSK3 promoter complemented the bsk3-1 mutant (Fig 3D).bsk3-1 plants did not exhibit obvious shoot growth defects.However, bsk3-1 transgenic plants expressing BSK3 R156K -HA or BSK3 G226E -HA, but not BSK3 G238S -HA, exhibited slightly smaller rosette leaves (Fig 3E).Western blot analyses showed that bsk3-1 BSK3 R156K -HA and bsk3-1 BSK3 G226E -HA plants exhibited high levels of protein expression, while bsk3-1 BSK3 G238S  We noticed that bsk3-1 BSK3 G238S -HA plants expressed low levels of BSK3 G238S -HA protein.Twenty-six independent transgenic lines were screened, and Lines 8 and 26 exhibited highest protein expression.However, BSK3 G238S -HA protein levels were still much lower than those of BSK3 R156K -HA and BSK3 G226E -HA proteins (Fig 3F ), suggesting that the G238S mutation may affect BSK3 protein stability.To test this hypothesis, we analyzed bsk3-1 transgenic plants expressing BSK3-HA or BSK3 G238S -HA under the control of the native BSK3 promoter.Semi-quantitative RT-PCR showed that bsk3-1 BSK3-HA lines 5 and 24 exhibited similar levels of transcripts as those of bsk3-1 BSK3 G238S -HA lines 8 and 26 (Fig 3I).However, western blot analyses showed that BSK3 G238S -HA protein levels were much lower than those of BSK3-HA protein (Fig 3I).Together, these results suggest that the G238S mutation may indeed reduce BSK3 protein stability.

BSK3 physically interacts with BRI1 receptor kinase, BSU1 phosphatase, BIN2 kinase, and the BSK family members
BSK3 was previously shown to be a substrate of BRI1 kinase in vitro [15], however, the precise function of BSK3 in BR signaling is not known.To understand the role of BSK3 in BR signaling, we sought to identify BSK3 interactors by testing the interactions between BSK3 and the known BR signaling regulators: BRI1 receptor kinase, CDG1 kinase, BSU1 phosphatase, and BIN2 kinase.To know whether BSK3 can form a homodimer or heterodimer with the BSK family members, we also tested BSK3/BSK3 and BSK3/BSK1 protein interactions.We generated transgenic Arabidopsis plants co-expressing BSK3-HA and BRI1-GFP, CDG1-EYFP, BSU1-EYFP, BIN2-EYFP, BSK3-GFP, or BSK1-EYFP proteins.We immunoprecipitated BSK3-HA protein using an anti-HA antibody and detected the co-immunoprecipitated (co- We were curious whether BSK3 is a substrate of BIN2 kinase, and therefore searched for the GSK3 consensus phosphorylation motifs (S/T-X-X-X-S/T, X is any amino acid) in BSK3.BSK3 contains nine putative GSK3 consensus phosphorylation motifs (Fig 4F ), suggesting that BIN2 may be able to phosphorylate BSK3.To determine whether BSK3 is a BIN2 kinase substrate, we performed kinase assays using GST-BSK3, kinase active GST-BIN2, and kinase dead GST-BIN2 K69R proteins.A previous study showed that GST protein is not phosphorylated by BIN2 kinase, suggesting that it is not a BIN2 kinase substrate [21].BSK3 phosphorylation was observed when GST-BSK3 was incubated with GST-BIN2, but not with GST-BIN2 K69R (Fig 4G ), indicating that BIN2 kinase phosphorylates BSK3.Our results are consistent with the previous finding that BIN2 kinase phosphorylates BSK3 in vitro [30].Together, these findings suggest that BSK3 is a substrate of BIN2 kinase.

BSK3 is broadly expressed throughout plant development
To understand BSK3 function in BR-mediated plant growth and developmental processes, we examined BSK3 gene expression using the GUS (β-glucuronidase) reporter gene driven by the native BSK3 promoter.BSK3 was expressed in the cotyledons, hypocotyls, and roots of etiolated and light-grown seedlings (Fig 6A -6C).Interestingly, roots exhibited higher BSK3 expression than shoots in light-grown seedlings (

Increased dosage of BSK3 activates BR signaling, enhances plant growth, and confers defective shoot organ separation
To know the effects of BSK3 gain-of-function on BR signaling and plant growth and development, we generated transgenic Arabidopsis plants that express an extra copy of BSK3 gene under the control of the native BSK3 promoter (BSK3pro:BSK3-HA).We screened forty-two independent transgenic lines, and lines 6 and 38 expressing high levels of BSK3-HA protein were selected for phenotypic analyses (   To know whether increased BSK3 expression could suppress the growth defects of bri1-801 (GABI_227B07) and bin2-1 mutants, we crossed the BSK3pro:BSK3-HA line 38 transgene into these mutants.Western blots using an anti-BRI1 antibody were not able to detect the fulllength BRI1 protein in bri1-801 (S7B Fig) .If a truncated BRI1 protein is produced, this mutant protein would lack the region after LRR19, including the BR-binding island domain, transmembrane domain, and kinase domain [46].Therefore, this mutant BRI1 protein cannot perceive BRs to initiate BR signaling.Like the bri1-116 null mutant [4,24], bri1-801 plants exhibited severe growth defects (S7C-S7F Fig) .All these results suggest that bri1-801 is a null mutant.Consistent with the previous findings that BSK3 overexpression from the 35S promoter partially suppresses the dwarf phenotype of bri1-116 [15], BSK3-HA expression from the native BSK3 promoter also partially suppressed the dwarf phenotype of bri1-801 (S7C and S7F Fig) .In addition, our results further revealed that BSK3-HA expression partially suppressed all examined aspects of the growth defects of bri1-801, including root growth, shoot growth, and male fertility (S7C-S7F Fig) .These results indicate that BSK3 can activate BR signaling without a functional BRI1 receptor, and it is an important player that regulates BRI1mediated plant growth and developmental processes.
The homozygous bin2-1 mutant exhibits severe bri1-like growth defects, such as dwarf and male sterility [34].A previous study reported that BSK3 overexpression from the 35S promoter could not suppress the dwarf phenotype of bin2-1 [15].However, BSK3-HA expression from the native BSK3 promoter partially suppressed the growth defects of bin2-1.Etiolated bin2-1 BSK3-HA seedlings exhibited longer hypocotyls than those of bin2-1 seedlings (Fig 7F and  7G), and mature bin2-1 BSK3-HA plants were bigger than bin2-1 plants (Fig 7F and 7H).Although smaller than those of wild-type, the siliques of bin2-1 BSK3-HA plants were larger than those of bin2-1 plants and could set seeds (Fig 7H).Our findings demonstrate that BSK3-HA expression can partially rescue the growth defects of the bin2-1 mutant, including shoot growth and male fertility.
Our findings revealed that BSK3 is involved in upregulating BSU1 protein levels to activate BR signaling.To determine whether BRs can upregulate BSU1 protein levels, we treated lightgrown BSU1-EYFP seedlings with 0.01 and 0.

Discussion
Since their discovery in 2008 [15], the functions of the BSK family proteins in the BR signaling pathway have remained elusive.In this study, we focus on BSK3 and elucidate its function in BR signaling and plant growth and development.Our genetic screen identified a G2R missense mutation in the bsk3-2 mutant, causing BSK3 loss of function and decreased BR responses (Fig 1A -1D).Interestingly, the G2R mutation blocked BSK3 myristoylation and plasma membrane localization (Fig 2B -2D).These results suggest that plasma membrane localization is required for BSK3 function in BR signaling.In addition to the G2R mutation, our genetic screen also identified three additional missense mutations (R156K/bsk3-3, G226E/bsk3-4, and G238S/ bsk3-6) in BSK3 kinase domain, causing BSK3 loss-of-function and decreased BR responses (Figs 1D and 3A).The arginine residue corresponding to BSK3 R156 is absolutely conserved in all BSK family members, while the glycine residue corresponding to BSK3 G226 is also highly conserved in all BSK family members except BSK11 (S2 Fig) .Expression of BSK3 R156K -HA and BSK3 G226E -HA proteins under the control of the native BSK3 promoter in the bsk3-1 mutant conferred defective shoot growth, including smaller rosette leaves (Fig 3E ), reduced plant height (Fig 3G ), and smaller siliques (Fig 3H).These results suggest that BSK3 R156K -HA and BSK3 G226E -HA proteins may have dominant-negative effects, which interfere with the function of additional BSK family members.These mutant proteins may be non-functional and sequester BSK interactors, causing BSK loss-of-function.Unlike R156K and G226E, the G238S mutation appeared to destabilize BSK3 protein to reduce its levels in the bsk3-6 mutant, causing BSK3 loss-of-function (Fig 3I).The glycine residue corresponding to BSK3 G238 is absolutely conserved in all BSK family members (S2 Fig) .It will be interesting to know whether this glycine regulates the protein stability of other BSK family members.Together, our results demonstrate that the kinase domain is crucial for BSK3 function in BR signaling.
The TPR domain is a 34 amino acid structural motif involved in mediating protein-protein interaction and assembling a multiprotein complex, which has been found in a wide variety of proteins in all organisms [47][48][49].Previous studies have shown that TPR motifs are crucial for BSK1 and SSP/BSK12 functions in regulating plant defense responses and suspensor development during embryogenesis, respectively [29,41].A recessive R443Q missense mutation in the TPR motifs confers BSK1 loss-of-function and defective defense responses [41].TPR deletion completely inactivates SSP/BSK12, causing defective suspensor development during embryogenesis [29].More recently, TPR motifs were reported to negatively regulate OsBSK3 activity in BR signaling, as OsBSK3 TPR-Δ -GFP overexpression from the 35S promoter activates BR signaling more efficiently than OsBSK3-GFP overexpression [44].We demonstrate that the C-terminal three tandem TPR motifs are not essential for BSK3 function in BR signaling.However, TPR deletion impairs BSK3 function, suggesting that TPR motifs are required for the full function of BSK3 in BR signaling (Fig 5C).Therefore, different from the negative role of TPR motifs in the control of OsBSK3 activity in BR signaling [44], our results do not support a negative role of TPR motifs in regulating BSK3 activity in BR signaling.Interestingly, TPR deletion impairs BSK3/BSK3 interaction and BSK3/BSK1 interaction (Fig 5D ), suggesting that TPR motifs contribute to BSK homodimer/heterodimer formation.
GSK3 kinases are highly conserved in eukaryotes and regulate diverse physiological and developmental processes [50,51].In the BR signaling pathway, the GSK3-like kinase BIN2 phosphorylates BZR1 and BES1 transcription factors to inhibit BR signaling via multiple mechanisms, causing proteasomal degradation of BZR1 and BES1, inhibiting their DNA binding, and promoting their binding to 14-3-3 proteins to cause cytoplasmic retention of BZR1 and BES1 [19,21,52,53].We found that BSK3 directly interacted with BIN2 in Arabidopsis, and their interaction did not appear to be regulated by BRs (Fig 4C and 4D).Furthermore, BSK3 contained nine putative GSK3 consensus phosphorylation motifs and was a BIN2 kinase substrate (Fig 4F and 4G) [30].Interestingly, BIN2 phosphorylation of BSK3 enhanced BSK3 interactions with BSK1, BSK3, BRI1, and BSU1 (Fig 4H).These results suggest that BIN2 may have a positive role in BR signaling, promoting BSK homodimer/heterodimer formation, BSK3/BRI1 interaction, and BSK3/BSU1 interaction to enhance BR signaling.A similar positive role of GSK3 kinase in Wnt signaling was reported.GSK3 kinase phosphorylates and activates the Wnt co-receptor low-density lipoprotein receptor-related protein 6 (LRP6) to activate Wnt signaling [54].
The BSU1 protein family is composed of four members (BSU1, BSU1-LIKE 1/BSL1, BSL2, and BSL3) and positively regulate BR signaling [14,16,55].BSU1 phosphatase inactivates BIN2 kinase activity by dephosphorylating the phosphotyrosine residue pTyr200 of BIN2 to activate BR signaling [16].We and others have shown that BSK3 and BSK1 physically interact with BSU1 and maybe other BSU1 family members (BSL1, BSL2, and BSL3) to regulate BR signaling [16,55].However, the molecular mechanisms underlying BSK-mediated BSU1 activation is still not known.We found that BSK3-HA expression driven by the native BSK3 promoter upregulated BSU1-EYFP transcript levels via a post-transcriptional mechanism to increase BSU1 protein levels and thereby activate BR signaling (Fig 8B).These results reveal that an important mechanism to activate BR signaling is BSK3-mediated upregulation of BSU1 transcript and protein levels.The detailed mechanisms require further investigation.
Our genetic analyses showed that all five bsk3 loss-of-function mutants that we identified are caused by semidominant mutations (Fig 1 and S1 Fig).Interestingly, the bsk3-1 T-DNA loss-of-function mutant is also semidominant.Homozygous bsk3-1 seedlings exhibited stronger BR resistance than heterozygous bsk3-1 seedlings (Fig 1H and 1I), suggesting a dosedependent effect of BSK3 protein on BR responses.Expression of a kinase dead BSK3 K86R -HA protein under the control of the native BSK3 promoter fully rescued the root growth defect and partially rescued the BR resistance phenotype of the bsk3-1 mutant, suggesting that this mutant protein can still activate BR signaling to promote root growth (Fig 3D).These results provide experimental evidence to support that BSK3 may function as a scaffold protein to positively regulate BR signaling.Scaffold proteins are signal-organizing proteins that regulate a variety of signal transduction pathways.Their basic functions include assembling their interactors into specific protein complexes and localizing their interactors to specific subcellular compartments [56,57].As a scaffold protein in the BR signaling pathway, BSK3 may function as monomers, homodimers, and/or heterodimers with other BSK family members.BSK3 may sequester BSU1 phosphatase and BIN2 kinase at the inner face of the plasma membrane to enhance their interaction [16].BSK3 may also enhance the oligomerization of the BSU1 family proteins [58].In either case, BSU1 phosphatase may dephosphorylate the phosphotyrosine residue pTyr200 of BIN2 more efficiently to inactivate BIN2 kinase and thereby activate BR signaling.
Despite extensive efforts, we were not able to detect BSK3 autophosphorylation in kinase assay reactions containing either Mg 2+ or Mn 2+ (Fig 3B ), indicating that unlike BRI1 receptor kinase, BSK3 does not have kinase activity under our conditions.However, these results could not rule out that BSK3 may have kinase activity in planta.Supporting this hypothesis, the kinase dead K86R mutation impairs BSK3 function in BR signaling.Expressed at similar levels, BSK3 K86R -HA protein was less efficient to rescue the BR resistance phenotype of bsk3-1 seedlings than BSK3-HA protein (Fig 3C and 3D), suggesting that BSK3 kinase activity may be required for its full function in BR signaling.The mechanisms underlying BSK3 kinase activation in planta are not known.Based on the structural studies of BSK8 kinase domain [40], one possibility is that BSK3 binding to other proteins may reverse the catalytically inactive autoinhibition state of BSK3 to a catalytically active state.Alternatively, BSK3 phosphorylation by BRI1 receptor kinase may activate BSK3 kinase activity.This hypothesis is supported by the finding that OsBSK3 phosphorylation by OsBRI1 receptor kinase appears to enhance OsBSK3 autophosphorylation in vitro [44].Together, our findings suggest that BSK3 may function as a scaffold protein with possible kinase activity to regulate BR signaling.Future research to demonstrate BSK3 kinase activity, investigate BSK3 kinase activation mechanisms, and identify BSK3 physiological substrates in the BR signaling pathway will advance the understanding of BR signaling mechanisms.Alternatively, our results could not rule out that BSK3 may not function as a kinase.However, full BSK3 function in BR signaling requires ATP binding, which may modulate the scaffold function of BSK3.Supporting this hypothesis, some pseudokinases have been shown to bind ATP, which may have a structural or functional role in regulating pseudokinase activity in the absence of catalytic activity [59].
The BSK protein family is composed of twelve members.Previous studies using a loss-offunction approach to understand their role in BR-mediated plant growth and development were not successful [15,30].Despite reduced growth, bsk3/4/6/7/8 pentuple mutants do not exhibit dark green and rounded leaves, male sterility, and reduced silique growth, which are typical growth phenotypes of BR deficient and response mutants [3,[30][31][32][33].We performed phenotypic analyses on 10 bsk T-DNA insertion mutants (bsk1, 2, 3, 4, 5, 6, 8, 10, 11, and 12), including growth phenotypes and responses to 0.01 and 0.1 μM BL (S1 Table ).None of these mutants exhibited obvious growth phenotypes and BR resistance except that the bsk3-1 mutant exhibited slightly reduced root growth and BR resistance (Fig 1I and S6 Fig).Supporting an important function of BSK3 in root growth, BSK3 was highly expressed in seedling roots (Fig 6B).In addition, all five bsk3 mutants that we identified exhibited slightly reduced root growth (S6A and S6B Fig) .Our BSK3 gain-of-function approach revealed that in addition to root growth, BSK3 is also important for various shoot growth and developmental processes.BSK3-HA expression driven by the native BSK3 promoter conferred increased seedling hypocotyl growth and cauline leaf and inflorescence stem fusion (Fig 7A , 7B and 7E).In addition, BSK3-HA expression partially suppressed the growth defects of bri1-801 and bin2-1 mutants, including root growth, shoot growth, and male fertility (Fig 7F-7H and S7C-S7F Fig) .Interestingly, BSK3 R156K -HA or BSK3 G226E -HA expression under the control of the native BSK3 promoter in the bsk3-1 mutant conferred reduced rosette leaf growth, plant height, and silique growth (Fig 3E -3H).Together, our findings demonstrate that BSK3 plays an important role in BR-mediated plant growth and development, including root growth, shoot growth, and organ separation.Future research to generate mutant plants loss of all twelve BSK genes may provide further evidence in support of the crucial role of the BSK family proteins in BR-mediated plant growth and development.

Plant materials and growth conditions
All Arabidopsis thaliana mutants and transgenic lines were in the Columbia (Col) ecotype.Plants were grown under long-day conditions (16 h light/8 h dark) at 22 o C. Seeds were surface sterilized in a solution containing 30% bleach and 0.04% triton X-100 for 15 minutes, washed 3 times in sterile water, and cold treated for 3 days at 4 o C to synchronize germination.Seedlings were grown on ½ LS (Linsmaier and Skoog medium, Caisson Labs) plates containing 1% sucrose and 0.8% agar.

Genetic screen and positional cloning
EMS-mutagenized M2 seeds (Catalog nos.M2E-02-05 and M2E-01A-07, ~448100 M2 seeds from ~56012 M1 plants) and activation-tagged lines (Stock no.CS31100, ~62000 lines) were purchased from Lehle Seeds and Arabidopsis Biological Resource Center, respectively.Sterilized seeds were grown on ½ LS agar plates containing 0.1 μM brassinolide (BL) (Daiichi Fine Chemical).Seven-day-old light-grown BL resistant seedlings exhibiting long roots were selected.All candidate mutants were retested for BL resistance in the next generation.The semi-dominant brr1 mutant was crossed into the Landsberg erecta (Ler) ecotype to create a mapping population.DNA extractions were performed using a CTAB method [64] from BL sensitive F2 plants, and SSLP (simple sequence length polymorphism), INDEL (insertion-deletion polymorphism), and CAPS (cleaved amplified polymorphic sequence) markers were used to map the BSK3 mutation in the brr1 mutant by PCRs.

Co-immunoprecipitation (Co-IP) and western/far-western blots
Total protein extracts were prepared from Arabidopsis seedlings using an extraction buffer containing 50 mM Tris�HCl (pH 7.5), 150 mM NaCl, and 0.5% Igepal CA-630.Microsomal proteins were prepared from Arabidopsis seedlings or Nicotiana benthamiana leaves as described previously [65].The complete EDTA-free protease inhibitor cocktail (Roche Applied Science) was added to all buffers.All centrifugation steps were carried out at 4 o C. For protein fractionation to isolate microsomal proteins, total protein extracts were centrifuged for 30 minutes at 5000 × g, and supernatants were then centrifuged for 1 hour at 100000 × g.Co-IP and western blots were performed as previously described with slight modification [66].Protein G agarose beads (Roche Applied Science) were used to collect the immune complexes.The tris-buffered saline buffer (TBST, 0.05% tween 20, pH7.6) was used for western blots.Proteins were detected using the SuperSignal West Pico or Dura chemiluminescent substrates (Thermo Scientific).

RT-PCR
RNAs were prepared using the Spectrum™ plant total RNA kit (Sigma).An on-column DNase I digest set (Sigma) was used to remove contaminating DNA.Complementary DNAs (cDNAs) were synthesized using the SuperScript III first-strand synthesis system (Invitrogen).RT-PCR reactions were carried out using gene specific primers, and PCR products were analyzed by agarose gel electrophoresis.

In vitro myristoylation assays
BSK3 full-length cDNA without the stop codon was cloned into pTNT (HA) to generate a Cterminal HA epitope-tagged BSK3 construct.Site-directed mutagenesis PCRs were performed to create BSK3 G2A/G2R mutations.Proteins were synthesized using the TNT SP6 high-yield wheat germ protein expression system.[ 3 H]Myristic acid (tetradecanoic acid) (PerkinElmer) was added to the reactions at a concentration of 0.5 μCi/μl for radiolabelling.Reaction products were analyzed by SDS-PAGE, western blot, and fluorography.

In vitro kinase assays
BIN2 or BSK3 full-length cDNA was cloned into pENTR/D-TOPO.Similarly, BRI1 partial cDNA encoding the kinase domain containing a stop codon (BRI1-KD) was cloned into in the same vector.Site-directed mutagenesis PCRs were performed to create BSK3 K86R , BIN2 K69R , and BRI1-KD K911E kinase dead mutations.All DNA fragments were recombined into pTNT (GST) or pDEST15 (GST) to generate N-terminal GST-tagged fusion protein constructs.GST fusion proteins were synthesized using the TNT SP6 high-yield wheat germ protein expression system or expressed in E. coli BL21 (DE3) cells induced by 1 mM IPTG for 4 hours at 30 o C. All GST fusion proteins were purified using glutathione sepharose 4B beads according to the manufacturer's instructions.Purified GST fusion proteins were incubated in 50 μl kinase buffer (

GST pull-down assays
BSU1 full-length cDNA was cloned into pENTR/D-TOPO and recombined into pTNT (GST) to generate an N-terminal GST-tagged fusion protein construct.BSK1 full-length cDNA without the stop codon was cloned into pENTR/D-TOPO and recombined into pTNT (FLAG) to generate a C-terminal FLAG-tagged fusion protein construct.GST fusion proteins were synthesized using the TNT SP6 high-yield wheat germ protein expression system and purified using glutathione sepharose 4B beads.GST pull-down assays were performed in the PBS buffer (pH 7.4) containing 1 mM dithiothreitol and 0.1 to 1% igepal CA-630 at 4 o C.After 2 hours of incubation of GST fusion proteins and HA-or FLAG-tagged proteins, beads were washed 4 times with the same buffer.Finally, Beads were resuspended in 1x SDS-PAGE sample buffer and boiled for 5 minutes.The supernatants were analyzed by SDS-PAGE and western blots using anti-GST, anti-HA, and anti-FLAG antibodies.

Fig 2 .
Fig 2. BSK3 is an N-myristoylated protein that localizes to the plasma membrane.(A) A putative BSK3 Nmyristoylation sequence GGQCSSLSCCRNTSHKT predicted by NMT, the MYR predictor.(B) In vitro myristoylation assays.(C) Subcellular localization of BSK3 WT/G2A/G2R -mCitrine proteins.Root tip regions of 4-day-old light-grown seedlings were analyzed.(D) Western blots detecting BSK3 WT/G2A/G2R -mCitrine proteins.Proteins were extracted from 7-day-old light-grown seedlings.Twenty to thirty micrograms of total proteins (TP), soluable proteins (SP), and microsomal proteins (MP) were loaded.AHA (Arabidopsis H + -ATPases) proteins were detected by an anti-AHA antibody.(E) Subcellular localization of BSK3-GFP protein.Seedlings were grown on 2 μM BRZ in the light for 4 days, or 4-day-old light-grown seedlings were treated with 1 μM BL for 2 hours.Scale bars = 50 μm.(F) Western blots detecting BSK3-HA and BES1-HA proteins.Seedlings were grown on 2 μM BRZ for 6 days, or 6-day-old light-grown seedlings were treated with 1 μM BL for 2 hours.Thirty micrograms of total proteins were loaded.https://doi.org/10.1371/journal.pgen.1007904.g002 Fig).We performed PCR-based site-directed mutagenesis to create kinase dead BSK3 K86R mutation (Fig 3A and S2 Fig).GST-BRI1-KD protein exhibited kinase activity with an exposure time of two minutes, showing stronger autophosphorylation in the presence of Mn 2+ (Fig 3B).However, GST-BSK3 protein did not exhibit any autophosphorylation in the presence of either Mg 2+ or Mn 2+ despite an exposure time of three days (Fig 3B).These results indicate that unlike BRI1 receptor kinase, BSK3 does not have kinase activity under our conditions.

Fig 3 .
Fig 3. Kinase dead BSK3 K86R protein partially rescues the bsk3-1 mutant phenotypes.(A) BSK3 mutations in the kinase domain.The kinase domain (58-312 aa) of BSK3 was predicted by SMART (Simple Modular Architecture Research Tool, http://smart.embl-heidelberg.de/).(B) In vitro kinase assays.About 10-20 μg of GST fusion proteins were loaded for kinase assays.Dried protein gels were exposed to films for 2 minutes (GST-BRI1-KD and GST-BRI1-KD K911E ) or 3 days (GST-BSK3 and GST-BSK3 K86R ) at-80 o C. (C) Western blot analyses of BSK3-HA and -HA plants exhibited low levels of protein expression (Fig 3F).Consistent with the slight growth defect of rosette leaves, mature bsk3-1 BSK3 R156K -HA and bsk3-1 BSK3 G226E -HA plants were shorter than bsk3-1 plants (Fig 3G) and had smaller siliques (Fig 3H), indicating defective shoot and silique growth.The above results suggest that BSK3 R156K and BSK3 G226E mutations may have dominant-negative effects, interfering with the function of additional BSK members in BR signaling.

Fig 6 .
Fig 6.BSK3 is broadly expressed.(A) A 6-day-old etiolated seedling.The apical hook of the seedling was manually opened to better observe cotyledon GUS staining.(B and C) Five-day-old light-grown seedlings.(D) Lateral roots of a 9-day-old seedling.(E) Rosette leaves of a 12-day-old plant.(F) Cauline leaf of a 24-day-old plant.(G) An open flower.(H) A close look at an anther shown in (G).(I) Siliques.GUS staining was performed at 37 o C for 1 hour (B) or 24 hours (A and C-I).https://doi.org/10.1371/journal.pgen.1007904.g006