The Arabidopsis Homolog of the Mammalian OS-9 Protein Plays a Key Role in the Endoplasmic Reticulum-Associated Degradation of Misfolded Receptor-Like Kinases

Abstract

The endoplasmic reticulum-associated degradation (ERAD) is a highly conserved mechanism to remove misfolded membrane/secretory proteins from the endoplasmic reticulum (ER). While many of the individual components of the ERAD machinery are well characterized in yeast and mammals, our knowledge of a plant ERAD process is rather limited. Here, we report a functional study of an Arabidopsis homolog (AtOS9) of an ER luminal lectin Yos9 (OS-9 in mammals) that recognizes a unique asparagine-linked glycan on misfolded proteins. We discovered that AtOS9 is an ER-localized glycoprotein that is co-expressed with many known/predicted ER chaperones. A T-DNA insertional atos9-t mutation blocks the degradation of a structurally imperfect yet biochemically competent brassinosteroid (BR) receptor bri1-9, causing its increased accumulation in the ER and its consequent leakage to the cell surface responsible for restoring the BR sensitivity and suppressing the dwarfism of the bri1-9 mutant. In addition, we identified a missense mutation in AtOS9 in a recently discovered ERAD mutant ems-mutagenized bri1 suppressor 6 (ebs6-1). Moreover, we showed that atos9-t also inhibits the ERAD of bri1-5, another ER-retained BR receptor, and a misfolded EFR, a BRI1-like receptor for the bacterial translation elongation factor EF-Tu. Furthermore, we found that AtOS9 interacted biochemically and genetically with EBS5, an Arabidopsis homolog of the yeast Hrd3/mammalian Sel1L known to collaborate with Yos9/OS-9 to select ERAD clients. Taken together, our results demonstrated a functional role of AtOS9 in a plant ERAD process that degrades misfolded receptor-like kinases.

INTRODUCTION

Upon entering the endoplasmic reticulum (ER), most membrane and secretory proteins are co-/post-translationally glycosylated at certain asparagine (Asn or N) residues catalyzed by oligosaccharyltransferase that transfers a preassembled three-branched Glc₃Man₉GlcNAc₂ (Glc, Man, and GlcNAc indicate glucose, mannose, and N-acetylglucosamine, respectively) from its lipid linker to nascent polypeptides (Kelleher and Gilmore, 2006). The N-linked Glc₃Man₉GlcNAc₂ is subsequently processed in the ER, and processed N-glycans are recognized by corresponding lectins, thus regulating protein folding, quality control (QC), intracellular trafficking, and degradation (Aebi et al., 2010). Rapid removal of the outer and middle Glc residues from the A-branch permits association of Glc₁Man₉GlcNAc₂-containing glycoproteins with two ER lectins, calnexin (CNX) and calreticulin (CRT), which recruit additional ER chaperones to assist folding (Helenius and Aebi, 2004; Williams, 2006), whereas slow trimming of the last Glc liberates a maturing glycoprotein from CNX/CRT. A correctly folded protein exits the ER to continue its secretory journey (Hauri et al., 2002), whereas a mis-/incompletely folded glycoprotein is recognized and reglucosylated by an ER luminal protein-folding sensor known as UDP-Glc:glycoprotein glucosyltransferase (UGGT) (Arnold et al., 2000; Caramelo and Parodi, 2007), allowing CNX/CRT reassociation for additional rounds of chaperone-assisted folding. A glycoprotein that fails to acquire its native structure within a given time window is removed from the ER by a unique mechanism known as ER-associated degradation (ERAD) that involves retrotranslocation into the cytosol for proteasome-mediated degradation (Vembar and Brodsky, 2008).

A crucial decision in the N-glycan-mediated ERQC is to terminate futile folding cycles and to divert a terminally misfolded glycoprotein into an ERAD pathway. It is now believed that sequential trimming of the terminal α1,2 Man residue from the B and C-branches by the ER α1,2 mannosidase 1 (Mns1 in yeast and ERManI in mammals) and homologous to α-mannosidase 1 (Htm1) (ER-degradation enhancing α-mannosidase-like proteins (EDEMs) in mammals), respectively, is required to generate an N-glycan ERAD signal, Man₇GlcNAc₂ with a free α1,6 Man on the C-branch (Quan et al., 2008; Clerc et al., 2009). This unique N-glycan is recognized by a different ER luminal lectin known as Yos9 (OS-9 in mammals) containing a glycan-binding Man-6-phosphate receptor homology (MRH) domain (Hosokawa et al., 2010). Yos9/OS-9 recruits an ERAD client through its interaction with Hrd3 (Sel1L in mammals) (Denic et al., 2006; Gauss et al., 2006), a type I membrane protein with a large luminal domain (Hampton et al., 1996) that can directly recruit misfolded ERAD clients, to a membrane-embedded ubiquitin E3 ligase Hrd1 for ubiquitination and subsequent retrotranslocation into the cytosolic proteasome for degradation (Smith et al., 2011).

Our knowledge of ERQC/ERAD mainly came from genetic/biochemical studies using yeast and cultured mammalian cells. Although similar processes were known to operate in plants (Ceriotti and Roberts, 2006; Liu et al., 2011), our understanding of their biochemical mechanism(s) and molecular components is rather limited. This is largely due to the lack of convenient model proteins for forward and reverse genetic studies in model plants. Recent studies identified several excellent model proteins to study ERQC/ERAD in Arabidopsis (Saijo, 2010), including two mutant variants of BRASSINOSTEROID-INSENSITIVE 1 (BRI1), a well-studied leucine-rich-repeat receptor-like kinase (LRR-RLK) that functions as a cell-surface receptor for the plant steroid hormone brassinosteroid (BR) (Li and Chory, 1997; Kinoshita et al., 2005), and EF-Tu Receptor (EFR), a BRI1-like LRR-RLK that recognizes the bacterial translation elongation factor EF-Tu to induce a plant immunity response (Zipfel et al., 2006). We discovered that a Cys⁶⁹–Tyr mutation in bri1-5 and a Ser⁶⁶²–Phe mutation in bri1-9 result in ER retention and subsequent ERAD of the two mutant BR receptors, causing a severe BR-insensitive dwarf phenotype in Arabidopsis (Jin et al., 2007; Hong et al., 2008; Jin et al., 2009). A large-scale genetic screen for bri1-9 suppressers led to identification of EMS-mutagenized bri1 suppressor 1 (EBS1), the Arabidopsis UGGT homolog, and EBS2, a land plant-specific CRT known as CRT3 (Jin et al., 2007, 2009). It was believed that the Ser⁶⁶²–Phe mutation introduces a minor structural defect into bri1-9, which is recognized by EBS1 that adds a terminal Glc residue back to the N-glycans of bri1-9 to permit bri1-9 binding to EBS2, thus retaining the mutant receptor in the ER. Loss-of-function ebs1 or ebs2 mutations compromise this Arabidopsis ERQC system, causing escape of bri1-9 to the cell surface, where the mutant BR receptor activates BR signaling to suppress the bri1-9 dwarfism (Jin et al., 2007, 2009).

Further studies of the three Arabidopsis model proteins showed that their protein abundance could be significantly increased by treatment with kifunensine (Hong et al., 2008, 2009; Nekrasov et al., 2009; Saijo et al., 2009), a widely used inhibitor of α1,2 mannosidases (Elbein et al., 1990) that include Mns1/ERManI and Htm1/EDEMs. These findings not only confirmed the existence of an ERAD process in Arabidopsis, but also suggested involvement of Man-trimming steps in generating an ERAD N-glycan signal. To fully understand the Arabidopsis ERAD process, we performed an immunoblot-based secondary screen for ebs mutants with increased bri1-9 abundance and isolated ebs3, ebs4, ebs5, and ebs6 as ERAD mutants (Hong et al., 2009; Su et al., 2011). We discovered that EBS4 encodes an ER-localized mannosyltransferase involved in generating the C-branch of Glc₃Man₉GlcNAc₂ and that loss-of-function ebs4 mutations cause transfer of truncated N-glycan precursors to nascent polypeptides, which can not be processed to generate the α1,6 Man-exposed N-glycan signal to mark bri1-9 for ERAD (Hong et al., 2009). Recent cloning of the EBS5 gene and several reverse genetic studies identified the Arabidopsis homolog of the Hrd3/SEl1L, revealed a redundant role for the two Arabidopsis Hrd1 homologs, AtHrd1A and AtHrd1B, and implicated an ER-localized ubiquitin conjugating enzyme UBC32 in degrading the two mutant BR receptors (Liu et al., 2011; Su et al., 2011; Cui et al., 2012).

In the current study, we took both the forward and reverse genetic approaches to define the functional role of the Arabidopsis homolog of the Yos9/OS-9 proteins (named hereinafter as AtOS9) in the ERAD of two mutant bri1 proteins and the misfolded EFR (in the absence of EBS1/UGGT). Our study thus discovered the fourth component of the Arabidopsis ERAD machinery that removes terminally misfolded glycoproteins.

RESULTS

At5g35080 Encodes an Arabidopsis Homolog of Yos9/OS-9

Using Yos9 and the human OS-9 as query, BLAST searches against the entire Arabidopsis genome identified a potential Yos9/OS-9 homologous gene, At5g35080, which consists of eight exons and seven introns and encodes a polypeptide (named as AtOS9) of 282 amino acid residues with a predicted hydrophobic signal sequence at its N-terminus (Supplemental Figure 1). AtOS9 is significantly smaller than its counterparts in yeast and mammals due to its lacking of the large C-terminal extensions of Yos9/OS-9 but is predicted to contain the MRH domain (from residue 110 to 243, Supplemental Figure 1), which is known to be crucial for the ERAD lectin function in yeast and mammals (Hosokawa et al., 2010). The predicted MRH domain shares 29 and 36% identity with those of Yos9 and OS-9 and contains the eight amino acid residues that are directly involved in sugar binding (Satoh et al., 2010) (Figure 1A).

Figure 1.

AtOS9 Is a Likely Homolog of Yos9/OS-9 and Localizes in the ER.

(A) Alignment of the MRH domains between Yos9/OS-9 and their plant homologs. Yos9 (accession number: NP_010342), human Os-9 (NP_006803), and predicted Yos9/OS-9 homologs from Arabidopsis (AtOS9, NP_568525), Physcomitrella patens (PpOS9, XP_001756223), Oryza sativa (OsOS9, NP_001174928), Zea mays (ZmOS9, NP_001149187), and mouse (MmOS9, NP_001164497) were aligned using the ClustalW program. The aligned amino acid sequences were color shaded by the Boxshade program at the Mobyle portal (http://mobyle.pasteur.fr). The regions containing the predicted MRH domains are shown here, while the alignment of full-length proteins is presented in Supplemental Figure 1. Identical residues are colored in red while similar residues are shaded in cyan. Stars indicated amino acids directly involved in N-glycan binding (Satoh et al., 2010).

(B, D) Immunoblot analysis of AtOS9. Equal amounts of total proteins extracted from 2-week-old seedlings treated with or without 5 μg ml⁻¹ TM were treated without or with EndoH, separated by 10% SDS–PAGE, and analyzed by immunoblot using an anti-AtOS9 antibody. Coomassie blue staining of the small subunit of the Arabidopsis Rubisco (RbcS) serves as a loading control.

(C) Confocal microscopic analysis of AtOS9. Shown from left to right are fluorescence patterns of the AtOS9–GFP fusion protein (green), the ER-localized HDEL-tagged RFP (red), and the superimposed image of the green and red fluorescent signals in Agrobacterium-infiltrated tobacco leaf epidermal cells.

Several previous genome-wide gene expression analyses showed that At5G35080 was up-regulated by ER-stresses (Martinez and Chrispeels, 2003; Kamauchi et al., 2005) and was co-expressed with many Arabidopsis genes encoding known or putative ER chaperones including CNXs/CRTs and protein disulfide isomerases (ATTED-II; http://atted.jp/; Supplemental Figure 2). Consistently with these gene expression studies, our immunoblot analysis revealed that a short-term treatment of Arabidopsis seedlings with tunicamycin (TM), an N-glycan biosynthesis inhibitor that causes severe ER-stresses (Elbein, 1981), significantly increased the AtOS9 abundance, whereas a prolonged TM treatment led to detection of a fast-moving AtOS9 band on protein gel (Figure 1B). The molecular weight difference between the fast- and slow-moving AtOS9 bands is ∼ 6 kD, which is consistent with its three predicted N-glycan sites (Asn⁹⁴, Asn¹⁶⁹, and Asn¹⁹⁰; see Supplemental Figure 1).

Unlike the yeast Yos9, AtOS9 lacks the His-Asp-Glu-Leu (HDEL) ER retrieval sequence, but our confocal analysis of an AtOS9 fusion protein tagged at its C-terminus with green fluorescent protein (AtOS9-GFP), which was transiently expressed in tobacco leaves, indicated that AtOS9 is an ER-localized protein because its fluorescent pattern overlapped with the fluorescent signal of a widely used ER marker protein, a HDEL-tagged red fluorescent protein (RFP-HDEL) (Figure 1C). The ER localization of AtOS9 was further confirmed by a simple biochemical assay using the endoglycosylase H (Endo H) that cleaves high-mannose (H)-type N-glycans of ER-localized glycoproteins but cannot remove Golgi-processed complex (C)-type N-glycans (Robbins et al., 1984). As shown in Figure 1D, the N-glycans of AtOS9 can be efficiently cleaved by Endo H, converting the slow-moving AtOS9 band to a fast-moving band that exhibits the same electrophoretic mobility as the non-glycosylated AtOS9 band of the TM-treated seedlings.

A T-DNA Insertional Mutation in AtOS9 Prevents the ERAD of bri1-9

To investigate the physiological functions of AtOS9, we identified a T-DNA insertional mutant (SALK_029413, named hereinafter as atos9-t) from the Ecker SIGNAL T-DNA express database (Alonso et al., 2003), which carries a T-DNA insertion in the middle of the predicted sixth exon (Figure 2A). An immunoblot analysis with an affinity-purified anti-AtOS9 antibody, which easily detected AtOS9 in seedlings of both wild-type and bri1-9, failed to detect its presence in the T-DNA insertional mutant (Figure 2B), suggesting that atos9-t is likely a null mutant. Although the atos9-t mutant is indistinguishable from the wild-type control when grown under standard laboratory conditions (Figure 2C), it was hypersensitive to TM (Figure 2D), implying a role in coping with ER-stresses that often lead to overaccumulation of misfolded proteins in the ER.

Figure 2.

Identification of a T-DNA Insertional atos9-t Mutant Hypersensitive to TM.

(A) Schematic presentation of the AtOS9 gene structure. The thin lines represent introns, black bars denote amino acid-encoding exons, and white bars indicate untranslated regions of the first and last exons. The position and orientation of the inserted T-DNA are indicated.

(B) Immunoblot analysis of AtOS9. Equal amounts of total proteins extracted from seedlings of wild-type, atos9-t, and bri1-9 were separated by 10% SDS–PAGE and analyzed by immunoblot using an anti-AtOS9 antibody. Asterisk indicates a non-specific band for loading control.

(C) Images of 4-week-old soil-grown seedlings of wild-type and atos9-t.

(D) Images of 4-week-old seedlings of wild-type and atos9-t mutant grown on TM-containing ½ MS medium.

To directly test a role of AtOS9 in the Arabidopsis ERAD system, we crossed the atos9-t mutation into the bri1-9 mutant, whose dwarf phenotype was previously shown to be caused by ER retention and ERAD of the mutant BR receptor bri1-9 (Jin et al., 2007; Hong et al., 2009; Jin et al., 2009). If AtOS9 was a component of the Arabidopsis ERAD machinery that degrades the mutant BR receptor, we would expect that the atos9-t mutation should block the ERAD of bri1-9 and significantly increase the bri1-9 abundance. Indeed, as shown in Figure 3A, the bri1-9 protein abundance is significantly higher in atos9-t bri1-9 than in bri1-9. To eliminate the possibility that the elevated bri1-9 level in atos9-t bri1-9 is caused by increased biosynthesis of the mutant BR receptor, we performed a cycloheximide (CHX, a widely used protein biosynthesis inhibitor) chase experiment. We treated 2-week-old seedlings of both bri1-9 and atos9-t bri1-9 mutants with 180 μM CHX and analyzed the bri1-9 abundance by immunoblot with an anti-BRI1 antibody (Mora-Garcia et al., 2004). As shown in Figure 3B, the bri1-9 protein is relatively unstable, with a half-life of ∼6 h. By contrast, a significant amount of the mutant BR receptor remained in the atos9-t bri1-9 double mutant 24 h after the CHX treatment. We concluded that the increased bri1-9 abundance in the atos9-t bri1-9 mutant is likely caused by reduced degradation rather than by increased biosynthesis, supporting a functional role of AtOS9 in the ERAD of the mutant BR receptor.

Figure 3.

The atos9-t Mutation Inhibits the ERAD of bri1-9 and Suppresses the bri1-9 Dwarfism.

(A) Immunoblot analysis of the BRI1/bri1-9 abundance. Equal amounts of total proteins extracted from 2-week-old seedlings were treated with or without Endo H, separated by 7% SDS–PAGE, and analyzed by immunoblot with anti-BRI1 antibody.

(B) Immunoblot analysis of the bri1-9 stability. Two-week-old seedlings were transferred into liquid ½ MS medium containing 180 μM CHX. Equal amounts of seedlings were removed at indicated incubation times to extract total proteins in 2×SDS sample buffer, which were subsequently analyzed by immunoblot with the anti-BRI1 antibody.

(C–E) Images of 4-week-old soil-grown plants (C), 5-day-old dark-grown seedlings (D), and 2-month-old mature plants (E) of wild-type, bri1-9, and atos9-t bri1-9.

(F) The root-growth inhibition assay. Root lengths of 7-day-old seedlings grown on BL-containing medium were measured and presented as the relative value of the average root length of BL-treated seedlings to that of untreated seedlings of the same genotype. Each data point represents the average of ∼40 seedlings of duplicated experiments. Error bars denote standard error.

(G) Immunoblot analysis of the BES1 phosphorylation status. Equal amounts of total proteins extracted from 2-week-old seedlings treated with or without 1 μM BL for 1 h were separated by 10% SDS–PAGE and analyzed by immunoblot using an anti-BES1 antiserum. In (A), (B), and (G), coomassie blue staining of RbcS serves as a loading control.

The atos9-t mutation also nicely suppressed the dwarf phenotype of the BR receptor mutant. Compared to bri1-9, the atos9-t bri1-9 double mutant has a much larger rosette with longer petioles at the seedling stage, longer hypocotyls in the dark, and taller inflorescence stems when grown in soil (Figure 3C–3E). We suspected that this is likely caused by leakage of a small amount of bri1-9 protein from the ER as a result of saturating the bri1-9 retention system by overaccumulated mutant BR receptor in the folding compartment. If this were true, we would expect that the atos9-t mutation should at least partially restore BR sensitivity to the BR-insensitive dwarf mutant. As shown in Figure 3F, increasing concentrations of brassinolide (BL, the most active BR) had little effect on the root elongation of the bri1-9 seedlings, but inhibited the root growth of the wild-type as well as the atos9-t bri1-9 mutant (albeit to a lesser degree). The restored BR sensitivity can also be detected biochemically. A 1-h treatment of 2-week-old seedlings with 1 μM BL had little effect on the phosphorylation status of BES1 (Yin et al., 2002; Zhao et al., 2002), a very robust biochemical marker of active BR signaling (Mora-Garcia et al., 2004), but led to rapid dephosphorylation of BES1 in both the wild-type and the atos9-t bri1-9 mutant (Figure 3G).

Additional support for our ‘leakage’ theory came from an immunoblot experiment with total protein extracts treated with Endo H. If the restored BR sensitivity was caused by a small amount of leaked mutant receptor from the ER, we should be able to detect their presence by the Endo H assay as their N-glycans are no longer the H-type and are therefore resistant to the Endo H-catalyzed deglycosylation. Indeed, a small but significant amount of bri1-9 carrying the Endo-H-resistant C-type N-glycan was detected in the atos9-t bri1-9 mutant (Figure 3A). Consistently, overexpression of EBS2, a land plant-specific CRT that functions as a rate-limiting factor for the ER retention of bri1-9 (Jin et al., 2009), nullified the suppressive effect of the atos9-t mutation on the bri1-9 mutant, as the gEBS2 atos9-t bri1-9 transgenic mutants are morphologically similar to the bri1-9 single mutant (Supplemental Figure 3A). An immunoblot assay showed that EBS2 overexpression had little effect on the bri1-9 abundance but significantly reduced the amount of bri1-9 proteins carrying the Endo H-resistant C-type N-glycan (Supplemental Figure 3B).

Further Genetic Support for a Role of AtOS9 in the ERAD of bri1-9

The involvement of AtOS9 in the ERAD of bri1-9 was further confirmed by two additional experiments. The first was a complementation assay to ensure that the mutation for suppressing the bri1-9 dwarfism is indeed the atos9-t mutation instead of other potential mutations in the T-DNA line. We introduced a genomic AtOS9–GFP transgene, which carries the native promoter, all seven introns, exons 1–7, the amino acid-coding segment of the eighth exon with its last codon fused to the open reading frame of a GFP gene, and the 3'-terminator of the pea (Pisum sativum) rbsc E9 gene, into the atos9-t bri1-9 mutant. As shown in Figure 4A and 4B, the introduction of the gAtOS9–GFP transgene led to production of the AtOS9–GFP fusion protein (similar to or higher than the AtOS9 level in bri1-9) and restored the bri1-9-like dwarfism to the atos9-t bri1-9 double mutant. Consistently with the observed phenotypic rescue, immunoblot analysis revealed that the bri1-9 abundance was significantly reduced to that of the bri1-9 mutant in the transgenic gAtOS9–GFP atos9-t bri1-9 lines (Figure 4C). The second support for an essential role of AtOS9 in the ERAD of bri1-9 came from our mapping and sequence analysis of the ebs6-1 mutant that was previously identified by our secondary screen for Arabidopsis ERAD mutants (Hong et al., 2009). PCR-based genetic mapping with genomic DNAs of ∼100 F₂ seedlings derived from a mapping cross of the ebs6-1 bri1-9 mutant in a Columbia ecotype (Col-0) with a bri1-9 mutant in the Wassilewskija-2 (Ws-2) ecotype (Noguchi et al., 1999) located the EBS6 locus in a genomic region that contains AtOS9 gene. Subsequent sequence analysis revealed a G-to-A single-nucleotide polymorphism (SNP) in the AtOS9 gene, which changes an absolutely conserved glycine residue (Gly¹⁹¹) to glutamic acid (Glu) (Supplemental Figure 4A). This detected SNP was subsequently converted to a derived cleaved amplified polymorphic sequence (dCAPS) marker (Neff et al., 1998) that was used to confirm a tight genetic linkage between the G–A mutation and the ebs6-1 bri1-9 phenotype in several F₂ mapping populations. Interestingly, despite the fact that the mutated Gly¹⁹¹ residue is absolutely conserved among Yos9, mammalian OS-9s, and several predicted plant OS9s, the ebs6-1 mutation is a rather weak suppressor of the bri1-9 mutant, as the ebs6-1 bri1-9 mutant is only slightly larger and has a less compact rosette with a slightly taller stature compared to the bri1-9 mutant (Figure 4D and 4E). Immunoblot analysis showed that the bri1-9 abundance and the amount of the Endo H-resistant bri1-9 are lower in ebs6-1 bri1-9 compared to atos9-t bri1-9 (Figure 4F). These results suggested that Gly¹⁹¹ is not essential for the biochemical function of AtOS9 but might affect its structure, as the AtOS9 abundance is significantly reduced in the ebs6-1 bri1-9 mutant (Supplemental Figure 4B).

Figure 4.

The atos9-t Mutation Blocks the ERAD of the Mutant BR Receptor bri1-9.

(A) Phenotypic comparison between bri1-9, atos9-t bri1-9, and three representative gAtOS9–GFP atos9-t bri1-9 transgenic lines expressing a genomic AtOS9–GFP transgene.

(B, C) Immunoblot analysis of AtOS9/AtOS9–GFP (B) and bri1-9 (C). Equal amounts of total proteins extracted from 2-week-old seedlings were separated by 10% SDS–PAGE and analyzed by an anti-AtOS9 antibody (B) or by an anti-BRI1 antibody (C).

(D, E) Images of 4-week-old (D) and 2-month-old (E) soil-grown plants of wild-type, bri1-9, and ebs6-1 bri1-9.

(F) Immunoblot analysis of the BRI1/bri1-9 abundance. Equal amounts of total proteins extracted from 2-week-old seedlings were treated with or without Endo H, separated by 7% SDS–PAGE, and analyzed by immunoblot with anti-BRI1 antibody. In (B), (C), and (F), coomassie blue staining of RbcS serves as a loading control.

The atos9-t Mutation Also Blocks the Degradation of Other ER-Retained Receptors

To determine whether AtOS9 is a general ERAD component rather than a specific component involved only in the ERAD of bri1-9, we crossed the atos9-t mutation into another weak bri1 mutant, bri1-5, whose BR-insensitive dwarf phenotype is also caused by ER retention and subsequent ERAD of the mutant BR receptor bri1-5 that carries a Cys⁶⁹–Tyr mutation (Hong et al., 2008). As shown in Figure 5A, the atos9-t mutation significantly increased the bri1-5 abundance and caused accumulation of a detectable amount of bri1-5 carrying the Endo H-resistant C-type N-glycan indicative of ER escape. Consistently, the atos9-t mutation suppresses multiple growth defects of the bri1-5 mutant. Compared with the bri1-5 mutant, the rosette of atos9-t bri1-5 mutant was much larger and its dark-grown hypocotyl and mature inflorescent stem are much taller (Figure 5B–5D).

Figure 5.

The atos9-t Mutation Inhibits the ERAD of bri1-5 and Misfolded EFR.

(A) Immunoblot analysis of bri1-5. Total proteins extracted from 2-week-old seedlings were treated with or without Endo H, separated by 7% SDS–PAGE, and analyzed by immunoblot with anti-BRI1 antibody.

(B–D) Images of 4-week-old soil-grown plants (B), 5-day-old dark-grown seedlings (C), and 2-month-old mature plants (D) of wild-type, bri1-5, and atos9-t bri1-5.

(E) Immunoblot analysis of EFR. Total proteins extracted from 2-week-old seedlings were treated with or without Endo H, separated by 7% SDS–PAGE, and analyzed by immunoblot with an anti-EFR antibody. In both (A) and (E), coomassie blue staining of RbcS serves as a loading control.

Several recent studies have shown that the correct folding of EFR, a BRI1-like LRR-RLK that functions as a cell-surface receptor for the bacterial translation elongation factor EF-Tu to elicit a plant immunity response (Zipfel et al., 2006), requires the UGGT-mediated CNX/CRT folding cycle (Li et al., 2009; Lu et al., 2009; Nekrasov et al., 2009). Loss-of-function mutations in UGGT or CRT3 result in misfolding of EFR and its subsequent ER retention and eventual degradation by an ERAD process. To investigate whether AtOS9 is also involved in degrading a misfolded EFR, we crossed the atos9-t mutation into a T-DNA insertional uggt mutant, ebs1-6 (Jin et al., 2007), and performed an immunoblot experiment using an anti-EFR antibody. Consistent with several earlier studies (Li et al., 2009; Lu et al., 2009; Nekrasov et al., 2009), no EFR protein was detected in ebs1-6 bri1-9, but the introduction of the atos9-t mutation elevated the EFR protein abundance to the same level as that of the bri1-9 or the atos9-t bri1-9 mutant (Figure 5E), thus supporting a general role of AtOS9 in the Arabidopsis ERAD process. It should be interesting to note that, unlike the two mutant BR receptors, the misfolded EFR protein is not significantly accumulated in the atos9-t ebs1-6 mutant (likely due to its lower biosynthesis rate compared to BRI1) and thus unlikely saturates its ER retention system to allow a small fraction of EFR protein to escape the ER. Consistent with this interpretation, the atos9-t ebs1-6 bri1-9 mutant remains insensitive to elf18, a biologically active epitope derived from the bacterial EF-Tu (Kunze et al., 2004), as seedlings of both ebs1-6 bri1-9 and atos9-t ebs1-6 bri1-9 mutants grew quite well on medium containing 100 nM elf18 that inhibits the growth of the atos9-t bri1-9 seedlings (Supplemental Figure 5).

Functional Study of Importance of Several Conserved Residues in the MRH Domain

To investigate the role of the MRH domain of AtOS9 in mediating the ERAD of bri1-9, we generated a p35S:AtOS9 transgene driven by the strong and constitutively active 35S promoter, mutated the resulting transgene by changing Tyr¹³², Gln¹⁴², and Glu²²¹ into Ala, Glu, and Asn, respectively (see Supplemental Figure 1 for the positions of mutated residues), and transformed the wild-type and mutant p35S:AtOS9 transgenes individually into the atos9-t bri1-9 mutant. All three conserved residues were previously shown to be crucial for the efficient ERAD of a well-studied ERAD client protein carboxpeptidase Y (CPY*, where * indicates an ER-retained mutant of the vacuolar enzyme) in yeast (Szathmary et al., 2005). As shown in Figure 6, both Gln¹⁴²–Glu and Glu²²¹–Asn mutations destroyed the ability of the p35S:AtOS9 transgene to rescue the atos9-t mutation, which is consistent with a recent structural study showing direct involvement of the two conserved residues in binding an α1,6 Man-exposed N-glycan (Satoh et al., 2010). Surprisingly, the Tyr¹³²–Ala mutation had no detectable effect on the AtOS9 function (Figure 6), whereas the exact same mutation completely destroyed the biochemical function of Yos9 in yeast cells (Szathmary et al., 2005). We thus concluded that the MRH domain is crucial for the biochemical function of AtOS9 in the Arabidopsis ERAD mechanism.

Figure 6.

The MRH Domain Is Important for the AtOS9 Function.

Shown here are images of representative transgenic atos9-t bri1-9 lines containing an empty vector or expressing the wild-type or a mutant p35S:AtOS9 transgene carrying one of three indicated single amino acid changes.

Genetic and Biochemical Interaction between AtOS9 and EBS5

It is well known that Yos9/OS-9 interacts directly with Hrd3/Sel1L to bring a terminally misfolded glycoprotein onto an ER membrane-anchored E3 ligase Hrd1, with the former recognizing a unique N-glycan signal while the latter detects a structural alteration (Denic et al., 2006; Gauss et al., 2006). The Arabidopsis genome has two Hrd3/Sel1L homologous genes, but only one (known as EBS5 or Sel1A) is functionally involved in the ERAD of the two mutant BR receptors while the other is a non-functional pseudogene (Liu et al., 2011; Su et al., 2011). To test whether AtOS9 directly interacts with EBS5, we performed an in vitro pull-down assay using E. coli-expressed fusion proteins of the luminal domain of EBS5 tagged at its N-terminus with maltose binding protein (MBP–EBS5) and the full-length AtOS9 tagged at its N-terminus with glutathione S-transferase (GST–AtOS9) or just the GST itself. As shown in Figure 7A, MBP–EBS5, immobilized on the amylose resin, was able to pull down only the GST–AtOS9 fusion protein, but not the GST tag itself, suggesting the two ERAD client recognition factors are capable of direct physical interaction. Our results are consistent with a recent study that demonstrated co-immunoprecipitation of AtOS9 and EBS5 (also known as Sel1A; Liu and Howell, 2010) when the two proteins were transiently expressed in tobacco leaves (Huttner et al., 2012).

Figure 7.

AtOS9 Interacts Biochemically and Genetically with EBS5.

(A) An in vitro pull-down assay. Equal amounts of the MBP–EBS5 fusion protein still bound to amylose resin were incubated with GST–AtOS9 or GST. After extensive washing, the proteins remained on the resin were dissolved in 2×SDS sample buffer, separated by 10% SDS–PAGE, and analyzed by immunoblot with an anti-GST antibody. Commassie blue staining of MBP–EBS5 shows the amounts of MBP–EBS5 used in the binding assay.

(B) Immunoblot analysis of AtOS9. Equal amounts of total proteins extracted from 2-week-old seedlings were separated by 10% SDS–PAGE and analyzed by immunoblot with an anti-AtOS9 antibody. Star indicates a cross-reacting band used for the loading control.

(C) Immunoblot analysis of the AtOS9 stability. Two-week-old seedlings were transferred into liquid ½ MS medium containing 180 μM CHX. Equal amounts of seedlings were removed at indicated incubation times to extract total proteins in 2×SDS sample buffer, which were subsequently analyzed by immunoblot with the anti-AtOS9 antibody. Coomassie blue staining of RbcS was used for a loading control.

(D) Shown here (from left to right) are images of 4-week-old soil-grown plants of wild-type, atos9-t ebs5-3 bri1-9, bri1-9, ebs5-3 bri1-9, and atos9-t bri1-9.

A potential AtOS9–EBS5 biochemical interaction was supported by an immunoblot study that revealed reduced AtOS9 abundance in an ebs5 mutant (Figure 7B) but detected normal EBS5 level in the atos9-t mutant (Supplemental Figure 6). Recent studies in yeast showed that Yos9 joins the ER membrane-embedded E3 ligase Hrd1 complex via its physical binding to Hrd3 (the yeast homolog of the Arabidopsis EBS5) (Denic et al., 2006; Gauss et al., 2006). We suspected that the stability of AtOS9 is reduced in the absence of the membrane-anchored linker protein EBS5 but should not be altered in the absence of the Hrd1 E3 ligase itself. Consistent with our prediction, a CHX decay assay demonstrated that the reduced AtOS9 level in the ebs5-3 mutant was indeed caused by increased degradation rather than decreased biosynthesis of AtOS9 (Figure 7C) but that simultaneous elimination of two Arabidopsis Hrd1 homologs had little effect on the stability of AtOS9 (Supplemental Figure 7). A requirement of EBS5 for the stability of AtOS9 is consistent with an earlier yeast study showing that Hrd3 is required for the stability of an HDEL-deleted form of Yos9 that depends on Hrd3 binding for its ER localization (Gauss et al., 2006).

The AtOS9–EBS5 interaction was further supported by our genetic studies. We reasoned that, if AtOS9 and EBS5 are physically and mechanistically linked in the same ERAD pathway, simultaneous elimination of both genes should exhibit no additive effect on suppressing the bri1-9 mutation compared to either single mutation. Indeed, as shown in Figure 7D, the atos9-t ebs5-1 bri1-9 triple mutant is morphologically indistinguishable from the atos9-t bri1-9 or ebs5-1 bri1-9 double mutant. Consistently, overexpression of EBS5 was not able to suppress the atos9-t mutation while overexpression of AtOS9 failed to compensate for the EBS5 deficiency (Supplemental Figure 8). These genetic results indicated that the efficient ERAD of bri1-9 requires both AtOS9 and EBS5.

DISCUSSION

In this study, we demonstrated a functional role of AtOS9 in the ERAD of three mutant/misfolded receptor-like kinases. AtOS9 is the only Arabidopsis protein that exhibits significant sequence homology with Yos9/OS-9 proteins and contains the highly conserved MRH domain known to bind α1,6 Man-exposed N-glycan. Consistent with previous genome-wide gene expression analyses, which revealed co-expression patterns between AtOS9 and other known/annotated ER chaperones, our immunoblot assay showed that the AtOS9 protein abundance is up-regulated by a treatment with TM known to induce an unfolded protein response (UPR) that increases the production of various ER chaperones and folding catalysts. Despite its lacking the C-terminal HDEL ER retrieval sequence, our confocal microscopic analysis coupled with the EndoH assay demonstrated that AtOS9 is an ER-localized protein. Furthermore, a T-DNA insertional mutation in AtOS9 resulted in a hypersensitivity to TM. More importantly, the atos9-t mutation caused significant increased stability of two mutant BR receptors and a misfolded EFR, thus providing the most direct genetic evidence for its involvement in the Arabidopsis ERAD system and indicating that AtOS9 is an evolutionarily conserved component of the eukaryotic ERAD machinery. Our discovery of the critical role of AtOS9 in the Arabidopsis ERAD mechanism was confirmed by a recent independent study (Huttner et al., 2012). It should be interesting to note that overexpression of Yos9 or the human OS-9, driven by the strong and constitutively active 35S promoter, failed to complement the atos9-t mutation (Supplemental Figure 9), suggesting that the C-terminal extensions of the yeast and mammalian proteins might interfere with their biochemical functions in Arabidopsis. It will be interesting to test whether the truncated version of the Yos9 or OS-9 (lacking their C-terminal extensions) is able to substitute AtOS9 to recognize the three tested misfolded receptor kinases in Arabidopsis.

The MRH domain is known to be essential for the ERAD lectin function for both Yos9 and mammalian OS-9 (Hosokawa et al., 2010; Mikami et al., 2010). A recent structural study revealed a flatten β-barrel structure composed of two eight-stranded anti-parallel β-sheets for the MRH domain of the human OS-9 and identified a total of eight amino acids (Trp¹¹⁷, Trp¹¹⁸, Gln¹³⁰, Asp¹⁸², Leu¹⁸³, Arg¹⁸⁸, Tyr²¹⁸, and Glu²¹²), which are directly involved in binding α1,6 Man-exposed N-glycans with the Trp¹¹⁷Trp¹¹⁸ motif specifically recognizing two covalently linked α1,6 Man residues (Satoh et al., 2010). Our sequence analysis revealed that the eight sugar-binding residues are absolutely conserved in AtOS9 and several other plant Yos9/OS-9 homologs (Figure 1), while our site-directed mutagenesis experiment confirmed the essential role of the MRH domain in the ERAD of bri1-9 as mutations of two absolutely conserved sugar-binding residues, Gln¹⁴² and Glu²²¹ (equivalent of Gln¹³⁰ and Glu²¹² in human OS-9), completely destroyed the AtOS9 function in the ERAD of bri1-9 (Figure 6). It is important to note that the Trp¹¹⁷ residue of the human OS-9 (the equivalent of the Trp¹²⁹ residue of AtOS9) directly binds the first α1,6 Man residue (Satoh et al., 2010); however, recent studies showed that the other α1,6 Man can also function as an ERAD signal to tag a misfolded glycoprotein (Quan et al., 2008; Clerc et al., 2009). It is therefore interesting to test whether the Trp¹²⁹ residue is absolutely required for the AtOS9 function in recognizing bri1-9 that carries shorter N-glycans exposing a different α1,6 Man residue that is predicted to bind the neighboring Trp¹³⁰ residue. In addition to such a genetic experiment, it is also important to perform biochemical experiments using recombinant AtOS9 proteins and synthetic N-glycan analogs to demonstrate a direct binding between AtOS9 and α1,6 Man-exposed glycans (Hosokawa et al., 2009).

Our biochemical and genetic studies suggested that AtOS9 functions together with EBS5, the Arabidopsis homolog of the yeast Hrd3/mammalian Sel1L (Su et al., 2011), in bringing in a misfolded glycoprotein for ERAD. First, we showed that AtOS9 could physically bind EBS5 in a simple in vitro pull-down assay, which is consistent with a recent independent study revealing an AtOS9–EBS5/Sel1A interaction in tobacco leaf cells (Huttner et al., 2012). Second, we discovered that the stability of AtOS9 depends on the presence of EBS5. This finding seems to be consistent with an earlier yeast study that revealed a requirement of Hrd3 (the yeast equivalent of EBS5) for the stability of a HDEL-lacking form of Yos9 (Gauss et al., 2006), which suggested that the Yos9–Hrd3 binding is required to retain the HDEL-deleted Yos9 in the ER. Because AtOS9 lacks the HDEL ER retrieval sequence, its ER retention might also depend on the AtOS9–EBS5 interaction. In addition, we found that an atos9-t ebs5-3 bri1-9 triple mutant was morphologically indistinguishable from either atos9-t bri1-9 or ebs5-3 bri1-9 double mutant (Figure 7D) and that overexpression of AtOS9 and EBS5 failed to compensate for the deficiency of EBS5 and AtOS9 in the ERAD of bri1-9, respectively (Supplemental Figure 8). All these results strongly support a plant ERAD model in which AtOS9 and EBS5 are physically and mechanistically linked in the same ERAD machinery that degrades the misfolded BR receptor in Arabidopsis. Further biochemical experiments, such as co-immunoprecipitation and purification of the AtOS9/EBS5-containing protein complex, are needed to prove that both AtOS9 and EBS5 are two integral components of the ER membrane-embedded AtHrd1 E3 ligase complex and to identify additional components of the Arabidopsis ERAD machinery.

METHODS

Plant Materials and Growth Conditions

All Arabidopsis mutants and transgenic lines used in this study are in the Col-0 ecotype, except bri1-5 (Ws-2) and bri1-9 (Ws-2) for genetic analyses and mapping. The T-DNA insertional mutant atos9-t (SALK_029413) was obtained from the Arabidopsis Biological Resource Center at Ohio State University and crossed with bri1-9 or bri1-5 mutant, while the ebs6-1 mutant was isolated in a previous genetic screen for suppressors of the bri1-9 mutant (Hong et al., 2009). Methods for seed sterilization and conditions for plant growth were described previously (Li et al., 2001). The root-growth inhibition assay on BR-containing medium was performed as previously described (Clouse et al., 1996).

Map-Based Cloning of the EBS6 Gene

The ebs6-1 bri1-9 (ecotype Col-0) was crossed with a bri1-9 (ecotype Ws-2; Noguchi et al., 1999), and the resulting F1 plants were self-fertilized to generate several F2 mapping populations. Genomic DNAs from segregating F2 seedlings exhibiting the ebs6-1 bri1-9-like morphology was extracted as previously described (Li and Chory, 1998) and used for PCR-based molecular mapping using known simple sequence-length polymorphism markers (see Supplemental Table 1 ).

Construction of Plasmids and Generation of Transgenic Plants

A 3.5-kb genomic fragment of At5g35080 lacking the predicted stop codon and its 3'-untranscribed/untranslated region was amplified from the genomic DNA of wild-type Arabidopsis plants and fused to a DNA fragment amplified from a previously described pBRI1:BRI1–GFP plasmid (Friedrichsen et al., 2000) that contains a synthetic gene for the green fluorescent protein (GFP) and the 3'-terminator of the pea (Pisum sativum) rbsc E9 gene (see Supplemental Table 1 for primers used for DNA amplification). They were subsequently cloned into the pPZP222 binary plasmid (Hajdukiewicz et al., 1994) to create pPZP222-gAtOS9–GFP. The full-length cDNA of AtOS9 was amplified by reverse transcription-PCR and subsequently cloned into the pENTR/D-TOPO vector (Invitrogen). The resulting pENTR-cAtOS9 plasmid was then mobilized using the Gateway LR Clonase (Invitrogen) into a gateway destination pMDC83 vector (Curtis and Grossniklaus, 2003) following the manufacturer’s suggested protocol to create a p35S:AtOS9 plasmid. Similar approaches were used to create p35S:Yos9 and p35S:HsOS-9 plasmid DNAs. Site-directed mutagenesis with the QuikChange XL site-directed mutagenesis kit (Stratagene) was performed on the p35S:AtOS9 plasmid to generate p35S:AtOS9(Tyr¹³²-Glu), p35S:AtOS9(Gln¹⁴²-Glu), and 35S:AtOS9(Glu²²¹-Asn) mutant plasmids. The construction of a genomic gEBS2 transgene and a p35S:EBS5 transgene was previously described (Jin et al., 2009; Su et al., 2011). Each created plasmid was sequenced to ensure no error within the PCR-amplified DNA fragment or at a cloning junction. These plasmids were individually transformed into the Agrobacterium GV3101 strain, and the resulting Agrobacterial strains were subsequently used to transform the atos9-t bri1-9 mutant by the vacuum-infiltration method (Bechtold and Pelletier, 1998).

Transient Expression and Confocal Microscopic Analysis of a GFP-Tagged AtOS9 Fusion Protein in Tobacco Leaves

The pPZP222–AtOS9:GFP, pSITE03–ER–RFP (encoding a red fluorescent protein (RFP) tagged at its C-terminus with the HDEL ER retrieval motif; Chakrabarty et al., 2007)), and p35S:p19 (encoding the p19 protein of tomato bushy stunt virus for suppressing gene silencing; Voinnet et al., 2003) plasmids were co-transformed into leaves of 3-week-old tobacco (Nicotiana benthamiana) plants via an Agrobacterium-mediated infiltration method (Voinnet et al., 2003). The localization patterns of AtOS9–GFP and the ER-localized RFP–HDEL in the co-infiltrated tobacco leaves were examined by using a Leica confocal laser-scanning microscope (TCS SP5 DM6000B) with an HCX PL APO CS 63X 1.30 glycerin lens and LAS AF software (Leica Microsystems). The GFP or RFP signal was excited by using the 488 or 543-nm laser light, respectively.

Expression of a Truncated GST–AtOS9 Fusion Protein and Generation of an Anti-AtOS9 Antibody

A 600-bp cDNA fragment encoding the N-terminal 200 amino acids of AtOS9 was cloned into pGEX–KG (Guan and Dixon, 1991) and subsequently transformed into BL21 competent cells (Novagen). The GST fusion protein was induced and purified according to the manufacturer’s recommended protocols. The purified GST–AtOS9 protein was sent to Pacific Immunology for custom antibody production, and a resulting antiserum was affinity-purified using the GST–AtOS9 fusion protein covalently immobilized on the Aminolink Plus coupling resin (Thermo Scientific) by the manufacturer’s suggested procedures.

Protein Extraction and Immunoblot Analyses (T7765, Sigma)

Two or 4-week-old Arabidopsis seedlings treated with or without CHX (Sigma), BL (Chemiclones, Inc., Canada), or TM (T7765, Sigma) were grounded in liquid N₂, dissolved in 2×SDS buffer (100 mM Tris-HCl, pH 6.8, 4% (w/v) SDS; 0.2% (w/v) bromophenol blue, 20% (v/v) glycerol, and 200 mM β-mercaptoethanol), and boiled for 10 min. After 10-min centrifugation, supernatants were used directly for immunoblot analysis or incubated with or without 1000 U of Endo Hf in 1×G5 buffer (New England Biolabs) for 1 h at 37°C, and the treated proteins were separated by 7 or 10% SDS–PAGE and analyzed by coomassie blue staining or by immunoblot with antibody raised against BRI1 (Mora-Garcia et al., 2004), BES1 (Mora-Garcia et al., 2004), EBS5 (Su et al., 2011), EFR, or AtOS9.

An In Vitro Pull-Down Assay

The entire open reading frame of AtOS9 was obtained by RT–PCR amplification from total RNAs isolated from wild-type Arabidopsis plants and cloned into the bacterial expression vector pGEX–KG (Guan and Dixon, 1991). The induction and purification of the GST-tagged full-length AtOS9 protein were carried out as described above for a truncated AtOS9 fusion protein used for raising the anti-AtOS9 antibody, while the induction and purification of an MBP–EBS5 fusion protein were performed as previously described (Su et al., 2011). To perform the pull-down assay, purified soluble GST and GST–AtOS9 protein were incubated with equal amounts of MBP–EBS5 still bound to the amylose resin (New England Biolabs) for 2 h at 4°C. After washing for four times (5 min each) in PBS (10 mM phosphate buffer, 7.4 pH , 137 mM NaCl, 2.7 mM KCl) containing 1.0% (v/v) Triton X-100, the proteins still remained on the resin were dissolved in 50 μl 2×SDS sample buffer, separated by 10% SDS–PAGE, and analyzed by immunoblot using an anti-GST antiserum.

SUPPLEMENTARY DATA

Supplementary Data are available at Molecular Plant Online.

FUNDING

This work was partly supported by grants from National Institutes of Health (GM060519) and National Science Foundation (IOS 1121496) to J.L.