Copyright � 2010 by the Genetics Society of AmericaDOI: 10.1534/genetics.109.111898

Intragenic Suppression of a Trafficking-Defective BrassinosteroidReceptor Mutant in Arabidopsis

Youssef Belkhadir,‡‡,1,2,3 Amanda Durbak,†,3 Michael Wierzba,*,3 Robert J. Schmitz,*,1

Andrea Aguirre,* Rene Michel,* Scott Rowe,* Shozo Fujioka‡ and Frans E. Tax*,†,4

*Department of Molecular and Cellular Biology and †Department of Plant Sciences, University of Arizona, Tucson, Arizona 85721,‡RIKEN Advanced Science Institute, Saitama 351-0198, Japan and ‡‡Plant Biology Laboratory, The Salk Institute for

Biological Studies, La Jolla, California 92037

Manuscript received February 23, 2010Accepted for publication May 1, 2010

ABSTRACT

The cell surface receptor kinase BRASSINOSTEROID-INSENSITIVE-1 (BRI1) is the major receptor forsteroid hormones in Arabidopsis. Plants homozygous for loss-of-function mutations in BRI1 display areduction in the size of vegetative organs, resulting in dwarfism. The recessive bri1-5 mutation producesreceptors that do not accumulate to wild-type levels and are retained mainly in the endoplasmicreticulum. We have isolated a dominant suppressor of the dwarf phenotype of bri1-5 plants. We show thatthis suppression is caused by a second-site mutation in BRI1, bri1-5R1. The bri1-5R1 mutation partiallyrescues the phenotypes of bri1-5 in many tissues and enhances bri1-5 phenotypes above wild-type levels inseveral other tissues. We demonstrate that the phenotypes of bri1-5R1 plants are due to both increased cellexpansion and increased cell division. To test the mechanism of bri1-5 suppression, we assessed whetherthe phenotypic suppression in bri1-5R1 was dependent on ligand availability and the integrity of thesignaling pathway. Our results indicate that the suppression of the dwarf phenotypes associated withbri1-5R1 requires both BR biosynthesis and the receptor kinase BRI1-ASSOCIATED KINASE-1 (BAK1).Finally, we show that bri1-5R1 partially restores the accumulation and plasma membrane localization ofBRI1. Collectively, our results point toward a model in which bri1-R1 compensates for the protein-foldingabnormalities caused by bri1-5, restoring accumulation of the receptor and its delivery to the cell surface.

STEROIDS are important small-molecule hormonesthroughout the plant and animal kingdoms. The

structure of the first plant brassinosteroid (BR) wasdetermined almost 30 years ago (Grove et al. 1979),and the first major bioactivities associated with BRapplication were the stimulation of cell elongation andcell division (Mandava 1988). BRs are widely distrib-uted in the plant kingdom (Bishop 2003). BRs gainedattention as a major class of plant hormones after theanalysis of Arabidopsis mutants deficient in BR bio-synthesis (reviewed in Clouse and Feldmann 1999).These mutants showed major developmental defectsincluding dwarfism, dark-green leaves, male sterility,and delays in flowering and senescence. The dwarfism,which was mainly due to shorter organs such as leaves

and stems, could be partly reversed by supplementingmutant plants with end products of BR biosynthesis(Szekeres et al. 1996).

While steroids serve as signaling molecules in bothplants and animals, the mechanisms through which theyare perceived appear to be distinct. The primary animalsteroid receptors are transcription factors, such as theestrogen receptor, whose ligand binding triggers DNAassociation and the induction or repression of specificgenes (Rosenfeld and Glass 2001). The nuclearhormone receptors found in animals are likely notinvolved in plant steroid perception, as this family wasnot identifiable in the Arabidopsis genome (Arabidopsis

Genome Initiative 2000). Rather, analysis of steroidresponses in plants has revealed that the major mech-anism of steroid perception is via plasma membrane-spanning receptors (Li and Chory 1997; Li et al. 2002;Nam and Li 2002).

The cellular activities of BRs are mediated by theirinteraction with a specific cell-surface receptor thatbelongs to the leucine-rich repeat–receptor-like kinase(LRR–RLK) family, a group of structurally related plantreceptors (Dievart andClark 2003).BRASSINOSTEROIDINSENSITIVE-1 (BRI1) was identified through geneticscreens for Arabidopsis mutants displaying a dwarfstature that could not be rescued by exogenous BR

Supporting information is available online at http://www.genetics.org/cgi/content/full/genetics.109.111898/DC1.

1Present address: Plant Biology Laboratory, The Salk Institute forBiological Studies, La Jolla, CA 92037.

2Present address: Australian Research, Council Centre of Excellence inPlant Energy Biology, University of Western Australia, Crawley, WA 6009,Australia.

3These authors contributed equally to this work.4Corresponding author: Department of Molecular and Cellular Biology,

1007 E. Lowell, University of Arizona, Tucson, AZ 85721-0106.E-mail: fetax@u.arizona.edu

Genetics 185: 1283–1296 (August 2010)

treatment (Clouse et al. 1996; Li and Chory 1997). BRI1contains an extracellular domain (ECD) composed of anN-terminal signal sequence followed by 25 LRRs (Figure1C; Li and Chory 1997). LRRs form a highly reiterativestructure shaped like a horseshoe, providing surfaces forinteraction with ligands or other molecules (Bell et al.2003). BRI1 co-immunoprecipitates with active BRs, andphotoaffinity experiments demonstrate that the bindingsite for BRs is an atypical LRR, LRR22, and a 70-amino-acid ‘‘island’’ found between LRRs 21 and 22 (Kinoshita

et al. 2005). The ECD of BRI1 is followed by a singletransmembrane pass and a cytoplasmic Ser/Thr kinasedomain (Figure 1B), the latter of which is maintained inits basal resting state through the activities of both its C-terminal tail and a membrane-associated protein, BRI1Kinase Inhibitor-1 (BKI1) (Friedrichsen et al. 2000;Wang et al. 2005a,b; Wang and Chory 2006). Recently,the kinase domain of BRI1 has been shown to belong tothe superfamily of dual-specificity kinases and is able todisplay, in addition to its Ser/Thr kinase activity, a Tyrkinase activity in planta (Oh et al. 2009). The Tyr kinaseactivity of BRI1 is required for full receptor function andacts in an additive fashion to the Ser/Thr kinase activity.Thus, the early activation of the pathway depends on theordered and regulated autophosphorylation of BRI1 byboth its Ser/Thr- and Tyr-kinase activity, suggesting thatdistinct phosphorylation states may fine-tune the re-ceptor signaling capacity.

In vivo cellular biology approaches showed that theBRI1 protein is localized primarily to the plasmamembrane, where the perception of steroids is likelyto occur (Friedrichsen et al. 2000; Kinoshita et al.2005; Geldner et al. 2007). In addition to its plasmamembrane localization, BRI1 is present in intracellularendosomal compartments (Russinova et al. 2004;Geldner et al. 2007). While the distribution of BRI1in the endosomal pools is not affected either byexogenous treatment with BRs or by endogenous BRdepletion, several lines of evidence indicate that BRI1has the ability to signal from these intracellular com-partments. This intracellular localization of the recep-tor is reminiscent of TGF-b signaling in animals whereendocytosis of the TGF-b receptor is necessary for signaltransmission (Raikhel and Hicks 2007). Taken to-gether, the BR insensitivity of bri1 mutants, the plasmamembrane localization of the receptor, and the ability ofBRs to bind to the extracellar LRRs demonstrate thatBRs are sensed by the LRR module of BRI1 at the cellsurface (Belkhadir and Chory 2006). However, itremains unclear how the binding of BRs regulates thephosphorylation state of the receptor.

Initially, the only types of mutants identified inscreens for BR-insensitive plants were loss-of-functionmutations in BRI1. This suggests that additional positiveregulators in this pathway are lethal or redundant. Bothsecond-site suppressor screens and biochemical methodswere used to bypass these genetic limitations to iden-

tify other BR-signaling components. BRI1-ASSOCIATEDKINASE-1 (BAK1) was one locus identified throughalternate approaches (Li et al. 2002; Nam and Li

2002). A yeast two-hybrid screen using the BRI1 kinasedomain as a prey identified BAK1 as a BRI1 interactor(Nam and Li 2002). In addition, BAK1 was also found tobe a dose-responsive suppressor of the weak BRI1 allelebri1-5 in an activation-tagging screen (Li et al. 2002). Theweak dwarf phenotype associated with loss-of-functionbak1 alleles is likely due to the overlapping functionsof its paralogs, a small subfamily of five LRR–RLKscalled the SOMATIC EMBRYOGENESIS RECEPTORKINASES (SERKs) (Hecht et al. 2001; Albrecht et al.2008). A striking structural difference between BRI1and BAK1 lies in the number of LRRs that they possess.BAK1 has only five extracellular LRR repeats and noobvious platform for a BR-binding domain (Belkhadir

and Chory 2006). Furthermore, the binding of BRs toBRI1 has been shown to be BAK1 independent. Thebinding of BRs to preformed BRI1 oligomers is imme-diately followed by the activation of the kinase domain,leading to autophosphorylation of the receptor andsubsequent phosphorylation of BAK1 after the displace-ment of BKI1 from the plasma membrane (Wang et al.2005a,b; Wang and Chory 2006).

Currently, the only known downstream factors ofBRI1–BAK1 complexes are the BRASSINOSTEROIDSIGNALING KINASES (BSKs) (Tang et al. 2008).Recent models propose that, in the presence of BRs,the phosphorylation of BSK1 by BRI1 promotes thebinding of BSK1 to the phosphatase BRI1 Suppressor-1(BSU1). Consequently, BSU1 is able to inactivate theBRASSINOSTEROID INSENSITIVE-2 (BIN2) kinase (Li

and Nam 2002; Vert and Chory 2006) by targetingBIN2 for dephosphorylation (Kim et al. 2009). TheBSU1-dependent inactivation of BIN2 allows the tran-scription factors BRI1 EMS SUPPRESSOR-1 (BES1) andBRASSINAZOLE RESISTANT-1 (BZR1) to activategrowth responses (Vert and Chory 2006). BZR1 playsa critical role in feedback signaling by repressing BRbiosynthetic genes transcriptionally in response to BRperception (He et al. 2005). Conversely, in the absenceof BRs, BIN2 phosphorylates BES1and BZR1, whichtargets them for degradation (Yin et al. 2002; He et al.2005).

While there is strong evidence that the kinasedomains of BRI1 and BAK1 physically interact andphosphorylate each other, the interactions betweentheir ECDs are less clear. Many LRR–RLKs contain twopairs of cysteines in their ECDs: one at their aminoterminus and one near the transmembrane domain.These have been proposed to function in the homo-dimerization or heterodimerization of RLKs (Dievart

and Clark 2003), although there is no direct evidencefor this yet. Recent findings favor a model in which thedisulphide-bonded cysteine caps flanking both ends ofan LRR platform would function by protecting the

1284 Y. Belkhadir et al.

exposed edges of the hydrophobic core formed by thecanonical LRRs and by helping the LRRs to fold alongan N-terminal polarized pathway in which the N-terminalcysteine-rich capping domain is used as a nucleationpoint for proper folding to proceed (Courtemanche

and Barrick 2008; Truhlar and Komives 2008).The relative lack of missense mutations in the ECD of

BRI1 as compared to its kinase domain suggests thatthere are fewer key individual amino acids in the ECD ofBRI1 compared to the cytosolic domain. Alternatively,the LRR platform, because of its versatility, may tolerateamino acid substitution better than the kinase domain.The bri1-9 mutation, which confers weak insensitivity toBRs, is located in the 22nd LRR and may interfere withligand binding (Noguchi et al. 1999; Kinoshita et al.2005). However, analysis of an allele-specific suppressorof bri1-9 revealed that BRI1-9 is a functional receptor,but is retained in the Endoplasmic Reticulum (ER) bythe collective action of a plant-specific calreticulin(CRT3) ( Jin et al. 2009) and an a1,6 mannosyltransfer-ase (Hong et al. 2009) for targeted degradation by aproteasome-mediated process ( Jin et al. 2007; Hong

et al. 2008, 2009).To identify essential regions of BRI1, we have per-

formed a suppressor screen with bri1-5, a hypomorphicmutation in BRI1 that changes one of the pairedcysteines in the amino-terminus of BRI1 to a tyrosine.bri1-5, like bri1-9, encodes a receptor variant that doesnot accumulate to wild-type levels and is retained in theER, likely by similar processes ( Jin et al. 2007; Hong et al.2008, 2009). In this study, we show that one of oursuppressor mutants, bri1-5R1, is an intragenic revertantof bri1-5. bri1-5R1 is a mutation in the first LRR of BRI1,only 20 amino acids from the bri1-5 mutation. Ouranalysis of bri1-5 and bri1-5R1 phenotypes reveals thatbri1-5R1 plants display many intermediate phenotypescompared to wild-type and bri1-5 plants. However, bri1-5R1 also confers several gain-of-function phenotypes,including increased pedicel length. We demonstratethat the bri1-5R1 phenotypes require BR biosynthesisand the coreceptor BAK1. Our results suggest that bri1-5R1 acts primarily by allowing the trafficking-defectivereceptor to exit the ER and to reach the cell surfacewhere it can perceive its ligand.

MATERIALS AND METHODS

Isolation, mapping, and sequencing of bri1-5R1: Mutagen-esis: Seeds homozygous for bri1-5 in the Wassilewskija-2 (Ws-2)accession were mutagenized with EMS as described previously(Zhao et al. 2002)

Mapping: One revertant strain (4-8) was intermediate inheight between bri1-5 and Ws-2 plants. The strain 4-8 wasoutcrossed five times to segregate out additional mutationsand crossed to bri1-9 plants in the Columbia accession forRFLP mapping. DNA was isolated from F2 plants resembling4-8 and tested using nga 1107, an RFLP marker closely linkedto BRI1 (Bell and Ecker 1994).

Complementation test with bri1-4: To test whether 4-8 was anallele of BRI1, we crossed 4-8 plants to plants heterozygous forbri1-4, a null allele of BRI1 (Noguchi et al. 1999). Half of the F1

plants were wild type in appearance (bri1-5 4-8/1) and halfshowed the 4-8 phenotype (bri1-5 4-8/bri1-4). The genotypes ofthe two phenotypic classes were confirmed with PCR (seebelow).

Sequencing of bri1-5R1: DNA sequencing, including PCRfragments and primers, was performed as described by(Noguchi et al. 1999). The bri1-5R1 mutation generates anew EcoRI site, which was used to distinguish between mutantand wild-type alleles.

Phenotypic comparison of bri1-5 and bri1-5R1 to wild-typeplants: Growth of plants: Approximately five seeds were plantedin round pots (5 cm in diameter) with soil (4 parts Sunshinemix 3, Sungro Horticulture:1 part vermiculite, CAS# 1318-00-9, Therm-O-Rock West) presoaked in water. Flats containingthe pots were covered in plastic wrap and cold treated for 3–4 days before transfer to a Conviron MTR30 growth chamber[16 hr of light and 8 hr of dark at 22�, with 75% humidity, and alight intensity of �120 mE/m2 from cool-white fluorescenttube lamps (Philips F96/CW/VHO) supplemented by incan-descent bulbs]. Plastic wrap was removed after the seedlingswere established (after 3–5 days), and one representativeseedling was allowed to grow per pot. The pots were sub-irrigated with Hoagland’s nutrient solution as necessary.When the plants were 5 weeks of age, some of the morpho-logical traits listed in Table 1 were measured. Other traits weremeasured after the plants had ceased growth. Plant height wasmeasured to the nearest millimeter, and the length of siliques,the distance between siliques on the main inflorescence, andthe length and width of leaves were measured to the nearesthalf-millimeter using a ruler. Differences among genotypeswere tested using analysis of variance (ANOVA).

Seedling responses to BRs and other chemicals: Plate assays wereperformed using surface-sterilized seeds, which were soaked in70% ethanol and 0.1% Triton X-100 for 10 min and thenrinsed twice with 95% ethanol. Seedlings were grown on 1%agar plates with 0.53 Murashige and Skoog (MS) media and0.5% 2-(N-morpholino)ethane sulfonic acid, pH 5.7. Plateswere sealed with Micropore surgical tape (3M), stratified inthe dark at 4� for 72 hr and grown vertically at 22�, with 16 hrlight/8 hr dark, under cool-white fluorescent tubes (PhillipsAlto F40CW/RS/EW) with an average output of 90 mE/m2.For root-length assays, plates were scanned with a HPScanJet5370C and roots were measured using National In-stitutes of Health (NIH) ImageJ (NCBI). BR root-growthinhibition for Ws-2, bri1-5, and bri1-5R1 was assayed as inClouse et al. (1996) on plates containing 0, 5 3 10�5, 1 3 10�4,5 3 10�4, 1 3 10�3, 5 3 10�3, 1 3 10�2, 5 3 10�2, 0.1, 0.5, and1 mm epi-brassinolide (epiBL) (Sigma E1641), measuring after7 days of growth; n . 20 for each data point 6 1 SD. Seedlingresponses to brassinazole and kifunensine (Kif) were assayedas described (Wang et al. 2002; Jin et al. 2007).

Histological analysis: Stem and pedicel tissue was collectedfrom 3-week-old plants. Stem samples were collected frombetween the first and second siliques, and pedicels werecollected from the first and second siliques. The samples wereembedded in 1% agarose and fixed in 2% gluteraldehyde in13 phosphate-buffered saline at room temperature for 3 hr.The samples were then dehydrated using an ethanol seriesof 10, 20, 30, 40, 50, 60, 70, 80, and 90% ethanol for 15min, followed by three washes in 100% ethanol. After de-hydration, LR White embedding resin (Electron MicroscopySciences, Hatfield, PA) was added in 1/4 volume to 100%ethanol every 4 hr, followed by three changes of 100% LRWhite. After infiltration, the samples were embedded ingelatin capsules or BEEM narrow-neck molds (Electron Mi-

bri1-5 Intragenic Suppressor 1285

croscopy Sciences) and baked at 61� overnight. Sections�2- to7-mm thick were created using a Sorvall MT2B Ultramicro-tome, stained with 0.1% methylene blue, and visualized under35 magnification on a Zeiss Axioplan compound microscopeequipped with a QImaging Micro Publisher 3.3 RTV camera.Measurements and counts from 10 plants were performedusing NIH ImageJ Software (NCBI). All graphs were preparedusing Microsoft Excel, and error bars represent 1 SD aboveand below the mean. Statistical differences in average celllengths were analyzed using ANOVA.

Analysis of BR levels: Aerial parts of soil-grown plants (5 weeksold, �30 g fresh weight; see Phenotypic comparison of bri1-5 andbri1-5R1 to wild-type plants, above, for growing conditions) wereharvested and lyophilized. The tissues were extracted twicewith 500 ml of methyl alcohol. Deuterium-labeled internalstandards were added to the extracts. Purification andquantification of sterols and BRs were carried out accordingto the method described by (Fujioka et al. 2002).

Protein analysis and EndoH treatment: Two-week-old seedlingswere first ground in liquid nitrogen with a mortar and pestle.This material was then homogenized by alternate rounds ofPolytron (Kinematica) in 2 ml of sterile buffer (20 mm Tris–HCl, pH 8.0, 0.33 m sucrose, 10 mm EDTA, 5 mm dithiothrei-tol, and plant protease inhibitor cocktail (Roche)/1 g oftissue). Debris was removed by centrifugation at 2000 3 g for10 min at 4�. The supernatant of the 2000 3 g spin was thencentrifuged at 20,000 3 g for 45 min. For detection of BRI1,the pellet from the 20,000 3 g spin was extracted in 2.53

Laemli buffer. For immunodetection, samples were electro-phoresed on 6% SDS–polyacrylamide gels. Western blots wereperformed using standard methods and detected with ECL1

(Amersham). Proteins were subjected to deglycosylation assaysby treatment with endoglycosidase H by following the manu-facturers’ instructions.

Genetic interactions with BR biosynthetic and signalingmutants: bri1-5R1 plants were crossed with plants containingmutations in bin2 (dwf12-1D; Choe et al. 2002), dwf4-1 (Choe

et al. 1998), cpd-5 (Choe 2004), and bak1-2 (Li et al. 2002) toassay for genetic interactions. The F1 generation was screenedvia PCR for plants that were doubly heterozygous to ensurethat the cross was successful, and these plants were then selfedto obtain double homozygous mutants. For dwf4-1 and cpd-5,which are unlinked to bri1-5R1, strains homozygous for bri1-5R1 were identified, and phenotypes of the progeny that weresegregating for dwf4-1 or cpd-5 were monitored. Since BAK1and BIN2 are linked to BRI1, the progeny from these crosseswere planted and then screened for F2 plants that wereheterozygous for one mutation and homozygous for the otherusing PCR (see below). These F2 plants were self-crossed, andprogeny seeds were collected and planted to obtain doublehomozygous mutants. Double mutants were confirmed byPCR and digestion of products. For identifying insertionmutants, PCR was performed using two primers that amplifythe genomic DNA and a third border primer that amplifies theT-DNA/plant DNA junction. For point mutants, cleaved ampli-fied polymorphic sequences (CAPS) or derived CAPS (dCAPS)primers were made using the dCAPS program (Neff et al. 2002),and PCR-amplified DNA was cut with an enzyme that distin-guishes mutants from wild type. The progeny of the doublehomozygous mutants were then characterized phenotypicallyand photographed. All mutations except cpd-5 were originallyisolated in the Ws-2 accession. Our results with cpd-5 and dwf4-1were very similar, indicating that the Enkheim accession back-ground of cpd-5 did not influence these genetic interactions.

PCR primers and CAPS enzymes: The following primers andenzymes were used in our study:

bak1-2 primers: sense—59-CTTCCAAGTCTTAATCTGATGGGCCTTTA-39; anti-sense—59-GCAGGTGATGGCGGTGTAGGAGAGATAGG-39; JL-202 (left border)—59-CATTTTATAATAACGCTGCGGACATCTAC-39;

bri1-4 CAPS primers: sense—59-TGTTTCTCTCAAACTCACACATCAA-39; anti-sense—59-CGAATTGGTTACTAGAGATGTTCAA-39; bri1-4 contains a 10-bp deletion, 458 bp down-stream of ATG, and results in the loss of a BsmAI site;

bri1-5 dCAPS primers: sense—59-CCGTGTACTTTCGATGGCGTTACCT-39; anti-sense—59-CAAGCTGGTTAAAGAAGCAGAGCAC-39; bri1-5 contains the mutation G206A (C69Y).The underlined base in the sense primer is an introduceddCAPS mismatch causing the gain of a PstI site in wild type,but not in bri1-5.

bri1-5R1 CAPS primers: sense—59-TAATCAGAAGAAGAGGTAAC-39; antisense—59-TTCCCGGAGATGTCAAGATG-39;bri1-5R1 contains the mutation G260A (G87E) and resultsin the gain of an EcoRI site;

bin2-1 CAPS primers: sense—59-TCTTGGTCAGGTAAACAATTCTTTCAGT-39; antisense—59-AAAGAAACTGAAACAAGAACACATGCAA-39; bin2-1 contains the mutation G989A(E264K) and results in the loss of an MboII site.

RESULTS

Identification of an intragenic suppressor mutationin the BRI1 gene: To understand how the BRI1 receptorkinase connects BR perception to cellular responses, wecarried out a suppressor screen by looking for revertantsof the dwarf phenotype of plants homozygous for theweak bri1-5 mutation. In addition to identifying fiveextragenic suppressors in BES1 (Zhao et al. 2002), wealso found a suppressor, initially named 4-8, with aunique phenotype. This phenotype includes a semi-dwarf stature intermediate between that of bri1-5 plantsand that of the wild type, rounded leaves, and longpedicels (Figure 1A and Table 1). Plants derived fromseeds of the original 4-8 strain were outcrossed to plantsof the Wassilewskija-2 (Ws-2) accession. The F1 progenyfrom these crosses were wild type in appearance, in-dicating that 4-8 is recessive. Plants resembling theoriginal 4-8 plants were found in an �1:3 ratio (nine4-8:24 wild type) in the F2 from these crosses, and noplants resembling bri1-5 were identified. This indicatedthat the 4-8 mutation was homozygous in the originalstrain and suggested that 4-8 was linked to bri1-5.

To more precisely determine the linkage of the 4-8mutant phenotypes to bri1-5, 4-8 mutants were crossedto plants homozygous for the weak bri1-9 mutation inthe Columbia accession ( Jin et al. 2007). The F1 fromthis cross phenotypically resembled 4-8 plants, indicat-ing that 4-8 is dominant to the bri1-9 allele. Analysis ofDNA from 40 F2 plants with a bri1-9 phenotype identi-fied only one recombinant chromosome for the markernga1107, which is tightly linked to BRI1.

To test whether 4-8 was an allele of BRI1, we performeda complementation test with a null allele for BRI1. Plantshomozygous for 4-8 were crossed to plants heterozygousfor the bri1-4 mutation, a 10-bp deletion in the extracel-

1286 Y. Belkhadir et al.

lular domain of BRI1 (Noguchi et al. 1999). Approxi-mately half of the progeny from this cross are expected tobe 4-8/1, with the remainder being 4-8/bri1-4. Half ofthe progeny were wild type in appearance, and theother half resembled 4-8, indicating that 4-8 fails tocomplement bri1-4 and is an allele of BRI1. PCR analysisof the 10-bp deletion in bri1-4 and of the 4-8 mutationconfirmed that plants resembling 4-8 were 4-8/bri1-4.

We sequenced the BRI1 coding sequence in 4-8 andidentified, in addition to the bri1-5 mutation, a mutationin the first LRR (G260A, Gly87Glu). The first LRR inBRI1 has been described as a partial LRR because it doesnot contain all of the residues typical of the LRRconsensus for BRI1 (Li and Chory 1997). The glycinealtered in bri1-5R1 is highly conserved in the LRRs ofBRI1 and other prototypical RLKs (Figure 1D). This newmutation was named bri1-R1, and when present togetherwith the original bri1-5 mutation, is called bri1-5R1; thisterminology will be used here to refer to the revertantstrain.

Phenotypic characterization of bri1-5R1 plants: Thebri1-5R1 mutant was selected as a revertant because of itsincreased size relative to bri1-5 (Figure 1A). Beforefurther investigating the morphology of bri1-5R1 plants,we performed five backcrosses to the wild-type Ws-2accession to remove possible background mutations.Progeny of the final backcross plants were grown withbri1-5 and Ws-2 plants under controlled conditions inwhich plant organ sizes were measured and compared(Table 1). In general, bri1-5R1 plants were intermediatebetween bri1-5 and Ws-2 plants for specific traits such asplant height, distance between internodes, siliquelength, and seed number per seedpod. However, bri1-5R1 plants had longer pedicels than bri1-5 or Ws-2plants. bri1-5R1 plants had slightly more sepals andpetals than both bri1-5 and Ws-2 plants and also pro-duced more valves in the mature fruit than Ws-2 plantsproduced, but a similar number of valves were alsogenerated by bri1-5 plants.

The increased size of bri1-5R1 plants compared tobri1-5 plants appeared to be due to an increase in thelengths of internodes. To test if this increase was dueto increased cell expansion, cell division, or both,transverse sections of stems were made, and epider-mal cell length was measured. We found that themean stem epidermal cell length in the bri1-5R1plants was statistically shorter than Ws (0.0654 6

0.0231 mm vs. 0.109 6 0.0518 mm) and longer thanbri1-5 (0.0371 6 0.0148 mm, P , 0.01, one-wayANOVA), indicating that bri1-5R1 is intermediate incell length between Ws-2 and bri1-5; however, weobserved a large variation in cell length in both Ws-2and bri1-5R1 (Figure 2, A–C). Therefore, we per-formed a more detailed analysis of the distributionof cell length and found that, while the range of celllengths observed in both Ws-2 and the mutants variedgreatly, the majority of cells in bri1-5R1 were interme-

diate in length between bri1-5 and Ws-2 and that bri1-5R1 never produces cells as long as Ws-2 (Figure 2G).These results suggest that the size increase seen instems from bri1-5R1 plants is caused by the reversionof the cell expansion defects of bri1-5.

bri1-5R1 mutants also generated significantly longerpedicels as compared to Ws-2 (Table 1). To determinewhether this phenotype is also due to an increase in cellexpansion, we sectioned and measured pedicel epider-mal cell length (Figure 2D, E and F). We found that,while there was no significant difference in average celllength between Ws-2 (0.0759 6 0.0406 mm) and bri1-5(0.0747 6 0.0363 mm, P , 0.01 one-way ANOVA), theaverage cell length in bri1-5R1 pedicels was significantlyhigher than Ws-2 and bri1-5 (0.0912 6 0.0433 mm, P ,

0.001 one-way ANOVA). However, when the distributionof cell lengths was examined for pedicel cells, we foundthat it was only slightly skewed toward having a greaternumber of more expanded cells in bri1-5R1 as opposedto Ws-2 and bri1-5 (Figure 2H). These results suggestthat, unlike stems, the increased pedicel length ob-served in bri1-5R1 is mostly due to an increase in celldivision, possibly aided by a small increase in cellexpansion.

bri1-5R1 plants are insensitive to BR: Because bri1-5R1 plants show reversion of many bri1-5 phenotypes,we tested if bri1-5R1 plants have increased sensitivity toepiBL, as compared to bri1-5 plants. In a root-growthinhibition assay, Ws-2 plants showed a typical responseto epiBL, with low concentrations from 0.05 nm to0.01 mm showing increased primary root length andwith concentrations .0.01 mm showing an inhibition ofprimary root length. Like bri1-5, bri1-5R1 plants showedreduced primary root growth in the absence of exog-enous epiBL (Figure 3A). Unlike Ws-2, both bri1-5 andbri1-5R1 showed no inhibition of root growth withinthe concentrations tested, and at concentrations .0.01mM, bri1-5R1 plants showed a trend toward increasedroot growth in response to epiBL treatment (Figure3A).

bri1-5R1 mutants have partially restored levels ofbrassinosteroid intermediates: Mutants in many steroidsignaling genes have been found to accumulate ste-roids, suggesting that perception is linked via feedbackregulation to biosynthesis (Noguchi et al. 1999; Choe

et al. 2002; Wang et al. 2002). To test whether thereversion of the dwarf phenotype of bri1-5 was accom-panied by a corresponding change in the regulation ofsteroid synthesis, we examined the levels of BR bio-synthetic intermediates in bri1-5R1 plants, bri1-5 plants,and Ws-2 plants (Figure 3B; supporting information,Table S1). The values for bri1-5 and Ws-2 vary slightlyfrom the values reported previously (Noguchi et al.1999), although the general pattern is very similar. Thedifferences may be due to alterations in growth con-ditions. Brassinolide, the end product of the pathway,was still detectable in bri1-5R1 plants, but was reduced

bri1-5 Intragenic Suppressor 1287

Figure 1.—A comparison of bri1-5R1 phenotypes with other mutants in steroid signaling, genetic interactions with steroid-sig-naling mutants and a comparison of other known mutations in extracellular LRRs of RLKs. (A) Photographs of 6-week-old Ws-2,bak1-2, bri1-5R1, bri1-5, bak1-2 bri1-5R1, bak1-1 bri1-5, bin2-1, bin2-1 bri1-5R1, and bri1-4 plants. bak1-1 bri1-5 plants were described inLi et al. (2002). (B) Schematic of the domain organization of BRI1. The locations of the missense mutations bri1-5, bri1-R1, andbri1-9 and the deletion mutation bri1-4 in the extracellular domain of BRI1 are indicated by triangles. (C) Location of mutations inextracellular LRRs from Arabidopsis RLKs. The BRI1 consensus sequence is from Li and Chory (1997), the clv1 mutations (D toN, G to E) from Dievart et al. (2003), FLS2 from Gomez-Gomez and Boller (2000), elg from Whippo and Hangarter (2005),and bri1-9 from Noguchi et al. (1999). (D) Clustering of the bri1-5R1 mutation in the highly conserved N-terminal Cys cappingdomain. (Top) Alignment of the N-terminal Cys capping domains of AtBRI1 and bri1-5R1. (Middle) Alignment of the N-terminalCys capping domains of BRI1 orthologs across dicots (Middle top) and monocots (Middle bottom), highlighting the residuesmutated in our bri1-5R1 allele in red. Proteins compared were the following: dicots—LeBRI1 (tomato), SbBRI1 (potato), NbBRI1

1288 Y. Belkhadir et al.

approximately sixfold from the level found in bri1-5plants. bri1-5R1 plants displayed lower levels of 6-oxoBRsthan did bri1-5 plants, but around the same levels of 6-deoxoBRs as bri1-5 plants displayed. Since BR signalingrepresses BR synthesis, we conclude from these resultsthat the bri1-5R1 mutation increases the signalingcapacity of BRI1-5.

Our phenotypic results together with our BR meas-urements suggested that the phenotypes of bri1-5R1mutants result from increased signaling throughBRI1-5. We hypothesized that this increased signalingcould be due to (1) ligand-independent activation ofBRI1, (2) a bypass of BAK1 function, or (3) re-accumulation of BRI1-5 through ER exit.

BR biosynthesis is required for bri1-5R1 suppressionof bri1-5 morphology: We tested whether or not thephenotypic suppression of dwarfism observed in bri1-5R1 plants was dependent on the availability of BRs.First, we disrupted the BR biosynthetic pathway byusing a genetic approach. We constructed strains thatwere homozygous for bri1-5R1 and for cpd-5 or dwf4-1,loss-of-function alleles for two different enzymes in theBR biosynthetic pathway (Choe et al. 1998; Choe

2004). The triple mutants were indistinguishable fromthe cpd-5 and dwf4-1 single mutants (Figure 1A),indicating that bri1-5R1 phenotypes do require wild-

type BR biosynthesis. Second, we depleted the endog-enous pools of BRs by using a chemical inhibitor. Weused brassinazole (BRZ), a triazole compound thatspecifically blocks BR biosynthesis by acting on theproduct of the DWF4 gene (Asami et al. 2001). Wetested whether bri1-5R1 plants were BRZ insensitive byassessing their ability to de-etiolate. bri1-5R1 plantsgrown in the dark in the presence of 1 mM of BRZ hadshort hypocotyls and looked like bri1-5-treated seed-lings. At the same dose, Ws-2 plants also displayedshorter hypocotyls compared to untreated plants butappeared to be more resistant to BRZ when comparedto bri1-5 and bri1-5R1 (Figure 4).

An intact signaling pathway is required for bri1-5R1suppression of bri1-5 morphology: To test the relation-ship between bri1-5R1 phenotypes and BAK1, we crossedbak1-2 into bri1-5R1. Triple mutants for bri1-5R1 andbak1-2 resembled bri1-5 single mutants in stature (Figure1A, Table 1), indicating that BAK1 is required for thesize reversion of bri1-5 to bri1-5R1 and that bri1-5R1 doesnot act to bypass BAK1. The suppression of the longerbri1-5R1 internodes by bak1-2 is consistent with a re-quirement for BAK1 in bri1-5R1 reversion of bri1-5.However, the bri1-5R1 long pedicel phenotype was onlypartially suppressed in the triple bri1-5R1 bak1-2 mutant(Table 1).

TABLE 1

Morphometric analysis of Ws-2, bri1-5, bri1-5R1, bak1-2, and bri1-5R1bak1-2 mutants

Attribute Ws-2 bri1-5 bri1-5R1 bak1-2 bri1-5R1bak1-2

Heighta (in cm, 7 weeks old, n . 18) 34.9 6 4.64 9.88 6 1.32 21.9 6 3.18 36.2 6 5.38 9.79 6 1.85Distance between siliques on

main inflorescencea (in cm, n . 50)0.89 6 0.4 0.26 6 0.18 0.60 6 0.49 1.07 6 0.69 0.31 6 0.26

Length of pedicelsb (in cm, n . 50) 1.21 6 0.25 1.16 6 0.17 1.94 6 0.23 0.91 6 0.15 1.48 6 0.14Length of siliques (in mm, n ¼ 50) 11.9 6 0.83 7.08 6 0.43 10.2 6 1.4No. of rosette leaves (n ¼ 20) 5.85 6 1.0 7.65 6 1.17 8.35 6 0.88Width of rosette leaves (in cm, n . 115) 0.86 6 0.35 1.0 6 0.35 0.93 6 0.29Length of rosette leaves (in cm, n . 115) 2.6 6 0.8 1.8 6 0.6 2.6 6 0.8No. of cauline leaves (n ¼ 20) 1.7 6 0.8 2.2 6 1.03 3.3 6 0.74Width of cauline leaves (in cm, n . 39) 0.49 6 0.3 0.45 6 0.2 0.43 6 0.14Length of cauline leaves (in cm, n . 39) 1.9 6 0.44 1.84 6 0.44 2.2 6 0.51No. of seeds (n ¼ 50 siliques) 50.0 6 4.8 30.3 6 3.6 40.4 6 5.8No. of floral organs (n ¼ 200):

Sepals** 4.02 6 0.12 4.02 6 0.14 4.22 6 0.44Petals** 4.02 6 0.14 4.02 6 0.14 4.13 6 0.40Anthers* 5.88 6 0.39 5.96 6 0.23 6.04 6 0.34Valves*** 2.01 6 0.08 2.22 6 0.47 2.21 6 0.40

Each value represents the mean 6 SD. *P , 0 .05, one-way ANOVA for all three genotypes; **P , 0.5, one-way ANOVA that bri1-5R1 is different from Ws-2 and bri1-5; ***P , 0.05, one-way ANOVA that bri1-5R1 and bri1-5 are different from Ws-2.

a P , 0.05, one-way ANOVA that Ws-2 and bak1-2 are not different, bri1-5 and bri1-5R1 bak1-2 are not different, and bri1-5R1 isdifferent from all others.

b P , 0.05, one-way ANOVA that Ws-2 and bri1-5 are not different from each other, but all the rest are different.

(Nicotiana benthamiana, tobacco), and NtBRI1 (Nicotiana tabacum, tobacco); monocots—HvBRI1 (barley), TaBRI1 (wheat), andOsBRI1(rice). (Bottom) Alignment of the N-terminal Cys capping domains of other closely related LRR–RLKs in Arabidopsis,including BAK1, a coreceptor for BRI1. AtBRL1 and BRL3 are BRI1 paralogs and CLV1 and FLS2 are phylogenetically relatedto BRI1. Residues related to this study are shown in red.

bri1-5 Intragenic Suppressor 1289

These results suggested that bri1-5R1 suppresses bri1-5phenotypes by increasing signaling through the BRresponse pathway. To further test the mechanism ofsuppression by bri1-5R1, we tested whether bin2 mutantscould block the morphological suppression mediatedby bri1-5R1. The triple mutant bri1-5R1 bin2 plantsresembled bin2 plants (Figure 1) or bin2 bri1-5 (Li andNam 2002) plants, indicating that the mutated BRI1-5R1 receptor signals through BIN2. Thus, the ability ofbri1-5R1 to suppress the phenotypes associated with bri1-5 relies on both the availability of BRs and the propertransduction of BR signals.

bri1-5R1 acts as a suppressor by modulating BRI1-5stability or accumulation: bri1-5 encodes a partially

functional receptor that does not accumulate to wild-type levels because it is retained in the ER where it issubjected to the endoplasmic reticulum–associateddegradation (ERAD) process (Hong et al. 2009). Thus,to further investigate the recovery of BRI1-mediatedsignaling in bri1-5R1 plants, we tested whether theinherent instability of the BRI1-5 protein is suppressedby bri1-5R1. We monitored the accumulation of BRI1 inthe Ws-2 accession and in bri1-5 and bri1-5R1 plants byusing an affinity-purified anti-BRI1 antibody. BRI1accumulated to readily detectable levels in Ws-2, butwas severely reduced in bri1-5 (Figure 5A). BRI1accumulation was restored to almost wild-type levels inbri1-5R1 plants, which indicates that the levels of BRI1

Figure 2.—Analysis of cell division and cellexpansion in bri1-5 and bri1-5R1 mutants. Trans-verse sections of Ws-2 (A), bri1-5R1 (B), and bri1-5 (C) stems and Ws-2 (D), bri1-5R1 (E), and bri1-5(F) pedicels stained with 0.1% methylene blueto visualize cells. Bars, 0.1 mm. Due to the varia-tion in cell length, we analyzed the distributionof cell lengths for Ws-2, bri1-5, and bri1-5R1 stemepidermal cells (G) and pedicel epidermal cells(H). In the stems, a majority of cells in bri1-5plants were very short, whereas in bri1-5R1 mostwere intermediate, and in Ws-2, the cells weremore evenly distributed with more cells beinglonger in length, supporting the hypothesis thatthe height differences observed in these plantsare due to cell expansion defects. In the pedicels,no significant difference was observed in the dis-tribution between Ws-2 and bri1-5 plants, whilethe bri1-5R1 genotype was found to have morecells of longer length.

1290 Y. Belkhadir et al.

are rate limiting for proper BR responses and that bri1-R1 suppresses bri1-5 by allowing the steady-state level ofBRI1-5 to increase.

bri1-5R1 acts as a suppressor by modulating thelocalization of BRI1-5: Intuitively, BRI1 exit from theER, atop general accumulation, is necessary for BRsignaling from the cell surface. To assess whetherreceptor localization contributes to the restorationof BR signaling in bri1-5R1 plants, we monitoredBRI1-5R1 exit from the ER by exploiting the fact thatBRI1 is a glycoprotein. We used receptor sensitivity toendoglycosidase H (EndoH) as a reporter for putativelocalization. Typically, plasma membrane-associatedglycoproteins are resistant to EndoH digestion whereasER-retained glycoproteins are sensitive (Hong et al.2009). We isolated and treated microsomal extractsfrom wild-type and bri1-5 and bri1-5R1 plants withEndoH. The treated bri1-5 protein extracts produced asingle band of high electrophoretic mobility whenprobed with anti-BRI1 antibodies, consistent with thepreviously reported ER retention of BRI1-5 (Hong et al.2009) (Figure 5B). The EndoH treatment of wild-typeextracts also produced one single band, but of low

electrophoretic mobility. This electrophoretic pattern isconsistent with the proper delivery of wild-type recep-tors to the cell surface. The treatment of bri1-5R1protein extracts with EndoH produced two bands,indicating that in bri1-5R1 plants the receptor exits theER, but is also retained in the ER (Figure 5C). While thevast majority of the protein appeared to be EndoHresistant, a significant fraction was EndoH sensitive.This is consistent with a dual localization of the receptorat the ER and at the plasma membrane in bri1-5R1mutant plants.

Kifunensine does not enhance the ability of bri1-5R1mutants to elongate: During the maturation of glyco-proteins, the trimming of N-linked glycans acts as amolecular timer that monitors proper protein folding.In the ER and Golgi, terminally misfolded glycoproteinsbecome fully trimmed by 1,2-mannosidases, whichtargets them for destruction. When these mannosidasesare inhibited, for example, by the mannosidase in-hibitor Kif, more protein can escape ER quality controland reach the plasma membrane. We believed that thiseffect would promote a more wild-type receptor distri-bution in bri1-5R1 plants, allowing us to see whether

Figure 3.—bri1-5 and bri1-5R1 responses to BRs. Accumulation of BRs is reduced in bri1-5R1 mutants compared to bri1-5 mutants.(A) BR root-growth inhibition for Ws-2, bri1-5, and bri1-5R1 on plates containing 0, 5 3 10�5, 1 3 10�4, 5 3 10�4, 1 3 10�3, 5 3 10�3, 1 310�2, 5 3 10�2, 0.1, 0.5, and 1 mm epi-brassinolide measured after 7 days of growth. n . 20 for each data point 6 1 SD. (B) Purificationand quantification of sterols and BRs from aerial parts of 5-week-old soil-grown plants carried out according to the method describedby Fujioka et al. (2002). See Table S1; for raw data. All values are in nanograms/gram fresh weight.

bri1-5 Intragenic Suppressor 1291

bri1-5R1 is a hypermorphic mutation. Validating ourapproach, the Kif treatment of bri1-5 seedlings both letsmore BRI1-5 receptor escape the ER and promotesseedling elongation (Figure 5, C and D). However,applying Kif to bri1-5R1 seedlings did not cause addi-tional or faster elongation. Thus, bri1-5R1 does notappear to act as a gain-of-function in the control of thede-etiolation process mediated by the BR signalingpathway.

DISCUSSION

The binding of BR to the LRRs of BRI1 most likelyoccurs on the cell surface. This cell surface receptor–ligand interaction activates a fairly well-defined intra-cellular signal transduction cascade that primarilypromotes cell elongation. The analysis of bri1 mutantsblocked at successive steps of the secretory pathway,such as bri1-5 and bri1-9, has shown that BRI1 abundanceat the cell surface is critical for optimal BR perceptionand hence for BR signaling. The delivery of newlysynthesized BRI1 receptors to the plasma membrane is

controlled by the activity of at least three independent,but interrelated, protein-folding quality control path-ways. Here, we report that the aberrant accumulationand cellular distribution of the BRI1-5 protein, which isin abnormally low quantities at the plasma membraneand largely accumulates in the ER, is restored to almostwild-type levels by an intragenic suppressor mutationthat we named bri1-R1. We show that the restoration ofboth protein levels and protein localization is correlatedwith the suppression of the morphological phenotypesassociated with bri1-5. Interestingly, this intragenic re-version also allows discrete but important organ-specificgain-of-function phenotypes to take place. We proposethat bri1-R1 suppresses bri1-5 (1) by antagonizing theactivities of three ER resident protein-folding pathwaysby an unknown mechanism and (2) by increasing thesignaling activity of the BR pathway above the wild-typethreshold in specific cell types.

LRRs as a mutational target in BRI1: In bri1-5, thesecond Cys residue of the N-terminal conserved Cys pairis substituted by a Tyr (C69Y). This mutation is associ-ated with morphological phenotypes reminiscent of BRbiosynthetic mutants, indicating that these cysteinesplay an important role in vivo. The Cys capping domainsare a feature common to many plant extracellular LRRproteins. Initially, this pair of cysteines was proposed tobe involved in the formation of disulfide bonds that aidin homo- or heterodimerization of receptor-like kinases.This model remains attractive yet speculative (Dievart

and Clark 2003). An emerging model, based mainly onin vitro folding studies, suggests that these Cys pairs areinvolved in forming specific disulfide bonds to helpnucleate and stabilize the fold of the first LRR into anenergetically favorable conformation. Consistent withthis model, the C69Y mutation in bri1-5 forces BRI1 tolocalize in the ER because of its improper folding. Here,we demonstrate that the mutation of a conservedglycine to a glutamic acid (bri1-R1; G87E) located inthe closest LRR to the amino-terminal paired cysteineslargely compensates for bri1-5 defects. On the basis of acomparative and quantitative morphometric analysisthat we conducted on wild-type, bri1-5, and bri1-5R1plants (Table 1), we show that most phenotypes of bri1-5R1 plants are intermediate between bri1-5 and wild-type plants (Table 1), consistent with the second-sitemutation restoring receptor localization and not en-hancing receptor activity.

Little is known about the functions of the LRRs inBRI1. In contrast, there are .15 mutations isolated inthe smaller kinase domain and �10 in the islanddomain, which is 70 amino acids long (Li and Chory

1997; Noguchi et al. 1999; Friedrichsen et al. 2000).Thus, in formal genetic terms, the LRRs appear to bealmost dispensable for function. The only other LRRmissense mutation identified in BRI1 is bri1-9, a weakmutant caused by a missense mutation in the 22nd LRR(Figure 1, B and C; Noguchi et al. 1999). Thus, bri1-R1 is

Figure 4.—BR biosynthesis is required for bri1-5R1 etiola-tion phenotypes. (A) Morphology of 5-day-old dark-grownseedlings of Ws-2 (wild type) and isogenic bri1-5 and bri1-5R1grown on half-strength MS medium in the absence (�) orpresence (1) of 1 mm BRZ. This experiment is representativeof three independent replicates. (B) Lengths of 5-day-olddark-grown seedling hypocotyls in the absence or presenceof 1 mm BRZ. Means and standard deviations were calculatedfrom 40 seedlings. This experiment is representative of threeindependent replicates.

1292 Y. Belkhadir et al.

the second mutation found by forward genetics in theLRRs of BRI1, which comprise more than half of thetotal length of the BRI1 protein. It is likely thatadditional missense mutations in LRRs have not beenobtained because they confer weak loss-of functionphenotypes that would have been missed in the strin-gent genetic screening conducted to date. Furthermore,it is reasonable to speculate that gain-of-functionproperties could be revealed only in sensitized geneticbackgrounds such as the bri1-5 and bri1-9 mutants. Clearevidence for direct LRR–ligand interactions is knownfor at least two other LRR–RLK plant perception sys-tems, CLAVATA1 (CLV1) and FLAGELLIN SENSITIVE-2(FLS2), which, respectively, play roles in the mainte-nance of stem cells and plant immunity. However, therole of the LRR module in these perception systems israther elusive. Interestingly, mutations in LRRs do notconfer similar properties; the identical amino acidsubstitution to elg, an activated variant of BAK1, in anLRR of CLV1 causes a dominant negative phenotype(Dievart et al. 2003; Whippo and Hangarter, 2005)(Figure 1B). Thus, similar mutations in similar LRRs cantranslate into radically opposed effects. The identifica-tion of bri1-R1 as an intragenic revertant localized in thecanonical LRR module of BRI1 is quite unique. Wepresent here the first report of an intragenic reversionin a member of the LRR–RLK superfamily. As such, ourfindings have implication for a better understanding ofLRR–RLK signaling pathways in plants.

Identical or similar mutations to bri1-R1 occur asnatural polymorphic variants in several plant species:In Arabidopsis, the glycine residue mutated in bri5-R1 isremarkably invariant in other members of the BRreceptor family (BRL1 and BRL3) as well as in otherLRR–RLKs, including BAK1, SERK1, FLS2, and CLV1(Figure 1D). Orthologs of BRI1 have been isolated frommany plant species from dicots such as tomato tomonocots such as rice, barley, and wheat. Strikingly,with the exception of Arabidopsis, it seems that otherplant species have evolved BR receptors in which theconserved Gly residue affected by bri1-R1 is substitutedfor an Asp or a Glu residue. Thus, our intragenicsuppressor, which changes the same amino acid asfound in the OsBRI1 sequence, offers a snapshot ofgenetic variation in phylogenetically related BR recep-tors. The potentially hypermorphic effects observed inbri1-5R1 plants (Table 1) are not inconsistent with anadaptive phenomenon where the genetic assimilationof an Asp or a Glu, instead of a Gly, at a specific positionwould allow environmental information to be embod-ied in new traits. It is tempting to speculate that thisgenetic variation would account, to some extent, forchanges in the ability of various plants to elongate.Whether the latter is relevant for the functional di-versification of BR receptors in land plants is an openquestion that needs to be experimentally tested.

The morphology of bri1-5R1 plants is suggestive ofgain-of-function phenotypes: The amino acid sequence

Figure 5.—BRI1 re-accumu-lates at the plasma membrane inbri1-5R1. (A) Total proteinextracts were prepared fromwild-type Ws-2 (wild type) and iso-genic bri1-5 and bri1-5R1 plants.These extracts were subjected toan anti-BRI1 protein gel blot.Equal loading for all protein sam-ples was ensured by protein quan-tification before loading. Thenonspecific band detected belowBRI1 was used to demonstrateequal loading in each lane aftertransfer. This experiment is indic-ative of four independent repli-cates. (B) Microsomal proteinextracts prepared from Ws-2 (wildtype) and isogenic bri1-5 and bri1-5R1 plants were treated with ei-ther water alone or endoglycosi-

dase H (EndoH; NEB). Samples were collected for anti-BRI1 immunoblot analysis 3 hr after treatment (similar results wereseen at 24 hr). Equal loading for all protein samples was ensured by protein quantification of microsomal pellets after resuspen-sion. PM, plasma membrane; ER, endoplasmic reticulum. This experiment is indicative of three independent replicates. (C) Pro-tein samples of bri1-5 plants from B were used separately for protein blot analysis. The samples are overrepresented by threefoldand were subjected to higher exposure time. (D) (Top) Morphology of 5-day-old dark-grown seedlings of Ws-2 (wild type) andisogenic bri1-5 and bri1-5R1 grown on half-strength MS medium in the absence (�) or presence (1) of 10 mm kif (Calbiochem,EMD Chem). (Bottom) Lengths of 5-day-old dark- grown seedling hypocotyls grown in the absence or presence of 10 mm kif.The mean and standard deviation were calculated from 30 seedlings. This experiment is representative of two independentreplicates.

bri1-5 Intragenic Suppressor 1293

analysis of BRI1 orthologs revealed that the bri1-R1mutation occurs as a natural variant in many plantspecies. Here we demonstrate that the bri1-R1 mutationis able to restore BR-signaling responses to plants withthe weak bri1-5 mutation. Our analysis of bri1-5 and bri1-5R1 morphologies revealed three classes of phenotypes.The first class appeared to be bri1-5 specific (theadditional valves in the fruit), while the second classwas already present in bri1-5 and enhanced in bri1-5R1(the increased pedicel length and the extra anthers).The third class is bri1-5R1 specific, and as such,represents putative gain-of-function phenotypes (moresepals and petals). We anticipate that if bri1-R1 indeed isa gain-of-function mutant, it would display phenotypessimilar to those found in plants overexpressing BRI1 orother positive regulators of BR signaling, includinglonger leaves and dramatically elongated petioles. Onthe other hand, it is possible that the phenotypesspecific to bri1-R1 in the bri1-5 genetic background willsimply be more pronounced in a wild-type background.Interestingly, activation-tagging screens have identifiedseveral dominant extragenic suppressors of bri1-5, in-cluding BAK1, BRL1, and BRS1 (Li et al. 2001, 2002;Zhou et al. 2004). The morphometric analysis of bri1-5R1 plants revealed that, on average, traits that areaffected in bri1-5R1 such as pedicel length, leaf width,and sepal and petal number were not altered in theextragenic suppressors. Thus, in comparison to theseextragenic suppressors, the intragenic suppressor, bri1-R1, appears to act differently, perhaps by modulatingthe signaling competency of BRI1. The phenotypicanalysis of transgenic plants in which a variant of BRI1harboring the bri1-R1 mutation but without bri1-5transformed into bri1-null plants will help resolve thisissue.

bri1-5R1 further implicates the BR-signaling pathwayin the control of the plant cell cycle: The increased sizeof bri1-5R1 plants compared to bri1-5 plants appeared tobe due to an increase in the lengths of internodes. Togain insight into the physiological and cellular mecha-nisms underlying the reversion of the phenotypesassociated with bri1-5, we performed a histologicalanalysis. Our transverse stem sections indicate that thesuppression of the morphological phenotypes is drivenby a significant increase in cell elongation (Figure 2A).The latter was expected because BRs act on plantdevelopment mainly by controlling the ability of a cellto elongate. To our surprise, we noted that the increasein pedicel length in bri1-5R1 plants was due to anincrease in cell division possibly aided by a smallcontribution from cell elongation (Figure 2, B and C).Our results demonstrate that the BR-signaling pathwayis able to modulate the cell cycle, although we do notknow where the input from BR signaling connects to cellcycle regulation.

bri1-5R1 restores the flux of the BR biosynthetic andsignaling pathway concomitantly: The perception of

BRs by BRI1 at the cell surface alters the expression ofhundreds of genes through the balanced activitiesof BES1 and BZR1, two plant-specific transcriptionfactors. A significant subset of these BR responsivegenes encodes key enzymes required for the properbiosynthesis of BRs. There at least two epistatic read-outsavailable to measure the signaling flux of the BR-signaling pathway: (1) the dephosphorylation of BES1and BZR1 and (2) the feedback regulation of thebiosynthetic pathway by the signaling pathway. Tocoordinate BR homeostasis and signaling, BZR1 re-presses the expression of the BR biosynthetic genes CPDand DWF4 by directly binding to their promoters.Accordingly, when signaling occurs, the transcriptionof CPD and DWF4 is dramatically reduced by the actionof BZR1. Thus, the activation of signaling can bemonitored by measuring the transcriptional activitiesof DWF4 and/or CPD. Here, we measured the activationof the BR-signaling pathway by monitoring the flux ofthe BR biosynthetic pathway, the most downstream andrelevant read-out of the feedback regulatory loop. Aspreviously shown, our BR quantification showed theincreased accumulation of BR intermediates in bri1-5plants (Figure 3B), an indication of increased BRmetabolism and hence of reduced signal transduction.On average, the BR metabolic pathway in bri1-5R1 plantswas almost identical to that of wild-type plants. Weassume that the restoration of wild-type endogenous BRaccumulation in bri1-5R1 is probably achieved throughthe successful perception and transduction of BRsignals. Our BR measurements did not allow us to assesswhether bri1-5R1 was a real gain-of-function because wemeasured the bulk accumulation of BRs in whole aerialparts of plants, and our measurements lacked thesensitivity required to detect tissue-specific variation inBR pools. This is consistent with the discrete but organ-specific gain-of-function phenotypes associated with thebri1-5R1 mutation. For the same reason, we anticipatethat the phosphorylation status of BES1 and/or theexpression of CPD or DWF4 would have been as in-formative as our BR measurements in monitoring thelocalized BR responses in actively growing tissues of bri1-5R1 plants. The analysis of organ-specific expressionpatterns of the additional components of the BRI1signaling pathway will be important for uncovering themechanisms responsible for these phenotypes.

bri1-5R1 antagonizes the activities of the unfoldedprotein response: The Unfolded Protein Response(UPR) is a system capable of discriminating native andnonnative proteins by recognizing several commonfeatures of improperly/incompletely folded proteins,such as unpaired Cys residues, immature Asn-linkedglycans, and exposed hydrophobic amino acids. TheUPR is mediated by both the endoplasmic reticulum–mediated quality control (ERQC) and the ERADmachineries. Our endoglycosidase-H digests, togetherwith our ERAD inhibition assays using Kif, are consistent

1294 Y. Belkhadir et al.

with previous reports showing that BRI1-5 is immaturelyglycosylated and retained intracellularly by a functionalcheckpoint. Our findings show that bri1-R1 restores thefunctionality of the BR-signaling pathway probably bypromoting the escape of the BRI1 receptor from the ER.This scenario has three consequences: (1) the mutantBRI1 is no longer subject to the ERAD machinery andre-accumulates to almost wild-type levels, (2) terminalglycosylation can proceed further in the Golgi appara-tus and, consequently, (3) the receptor re-accumulatesat the cell surface. It is unclear how this intragenicmutation allows the delivery of a defective receptor fromthe ER to the cell surface. Our first hypotheses put fortha model in which bri1-R1 compensates for the structuraldefects associated with the bri1-5 mutation. This com-pensation subsequently allows the successful bypass ofthe Bip-dependent, the Thiol-mediated, and the UGGT-based retention mechanisms, which are fundamental tothe ERQC activities. Because the perturbation of onlyone of these processes is necessary and sufficient for ERretention, our findings indicate that a single amino acidsubstitution in the first LRR of BRI1 is capable ofantagonizing the activity of three protein quality controlpathways.

The genetic interaction between bri1-5R1 and BAK1 isconsistent with our former hypothesis. We cannot ex-clude models in which bri1-R1 acts as a suppressor byincreasing interactions within the BRI1 extracellulardomain, between BRI1 monomers, or between BRI1and BAK1. Because BRI1 is able to associate with itselfand BAK1, we speculate that the so-called secondaryquality control of the ER could be involved in the ERretention of BRI1-5. The secondary quality controlensures the proper oligomerization of multisubunitreceptors in the ER via a mechanism involving thecoatamer proteins (COPI). While highly speculative, itis possible that the assembly of BRI1 and BAK1 higher-order complexes is monitored at the ER level. Theoligomeric BR receptors assembled in the proper con-formation could escape the ER, whereas improperconformations created by unpaired cysteine residues willresult in weaker oligomeric interactions and subsequentER retention. In bri1-5R1 plants, a stronger interactionbetween the ECDs of BAK1 and BRI1 could moreefficiently bury the structural defects of BRI1-5, allowingmore efficient relocalization to the plasma membrane. Inthe process, a more efficient BRI1/BAK1 oligomer wouldbe expressed at the cell surface where it could enhancethe signaling competency of the pathway. Analysis of thephysical interaction between BRI1 monomers and BRI1and BAK1, perhaps with the bri1-5R1 mutation, will berequired to ultimately determine the extracellular inter-actions of these two receptor kinases.

We thank Jianming Li for providing bri1-9 seeds in the Columbiaaccession and Sandra Bohn for help with the statistical analysis. Wealso thank Suguru Takatsuto ( Joetsu University of Education) forsupplying deuterium-labeled internal standards. Initial support for

this project came from the U. S. Department of Agriculture (USDA)(grant no. 97-353044708) and from the National Science Foundation(NSF) (grant nos. IBN-0347675 and MCB-0418946). This work was alsosupported in part by a Grant-in-Aid for Scientific Research (B) fromthe Ministry of Education, Culture, Sports, Science and Technology ofJapan to S.F. (grant no. 19380069). Y.B. was a Howard Hughes MedicalInstitute Fellow of the Life Sciences Research Foundation and wassupported by a grant from the USDA to Joanne Chory. Y.B. is also arecipient of the Alain Philippe Foundation. A.A. was a Minority Accessto Research Careers trainee, a program funded by the Nationalinstitutes of Health (grant no. T34 GM008718). R.J.S. and R.M. weresupported by the Undergraduate Biology Research Program (NSFgrant no. DBI-0242842). A.D. was supported by an IntegrativeGraduate Education and Research Traineeship fellowship from theNSF (grant no. DGE-0114420).

LITERATURE CITED

Albrecht C., E. Russinova, B. Kemmerling, M. Kwaaitaal, S. C.De Vries, 2008 Arabidopsis SOMATIC EMBRYOGENESISRECEPTOR KINASE proteins serve brassinosteroid-dependentand -independent signaling pathways. Plant Physiol. 148: 611–619.

Arabidopsis Genome Initiative, 2000 Analysis of the genomesequence of the flowering plant Arabidopsis thaliana. Nature408: 796–815.

Asami, T., M. Mizutani, S. Fujioka, H. Goda, Y. K. Min et al.,2001 Selective interaction of triazole derivatives with DWF4, acytochrome P450 monooxygenase of the brassinosteroid biosyn-thetic pathway, correlates with brassinosteroid deficiency in planta.J. Biol. Chem. 276: 25687–25691.

Belkhadir, Y., and J. Chory, 2006 Brassinosteroid signaling: aparadigm for steroid hormone signaling from the cell surface.Science 314: 1410–1411.

Bell, C. J., and J. R. Ecker, 1994 Assignment of 30 microsatelliteloci to the linkage map of Arabidopsis. Genomics 19: 137–144.

Bell, J. K., G. E. Mullen, C. A. Leifer, A. Mazzoni, D. R. Davies

et al., 2003 Leucine-rich repeats and pathogen recognition intoll-like receptors. Trends Immunol. 24: 528–533.

Bishop, G.J., 2003 Brassinosteroid mutants of crops. J. Plant GrowthRegul. 22: 325–335.

Choe, S., 2004 Brassinosteroids biosynthesis and metabolism, pp.156–178 in Plant Hormones: Biosynthesis, Signal Transduction,Action! edited by P. Davies. Kluwer Academic, Dordrecht, TheNetherlands.

Choe, S., B. P. Dilkes, S. Fujioka, S. Takatsuko, A. Sakurai et al.,1998 The DWF4 gene of Arabidopsis encodes a cytochromeP450 that mediates multiple 22alpha-hydroxylation steps in bras-sinosteroid biosynthesis. Plant Cell 10: 231–243.

Choe, S., R. J. Schmitz, S. Fujioka, S. Takatsuko, M. O. Lee et al.,2002 Arabidopsis brassinosteroid-insensitive dwarf12 mutantsare semidominant and defective in a glycogen synthase kinase3beta-like kinase. Plant Physiol. 130: 1506–1515.

Clouse, S. D., and K. A. Feldmann, 1999 Molecular genetics ofbrassinosteroid action, pp. 163–190 in Brassinosteroids Steroidal PlantHormones, edited by A. Sakurai, T. Yokota and S. D. Clouse.Springer-Verlag, Berlin/Heidelberg, Germany/New York.

Clouse, S. D., M. Langford and T. C. McMorris, 1996 A brassinos-teroid-insensitive mutant in Arabidopsis thaliana exhibits multipledefects in growth and development. Plant Physiol. 111: 671–678.

Courtemanche, N., and D. Barrick, 2008 The leucine-rich repeatdomain of Internalin B folds along a polarized N-terminal path-way. Structure 16: 705–714.

Dievart, A., and S. E. Clark, 2003 Using mutant alleles todetermine the structure and function of leucine-rich repeatreceptor-like kinases. Curr. Opin. Plant Biol. 6: 507–516.

Dievart, A., M. Dalal, F. E. Tax, A. D. Lacey, A. Huttley et al.,2003 CLAVATA1 dominant-negative alleles reveal functionaloverlap between multiple receptor kinases that regulate meri-stem and organ development. Plant Cell 15: 1198–1211.

Friedrichsen, D. M., C. A. Joazeuro, J. Li, T. Hunter and J. Chory,2000 Brassinosteroid-insensitive-1 is a ubiquitously expressed

bri1-5 Intragenic Suppressor 1295

leucine-rich repeat receptor serine/threonine kinase. PlantPhysiol. 123: 1247–1256.

Fujioka, S., S. Takatsuko and S. Yoshida, 2002 An early C-22oxidation branch in the brassinosteroid biosynthetic pathway.Plant Physiol. 130: 930–939.

Geldner, N., D. L. Hyman, X. Wang, K. Schumacher and J. Chory,2007 Endosomal signaling of plant steroid receptor kinaseBRI1. Genes Dev. 21: 1598–1602.

Gomez-Gomez, L., and T. Boller, 2000 FLS2: an LRR receptor-likekinase involved in the perception of the bacterial elicitor flagellinin Arabidopsis. Mol. Cell 5: 1003–1011.

Grove, M. D., D. F. Spencer, W. K. Rohwedder, N. Mandava, J. F.Worley et al., 1979 Brassinolide, a plant growth promotingsteroid isolated from Brassica napus pollen. Nature 281: 216–217.

He, J. X., J. M. Gendron, Y. Sun, S. S. Gampala, N. Gendron et al.,2005 BZR1 is a transcriptional repressor with dual roles in bras-sinosteroid homeostasis and growth responses. Science 307:1634–1638.

Hecht, V., J. P. Vielle-Calzada, M. V. Hartog, E. D. Schmidt,K. Boutilier et al., 2001 The Arabidopsis SOMATIC EMBRYO-GENESIS RECEPTOR KINASE 1 gene is expressed in developingovules and embryos and enhances embryogenic competence inculture. Plant Physiol. 127: 803–816.

Hong, Z., H. Jin, T. Tzfira and J. Li, 2008 Multiple mechanism-mediated retention of a defective brassinosteroid receptor inthe endoplasmic reticulum of Arabidopsis. Plant Cell 20:3418–3429.

Hong Z., H. Jin, A.-C. Fitchette, Y. Xia, A. M. Monk, T. Faye et al.,2009 Mutations of an alpha-1,6 mannosyltransferase inhibitendoplasmic reticulum–associated degradation of defectivebrassinosteroid receptors in Arabidopsis. Plant Cell 21: 3792–3802.

Jin, H., Z. Yan, K. H. Nam and J. Li, 2007 Allele-specific suppressionof a defective brassinosteroid receptor reveals a physiological roleof UGGT in ER quality control. Mol. Cell 26: 821–830.

Jin, H., Z. Hong, W. Su and J. Li, 2009 A plant-specific calreticulin isa key retention factor for a defective brassinosteroid receptor inthe endoplasmic reticulum. Proc. Natl. Acad. Sci. USA 106:13612–13617.

Kim, T.-W., S. Guan, Y. Sun, Z. Deng, W. Tang et al., 2009 Brassinsteroidsignal transduction from cell surface receptor kinase to nuclear tran-scription factors. Nat. Cell Biol. 11: 1254–1260.

Kinoshita, T., A. Cano-Delgado, H. Seto, S. Hiranuma, S. Fujioka

et al., 2005 Binding of brassinosteroids to the extracellulardomain of plant receptor kinase BRI1. Nature 433: 167–171.

Li, J., and J. Chory, 1997 A putative leucine-rich repeat receptorkinase involved in brassinosteroid signal transduction. Cell 90:929–938.

Li, J., and K. H. Nam, 2002 Regulation of brassinosteroid signalingby a GSK3/SHAGGY-like kinase. Science 295: 1299–1301.

Li, J., K. A. Lease, F. E. Tax and J. C. Walker, 2001 BRS1, a serinecarboxypeptidase, regulates BRI1 signaling in Arabidopsis thali-ana. Proc. Natl. Acad. Sci. USA 98: 5916–5921.

Li, J., J. Wen, K. A. Lease, J. T. Doke, F. E. Tax et al., 2002 BAK1, anArabidopsis LRR receptor-like protein kinase, interacts withBRI1 and modulates brassinosteroid signaling. Cell 110: 213–222.

Mandava, N., 1988 Plant growth-promoting brassinosteroids.Annu. Rev. Plant Physiol. Plant Mol. Biol. 39: 23–52.

Nam, K. H., and J. Li, 2002 BRI1/BAK1, a receptor kinase pairmediating brassinosteroid signaling. Cell 110: 203–212.

Neff, M. M., E. Turk and M. Kalishman, 2002 Web-based primerdesign for single nucleotide polymorphism analysis. TrendsGenet. 18: 613–615.

Noguchi, T., S. Fujioka, S. Choe, S. Takatsuto, S. Yoshida et al.,1999 Brassinosteroid-insensitive dwarf mutants of Arabidopsisaccumulate brassinosteroids. Plant Physiol. 121: 743–752.

Oh, M.H., X. Wang, U. Kota, M. B. Goshe, S. D. Clouse et al.,2009 Tyrosine phosphorylation of the BRI1 receptor kinaseemerges as a component of brassinosteroid signaling in Arabi-dopsis. Proc. Natl. Acad. Sci. USA 106: 658–663.

Raikhel, N, and G. Hicks, 2007 Signaling from plant endosomes:compartments with something to say! Genes Dev. 21: 1578–1580.

Rosenfeld, M. G., and C. K. Glass, 2001 Coregulator codes oftranscriptional regulation by nuclear receptors. J. Biol. Chem.276: 36865–36868.

Russinova, E., J. W. Borst, M. Kwaaitaal, A. Cano-Delgado, Y. Yin

et al., 2004 Heterodimerization and endocytosis of Arabidopsisbrassinosteroid receptors BRI1 and AtSERK3 (BAK1). Plant Cell16: 3216–3229.

Szekeres, M., K. Nemeth, Z. Koncz-Kalman, J. Mathur, A.Kauschmann et al., 1996 Brassinosteroids rescue the deficiencyof CYP90, a cytochrome P450, controlling cell elongation andde-etiolation in Arabidopsis. Cell 85: 171–182.

Tang, W., T.-W. Kim, J. A. Oses-Prieto, Y. Sun, Z. Deng et al.,2008 BSKs mediate signal transduction from the receptorkinase BRI1 in Arabidopsis. Science 321: 557–560.

Truhlar, S. M., and E. A. Komives 2008 LRR domain folding: Justput a cap on it! Structure 16: 655–657.

Vert, G., and J. Chory, 2006 Downstream nuclear events in brassi-nosteroid signalling. Nature 441: 96–100.

Wang, X., and J. Chory, 2006 Brassinosteroids regulate dissocia-tion of BKI1, a negative regulator of BRI1 signaling, from theplasma membrane. Science 313: 1118–1122.

Wang, X., X. Li, J. Meisenhelder, T. Hunter, S. Yoshida et al.,2005a Autoregulation and homodimerization are involved inthe activation of the plant steroid receptor BRI1. Dev. Cell 8:855–865.

Wang, X., M. B. Goshe, E. J. Soderblom, B. S. Phinney, J. A. Kuchar

et al., 2005b Identification and functional analysis of in vivophosphorylation sites of the Arabidopsis BRASSINOSTEROID-INSENSITIVE1 receptor kinase. Plant Cell 17: 1685–1703.

Wang, Z. Y., T. Nakano, J. Gendron, J. He, M. Chn et al.,2002 Nuclear-localized BZR1 mediates brassinosteroid-inducedgrowth and feedback suppression of brassinosteroid biosynthesis.Dev. Cell 2: 505–513.

Whippo, C. W., and R. P. Hangarter, 2005 A brassinosteroid-hypersensitive mutant of BAK1 indicates that a convergence ofphotomorphogenic and hormonal signaling modulates photot-ropism. Plant Physiol. 139: 448–457.

Yin, Y., Z. Y. Wang, S. Mora-Garcia, J. Li, S. Yoshida et al.,2002 BES1 accumulates in the nucleus in response to brassinos-teroids to regulate gene expression and promote stem elonga-tion. Cell 109: 181–191.

Zhao, J., P. Peng, R. J. Schmitz, A. D. Decker, F. E. Tax et al.,2002 Two putative BIN2 substrates are nuclear componentsof brassinosteroid signaling. Plant Physiol. 130: 1221–1229.

Zhou, A., H. Wang, J. C. Walker and J. Li, 2004 BRL1, a leucine-richrepeat receptor-like protein kinase, is functionally redundant withBRI1 in regulating Arabidopsis brassinosteroid signaling. Plant J.40: 399–409.

Communicating editor: B. Bartel

1296 Y. Belkhadir et al.

Supporting Information http://www.genetics.org/cgi/content/full/genetics.109.111898/DC1

Intragenic Suppression of a Trafficking-Defective Brassinosteroid Receptor Mutant in Arabidopsis

Youssef Belkhadir, Amanda Durbak, Michael Wierzba, Robert J. Schmitz, Andrea Aguirre, Rene Michel, Scott Rowe, Shozo Fujioka and Frans E. Tax

Copyright © 2010 by the Genetics Society of America DOI: 10.1534/genetics.109.111898

Y. Belkhadir et al. 2 SI

TABLE S1

Accumulation of BRs is reduced in bri1-5R1 mutants compared to bri1-5 mutants

Ws-2 bri1-5 bri1-5R1

24-Methylenecholesterol 4640 3130 4120

Campesterol 41200 28800 34900

Campestanol 877 834 1000

6-Oxocampestanol 31.4 46.1 47.7

6-Deoxocathasterone 1.02 1.21 1.83

6-Deoxoteasterone 0.08 0.1 0.05

3-Dehydro-6-deoxoteasterone 0.18 0.21 0.18

6-Deoxotyphasterol 1.31 1.09 1.3

6-Deoxocastasterone 2.68 4.39 4.81

Cathasterone not detected not detected not detected

Teasterone 0.01 0.03 0.01

Typhasterol 0.2 1.71 0.56

Castasterone 0.32 9.01 1.5

Brassinolide not detected 2.32 0.31

All values ng/g fresh weight