The Capture of a Disabled Proteasome Identifies Erg25 as a Substrate for Endoplasmic Reticulum Associated Degradation

Studies in the yeast Saccharomyces cerevisiae have helped define mechanisms underlying the activity of the ubiquitin–proteasome system (UPS), uncover the proteasome assembly pathway, and link the UPS to the maintenance of cellular homeostasis. However, the spectrum of UPS substrates is incompletely defined, even though multiple techniques—including mass spectrometry (MS)—have been used. Therefore, we developed a substrate trapping proteomics workflow to identify previously unknown UPS substrates. We first generated a yeast strain with an epitope tagged proteasome subunit to which a proteasome inhibitor could be applied. Parallel experiments utilized inhibitor insensitive strains or strains lacking the tagged subunit. After affinity isolation, enriched proteins were resolved, in-gel digested, and analyzed by high resolution liquid chromatography-tandem mass spectrometry. A total of 149 proteasome partners were identified, including all 33 proteasome subunits. When we next compared data between inhibitor sensitive and resistant cells, 27 proteasome partners were significantly enriched. Among these proteins were known UPS substrates and proteins that escort ubiquitinated substrates to the proteasome. We also detected Erg25 as a high-confidence partner. Erg25 is a methyl oxidase that converts dimethylzymosterol to zymosterol, a precursor of the plasma membrane sterol, ergosterol. Because Erg25 is a resident of the endoplasmic reticulum (ER) and had not previously been directly characterized as a UPS substrate, we asked whether Erg25 is a target of the ER associated degradation (ERAD) pathway, which most commonly mediates proteasome-dependent destruction of aberrant proteins. As anticipated, Erg25 was ubiquitinated and associated with stalled proteasomes. Further, Erg25 degradation experimental and control strains treated with a proteasome inhibitor, and label-free differential MS was used to compare isolated proteins from each condition. Data obtained in the presence of a proteasome inhibitor identified known UPS substrates along with an ER resident protein, Erg25, which had not previously been directly characterized as a UPS target. Subsequent biochemical and genetic analyses confirmed that Erg25 associates with the proteasome, is polyubiquitinated, and is stabilized in a yeast strain lacking ERAD-associated E3 ligases, thereby establishing Erg25 as a bona fide ERAD substrate. Because this enzyme plays a critical role in the biosynthesis of a yeast sterol 29 , our data expand the number of regulated, wild-type enzymes directed to the ERAD pathway and highlight the ER as a dynamic regulator of both protein synthesis and degradation.


INTRODUCTION
The 26S proteasome, a multi-catalytic cytosolic protease, serves as the major proteolytic factory in eukaryotes. The 26S particle is formed by two species, a 20S core particle, which houses pairs of trypsin, chymotrypsin, and caspase-like enzymes, and two 19S "caps" (or PA700 particles) that abut each end of the core particle 1,2 . Based on its abundance and robust activity, the targeting and destruction of substrates to this protease are tightly controlled because the vast majority of proteasome substrates are covalently modified by an isopeptide ubiquitin (Ub) linkage. Ub is added onto a target via a cascade of Ub activating, conjugating, and ligating enzymes, and covalent Ub appendages are formed between the C-terminus of Ub and most commonly the e amino group on a Lys in a protein substrate or on a previously attached Ub 3 . Although numerous varieties of poly-Ub "chains" have been described-due to the presence of seven internal Lys residues in Ub as well as the N-terminus-the most common poly-Ub linkage recognized by the proteasome is formed by the sequential addition of the Cterminus of Ub onto a Lys in Ub at position 48 [4][5][6] . Early results indicated that a poly-Ub chain of at least four Ub substituents is required for proteasome recognition 7 .
As a result of fine-tuned mechanisms underlying the recognition and Ub-tagging of example, Stable Isotope Labeling by/with Amino acids in Cell culture (SILAC) of yeast containing or lacking components of the ER-associated ubiquitination machinery led to the isolation of Erg1, an enzyme in the ergosterol pathway, as a regulated ERAD substrate 23 . A homolog of Erg1 that acts in the sterol biosynthetic pathway in higher cells was targeted by homologous components of the ubiquitination machinery in human cells, validating the power of the yeast-based screen. Moreover, the capture of ubiquitinated proteins after treating human cells with a retrotranslocation inhibitor and SILAC analysis identified other ERAD substrates 24 . Many of these more recently identified substrates are enzymes whose steady-state levels are controlled by ERAD [25][26][27] . Nevertheless, there are undoubtedly undiscovered ERAD substrates due to the fact that probably >7,000 proteins in a human cell associate at some point during their biogenesis with the ER 28 . In addition, due to the expansion of genome datasets, a plethora of mutations and polymorphisms have been identified in secreted and membranes proteins, and many of these variants may also be targeted for ERAD.
Together, there is a growing appreciation that metabolic pathway components residing in the ER as well as aberrant proteins are subject to ERAD.
In this study, we wished to identify previously uncharacterized UPS substrates using an approach that, to our knowledge, had not previously been pursued in yeast. First, we created a yeast strain that lacked a drug efflux pump so a specific inhibitor of the proteasome could be applied. Next, a proteasome subunit was tagged with an epitope and the gene encoding the modified subunit was integrated into the chromosome in the drug-sensitive yeast strain. We then performed a series of pull-down assays using 8 experimental and control strains treated with a proteasome inhibitor, and label-free differential MS was used to compare isolated proteins from each condition. Data obtained in the presence of a proteasome inhibitor identified known UPS substrates along with an ER resident protein, Erg25, which had not previously been directly characterized as a UPS target. Subsequent biochemical and genetic analyses confirmed that Erg25 associates with the proteasome, is polyubiquitinated, and is stabilized in a yeast strain lacking ERAD-associated E3 ligases, thereby establishing Erg25 as a bona fide ERAD substrate. Because this enzyme plays a critical role in the biosynthesis of a yeast sterol 29 , our data expand the number of regulated, wild-type enzymes directed to the ERAD pathway and highlight the ER as a dynamic regulator of both protein synthesis and degradation.
genotypes of the resulting strains were confirmed by PCR amplification of the NAT cassette at the PDR5 locus. The Δpdr5 phenotype was confirmed by assessing the inhibition of the degradation of a known proteasome substrate, NBD2*, in response to MG132 treatment, as described 34 . A plasmid containing a GFP-tagged control protein, SZ*, was used as described. 35 Unless indicated otherwise, all yeast manipulations and growth conditions employed standard methods 36 .

Pre8 affinity purification
The following yeast strains were used for the MS analysis: (1) those containing a wildtype copy of PDR5 and an untagged proteasome subunit (PRE8 cells), (2) those containing the proteasome-tagged subunit as well as the wild-type copy of PDR5 (PRE8:GFP cells), and (3) those lacking PDR5 but harboring the tagged proteasome subunit (PRE8:GFP ∆pdr5; Fig. 1A). All three strains also expressed NBD2* (see above) from an introduced vector. The strains were grown to late log phase in selective media at 30°C and MG132 (Selleckchem, Houston, TX) was added to a final concentration of 100 µM to all cultures for 1 hr prior to harvesting. Approximately 20 OD 600 units of cells were used for each GFP affinity purification. After incubation, cells were harvested by centrifugation at 2000g for 5 min and the cell pellets were washed twice with 5 ml of ice-cold PBS. Cell lysis was carried out in 1 mL of ice-cold lysis buffer (10 mM TrisCl, pH 7.5, 150 mM NaCl, 0.5 mM EDTA, 5 mM MgCl 2 , 10 mM ATP, 0.5% NP-40, and an ATP regeneration mix 37 ) in the presence of freshly added protease and phosphatase inhibitors (Thermo Fisher Scientific) using the FastPrep24 instrument and by guest on September 8, 2020 Lysing Matrix C (MP Biomedical, LLC, Santa Ana, CA, USA). The disruption was achieved with 3 rounds of vigorous Vortex mixing at a speed setting of 6 meters/sec for 60 sec. After centrifugation for 5 min at 4 °C at 16000g, the supernatants were transferred to new Eppendorf tubes containing 20 μL GFP-Trap®_A bead slurry (ChromoTek GmbH, 82152 Planegg-Martinsried, Germany) pre-washed with washing buffer (10 mM TrisCl, pH 7.5, 150 mM NaCl, 0.5 mM EDTA). After 2 hr of end-over-end rotation at 4 °C, the GFP-Trap®_A beads were collected by centrifugation at 2000g for 2 min and then washed twice with 500 µL of ice-cold washing buffer. Bound proteins were eluted at 95°C in 25 µL 2x NuPAGE™ LDS Sample Buffer (Thermo Fisher Scientific) for 10 min. All 18 GFP affinity purifications (6 purifications per transformed cell type; Fig. 1A) were performed simultaneously on the same day.

In-gel trypsin digestion
In-gel trypsin digestion was carried out as previously described 38 . In brief, eluates from the affinity purification were resolved by SDS-PAGE (NuPAGE™ 4-12% Bis-Tris Protein Gels, 1.0 mm, 10-well; Thermo Fisher Scientific) at 150 V for 10 min and stained with SimplyBlue TM SafeStain (Thermo Fisher Scientific). The stained region was excised, washed with HPLC water, and destained with 50% acetonitrile (ACN)/25 mM ammonium bicarbonate until no visible staining remained. Gel slices were dehydrated with 100% ACN and reduced with 10 mM dithiothreitol (DTT) at 56°C for 1 hr, which was followed by alkylation with 55 mM iodoacetamide (IAA) at room temperature for 45 min in the dark. The gel pieces were then dehydrated with 100% ACN to remove excess by guest on September 8, 2020 DTT and IAA. Next, 50 µL of 20 ng/ µL trypsin in 25 mM ammonium bicarbonate was added for overnight digestion at 37°C. The resulting tryptic peptides were extracted with 70% ACN/5% formic acid (FA), vacuum dried, and reconstituted in 18 µL 0.1% FA. A pooled instrument control sample was generated by combining equal volumes of the individual digests and analyzed repeatedly to monitor nLC-MS/MS performance throughout the duration of the experiment. To minimize systemic bias, a randomized block design was used to balance the sample analysis order by sample type (Supplemental Table S1). Reverse-phase separation was achieved with a 66 min linear gradient composition of 2-35% ACN/0.1% formic acid and a flow rate of 300 nL/min. A positive electrospray ionization voltage of 1.9 kV and a capillary temperature of 275°C was used to ionize and desolvate the eluted peptides, respectively. The mass spectrometer was operated in the data-dependent acquisition mode that records one high-resolution mass spectrum (MS1) followed by low-resolution tandem mass spectra (MS2) for each of the 13 most by guest on September 8, 2020 abundant precursor ions detected in the MS1. All high-resolution MS1 spectra were acquired with a m/z range of 375-1800 Da, an Orbitrap resolution setting of 60,000, and an automatic gain control (AGC) target value of 1.0E06. Low resolution MS2 spectra were acquired using collision-induced dissociation (CID) and an AGC target value of 5.0E06. The acquisition of replicate MS2 spectra was minimized using a dynamic exclusion time of 90 sec.

Label-free differential mass spectrometry
The nLC-MS/MS data were analyzed with the MaxQuant software suite (version 1.6.6.0) 30,31 . The Andromeda protein identification search engine and SwissProt 20 min and match time window at 1.5 min. A 1% false discovery rate (FDR) was used to filter the peptide identification results. Quantification of the MS1 peptide intensity was performed for all identified peptides following retention time alignment and peak matching across samples, and these values are provided in Supplemental Tables S2,   S3, and S4 (protein groups with single peptide identification were excluded). The mass spectrometry proteomics data have been deposited to the ProteomXchange 39 consortium via the MassIVE partner repository (also see "Data Availability" section, below).
In general, MaxQuant-calculated protein and peptide intensity values, statistical tests, fold-change cut-offs, and practical considerations were used to select Pre8 interacting partners, including proteasome particles, proteasome partners, and UPS substrates from non-specific contaminants. Proteins with a single identified peptide were excluded from further data analysis. First, log2-transformed protein intensity values were subject to a Student's t-test and a p-value cut-off were used to evaluate "protein-level" expression differences between experimental and control samples. Next, proteins with a protein-intensity fold-change greater than 2 when comparing the PRE8:GFP or PRE8:GFP ∆pdr5 to PRE8 protein intensity values were selected for additional analysis.
Proteins with a protein-intensity fold-change greater than 2 and p value less than 0.01 when comparing the PRE8:GFP or PRE8:GFP ∆pdr5 to Pre8 protein intensity values were selected for additional analysis. A protein was considered a bona fide Pre8 interacting partner if two or more peptides had p-values < 0.01 and fold-change > 2 based on peptide intensity values. To select putative UPS substrates, protein intensity by guest on September 8, 2020 values of Pre8-interacting partners in each sample were normalized by its Pre8 protein intensity, and a Student's t-test on log2-transformed Pre8-adjusted protein intensity values was used to compare PRE8:GFP samples and PRE8:GFP ∆pdr5 samples. A pvalue cut-off < 0.05 and fold change greater than 1.5 was used for UPS substrate selection.

Immunoprecipitation assays
The GFP-tagged Pre8 strain lacking PDR5, described above, was transformed with pIU2593 (2μ URA P ADH1 -ERG25-HA) 40 , a kind gift from the laboratory of Drs. Jack Kaplan and Diane Ward (University of Utah), or a vector control (pRS416) 41 . As a control, a ∆pdr5 strain lacking GFP-tagged Pre8 was transformed with pIU2593.
Overnight yeast cultures were grown at 30°C in selective media overnight, diluted into 30 ml of media the next morning, grown for 2-3 hr and then incubated for 1 hr with 50 µM MG132 (to stabilize the substrate) prior to harvesting ~30 OD 600 of log phase cells by centrifugation (see above). Cell pellets were resuspended in 500 µL of buffer (50 mM Tris-Cl, pH 7, 150 mM NaCl, 1% TX-100, plus Roche Complete EDTA-Free PI cocktail and 10 mM N-Ethylmaleimide (NEM)) and disrupted by the addition of glass beads and agitation on a Vortex mixer four times for 1 min. The lysate was cleared by centrifugation at ~16,000g for 10 min at 4 °C in microfuge, and the supernatant was incubated with 25 µL anti-HA affinity matrix (Roche) overnight on a rotator at 4°C. The anti-HA affinity matrix was washed three times with buffer, and the immunoprecipitated proteins were eluted with SDS-PAGE sample buffer plus NEM and sequentially by guest on September 8, 2020 incubated at 50°C (for anti-HA) and then at 90°C (for anti-GFP). Protein elution was followed by SDS-PAGE and western blot analysis. Erg25 was detected with anti-HA HRP-linked antibody (Roche) and Pre8 was detected with anti-GFP and anti-mouse IgG HRP linked secondary antibody (Cell Signaling). Image analysis was performed using a BioRAD Universal Hood II Imager, and signals were quantified using ImageJ 1.51h software (National Institutes of Health).
Assays to detect the levels of ubiquitinated Erg25 were performed essentially as described 42 . A yeast strain lacking Pdr5 (Δpdr5; Open Biosystems, Thermo Fisher) was transformed with pIU2593 (Erg25-HA; see above) and pRS423mycUbiqutin (2µ HIS3 PAGE sample buffer plus 10 mM NEM and heated to 50°C for 10 min prior to analysis by SDS-PAGE and western blotting. Erg25 was detected as described above, and ubiquitin was detected by immersing the blot in a boiling water bath for 10 min prior to blocking in a milk solution and incubation with anti-ubiquitin antibody (P4D1, Santa Cruz Biotechnology). Image analysis and quantification were performed as described above.
Yeast strains were transformed with pIU2593 (see above), NBD2* 34 (also see above) or left untransformed, and cultures were grown in the appropriate media at 30°C before cycloheximide chases were performed as described 46 . The CDC48 and cdc48-2 strains were grown at room temperature and then incubated at 39°C degrees for 2 hrs prior to addition of cycloheximide to induce the thermosensitive phenotype. The Δpdr5 and isogenic wild type strains were pre-treated with 50 µM MG132 or DMSO for 1 hr or, where indicated, with 40 µM fluconazole (Sigma) for 2 hr prior to the addition of cycloheximide. Yeast cell lysates were generated using alkaline lysis followed by trichloroacetic acid precipitation 47 . Protein pellets were resuspended in SDS-PAGE sample buffer, heated at 50°C for 10 min, and subject to SDS-PAGE and western blotting. Western blot detection of Erg25-HA, imaging and quantitation were performed as described above. Native Erg25 was detected using anti-Erg25 immunosera (a kind gift from Dr. Jack Kaplan and Diane Ward at the University of Utah), and an anti-rabbit by guest on September 8, 2020 immunoglobulin G (IgG) HRP-conjugated secondary antibody (Cell Signaling Technology).

RESULTS
The goal of this study was to develop an alternate approach to identify UPS substrates in yeast that might have been missed through other methods. This pursuit is warranted based on the fact that each attempt to identify UPS substrates-using a variety of methods-has provided a list of non-overlapping substrates compared to prior attempts.
Thus, each experimental regimen most likely exhibits unique attributes and, in turn, suffers from deficiencies. We therefore developed a new substrate capture assay in yeast in which proteasome particles, associated partners, and UPS substrates could all be isolated when the proteasome was either active or inactive. We reasoned that substrates would be rapidly degraded when the proteasome was active, but these substrates would instead associate with the proteasome was it was inactive. In contrast, core proteasome components would be isolated regardless of whether the proteasome was active or inactive. As a negative control, a pull-down was also conducted in which an epitope on a proteasome subunit for antibody-based isolation was absent.
We first utilized a strain in which the PRE8 chromosomal locus in a wild type yeast strain was replaced with Pre8-GFP to allow for detection and isolation 48 . PRE8 encodes an a subunit in the proteasome core particle and resides near a component in the 19S by guest on September 8, 2020 cap that stabilizes the holoenzyme, i.e., the 26S particle 49 . Next, we deleted the PDR5 gene in this strain (see Experimental Procedures) since loss of the encoded drug efflux pump allows for controlled proteasome inhibition when a specific inhibitor is added, such as MG132 50,51 . A wild-type strain in which wild-type PRE8 and PDR5 loci were maintained in the chromosome was used as a negative control (Fig. 1A).
Next, we confirmed that the constructed strains behaved as expected. First, we noted that each yeast strain grew identically (data not shown), and we detected the GFPtagged Pre8 subunit in yeast in which this modified subunit was integrated (Fig. 1B, lane   1). The specificity of the signal was confirmed by introducing a GFP-tagged protein into a wild-type strain (lane 3). 35 In contrast, cells containing the endogenous, unaltered copy of Pre8 lacked the GFP signal (lane 2). Second, we confirmed that the degradation of a known UPS substrate, NBD2* 34 , was slowed in cells lacking the Pdr5 drug efflux pump after a short pre-incubation with MG132, as expected (Fig. 1C).
After confirming the expected behavior of the strains, we followed the protocol outlined in Fig. 2A in which the indicated three strains were treated with MG132 for 1 hr and the proteasome was immunoprecipitated via GFP-affinity capture (see Experimental Procedures). NBD2* was also expressed from a transformed vector in all three strains to ensure quality control. Mass spectrometry identified 4166 peptides belonging to 1039 proteins, with 575 of them with a minimum of 2 peptides (Supplemental Table S2, S3, and S4). We found 84 and 144 proteins significantly enriched in the pull-down samples by guest on September 8, 2020 from PRE8:GFP and PRE8:GFP ∆pdr5 cells respectively, resulting in a combined total of 149 proteins as Pre8-interacting partners (Supplemental Table S5).
The robustness of GFP-affinity capture in isolating proteasomes was confirmed by the identification of all 33 proteasome subunits, including the 14 core subunits as well as the entire proteasome lid and base 8 . In addition, high resolution MS based quantification indicated that these subunits were significantly enriched in samples expressing GFPtagged Pre8 compared to untagged-Pre8 (Fig. 2B). These results validate the efficacy of the pull-down protocol and mass spec detection. We further validated the experimental design by asking whether the known UPS substrate, NBD2*, was enriched in the presence of MG132 in the PRE8:GFP tagged strain lacking the Pdr5 drug efflux pump. As shown in Fig. 2C, the intensity of NBD2* rose about 2.5-fold when the proteasome was inhibited. We also detected two known proteasome assembly chaperones as being stabilized in the presence of MG132 in the Pre8:GFP tagged strains. These proteins, Add66 (Pba2) and Pba1, assist the formation of the 20S core particle and are degraded after aiding in the assembly process (Fig. 2D). [52][53][54] Together, these data support our development of a new method to isolate previously unknown proteasome substrates.
We next more closely examined the list of proteins that were statistically enriched in replicate samples from PRE8:GFP ∆pdr5 cells compared to PRE8:GFP cells that had been treated with MG132. As presented in Table 1, a total of 27 proteins were identified, which in principle represent potential proteasome substrates. One protein, which had by guest on September 8, 2020 not previously been identified as a UPS substrate but exhibited a >5-fold enrichment and the third lowest p-value was Erg25. Erg25 is a known ER resident protein that converts dimethylzymosterol to zymosterol 29 . Zymosterol is an ergosterol precursor (see below), and ergosterol is the primary sterol in the yeast plasma membrane where it plays an analogous role to cholesterol in higher eukaryotic cells. Because other enzymes required for the synthesis of ergosterol were identified as ERAD substrates 23, 55-57 , we focused our subsequent efforts on Erg25 and asked whether this enzyme is also an ERAD substrate.
We first wished to confirm that Erg25 is proteasome-associated and therefore performed a coimmunoprecipitation experiment in PRE8:GFP yeast expressing an HA epitope-tagged form of Erg25. When Erg25 was immunoprecipitated using anti-HA antibody, GFP tagged Pre8 coprecipitated under nondenaturing conditions after treatment with MG132 (Fig. 3A, lane 3). In contrast, coprecipitation was absent in cells lacking Pre8-GFP or Erg25-HA (lanes 1 and 2). These results confirmed proteasome-Erg25 interaction, as suggested by the mass spec analysis.
Based on these data, we predicted that Erg25 is ubiquitinated and that this modification directs the enzyme to the proteasome. To this end, we precipitated HA-tagged Erg25 species in the presence or absence of MG132 in cells lacking the Pdr5 drug efflux pump. The cells also contained a plasmid that increased the steady-state levels of Ub, which aids in the detection of this transient modification 43 . After immunoprecipitating Erg25, we examined the isolated proteins for the presence of a polyubiquitin chain by by guest on September 8, 2020 western blot analysis. As shown in Fig. 3B, the level of ubiquitinated Erg25 increased when proteasome function was impaired, i.e., when cultures were incubated with MG132 prior to cell lysis (lane 3). Next, to more directly determine whether Erg25 is degraded in a proteasome-dependent manner, we measured Erg25 stability by performing a cycloheximide chase analysis when the proteasome was inhibited in a pdr5D background in the presence or absence of MG132. As shown in Fig. 3C, Erg25 degradation was slowed in the presence of MG132. Combined with the fact that Erg25 is an ER-resident protein 29 , these data establish Erg25 as an ERAD substrate.
Next, because ERAD substrates are ubiquitinated by ER-associated E3 ubiquitin ligases, we measured the rate of Erg25 turnover in a cycloheximide chase assay in the absence of one or both of the canonical ERAD-associated E3 ligases, Hrd1 and Doa10 ( Fig. 4A) 19,58 . Based on a mass spec analysis, Christiano et al. previously found that deletion of Hrd1 somewhat increased Erg25 half-life 59 ; however in our hands, the rate of Erg25 degradation was not significantly affected in either of the single E3 mutants (Δhrd1 or Δdoa10). In contrast, deletion of both Hrd1 and Doa10 almost completely stabilized Erg25 (Fig. 4A). Most commonly, Doa10 and Hrd1 target distinct groups of ERAD substrates for degradation, which is dependent on where the misfolded lesion resides 60 , yet other cases exist in which both enzymes are required to target a specific membrane protein for ERAD 46 . The difference between our data and the previous data might reflect the unique strain backgrounds used or methodology. Regardless, these data further support the targeting of Erg25 to the ERAD pathway.
by guest on September 8, 2020 After ubiquitination of ERAD substrates at the ER membrane, Cdc48, a AAA+-ATPase, provides the energy required to retrotranslocate ERAD substrates from the ER to the cytosol. Cycloheximide chases performed in either wild type (CDC48) yeast or yeast containing a temperature sensitive Cdc48 allele, cdc48-2, revealed that Erg25 degradation was significantly inhibited in the cdc48-2 strain when cells had been preshifted to a non-permissive temperature (Fig. 4B). These data strengthen our conclusion that Erg25 I a proteasome dependent, ERAD substrate.
Previous work established that the ERAD of two other enzymes in the ergosterol biosynthetic pathway, Hmg2 and Erg1 (see Fig. 5A), is regulated by the level of biosynthetic intermediates 23,61,62 . For example, the accumulation of lanosterol triggers Erg1 degradation, therefore blocking the synthesis of lanosterol precursors 23 . To ask whether the degradation of Erg25 is also regulated by biosynthetic intermediates, we measured Erg25 stability in the presence or absence of fluconazole, which inhibits Erg11 (Fig. 5A) 63 . As shown by cycloheximide chase analysis, pretreatment of yeast with fluconazole slowed Erg25 turnover (Fig. 5B). To confirm that the regulation of Erg25 stability is impacted by ergosterol pathway intermediates, we assessed Erg25 degradation in yeast lacking either Erg2 or Erg3. Unlike fluconazole, Erg2 and Erg3 act downstream of Erg25 in the ergosterol pathway. Surprisingly, deletion of either Erg2 or Erg3 also stabilized Erg25 (Fig. 5B), implying that accumulation of downstream intermediates positively affects Erg25 stability. Because prior work reported that deletion of Erg6 had no effect on proteasome-dependent degradation 50 , we do not believe that inhibiting the ergosterol pathway negatively affects ERAD generally. It is by guest on September 8, 2020 also possible that these perturbations affect ER homeostasis, stabilizing Erg25. Future studies will better characterize how the Erg25 degradation pathway is regulated.
Nevertheless, these data indicate that Erg25 is a regulated ERAD substrate, one that responds to the levels of sterol intermediates in the ER membrane.

DISCUSSION
MS has been used extensively to identify proteasome components along with the spectrum of UPS targets [64][65][66] . The experimental system outlined in the current study was designed so that core components of the proteasome as well as UPS targets could be identified in parallel by MS and so that non-specific partners could be readily excluded.
Among the 27 proteasome-associated proteins enriched in the PRE8:GFP ∆pdr5 cells compared to the PRE8:GFP cells treated with MG132 were known UPS substrates. For example, Pba1 and Add66 are yeast homologs of a proteasome assembly chaperone complex that are degraded after aiding 20S core particle assembly 52,53 . Interesting, Rad23 and Dsk2, were also isolated as putative UPS substrates in our analysis. Dsk2 and Rad23 function redundantly to deliver ubiquitinated substrates to the proteasome during ERAD 67, 68 , but Rad23 is known to escape degradation because it lacks a proteasome initiation region 69 . We propose that their enrichment as proteasome partners when degradation is disabled reflects a build-up of ferried substrates to the proteasome. Cdc48 is another factor that escorts ubiquitinated substrates to the proteasome 70 . Cdc48 was similarly enriched in the MG132-sensitive cells (Table I).
by guest on September 8, 2020 Other enriched proteins include Blm10 and Ecm29. These known proteasomeassociated proteins respectively open the aperture to the proteolytic chamber in the 20S core particle, which augments substrate entry 71,72 , and inhibit the ATPase and proteolytic activity of the proteasome 73 . A structural analysis of Blm10 bound to the proteasome suggests that it favors peptide entry and proteolysis in the 20S core 74,75 .
Our data suggest, under specific conditions, that it might also become a proteasome substrate. In turn, Ecm29 binds more avidly to the 26S particle than to dissociated subunits 76,77 . Because small molecule proteasome inhibitors, such as MG132, stabilizes reconstituted 26S particles 76 , the enrichment of Ecm29 with stalled proteasomes hints that this protein could similarly be an proteasome substrate. Finally, we note that Vph1 was identified as a high-confidence proteasome-targeted substrate. Cooper and colleagues previously reported that Vph1 is a robust ERAD substrate when a specific assembly chaperone is absent 78 . The data in our paper suggest that even when the chaperone (Vma22) is present, a portion of this topologically complex membrane protein is targeted for ERAD. These data highlight the fact that membrane protein folding in the ER is faulty, as observed for numerous other substrates 16 .
The strain used for these studies, which allows for treatment with a proteasome inhibitor and the facile isolation of proteasomes, will prove valuable for future work in the field.
For example, genes encoding specific proteins that target ubiquitinated substrates to the proteasome can be deleted in our strain. A comparison of isolated proteins, which takes advantage of the Pre8:GFP chimera, might identify specific substrates of these targeting factors 79 . In addition, numerous post-translational modifications have been by guest on September 8, 2020 identified on proteasome substrates that regulate their degradation 80 . The isolation of proteasome partners in strains deleted for genes encoding factors that add or remove these modifications may also identify new clients. Finally, the isolation of proteasomes after subcellular fractionation, e.g., of mitochondria, could identify new UPS substrates that are compartment-specific.
Prior to our work, Erg25 had not been characterized as an ERAD substrate. This is somewhat surprising given the range of MS studies performed to identify ERAD substrates and the fact that several other components of the ergosterol biosynthetic pathway are regulated by ERAD. An analysis of the stability of the human Erg25 homolog 81 , which is 38% identical to the S. cerevisiae isoform, is now warranted.
Interestingly, yeast Erg25 was identified in three other proteomic studies. First, the isolation of organelle membranes and immunoprecipitation from yeast expressing an epitope-tagged ubiquitin species in an ERAD mutant uncovered putative membraneassociated ERAD substrates 57 . Erg25 was absent from the list of 83 putative substrates, but a ubiquitinated Lys in the protein (IPSAK*EQLYCLK) was detected in the larger pool of modified proteins. Other proteins in the ergosterol pathway, Erg1, Erg9, and Erg27, were also detected in this analysis, suggesting that these enzymes may also be ERAD substrates. Indeed, Erg1 was later confirmed as a regulated ERAD substrate by Carvalho and colleagues 23 . Additional experiments-such as those performed in the present study-are needed to determine whether Erg9 and Erg27 are targeted for ERAD. Second, Erg25 was detected in a SILAC analysis comparing yeast containing or lacking components of the ubiquitin conjugation machinery. However, Erg25 failed to by guest on September 8, 2020 meet the criteria required to classify the protein as a bona fide ERAD client 55 . Finally, Erg25, as well as Erg1, Erg3, Erg5, and ERg11 were classified as proteins with short half-lives relative to the proteome as a whole, and that the Erg25, Erg3 and ERg5 halflives increased in a Hrd1 deletion strain 59 . However, in that study, the role of the ERAD pathway in targeting Erg25 was not directly tested, nor was the potential role of sterol synthesis examined. Overall, these and other data are consistent with the fact that each MS protocol exhibits unique strengths with regard to its ability to isolate intended targets (also see Introduction). We therefore suggest that an identification of the spectrum of ERAD and UPS substrates has and will continue to benefit from diverse approaches.