Quality control of inner nuclear membrane proteins by the Asi complex

See allHide authors and affiliations

Science  07 Nov 2014:
Vol. 346, Issue 6210, pp. 751-755
DOI: 10.1126/science.1255638

Trashing misfolded membrane proteins

Proteins move to and from the inner nuclear membrane (INM) from the rest of the endoplasmic reticulum through the nuclear pores. This movement is tightly controlled. Consequently, the INM accumulates a specific set of proteins required for a variety of functions, including chromosome organization and transcriptional control. But when INM proteins misfold, how are they eliminated? Foresti et al. addressed this question in yeast and found that a previously elusive branch of the endo–plasmic reticulum–associated degradation system was key (see the Perspective by Shao and Hegde).

Science, this issue p. 751; see also p. 701


Misfolded proteins in the endoplasmic reticulum (ER) are eliminated by a quality control system called ER-associated protein degradation (ERAD). However, it is unknown how misfolded proteins in the inner nuclear membrane (INM), a specialized ER subdomain, are degraded. We used a quantitative proteomics approach to reveal an ERAD branch required for INM protein quality control in yeast. This branch involved the integral membrane proteins Asi1, Asi2, and Asi3, which assembled into an Asi complex. Besides INM misfolded proteins, the Asi complex promoted the degradation of functional regulators of sterol biosynthesis. Asi-mediated ERAD was required for ER homeostasis, which suggests that spatial segregation of protein quality control systems contributes to ER function.

Misfolded proteins in the membrane and lumen of the endoplasmic reticulum (ER) are eliminated by ER-associated protein degradation (ERAD) (1, 2). In Saccharomyces cerevisiae, two ubiquitin ligase complexes, Hrd1 and Doa10, are involved in ERAD and show different specificity for misfolded proteins (15). Proteins with a misfolded domain in the cytosol (ERAD-C substrates) are targeted to degradation by the Doa10 complex, whereas proteins with misfolded domains in the membrane (ERAD-M) or lumen (ERAD-L) of the ER are ubiquitinated by the Hrd1 complex. The ubiquitin-conjugating (E2) enzyme Ubc7 is part of both complexes, and consequently ubc7Δ cells are defective in all ERAD branches. Besides misfolded proteins, ERAD is involved in the degradation of specific folded proteins, such as the sterol biosynthetic enzyme HMG-CoA reductase, in response to certain physiological stimuli (2).

To identify endogenous substrates of the different ERAD branches in S. cerevisiae, we compared the proteomes of doa10Δ, hrd1Δ, and ubc7Δ cells with the use of SILAC (stable isotope labeling by amino acids in culture) followed by quantitative proteomics (6) (table S1). This analysis revealed a set of proteins whose steady-state levels in doa10Δ or hrd1Δ were comparable to those in wild-type (wt) cells but were higher in ubc7Δ mutants relative to wt cells (table S2). Among these proteins were Erg11 (lanosterol 14-α-demethylase) and Nsg1, both involved in ergosterol synthesis (Fig. 1A and table S2). Given that Erg11 is an ER membrane protein and that sterol biosynthesis is regulated by ERAD (2), we tested whether Erg11 was an ERAD substrate. In wt cells, a plasmid-borne, functional Erg11 fused to hemagglutinin (Erg11-HA) was degraded with a half-life of ~60 min (Fig. 1B and fig. S1A). We observed similar kinetics of degradation in doa10Δ, hrd1Δ, or even doa10Δ hrd1Δ double mutant cells (Fig. 1B and fig. S1B). In agreement with the SILAC analysis, the degradation of Erg11-HA was delayed in ubc7Δ, indicating that this E2 works with an uncharacterized ubiquitin ligase (E3). The degradation of Erg11 was further reduced by simultaneous mutation of UBC7 and UBC4, indicating that these E2s have redundant roles in Erg11 ERAD (fig. S3).

Fig. 1 Erg11 is degraded by a novel ERAD branch.

(A) Erg11 abundance in the indicated mutants relative to wt cells, as detected by mass spectrometry upon SILAC labeling. Cells were grown in the presence of either heavy (wt) or light (mutants) l-lysine. Note that high steady-state levels of Erg11 result in low heavy/light ratios. (B) Degradation of Erg11-HA after inhibition of protein synthesis by cycloheximide in wt cells or in cells with the indicated deletions. Cell extracts were analyzed by SDS–polyacrylamide gel electrophoresis and blotting. The graph shows the average of at least three experiments; error bars represent the standard deviation of the mean. (C) Degradation of Erg11-HA and Erg1 in wt or temperature-sensitive cdc48-3 cells either at the permissive temperature (25°C) or after a 2-hour shift to the restrictive temperature (37°C). Samples were analyzed as in (B).

The ERAD of all known membrane-bound substrates requires the Cdc48 adenosine triphosphatase (ATPase). Inactivation of Cdc48 function in cells bearing the temperature-sensitive cdc48-3 allele showed impaired degradation of Erg11-HA (Fig. 1C). These findings suggest the existence of an ERAD branch independent of the known E3s Hrd1 and Doa10 but requiring Ubc7 and Cdc48.

Besides the canonical ERAD E3s, there is a third ER integral membrane E3. This is composed of two paralog proteins, Asi1 and Asi3, localized to the inner nuclear membrane (INM) (7, 8). The INM is connected to the rest of the ER membrane at nuclear pores, which restrict the exchange of proteins between the nucleus and cytoplasm and between the INM and the rest of the ER (9). Consequently, the INM defines a specialized ER subdomain enriched in proteins with nuclear functions. Indeed, ASI1, ASI3, and several ASI (amino acid signaling–independent) mutants were originally identified as defective in transcriptional repression of amino acid permeases (AAPs) (10). The effect of Asi1 and Asi3 on AAP expression appears to be achieved by controlling the binding of the transcription factors Stp1 and Stp2 to the promoter of AAP genes without affecting their localization or stability (7, 8). This repression of AAPs requires Asi2, another INM protein (8).

Deletion of ASI1, ASI2, or ASI3 (but not of other ASI genes) blocked the degradation of Erg11 and Nsg1 (Fig. 2A and figs. S1B, S2, and S4). Moreover, SILAC analysis showed higher levels of Erg11 in asi1Δ, asi2Δ, and asi3Δ (Fig. 2B). This effect was specific for Erg11, as the degradation (Fig. 2A) and steady-state levels (Fig. 2B) of Erg1, a Doa10 substrate (11), were mostly unaffected. Thus, the Asi and Doa10 branches of ERAD independently affect different steps in sterol biosynthesis.

Fig. 2 The INM proteins Asi1, Asi2, and Asi3 define a novel ERAD branch required for Erg11 degradation.

(A) Degradation of Erg11-HA and Erg1 in cells with the indicated genotype were analyzed as in Fig. 1B. (B) Erg11 and Erg1 abundance in the indicated mutants relative to wt cells, as detected by mass spectrometry upon SILAC labeling. (C) The degradation of endogenous Erg11-Flag in wt or asi1Δ cells bearing an empty plasmid (ϕ), a plasmid-borne Asi1wt, or a plasmid-borne Asi1C583-585S mutant. Samples were analyzed as in Fig. 1B.

The ubiquitin ligase activity of Asi1 is required for ERAD because a RING domain mutant, asi1(C583,585S), while expressed at wt levels (7), was defective in Erg11 degradation (Fig. 2C). Thus, Asi1, Asi2, and Asi3, together with the E2 Ubc7 and the Cdc48 ATPase, define a branch of ERAD involved in the degradation of Erg11 and Nsg1.

To characterize the mechanism of Erg11 ERAD, we evaluated its interaction with Asi1, Asi2, and Asi3, which were expressed as functional Flag-tagged proteins (fig. S5). Erg11-HA coprecipitated with Asi1, Asi2, and Asi3, whereas the Doa10 substrate Erg1 did not (Fig. 3A). Asi1, Asi2, and Asi3 comigrate upon sucrose gradient centrifugation (8). Moreover, Asi1 coprecipitates with Asi3 (8). We tested whether Asi2 was part of this complex. Affinity isolation of tagged Asi1, Asi2, or Asi3 led to coprecipitation of the other two proteins (Fig. 3B and table S3). Thus, Asi1, Asi2, and Asi3 assemble into the Asi complex. However, Asi2 was dispensable for the interaction between Asi1 and Asi3, which form the core of the Asi complex (Fig. 3C).

Fig. 3 Asi1, Asi2, and Asi3 assemble into the Asi complex and interact with the substrate Erg11-HA.

(A) Flag-tagged Asi proteins were purified from digitonin-solubilized membranes of cells expressing Erg11-HA and analyzed by blotting. Asterisk indicates the heavy chain used for immunoprecipitation (IP). (B) Flag-Asi1, HA-Asi2, and GFP-Asi3 were purified from detergent extracts and eluted proteins were analyzed as in (A). Under Lysate, I and U correspond to input and unbound, respectively. The input corresponds to 10% of the total extract used for IP. (C) Flag-Asi1 was purified from cells with the indicated genotype, and eluted proteins were analyzed as in (A).

All three ERAD E3s in yeast control the turnover of sterol biosynthetic enzymes, which suggests that sterol regulation is an ancient function of ERAD (2). The degradation of Hmg2 and Erg1 by Hrd1 and Doa10, respectively, is regulated by the levels of specific sterol metabolites as part of homeostatic feedback mechanisms (3, 11). However, manipulation of sterol intermediates did not have a major effect on Erg11 degradation (fig. S6). We then tested whether Asi-mediated ERAD was important to exclude Erg11 from the INM, where the Asi complex resides (7, 8). In wt cells, endogenous Erg11 fused to green fluorescent protein (GFP) was all over the ER (fig. S7A). In asi1Δ cells, Erg11 levels were higher, as expected, with the protein concentrated at the nuclear ER including both the inner and outer nuclear membranes (fig. S7, A to C). Furthermore, Erg11 overexpression was toxic to asi1Δ cells (fig. S7D). Thus, ERAD is involved in restricting sterol synthesis to specific ER subdomains.

Because the INM is not involved in protein biogenesis and consequently not prone to accumulate misfolded proteins, we wondered whether the Asi complex plays any role in protein quality control. Asi complex mutations did not affect the degradation of several ERAD substrates (fig. S8). Multiple E3s can contribute to the degradation of some ERAD substrates (12); therefore, we tested whether ASI1 mutations, when combined with hrd1Δ or doa10Δ, would further stabilize several substrates (table S4). Only Sec61-2, a translocon subunit with a G213D mutation, was further stabilized. The misfolding of this ERAD-M substrate is induced by high temperature (37°C) and consequently can be spatially and temporally uncoupled from Sec61-2 biogenesis. Whereas in hrd1Δ the degradation of Sec61-2 was slowed down as expected (4), in hrd1Δasi1Δ or hrd1Δasi3Δ mutants it was severely impaired (Fig. 4A and fig. S9). This is consistent with the observation that Sec61 can normally travel through the INM (13), although it functions only in ER membranes exposed to the cytoplasm (14). Compared to hrd1Δ single mutants, hrd1Δasi2Δ cells showed only a slight delay in Sec61-2 degradation (fig. S9B). In contrast, mutations in ASI1, ASI2, or ASI3 had indistinguishable effects on Erg11 and Nsg1 degradation (Fig. 2A and fig. S4). This different requirement for Asi components suggests that the complex can operate in distinct modes, perhaps according to the folding state of the substrate. Thus, the Asi complex contributes to the degradation of certain ERAD-M substrates, likely as they travel through the INM.

Fig. 4 Quality control of INM by the Asi complex is required for ER homeostasis.

(A) The degradation of sec61-2-HA analyzed in cells with the indicated genotype. Misfolding of Sec61-2 was induced by shifting the cells to 37°C for 40 min before cycloheximide addition. Samples were analyzed as in Fig. 1B. (B) The degradation of GFP-h2NLS-L-(Sec61-2) and derivatives was analyzed as in (A) but using antibodies to GFP. (C) Quantification of cycloheximide shut-off experiments performed as in (B). The average of at least three experiments was used; error bars represent SD of the mean. (D) Cells with the indicated genotype were grown for 3 days at 30°C either in synthetic complete (SC) media or SC supplemented with 5-fluoroorotic acid (5-FOA), which generates a toxic compound in URA3-expressing cells. Note that the triple mutant asi1Δire1Δhrd1Δ cannot grow on plates containing 5-FOA, indicating that loss of a functional Asi1 copy is incompatible with survival in this genetic background.

To directly test the role of the Asi complex in INM quality control, we manipulated the relative distribution of Sec61-2 between the INM and the rest of the ER. A long unstructured polypeptide segment appended to a high-affinity nuclear localization signal (NLS) targets several proteins to the INM (15) and can relocalize certain ER membrane proteins to the INM, such as Sec61 (15). We generated GFP-h2NLS-L-(Sec61-2), a Sec61-2 derivative fused at the N terminus to a high-affinity NLS, a long unstructured domain (L), and GFP. In wt cells, GFP-h2NLS-L-(Sec61-2) displayed an extremely short half-life, whereas in hrd1Δasi1Δ mutants it was strongly stabilized (Fig. 4B) and accumulated at the nuclear envelope (fig. S10A). Preventing GFP-h2NLS-L-(Sec61-2) from targeting the INM, either by removing the NLS or by shortening the linker (15), increased the dependency of its degradation on Hrd1 while decreasing the requirement for Asi1 (Fig. 4, B and C, and fig. S10A). The degradation of all chimeric constructs was due to the Sec61-2 mutation, as equivalent chimeras to wt Sec61 were stable (fig. S10B). Thus, the quality control of misfolded ER membrane proteins is spatially segregated, with the Asi complex acting on substrates localized to the INM.

To assess how general this role of the Asi complex in quality control is, we took advantage of the genetic interaction between ERAD and the unfolded protein response (UPR), a rectifying transcriptional program triggered by an increased ER load of misfolded proteins (16, 17). UPR mutants, such as ire1Δ, depend on ERAD to survive, particularly under stress conditions such as high temperature (16, 17). Asi complex mutations in UPR-deficient cells did not affect cell fitness (Fig. 4D), consistent with the observation that single Asi complex mutants efficiently degrade all tested ERAD substrates (fig. S8). In contrast, cells lacking Ire1, Hrd1, and an Asi complex core component were not viable (Fig. 4D and fig. S11). In agreement with the less prominent role of Asi2 in the degradation of misfolded Sec61-2 (fig. S9B), hrd1Δire1Δasi2Δ mutants had a milder growth defect (fig. S11). Thus, the Asi complex plays an important function in protein quality control and contributes to ER homeostasis.

Our results identify an ERAD branch defined by the Asi complex. Like the other branches, Asi mediates the degradation of both misfolded proteins and regulators of sterol biosynthesis. A distinctive feature of the Asi complex is its INM localization (fig. S12). Doa10 localizes both to the INM and the rest of the ER, and can target substrates in the nucleoplasm as well as in the cytoplasm (13). Because the ER encloses a continuous luminal space, the Hrd1 complex can target all misfolded luminal proteins. In contrast, INM misfolded proteins are not accessible to Hrd1, because it is excluded from this ER subdomain (13). Sequence analysis reveals obvious Asi complex homologs only in fungi. If a functional homolog of the Asi complex exists in higher eukaryotes, it might become critical in postmitotic cells that do not undergo frequent nuclear envelope disassembly and reassembly events, known to “refresh” the pool of INM proteins (9).

Asi1, Asi2, and Asi3 control AAP gene expression in response to amino acid availability, a process mediated by the transcription factors Stp1 and Stp2 (7, 8, 10). Even if the turnover of Stp1 or Stp2 is not affected by Asi complex mutations, Asi1 ubiquitin ligase activity is essential to control the response to amino acids (7). Intriguingly, the Cdc48 ATPase complex inhibits transcription in a proteolysis-independent manner by preventing ubiquitinated transcription factors from binding to their target sequences (18). Whether these functions of Cdc48 and Asi complexes in gene regulation are linked is unclear. However, an appealing possibility is that the ERAD machinery at the INM has been co-opted to perform additional functions, such as controlling transcription factor activity.

Supplementary Materials

Materials and Methods

Figs. S1 to S12

Tables S1 to S7

References (1925)

References and Notes

  1. Acknowledgments: Mass spectrometric measurements and data analysis were performed in the CRG/UPF Proteomics Unit (part of the ProteoRed, Instituto de Salud Carlos III). We thank M. Hochstrasser, M. Mendoza, and L. Veenhoff for reagents; C. Chiva and E. Sabido for help with the SILAC; R. Oliete for Erg11 and Nsg1 plasmids; and V. Goder, A. Ruggiano, and J. Valcarcel for critical reading of the manuscript. This work was supported by CRG, an International Early Career Award from the HHMI, the EMBO Young Investigator Program, grants from the Spanish MCCIN and ERC (P.C.), and a fellowship from the Spanish MCCIN (V.R.-V.).
View Abstract

Navigate This Article