A cytosolic degradation pathway, prERAD, monitors pre-inserted secretory pathway proteins

Cellular quality control has an essential role in shaping and maintaining a functional cellular proteome (Princiotta et al., 2003; Schubert et al., 2000). One of the best-characterized quality control triaging systems is that of the endoplasmic reticulum (ER), which comprises the central folding site of the secretory pathway. Disposal of misfolded ER proteins occurs through the ER-associated degradation (ERAD) pathway (Brodsky and McCracken, 1997), which involves the recognition of misfolded substrates by the membrane-embedded ERAD complexes, retro-translocation to the cytosol, ubiquitylation and handover to the proteasome for degradation (Hebert and Molinari, 2007; Meusser et al., 2005).

Two key pathways mediate ERAD for yeast secretory pathway proteins and they are centered around the Hrd1 and the Doa10 E3 ubiquitin ligases (Carvalho et al., 2006; Huyer et al., 2004; Vashist and Ng, 2004). The Hrd1 complex takes part in the degradation of proteins that bear misfolded lesions in the ER lumen or membrane, and is therefore also referred to as the ERAD-L/M pathway. The Doa10 complex recognizes transmembrane proteins with misfolded cytosolic domains, and is known as the ERAD-C pathway. Some auxiliary factors are common to both ERAD-L/M and ERAD-C pathways, such as the Cdc48 complex, which generates the driving force that is required for membrane extraction in both pathways (Rabinovich et al., 2002 Ravid et al., 2006).

To date, quality control in the secretory pathway has primarily been studied in the context of proteins that have successfully translocated into the ER. However, a large fraction of the secretome fails to engage the signal recognition particle (SRP), and is temporarily found in the cytosol prior to its translocation (Ast et al., 2013; Chen et al., 1993; Hessa et al., 2011; Shao and Hegde, 2011). We reasoned that additional cytosolic monitoring measures might be in effect during this stage of SRP-independent substrate biosynthesis. Here, we used an unbiased, systematic genetic screen with an SRP-independent substrate to reveal that this is indeed the case. If SRP-independent substrates do not enter the ER, they are tagged for degradation on the cytosolic face of the ER leaflet in a pathway we term prERAD. prERAD relies on opposing forces of ubiquitylation and deubiquitylation machineries, ensuring that nascent SRP-independent proteins are provided sufficient time to translocate but are efficiently cleared from the membrane if they fail to do so.

Proteins that undergo SRP-independent targeting to the ER must transiently reside in the cytosol. Given that such secretory proteins contain hydrophobic targeting signals, should they remain within the cytosol, they would needlessly engage the cytosolic folding machinery or possibly generate cytosolic aggregates. We therefore hypothesized that monitoring measures must be in place to clear the cytosol of proteins that have reached the ER membrane but have not translocated.

To uncover which proteins might mediate such pre-insertional degradation we performed an unbiased and systematic genetic screen in which we visualized the fluorescently tagged RFP–Gas1, an SRP-independent substrate of the glycosylphosphatidylinositol (GPI) family (Ast et al., 2013; Ng et al., 1996) on the background of mutations in 210 proteins that are affiliated with quality control (Fig. 1A). Although the majority of mutations did not affect the localization of RFP–Gas1, we identified nine mutations that generated a mislocalized pattern (Fig. 1B). These mutants could be grouped into three functional categories: those that affected the proteasome, the cytosolic arm of ER-associated degradation (ERAD-C) or protein deubiquitylation (DUBs).

To date, ERAD has been primarily studied in the context of misfolded secretory pathway proteins that have translocated into the ER (Hampton, 2002; Meusser et al., 2005). However, translocated Gas1 is completely luminal (Conzelmann et al., 1988), and would not be available to the cytosolic-sensing ERAD-C pathway. Moreover, the absence of the luminal arm of ERAD (ERAD-L) had no affect on the localization of RFP–Gas1, precluding an indirect effect on RFP–Gas1 localization in strains lacking functional ERAD pathways (supplementary material Fig. S1). Therefore, we hypothesized that ERAD-C could be interacting with the pre-inserted (cytosolic) form of RFP–Gas1.

To measure which form of RFP–Gas1 was being stabilized by deletions in the ERAD-C pathway, we halted translation with cycloheximide and followed the remaining protein pool by western blot analysis (Fig. 2A). In wild-type (WT) cells, we found that the cytosolic form of RFP–Gas1 disappeared after 60 minutes of translational halt. When translocation was attenuated, by utilizing the sec61-DAmP background, nearly all of RFP–Gas1 was found in the cytosol at the start of the experiment. Within 60 minutes, this cytosolic protein pool had been cleared, indicating that a pre-insertional degradation pathway was indeed in effect. In Δdoa10 cells, the amounts of cytosolic RFP–Gas1 was both elevated and stabilized over the time course of 60 minutes, suggesting that ERAD-C is indeed responsible for tagging the cytosolic form of RFP–Gas1 for degradation. Moreover, it appears that the translocation pathway is independent of the pre-insertional degradation pathway, as mature RFP–Gas1 is present in normal levels in Δdoa10 cells at the initial time point.

We hypothesized that Doa10 might degrade a subpopulation of the pre-inserted RFP-Gas1 for two reasons: (1) degradation is the result of protein folding, as folded proteins have lost their translocation competence; or (2) degradation is triggered by the cytosolic concentration or dwell time of the substrates. To differentiate between these two scenarios, we attenuated both translocation and degradation using a sec61-DAmP/Δdoa10 double mutant. Interestingly, we saw that at the initial time point, there is more mature protein when compared to the single mutant (sec61-DAmP). Furthermore, following 60 minutes of cycloheximide treatment, the majority of the cytosolic RFP–Gas1 had been translocated and trafficked onwards to the Golgi and cell membrane. Thus, it seems that the cytosolic pool of RFP–Gas1 is competent for translocation. When we imaged RFP–Gas1 in cells that were attenuated for both translocation and ERAD-C (supplementary material Fig. S2), we saw that the absence of the E3 ubiquitin ligase, DOA10, altered RFP–Gas1 localization from cytosolic inclusion bodies to an ER pattern. This phenomenon is in line with findings that trafficking to some inclusion bodies depends on ubiquitylation (Kaganovich et al., 2008). We therefore suggest that Doa10 can tag pre-inserted RFP–Gas1 for degradation, in a process that we term prERAD. prERAD does not appear to be based on the folding state or translocational competence of the substrate, but rather on its cytosolic occupancy time.

We next set out to test the protein motifs that engage prERAD. We reasoned that hydrophobic motifs might flag SRP-independent proteins as mislocalized, should they be exposed in the cytosol. Our previous work has indicated that although all SRP-independent proteins have hydrophobic targeting signals in the form of a signal sequence, the subgroup of GPI anchor proteins have an additional hydrophobic patch at their C-terminus, their GPI-anchoring sequence (Ast et al., 2013). As our model SRP-independent protein Gas1 is one such GPI-anchored protein, we set out to examine which of its hydrophobic domains direct Doa10-dependent degradation. To this end, we tagged GFP with either the signal sequence or GPI anchoring sequence of Gas1, and measured the stability of these fusion proteins in the cytosol. Although both GFP and the signal-sequence-bearing GFP </emph>remained stable in the cytosol (Fig. 2B; supplementary material Fig. S3A), regardless of the presence or absence of Doa10, the anchor-sequence-bearing GFP was degraded in the cytosol in a Doa10-dependent manner.

We next analyzed two additional fluorescently tagged GPI-anchored proteins, Ccw14 and Tos6, in WT and Δdoa10 cells (Fig. 2C). In control cells, Gas1, Ccw14 and Tos6 all localized to the cell surface, whereas in the absence of Doa10, all three proteins accumulated intracellularly in addition to their normal localization at the cell surface. In contrast to Gas1, Ccw14 and Tos6 appeared to aggregate in Δdoa10 cells, possibly owing to a dosage difference, as they were expressed from a multi-copy plasmid whereas Gas1 was expressed from a chromosomal copy. In addition, cycloheximide assays on YFP–Ccw14 confirmed the specific dependence on Doa10 and not on Hrd1 for cytosolic clearance (Fig. 2D). Thus, it appears that prERAD is indeed required for the elimination of pre-inserted GPI-anchored proteins. In contrast, when we analyzed the cytosolic clearance of a protein bearing only a signal sequence, CPY (also known as Prc1), we could not detect any pre-insertional degradation following a translational halt (supplementary material Fig. S3B). These findings indicate that the prERAD pathway recognizes the hydrophobic C-terminal patches displayed by GPI-anchored proteins.

The Cdc48 complex has previously been shown to generate the mechanical force needed to extract ERAD-C substrates from the membrane (Rabinovich et al., 2002). We rationalized that because prERAD deals with pre-inserted proteins there would be no need for Cdc48 function in this pathway. To test this hypothesis, we expressed YFP–Ccw14 in temperature-sensitive strains of the Cdc48 complex, namely cdc48^ts and ufd1^ts, grown at the restrictive temperature. Indeed, neither Cdc48 nor Ufd1 were required for the degradation of pre-inserted YFP–Ccw14, although their inactivation significantly attenuated the degradation of the misfolded luminal protein CPY* (Fig. 2D). The lack of dependence on the Cdc48 complex further demonstrates that these cytosolic proteins represent a bona fide pre-insertional protein population. Moreover, these results indicate that only a subset of proteins comprising the ERAD-C complex are required to carry out prERAD.

Finally, we set out to understand how the regulatory aspect of prERAD is maintained. A key force in any such pathway is a negative regulator, such as a deubiquitylation enzyme, ensuring that translocation substrates are not immediately degraded. Indeed, of the 20 DUB deletion strains present in our screen, two affected RFP–Gas1 localization, UBP1 and UBP11. We set out to further test the effect of these DUBs on RFP–Gas1 by overexpressing them (Fig. 3A). The overexpression of UBP11 had no effect on the localization of RFP–Gas1, which makes sense in light of its mitochondrial localization (supplementary material Fig. S4). The mitochondrial localization of Ubp11 raises the possibility that the RFP–Gas1 puncta generated in Δubp11 were a secondary effect of an aberrant mitochondrial quality control overloading the cell and would be interesting to follow up on. In contrast, the overexpression of the ER-localized UBP1 (Schmitz et al., 2005) phenocopied the loss of DOA10, resulting in accumulation of RFP–Gas1 on the ER surface. Indeed, the amounts of pre-inserted RFP–Gas1 were elevated in cells overexpressing UBP1, as could be measured by western blot (Fig. 3B). In fact, the overexpression of UBP1 elevated the fraction of pre-inserted:inserted RFP–Gas1 by over twofold, when compared to WT cells (P<0.025) (Fig. 3C). Thus, it appears that DOA10 and UBP1 mediate opposing forces in prERAD, working concurrently to fine-tune the amount of pre-inserted proteins at the ER surface.

Although SRP-independent targeting and translocation is an efficient process (Fig. 4), the newly identified cytosolic degradation mechanism, prERAD, ensures the clearance of pre-inserted proteins. prERAD is regulated by opposing forces of Doa10-dependent ubiquitylation and Ubp1-mediated deubiquitylation. It is tempting to suggest that the prERAD pathway monitors the extent of time a pre-inserted substrate has remained in the cytosol through ubiquitin chain length, along the lines of the calnexin–calreticulin cycle (Caramelo and Parodi, 2008). However, it is also possible that prERAD is activated when the pre-inserted protein concentration has accumulated above a given threshold.

Although Doa10 has been extensively implicated in the degradation of translocated malfolded ER proteins, there have also been reports of Doa10-dependent degradation of cytosolic and nuclear factors (Furth et al., 2011; Hochstrasser et al., 1991; Metzger et al., 2008; Ravid et al., 2006). Our findings add an additional group of substrates for this important E3 ligase, namely pre-inserted GPI-anchored proteins. It appears that Doa10 reacts to the cytosolic presence of the hydrophobic GPI-anchoring sequence because both various substrates bearing a GPI-anchoring sequence and the isolated sequence itself demonstrated Doa10-dependent degradation.

As SRP-independent substrates must remain in a loosely folded confirmation prior to their translocation (Tsukazaki et al., 2008; Van den Berg et al., 2004), pre-insertional degradation cannot hinge on the distinction between folding and misfolding. To overcome this challenge, it seems that immediate degradation is attenuated by the action of the ER-localized DUB Ubp1, whose overexpression rescues the pre-inserted form of RFP–Gas1. It should be noted that the exact role of Ubp1 in prERAD remains elusive – although it is appealing to suggest that Ubp1 is directly deubiquitylating pre-inserted substrates, it is also possible that Ubp1 serves to regulate other factors (Bernardi et al., 2013; Liu et al., 2014). This balance between E3 ubiquitin ligase and DUBs is not unique to our system, and has been recently shown to fine-tune degradation in the case of CD4 (Zhang et al., 2013). Furthermore, in the case of pre-inserted tail-anchored proteins, which bear a C-terminal transmembranal domain, antagonizing forces of protein ubiquitylation and protection appear to be at play in the cytosol (Leznicki and High, 2012). Little is known regarding Ubp1 – it has been shown to localize to both the ER and cytosol, and has been implicated in the stabilization of Ste6 and Golgi proteins (Poulsen et al., 2012; Schmitz et al., 2005). Thus, it seems that Ubp1 is involved in shaping and regulating the degradation of proteins within the secretory pathway.

The discovery of such a pre-insertional degradation pathway raises questions as to when and to what extent it is required. Although the translocation of SRP-independent substrates is fairly efficient, upon ER stress, Gas1 will undergo cytosolic degradation if the unfolded protein response is not effectively activated (Liu and Chang, 2008). Similar findings in higher eukaryotes have revealed that, during ER stress, the translocation of proteins bearing a mildly hydrophobic signal sequence is attenuated, and these proteins are cleared from the cytosol in a proteasome-dependent manner (Kang et al., 2006; Rutkowski et al., 2007). Furthermore, it has been previously shown that in unstressed mammalian systems, the efficiency of substrate translocation can range from 95% to 60% (Levine et al., 2005). This would indicate that, at any given time, the cell must identify and degrade a significant untranslocated cytosolic protein pool. Although secretory pathway degradation has long been thought to take place predominantly within the ER following translocation, this emerging understanding of prERAD highlights the multiple checkpoints that must be in place at every stage of secretory pathway function.

All yeast strains in this study are based on the BY4741 laboratory strain (Baker Brachmann et al., 1998). General laboratory strains and strains created in this study are listed in supplementary material Table S2. Unless otherwise stated, strains harboring a deletion in a specific open reading frame (ORF) were taken from the yeast deletion library (Giaever et al., 2002) and verified, whereas strains harboring a hypomorphic allele of an essential gene were taken from the DAmP (Decreased Abundance by mRNA Perturbation) library (Breslow et al., 2008). To construct the quality control mutation library, YeastMine was queried for genes that bear GO terms related to quality control (i.e. Ubiquitylation, Degradation, Proteasome and Microautophagy), resulting in a list of 210 genes found in supplementary material Table S1. In order to construct the SRP-independent query strain, RFP–Gas1, kindly provided by Howard Riezman (Department of Biochemistry, University of Geneva, Switzerland), was inserted into the URA3 locus, including sequences from ∼800 bp upstream and ∼300 bp downstream of the Gas1 ORF. Primers and plasmids utilized in this study are listed in supplementary material Tables S3 and S4, respectively. To study their effect, cdc48-2 and ufd1-1, kindly provided by Randy Hampton (Division of Biological Sciences, UC San Diego, CA) were grown at the restrictive temperature of 37°C for 4–6 hours prior to any experiment.

Cultures were grown at 30°C in either rich medium [1% Bacto-yeast extract (BD), 2% Bacto-peptone (BD) and 2% dextrose (Amresco)] or synthetic medium [0.67% yeast nitrogen base with ammonium sulfate and without amino acids (CondaPronadisa) and 2% dextrose (Amresco), containing the appropriate supplements for plasmid selection] (Sherman, 1991). When needed as selection markers, G418 (200 µg/ml, Calbiochem) or Nourseothricin (Nat) (200 µg/ml WERNER BioAgents) were added. In cases where G418 was required in a synthetic medium, yeast nitrogen base without ammonium sulfate (CondaPronadisa) was added and supplemented with monosodium glutamate (Sigma) as an alternative nitrogen source.

All genetic manipulations were performed using Synthetic Genetic Array (SGA) techniques to allow efficient introduction of RFP–Gas1 into systematic yeast libraries. SGA was performed as previously described (Cohen and Schuldiner, 2011; Tong and Boone, 2006). Briefly, using a RoToR bench top colony arrayer (Singer Instruments, UK) to manipulate libraries in 384-colony high-density formats, haploid strains from opposing mating types, each harboring a different genomic alteration, were mated on rich-medium plates. Diploid cells were selected on plates containing all selection markers found on both parent haploid strains. Sporulation was then induced by transferring cells to nitrogen-starvation plates. Haploid cells containing all desired mutations were selected for by transferring cells to plates containing all selection markers alongside the toxic amino acid derivatives Canavanine and Thialysine (Sigma-Aldrich) to select against remaining diploids. Each SGA procedure was validated by inspecting representative strains for the presence of RFP–Gas1 and for the correct genotype using PCR.

Microscopic screening was performed using an automated microscopy set-up as previously described (Breker et al., 2013; Cohen and Schuldiner, 2011). Briefly, liquid cultures were grown overnight in minimal medium, in a shaking incubator (LiCONiC Instruments) at 30°C. Cells were grown to logarithmic stage and were transferred onto glass-bottomed 384-well microscope plates (Matrical Bioscience) coated with Concanavalin A (Sigma-Aldrich). The strains were imaged with an automated inverted fluorescent microscopic ScanR system (Olympus), equipped with a cooled CCD camera. Images were acquired using a 60× air lens; excitation was at 555 nm (28-nm bandpass) and emission at 617 (73-nm bandpass). After acquisition, images were manually reviewed using the ScanR analysis program. Images were processed by the Adobe Photoshop CS3 program for slight contrast and brightness adjustments.

Manual microscopy was performed using an Olympus IX71 microscope controlled by the Delta Vision SoftWoRx 3.5.1 software with a ×60 oil lens. Images were captured by a Photometrics CoolSNAP HQ camera with excitation at 490 nm (20-nm bandpass) and emission at 528 (38-nm bandpass) for GFP, or excitation at 555 nm (28-nm bandpass) and emission at 617 (73-nm bandpass) for mCherry and RFP. Images were transferred to Adobe Photoshop CS3 for slight contrast and brightness adjustments.

Yeast protein extraction and cycloheximide assays were performed as previously described (Bhamidipati et al., 2005). In brief, mid-logarithmic yeast cells at an optical density at 600 nm (OD₆₀₀) of ∼2.5 were treated with 150 µg/ml of cycloheximide (Sigma) for the length of time indicated. Cells were then harvested, and re-suspended in 10% trichloroacetic acid (TCA) on ice for 20 minutes. Following this incubation, cells were centrifuged for 15 minutes at 20,000 g at 4°C, and the supernatant was removed. The pellet was washed in acetone, and resuspended in 100 µl of loading buffer (0.05 M Tris-HCl pH 6.8, 10% glycerol, 2% SDS, 5% β-mercaptoethanol, 0.1% Bromophenol Blue). 100 µl of glass beads (Scientific Industries Inc.) were added to the loading dye, and the samples were bead-beaten for 5 minutes at 4°C. The samples were then incubated at 95°C for 5 minutes, and centrifuged for 5 minutes at 6000 g at room temperature. 20 µl from the supernatant of the samples was resolved on 7.5 or 10% polyacrylamide gels, and probed with antibodies against RFP (ab62341, Abcam), GFP (ab290, Abcam), histone H3 (ab1791, Abcam), HA (MMS-101P, Covance), CPY (ab113685, Abcam) or PGK (459250, Life Technologies). Secondary antibodies consisted of goat anti-rabbit or anti-mouse Ig conjugated to IRDye800 or to IRDye680 (LI-COR Biosciences), and were scanned for infrared signal using the Odyssey Imaging System (LI-COR Biosciences). Protein amount was quantified using the Image Studio software (LI-COR Biosciences), and was tested for statistically significant variation using Student's t-test.

We wish to thank Howard Riezman (Department of Biochemistry, University of Geneva, Switzerland), Jeffrey Gerst (Department of Molecular Genetics, The Weizmann Institute of Science, Israel), Jodi Nunnari (College of Biological Sciences, UC Davis, CA) and Jeffrey Brodsky (Department of Biological Sciences, University of Pittsburgh, PA) for their generosity in sharing plasmids, and Randy Hampton (Division of Biological Sciences, UC San Diego, CA) for providing us with temperature-sensitive strains. We are grateful to Shachar Dagan (Department of Biological Chemistry, The Weizmann Institute of Science, Israel), Idan Frumkin (Department of Molecular Genetics, The Weizmann Institute of Science, Israel), Ofer Moldavski (Department of Molecular Genetics, The Weizmann Institute of Science, Israel) and Tommer Ravid (Department of Biological Chemistry, The Hebrew University of Jerusalem, Israel) for critical reading of the manuscript.