Competition between Der1 and ERAD-M substrates controls Hrd1 complex function

Significance

This study uncovers how the highly conserved Hrd1 ubiquitin ligase complex selectively degrades misfolded integral membrane proteins during endoplasmic reticulum-associated degradation (ERAD). Using deep mutational scanning, we identified key Hrd1 residues that altered its ability to degrade all integral membrane substrates that were tested. The identified residues are located at an interface with Der1, a Derlin protein important for degradation of lumenal ERAD substrates, revealing dynamic interactions required for different ERAD functions. This insight highlights the importance of protein complex interactions in ERAD function and demonstrates that the Hrd1 complex modulates its composition to handle different types of protein targets.

Abstract

Endoplasmic reticulum–associated degradation (ERAD) is a quality control process which removes misfolded proteins from the ER. The central component of the most conserved ERAD system is an integral membrane ubiquitin ligase called Hrd1. The Hrd1 ligase functions within a complex to mediate the recognition and ubiquitination of both soluble, lumenal substrates and integral membrane substrates, all of which are ultimately targeted for degradation by the cytosolic proteasome. Here, we used deep mutational scanning to identify Hrd1 residues exclusively involved in the degradation of integral membrane substrates. We report single residue Hrd1 variants that are broadly deficient in the degradation of all integral membrane substrates tested. Using in vivo assays to characterize Hrd1 variant deficiency, we explain how integral membrane substrates compete with other complex components to control Hrd1 function. This work reveals competition for the retrotranslocon cavity between both lumenal and membrane substrate degradation paths and highlights Hrd1 complex assembly as the primary determinant for tuning ERAD function.

In eukaryotes, approximately a quarter of the genome encodes for integral membrane proteins, most of which are integrated into the membrane at the endoplasmic reticulum (ER) (1, 2). Membrane proteins exhibit diverse topologies and require a complex translocon system for membrane integration and numerous intramembrane chaperones to assemble properly (3, 4). Even with these specialized systems, many membrane proteins fail to properly fold and can contribute to proteotoxicity if not promptly degraded. At the ER, misfolded membrane proteins are targeted for proteasomal degradation through the conserved quality control system known as ER-associated degradation (ERAD) [recent review here (5)].

Work in Saccharomyces cerevisiae has established the existence of at least four general ERAD pathways, each characterized by the location of the substrate’s misfolded lesion, whether in the ER lumen, within the ER membrane, on the cytosolic side of the ER membrane, or at the inner nuclear membrane (pathways defined as ERAD-L, -M, -C, or -INM, respectively). These pathways employ three distinct integral membrane ubiquitin ligase complexes, which function to recognize and polyubiquitinate their substrates on the cytosolic side of the ER membrane. The Hrd1 ligase degrades ERAD-L and ERAD-M substrates; ERAD-C substrates require Doa10; and ERAD-INM substrates utilize the Asi complex in yeast (6–11). Regardless of the ligase required for recognition, all ERAD substrates are eventually degraded by the cytosolic proteasome.

Among the ERAD pathways, the Hrd1 complex stands out because of its ability to degrade topologically distinct ERAD-L and ERAD-M substrates. However, based on the cellular machinery required, the mechanisms for their degradation are unique. For lumenal substrate degradation, Hrd1 functions in a complex with three additional membrane proteins, Hrd3, Der1, and Usa1, and a lumenal protein, Yos9 (11, 12). In most cases, all components are required for degrading lumenal proteins. Biochemical, genetic, and structural data suggest that the transmembrane regions of Hrd1 and Der1 together complete the “retrotranslocon,” a channel allowing for the movement of substrate through the membrane (13–16). Usa1 scaffolds Hrd1 and Der1 interactions and mediates higher-order oligomerization of the complex (17). Hrd1 interacts with Hrd3, a single-pass membrane protein that helps control Hrd1 ubiquitination activity, recruits substrates, and bridges an interaction with the substrate-recruiting protein, Yos9 (9, 18–21). Together, the Hrd1 complex functions by recruiting soluble, lumenal substrates, retrotranslocating substrate (movement from the ER lumen to the cytosol), and ubiquitinating substrate. After substrate is accessible and ubiquitinated on the cytosolic side of the membrane, all ERAD pathways converge to use the AAA-ATPase complex Cdc48, Ufd1, and Npl4 to extract (or “dislocate”) the polyubiquitinated substrate into the cytosol for subsequent proteasomal degradation (22–25).

In contrast, the mechanisms for degradation of ERAD-M substrates are not well understood, although clearly disparate. Based on topology, ERAD-M substrates already have cytosolically exposed regions, so retrotranslocation is not required; although, this may vary with the presence of additional structural features, such as large lumenal domains, in certain substrates (26). Instead, the presumed degradation steps are only recognition, ubiquitination, extraction from the membrane, and proteasomal degradation. In this case, only Hrd1 and Hrd3 are required for substrate degradation, except with specific glycosylated ERAD-M substrates that show slight dependence on Yos9 and Usa1 (17, 20). Der1 appears to be the only complex component not required for ERAD-M substrate degradation, possibly because retrotranslocation is not required (11, 27, 28). How ERAD-M substrates are recognized for degradation is unclear because only a few Hrd1 mutations are known that affect individual ERAD-M substrates, rather than all ERAD-M substrates (29). It is plausible that substrates are directly recognized by Hrd1 within the ER membrane (29) or that they are indirectly recruited and presented to Hrd1 for ubiquitination by an unknown, additional factor(s) (30). In addition, ERAD-M substrates pose challenges during their removal from the membrane, as their hydrophobic transmembrane segments must be extracted from the lipid bilayer environment. The Cdc48 ATPase complex provides force to extract proteins, but other factors have also been suggested to help overcome the significant energetic barrier. Both Hrd1 and the rhomboid pseudoprotease, Dfm1, have been genetically and biochemically implicated in facilitating ERAD-M extraction (31). Currently, the specific mechanisms of extraction are unknown and additional pathways of extraction may exist (32).

Here, to identify residues critical for the degradation of integral membrane substrates we employed deep mutational scanning (DMS) of Hrd1. We found mutations of specific, hydrophobic residues that broadly reduced ERAD-M function across all tested substrates, but not ERAD-L substrates. Surprisingly, these residues were spatially clustered around the lumenal side of Hrd1 transmembrane (TM) segments 3 and 8 near the retrotranslocon cavity and posited site of Der1 interaction (14, 33, 34). Using flow cytometry, coimmunoprecipitation, and site-specific crosslinking assays, we demonstrated that Hrd1 TM3 and TM8 variants formed altered associations with other Hrd1s and Der1, thereby tuning Hrd1 complex formation and specificity. With specific Hrd1 variants, removal of Der1 rescued ERAD-M function by increasing substrate interaction, indicating competition between Der1 and ERAD-M substrate for access to Hrd1. In alignment with this model, we observed that overexpression of Der1 slowed ERAD-M and biased the Hrd1 complex toward ERAD-L. Altogether, we propose that integral membrane substrates directly interact with the Hrd1 retrotranslocon and Der1 acts as a competitive inhibitor with ERAD-M substrates. For ERAD-M function, Der1 must be displaced from Hrd1 to allow ERAD-M substrate access for recognition and ubiquitination. This study provides insight into the mechanics of Hrd1 function in ERAD-M degradation and suggests that Hrd1 complexes must be dynamic in their composition to accomplish the degradation of both lumenal and integral membrane substrates.

Results

DMS Of the Hrd1 Ubiquitin Ligase.

We previously developed a DMS platform to determine the role of individual amino acid substitutions on the function of the ubiquitin ligase Hrd1 (35). The platform relied on fluorescent ERAD reporter substrates: an integral membrane substrate (ERAD-M: HMG-CoA Reductase, Hmg2-RFP) and a lumenal substrate (ERAD-L: Carboxypeptidase Y*, GFP-CPY*). To understand how the Hrd1 system degrades integral membrane protein substrates, we transformed Hrd1 variant libraries into cells expressing the reporters and used fluorescence-activated cell sorting (FACS) to collect cells expressing Hrd1 variants specifically unable to degrade the integral membrane substrate Hmg2 (SI Appendix, Fig. S1A).

Using FACS, we isolated populations of Hrd1 variants defective in degrading ERAD-M substrates from the libraries where we mutagenized amino acids Met1-Phe440, but not the library that mutagenized amino acids Ala441-Ile551 (SI Appendix, Fig. S1B). This was consistent with the previous study that demonstrated the Hrd1 C-terminal region (Gly408-Ile551) was not involved in degradation of ERAD-M substrates (35). We collected cells from the “M-defective” population, cultured these cells, and validated the ERAD-M defective phenotype (SI Appendix, Fig. S1B). To understand the Hrd1 variants incapable of degrading the ERAD-M substrate, we extracted DNA from the sorted populations and used Illumina sequencing to identify the Hrd1 mutations. We focused on only single amino acid substitutions from the sequencing data to eliminate complicated phenotypes of multiple mutations. We visualized our results by plotting the enrichment or depletion of individual amino acid substitutions relative to the input library in a heatmap showing the amino acid position (x-axis) versus substitution (y-axis) (SI Appendix, Fig. S1C). The false discovery rate was log transformed from 0.1 to 100% to linearly adjust the transparency from 0 to 90% and visually eliminate false positives from the screen (Fig. 1A).

Fig. 1.

DMS of the Hrd1 ubiquitin ligase for membrane substrate defective mutants. (A, Top) Topology diagram of Hrd1 with transmembrane segments shown as TM1-8. Colors indicate the cytosolic (blue), lumenal (magenta), and cytosolic RING domain (green). (Bottom) DMS results of cells sorted into the ERAD-M defective bins were displayed as a heatmap showing single codon changes that were enriched (red) or depleted (blue) compared to the input library. Color transparency was adjusted based on FDR. FDR below 0.1% were set to 0% transparent and FDR values between 0.1 to 100% were used to adjust transparency from 0% (opaque) to 90% transparent. Individual amino acids are on the y-axis, and the Hrd1 amino acid position is on the x-axis. Dark gray boxes indicate the wild-type amino acid and light gray boxes indicate a lack of sequencing coverage. Note for Hrd1 residues 441 to 551, no ERAD-M defective cells were isolated but are shown as fully depleted for clarity. (B) Flow cytometry analysis of hrd1Δ cells expressing the indicated ERAD substrates and Hrd1(WT), inactive Hrd1(C399S), or Hrd1 variants. Cells were treated for 4 h with 0.1% ethanol (vehicle, solid black line with no fill) or chased for 4 h with cycloheximide or zaragozic acid (Hmg2 only) (solid black line with gray fill). The dashed line highlights the position of Hrd1(C399S) after a 4 h chase. The results are representative of three independent replicates. Cells were analyzed by flow cytometry in separate experiments and a representative set of controls (Hrd1 WT and C399S) are displayed for all experiments. (C) Quantification of (B). Hrd1(WT) is set to 1 (full function, black) and inactive Hrd1(C399S) is set to 0 (no function, white). Values outside of the range were set to 0 or 1 for clarity. For full range of data, see Dataset S2. (D) Structural prediction of Hrd1 (AlphaFold: AF-Q08109-F1-v4) highlighting localization of ERAD-M deficient variants identified in this study. Hrd1 backbone is shown in blue for amino acids 1 to 330. The dashed blue line represents a disordered cytosolic loop between transmembrane segments 6 and 7 that is predicted with low confidence. Residues shown as red spheres indicate ERAD-M defective mutations from (B and C). Arrows indicate the Hrd1 aqueous retrotranslocon cavity and lateral entry gate formed by TMs 3 and 8.

This analysis illuminated mutations that inhibited ERAD-M substrate degradation throughout the transmembrane region of Hrd1, consistent with previous studies (29). The most substantial enrichment of mutations in this class were localized on transmembrane segments 3, 4, 6, and 8, with substitutions to nearly all classes of amino acids (charged, polar, aromatic, and hydrophobic). This deep mutational scan of the Hrd1 ligase allowed identification of hotspots in the transmembrane region of Hrd1 that were important for ERAD-M degradation.

Identification of Hrd1 Mutants Unable to Degrade ERAD-M Substrates.

We focused on validating the mutations within the transmembrane region of Hrd1 that were enriched at least 30 times compared to the input library and had less than a 1% false discovery rate. To confirm these hits (49 substitutions in total) maintained the wildtype-like ability to degrade lumenal substrates but had impaired Hmg2 degradation, we integrated individual fluorescent ERAD substrates (Hmg2, CPY*, or Pep4*) in hrd1Δ cells and complemented the cells with either wild-type Hrd1 (WT), catalytically inactive Hrd1 (C399S), or the individual Hrd1 mutants. We used flow cytometry to determine the Hrd1 variant’s ability to degrade lumenal and integral membrane substrates under two conditions (SI Appendix, Fig. S2 A–E). First, we used a saturated chase which was equivalent to that used for the FACS experiments. We found that the majority of Hrd1 variants pulled from the DMS approach were defective in degradation of Hmg2 under these conditions (SI Appendix, Fig. S2 A and E). Next, we used a mid-log chase assay, where substrate degradation could be observed after addition of compounds to either trigger degradation, or stop new translation [SI Appendix, Fig. S2 B and E with zaragozic acid for Hmg2 (36) or cycloheximide for CPY* and Pep4*].The addition of zaragozic acid accumulates geranylgeranyl pyrophosphate, the endogenous signal for driving robust Hmg2 degradation in cells (36). Thus, the mid-log assay more accurately replicated physiological Hmg2 degradation. When we evaluated our mutants using the mid-log chase assay, we confirmed 12 Hrd1 variants that showed an inability to effectively degrade Hmg2 but retained ability to degrade ERAD-L substrates (Q9L, I91R, F95D, F95N, F95Q, F95K, S97I, L114P, M298D, L299D, K301D, and D302K) (SI Appendix, Fig. S2 D and E).

As Hrd1 mutations previously reported to disrupt ERAD-M degradation were specific to individual substrates (29), we wondered whether the substitutions we isolated as Hmg2 defective were specific to Hmg2 or were more broadly impaired in the degradation of multiple ERAD-M substrates. Therefore, we tested the Hrd1 variants against three additional membrane substrates [6myc-Hmg2, Hmg2-NR1, and Pdr5* (9, 37, 38)] and two ERAD-L substrates (CPY* and Pep4*). These mutations showed no defect in degradation of either lumenal ERAD substrate tested. In fact, several mutations (I91R, F95Q, and F95K) appeared to have enhanced lumenal substrate degradation capability (SI Appendix, Fig. S2F). However, we found six Hrd1 variants (I91R, F95D, F95K, M298D, L299D, and D302K) that were able to degrade lumenal substrates but were deficient in degrading all ERAD-M substrates tested (Fig. 1 B and C). Strikingly, these residues clustered specifically to the lumenal side of transmembrane segments 3 and 8 (Fig. 1D). Based on available structural models, these two helices are positioned at the proposed lateral gate of the Hrd1 retrotranslocon [Fig. 1D, (13, 14)].

Importantly, we ruled out Hrd1 protein level as the primary factor in Hmg2 degradation by measuring the steady-state protein levels of each Hrd1 substitution using immunoblotting (SI Appendix, Fig. S2 C, G, and H). While specific substitutions destabilized Hrd1, there was no correlation between mutant Hrd1 protein level and Hmg2 degradation (Pearson correlation = 0.05, SI Appendix, Fig. S2H).

Der1 Deletion Rescues ERAD-M Function of Hrd1 TM3 and TM8 Mutants.

The ERAD-M deficient Hrd1 variants isolated in this study, in combination with existing structural models (13, 14) and crosslinking data (14, 33, 34), support that Der1 and integral membrane substrates require access to the same interface of Hrd1 formed by transmembrane segments 3 and 8 (Fig. 2 A and B and SI Appendix, Fig. S3A). Importantly, this means that Der1 and ERAD-M substrates would compete for Hrd1 interaction and require the displacement of Der1 from Hrd1 for ERAD-M and vice-versa for ERAD-L. We hypothesized that the ERAD-M deficient Hrd1 variants may be unable to degrade ERAD-M substrates because of increased association with Der1. If this idea was correct, we expected to see improved degradation of ERAD-M substrates in der1Δ cells.

Fig. 2.

Hrd1 variants can function in ERAD-M in the absence of Der1. (A) Space-filling structural prediction of Hrd1 (AlphaFold: AF-Q08109-F1-v4) highlighting overlap between ERAD-M deficient variants and the Der1 interaction site. Hrd1 amino acids 1 to 330 are displayed in blue. Residues shown in light gray represent photo-crosslinking data between Hrd1 and Der1 (14, 34). Residues shown in red indicate ERAD-M deficient variants identified in this study. Residues colored pink represent overlap between ERAD-M deficient variants at positions that also photo-crosslink to Der1. Arrows indicate positions of variants found to be rescued by Der1 deletion (Ile91, Phe95, and Asp302) in (D) and localization of the aqueous retrotranslocon cavity. (B) Space-filling structural model of the Hrd1–Der1 complex [PDB: 6VJZ (14)]. Hrd1 is shown in blue, Der1 is shown in green. Color scheme matches (A), with the addition of residues in dark gray, which represent Der1 sites that photo-crosslink to Hrd1 (34). (C) Zaragozic acid chase following Hmg2-RFP degradation by flow cytometry in actively growing hrd1Δ cells (Strain: hrd1Δ, black lines) and hrd1Δder1Δ cells (Strain: hrd1Δder1Δ, magenta lines) expressing wild-type Hrd1 (WT) or catalytically inactive Hrd1 (C399S). Results are displayed as individual values from at least three replicates. (D) As in (C) but with cells containing the indicated Hrd1 variants. (E) Immunoblot for wild-type Hrd1 and Der1 stability during cycloheximide chase in hrd1Δder1Δ cells with Hrd1-3xFLAG and Der1-HA or empty vector (der1Δ). Stain free shows total protein loaded per lane. These immunoblots are representative of three independent replicates. (F) As in (E), but with cells containing the indicated Hrd1 variants. (G) As in (E), but with cells expressing an ERAD-M substrate, Hmg2-3xV5, and subjected to zaragozic acid and cycloheximide chase. These immunoblots are representative of two independent replicates. (H) As in (G), but cells were complemented with indicated Hrd1 variants.

To test this, we compared the degradation rate of the ERAD-M substrate Hmg2 in hrd1Δ and hrd1Δder1Δ cells that were expressing either wild-type Hrd1, Hrd1 C399S, or the Hrd1 variants found to be defective for degradation of multiple ERAD-M substrates (I91R, F95D, F95K, M298D, L299D, D302K) (Fig. 2 C and D, black lines for hrd1Δ cells, magenta lines for hrd1Δder1Δ cells). In der1Δ cells expressing the Hrd1 variants I91R, F95K, and D302K, we observed ERAD-M function was partially restored, consistent with the ERAD-M dysfunction caused by increased interactions with Der1 (Fig. 2D). This effect was still observed when geranylgeranyl pyrophosphate was added directly to cells (SI Appendix, Fig. S4A). Similarly, when we followed degradation of the ERAD-M substrate Pdr5*, we observed similar trends as with Hmg2 degradation (SI Appendix, Fig. S4B). This rescue was specific to ERAD-M function, as the ERAD-L substrate Pep4* was stabilized in der1Δ cells, as expected based on the important role for Der1 in ERAD-L (SI Appendix, Fig. S4C, black lines for hrd1Δ cells, magenta lines for hrd1Δder1Δ cells). Noticeably, the three Hrd1 variants that exhibited rescued function in the absence of Der1 (I91R, F95K, and D302K) all had substitutions to positively charged amino acids. Conversely, Hrd1 variants F95D (TM3), M298D and L299D (TM8), which had substitutions of negative charges, were not rescued in der1Δ cells (Fig. 2D). As these residues map to the lumenal side of the Hrd1 transmembrane region near the site of Der1 interaction and substrate engagement, introduction of charge at this position may lead to altered association of Der1 or substrate with the complex.

We selected the Hrd1 TM3 variants F95D and F95K to further confirm these results by immunoblotting, as they exhibited opposite phenotypes in response to Der1 deletion (F95D: der1Δ unresponsive, F95K: der1Δ rescued). Hrd1 WT-3xFLAG, F95D-3xFLAG, and F95K-3xFLAG were expressed in hrd1Δder1Δ cells with an integration of Der1 with a single HA epitope (Der1-HA) or an empty vector control (EV) (39). We subjected these cells to cycloheximide chase and followed the stability of Hrd1 WT, F95D, F95K, and Der1-HA by immunoblotting (Fig. 2 E and F). Although previously unreported, we found that Hrd1 WT was modestly destabilized in der1Δ cells (Fig. 2E) and the F95D variant was increasingly unstable compared to Hrd1 WT. Conversely, F95K was stable regardless of Der1 expression. Instead, Der1 was unstable in F95K cells, suggesting that Der1 acts as a constitutive substrate for this variant and would be in direct competition with other ERAD-M substrates, which are inefficiently degraded (Fig. 2F).

Next, we expressed Hmg2-3xV5 in these cells and performed a zaragozic acid chase to follow ERAD-M function (Fig. 2 G and H). In alignment with our flow cytometry data, Der1 deletion rescued ERAD-M function of Hrd1 F95K but not F95D (Fig. 2H). Together, these data support a competition between Der1 and ERAD-M substrates for Hrd1 where removal of Der1 from the system restores ERAD-M function.

Der1 and ERAD-M Substrates Compete for Hrd1 Interaction.

Based on the hypothesis that ERAD-M deficient Hrd1 variants were unable to degrade ERAD-M substrates from increased association with Der1, we selected Hrd1 F95D (der1Δ unresponsive) and Hrd1 F95K (der1Δ rescued) and performed coimmunoprecipitation assays to evaluate their association with wild-type Hrd1 and Der1. Hrd1 WT-3xFLAG, F95D-3xFLAG, and F95K-3xFLAG were integrated into hrd1Δder1Δ cells under the Hrd1 endogenous promoter. To test Hrd1 self-association, we expressed a second copy of wild-type Hrd1 with a 3xHA epitope (Hrd1(WT)-3xHA), and to observe Hrd1–Der1 interactions we used Der1 with a single HA epitope (Der1-HA). Strains expressing Hrd1(WT)-3xHA represent an approximate twofold overexpression of Hrd1. We immunoprecipitated Hrd1-3xFLAG and confirmed coimmunoprecipitation of wild-type Hrd1-3xHA and Der1-HA (Fig. 3A, quantification in Fig. 3 B and C). Wild-type Hrd1-3xFLAG interacted with Hrd1-3xHA similarly in the presence or absence of Der1. The Hrd1-3xFLAG interaction with Der1 was also largely unaffected by the second copy of Hrd1 (Hrd1-3xHA, Fig. 3A) reflective of the oligomeric Hrd1 complexes that are proposed to exist in vivo (17, 40).

Fig. 3.

Hrd1 TM3 variants have altered complex and substrate association. (A) Coimmunoprecipitation of Hrd1 complexes. Hrd1(WT)-3xFLAG, and the indicated variants, were integrated under the endogenous Hrd1 promoter in hrd1Δder1Δ cells with wild-type Hrd1(WT)-3xHA and/or Der1-HA. Cells were lysed and Hrd1-3xFlag was immunoprecipitated with anti-FLAG antibodies and probed for interacting proteins. Input represents 1.5% of the cleared lysate. Additional controls for specificity are shown in (E). These immunoblots are representative of three independent replicates. (B) Quantification of coimmunoprecipitated Hrd1 complex components from (A). The amount of Hrd1-3xFLAG eluted in each sample was normalized to the amount eluted in lane 4, which was set to 1. The amount of Hrd1(WT)-3xHA in each elution was normalized to the amount of immunoprecipitated Hrd1-3xFLAG in that sample, such that Hrd1(WT)-3xHA eluted in lane 4 were also set to 1, for comparison of relative amounts bound across all samples. The bar graph represents the mean and error bars represent the SD. (C) As in (B), but for Der1-HA. (D) Self-association of Hrd1 and Hrd1 variants. Same as in (A) but instead of wild-type Hrd1-3xHA, cells contained a second copy of the indicated Hrd1 variant used as the bait for FLAG pulldown, notated as Hrd1-3xHA. Hrd1 homo-oligomers were immunoprecipitated with anti-FLAG antibodies. Input represents 1.5% of the cleared lysate. These immunoblots are representative of four independent replicates. (E) Quantification of self-associated Hrd1 and Hrd1 variants from (D). The amount of Hrd1-3xFLAG eluted in each sample was normalized to the amount eluted in lane 1, which was set to 1. The amount of Hrd1-3xHA and Der1-HA in each elution was then normalized to the amount of immunoprecipitated Hrd1-3xFLAG in that sample, for comparison of relative amounts bound across all samples. The bar graph represents the mean and error bars represent the SD. (F) As in (E), but for Der1-HA. (G) Coimmunoprecipitation of an ERAD-M substrate with Hrd1. hrd1Δder1Δ cells expressing either Hrd1(WT)-3xFLAG or untagged Hrd1 were transformed with wild-type Hrd1(WT)-3xHA, Der1-HA, and 6myc-Hmg2-3xV5 (a constitutive ERAD-M substrate). 6myc-Hmg2-3xV5 was coimmunoprecipitated with Hrd1(WT)-3xFLAG and was specific to FLAG pulldown. Endogenous Ubx3 did not coimmunoprecipitate with Hrd1, as expected, providing an additional control for pulldown specificity (41). (H) Coimmunoprecipitation of an ERAD-M substrate with Hrd1 complexes. The yeast strains in (A) were transformed with 6myc-Hmg2-3xV5. Hrd1–substrate complexes were immunoprecipitated with anti-FLAG antibodies. Input represents 1.5% of the cleared lysate. Additional controls for specificity are shown in (G). These immunoblots are representative of three independent replicates. (I) Quantification of coimmunoprecipitated Hrd1 complex components and substrate from (H). All samples were normalized to the amount of coimmunoprecipitated proteins in elution lane 4, which were set to 1, as described in (B). Results are displayed as the mean and error bars represent the SD. (J) As in (I), but for Der1-HA. (K) As in (I), but for the ERAD-M substrate 6myc-Hmg2.

By comparison, the Hrd1 TM3 mutants displayed perturbed association with the Hrd1 complex. Hrd1 F95D associated with wild-type Hrd1-3xHA whether Der1 was present or not (Fig. 3 A–C). In the absence of wild-type Hrd1, Hrd1 F95D interacted more strongly with Der1, but this interaction was disrupted when wild-type Hrd1 was expressed (Fig. 3 A–C). We interpreted this result to mean F95D has a preference to form homo-oligomers over interactions with Der1. This would explain why in the presence of Hrd1-3xHA, Hrd1 F95D interaction with Der1 was reduced, presumably through competition by additional wild-type Hrd1.

Conversely, F95K Hrd1 exhibited diminished ability to interact with both wild-type Hrd1 and Der1 under all conditions (Fig. 3 A–C). This is congruent with Der1 acting as a substrate of F95K Hrd1 and having lower steady state levels available for coimmunoprecipitation within these cells (Fig. 2F).

To confirm the homo-oligomeric states of Hrd1 F95D and F95K, we repeated coimmunoprecipitation in strains where the 3xHA-tagged wild-type Hrd1 was replaced with a second copy of the same Hrd1 variant used as bait (Fig. 3 D–F, notated as Hrd1-3xHA). These strains represent a twofold overexpression of either wild-type Hrd1 or the respective mutants. Hrd1 F95D strongly coimmunoprecipitated the HA-tagged version of itself (Fig. 3 D–F), supporting an increased propensity to self-associate. Meanwhile, Hrd1 F95K only weakly interacted with itself (Fig. 3 D–F).

Next, we expressed the constitutive ERAD-M substrate 6myc-Hmg2-3xV5 in these cells to probe for Hrd1–substrate interaction and Hrd1 complex composition in the presence of an ERAD-M substrate. Hrd1 coimmunoprecipitated 6myc-Hmg2, demonstrating that Hrd1 interacts with substrate, although this could be indirect (Fig. 3G). Remarkably, when cells expressing 6myc-Hmg2 also expressed Der1, the amount of substrate coimmunoprecipitated with wild-type Hrd1 was reduced by nearly 50% (Fig. 3 H–K). This result provides supporting evidence that Der1 competes with ERAD-M substrates for access to wild-type Hrd1.

Hrd1 F95D was able to interact with substrate and this interaction was not affected by expression of Der1 or wild-type Hrd1 (Fig. 3 H–K). The associations with Hrd1 and Der1 were also unaffected by expression of an ERAD-M substrate (compare Fig. 3 A and B with Fig. 3 H–K). These results were in alignment with our previous data for the F95D mutant, which was unresponsive to Der1 deletion (Fig. 2 D and H).

Hrd1 F95K exhibited a loss of substrate association in cells expressing Der1 compared to der1Δ cells (Fig. 3 H–K). Unlike wild-type Hrd1 and Hrd1 F95D, Der1 was not displaced by Hrd1 or substrate from Hrd1 F95K (Fig. 3J). In addition, substrate was displaced by Der1 directly. We propose that rather than self-associating to form larger order oligomers, F95K preferentially interacts with Der1, thus blocking substrate engagement and ERAD-M function. This aligns with our genetic data wherein Der1 deletion was able to rescue ERAD-M function of this mutant but Der1 was still required for ERAD-L activity (Fig. 2 D and H).

Finally, to investigate direct interactions between Hrd1 and a membrane substrate, we performed site-specific in vivo photo-crosslinking in hrd1Δder1Δ cells expressing the ERAD-M substrate 6myc-Hmg2-mScarlet-I (42, 43). Benzoyl-phenylalanine photoreactive probes were incorporated into FLAG-tagged Hrd1 at sites corresponding to ERAD-M deficient variants (I91, F95, M298, L299, and D302) using amber codon suppression. Following UV-irradiation, photo-crosslinked complexes were analyzed by immunoprecipitation of Hrd1-3xFLAG and immunoblotting. Consistent with prior studies, residues in TM3 showed the most prominent Hrd1 crosslinks (SI Appendix, Fig. S5A) (14, 40). We observed what we interpreted as intramolecular crosslinks at F95 and M298, in alignment with structural models positioning these residues ~5 to 6 Å apart across the narrowest region of the lateral gate (arrowheads in Bottom panel of SI Appendix, Fig. S5B) (14). Despite their close spatial proximity, each Hrd1 variant exhibited a distinct crosslinking pattern, supporting the site-specific nature of Hrd1 interactions and the potential role of these residues in tuning Hrd1 stoichiometry. Importantly, each of the tested Hrd1 positions appear to directly contact 6myc-Hmg2, as evidenced by the UV-dependent loss of 6myc-Hmg2 (SI Appendix, Fig. S5A). Collectively, these results provide supporting evidence that Hrd1 contacts ERAD-M substrates at the same interface as Der1.

Overexpression of Der1 Improves ERAD-L and Slows ERAD-M.

Altogether, our data support a model where the Hrd1 retrotranslocon directly engages integral membrane substrates during ERAD-M. This hinges on the displacement of Der1, which otherwise acts as a competitive inhibitor for substrate access to Hrd1. We reasoned that if Hrd1 TM3 variants are rescued by der1∆, then overexpression of Der1 should result in bias toward Hrd1–Der1 complexes, thus slowing or inhibiting ERAD-M function even further.

To control overexpression of Der1, we integrated Der1-HA under the copper-inducible CUP1 promoter (Der1_CUP) in hrd1Δder1Δ cells (Fig. 4A). Der1_CUP cells were also transformed with Hrd1 WT, C399S (catalytically inactive) or Hrd1 TM3 mutants, F95D (der1Δ unresponsive) and F95K (der1Δ rescued). To monitor ERAD function, we coexpressed GFP-CPY* and Hmg2-RFP, induced Der1_CUP expression, and followed substrate degradation during cycloheximide chase by flow cytometry. We noted that with coexpression of GFP-CPY* the steady state levels of the ERAD-M substrates increased, which already suggested competition between functions (SI Appendix, Fig. S6A). Der1 overexpression slightly enhanced ERAD-L degradation in cells with Hrd1 F95K, with the largest effects seen 1 to 2 h after cycloheximide addition (Fig. 4 B, Bottom Left). Conversely, ERAD-L remained unchanged in Der1_CUP cells containing Hrd1 F95D, a mutant which was previously unaffected by der1Δ (Fig. 4 B, Left Middle), supporting the idea that these variants tune Hrd1 complex specificity primarily through Der1 association.

Fig. 4.

Der1 competes with ERAD-M substrates for Hrd1. (A) Expression of Der1-HA was determined by immunoblotting (Left). hrd1Δder1Δ cells expressing Der1-HA from the copper-inducible CUP1 promoter or the endogenous Der1 promoter were treated with copper sulfate at the indicated concentrations overnight (approximately 16 h), subcultured into media containing a consistent concentration of copper sulfate and grown to log-phase (approximately 4 h). Total protein was visualized by stain-free technology as a loading control. Quantification of Der1-HA expression (Right) was normalized against expression from the endogenous promoter, which was set to 1. (B) Effect of Der1 overexpression on ERAD-L and ERAD-M function. hrd1Δder1Δ cells expressing Der1 from its endogenous promoter (Der1 WT, black lines) or the inducible CUP1 promoter for overexpression (Der1 OE, magenta lines), along with wild-type Hrd1 or the indicated variants were treated with 500 µM copper sulfate as in (A). GFP-CPY* and Hmg2-RFP were coexpressed and their degradation was followed simultaneously by flow cytometry during a mid-log chase. The starting amount of substrate was normalized to 1 for each sample. The results are presented as the mean and SD of three independent biological replicates. (C) As in (B), but with only Hmg2-RFP. (D) Multiple sequence alignment of selected Hrd1 and gp78 protein sequences from the indicated species. The alignment is trimmed to show transmembrane segment 3 based on S. cerevisiae Hrd1 structural models. Residues highlighted in light blue are conserved from the S. cerevisiae Hrd1 sequence. (Top) Heat maps represent conservation score of each residue across all aligned sequences and conservation score specifically within gp78 sequences. Arrows indicate positions more highly conserved in gp78 compared to overall conservation. ERAD-M deficient variants identified in this study are annotated above arrows. A conservation score of 1 indicates perfect conservation of a residue across all sequences aligned, a conservation score of 0 indicates no conservation between any sequences. (E) Model for Hrd1 interactions tuning substrate specificity of the complex. Der1 occupies Hrd1 for ERAD-L function, but this complex is ERAD-M incompetent. For ERAD-M, substrates directly contact the Hrd1 retrotranslocon for recognition and ubiquitination. Thus, the ERAD-L and ERAD-M competent complexes are in direct competition for access to Hrd1. Hrd1 also associates with itself to regulate its own activation and degradation. Dynamic flux between these complex arrangements allows for the processing of distinct types of ERAD substrates by Hrd1.

On the other hand, with Der1 overexpression ERAD-M function was further inhibited in Hrd1 F95K cells but was largely unchanged in cells expressing Hrd1 WT and F95D (Fig. 4B, second column). When Hmg2-RFP was independently expressed, Der1 overexpression inhibited ERAD-M substrate degradation even more robustly (Fig. 4C). Similar results were observed with other ERAD-M substrates, Pdr5* (SI Appendix, Fig. S6 C and D) and 6myc-Hmg2 (SI Appendix, Fig. S6 F and G). This is in alignment with increased competition between Der1 and ERAD-M substrates highlighted by the F95K variant and demonstrates a general preference of the ERAD system for lumenal substrates when Der1 is in excess.

Taken together, our results support a model where competition between Der1 and ERAD-M substrates is a central determinant of Hrd1 complex function.

ERAD-M Deficient Variants Are Differentially Conserved in Hrd1/SYVN1 and gp78/AMFR.

In higher eukaryotes, Hrd1 has diverged into two distinct E3 ligases, Hrd1 (also called Synoviolin/SYVN1) and gp78 (autocrine motility factor receptor/AMFR) (44–46). To assess the evolutionary conservation of the residues we identified to be critical for Hrd1 ERAD-M function in S. cerevisiae, we performed multiple sequence alignment (MSA) using MSAProbs analysis from the MPI Bioinformatics Toolkit (47). The alignment revealed distinct patterns of sequence conservation within the Hrd1 and gp78 transmembrane regions. This conservation supports a central role for these positions in determining functionally distinct roles (SI Appendix, Fig. S7C). The TM3 and TM8 variants identified as broadly ERAD-M deficient in this study, specifically F95, M298, and L299D, exhibited higher conservation in gp78, suggesting that the nature of the residues was required during the emergence of gp78 to support its functional specialization (Fig. 4D and SI Appendix, Fig. S7A).

To further investigate this evolutionary relationship, we constructed a maximum likelihood phylogenetic tree using the unweighted pair group method with arithmetic mean (UPGMA), implemented via the phangorn R package (SI Appendix, Fig. S7B) (48). The resulting tree revealed distinct evolutionary clades for Hrd1 and gp78 sequences where the function of Hrd1 and gp78 appears to have diverged after the evolution of multicellularity. Together, sequence and phylogenetic analysis suggest that gp78 emerged through functional divergence, and Hrd1 and gp78 have since followed unique evolutionary trajectories, likely adapting to a specialized role in higher eukaryotes (46, 49).

Discussion

In this study, we identified a mechanism for how the Hrd1 ubiquitin ligase is tuned for degradation of ERAD-L and ERAD-M substrates. Using our established DMS platform, we identified single residue Hrd1 variants that were deficient in the degradation of all model ERAD-M substrates tested (Fig. 1). The mutations were primarily substitutions of natively hydrophobic residues for charged amino acids and clustered near the proposed Hrd1 retrotranslocon cavity (Fig. 2). Strikingly, we found that certain Hrd1 variants were able to degrade ERAD-M substrates only in the absence of Der1 (Fig. 2). These variants exhibited aberrant associations with either Hrd1 or Der1, as well as reduced association with substrate in the presence of Der1 (Fig. 3). Finally, the overexpression of Der1 accelerated ERAD-L degradation and stabilized ERAD-M substrates, altogether supporting a model where Der1 is in direct competition with ERAD-M substrates for Hrd1 (Fig. 4).

In previous studies aimed at understanding ERAD-M, Hrd1 variants were reported that affected individual ERAD-M substrates (29, 30). However, no Hrd1 variants were reported that were broadly defective in degrading multiple substrates, making it difficult to elucidate any generalizable mechanism for how Hrd1 facilitates ERAD-M function. Here, we identified six individual substitutions at five unique positions in Hrd1 (I91R, F95D, F95K, M298D, L299D, and D302K) that resulted in broad ERAD-M impairment (Fig. 1 B and C). Spatially, these residues were clustered on the lumenal side of transmembrane segments 3 and 8, the two helices positioned at the lateral gate of the Hrd1 aqueous cavity (Fig. 1D) (13, 14). This provides strong genetic evidence for the universal importance of this interface in ERAD-M function, rather than only ERAD-L, as previously proposed (13–15, 33, 40, 50).

Intriguingly, structural and biochemical crosslinking data have corroborated extensive contacts between Hrd1 and Der1 at this same interface during ERAD-L (Fig. 2 A and B). Specifically, using either cysteine mutants or genetically encoded benzoyl-phenylalanine, Hrd1–Der1 crosslinks have been observed at Hrd1 TM3 (residues F83, F88, I91, L94, S98) and TM8 (residue W305) (14, 33). Der1 is not required for ERAD-M function and our previous results suggest that Der1 is likely to be positioned where Hrd1 is capable of ubiquitinating any accessible lysines in interacting proteins (35). Taken with the clustering of our ERAD-M deficient Hrd1 variants and site-specific in vivo crosslinking data, ERAD-M substrates likely also contact Hrd1 at the retrotranslocon for subsequent ubiquitination (SI Appendix, Fig. S5). Based on the structural models that exist (13, 14), this would require the displacement of Der1, thus Hrd1 TM3 and TM8 variants with increased interactions with Der1 would prevent ERAD-M substrates from effectively competing for the congruent Hrd1 site. In support of this model, we found that the I91R, F95K, and D302K mutants were rescued by Der1 deletion (Fig. 2 D and H and SI Appendix, Fig. S4 A and B). Curiously, these variants all represented substitution of positive charge on the lumenal site of retrotranslocon gate. Introduction of the opposite charge in similar positions (F95D, M298D, L299D) was also deleterious for ERAD-M but through a different mechanism, as these variants were unresponsive to Der1 deletion (Fig. 2 D and H and SI Appendix, Fig. S4 A and B). Instead, these mutants have an increased propensity to self-associate and/or weakened substrate binding/positioning within the aqueous cavity, potentially contributing to defects in ubiquitination or membrane extraction (Fig. 3 D–F).

Based on these results, we propose a model where assembly of the Hrd1 complex is a key determinant in its function for both lumenal and integral membrane substrates (Fig. 4E). Access of ERAD-M substrates to the Hrd1 retrotranslocon hinges on competition with Der1, which otherwise occupies Hrd1 at the same position in an ERAD-M incompetent conformation. Therefore, dynamic exchange of Hrd1 occupancy controls the degradation of distinct classes of substrates.

Because the Hrd1 complex is dynamic and exists in multiple stoichiometries, the oligomeric state of Hrd1 during ERAD-M remains unclear (13, 14, 17, 33, 40, 50–52). In cells, steady state levels of ERAD-M substrates were higher when an ERAD-L substrate was coexpressed (SI Appendix, Fig. S6A). Conversely, the coexpression of an ERAD-M substrate had no effect on steady state levels or rate of degradation for the ERAD-L substrate (Fig. 4B and SI Appendix, Fig. S6A), suggesting that the Hrd1–Der1 heterodimeric complex may represent the predominant state of Hrd1 in cells, and during heightened ERAD-L burden these complexes are not rearranged efficiently to support ERAD-M function. Additionally, the overexpression of Der1 further increased ERAD-L function while abrogating ERAD-M in the presence of wild-type Hrd1 or the Der1-responsive variant, F95K (Fig. 4 B and C and SI Appendix, Fig. S6 B–G). This was also observed previously, wherein the constitutive overexpression of Der1 by the strong TDH3 reporter slowed the degradation of Hmg2-GFP; our results allow us to put this observation into context (53). Together, we take this to mean that the formation of Hrd1–Der1 heterodimeric complexes is dependent on Der1 availability and that this complex is in direct competition with ERAD-M substrates (Fig. 4E).

Alternatively, multiple lines of evidence support the existence of Hrd1 homodimers and even higher oligomeric states, both in vivo and in vitro (13, 14, 17, 33, 40, 50–52). Previous crosslinking experiments in yeast indicated that Hrd1–Hrd1 dimers also form through interactions between the juxtaposed lateral gates. These studies reported self-association crosslinks at F88, I91, F95, and S98—positions that overlap with our identified TM3 mutants (I91R, F95D/K) and the putative Der1 interaction face, providing further evidence for dynamic regulation of Hrd1 complex stoichiometry at this site. Whether these oligomers represent the minimal unit required for ERAD-M function, or represent inert or regulatory forms of the Hrd1 complex, is still unclear. Certainly, oligomeric state seems to be a determinant of the autoubiquitination activity of Hrd1, as Hrd1–Hrd1 crosslinks increased when ligase activity was genetically abrogated (33) and monomeric Hrd1 reconstituted in a minimal proteoliposome system displayed little autoubiquitination activity compared to proteoliposomes containing two or more Hrd1 proteins (52) (Fig. 4E, Hrd1:Hrd1 “Activation”). That said, we cannot rule out Hrd1 monomers being sufficient for ERAD-M. If autoubiquitination in trans releases monomeric Hrd1 (33), it is improbable that Hrd1 exists indefinitely in this state with its aqueous cavity exposed to the hydrophobic core of the phospholipid bilayer. Instead, we suggest lone Hrd1s are in a constant state of “sampling” other membrane proteins in search of substrates (Fig. 4E, Hrd1 “Transient/sampling”). Our data support that, in any given cell, all conformations (monomeric Hrd1, Hrd1–Hrd1 oligomers, and Hrd1–Der1) likely exist in some equilibrium (Fig. 3). Importantly, altering or shifting this balance would allow the ERAD system a way to respond to changes in substrate burden.

In support of dynamic complex rearrangement, we found Hrd1 stability to be partially Der1-dependent. Hrd1 was stable in wild-type cells but became somewhat destabilized in the absence of Der1 (Fig. 2E), although this was not as dramatic as in hrd3Δ cells (19, 20, 53, 54). Under these conditions, Hrd1 could only exist in a monomeric or homo-oligomerized state. This might help explain the observed bias for ERAD-L when Hrd1–Der1 complexes are stabilized (Fig. 4B), while uncommitted Hrd1s undergo more dynamic regulation. The Der1-unresponsive Hrd1 variant F95D was generally less stable, even around Der1, consistent with our previous results that this mutant exhibited a preference for self-association over Der1 interaction (Figs. 2 F and H and 3 D–F). Meanwhile, Hrd1 F95K minimally interacted with itself and was mostly stable regardless of Der1 expression (Figs. 2 F and H and 3 D–F). This corroborates a role for Hrd1 stoichiometry in regulating the stability of discrete populations of Hrd1, and by extension, substrate selectivity (Fig. 4E).

Importantly, these mechanisms appear to be conserved in mammals, where Hrd1 diverged into two ligases (Hrd1/SYVN1 and gp78/AMFR), which preferentially target lumenal and integral membrane substrates, respectively (for review refs. 5 and 55). Phylogenetic analysis showed that this divergence took place after the most recent common ancestor between fungi and animals (SI Appendix, Fig. S7B). In mammals, as in yeast, misfolded lumenal substrates are exclusively processed by Hrd1, while gp78 mediates the degradation of most integral membrane substrates, including 3-hydroxy-3-methylglutaryl-CoA reductase (HMGCR, the homolog of Hmg2) (56–58) and a disease relevant cystic fibrosis transmembrane conductance regulator point mutant (CFTRΔF508, an ABC transporter comparable to Pdr5) (38, 59, 60). While mammalian Hrd1 and gp78 are orthologous to the yeast Hrd1, sequence analysis revealed that the mutations in yeast Hrd1 that specifically disrupt ERAD-M have higher conservation at equivalent positions in gp78, consistent with these regions being critical to ERAD of integral membrane substrates across eukaryotes (Fig. 4D and SI Appendix, Fig. S7 A and C). Of note, gp78 is believed to operate primarily independently of Derlins (49, 56, 61), which might imply that the cosegregation of these residues to enhance substrate selection while minimizing competition during ERAD-M.

Taken together, this study demonstrates a mechanic underlying Hrd1 ligase-mediated ERAD-M. Using DMS, we identified key Hrd1 variants broadly impaired in degrading all model ERAD-M substrates. The mutations clustered at the retrotranslocon interface, highlighting a critical role for this region in substrate recognition and degradation. We observed competition between Der1 and ERAD-M substrate for access to Hrd1 that appears to be a central determinant in ERAD complex function. Thus, dynamic shifts in Hrd1 complex assembly modulated, in part, by Der1 availability dictate Hrd1’s functional states and subsequent substrate preference. Finally, our Hrd1 mutants that disrupt ERAD-M specifically align with the evolutionary divergence of Hrd1 and gp78, where gp78 has a strong conservation at these positions. Altogether, these findings underscore the centrality of the Hrd1 retrotranslocon in all ERAD function and shed light on conserved molecular features driving selective substrate degradation pathways.

Methods

Strains and Plasmids.

Yeast strains used in this study were purchased from Horizon Discovery Ltd. and are derivatives of BY4741 (MATa his3Δ1 leu2Δ0 met15Δ0 ura3Δ0) or BY4742 (MATα his3Δ1 leu2Δ0 lys2Δ0 ura3Δ0) (SI Appendix, Table S1). Yeast were cultured at 30 °C in synthetic dropout media [0.17 % (w/v) yeast nitrogen base (Becton, Dickinson and Company), 0.5 % (w/v) ammonium sulfate (Fisher Scientific), 2 % (w/v) glucose (Sigma)]. Synthetic complete media contains the following individual nutrients: adenine sulfate (20 mg/L), uracil (20 mg/L), L-tryptophan (20 mg/L), L-histidine (20 mg/L), L-arginine (20 mg/L), L-methionine (20 mg/L), L-tyrosine (30 mg/L), L-leucine (60 mg/L), L-isoleucine (30 mg/L), L-lysine (30 mg/L), L-phenylalanine (50 mg/L), L-glutamic Acid (100 mg/L), L-aspartic Acid (100 mg/L), L-valine (150 mg/L), L-threonine (200 mg/L), and L-serine (400 mg/L). Yeast were transformed using standard LiAc/PEG methods (62). Combinational knockout strains were generated by transformation of a PCR-amplified cassette containing antibiotic resistance with homology to genomic DNA or by genetic crosses and sporulation. Genetic manipulations were verified by PCR. Plasmids were constructed using either standard restriction cloning or NEBuilder HiFi DNA assembly (New England Biolabs) and propagated in DH5α Escherichia coli. For most in vivo experiments, we used custom integrating plasmids targeted to the leu2Δ0, his3Δ1, or ura3Δ0 loci in BY4741/BY4742 strains (39). Centromeric plasmids were used only where specified in figure legends (63). All plasmids used in this study are in (SI Appendix, Table S2).

Hrd1 Steady State Collection.

To follow Hrd1 expression at steady state, Hrd1-3xFlag (or the indicated Hrd1 variants) was expressed from the native Hrd1 promoter in hrd1Δ cells grown with shaking at 30 °C to mid-log phase, OD₆₀₀ of 0.4 to 1.0. Cells were pelleted, frozen on dry ice, and stored at −80 °C until lysis. Samples were resuspended in lysis buffer (10 mM 3-(N-morpholino)propanesulfonic acid (MOPS), pH 6.8, 1 % sodium dodecyl sulfate (SDS), 8 M urea, 10 mM ethylenediaminetetraacetic acid (EDTA), fresh protease inhibitors [1 mM Phenylmethylsulfonyl fluoride (PMSF), 1.5 µM pepstatin A)] with 0.1 mm glass beads (BioSpec), at 25 OD₆₀₀/mL. Samples were vortexed for 2 min then diluted with an equal volume of urea sample buffer (125 mM trisaminomethane (Tris), pH 6.8, 4 % SDS, 8 M urea, 10 % β-mercaptoethanol). Samples were incubated at 65 °C for 5 min before separation by SDS-polyacrylamide gel electrophoresis (SDS-PAGE), transfer to a polyvinylidene difluoride (PVDF) membrane, immunoblotted with anti-DYKDDDK (Genscript) and Mouse IgG HRP-linked whole Ab (Cytiva), then imaged using chemiluminescence (ECL Select, Cytiva) with a ChemiDoc MP (Bio-Rad). Band intensities were quantified using ImageJ (NIH), normalized to total protein in the sample, using Bio-Rad Stain Free Dye (StainFree). Hrd1 variant expression levels were normalized to wild-type Hrd1. All individual Hrd1 protein level values can be found in (Dataset S1). For statistical analysis one-way ANOVA tests were performed and P values were derived using the Brown–Forsythe and Welch multiple comparisons test against Hrd1(WT).

Immunoblot-Based Degradation Assays.

To follow degradation using immunoblotting, Hrd1-3xFlag (or indicated Hrd1 variants) was integrated at the his3 locus under the native Hrd1 promoter, Hmg2-3xV5 at the leu2 locus under the strong CCW12 promoter, and Der1-HA under its endogenous promotor at the ura3 locus in hrd1Δder1Δ cells. Cells were grown in synthetic dropout media with shaking at 30 °C to mid-log phase. Cells were pelleted and resuspended in synthetic dropout media containing 10 µg/mL zaragozic acid (for Hmg2 degradation assays) or 50 µg/mL cycloheximide. Cells were grown with shaking at 30 °C and 2 OD₆₀₀ samples were collected at the indicated time points by pelleting cells at 4 °C. Samples were processed for SDS-PAGE and immunoblotting with anti-DYKDDDK (Genscript), anti-HA (clone 3F10, Roche), and anti-V5 (A01724, Genscript) as described above.

Flow Cytometry-Based Degradation Assays.

Cells were inoculated from transformation plates and cultured at 30 °C with shaking in synthetic dropout liquid media in 96 deep-well plates until cells entered log-phase (<1.5 OD₆₀₀/mL). Cells were pelleted and resuspended in fresh synthetic dropout media supplemented with either cycloheximide (at 50 µg/mL), zaragozic acid (at 10 µg/mL, for Hmg2), geranylgeranyl pyrophosphate (at 1.5 µM, for Hmg2), or ethanol (0.1 % as a vehicle control) for the indicated times. At the end of the time course, cells were pelleted, washed with ice-cold phosphate-buffered saline [PBS; 137 mM NaCl, 2.7 mM KCl, 10 mM sodium phosphate buffer, and 1.8 mM KH₂PO₄ (pH 7.4)], resuspended in PBS containing 1 µM SytoxBlue (viability dye; Invitrogen). Cells were kept at 4 °C during acquisition on a MACSQuant VYB (MACSQuantify software; Miltenyi Biotec). On the MACSQuant VYB we used the 561 nm laser for forward/side scatter (FSC/SSC) to identify single cells, and SytoxBlue fluorescence from the 405 nm laser with a 450 nm/50 nm bandpass filter set was used to exclude dead cells. GFP fluorescence was measured from the 488 nm laser with a 525 nm/50 nm bandpass filter set, and mScarlet-I fluorescence from the 561 nm laser with a 615 nm/20 nm bandpass filter set. FlowJo V10.7.1 (FlowJo LLC) was used to analyze FCS files (version 3.1) on FSC/SSC to eliminate debris, gate on single cells, and eliminate dead cells. Median GFP, or mScarlet-I, fluorescence values from at least 10,000 cells passing FSC/SSC and viability gates were exported and used to quantify Hrd1 variant function. For quantification, we determined the fraction of substrate remaining after the treatment, compared to the vehicle control, which was set to “1” for each variant. Steady state levels of substrate were compared to their expression in wild-type Hrd1 strains with the vehicle control set to “1” and Hrd1(C399S) set to “0.” Values for individual samples can be found in (Datasets S2 and S3). For saturated chases, single cells were inoculated in synthetic dropout media and grown overnight (~14 h). Cells were diluted 1 to 100 in fresh synthetic dropout media and cultured for ~24 h to enter a “saturated chase” and processed as described above. When cells were analyzed by flow cytometry in separate experiments, a representative set of controls (Hrd1 WT and C399S) are displayed for all experiments in figures. All quantifications are included in Datasets S2, S6, and S7, including the control values from individual experiments.

Library Generation.

Tiling primers mutagenesis of Hrd1 was performed as previously described (35, 64). See SI Appendix, Supporting Text and Table S3.

FACS.

FACS was performed as previously described (35). See SI Appendix, Supporting Text and Table S3.

Illumina Sequencing and Data analysis.

DNA libraries were prepared for amplicon sequencing and analysis was performed as described (35). All code and upstream processing are available (https://github.com/baldridge-lab/hrd1_dms_2023) (35). See SI Appendix, Supporting Text, Tables S7-S8, and Datasets S4-S5.

Coimmunoprecipitation of the Hrd1 Complex.

Hrd1-3xFLAG (or Hrd1 variants) were integrated at the his3 locus, 6myc-Hmg2-3xV5 at the leu2 locus, and Hrd1-3xHA and/or Der1-HA at the ura3 locus in hrd1Δder1Δ cells. Cells were cultured to mid-log phase in synthetic dropout media, pelleted, resuspended in ice-cold IP buffer (50 mM HEPES, pH 7.4, 150 mM NaCl) with fresh protease inhibitors (1 mM PMSF,1.5 μM pepstatin A) at 25 OD₆₀₀/mL, and flash frozen in liquid nitrogen to form yeast “balls” prior to cryogenic lysis by freezer-mill (SPEX SamplePrep). Cells were lysed with 2 min of grinding at five cycles per second for five rounds with 2 min of cooling time between each round. Cell powders from 25 OD₆₀₀ were thawed on ice and centrifuged at 300× g for 5 min at 4 °C to clear the lysate of unbroken cells. The cleared lysate was centrifuged at 20,000× g for 15 min at 4 °C to collect the microsomal fraction. Microsome pellets were resuspended and solubilized by rotating for 1 h at 4 °C in IP buffer supplemented with 1% decyl maltose neopentyl glycol (DMNG, Anatrace) and fresh protease inhibitors. Following solubilization, insoluble material was pelleted at 21,000× g for 30 min at 4 °C. Input samples were collected prior to diluting the solubilized proteins 1:5 in IP buffer to a final concentration of 0.2% DMNG. The solubilized proteins were then mixed with 20 μL of anti-FLAG M2 magnetic beads (Sigma) and rolled for 3 h at 4 °C. The beads were washed two times with equal volume of IP buffer containing 0.2% DMNG, followed by four washes with equal volume of IP buffer containing 0.1% DMNG. The proteins were eluted with 2× SDS sample buffer. The samples were analyzed by SDS-PAGE and immunoblotting with the inputs loaded at 1.5%. For blot quantification, band intensities were measured using ImageJ (NIH). Bound proteins were normalized to the amount of Hrd1-3xFLAG eluted in each sample. All bound proteins were set to “1” in the hrd1Δder1Δ +Hrd1-3xFLAG (WT), Hrd1-3xHA, Der1-HA sample (Fig. 3A, elution lane 4; Fig. 3D, elution lane 1; Fig. 3H, elution lane 4) to compare changes in relative amounts of bound proteins across samples.

Site-Specific In Vivo Photo-Crosslinking.

Site-specific in vivo photo-crosslinking was carried out as described (40), with the following modifications. For substrate crosslinking, hrd1Δder1Δ cells were transformed with three plasmids. The first plasmid encoded a modified tRNA-synthase, which charges tRNA with the photoreactive amino acid analog benzoyl-phenylalanine (Bpa, Cayman Chemical), as well as a tRNA that suppresses the amber stop codon (42, 43). The second plasmid contains the ERAD-M substrate 6myc-Hmg2-mScarlet, and the third contains Hrd1-3xFLAG with an amber stop codon at the indicated positions. Cells were grown overnight (16 to 20 h) at 30 °C with shaking to log phase in synthetic dropout media supplemented with 0.5 mM Bpa. Cells were harvested (50 OD₆₀₀), washed with ice-cold water, and resuspended in 2 mL of ice-cold IP buffer. One-half of the cells were left on ice while the other half was transferred to a 12-well plate and exposed to UV irradiation (λ= 365 nm) for 45 min with a Spectroline ENF-260C lamp at 4 °C. Cells were collected and subjected to cryogenic lysis and immunoprecipitation, as described above. Eluted proteins were separated by SDS-PAGE and analyzed by immunoblotting with anti-DYKDDDK (Genscript) and anti-mCherry (ab167453, Abcam) antibodies.

CUP1 Promoter-Driven Der1 Overexpression.

Der1-HA was integrated behind the CUP1 promoter at the ura3 locus in hrd1Δder1Δ cells expressing Hrd1 and Hrd1 variants under their native promoter. Individual colonies were picked into synthetic dropout media and allowed to saturate for 24 h with shaking at 30 °C. Cells were subcultured 1:100 in synthetic dropout media spiked with 0 µM, 100 µM, 250 µM, or 500 µM CuSO₄ and grown overnight, around 16 h. The next morning, cells were subcultured 1:50 in synthetic dropout media containing a consistent amount of CuSO₄ and grown to mid-log phase, taking between 4 to 6 h. 10 OD₆₀₀ of cells were harvested by centrifugation at 3,000× g for 5 min at 4 °C. Cells were washed once in 1 mL of ice-cold water and resuspended at 30 OD₆₀₀/mL in spheroplasting buffer (50 mM HEPES pH 7.4, 150 mM NaCl, 1 M sorbitol, 10 mM DTT) and incubated for 10 min on ice. Cells were collected by centrifugation at 3,200× g for 5 min at 4 °C, washed once with an equal volume of spheroplasting buffer, and resuspended in spheroplasting buffer supplemented with 40 µg/mL zymolyase 100 T. Samples were incubated for 1 h at 30 °C with shaking. The resulting spheroplasts were pelleted at 3,200× g for 5 min at 4 °C and resuspended to 20 OD₆₀₀/mL in IP buffer (50 mM HEPES pH 7.4, 150 mM NaCl, 1 mM PMSF, 1.5 µM pepstatin A, 20 µM bortezomib), and lysed with 10 strokes of a type B pestle in a 2 mL Dounce homogenizer. Lysates were cleared by centrifugation at 300× g for 5 min at 4 °C and collected into fresh tubes. Cleared lysates were centrifuged at 20,000× g for 20 min at 4 °C to isolate the microsome fraction. Microsomes were resuspended in 50 µL of 2× SDS loading buffer containing 10% BME. 10 µL (2.0 OD₆₀₀ cells) sample was loaded onto SDS-PAGE gel and subjected to immunoblotting for Der1-HA (anti-HA) as described above. Alternatively, cells for flow cytometry were grown to mid-log in 500 µM CuSO₄ and subjected to cycloheximide chase, as described.

MSA and Phylogenetic Analysis.

Hrd1 and gp78 sequences were trimmed to represent the transmembrane region based on AlphaFold (65) structural prediction models and secondary structure prediction by Ali2D from the MPI Bioinformatics Toolkit (https://toolkit.tuebingen.mpg.de/) using default parameters (30% identity cutoff to invoke a new PSIPRED run) (47). MSA was performed by MSAProbs, also from the MPI Bioinformatics Toolkit. Conservation scores were calculated using the “conserv” function from the “bio3d” package in R (version 2.4-5) (66). The MSA was converted into a distance matrix and a phylogenetic tree was built using the UPGMA, followed by maximum likelihood analysis (best fit model “LG”) using the “phangorn” R package (version 2.12.1) (48). Tree was visualized using the “ggtree” (version 3.20) from Bioconductor (67).

Supplementary Material

Appendix 01 (PDF)

Dataset S01 (XLSX)

Dataset S02 (XLSX)

Dataset S03 (XLSX)

Dataset S04 (XLSX)

Dataset S05 (XLSX)

Dataset S06 (XLSX)

Dataset S07 (XLSX)

Data, Materials, and Software Availability

Illumina sequencing data have been deposited in NCBI SRA (PRJNA1265229) (68).