PAWH1 and PAWH2 are plant-specific components of an Arabidopsis endoplasmic reticulum-associated degradation complex

Endoplasmic reticulum-associated degradation (ERAD) is a unique mechanism to degrade misfolded proteins via complexes containing several highly-conserved ER-anchored ubiquitin ligases such as HMG-CoA reductase degradation1 (Hrd1). Arabidopsis has a similar Hrd1-containing ERAD machinery; however, our knowledge of this complex is limited. Here we report two closely-related Arabidopsis proteins, Protein Associated With Hrd1-1 (PAWH1) and PAWH2, which share a conserved domain with yeast Altered Inheritance of Mitochondria24. PAWH1 and PAWH2 localize to the ER membrane and associate with Hrd1 via EMS-mutagenized Bri1 Suppressor7 (EBS7), a plant-specific component of the Hrd1 complex. Simultaneously elimination of two PAWHs constitutively activates the unfolded protein response and compromises stress tolerance. Importantly, the pawh1 pawh2 double mutation reduces the protein abundance of EBS7 and Hrd1 and inhibits degradation of several ERAD substrates. Our study not only discovers additional plant-specific components of the Arabidopsis Hrd1 complex but also reveals a distinct mechanism for regulating the Hrd1 stability.

E ndoplasmic reticulum-associated degradation (ERAD) is an integral part of the ER-mediated protein quality control system, which constantly monitors the folding status of secretory and membrane proteins, repairs misfolding proteins, and degrades irreparable terminally misfolded proteins 1 . ERAD is a highly conserved degradation mechanism that involves substrate recognition, ubiquitination at the cytosolic surface of the ER membrane, retrotranslocation through ER membraneembedded retrotranslocons, and eventual degradation by cytosolic proteasome 2 . The ERAD machinery builds around several ER membrane-anchored ubiquitin (E3) ligases that recognize different types of ERAD clients carrying structural defects in their luminal domains, transmembrane domains, or cytosolic domains (known as ERAD-L, ERAD-M, and ERAD-C substrates, respectively) 3 . One of the well-studied ERAD system is a multiprotein complex centered around an ER membrane-anchored ubiquitin E3 ligase known as HMG-CoA reductase degradation1 (Hrd1) in yeast 4 [HRD1 and glycoprotein 78 (gp78) in mammals 5,6 ]. The Hrd1 complex is known to degrade ERAD-L and ERAD-M substrates and contains several other highly conserved membrane and luminal proteins 3 . They include Hrd3 7 [Sel1L (Suppressor of lin-12-like) in mammals 8 ], Yos9 (Yeast OS-9 homolog) 9,10 [OS-9 (Osteosarcoma amplified 9), and XTP3-B (XTP3-transactivated protein B) in mammals 11,12 ], Usa1 13 [U1 SNP1-associated protein 1; HERP for Homocysteine-induced ER Protein in mammals 14 ], Der1 (Degradation in the endoplasmic reticulum1) 15 [DERLIN1-3 (Der-like domain-containing 1-3) in mammals 16 ]. The Hrd1 complex also includes one or more ubiquitinconjugating enzyme (E2), such as the ER membrane-anchored UBC6 (Ubiquitin-Conjugating 6) [Ube2j2 (Ubiquitin-conjugating enzyme E2 j2) in mammals 17 ] and a cytosolic E2 UBC7 (Ube2g2 in mammals 18 ) with its ER membrane-anchored recruiter Cue1 (Coupling of Ubiquitin conjugation to ER degradation that has no mammalian homolog) 19 . Biochemical and genetic studies in yeast and mammalian cells have shown that terminally misfolded glycoproteins with a unique asparagine-linked glycan (N-glycan) structure carrying an exposed α1,6-mannose residue are recruited to Hrd1 through a bipartite recruitment mechanism that uses Yos9/Os-9 to bind the α1,6-mannose-exposed N-glycans and Hrd3/Sel1L to bind exposed hydrophobic amino acid patches of misfolded glycoproteins 20 . A recruited ERAD client is subsequently ubiquitinated and retrotranslocated, which is likely mediated by Hrd1 in yeast 21 , into the cytosol where the ubiquitinated ERAD substrate is degraded by the 26S proteasome. In both yeast and mammalian cells, deleting Hrd3/Sel1L significantly reduces the stability of Hrd1/HRD1 7,22,23 . Interestingly, while the Δhrd3-induced autodegradation requires Usa1 in yeast cells 24 , the loss-of-Sel1L-caused HRD1 instability does not involve HERP in mammalian cells 23,25 .
Recent studies have revealed that Arabidopsis has a similar Hrd1-mediated ERAD machinery 26 . This system is known to degrade two ER-retained mutant forms of the brassinosteroid (BR) receptor BRASSINOSTEROID-INSENSITIVE 1 (BRI1), bri1-5 and bri1-9 27,28 , a misfolded conform of an Arabidopsis immunity receptor EF-Tu Receptor (EFR) produced in an Arabidopsis mutant defective in an ER-luminal protein folding sensor 29,30 , several engineered ERAD substrates [31][32][33][34] , and an ERAD component 35 . These studies not only identified conserved ERAD components in Arabidopsis, such as Hrd1a and Hrd1b 35,36 , EBS5 (EMS-mutagenized bri1 suppressor 5; also known as HRD3A) 34,36 , EBS6 (also known as AtOS9) 37,38 , and UBC32 (Ubiquitin Conjugase 32) 33 , which are homologs of the yeast Hrd1, Hrd3, Yos9, and Ubc7, respectively, but also discovered a plant-specific component known as EBS7 that regulates the protein stability of Hrd1a 39 . Analysis of the Arabidopsis genome fails to discover Arabidopsis homologs of Cue1 and Usa1 but identified three Arabidopsis homologs of Der1 [(Der1, Der2.1, and Der2.2) 40 . Compared to what has learnt from the yeast and mammalian systems, our knowledge of the plant Hrd1 ERAD complex remains limited 26 . For example, little is known about the composition and organization of the Arabidopsis Hrd1complex, and it remains to be determined if the Arabidopsis Hrd1 ERAD complex contains a Der1/Derlin homolog. We also know little about how the protein stability and biochemical activity of Hrd1 are regulated in plants and whether or not the Arabidopsis Hrd1 is also involved in retrotranslocating its ERAD clients. In order to expand our understanding of the plant ERAD mechanism, we took a proteomic approach with independently generated transgenic lines expressing epitope-tagged Hrd1a/EBS7 to identify proteins that interact with both EBS7 and Hrd1a. Our subsequent biochemical and genetic studies demonstrated that two paralogous Arabidopsis proteins, named hereinafter as PAWH1 and 2 for Protein Associated With Hrd1, are crucial core components of the Arabidopsis Hrd1-containing ERAD machinery and are required to maintain the protein stability of both EBS7 and Hrd1a.

Results
Identification of PAWH1 and PAWH2 via a proteomic approach. To identify additional components of the Arabidopsis Hrd1-complex, we employed a proteomic approach of immunoprecipitation coupled with mass spectrometry. We generated several transgenic lines expressing a fusion protein of green fluorescent protein (GFP) with the Arabidopsis Hrd1a (Hrd1a-GFP) 36 , one of the two Arabidopsis Hrd1 homologs, or MYC/ HA-tagged EBS7, a newly identified plant-specific component of the Arabidopsis ERAD machinery crucial for maintaining the Hrd1a stability 39 . Our previous studies have shown that simultaneous elimination of the two Arabidopsis Hrd1 homologs, Hrd1a and Hrd1b, or loss-of-function mutations in EBS7, inhibit ERAD of bri1-5 and bri1-9, two ER-retained mutant variant of the BR receptor BRI1 carrying a Cys 69 -Tyr mutation and a Ser 662 -Phe mutation in its extracellular domain, respectively 28,41 . Consequently, a small percentage of ER-accumulated bri1-5 and bri1-9 proteins leak out of the ER, likely due to saturation of their retention mechanisms, to reach the PM where the two mutant receptors bind extracellular BRs to promote plant growth, resulting in phenotypic suppression of the corresponding bri1-5 and bri1-9 dwarf mutants 36,39 . The three transgenes were able to rescue the corresponding hrd1a hrd1b bri1-9 and ebs7-3 bri1-5 mutants ( Supplementary Fig. 1), respectively, indicating that all three tagged proteins are physiologically functional. We used one representative transgenic line for each transgene to extract total proteins or microsomal proteins and subsequently performed immunoprecipitation (IP) experiments with antibody-conjugated beads. The resulting immunoprecipitates were analyzed by liquid chromatography coupled with tandem mass spectrometry (LC-MS/MS) to identify proteins that were coimmunoprecipitated with the GFP-fused Hrd1a or MYC/HA-tagged EBS7. We also included the non-transgenic wild-type plant as our negative control to eliminate proteins that bound non-specifically to antibody-conjugated beads. Comparison of the five sets of coimmunoprecipitated proteins identified nine common proteins (Fig. 1a, b), including Hrd1a, EBS7, and a previously demonstrated component of the Arabidopsis ERAD machinery, EBS5 36 (also known as HRD3A 34 or SEL1L 38 that is the Arabidopsis homolog of the yeast Hrd3 and mammalian Sel1L). It is interesting to note that the three IP experiments with total proteins also identified Hrd1b and EBS6 (the Arabidopsis homolog of Yos9/OS-9 37,38 ; Supplementary Fig. 2), suggesting the presence of multimeric Hrd1 in the Arabidopsis Hrd1 complex. The identification of known ERAD components as the abundant interacting proteins of both Hrd1a and EBS7 indicated success of our proteomic approach, whereas the failure to detect Hrd1b and EBS6 in the anti-GFP immunoprecipitates of the microsomal preparation was likely caused by low recovery of the immunoprecipitated proteins, evidenced by lower coverage of Hrd1a and EBS5 compared to a similar coimmunoprecipitation (coIP) experiment using the total proteins ( Supplementary Fig. 2). Our analysis also identified two highly homologous proteins, At4g17420 (285 amino acids) and At5g47420 (282 amino acids) that were previously annotated as tryptophan RNA-binding attenuator protein-like proteins (TRAPs) and were renamed hereinafter as Protein Associated With Hrd1-1 (PAWH1) and PAWH2, respectively ( Fig. 1b and Supplementary Fig. 2). A simple BLAST search revealed that PAWH1 and PAWH2 are highly conserved in the plant kingdom and contain AIM24 domain ( Supplementary Fig. 3), which is originally discovered in the yeast mitochondria AIM24 (Altered Inheritance of Mitochondria protein 24) recently implicated in stabilizing the mitochondria contact site complex and the respiratory chain supercomplexes 42 . The four other proteins recovered in all 4 IP-MS experiments include a jacalin-related lectin (At3G16460), a calcium-dependent lipid-binding protein (At3g14590), a germinlike protein (At5g20630), and one of the three Arabidopsis translational initiation factor EIF4As (At1g54270). Further studies are needed to determine whether they are bona fide components of the Arabidopsis Hrd1 complex.
Analysis of PAWH interactions with EBS7 and Hrd1a. To test if PAWHs directly interact with Hrd1a and EBS7, we performed three experiments. The first one was a simple yeast two-hybrid assay with the predicted AIM24 domain-containing N-terminal 260 amino acids (AAs) of PAWH1 (PAWH1-N260) or 257 AAs of PAWH2 (PAWH2-N257) and the N-terminal soluble domain of EBS7 containing its N-terminal 142 AAs (EBS7-N142) or the cytoplasmic RING finger domain of Hrd1a (Hrd1a-CD). These assays showed that while PAWH1-N260 (also PAWH2-N257) interacted well with the EBS7-N142 fragment, it failed to interact with Hrd1a-CD that was previously shown to interact with the EBS7-N142 fragment 39 (Supplementary Figs. 4 and 5a). The second experiment was a transient bimolecular fluorescence complementation assay (better known as BiFC 43 ) in tobacco leaf epidermal cells using the full-length PAWH1 and PAWH2 fused at their N-termini with the N-terminal half of the yellow fluorescent protein (nYFP-PAWH1 and nYFP-PAWH2) and the fulllength EBS7 fused at its N-terminus with the C-terminal half of YFP (cYFP-EBS7) or the full-length Hrd1a fused with cYFP at its C-terminus (Hrd1a-cYFP). While yellow fluorescent signals were detected in tobacco leaf cells coexpressing nYFP-PAWH1 or nYFP-PAWH2 with cYFP-EBS7 (Fig. 1c), no fluorescent signal was detected in tobacco leaf cells coexpressing nYFP-PAWH1 or nYFP-PAWH2 with Hrd1a-cYFP ( Supplementary Fig. 5b). In addition, we performed a coIP experiment with total proteins from the wild-type Arabidopsis seedlings using anti-EBS7 antibody 39 , a custom-made anti-Hrd1a antibody ( Supplementary  Fig. 6), and a custom-made anti-PAWH antibody that could identify both PAWH1 and PAWH2 proteins (see Supplementary  Fig. 11d). As shown in Fig. 1d, the anti-EBS7 antibody not only immunoprecipitated the endogenous EBS7 protein but also brought down Hrd1a and the two PAWH proteins. By contrast, neither Hrd1a nor PAWHs were detected in a similar coIP assay with the total proteins extracted from the null ebs7-3 mutant 39 (Fig. 1d). These results suggested that PAWH1/2 could directly AT4G29960 EBS7 AT1G18260 EBS5 AT3G16090 Hrd1a AT4G17420 PAWH1 AT5G47420 PAWH2 AT3G16460 Mannose-binding protein AT3G14590 Ca 2+ -dependent lipid-binding protein ARTICLE bind EBS7 but not Hrd1a and that the coIP-detected Hrd1a-PAWH1/2 association is likely mediated by EBS7 known to interact directly with Hrd1a 39 .
PAWHs are ER membrane proteins and induced by ER stress.
Both PAWH genes are widely expressed in various tissues/organs throughout the Arabidopsis development with PAWH1 exhibiting high expression at later stages of seed development and reaching its highest expression level in dry seed ( Supplementary  Fig. 7). The two PAWH genes were previously shown to coexpress with genes known/predicted to be involved in protein folding and/or protein quality control ( Supplementary Fig. 8) 44 . Many PAWH-coexpressed genes were known to be induced by ER stress 45 . To test if PAWH transcripts and PAWH proteins are also induced by ER stress, we treated the wild-type seedlings with or without tunicamycin (TM), a widely used ER stress inducer that inhibits the first biosynthetic step of the N-glycan precursor, and used the treated seedlings to examine the abundance of PAWH transcripts and PAWH proteins by RT-PCR and immunoblotting, respectively. We found that the TM treatment increased the mRNA levels of both PAWH genes and elevated the protein abundance of both PAWHs ( Supplementary Fig. 9). Both PAWHs lack the N-terminal signal peptide or the Cterminal ER retrieval motifs but were predicted to contain a potential transmembrane (TM) segment near their C-termini ( Supplementary Fig. 10). To directly determine their subcellular localization, we generated GFP-fusion transgenes for the two PAWH genes and transiently expressed them in tobacco leaf epidermal cells. Confocal microscopy analysis of agro-infiltrated tobacco leaves revealed that the green fluorescent signals of the GFP-PAWH1/2 fusion proteins overlapped with the fluorescent signals of the red fluorescent protein (RFP) tagged at its Cterminus with the HDEL (histidine-aspartate-glutamate-leucine) ER retrieval motif (Fig. 2a), a widely used ER-localized marker, strongly suggesting that PAWH1/2 are localized to the ER. A further confirmation was provided by sucrose density-gradient centrifugation with Arabidopsis microsomal proteins in the presence or absence of Mg 2+ . Because Mg 2+ is required for polyribosome binding to the ER and that the density of the ribosome-bound ER membrane is higher than that of the ribosome-free ER or other microsomal membranes, an ERlocalized protein should undergo a diagnostic Mg 2+ -dependent shift from lower density to higher density on a sucrose gradient 46 . Figure 2b shows that PAWHs exhibited a similar Mg 2+ -dependent density shift as BiPs (binding immunoglobulin proteins, ERlocalized Heat Shock Protein 70), but differs from BRI1 known to be localized to the PM 47 . Consistent with the predicted Cterminal TM segments ( Supplementary Fig. 10), immunoblot analysis of PAWHs in soluble and insoluble fractions of resuspended Arabidopsis microsomal pellets in 0.1 M NaCl, 0.1 M Na 2 CO 3 , 1% (v/v) Triton X-100, or 1% (v/v) Nonidet P-40 solution showed that only the two nonionic detergents could release the two PAWHs from microsomes, suggesting that both PAWHs are likely anchored to the ER membrane (Fig. 2c).

Mutating both PAWHs stabilizes several ERAD substrates.
To investigate if the two PAWHs are involved in an Arabidopsis ERAD process, we obtained T-DNA insertional mutants for the two PAWH genes from the Arabidopsis Biological Resource Center (http://abrc.osu.edu/) 48 (Supplementary Fig. 11a, b). RT-PCR analysis indicated that the two mutants failed to produce the full-length transcript of the corresponding PAWH gene while immunoblot analysis failed to detect one of the two crossingreacting bands of the wild-type plant ( Supplementary Fig. 11c, d), indicating that the two T-DNA insertional mutants are null mutants. We crossed each mutant into bri1-9 and found that the resulting pawh1 bri1-9 and pawh2 bri1-9 mutants were morphologically indistinguishable from bri1-9 ( Fig. 3a-c). A further cross between the two double mutants generated a pawh1 pawh2 bri1-9 triple mutant that was morphologically similar to the wildtype (Fig. 3a-c), indicating that simultaneous elimination of PAWH1 and PAWH2 suppressed the bri1-9 dwarf phenotype. Consistent with the phenotypic suppression, the pawh1 pawh2 double mutation partially restored the BR sensitivity to the bri1-9 mutant revealed by the BR-triggered root growth inhibition assay 49 ( Supplementary Fig. 12) and the BR-induced BES1 dephosphorylation assay 50 (Fig. 3d). Importantly, immunoblot analysis showed that the pawh1 pawh2 double mutation greatly increased the bri1-9 abundance, which was much higher than the abundance of the wild-type BRI1 that undergoes a phosphorylation-dependent endocytosis and ubiquitin-mediated degradation 51 . We also treated total proteins with endoglycosidase H (Endo H) that cleaves high mannose-type N-glycans of ER-localized/retained glycoproteins but does not cut Golgiprocessed N-glycans of glycoproteins. We found that the pawh1 pawh2 mutation increased abundance of the Endo H-resistant form of bri1-9 ( Fig. 3e), suggesting increased presence of bri1-9 on the PM to promote growth. Indeed, our transgenic experiment showed that overexpression of EBS2 known to retain bri1-9 in the ER 52 , significantly decreased the amount of the Endo H-resistant form of bri1-9 and nullified the suppressive effect of the pawh1 pawh2 double mutation on the bri1-9 dwarfism while having a little effect on the total amount of bri1-9 proteins ( Supplementary  Fig. 13). To determine the biochemical basis of increased bri1-9 abundance in the pawh1 pawh2 bri1-9 mutant, we performed a cycloheximide (CHX)-chasing experiment and found that increased bri1-9 abundance is caused by decreased protein degradation rather by increased protein biosynthesis (Fig. 3f). A subsequent rescue experiment showed that the effect on growth and biochemical phenotypes were indeed caused by the double pawh1 pawh2 mutation ( Supplementary Fig. 14). Taken together, these experiments demonstrated that PAWH1 and PAWH2 are crucial components of the Arabidopsis ERAD machinery that degrades the ER-retained mutant bri1-9 receptor. A similar set of experiments revealed that simultaneous elimination of the two PAWHs by the T-DNA insertional mutations or by CRISPR/Cas9-mediated genome editing inhibited the degradation of bri1-5 and suppressed the dwarf phenotype of the corresponding bri1-5 mutant (Supplementary Figs. 15 and 16). We also tested if the pawh1 pawh2 double mutation could inhibit degradation of other misfolded Arabidopsis proteins. Previous studies showed that the Arabidopsis EF-Tu receptor (EFR), a cell surface-localized receptor that detects and binds a small peptide derived from the highly conserved translation factor EF-TU (elongation factor thermo unstable) of pathogenic bacteria 53 , becomes misfolded and degraded in an Arabidopsis mutant defective in an ER-luminal protein homologous to the mammalian UDP-glucose:glycoprotein glucosyltransferase 29,30 (also known as EBS1 41 ). We crossed the pawh1 pawh2 mutation into an ebs1 mutant and analyzed the EFR abundance in ebs1, pawh1 pawh2, ebs1 pawh1 pawh2, and their wild-type control. Consistent with the previous findings 29, 30 , EFR was non-detectable in ebs1 but its level was greatly increased in the ebs1 pawh1 pawh2 triple mutant ( Supplementary Fig. 17), demonstrating that the two PAWHs are also involved in degrading a misfolded EFR. We thus concluded that PAWH1 and PAWH2 are likely general components of the Arabidopsis Hrd1-containing ERAD complex.
The loss of PAWHs activates UPR and reduces stress tolerance. Previous studies have shown that loss-of-function ERAD  [36][37][38][39] , a highly conserved ER stress response pathway that upregulates production of ER chaperones and ERAD components to maintain proteostasis 54,55 . To examine if the pawh1 pawh2 mutation also activates UPR, we performed an immunoblot analysis with total proteins of the wild-type and pawh1 pawh2 mutant treated with or without TM for variable durations. Figure 4a shows that the protein abundance of BiPs, protein disulfide isomerases (PDIs), calreticulins/calnexins (CRT/CNXs), EBS5, and EBS6 was higher   Fig. 4 The pawh1 pawh2 mutation activates UPR and reduces stress tolerance. a Immunoblot analysis of various ER-localized proteins in seedlings treated with TM for different durations. The faster moving bands detected by anti-PDI and anti-EBS5 antibody are non-glycosylated forms of PDIs and EBS5, respectively. The star indicates a non-specific band. The lower strip is the Ponceau Red-stained RbcS bands used as the loading control. b, c The left are photographs of 15-day-old Arabidopsis seedlings grown on ½ MS medium supplemented with or without 1.5 mM DTT (b) or 150 mM NaCl (c), while the right are the bar graphs showing the percentage of three kinds of seedlings: blue (alive), brown (dying), and yellow (dead). Scale bar = 1 cm. The stress sensitivity experiments were repeated three times with~100 seedlings/each for the DTT assay and~50 seedlings/each for the NaCl experiment with individual data points shown as open circles distributed above or below the average values of the three repeats. Source data are provided as a Source Data file in pawh1 pawh2 seedlings treated with or without TM than the corresponding wild-type seedlings, indicating that the pawh1 pawh2 mutation constitutively activates the UPR pathway. The detected enhancement of UPR prompted us to test if the pawh1 pawh2 double mutation affects the plant stress tolerance, which is known to involve the UPR pathway 55 . We germinated seeds and grew seedlings on ½ MS medium containing 1.5 mM dithiolthreitol (DTT, a widely used ER stress-inducer that interferes with oxidative protein folding) and 150 mM NaCl known to cause salt stress in Arabidopsis. As shown in Fig. 4b, c, the percentages of dying and dead seedlings on DTT/NaCl-containing medium were markedly higher for the pawh1 pawh2 mutant than the wildtype control, indicating that the pawh1 pawh2 double mutation compromises the plant stress tolerance.
Both PAWH1 and PAWH2 are unstable in the ebs7-1 mutant. Given the facts that EBS7 interacts with both Hrd1a and PAWHs and that ebs7 mutations greatly destabilize Hrd1a 39 , we were interested in examining the impact of the ebs7-1 mutation on the PAWH1/2 abundance as well as the impact of the pawh1 pawh2 mutation on the protein stability of both EBS7 and Hrd1a. We also included three other ERAD mutants, ebs5 bri1-9, ebs6 bri1-9, and hrd1a hrd1b bri1-9 in our experiment. Consistent with our earlier finding 39 , the missense ebs7-1 mutation (Ala 131 -Thr) caused disappearance of not only the mutated ebs7-1 protein but also Hrd1a (Fig. 5a). Interestingly, ebs7-1 also markedly reduced the protein abundance of the two PAWHs (Fig. 5a). By contrast, the ebs6 mutation seemed to have a marginal effect on the abundance of EBS7, PAWHs, or Hrd1a, whereas the ebs5 mutation increased the protein levels of EBS7 and PAWHs (Fig. 5a), which could be caused by non-significant and significant impact of the ebs6 and ebs5 mutations, respectively, on the transcript abundance of EBS7 and PAWHs (Supplementary Fig. 18). To determine if the ebs7-1-caused reduction of PAWH abundance is attributed to increased protein degradation, we first treated seedlings of wild-type and ebs7-1 with TM for 12 h to increase the PAWH1/2 abundance, subsequently treated these seedlings with CHX or MG132 (a widely used proteasome inhibitor) for different durations, and analyzed the PAWH abundance by immunoblotting. Figure 5b shows that while the CHX treatment had little impact on the PAWH1/2 abundance in wild-type seedlings, the same treatment led to a rapid decrease in the PAWH1/2 abundance in ebs7-1 seedlings. Consistently, a 12-htreatment of MG132 caused little change or slightly reduction in the PAWH abundance in wild-type seedlings, whereas the same MG132 treatment actually increased the PAWH1/PAWH2 protein levels in the ebs7-1 mutant (Fig. 5c). Together, these experiments indicated that ebs7-1 destabilizes the two PAWH proteins that are likely degraded via a proteasome-mediated process.
The pawh1 pawh2 mutation destabilizes EBS7 and Hrd1a. Our immunoblot assays of various ERAD mutants also revealed that the pawh1 pawh2 double mutation greatly reduced the protein abundance of both EBS7 and Hrd1a (Fig. 5a), while our quantitative real-time RT-PCR (qPCR) analyses showed that the pawh1 pawh2 double mutation slightly elevated the levels of the EBS7 mRNA but had a marginal impact on the Hrd1a transcript abundance (Supplementary Fig. 18). To determine if the observed reduction in EBS7 and Hrd1a protein abundance is due to increased protein degradation or decreased protein synthesis, we used the same TM-CHX/MG132 protocol described above to treat seedlings of the wild-type and the pawh1 pawh2 double mutant and analyzed the protein abundance of EBS7 and Hrd1a by immunoblotting. As shown in Fig. 6a, b, neither CHX or MG132 treatment had an observable effect on the protein abundance of EBS7 or Hrd1a in the wild-type seedlings. However, the CHX treatment caused a gradual reduction of EBS7 but very rapid disappearance of the TM-induced Hrd1a (Fig. 6c), whereas the MG132 treatment gradually stabilized EBS7 and caused rapid and significant stabilization of the ER-anchored E3 ligase (Fig. 6d). We concluded that the pawh1 pawh2 double mutation destabilizes both EBS7 and Hrd1a that are likely degraded via a proteasome-mediated mechanism. Because PAWHs fail to interact directly with Hrd1a (Supplementary Fig. 5) and the pawh1 pawh2 mutation greatly reduces the protein abundance of EBS7 (Fig. 5a), which is known to be important to maintain the Hrd1a stability 39 , we wondered whether the impact of the pawh1 pawh2 mutation on Hrd1a stability is a secondary effect caused by reduced EBS7 abundance. We therefore overexpressed EBS7 using the strong p35S promoter to drive the expression of HA-tagged EBS7 in the pawh1 pawh2 bri1-9 triple mutant. This experiment revealed that while the p35S::HA-EBS7 transgene did lead to increased EBS7 protein abundance in the triple mutant, it failed to stabilize Hrd1a or decrease the bri1-9 abundance (Supplementary Fig. 19), suggesting that PAWHs exert their direct impact on the ER-anchored E3 ligase via a yet unknown mechanism.
Our immunoblot analysis also revealed that the hrd1a hrd1b double mutation did not reduce but increased the abundance of EBS5, EBS6, EBS7, and the two PAWH proteins (Fig. 5a). Because our qPCR experiments showed that the hrd1a hrd1b double mutation significantly increased the transcript levels of both PAWH genes but had a marginal impact on the EBS7 transcript abundance ( Supplementary Fig. 18), we suspected that Hrd1a and/or Hrd1b might be the E3 ligase that ubiquitinates EBS7 in the pawh1 pawh2 mutant to promote EBS7 degradation. Consistent with this hypothesis, overexpression of a GFP-tagged Hrd1a driven by the strong p35S promoter not only reduced the bri1-9 abundance and complemented the phenotypes of the hrd1a hrd1b bri1-9 triple mutant, but also greatly lowered the protein levels of bri1-9, EBS7, and the two PAWHs (Supplementary Fig. 20). Because both EBS7 and PAWHs remained stable in the hrd1a hrd1b mutant background, we also used total proteins of the hrd1a hrd1b bri1-9 mutant to analyze the EBS7-PAWHs interaction and discovered that the hrd1a hrd1b mutation had no detectable impact on the EBS7-PAWHs interaction (Supplementary Fig. 21).

Discussion
In this study, we used a proteomic approach to identify two redundant proteins, PAWH1 and PAWH2, which were coimmunoprecipitated with GFP-tagged AtHrd1a and MYC/HA-tagged EBS7 expressed in their corresponding loss-of-function Arabidopsis mutants. Previous transcriptome analyses indicated that both PAWH genes were co-expressed with genes known or predicted to encode ER chaperone, folding catalysts, and ERAD components, while our current study confirmed that both PAWH1 and PAWH2 proteins were induced by chemically triggered ER stress. Despite the lack of an N-terminal signal peptide and known ER retention/retrieval sequence motifs, both PAWHs were found to be localized at the ER membrane, likely through its predicted C-terminal transmembrane segment, as evidenced by confocal microscopy, sucrose-gradient ultracentrifugation, and solubilization of microsomal proteins with different solvents and detergents.
In addition, our yeast two-hybrid assay, transient BiFC assay in tobacco leaves, and coIP experiments using transgenic Arabidopsis plants showed that PAWHs interact directly with EBS7 but The pawh1 pawh2 double mutation reduces the stability of EBS7 and Hrd1a. a-d Immunoblot analysis of EBS7 and Hrd1a in the wild-type (a, b) and pawh1 pawh2 double mutant (c, d). Total proteins were extracted from 10-day-old seedlings that were initially treated with 5 μg/mL TM for 12 h and subsequently treated with 180 μM CHX (a, c) or 80 μM MG132 (b, d) for different durations, separated by SDS-PAGE, and analyzed by immunoblotting using anti-EBS7 and anti-Hrd1a antibodies. In each immunoblotting experiment, the RbcS-labeled strip was used as a loading control. Asterisks and stars denote cross-reacting bands to anti-EBS7 and anti-Hrd1a antibodies, respectively, while arrows indicate the Hrd1a band. Source data are provided as a Source Data file associate with Hrd1a indirectly, most likely mediated by the previously demonstrated EBS7-Hrd1a interaction 39 . Furthermore, we showed that the pawh1 pawh2 double mutation constitutively activated the UPR pathway and greatly compromised plant tolerance against salt stress and DTT-triggered ER stress, similar to what were previously reported for Arabidopsis ERAD mutants [36][37][38][39] . More importantly, we have shown that the pawh1 pawh2 mutation blocked degradation of two ER-retained mutant BR receptors, bri1-5 and bri1-9, allowing them to accumulate and consequently leak out of the ER to reach the PM where the two mutant BR receptors could respond to BRs to stimulate plant growth. Our study also showed that the pawh1 pawh2 mutation also inhibited degradation of misfolded EFR in the ebs1 mutant background, strongly suggesting that the two redundant PAWH proteins are likely general components of the Arabidopsis Hrd1containing ERAD complex that degrades many misfolded proteins. Thus, our study, coupled with previously reported results, revealed that such an Arabidopsis ERAD complex consists of not only evolutionarily conserved components, including EBS5/ HRD3A 34,36 , EBS6/AtOS9 37,38 , Hrd1a/1b 36 , and their associated E2 conjugase UBC32 33 , but also plant-specific components, such as EBS7 39 and PAWH1/PAWH2 (Fig. 7).
Our results have shown that the plant-specific components, EBS7 and PAWHs, are important to maintain the protein stability of Hrd1a. In the absence of the two homologous PAWH proteins, the stability of EBS7 and Hrd1 was significantly reduced. Our previous study demonstrated that ebs7 mutations significantly reduce the Hrd1 stability 39 , raising a question if reduction of Hrd1 stability by the pawh1 pawh2 mutation is an indirect consequence of reduced abundance of EBS7 that directly binds PAWHs. However, our transgenic experiment showed that increased production of EBS7 driven by the p35S promoter failed to increase Hrd1a abundance, suggesting a direct role of PAWHs in maintaining the Hrd1a stability. This finding and our earlier results of the EBS7-AtHrd1a interaction demonstrated that EBS7 and PAWHs work together to regulate the stability of the ER membrane-anchored Hrd1a.
Studies in yeast and mammalian cells showed that Hrd3/Sel1L is essential to maintain the stability of Hrd1 and that mutations in Hrd3 in yeast or Sel1L in mammalian cells resulted in a significant reduction in the stability of the yeast or mammalian Hrd1 E3 ligase 7,22,23 . However, mutations of EBS5, the Arabidopsis homolog of Hrd3/Sel1L 36 , had no detectable impact on the Hrd1a stability. Our results thus revealed a novel mechanism that regulates the protein stability/activity of a plant ERAD E3 ligase, which is quite diverged from a conserved regulatory mechanism that controls the stability of yeast and mammalian Hrd1s. In yeast, the hrd3Δ-triggered Hrd1 loss is caused by autodegradation facilitated by the Hrd1's own catalytic RING domain 24 . It remains to be tested whether or not the disappearance of Hrd1a in ebs7 or pawh1 pawh2 mutant is caused by self-destruction of Hrd1a or by other ER-anchored E3 ligases such as SUPPRESSOR OF DRY2 DEFECTS1 56 , an Arabidopsis homolog of the yeast Degradation of alpha factor2-10 (Doa10) that is known to be involved in degrading misfolded transmembrane proteins carrying a cytosolic structural defect 3 or one of the three Arabidopsis RING finger proteins with membrane anchor (RMA1-3) 57 .
While loss-of-function mutations in EBS7 and PAWHs destabilize PAWHs/Hrd1a and EBS7/Hrd1a, respectively, the hrd1a hrd1b double mutation does not reduce but increases the protein abundance of both EBS7 and PAWHs. We hypothesize that this is caused by increased transcription and reduced protein degradation. The hrd1a hrd1b double mutation may cause accumulation of misfolded proteins in the ER and consequently activates the UPR pathway, which increases production of many chaperones, folding enzymes, and components of the ERAD machinery such as EBS7 and PAWHs. Moreover, the two functionally redundant Hrd1 homologs likely trans-ubiquitinate EBS7 to regulate its protein abundance via the so-called ER-tuning mechanism 35,58 to tightly and rapidly regulate the ERAD capacity to adapt to fluctuations in the amount of misfolded proteins accumulated in the ER, thus maintaining the ER proteostasis essential for cell survival. A further investigation is needed to fully understand the role of Hrd1 in such an ER-tuning regulatory mechanism.
Although our study demonstrated an essential role of PAWHs in maintaining the stability of Hrd1a, it remains unknown how the two PAWHs execute this function. Both PAWH1 and PAWH2 carry a highly conserved domain that was first discovered in the yeast AIM24 protein, a mitochondria protein tightly associated with the inner membrane and required for the ultrastructure, composition, and function of mitochondria 42,59 . It was previously thought that AIM24 is a fungal invention 60 , but the 32nd release of the Pfam database of protein families in Sept. 2018 (https://pfam.xfam.org) revealed a total of 5937 AIM24 domain-containing proteins from 3272 species in all three domains of life, with >93% of them containing only the AIM24 domain ( Supplementary Fig. 22). Despite wide occurrence of the AIM24 domain-containing proteins (a total of 4535 sequences) in bacteria, nothing is known about physiological function of any of these bacterial proteins. Interestingly, crystal structures of at least two bacterial AIM24 domain-containing proteins were solved: an unknown protein of 248 AAs (PA3696) from Pseudomonas aeruginosa strain PA01 (PDB: 1YOX), and a hypothetic protein of 243 AAs (SpyM3_0169) from Streptococcus pyogenes (PDB: 1PG6). Structural modeling at SWISS-MODEL server (https://swissmodel.expasy.org) using the structures of these two bacterial proteins as templates suggested that PAWH1/2 likely adopt a similar monomeric structure consisting of three structural repeats of beta-sandwich ( Supplementary Fig. 23a, b), each having two beta-sheets of 3/4 anti-parallel beta-strands. It is worth mentioning that each structure unit resembles the crystal structure of the bacterial tryptophan RNA-binding attenuation protein (PDB: 1WAP) 61 , explaining the original names of the two PAWHs as tryptophan RNA-binding attenuator-like proteins. Our structural modeling also suggested that PAWHs could form a homotrimeric subcomplex with a flat surface on the bottom ( Supplementary  Fig. 23c-f). It is important to note that the predicted C-terminal transmembrane α-helix is not included in the models and that the C-terminal α-helices shown in the two models correspond to the peptide fragment of 239-250 of PAWH1 (236-247 in PAWH2), which is unlikely a transmembrane domain due to its short length.  Fig. 7 A model of the Arabidopsis Hrd1 complex. An ERAD-L substrate is recognized and recruited by both EBS6/AtOS9 and EBS5/HRD3 to the ER membrane-anchored Hrd1 that works with the E2 UBC32 to ubiquitinate the ERAD client. EBS7 and PAWH1/2 interact with Hrd1 to regulate the stability and activity of the E3 ligase NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-11480-7 ARTICLE Such a modeled PAWH structure suggests that PAWHs likely rely on their flat ring-surface to interact with other proteins and use the protruding α-helices (corresponding to the consensus C-terminal transmembrane α-helical domain, see Supplementary Fig. 10) to attach the trimeric PAWH complex into the ER membrane. Further studies are needed to determine if the predicted C-terminal transmembrane domain is needed for the biological function of PAWHs, and more importantly, to identify additional PAWH-interacting proteins and/or crucial amino acid residues of PAWH1/2, which could help to understand the biochemical mechanism(s) by which PAWHs regulate the protein stability and catalytic activity of the plant Hrd1.

Methods
Plant materials and growth conditions. Most of the Arabidopsis wild-type, mutants, and transgenic lines are in the Columbia (Col-0) ecotype except for mutants and transgenic lines that carry the bri1-5 mutation, which are in the Wassilewskija-2 (Ws-2) ecotype. The Arabidopsis mutants used in this study include bri1-5 62 , bri1-9 41 , ebs1 41 , ebs5 36 , ebs6 37 , ebs7-1 /339 , and hrd1a hrd1b 36 . The T-DNA insertional mutants CS335767 (pawh1) and SALK_111654 (pawh2) were obtained from Arabidopsis Biological Resource Center (ABRC) at Ohio State University 48,63 . The pawh1-c and pawh2-c mutants were created by CRISPR/Cas9mediated genome editing in the bri1-5 mutant background (see below for the experimental details). Seed were surface sterilized using the ethanol-washing protocol 64 and seedling were grown at 22°C in growth chamber or growth room under long-day (16 h-light/8 h-dark) photoperiodic condition.
Chemical treatment of Arabidopsis seedlings. To study the stress tolerance, seeds were germinated, and young seedlings were grown on ½ Murashige and Skoog (MS, Sigma) medium supplemented with or without 1.5 mM DTT (Roche) or 150 mM NaCl for 15 days and the numbers of seedlings that were green (alive), yellowish (dying), or dead were counted and recorded. To analyze UPR activation, 10-day-old Arabidopsis seedlings were carefully transferred into liquid ½ MS medium supplemented with 5 μg/mL TM (BioMol) and were incubated for 0, 3, 6, 9, or 12 hours before samples were harvested into liquid nitrogen. To analyze the BR sensitivity, seeds were germinated, and seedlings were grown on ½ MS medium containing varying concentration of brassinolide (BL, Wako Chemical) for 7 days under a long-day (16-h-light/8-h-dark) photoperiodic growth condition. The seedlings were photographed, and their root length were measured digitally by ImageJ (https://imagej.nih.gov/ij/). Source data of the measurement are provided as a Source Data file. To study the BR-induced BES1 dephosphorylation, 10-day-old Arabidopsis seedlings were carefully transferred into liquid ½ MS medium supplemented with or without 1 μM BL, incubated for 2 hours, and subsequently harvested into liquid nitrogen. To study protein degradation, 10-day-old Arabidopsis seedlings were carefully transferred into liquid ½ MS medium containing 180 μM CHX (Sigma), incubated for varying durations, and harvested into liquid nitrogen for protein extraction. To study the impact of ebs7 and pawh1 pawh2 mutations on protein stability, 10-day-old Arabidopsis seedlings were pretreated with 5 μg/mL TM in liquid ½ MS medium for 12 hours before being treated with 180 μM CHX or 80 μM MG132 (Sigma).
Generation of transgene constructs and transgenic plants. A~4.1-kb genomic fragment for each PAWH gene was amplified from genomic DNAs of the wild-type Arabidopsis Col-0 seedlings using the gPAWH1 and gPAWH2 primer sets (see Supplementary Table 1) and cloned into pCambia1300 (https://cambia.org/ welcome-to-cambialabs/) to generate pCambia1300-gPAWH1 and pCambia1300-gPAWH2 transgenes. The p35S::Hrd1a-GFP transgene was created by cloning a 1,479-bp Hrd1a cDNA fragment amplified from the 1st strand cDNAs converted from total RNAs of the wild-type Arabidopsis seedlings using the Hrd1a-GFP primer set (Supplementary Table 1) into the pCambia1300-p35S::C-GFP vector. To create p35S::HA-EBS7 or p35S::MYC-EBS7 transgene, an 876-bp EBS7 cDNA fragment was amplified from the same Arabidopsis 1st cDNA preparation using the HA-EBS7 and MYC-EBS7 primer set (Supplementary Table 1) and cloned into the pCambia1300-p35S::N-HA or pCambia1300-p35S::N-MYC vector, respectively. To generate the pEFR::EFR-FLAG transgene, a 5.2-kb EFR genomic fragment (gEFR without its 3′-UTR sequence) was PCR amplified from the genomic DNAs of the wild-type Arabidopsis seedlings using the gEFR-FLAG primer set (Supplementary Table 1) and cloned into the pCambia1300-C-FLAG vector. The gEBS2 construct is a genomic transgene for EBS2 52 . All created transgenes were fully sequenced to ensure no PCR-introduced sequence error, and were individually transformed into the Agrobacterium tumefaciens strain GV3101 by electroporation and the resulting Agrobacterial strains were subsequently used to transform Arabidopsis plants using the floral-dipping method 65 .
CRISPR/Cas9-mediated genome editing to create pawh mutants. The CRISPR/ Cas9 vectors used to create pawh1/2-c mutation were provided by Dr. Jian-Kang Zhu 66 . The target sites for introducing mutations into PAWH1 and PAWH2 genes were selected using the web program CRISPR-PLANT (http://crispr.hzau.edu.cn/ cgi-bin/CRISPR/CRISPR), and the oligonucleotides for generating guide RNAs are listed in Supplementary Table 1. The CRISPR/Cas9 constructs were introduced into the Agrobacterium strain GV3101 by electroporation, which were subsequently used to transform bri1-5 mutant by the floral-dipping method 65 . T0 seeds were harvested, sterilized, and subsequently screened on ½ MS medium for T1 lines that carried the CRISPR/Cas9 constructs, which were subsequently screened for intended mutations of PAWH1/PAWH2 genes and verified in T2 and T3 generations. The verified pawh1-c bri1-5 and pawh2 bri1-5 mutants were crossed to get the pawh1-c pawh2-c bri1-5 triple mutant.
Expression of fusion proteins and generation of antibodies. The 1st cDNA preparation derived from total RNAs of wild-type Arabidopsis seedlings and the antigen-PAWH1 primer sets (Supplementary Table 1) were used to amplify a 480bp PAWH1 cDNA fragment corresponding to its N-terminal 160 AAs. The amplified fragment was cloned into pGEX-4T-3 (GE Healthcare) and pMALc-H 67 vectors, which were subsequently transformed into BL21-competent cells. The same cDNA preparation was also used to amplify a Hrd1a cDNA fragment encoding the N-terminal fragment of 73 AAs using the antigen-Hrd1a primer sets (Supplementary Table 1), and two copies of the amplified Hrd1a cDNA were cloned into the pGEX-4T-3 and pMALc-H vectors. The induction of GST and MBP fusion proteins and their subsequent purification using Glutathione Sepharose TM 4 Fast Flow beads (GE Healthcare) and Amylose Resin beads (New England Biolabs), respectively, were carried out following the manufacturers' recommended protocols. The purified MBP-PAWH1 and GST-Hrd1a fusion proteins were used to make custom antibody by Abmart (http://www.ab-mart.com.cn) while the purified GST-PAWH1 and MBP-Hrd1a fusion proteins were used to affinity-purify anti-PAWH and anti-Hrd1a antibodies from Abmart-generated anti-PAWH and anti-Hrd1a antisera, respectively, using an online protocol with nitrocellulose membrane (http://post.queensu.ca/~chinsang/lab-protocols/antibody-purification.html). The specificity of the purified anti-PAWH and anti-Hrd1a antibodies was analyzed by immunoblotting with total proteins extracted from 2-week-old seedlings of wild-type Arabidopsis plant and relevant mutants. Source data are provided as a Source Data file.
Yeast two-hybrid assay. The Clontech's yeast two-hybrid system was used to examine the interactions of PAWHs, EBS7, and Hrd1a in yeast cells (AH109). The 1st cDNA preparation described above was used to amplify cDNA fragments of the N-terminal 260 amino acids of PAWH1 (PAWH1-N260) or the N-terminal 257 amino acids of PAWH2 (PAWH2-N257) with the Y2H-PAWH1/2 primer sets (Supplementary Table 1). The PCR-amplified PAWH1/2 cDNA fragments were subsequently cloned into the vectors, pGADT7 (for expressing a fusion protein of the GAL4 activation domain with a target protein) and pGBKT7 (for producing a fusion protein of the GAL4 DNA-binding domain with a protein of interest). The pGADT7 and pGBKT7 constructs of EBS7-N142 and the catalytic domain of Hrd1a (Hrd1a-CD containing amino acids of 247-492) were described previously 39 . Yeast cells containing different combinations of pGADT7 and pGBKT7 plasmids were selected by growth on synthetic media lacking leucine and tryptophan. Several independent colonies were picked, resuspended in the liquid yeast growth medium, subsequently spotted via serial dilutions onto solid synthetic medium lacking leucine, tryptophan, and histidine for visual observation of their growth.
RNA isolation and analysis. Ten-day-old Arabidopsis seedlings grown on ½ MS medium supplemented with or without certain chemicals were harvested and ground in liquid nitrogen into a fine powder and their total RNAs were extracted using the RNeasy Plant Mini Kit (QIAGEN). One microgram of the purified total RNAs were treated with RNase-free DNase I (TIANGEN) and subsequently reverse transcribed into 1st strand cDNAs by the iScript TM cDNA synthesis Kit (Bio-Rad). The resulting cDNAs were used for classical PCR or quantitative real-time PCR (qPCR) analysis with gene-specific oligonucleotides listed in Supplementary  Table 1. The qPCR assays were performed on the CFX96 Real-Time System (Bio-Rad) with SYBR ® GREEN PCR Master Mix (Bio-Rad) following the manufacturer's instruction. Three biological replicates each with three technical repeats were conducted for each target mRNA and each sample. The ACTIN8 cDNA was used as an internal control. The classical RT-PCR assays were carried out on a C1000 Touch TM Thermal Cycler (Bio-Rad) with 2xHieff TM PCR Master Mix (Yeasen). The PCR amplified cDNA fragments were separated by agarose gel electrophoresis and visualized by Gel Doc TM XR+ Molecular Imager (Bio-Rad) with Image Lab TM software. The β-Tubulin was used as an internal reference. Source data are provided as a Source Data file.
Protein extraction and analyses. To isolate microsomal proteins, 1 g of 10-dayold Arabidopsis seedlings was harvested directly into liquid N 2 and immediately ground in liquid nitrogen into fine powder, dissolved in a homogenization buffer [50 mM Tris-HCl (pH 8.2), 20% (v/v) glycerol, 1 mM phenylmethylsulphonyl fluoride (PMSF, Sigma), 2 mM ethylenediaminetetraacetic acid (EDTA), 1 mM DTT, 2 protease inhibitor Cocktail Tablets (Roche) per 100 mL solution], and centrifuged for 10 min at 8000 × g at 4°C. The supernatant (total proteins) was centrifuged further at 100,000 × g at 4°C for 60 min to collect the microsomal pellet. The resulting pellet was dissolved in the homogenization buffer as the microsomal fraction. Both the total proteins and microsomal fraction were used to immunoprecipitate the GFP-tagged Hrd1a and MYC/HA-tagged EBS7 using anti-GFP mAb-agarose (D153-8, MBL) and anti-MYC/anti-HA-agarose affinity gels (Sigma) for mass spectrometric analysis of Hrd1/EBS7-associated proteins. The microsomal fraction of the wild-type seedlings was subsequently treated with solutions of 0.1 M NaCl, 0.1 M Na 2 CO 3 , 1% (v/v) Triton X-100 (Sigma) or 1% (v/v) Nonidet P-40 (Roche) at 4°C for 4 h. After treatment, the four fractions were centrifuged at 100,000 × g at 4°C for 60 min to collect the supernatants (soluble fraction) and the pellets, which were mixed with 2x SDS buffer, boiled for 10 min, separated by 12% SDS/PAGE, and analyzed by immunoblotting.
We used 10-day-old seedlings to perform the coIP assays 39 . In brief, 10-day-old Arabidopsis seedlings were harvested, ground into fine powder in liquid N 2 . Total proteins were extracted in the extraction buffer [50 mM Tris-HCl (pH 8.0), 150 mM NaCl, 5 mM EDTA, 0.1% (v/v) Triton X-100, 0.2% (v/v) Nonidet P-40 (Roche), 2 tablets of the protease inhibitor cocktail (Roche) per 100 mL] and the extracts were centrifuge at 14,000 × g for 15 min at 4°C. The resulting supernatants were incubated with anti-EBS7 antibody 39 followed by protein-A/G-agarose beads (Abmart) at 4°C for 8 h. The immunoprecipitates were washed five times with the extraction buffer, resuspended in 2X SDS sample buffer, boiled at 95°C for 10 min, and separated by SDS/PAGE. For other immunoblot assays, 10-day-old Arabidopsis seedlings treated with or without a chemical of interest were harvested into liquid N 2 and stored in −80°C freezer for later analysis or immediately ground into fine powder in liquid N 2 . The broken plant tissues were dissolved in 2x SDS sample buffer, boiled at 95°C for 10 min, and centrifuged for 10 min at the top speed in an Eppendorf 5424 centrifuge. The resulting supernatants were immediately used for immunoblotting analysis or incubated with or without 1000 U Endo-Hf in 1x G5 buffer (New England Biolabs) at 37°C for 1.5 h. The treated/ non-treated total proteins were separated by 7, 10, or 12% SDS-PAGE. For all the immunoblotting performed throughout this study, after transfer from SDS-PAGE gels to PVDF membranes (Bio-Rad), proteins were analyzed by immunoblot with antibodies against BES1 50 , BRI1 50 , BiP (at-95, Santa Cruz Biotechnology), maize-CRT 68 , PDI (Rose Biotechnology Inc), EBS5 36 , EBS6 37 , EBS7 39 , PAWHs (this study), Hrd1a (this study), GFP [living color® A.v.monoclonal antibody (JL-8), 632381, Takara], or HA (H9658, Sigma). The signals on those immunoblots were detected with horseradish peroxidase-conjugated secondary antibodies, ECL Select TM Western Blotting Detection Reagent (GE Healthcare), and Medical X-ray Processor (Kodak). The resulting signals on X-ray films were scanned and digitalized. Source data are provided as a Source Data file.
Sucrose density-gradient centrifugation. Sixteen grams of 10-day-old Arabidopsis seedlings were ground in liquid N 2 into a fine powder and immediately extracted by the homogenization buffer (see above) at 4°C. The protein extracts were first filtered through Miracloth (CalBiochem) to remove insoluble plant debris and subsequently centrifuged at 5000×g for 5 min at 4°C to remove cellular debris and organelles. The supernatant was centrifuged at 100,000×g for 45 min to pellet the microsomes, which was resuspended in 1 mL resuspension buffer [25 mM Tris-HCl (pH 7.5), 10% (w/v) sucrose, 1 mM PMSF, 2 mM EDTA, 1 mM DTT, 2 protease inhibitor Cocktail Tablets (Roche) per 100 mL]. The microsome resuspension was loaded onto the top of a 11-mL 20-50% (w/w) sucrose gradient in 10 mM Tris-HCl (pH 7.5), 2 mM EDTA, 1 mM DTT, 0.1 mM PMSF, and centrifuged at 100,000×g for 16 h. After centrifugation, 1 mL fractions (12 fractions) were hand collected, and 50 μL protein sample for each fraction was mixed with 2x SDS buffer, boiled for 10 min, separated by 10 and 12% SDS/PAGE, and analyzed by immunoblotting. For the Mg 2+ -plus experiments, 5 mM MgCl 2 was added to the buffers of homogenization, resuspension, and ultracentrifugation. Source data are provided as a Source Data file.
Confocal analysis of PAWH1/2 fusion proteins. To generate p35S::GFP-PAWH1 and p35S::GFP-PAWH2 transgenes for visualization of PAWH1/2 subcellular localization patterns, an 858-bp PAWH1 cDNA fragment and an 849-bp PAWH2 cDNA fragment were amplified from the 1st cDNA preparation of the wild-type Arabidopsis seedlings using the GFP-PAWH1 and GFP-PAWH2 primer sets (Supplementary Table 1) and subsequently cloned into the pCambia1300-p35S::N-GFP vector. The plasmid p35S::RFP-HDEL was the same as used in the EBS7 study 39 . To analyze the protein interaction using the bimolecular fluorescence complementation (BiFC) assay, an 858-bp PAWH1 cDNA or an 849-bp PAWH2 cDNA fragment amplified from the 1st strand cDNAs of wild-type Arabidopsis seedlings using the NER-PAWH1 or NER-PAWH2 primer set (see Supplementary Table 1) was cloned into the pVYNER vector 69 . The same 1st cDNA preparation was also used to amplify the full-length (876-bp) EBS7 cDNA using the CER-EBS7 primer set and the full-length (1479 bp) Hrd1a cDNA using the CER-Hrd1a primer set (Supplementary Table 1), which were subsequently cloned into the pVYCER and pVYCE vectors 69 . These plasmids and the corresponding empty vectors, after verifying no PCR-introduced error, were individually transformed into the Agrobacterium tumefaciens strain GV3101. The p35S::RFP-HDEL-carrying GV3101 cells were mixed with the p35S::GFP-PAWH1 or p35S::GFP-PAWH2-transformed GV3101 strain and co-transformed into leaves of 3-week-old tobacco (Nicotiana benthamiana) plants by the agro-infiltration method 70 . Similarly, a mixture of two Agrobacterium strains, one carrying a pVYNER plasmid and the other containing a pVYCER or pVYCE plasmid, was used to infiltrate young leaves of 3-week-old tobacco. Two days after infiltration, the transformed tobacco leaves were examined by confocal microscopy on a Leica SP8 (with LAS AF software, Leica Microsystems) for the subcellular localization patters of GFP-tagged PAWH1/2 and the ER-localized RFP-HDEL or the reconstituted YFP signal. GFP, YFP, and RFP were excited by using 488-, 514-, and 542-nm laser light, respectively.
Liquid chromatography-tandem mass spectrometry. Immunoprecipitated proteins were eluted from antibody-conjugated beads with 0.2 M Glycine (pH 2.5), dried, and solubilized in 8 M urea containing 50 mM iodoacetamide. The alkylated proteins were digested by sequencing-grade trypsin (Promega) in 25 mM NH 4 HCO 3 at an enzyme:substrate ratio of 1:100 (w/w) in a 37°C shaking incubator for 18 h. The tryptic peptides were collected after centrifugation at 13,523×g and freeze-dried with a refrigerated CentriVap concentrator (Labconco). Protein samples were reconstituted in 0.1% formic acid (FA) and analyzed by online nanoAcquity ultraperformance LC (Waters) coupled with an Orbitrap Fusion Tribrid mass spectrometer (Thermo Fisher Scientific). Nanospray was controlled by a PicoView Nanospray Source (PV550; New Objective) at a spray voltage of 1.9 kV. The peptides were trapped by a 2G-V/MT Trap symmetry C18 column (5 μm particles, 180 μm inner diameter × 20 mm length) at a flow rate of 3 μL min −1 for 3 min, and separated on a BEH130 C18 analytical column (1.7 μm particles, 100 μm inner diameter × 250 mm length) at a flow rate of 250 nL min −1 . Peptides were eluted using mobile phases consisting of solvent A (0.1% FA in water) and solvent B (0.1% FA in acetonitrile) through a linear gradient, and then 85% of solvent B at a duration of 90 min. Data-dependent MS/MS acquisition was performed following a full MS survey scan by Orbitrap at a resolution of 70,000 over the m/z range of 350-1800, and MS/MS measurements of the top 20 most intense precursor ions. The MS raw data from the LC-MS/MS analyses were separately converted to mascot generic format (MGF) files using the Proteome Discoverer 1.4 software (Thermo Fisher Scientific) and then searched against in-house databases using Mascot Daemon 2.4 (MatrixSciences). The search parameter for tryptic digestion was restricted to a maximum of two missed cleavages of proteins. Cysteine carbamidomethylation was designated as a fixed modification. Mass tolerances were set up to 20 ppm for the Orbitrap-MS ions, and 0.2 or 1 Da for ion-trap MS/MS fragment ions.
Reporting summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.