Cdc48 and Ubx1 participate in a pathway associated with the inner nuclear membrane that governs Asi1 degradation

The nuclear envelope comprises the outer nuclear membrane (ONM) and the inner nuclear membrane (INM). These two membranes are continuous but functionally separated by nuclear pore complexes (NPCs) anchored in the bridging membranes. In addition to providing the main route of soluble transport in and out of the nucleus, NPCs provide a physical barrier that must be overcome for the movement of membrane proteins from the ONM to INM. Consequently, the nuclear membranes have unique properties and protein compositions. The endoplasmic reticulum (ER) extends directly from the ONM, and thus, the ONM and ER are quite similar. By contrast, the composition of the INM is not well established. However, proteins localized to the INM are known to participate in a variety of essential cellular processes, including maintenance of the nuclear architecture, transcriptional regulation, chromatin organization and DNA repair (Heessen and Fornerod, 2007; Mekhail and Moazed, 2010).

The diverse protein quality control pathways in eukaryotic cells share the common feature of orchestrating the attachment of small protein tags to substrate proteins that function as targeting motifs for degradation. The most commonly used tag is ubiquitin (Varshavsky, 2012). Short-lived proteins, misfolded or damaged proteins, or functional proteins subject to downregulation are ubiquitylated by the sequential action of an E1 ubiquitin-activating enzyme, E2 ubiquitin-conjugating enzymes (Ubc) and E3 ubiquitin ligases. The reiteration of these enzymatic events leads to the polyubiquitylation of substrate proteins. Polyubiquitylated substrate proteins are targeted to one of two proteolytic compartments, either the vacuole (analogous to lysosomes of mammalian cells) or proteasomes. Vacuoles primarily receive degradation substrates through vesicle-mediated transport (Bryant and Stevens, 1998; Rendueles and Wolf, 1988). Proteasomes are large multicomponent proteases that assemble in the cytoplasm and nucleoplasm (Ciechanover, 2007; Hershko and Ciechanover, 1998).

Two ER-associated degradation (ERAD) complexes in yeast, named after their core E3 ligases Hrd1 and Doa10, respectively, represent well-characterized protein quality control systems that target misfolded ER and cytoplasmic proteins for degradation by cytoplasmic proteasomes (Carvalho et al., 2006; Denic et al., 2006; Hiller et al., 1996; Sommer and Jentsch, 1993; Werner et al., 1996). Hrd1 primarily operates with Ubc7, whereas Doa10 functions with both Ubc6 and Ubc7 (Bays et al., 2001; Plemper et al., 1999; Rubenstein et al., 2012; Swanson et al., 2001). Ubc6 is a C-terminal-tail-anchored membrane protein, whereas Ubc7 is recruited to the membranes through interactions with Cue1, a type-1 transmembrane protein (Biederer et al., 1997). In cue1Δ mutants, Ubc7 is targeted for proteasomal degradation by the action of the Ufd4 E3 ubiquitin ligase (Ravid and Hochstrasser, 2007). Ubc6 is itself a substrate for Doa10- and Ubc7-dependent ubiquitylation (Kreft and Hochstrasser, 2011; Swanson et al., 2001; Walter et al., 2001).

In addition to endogenous resident INM proteins, many membrane proteins that normally reside in other intracellular locations are apparently able to gain access to the INM by leaking past nuclear pore complexes (Foresti et al., 2014; Khmelinskii et al., 2014; Popken et al., 2015). Both resident and mislocalized INM proteins are subject to turnover (Khmelinskii et al., 2014; Omnus and Ljungdahl, 2014). Currently, two quality control pathways are known to operate in association with the INM in yeast (Deng and Hochstrasser, 2006; Foresti et al., 2014; Khmelinskii et al., 2014). Analogous to the ERAD complexes, the INM-associated degradation (INMAD) complexes are defined by their core E3 ubiquitin ligases. In addition to its well-established role in ERAD, Doa10 is able to pass the barrier of NPCs and function together with Ubc6 and Ubc7 in the INM to ubiquitylate multiple nuclear substrates, targeting them to nuclear proteasomes for degradation (Boban et al., 2014; Deng and Hochstrasser, 2006; Furth et al., 2011; Ravid et al., 2006). The second, and most recently characterized, INMAD pathway is defined by the Asi1–Asi3 E3 ubiquitin ligase complex (Khmelinskii et al., 2014).

Asi1, Asi2 and Asi3 were originally identified based on their functions as negative regulators of the Ssy1–Ptr3–Ssy5 (SPS) nutrient-sensing pathway in yeast (Forsberg et al., 2001; Ljungdahl and Daignan-Fornier, 2012). Asi1 and Asi3 are homologous proteins: Asi1 (624 residues) and Asi3 (669 residues) are 31% identical and 42% similar in their sequences (Boban et al., 2006; Forsberg et al., 2001; Zargari et al., 2007). Both proteins have two equally sized domains – an N-terminal domain with five hydrophobic segments and a nucleoplasmically oriented hydrophilic C-terminal domain with a conserved Zn²⁺-binding RING motif. Asi1 and Asi3 function together as components of the heterodimeric E3 ubiquitin ligase complex; the RING motifs of Asi1 and Asi3 interact with Ubc7 and Ubc6, respectively (Khmelinskii et al., 2014). Asi2 appears to function as a specificity factor or as an adaptor protein required for targeting substrates for Asi1–Asi3-dependent ubiquitylation (Khmelinskii et al., 2014). All three Asi proteins are required to catalyze the ubiquitylation of Stp1 and Stp2, the effector transcription factors of the SPS sensing pathway, when they inappropriately gain access to the nucleus in the absence of amino-acid-mediated induction (Khmelinskii et al., 2014; Omnus and Ljungdahl, 2014).

The Asi1–Asi3 complex ubiquitylates multiple integral membrane proteins when they mislocalize to the nuclear compartment (Foresti et al., 2014; Khmelinskii et al., 2014). Consistently, cells require a functional Asi1–Asi3 E3 ubiquitin ligase complex when misfolded proteins accumulate in the ER, a condition that presumably results in enhanced levels of mislocalized proteins in the INM. Strains carrying mutations in HRD1 and IRE2, with compromised ERAD and unfolded-protein response pathways, respectively, exhibit impaired growth at elevated temperatures (Friedlander et al., 2000). The introduction of an asi1Δ allele into an hrd1Δ ire1Δ strain results in synthetic lethality (Foresti et al., 2014). These finding indicate that the protein quality control activity of the Asi1–Asi3 INMAD E3 ubiquitin ligase is essential for cells to maintain and safeguard the integrity of the nuclear compartment.

The processes governing the intrinsic stability of endogenous INM proteins are not well established. To fill the gap in knowledge, we have used the Asi proteins as model INM proteins to identify the cellular components that determine their stability. Here, we report that Asi1 is degraded rapidly within the nucleus, and we identify several cellular components required for its degradation. Intriguingly, Asi1 turnover occurs largely independent of the two known INMAD E3 ubiquitin ligase complexes, Asi1–Asi3 and Doa10. The results indicate that the INM possesses additional and previously unidentified protein quality control systems, a finding that has important implications for understanding the diverse functions associated with this vital intracellular membrane.

The stability of Asi1 and Asi3 carrying the indicated number of hemagglutinin (HA) epitope tags at their C-termini was determined in the presence of translation inhibitor cycloheximide (CHX) (Fig. 1). Protein levels were analyzed by immunoblotting for HA. Asi1 (Asi1–3HA, pPL1141) exhibited a half-life of approximately 30 min (Fig. 1A); chromosomal Asi1–3HA exhibited similar turnover rates (data not shown). Asi3–6HA (chromosomal) was more stable with a half-life of greater than 90 min (Fig. 1B). We examined whether the different rates of turnover affected the steady-state levels of Asi1 and Asi3 expressed from in-locus endogenously promoted genes encoding Asi1–6HA and Asi3–6HA (Fig. 1C). Consistent with a more-rapid turnover, the steady-state levels of Asi1–6HA (chromosomal) were significantly lower than those for Asi3–6HA (chromosomal). We then tested whether Asi3 and Asi2 affected Asi1 turnover. The data indicate that Asi1 turnover is independent of both Asi2 and Asi3 (Fig. 1D). Based on this and previous data examining the turnover of Asi2 (Boban et al., 2014), it is clear that each of the Asi proteins exhibits unique turnover kinetics.

The Asi proteins have been identified as negative regulators of the SPS sensing pathway that are required to suppress gene expression in the absence of amino-acid-mediated induction (Boban et al., 2006; Forsberg et al., 2001; Zargari et al., 2007). We tested the possibility that Asi1 becomes targeted for downregulation in an SPS-sensor-dependent manner. The stability of Asi1 was examined in cells lacking Ssy5, the endoprotease component of the SPS sensor. In ssy5Δ cells, Stp1 and Stp2 cannot be proteolytically activated, and hence, cells are insensitive to induction by extracellular amino acids. The rate of Asi1 degradation remained the same, independent of Ssy5 (Fig. 1E); a result consistent with our previous finding that Asi2 degradation occurs independently of SPS sensor signaling (Boban et al., 2014). Thus, the activity of the plasma-membrane-localized SPS sensor does not appear to influence the stability and turnover of the Asi proteins.

We sought to define whether Asi1 turnover occurs in the vacuole or is proteasome dependent. First, we examined the stability of Asi1 in a pep4-null mutant strain with greatly diminished capacity to degrade vacuolar substrates (Ammerer et al., 1986; Woolford et al., 1986). Our results indicate that Asi1 degradation occurs independently of Pep4 (Fig. 2A); Asi1 exhibited similar stability in PEP4 and pep4Δ strains, indicating that Asi1 is not degraded in the vacuole. Next, we assessed whether Asi1 is turned over in a proteasome-dependent manner. Asi1 stability was assessed in cim3-1 mutants carrying temperature-sensitive mutations in RPT6, which encodes one of the ATPases present in the heterohexameric Rpt ring of the 19S regulatory particle of the proteasome (Ghislain et al., 1993; Schork et al., 1995). At non-permissive temperatures, the proteasomes of cim3-1 mutants exhibit greatly impaired degradative capacity. Consistent with being a proteasome substrate, Asi1 exhibited clearly enhanced stability when expressed in a cim3-1 mutant growing at the non-permissive temperature (Fig. 2B).

Most proteasomal substrates commonly require modification by polyubiquitin chains that act as a targeting signal recognized by the subunits of the 19S regulatory particle and additional shuttling receptors associated with the holo 26S proteasome (Chau et al., 1989; Ravid and Hochstrasser, 2008; Thrower et al., 2000). To examine the ubiquitylation status of Asi1, we performed immunoprecipitation experiments in cells overexpressing ubiquitin. Immunoblots for ubiquitin showed that the immunoprecipitated HA-tagged Asi1 is polyubiquitylated (Fig. 2C). Taken together, these observations suggest that the ubiquitin–proteasome system (UPS) dictates the stability of Asi1.

Immuno-electron microscopy studies have clearly shown that the vast majority of Asi1 (>80%) is localized to the INM (Boban et al., 2006). We addressed whether the rapid turnover of Asi1 is the consequence of a degradative pathway operating inside the nucleus. The stability of Asi1 was assessed in strains carrying the temperature sensitive sts1-2 mutation. Sts1 interacts with the nuclear importin Srp1 to mediate the transport of inactive precursors of the proteasomal complexes into the nucleus. Consistently, Sts1 is required for proper assembly of nucleus-localized proteasomes (Chen et al., 2011; Romero-Perez et al., 2007). In sts1-2 mutant cells shifted to the restrictive temperature, proteasomal activity in the nucleus gradually becomes depleted, resulting in the stabilization of nuclear substrates and the increased turnover of cytosolic substrates (Chen et al., 2011). Asi1 exhibited enhanced stability in an sts1-2 mutant incubated at the non-permissive temperature (Fig. 2D). The results indicate that the bulk of Asi1 is degraded by proteasomes within the nucleus.

To identify UPS components participating in the targeting of Asi1 for degradation, we examined the involvement of the E2 ubiquitin-conjugating enzymes Ubc6 and Ubc7. The stability of Asi1 was slightly enhanced in ubc6Δ cells and clearly enhanced in ubc7Δ cells (Fig. 3A). An additive effect on Asi1 stability was observed in ubc6Δ ubc7Δ double-mutant cells, consistent with a partial functional overlap. The involvement of Ubc6 and Ubc7 in the turnover of Asi1 is reflected by the enhanced steady-state levels of Asi1 in both ubc6Δ and ubc7Δ mutants (Fig. 3B). The deletion of UBC7 did not affect ASI1 mRNA expression (Fig. 3C), indicating that the increased level of Asi1 in ubc7Δ mutants is due to decreased protein turnover. The clear and distinct requirement for Ubc7 is consistent with its pronounced role in multiple protein quality control systems operating at the INM – e.g. the Asi1–Asi3-dependent degradation of soluble mislocalized transcription factors Stp1 and Stp2 (Khmelinskii et al., 2014), and the Doa10-dependent degradation of MATα2 repressor (Deng and Hochstrasser, 2006) and Asi2 (Boban et al., 2014).

Consistent with Asi1 being a resident of the INM, the turnover of Asi1 proceeded independently of the ERAD E3 ubiquitin ligase Hrd1 (Fig. 4A, upper panel). The stability of Asi1 was slightly enhanced in doa10Δ (Fig. 4A, middle panel) and hrd1Δ doa10Δ cells (Fig. 4A, lower panel). To more rigorously test the involvement of Doa10, we constructed a quadruple-mutant strain that lacked ASI1, ASI2, ASI3 and DOA10, and analyzed the turnover of Asi1 and Asi2. The data indicate that Asi1 degradation occurs rapidly and independently of functional Asi1–Asi3 and Doa10 E3 ubiquitin ligase complexes (Fig. 4B and D). By contrast, and consistent with our previous studies, Asi2 turnover is clearly dependent on Doa10 (Fig. 4C and E) (Boban et al., 2014). Taken together, these results demonstrate that the stability of Asi1 and Asi2 is governed by the action of different E3 ubiquitin ligases, and that the E3 ubiquitin ligase controlling Asi1 turnover is not one of the two characterized INMAD ligases. Consequently, additional E3 ubiquitin ligases must operate in the INM to target Asi1 for degradation.

To identify additional INMAD components that contribute to the rapid turnover of Asi1, we assessed the stability of Asi1 in a panel of 86 strains (Table 1, generously provided by Mark Hochstrasser, Yale University) carrying loss-of-function mutations in genes encoding known components that operate in multiple degradation pathways. A plasmid carrying Asi1–3HA (pPL1141) was introduced into each strain, and the stability of Asi1 was assessed 60 min after CHX was added. The mutant collection included 39 E3 ubiquitin ligases, including RING, HECT and PHD proteins, and F-box proteins. Compared to wild-type control strains, with the exception of Doa10, none of these candidates showed any significant effect on turnover of Asi1; consistent with the data shown in Fig. 4A, the doa10Δ mutant reproducibly exhibited slightly higher levels of Asi1 remaining at the 60 min time point. Notably, Asi1 turnover was independent of San1, an important non-membrane E3 ubiquitin ligase in the nucleus (Rosenbaum et al., 2011). Although the lack of San1 involvement was expected – i.e. Asi1 does not fit the profile of San1 substrates – we examined Asi1 turnover in asi2Δ san1Δ hrd1Δ doa10Δ and asi1Δ san1Δ hrd1Δ doa10Δ quadruple-mutant strains. Asi1 turnover occurred unimpeded (data not shown), providing independent confirmation that San1 does not recognize Asi1 as a ubiquitylation substrate.

Additionally, all of the 13 Ubc proteins were tested for their influence on Asi1 turnover, including a ubc4Δ ubc5Δ double-mutant strain. The experiments confirmed the strong involvement of Ubc7 (Fig. 5A), again providing independent confirmation of results from our more-directed approach (Fig. 3). Moreover, 35 accessory proteins of the UPS were examined. Cue1 was found to participate in the degradation of Asi1 (Fig. 5A). Cue1 is an integral protein that tethers Ubc7 to the membranes, activates the E2 ubiquitin-conjugating enzyme and prevents its degradation (Bazirgan and Hampton, 2008; Biederer et al., 1997; Kostova et al., 2009). Also, consistent with Asi1 being a proteasomal substrate (Fig. 2B), Asi1 is significantly stabilized in the strain lacking RPN4. RPN4 encodes a transcription factor that regulates the expression of proteasome subunit genes (Mannhaupt et al., 1999). Rpn4 function is well characterized and known to be required to maintain normal levels of intracellular proteolysis (Mannhaupt et al., 1999; Xie and Varshavsky, 2001).

Of the remaining mutants tested, we found that inactivation of the AAA ATPase Cdc48 and Ubx1(Shp1) strongly affected the stability of Asi1 (Fig. 5B and C, respectively). Cdc48 acts as a key regulator of the ubiquitin–proteasome pathway, interacting with a large number of interaction partners and co-factors that regulate its subcellular localization and substrate specificity (Buchberger, 2013; Meyer et al., 2012). The function of Ubx1 in yeast is not well established; however, Ubx2 has been implicated as an adaptor component of Cdc48 complexes involved in a variety of cellular processes (Bohm and Buchberger, 2013; Krick et al., 2010; Schuberth and Buchberger, 2005; Schuberth et al., 2004). The deletion of UBX1 did not affect ASI1 mRNA expression (Fig. 5C, right panel), indicating that the increased level of Asi1 in ubx1Δ mutants is due to decreased protein turnover. Our data demonstrate a new role for Cdc48 and Ubx1 as components of protein quality control pathway functioning at the INM.

The INM is the least-characterized membrane in eukaryotic cells, particularly with respect to the parameters affecting turnover of its integral membrane components (Boban and Foisner, 2016). To obtain mechanistic insight into the quality control systems operating in association with the INM, we have investigated the turnover of three native well-characterized and fully functional INM proteins as paradigmatic INMAD substrates – i.e. Asi1, Asi2 and Asi3 (Boban et al., 2006; Forsberg et al., 2001; Zargari et al., 2007). The Asi proteins exhibit distinct turnover rates – Asi1 t_1/2≤30 min, Asi2 t_1/2≈45 min and Asi3 t_1/2>90 min (Fig. 1, and see Boban et al., 2014). By examining the mechanisms that control Asi1 turnover, we have found that Asi1 is ubiquitylated in a unique quality control pathway associated with the INM (Figs 2, 3 and 5). Strikingly, Asi1 turnover occurs independently of the known INMAD pathways (Figs 1 and 4 and Table 1), indicating that the INM possesses additional quality control systems (Fig. 6).

The clear difference in rates of Asi1 and Asi3 turnover and the finding that their turnover occurs largely independent of each other are noteworthy. These proteins are homologous, share similar structural features – including a C-terminal RING domain – co-purify and function together as subunits of the Asi1–Asi3 E3 ubiquitin ligase complex (Foresti et al., 2014; Khmelinskii et al., 2014; Zargari et al., 2007). Thus, although the Asi1–Asi3 E3 ubiquitin ligase operates together with the E2 ubiquitin-conjugating enzymes Ubc6 and Ubc7, and plays an essential role in clearing the nuclear compartment of mislocalized soluble and integral membrane proteins (Khmelinskii et al., 2014), this heterodimeric E3 ubiquitin ligase complex does not autoregulate the stability of its individual subunits. Our finding that the steady-state levels of Asi1 are significantly lower than those of Asi3 (Fig. 1C) suggests that the levels of Asi1 are limiting in the formation of the catalytically active Asi1–Asi3 E3 ubiquitin ligase complexes.

RING-deficient mutant forms of Asi1 that lack the ability to bind to Zn²⁺ are inactive. The asi1-21HA and asi1-22HA alleles, encoding mutant proteins with two and four of the Zn²⁺-coordinating cysteine residues replaced with serine, respectively, precisely phenocopy the asi1Δ phenotype (Boban et al., 2006). Thus, an intact RING domain of Asi1 is an essential feature of the Asi1–Asi3 E3 ubiquitin ligase. We examined the turnover of the RING-mutant proteins and found that both are rapidly degraded in an asi1Δ mutant. Moreover, in control experiments, we found that their degradation occurred independently of Ubc7 (data not shown). Thus, the quality control pathway that controls the stability of the RING-deficient mutant proteins does not regulate the turnover of native and functional Asi1. Consequently, although consistent with our conclusion that Asi1 does not autoubiquitylate, the data regarding the turnover of the RING-mutant forms of Asi1 are difficult to properly interpret and unlikely to be relevant to furthering our understanding regarding the turnover of native Asi1.

In an effort to identify factors participating in native Asi1 turnover, we screened a collection of 86 mutant strains carrying defects in components of known protein degradation systems. This screen identified Ubc7, Cue1, Cdc48 and Ubx1 as being required for Asi1 turnover but failed to identify a primary E3 ubiquitin ligase (Table 1, Fig. 5). This negative result could be the consequence of the involvement of an essential ligase or redundancy – perhaps multiple ligases contribute to regulating Asi1 stability. Interestingly, the two previously identified INMAD quality control systems and the new system we have shown to exist in this work share common components – i.e. the E2 ubiquitin-conjugating enzyme Ubc7 and its co-factor Cue1. Ubc7 has multiple roles in INM protein quality control. Together with the Asi1–Asi3 E3 ubiquitin ligase complex, it mediates degradation of the soluble transcription factors Stp1 and Stp2 when these factors ectopically enter the nucleus (Khmelinskii et al., 2014). Also, Ubc7 works with Doa10 to target transcription factor Matα2 (Chen et al., 1993; Deng and Hochstrasser, 2006) and Asi2 (Boban et al., 2014) for degradation.

The AAA ATPase Cdc48 participates in distinct cellular processes, acting as a key regulator of the ubiquitin–proteasome pathway (Buchberger, 2013; Meyer et al., 2012). Ubiquitylated membrane proteins that have been tagged for degradation by proteasomes must be extracted, or dislocated, from the membrane. Although proteasomes could contribute to this process, Cdc48 is thought to be the most important facilitator of the dislocation of ubiquitylated substrate proteins. Cdc48 is directed to different degradative pathways within the cell, where its activity is regulated and is dependent on the action of adaptor proteins – i.e. UBX-domain-containing proteins (Buchberger et al., 2015; Neuber et al., 2005; Schuberth and Buchberger, 2008). In yeast, there are seven UBX proteins. Ubx2, the best-characterized UBX protein, resides in the ER where it is thought to recognize polyubiquitylated substrates exposed on the cytosolic face of the ER and to be in complex with Cdc48 (Cdc48^Ubx2), facilitating dislocation of the ubiquitylated proteins from the ER (Kolawa et al., 2013; Neuber et al., 2005; Schuberth and Buchberger, 2005). At the ER, Ufd1 and Npl4 are associated with Cdc48^Ubx2 (Jentsch and Rumpf, 2007). We found that Ubx1, the least-characterized UBX protein, is stringently required for Asi1 turnover. Analogous to Ubx2, Ubx1 could function by recruiting Cdc48 to the INM, forming a complex that facilitates the membrane dislocation of ubiquitylated Asi1 and potentially other ubiquitylated INM proteins, a requisite for their degradation by nuclear proteasomes.

Ubx1 has been implicated in regulating the activity of the protein phosphatase 1 (Glc7) during cell cycle progression, participating in the ubiquitin-dependent proteasomal degradation of ubiquitylated phosphorylated β-galactosidase and in Atg8-dependent piecemeal microautophagy of the nucleus (PMN) (Bohm and Buchberger, 2013; Krick et al., 2010; Schuberth et al., 2004). PMN is a process whereby parts of the nucleus are degraded by the vacuole (Pan et al., 2000; Roberts et al., 2003). Under nitrogen starvation, the yeast INM protein Src1 and ER and ONM protein Hmg1 are directed for vacuolar degradation through autophagic sequestration into double-membraned vesicles derived from the nuclear envelope (Mochida et al., 2015). Thus, perhaps under nutrient limitation or other cell stress, Asi1 is targeted through PMN to the vacuole in a Ubx1- and Atg8-dependent manner. However, in the absence of nutrient stress, as in the experimental conditions used in our studies, Ubx1 is likely to function together with Cdc48 to facilitate membrane dislocation of ubiquitylated substrates, as described in the preceding paragraph.

Based on cumulating information, the Asi1–Asi3 and Doa10 E3 ubiquitin ligase complexes clearly function as non-redundant quality control pathways within the INM. This notion is supported by a number of observations. The spectrum of proteins stabilized by mutations inactivating Asi1 and Asi3 exhibit extensive overlap (Khmelinskii et al., 2014), a finding consistent with the obligate functional interdependence in the Asi1–Asi3 E3 ubiquitin ligase complex. By contrast, there is little overlap with proteins that are stabilized by inactivating Doa10. Consequently, these E3 ubiquitin ligases have distinct substrate specificities. Also, there are no observable additive effects by combining asi1Δ with doa10Δ hrd1Δ on the stability of proteins exhibiting Asi1–Asi3 dependence; substrate proteins exhibit indistinguishable rates of turnover in asi1Δ and in the triple mutant asi1Δ hrd1Δ doa10Δ (Khmelinskii et al., 2014). Finally, null alleles of DOA10 do not induce constitutive SPS-sensor-regulated gene expression; apparently, Doa10 does not recognize the N-terminal nuclear degrons of Stp1 and Stp2 (Khmelinskii et al., 2014; Omnus and Ljungdahl, 2014). Thus, the characterized INMAD pathways are clearly separate from each other and also distinct from the well-characterized ERAD pathways defined by the Hrd1 and Doa10 E3 ubiquitin ligases that target ubiquitylated proteins for degradation by proteasomes in the cytoplasm (Fig. 6).

The multiple INMAD pathways controlling the turnover of INM proteins target their ubiquitylated substrates to nucleus-localized proteasomes (Fig. 3, and see Boban et al., 2014). Proteasomes account for 80–90% of protein turnover in rapidly growing yeast cells (Enenkel, 2014). The nucleus is enriched with proteasomes, indicating that nuclear protein quality control is important in proliferating yeast cells as it secures DNA stability and repair, DNA replication, chromosome segregation, transcription and gene silencing (Chowdhury and Enenkel, 2015; Wojcik and DeMartino, 2003). In this context, cytosolic proteasomes are indispensable for yeast growth, whereas nuclear proteasomes are essential for viability (Tsuchiya et al., 2013). Additionally, there is growing evidence that diverse cytosolic proteins are translocated to the nucleus and are degraded by nuclear proteasomes through chaperone-dependent delivery and ubiquitylation by the nuclear E3 ubiquitin ligase San1 (Guerriero et al., 2013; Heck et al., 2010; Park et al., 2013; Prasad et al., 2010). Hence, the discovery that multiple INMAD pathways exist bolsters the notion that the nuclear compartment is well equipped as a major site for protein quality control.

Extremely little is known regarding the turnover of INM proteins. Thus, our findings that multiple quality control INMAD pathways exist represent new and important insights into the biology of eukaryotic cells. In contrast to metazoan cells, yeast cells exhibit closed mitosis, meaning that the nuclear envelope remains intact throughout the cell cycle (Anderson and Hetzer, 2008). Thus, the yeast INMAD pathways enable cells to maintain the integrity of the INM by removing genuine INM proteins that become damaged or aged and by eliminating proteins that mislocalize to the INM. Perhaps one of the most important implications of recent findings regarding the biological function of INMAD pathways is that many membrane proteins that primarily localize elsewhere slip past the barrier function of nuclear pore complexes and gain access to the INM (Foresti et al., 2014; Khmelinskii et al., 2014; Popken et al., 2015). These unexpected observations undoubtedly provide the biological requirement for multiple quality control systems associated with this important membrane. Apparently, INMAD pathways provide cells with solutions to the inherently difficult challenge of creating fail-safe targeting mechanisms to ensure the proper composition of the nuclear compartment. Although it is presently unclear whether similar INMAD quality control pathways exist outside of yeast, our findings will hopefully encourage others to pursue the possibility that functional INMAD pathways exist and operate analogously in other organisms. Clearly, safeguarding the integrity of the INM must be a crucial and common concern shared by most, if not all, eukaryotic cells. The importance of membrane-associated protein degradation mechanisms and the large diversity of integral membrane RING domain proteins in mammalian cells suggest that such research will lead to fruitful findings.

Standard yeast growth media were used, including yeast-extract-peptone-dextrose (YPD) medium and ammonia-based synthetic complete dextrose (SC) and ammonia-based synthetic minimal dextrose (SD) medium, supplemented as required to enable the growth of auxotrophic strains, prepared as described previously (Andréasson and Ljungdahl, 2002). Antibiotic selections were made on solid YPD containing 200 mg/l G418 or 100 mg/l clonNAT.

Saccharomyces cerevisiae strains used in this study (Table S1) are isogenic descendants of the S288c-derived strain AA255/PLY115, except for sts1-2, provided by Kiran Madura (Robert Wood Johnson Medical School, Rutgers University, NJ), and cdc48-3 (W303-1B background) provided by Dr Mark Hochstrasser (Yale University, CT). The EUROSCARF deletion library used in our screen, and generously provided by Dr Hochstrasser, consists of BY4741-derived strains that are considered to be isogenic to the S288c genetic background. The temperature-sensitive strains cdc34-1, cdc53-1 and skp1-11 provided by Dr Bruno André (Université Libre de Bruxelles, Belgium) are isogenic descendants of the W303-1B and MT235 strain background. The plasmids used in this study are listed in Table S1. The sequences of mutagenic oligonucleotides and PCR primers for homologous recombination are available upon request.

Yeast strains were grown at 30°C, unless indicated otherwise, to OD₆₀₀ of 0.8–1. CHX (Sigma-Aldrich) was added to cultures (final concentration of 100 μg/ml), and samples were harvested at indicated time points. Total protein extracts were prepared under denaturing conditions using NaOH and trichloroacetic acid, as described previously (Boban et al., 2014). The following antibodies were used for immunoblot analysis: horseradish peroxidase (HRP)-conjugated rat monoclonal anti-HA (3F10, 1:5000; Roche Applied Science), mouse monoclonal anti-Pgk1 (22C5, 1:5000; Molecular Probes), mouse monoclonal anti-Dpm1 (5C5, 1:1000; Molecular Probes) and HRP-conjugated goat anti-mouse (1:5000; Roche Applied Science) antibodies. Enhanced chemiluminescence (SuperSignal West Dura Extended Duration Substrate, Thermo Scientific) signals were detected using a BIO-RAD ChemiDoc XRS+ system and quantified with BIO-RAD image Lab 3.0 build 11. The sum of the signal intensity of Asi1 bands was normalized to the signal of the stable protein control Dpm1 or Pgk1. In each experiment, average values, s.d. and P-values based on three independent samples were calculated.

Yeast lysates from strains expressing Asi1–3HA (pPL1136) or Asi1 (pAZ013) and carrying ubiquitin-overexpression plasmid pPL1307 (PCUP1-6HIS-UBI) were immunoprecipitated using anti-HA affinity matrix (rat monoclonal 3F10, Roche Applied Science), as described previously (Boban et al., 2014). Immunoblot analysis was performed using mouse monoclonal anti-ubiquitin (P4D1, 1:0000; Santa Cruz), HRP-conjugated rat monoclonal anti-HA (3F10, 1:5000; Roche Applied Science), mouse monoclonal anti-Pgk1 (22C5, 1:5000; Molecular Probes) and HRP-conjugated goat anti-mouse (1:5000; Roche Applied Science) antibodies.

Exponentially growing cells in SC medium (≈10⁷ cells) carrying pPL1141 (ASI1-3HA) were harvested by centrifugation. RNA was isolated using the RiboPure Yeast Kit (Ambion by Life Technologies) and treated with Turbo-DNase (Ambion by Life Technologies). One microgram of RNA was used for complementary DNA synthesis with oligo (dT)12-19 (Invitrogen by Life Technologies) using SuperScript III Reverse Transcriptase (Life Technologies). Quantitative reverse transcriptase PCR (qRT-PCR) reactions were prepared using Kapa SybrFast qPCR Master Mix (Kapa Biosystems). The absence of DNA contamination was confirmed using minus reverse transcriptase samples. The cDNA preparations were diluted 1:40, and 5 μl was used in a reaction volume of 20 μl. The levels of gene expression in three biological replicates were determined in three separate amplifications with triplicate technical replicates of each of the two genes analyzed using the comparative ΔC_T method (RotorGene 6000, Corbett Life Science). Relative levels of ASI1 messenger RNA were normalized with respect to the levels of the invariant reference gene TAF10. The following primer pairs were used: ASI1-3HA fwd 5′-GGCCCTGCCGATGTTTTGCC-3′ and ASI1-3HA rev 5′-GGGACGTCATAGGGATAGCC-3′; TAF10 fwd 5′-ATATTCCAGGATCAGGTCTTCCGTAGC-3′ and TAF10 rev 5′-CAACAACAACATCAACAGAATGAGAAGACTAC-3′.

Portions of the data have been discussed in the Ph.D. thesis of Marina Pantazopoulou (Protein Quality Control at the Inner Nuclear Membrane – The Asi complex in Saccharomyces cerevisiae, 2016, Stockholm University, Sweden). We thank Mark Hochstrasser, Bruno André and Kiran Madura for plasmids and yeast strains. Our Ljungdahl laboratory colleagues Antonio Martins, Andreas Ring and Kicki Ryman are gratefully acknowledged for contributions to the revision process.