The Ubiquitin–Proteasome System of Saccharomyces cerevisiae
Daniel Finley, Helle D. Ulrich, Thomas Sommer, Peter Kaiser

Abstract

Protein modifications provide cells with exquisite temporal and spatial control of protein function. Ubiquitin is among the most important modifiers, serving both to target hundreds of proteins for rapid degradation by the proteasome, and as a dynamic signaling agent that regulates the function of covalently bound proteins. The diverse effects of ubiquitylation reflect the assembly of structurally distinct ubiquitin chains on target proteins. The resulting ubiquitin code is interpreted by an extensive family of ubiquitin receptors. Here we review the components of this regulatory network and its effects throughout the cell.

THE modification of proteins by the covalent attachment of ubiquitin is a regulatory process whose influence is felt throughout the cell in all eukaryotes. Ubiquitylation targets proteins to the proteasome to be degraded, a process that dynamically sculpts the proteome, with hundreds of yeast proteins being rapidly and selectively degraded (Belle et al. 2006). However, many ubiquitin modifications act through nonproteolytic mechanisms, such as in DNA repair, chromatin dynamics, mRNA export, the extraction of proteins from multisubunit complexes, and the trafficking of membrane proteins. These differing fates of ubiquitylated proteins are controlled by the nature of the ubiquitin modification; a single ubiquitin is often insufficient to target the substrate to the proteasome, whereas substrates modified by a polyubiquitin chain can be preferentially targeted to the proteasome. Thus, the degree of processivity of a ubiquitin ligase is crucial in determining the consequences of the modification.

Ubiquitin is usually attached to protein lysine residues. Ubiquitin itself has seven lysines, all of which can be conjugated to a second ubiquitin molecule (Peng et al. 2003). This allows for the construction of topologically distinct polyubiquitin chains and a diversity of signaling modes beyond those associated with chain length. For example, Lys48-linked chains are critical for protein degradation, whereas Lys63-linked chains are used in DNA repair and the trafficking of membrane proteins. This flexibility in signaling is fundamental to the ubiquitin system and may account for the pervasive influence of ubiquitin in cellular regulatory pathways. Ubiquitin receptors, many of which display specificity or preference for ubiquitin chain linkage type or length, play a key role in decoding the signals embedded in the structure of ubiquitin chains (Dikic et al. 2009).

The ubiquitin–proteasome and autophagy systems represent the principal modes of breakdown of intracellular proteins in eukaryotes. Autophagy, the hydrolysis of intracellular proteins within the vacuole (Nakatogawa et al. 2009), will be described in another article from this series. Autophagy is responsible for the selective breakdown of whole organelles, such as mitochondria and peroxisomes, as well as at least one large protein complex, the ribosome, but in Saccharomyces cerevisiae, autophagy is otherwise thought to be nonselective as compared to proteasomal degradation. Rapid protein breakdown within the cytoplasm and nuclei of eukaryotic cells, as exemplified by substrates such as cyclins, is generally mediated by the proteasome.

The paradigm of ubiquitylation, in which a small protein with a β-grasp fold covalently modifies other molecules via its C-terminal glycine, extends to several ubiquitin-like proteins (UBLs), each with a dedicated conjugation machinery: Smt3 (this modification is known as SUMOylation), Rub1 (NEDDylation), Urm1 (urmylation), and the autophagy factors Atg8 and Atg12 (Hochstrasser 2009; Inoue and Klionsky 2010). The target of conjugation is not always a protein; Atg8 is conjugated to phosphatidylethanolamine, thus converting it from a soluble to a membrane-bound protein, and Urm1 acts both as a protein modifier and as a sulfur carrier to support thiolation of tRNAs. Because of space limitations we will discuss these modification pathways below only as they relate to ubiquitylation itself. Space restrictions also prevent us from giving a complete description of the ubiquitin–proteasome system in yeast, and we apologize for any gaps in coverage.

We begin this review by describing the various components of the ubiquitin–proteasome system, including the conjugation cascade, the deubiquitylating enzymes, the proteasome, and the ubiquitin-selective chaperone Cdc48. The nature of substrate recognition in this pathway is then discussed, with special emphasis on the selective modification and degradation of defective proteins. Finally, we consider the many specialized functions of ubiquitylation in the nucleus, endomembrane system, and other subcellular sites.

Ubiquitin–Protein Conjugation

Ubiquitylation reaction

Ubiquitin is typically linked to substrates through an isopeptide bond between the ε-amino group of a substrate lysine residue and the carboxyl terminus of ubiquitin. Ubiquitin conjugation involves the E1–E2–E3 cascade of enzymes (Figure 1A). The reaction is initiated by the ubiquitin-activating enzyme E1 (Uba1), which forms a high-energy thioester bond with the main-chain carboxyl group of the terminal glycine residue of ubiquitin. This step consumes ATP in forming a ubiquitin–adenylate intermediate with subsequent release of AMP and pyrophosphate. Activated ubiquitin is transferred to one of the ubiquitin-conjugating enzymes (E2 or Ubc enzymes) by transesterification. Finally, E3 enzymes (ubiquitin ligases) catalyze the formation of isopeptide bonds between ε-amino groups of lysine residues in substrate proteins and the activated carboxyl group of ubiquitin (Deshaies and Joazeiro 2009; Varshavsky 2012). Through successive rounds of conjugation, polyubiquitin chains are synthesized. Lysines are by far the most common acceptor sites but ubiquitin ligation to the N-terminal amino group in higher eukaryotes (Bloom et al. 2003; Ben-Saadon et al. 2004; Kirisako et al. 2006; Rahighi et al. 2009; Tokunaga et al. 2009), as well as to serine, threonine, or cysteine in both yeast and mammals, has been observed (Cadwell and Coscoy 2005; Ravid and Hochstrasser 2007; Wang et al. 2007; Shimizu et al. 2010). The conjugation machinery shows hierarchical organization with one or two E1s (one in yeast), multiple E2s (11 in yeast) (Table 1), and a large family of E3s (60–100 in yeast) (Table 2). E3s mediate the exquisite selectivity of ubiquitylation by direct interaction with substrates.

Figure 1 

Protein ubiquitylation. (A) Ubiquitin is activated by E1 in an ATP-dependent step, transferred to the active site cysteine in an ubiquitin-conjugating enzyme (E2), and covalently attached to substrate proteins. Substrate selection depends on ubiquitin ligases (E3). Conjugation of a single ubiquitin molecule generates monoubiquitylated proteins. Repeated rounds of ubiquitin activation and conjugation lead to multi- or polyubiquitylated proteins. (B) Different polyubiquitin chain topologies can be synthesized depending on the specific lysine residue in ubiquitin used for chain formation. Three of the eight possible unbranched chain topologies (K6, K11, K27, K29, K33, K48, K63, and linear chains), and only one type of the possible forked polyubiquitin chains are shown. (C) Structural model for synthesis of K63-linked polyubiquitin chains by Ubc13/Mms2. Mms2 positions the acceptor ubiquitin with K63 in proximity to the active site cysteine of Ubc13. Figure adapted with permission from Macmillan Publishers Ltd: Chan, N. L., and C. P. Hill, 2001 Nat. Struct. Biol. 8: 650–652.

View this table:
Table 1  Ubiquitin conjugating enzymes of Saccharomyces cerevisiae
View this table:
Table 2  Ubiquitin ligases and components of Saccharomyces cerevisiae

Topology of ubiquitin conjugates

Monoubiquitylation describes the attachment of a single ubiquitin molecule to a substrate protein, whereas the attachment of more than one ubiquitin is referred to as polyubiquitylation or multiubiquitylation (Figure 1A). Polyubiquitylation represents the characteristic degradation signal, synthesized through isopeptide bond formation between lysine residues on substrate-anchored ubiquitin molecules and activated free-ubiquitin moieties. In contrast, multiubiquitylation is the attachment of multiple single ubiquitin molecules to several acceptor lysine residues in one protein. Marking proteins for degradation by the proteasome is the primary function of most polyubiquitin chains. In contrast, multi- or monoubiquitylation often, but not always (Dimova et al. 2012), mediates proteasome-independent functions such as protein binding, subcellular localization, intracellular trafficking, and modulation of activity (Hicke 2001; Kravtsova-Ivantsiv et al. 2009; Ziv et al. 2011).

Polyubiquitin chain assembly involves the formation of ubiquitin–ubiquitin conjugates, and any of the seven lysines of ubiquitin (K6, K11, K27, K29, K33, K48, and K63) can serve as an isopeptide bond acceptor in yeast and mammals (Figure 1B; Peng et al. 2003; Tagwerker et al. 2006; Meierhofer et al. 2008; Xu et al. 2009; Komander and Rape 2012). The resulting chains may define distinct signals, though all of them except K63-linked chains appear to mark proteins for degradation by the proteasome (Meierhofer et al. 2008; Xu et al. 2009; Kim et al. 2011). Depending on the substrate, some K63 chains might do so as well (Saeki et al. 2009b). Given that K48 and K63 chains have long been considered as canonical chain topologies, quantitation of the different polyubiquitin chains in yeast revealed a surprisingly high abundance of unconventional linkages. In unperturbed cells, ∼29% of all ubiquitin–ubiquitin linkages are through K48, 16% through K63, and 28% through K11. The remaining topologies are less abundant, with K6 at 11%, K27 at 9%, and ∼3% each for K29 and K33 (Xu et al. 2009).

The concept of specific signaling functions mediated through different ubiquitin chain topologies emerged from the engineering of yeast cells expressing K48R or K63R ubiquitin mutants to prevent formation of K48- or hypothetical K63-linked ubiquitin chains, respectively (Finley et al. 1994; Spence et al. 1995). K48-linked chains were found to target proteins for degradation and to be essential for viability, whereas K63-chains were dispensable during unstressed growth and did not affect degradation of proteasome substrates, but were required for the DNA damage response. The specific functions of other chain topologies are less clear. Interestingly, preventing K11 chain formation in yeast by K11R-ubiquitin replacement results in hypersensitivity to the ER-stress inducers DTT and tunicamycin, indicating that K11 chains are important for the endoplasmic-reticulum–associated degradation (ERAD) pathway (Xu et al. 2009).

Methods for the detection and quantification of ubiquitin conjugates have recently been reviewed (Kim et al. 2011; Laney and Hochstrasser 2011).

Ubiquitin-activating enzyme

In yeast, a single E1 enzyme is responsible for activation of ubiquitin. E1 is encoded by the essential UBA1 gene (McGrath et al. 1991). Several temperature-sensitive uba1 alleles exist; uba1-206 is a tight mutant, and shows rapid depletion of ubiquitin conjugates at nonpermissive temperature as well as other phenotypes expected from a general block of ubiquitylation (Ghaboosi and Deshaies 2007).

Ubiquitin-conjugating enzymes

The first yeast E2 enzymes identified were Rad6/Ubc2 (Jentsch et al. 1987) and Cdc34/Ubc3 (Goebl et al. 1988). A total of 13 yeast UBC genes have been designated (Table 1), though further biochemical analyses revealed that Ubc9 and Ubc12 do not conjugate ubiquitin, but rather the ubiquitin-like proteins Smt3 (mammalian SUMO) and Rub1 (mammalian Nedd8), respectively (Johnson and Blobel 1997; Liakopoulos et al. 1998). Among the 11 genuine ubiquitin-conjugating enzymes only Cdc34/Ubc3 is essential for viability (Goebl et al. 1988). Temperature-sensitive cdc34 mutants arrest at the G1-to S-phase transition of the cell cycle due to a defect in degradation of the cyclin-dependent kinase inhibitor Sic1 (Schwob et al. 1994). Cdc34 has many other substrates and most are selected by the Skp1Cdc53–F-box (SCF) ubiquitin ligase family for which Cdc34 serves as the main, if not only, E2 enzyme (Petroski and Deshaies 2005). In addition, Cdc34 together with the ubiquitin ligase San1 functions in the nuclear protein quality control pathway (Gardner et al. 2005a; see below). Ubc1 is an alternative ubiquitin-conjugating factor for San1 (Gardner et al. 2005a).

Several other E2 enzymes are important for protein quality control pathways outside the nucleus. Ubc4 and Ubc5 are highly similar and function redundantly in conjugation of ubiquitin to abnormal proteins in the cytosol to induce their degradation by the proteasome (Seufert and Jentsch 1990). The double mutant is inviable in some genetic backgrounds (Panasenko et al. 2009; Stoll et al. 2011); in others it shows severe growth defects (Seufert and Jentsch 1990; Chen et al. 1993). Three E2 enzymes are involved in degradation of misfolded proteins from the endoplasmatic reticulum (ERAD pathway; see below): Ubc1, Ubc6, and Ubc7. Among these, only Ubc6 is directly anchored to the ER membrane by a C-terminal transmembrane region (Sommer and Jentsch 1993), whereas Ubc7 is recruited to the ER membrane and activated by ER-bound Cue1 (Biederer et al. 1997; Bazirgan and Hampton 2008). Ubc6 but not Ubc7 contributes significantly to total cellular protein modification with K11-linked polyubiquitin chains (Xu et al. 2009).

Multiple E2 enzymes can be involved in degradation of a single substrate, the MATα2 transcriptional regulator being a complex case in which four different UBCs have been implicated (Ubc4, Ubc5, Ubc6, and Ubc7) (Chen et al. 1993). However, many substrates may rely on a single E2. Ubiquitin-conjugating enzymes can also operate sequentially for efficient substrate polyubiquitylation, as demonstrated for Ubc1 and Ubc4 in polyubiquitylation of cell cycle regulators targeted by a ubiquitin ligase known as the anaphase promoting complex, or cyclosome (APC/C; Rodrigo-Brenni and Morgan 2007). Polyubiquitylation requires two distinct types of conjugation events: Attachment of the first ubiquitin to the substrate protein in an initial monoubiquitylation step, followed by cycles of ubiquitin chain elongation. In yeast, the rate-limiting monoubiquitylation step for APC/C substrates is catalyzed by Ubc4, whereas efficient ubiquitin chain synthesis requires Ubc1 (Rodrigo-Brenni and Morgan 2007). A C-terminal ubiquitin associated (UBA) domain that binds ubiquitin—a feature of Ubc1 not shared with any other yeast E2 (Merkley and Shaw 2004)—is required for optimal processivity of this reaction (Rodrigo-Brenni and Morgan 2007).

Some E2s are poised for synthesis of polyubiquitin chains. For example, heterodimeric E2s such as the yeast Ubc13/Mms2 complex synthesize polyubiquitin chains by transferring the thioester-bound donor ubiquitin from the catalytically active subunit (Ubc13) onto an acceptor ubiquitin that is noncovalently bound to a catalytically inactive UEV (ubiquitin E2 variant) binding partner (Mms2) (Hofmann and Pickart 1999; Eddins et al. 2006). Other E2s may transfer preassembled polyubiquitin chains onto substrates, as described for the mammalian Ube2g2 enzyme and its yeast ortholog Ubc7 (Li et al. 2007b; Ravid and Hochstrasser 2007). However, the same E2 can often catalyze both mono- and polyubiquitylation. For example, Rad6/Ubc2 catalyzes monoubiquitylation of the proliferating cell nuclear antigen PCNA (Hoege et al. 2002) and histone H2B (Robzyk et al. 2000), but forms polyubiquitin chains in the context of the N-end rule (Dohmen et al. 1991), a conserved pathway that relates protein stability to the identity of the amino terminal residue (Varshavsky 1992; Varshavsky 2011; Tasaki et al. 2012). Rad6/Ubc2 functions with different ubiquitin ligases in these pathways, and it appears that ligases and E2 enzymes, as well as the substrates themselves, can be determinants deciding between mono- or polyubiquitylation.

Ubiquitin-conjugating enzymes also help to define the linkage type during polyubiquitin chain synthesis. In particular, E2s dictate chain architecture when paired with really interesting new gene (RING) domain ubiquitin ligases, whereas HECT (homologous to E6-AP carboxy terminus) domain E3s override any intrinsic chain topology preference of E2s. Synthesis of polyubiquitin chains with specific architectures by RING–E3/E2 pairs requires positioning of the E2 such that the linkage-defining lysine residue in the acceptor ubiquitin is proximal to the charged active site cysteine of the E2. The best-studied example in yeast is Ubc13, which synthesizes K63-linked chains (Figure 1C). In the Ubc13/Mms2 heterodimer, Mms2 positions the acceptor ubiquitin so that only K63 is allowed to approach the active site cysteine of Ubc13 (Eddins et al. 2006). A related mechanism was demonstrated for the mammalian ubiquitin-conjugating enzyme Ube2S, where a ubiquitin-binding region in the E2 orients the acceptor ubiquitin for K11-selective chain synthesis (Wickliffe et al. 2011).

The ubiquitin-conjugating enzymes are at the center of the E1–E2–E3 cascade. They interact with E1 and E3 but also ensure unidirectional handoff of ubiquitin from E1 to the substrate. E1 and E3 use a shared binding site on E2s, preventing recharging of E2s while bound to E3s and forcing their dissociation before the next round of conjugation (Eletr et al. 2005). Directionality of ubiquitin transfer is ensured by E1-dependent ATP hydrolysis as well as the different affinities of charged and uncharged E2s for E1 and E3. The ubiquitin-activating enzyme E1 binds uncharged E2s with higher affinity than the E2∼Ub leading to release of the loaded E2∼Ub (Hershko et al. 1983; Pickart and Rose 1985). Similarly, E3s have somewhat higher affinity for E2∼Ub than for the uncharged E2, facilitating processive ubiquitin chain synthesis (Siepmann et al. 2003; Saha and Deshaies 2008).

Ubiquitin ligases

Ubiquitin ligases (E3s) form the largest group of proteins involved in ubiquitylation and they confer selectivity to the process. They bind E2s and substrate proteins to facilitate substrate-specific ubiquitylation. The first E3 identified was Ubr1, a mediator of the N-end rule pathway. Ubr1 binds protein substrates with different affinities based on their N-terminal amino acids (Bartel et al. 1990; Varshavsky 1992). Many other E3 enzymes were subsequently identified, all falling into two major classes: RING domain E3s (including the structurally related U-box domain E3s) and HECT domain E3s. Considering sequence features frequently associated with ubiquitin ligases, such as RING (like), F box, or HECT motifs, there are 60–100 putative E3s in yeast (Table 2). Most belong to the class of RING domain E3s, and only five HECT domain E3s are encoded in the yeast genome (Table 2). RING and HECT domain E3s follow distinct mechanisms to catalyze ubiquitylation (Figure 2). HECT domain E3s contain an active site cysteine within the HECT domain, which forms a thioester with ubiquitin received from an E2 prior to its transfer to the substrate (Scheffner et al. 1995). RING E3s do not form thioester intermediates; they instead facilitate ubiquitin transfer by positioning the charged E2∼Ub in proximity to the acceptor lysine in the substrate. In addition, RING domain ligases seem to activate E2s to facilitate ubiquitylation (Deshaies and Joazeiro 2009).

Figure 2 

HECT and RING E3 ubiquitin ligases. Substrate ubiquitylation with HECT E3s involves an E3∼Ub thioester intermediate. Ubiquitin is transferred from the HECT E3 to the substrate. RING E3s typically do not form thioester intermediates but promote ubiquitin conjugation by bridging the interaction between E2 and substrate proteins. RING E3s also stimulate E2 activity. A subclass of RING-based ligases, the RING-in-between-RING (RBR) proteins, function like RING/HECT hybrids and form thioester intermediates. This mechanism remains to be confirmed for putative yeast RBR ligases.

A subclass of RING domain E3s, the RING-in-between-RING (RBR) proteins, appear to function as RING/HECT hybrids (Wenzel and Klevit 2012). They bind E2s with one RING domain and stimulate the transfer of ubiquitin onto a conserved cysteine residue in the other RING domain, forming an E3∼Ub thioester before conjugation to the substrate (Wenzel et al. 2011). Homology searches revealed two putative RBR ligases in yeast (Hel1 and Itt1) (Eisenhaber et al. 2007). Whether they indeed form E3∼Ub intermediates is unknown.

Functional interaction between RING and HECT domain E3s has been demonstrated for the N-end rule pathway (Hwang et al. 2010). The RING-type Ubr1 and HECT-type Ufd4 ligases form a complex to enhance processivity of substrate ubiquitylation. A similar role for the Ubr1/Ufd4 complex in the ubiquitin-fusion degradation pathway has also been suggested (Hwang et al. 2010). Interestingly, the Ubr1/Ufd4 complex may function as an E3/E4 pair. E4 enzymes—a small subgroup of ubiquitin ligases—select substrate proteins based on their having been previously ubiquitylated, and E4s function to extend these ubiquitin chains (Koegl et al. 1999).

HECT ubiquitin ligases:

HECT domain E3s are named after their founding member E6AP, which ubiquitylates mammalian p53 in cells expressing the human papilloma virus protein E6. Yeast has five HECT E3s: Rsp5, Ufd4, Hul4, Hul5, and Tom1. The HECT domain is an ∼350-residue region consisting of the N-terminal lobe, which binds an E2, and the C-terminal lobe containing the active site cysteine, which forms a thioester intermediate with ubiquitin. N and C lobes are connected by a flexible hinge region (Huang et al. 1999). The five yeast HECT domain ubiquitin ligases function in diverse processes ranging from multivesicular body (MVB) sorting, endocytosis, histone degradation, and processing of ubiquitylated proteins (Hoppe et al. 2000; Shcherbik et al. 2003; Rape and Jentsch 2004; Crosas et al. 2006; Rotin and Kumar 2009; Singh et al. 2009).

The E3∼Ub thioester intermediate mediates E3-instructed ubiquitin chain assembly as demonstrated for Rsp5, which has been shown to dictate synthesis of K63-linked chains independently of the E2 enzymes used (Kim and Huibregtse 2009). Although the molecular mechanism is not known in detail, mutational studies suggest that the carboxy-terminal region of Rsp5 is involved in acceptor ubiquitin orientation to favor nucleophilic attack from lysine-63 in ubiquitin.

Rsp5 is the only yeast HECT E3 essential for viability in rich medium. Rsp5 is a particularly active E3 that mediates ubiquitylation of a large number of substrates and contributes to regulation of diverse biological pathways (Gupta et al. 2007; Rotin and Kumar 2009). Rsp5 is required for upregulation of expression of the fatty acid desaturase OLE1 by the homologous transcription factors Spt23 and Mga2, and accordingly the lethality of rsp5 mutants can be rescued by addition of oleic acid to the growth medium (Hoppe et al. 2000). While Spt23 and Mga2 are normally anchored in the ER membrane, Rsp5-mediated ubiquitylation induces proteasomal processing and release of transcriptional activation domains from these proteins (Hoppe et al. 2000; Shcherbik et al. 2003). Hul5, another HECT domain protein, is discussed below in the Proteasome section.

RING domain ubiquitin ligases:

There are 44 yeast proteins containing RING domains and two proteins of the U-box family, which are structurally related to RING E3s but do not bind zinc (Ufd2 and Prp19). Although conclusive biochemical evidence for ubiquitin ligase activity is not available for all RING domain proteins, most of them probably have this activity. The globular RING domains bind E2 enzymes (Zheng et al. 2000) and appear to stimulate ubiquitin transfer by induction of subtle structural changes (Ozkan et al. 2005). Substrate recruitment, the central function of ubiquitin ligases, is achieved either by substrate binding domains within the same polypeptide chain as the RING domain (single subunit RING E3s) or by engaging specialized substrate receptors to form multisubunit RING E3s (Deshaies and Joazeiro 2009). Examples of the former are the N-recognin Ubr1 (Bartel et al. 1990); the ubiquitin ligase Bre1 that together with Rad6/Ubc2 catalyzes histone H2B ubiquitylation (Wood et al. 2003); the regulator of nuclear protein quality control San1 (Gardner et al. 2005a); Rkr1/Ltn1, which ensures degradation of potentially cytotoxic translation products produced from mRNAs that lack stop codons (Bengtson and Joazeiro 2010); and the two RING E3s, Rad18 and Rad5, which catalyze mono- and polyubiquitylation of PCNA, respectively (Hoege et al. 2002; see below). Prominent members of the multisubunit RING E3s are the APC/C (Pesin and Orr-Weaver 2008) and the largest group of ligases, the modular cullin-RING ligases (CRLs) (Petroski and Deshaies 2005; Zimmerman et al. 2010; Duda et al. 2011). Although one subunit (Apc2) of APC/C contains a cullin-like domain, the overall ligase architecture is very different from that of true CRLs. Detailed studies of other RING domain proteins (Table 2) may identify additional multisubunit E3s as has been shown for the seven-subunit Gid (glucose-induced degradation-deficient) E3, which controls the metabolic switch between glycolysis and gluconeogenesis (Santt et al. 2008; Menssen et al. 2012).

APC/C:

APC/C is perhaps the most complex ubiquitin ligase. Its core is composed of 13 subunits (Apc1, Apc2, Cdc27, Apc4, Apc5, Cdc16, Cdc23, Apc9, Doc1, Apc11, Cdc26, Swm1, and Mnd2), with Apc11 being the RING domain component that binds Ubc1 and Ubc4, the two primary E2s functioning with yeast APC/C (McLean et al. 2011). The core APC/C associates with one of three activators that bind substrates and are crucial targets for APC/C regulation. Cdh1 and Cdc20 are activators controlling mitotic cell cycle progression and Ama1 recruits meiotic targets to APC/C (Visintin et al. 1997; Cooper et al. 2000).

Degradation of several important APC/C substrates ensures ordered progression through the steps of chromosome segregation. A cascade of mitotic events is unleashed by APC/C-mediated degradation of Pds1/securin to initiate the metaphase-to-anaphase transition (Cohen-Fix et al. 1996; Yamamoto et al. 1996). Pds1 is an inhibitor of Esp1/separase, a protease that cleaves the cohesin Scc1 to allow sister chromatid separation (Ciosk et al. 1998; Uhlmann et al. 1999). Clb2 and other B-type cyclins are degraded by APC/C from anaphase until the end of the subsequent G1 phase, which ensures a period of low cyclin-dependent kinase activity that is important for cytokinesis and the assembly of prereplication complexes (Irniger et al. 1995). Many other mitotic and meiotic regulators are APC/C substrates, and their degradation controls both normal mitotic processes and cell cycle checkpoint pathways (Pesin and Orr-Weaver 2008; McLean et al. 2011).

Tight regulation of Cdh1 and Cdc20 restricts APC/C activity to M phase and G1 of the mitotic cell cycle (Pesin and Orr-Weaver 2008). G2/M-phase–induced CDC20 expression, APC/C phosphorylation-dependent binding of Cdc20 (Rudner and Murray 2000; Rudner et al. 2000), combined with active Cdc20 degradation during G1 by APC/CCdh1, limit Cdc20 association with APC/C to M phase (Prinz et al. 1998; Foe et al. 2011). In contrast, Cdh1 levels are largely constant throughout the cell cycle, but binding to APC/C is prevented by Cdh1 phosphorylation during most of the cell cycle, except late M phase and G1 (Zachariae et al. 1998).

APC/C substrates share distinct degradation motifs, the most common being the classic destruction box (D box) and the KEN box (Glotzer et al. 1991; Pfleger and Kirschner 2000). Although APC/C activators play a crucial role in D-box and KEN-box recognition, the core subunit Apc10/Doc1 serves as a coreceptor in D-box recognition (Carroll et al. 2005; Da Fonseca et al. 2011). Regulation occurs at the level of activator abundance, phosphorylation of activators and core components, as well as binding of the APC/CCdc20 inhibitors Mad2 and Mad3 (McLean et al. 2011).

Cullin-RING ligases:

CRLs form the largest group of ubiquitin ligases in all eukaryotes. A typical CRL ligase consists of four subunits: the RING protein Hrt1/Rbx1/Roc1, a cullin, a linker protein, and one of many alternative substrate receptors (Petroski and Deshaies 2005; Zimmerman et al. 2010; Duda et al. 2011). CRLs are assembled on a central scaffold subunit, the cullin, three of which are found in budding yeast (Cdc53, Cul3, and Rtt101). The C-terminal regions of cullins bind the small RING domain subunit Hrt1 (Kamura et al. 1999; Ohta et al. 1999; Seol et al. 1999), which in turn recruits and activates the E2 Cdc34. The N-terminal regions of cullins interact with substrate receptor subunits (F box, SOCS box, or DCAF proteins), usually through linker proteins (Skp1, Elc1, and Mms1) (Figure 3). Depending on the cullin, different classes of CRLs are formed. Cdc53 and Cul3 are orthologs of human Cul1 and Cul3, respectively. Rtt101 does not show significant homology to any particular vertebrate cullin but is functionally similar to human Cul4. The canonical CRLs, the SCF ligases, are assembled onto Cdc53/Cul1.

Figure 3 

Cullin RING ligases (CRLs). A large class of multisubunit RING-based ligases is nucleated around cullins. Yeast has three classes of CRLs formed with the cullins Cdc53 (cullin 1), Cul3, and Rtt101 (functionally similar to human Cul4). The C-terminal regions of cullins bind the RING protein Hrt1/Rbx1/Roc1, and the N-terminal portions interact with specific adaptor proteins (Skp1, Elc1, and Mms1), which recruit substrate receptor proteins (F-box, SOCS-box, or DCAF proteins). Putative substrate receptors are listed in Table 3.

Proteins containing the F-box motif form substrate receptors of SCF ligases and recruit proteins with their C-terminal protein binding domains for ubiquitylation (Bai et al. 1996). Often substrate phosphorylation creates a binding surface that is recognized by the F-box subunit (Feldman et al. 1997; Skowyra et al. 1997). Yeast encodes 22 F-box proteins (Table 2), most of which form SCF ligases with distinct substrate specificities. Three F-box proteins (Cdc4, Met30, and Ctf13) are essential for viability in rich medium. Ctf13 likely does not form a conventional SCF E3, but is a structural component of the centromere binding complex CBF3 (Russell et al. 1999a). The best-studied yeast F-box proteins are Cdc4, Grr1, and Met30. The corresponding ligases SCFCdc4, SCFGrr1, and SCFMet30 each control ubiquitylation of cell cycle regulators and proteins involved in nutrient signaling and may thus be key factors for integration of cell cycle progression and nutrient status.

SCFCdc4 controls entry into S phase by degradation of the cyclin-dependent kinase inhibitor Sic1 (Feldman et al. 1997; Skowyra et al. 1997) and regulates the response to amino acid starvation through ubiquitylation and degradation of the transcription factor Gcn4 (Meimoun et al. 2000; Chi et al. 2001). SCFCdc4/Sic1 is probably the best-studied ligase/substrate pair, and much of our understanding about CRL function comes from biochemical characterization of Sic1 ubiquitylation. SCFGrr1 ubiquitylates the G1 cyclins Cln1 and Cln2 to control their abundance (Barral et al. 1995; Seol et al. 1999; Skowyra et al. 1999). Consequently, SCFGrr1 is an important regulator of cyclin-dependent kinase activity during G1. In addition, SCFGrr1 induces degradation of Mth1, which is critical for glucose sensing and adaptation to varying glucose concentrations (Flick et al. 2003). SCFCdc4 and SCFGrr1 have many additional substrates and functions (Benanti et al. 2007; Skaar et al. 2009).

Fewer substrates are currently known for SCFMet30, but their analyses have taught us about diversity and flexibility of ubiquitin signaling. SCFMet30 coordinates cell division with nutrient or heavy metal stress (Kaiser et al. 2006). One key substrate in this pathway is the transcription factor Met4, which is directly inactivated by modification with a K48-linked ubiquitin chain, but degradation is prevented because two ubiquitin binding motifs in Met4 shield the polyubiquitin chain from signaling degradation (Flick et al. 2006; Tyrrell et al. 2010). Although ubiquitylated Met4 is inactive as a transcription factor, it functions as a substrate receptor in the context of the extended SCFMet30/Met4 ubiquitin ligase to trigger ubiquitylation and degradation of several Met4 binding factors, including Met32, which induces cell cycle arrest when stabilized (Ouni et al. 2010). The dual function of Met4 as transcription factor and ubiquitin ligase component allows it to coordinate cell cycle progression with response to nutrient or heavy metal stress.

An interesting aspect of CRL regulation is a ubiquitin-like modification found on cullins. Cullins are covalently modified on a conserved lysine residue in the C-terminal region by the ubiquitin-like protein Rub1, the yeast ortholog of metazoan Nedd8 (Lammer et al. 1998; Liakopoulos et al. 1998). Cullin modification with Nedd8 induces a major conformational change such that the E2-binding interface of the RING component Hrt1 extends out from the cullin surface, remaining tethered only by a flexible linker region. This not only allows the E2 to closely approach the substrate, but also provides the flexibility to adopt different conformations necessary for polyubiquitin chain synthesis (Duda et al. 2008). Rub1 modification is not essential for viability of budding yeast, but it is required for robust CRL activity and is essential in other organisms (Willems et al. 2004).

Deubiquitylation

Deubiquitylating enzymes catalyze the hydrolysis of the isopeptide bonds that link ubiquitin to its targets (Reyes-Turcu et al. 2009). Twenty deubiquitylating enzymes (DUBs) are found in yeast (Table 3), falling into four families: the Usp family, including 16 members; the Otu family, with two members; and the JAMM and Uch families, with one member each. Additional paralogs exist, with specificity for ubiquitin-like proteins such as Smt3 and Rub1. Yuh1, the lone Uch-type DUB in yeast, may serve primarily in the removal of Rub1 from target proteins, although capable of deubiquitylation as well (Linghu et al. 2002). Most DUBs are thiol proteases, the only exception being Rpn11, a zinc metalloprotease (Verma et al. 2002; Yao and Cohen 2002). The three-dimensional structures of several DUBs from yeast and other organisms are available (Johnston et al. 1999; Hu et al. 2005; Li et al. 2007a; Sato et al. 2008; Reyes-Turcu et al. 2009; Köhler et al. 2010).

View this table:
Table 3  Deubiquitylating enzymes of Saccharomyces cerevisiae

The DUBs are highly diverse functionally, reflecting both their subcellular localization and their inherent substrate specificities. For example, Ubp8 is a component of the SAGA complex, a nuclear particle involved in chromatin remodeling (Henry et al. 2003). Ubp10 is also a specific regulator of nuclear processes such as the silencing of gene expression (see below). Other DUBs seem to function specifically on endosomes and multivesicular bodies, such as Doa4/Ubp4 (Luhtala and Odorizzi 2004; Amerik et al. 2006). One DUB, Ubp16, is thought to be an integral membrane protein and fractionates with mitochondria (Kinner and Kölling 2003). The enzymatic specificity of DUBs from yeast is only partially characterized (Amerik et al. 2000b; Schaefer and Morgan 2011). DUBs are presented with potential substrates that must number in the hundreds and possibly thousands, given the breadth of the ubiquitin pathway (Kim et al. 2011). Systematic identification of DUB substrates in yeast has not been attempted, and it is even unclear in general how rapidly ubiquitin modifications of protein substrates are reversed within cells.

DUB activity is required not only for the disassembly of ubiquitin–protein conjugates but also for biosynthetic processing of the Ubi1Ubi4 gene products, ubiquitin fusion proteins that are the sole source of ubiquitin in the cell. UBI1UBI3, which supply most of the ubiquitin in growing, unstressed cultures, encode ubiquitin as N-terminal fusions to ribosomal proteins L40 and S31 (Finley et al. 1989). UBI4, the stress-responsive ubiquitin gene (Finley et al. 1987), has a series of six tandem repeats of the ubiquitin coding sequence (Ozkaynak et al. 1984). DUB activity is essential to release ubiquitin from these precursor forms, as their C termini are blocked. It is not known which DUBs are responsible for these cleavage events, but they exhibit fast reaction kinetics, as observed for artificial ubiquitin–β-galactosidase fusion proteins (Bachmair et al. 1986).

An important function of the DUBs is to recycle ubiquitin by recovering it from ubiquitin–protein conjugates before the target protein is degraded. Defects in this process give rise to reduced ubiquitin levels and pleiotropic stress sensitivities. The main DUBs responsible for recovering ubiquitin from conjugates that are en route to being degraded are Ubp6, Rpn11, and Doa4 (Swaminathan et al. 1999; Amerik et al. 2000b; Leggett et al. 2002; Hanna et al. 2003, 2007; Chernova et al. 2003; Kimura et al. 2009). Ubp6 and Rpn11 rescue ubiquitin from degradation by the proteasome, and Doa4 releases ubiquitin from membrane proteins that are about to be internalized within multivesicular bodies en route to the lysosome. Both Ubp6 and Rpn11 can release ubiquitin from proteasome substrates in the form of unanchored chains (Verma et al. 2002; Yao and Cohen 2002; Hanna et al. 2006). If not promptly disassembled, such chains can inhibit the proteasome by competing with ubiquitin–protein conjugates for access to proteasomal ubiquitin receptors. Ubp14 is dedicated to breaking down such unanchored chains (Amerik et al. 1997). Its specificity is achieved by recognition of the free C terminus of the proximal ubiquitin of the chain, leading to allosteric activation and cleavage of the isopeptide bond joining the proximal ubiquitin to the penultimate member of the chain (Reyes-Turcu et al. 2009). Doa4 can also disassemble free chains and plays a major role in this process upon heat shock (Kimura et al. 2009).

DUBs often function within protein complexes, and in such cases are typically activated by incorporation into the complex. For example, Ubp6 and Rpn11 are thought to be active only when associated with the proteasome (Leggett et al. 2002; Verma et al. 2002), Ubp3 is activated by Bre5 (Cohen et al. 2003), and Otu1 functions in association with Cdc48 (Rumpf and Jentsch 2006). A particularly elegant example is the activation of Ubp8 as a DUB when it is incorporated into the SAGA complex (Köhler et al. 2010). Additional modes of DUB regulation are exemplified by the transcriptional induction of the UBP6 gene in response to reduced ubiquitin levels (Hanna et al. 2007); the inhibition of Doa4 by Rfu1, which is relieved upon heat shock (Kimura et al. 2009); and stimulation of Ubp3 activity by Hog1 kinase-dependent phosphorylation upon osmotic stress (Solé et al. 2011).

Some DUBs antagonize specific ubiquitin ligases. Ubp2 forms a complex with the ligase Rsp5, and deubiquitylates those proteins that Rsp5 modifies (Kee et al. 2005, 2006; Harreman et al. 2009). Other cases of DUB–ligase antagonism involve E4 enzymes. Thus, the E4 Ufd2 is antagonized by Otu1, with both residing on Cdc48 (Rumpf and Jentsch 2006), and the E4 Hul5 is antagonized by Ubp6, with both residing on the proteasome (Crosas et al. 2006). It would be interesting to understand why DUB–ligase pairs have evolved in these cases, since most ligases do not seem to be pitted against a specific DUB in this way.

Because of the abundance of DUBs in yeast, it is necessary to take precautions against postlysis deubiquitylation when assessing the role of ubiquitylation in any setting. DUBs that are thiol proteases are inactivated by the alkylating agent N-ethylmaleimide, but a zinc chelating agent such as o-phenanthroline is recommended in addition to neutralize the metalloprotease Rpn11 (Verma et al. 2002).

Proteasome

The proteasome has 33 distinct subunits (Table 4) and is the most complex protease known (Finley 2009). Its principal function is to degrade ubiquitin–protein conjugates. The proteasome is found in all eukaryotes and is highly conserved in evolution. Proteasomes are organized into two subassemblies, the 19S regulatory particle (RP) and the 20S core particle (CP). The RP recognizes substrates to be degraded, while the CP contains the proteolytic active sites. The proteolytic sites are sequestered within an interior space of the CP, ensuring that access to these sites is under strict control and nonspecific proteolysis is minimized (Figure 4). Substrates are routed from the RP to the CP through a narrow substrate translocation channel, which can exist in open and closed states (Figure 4). Globular proteins must be unfolded to traverse this channel. Unfolding is an active process mediated by the six distinct ATPases of the RP, Rpt1Rpt6, which form a heteromeric ring complex (Figure 5A). Simple methods are available for testing whether an unstable protein is degraded in a proteasome-dependent manner (Fleming et al. 2002; Liu et al. 2007).

View this table:
Table 4  Proteasome components and cofactors
Figure 4 

Proteasome core particle. (A) Space-filling exterior view of the CP, with subunits differentiated by color. Note the α7β7β7α7 organization. (B) Medial cut-away view of the CP, showing the interior cavity and active sites (red) sequestered within it. The substrate transloction channel is fully closed in the crystal structure of the free CP, but brackets indicate the approximate position of the channel in its open state. (C) Detail of the CP gate. The N-terminal tails of the α subunits, particularly α2, α3, and α4, as shown, block substrate access. The bodies of the α subunits are rendered in gray. Arrow indicates the movement of the tails that constitutes gate opening, a likely upward and outward migration (Förster et al. 2003). Images modified from Groll et al. 1997 and Tian et al. 2011, with permission.

Figure 5 

The proteasome holoenzyme. (A) Model of the Rpt ring of the proteasome in association with the yeast CP. Medial cut-away view, with the Rpt ring modeled from observations of the PAN ATPase from Archaea (adapted from Zhang et al. 2009b, with permission). The ATPase domain of the Rpt ring and the smaller OB domain above it both in blue. Coiled-coil elements (turquoise) emerge distally from the OB domain with their trajectory influenced by Pro91 (pink). The CP is in green, with proteolytic sites in red. Slice surfaces of the CP and Rpt ring are in black. The presumptive substrate translocation channel is demarcated with yellow lines: The entry port of the translocation channel is thought to be the OB ring, and substrates must migrate to the proteolytic active sites (red) to be hydrolyzed. The driving force for translocation is thought to be axial motions of the pore loops from the ATPase domain that line the translocation channel (gold rectangles). (B) Tilted view of the RP based on EM studies (Lander et al. 2012). The Rpt ring and CP are colored as in A. The DUB Rpn11 is in turquoise, with the presumptive substrate entry port directly beneath it (red-orange). The ubiquitin receptor Rpn13 is in orange. To its left is Ubp6 (approximate position), contacting Rpn1. To the right is Rpn10, with its Von Willebrand A (VWA) domain in yellow and its ubiquitin-binding UIM domain in red. All other RP subunits are in gray. Shown for comparison at upper right is free ubiquitin (pink). (C) Lateral view of the RP (derived from Lander et al. 2012). Highlighted are Rpn1 (red-orange), Rpn2 (pink), Rpn13 (orange), and Rpn10 (yellow). Lid subunits are in gray. B and C are from Tian et al. (2012), with permission.

Core particle

The CP is a barrel-like structure composed of four stacked rings of subunits (Groll et al. 1997). The two outer rings are known as α rings, the two inner rings as β rings (Figure 4). CP components are generally referred to as α1–α7 and β1–β7 (Table 4). The proteolytic activity of the proteasome resides in the β ring; subunits β1, β2, and β5 are proteolytically active and are founding members of the threonine class of proteases. In each case, the active site nucleophile is the N-terminal α-amino group of the main chain. β1, β2, and β5 are synthesized as proenzymes and cleaved upon CP assembly to reveal a threonine residue at the new N terminus (Chen and Hochstrasser 1996; Arendt and Hochstrasser 1997; Groll et al. 1997). The specificities of the β1, β2, and β5 active sites are trypsin-like, caspase-like, and chymotrypsin-like, in that they prefer basic, acidic, or hydrophobic residues, respectively, N-terminal to the scissile bond (Groll et al. 2005).

The α rings regulate substrate access into the inner chamber of the CP (Groll et al. 2000; Whitby et al. 2000; Bajorek et al. 2003). In the free form of the CP, the center of the α ring is occupied by N termini from all seven subunits, which converge into a defined but irregular structure that blocks substrate access to the chamber. Another important function of the α ring is to serve as a docking site for the RP and other regulators of the CP, such as Blm10. Both the RP and Blm10 activate the peptidase of the CP by shifting the α N termini away from the center of the α ring, and thus creating an opening for the passage of substrate (Finley 2009; Sadre-Bazzaz et al. 2010). The interfaces of the α subunits form seven pockets, which provide docking sites for the RP and Blm10 (Sadre-Bazzaz et al. 2010; Tian et al. 2011). The C termini of the Rpt proteins project into these pockets to stabilize the association between the RP and CP and drive opening of the CP channel (see below).

Regulatory Particle

Subunit organization of the regulatory particle:

The spatial organization of the RP has been resolved in recent electron microscopy studies (Lander et al. 2012; Lasker et al. 2012; Pathare et al. 2012; Sakata et al. 2012), as summarized in Figure 5. The RP is composed of the 10-subunit base and nine-subunit lid subassemblies (Table 4; Glickman et al. 1998; Finley 2009). The RP is anchored to the CP principally through the base, but the lid subunit Rpn6 also contacts the CP (Lander et al. 2012; Pathare et al. 2012). Dissociation of the RP into base and lid is observed upon purification of proteasomes from rpn10Δ mutants, or upon purification of wild-type proteasomes in the presence of high salt (Glickman et al. 1998; Saeki et al. 2000). Moreover, the base and lid are intermediates in RP assembly (see below). Thus, the base–lid dichotomy reflects the fundamental organization of the RP.

Unfolding of the protein substrate and its translocation into the CP are driven by ATP hydrolysis (Schrader et al. 2009; Sauer and Baker 2011; Smith et al. 2011a). The heterohexameric Rpt ring of the base represents the ancient core of the machinery that defines ATP-dependent proteases in all kingdoms of life (Figure 5A). The 13 additional components of the RP are peculiar to eukaryotes and seem designed in large part to recognize or process the ubiquitin component of the ubiquitin–protein conjugate, as discussed below. For example, two components of the base are ubiquitin receptors, and other components of the base, Rpn1 and Rpn2, are large subunits that serve as scaffolds (Figure 5C), allowing for the recruitment of a variety of factors, such as shuttling receptors (see below) with their cargo of ubiquitin–protein conjugates.

Substrate recognition:

Two subunits of the RP, Rpn10, and Rpn13, bind ubiquitin chains. Rpn10 binds via its α-helical Ubiquitin-Interacting Motif (UIM) element (Elsasser et al. 2004; Verma et al. 2004; Mayor et al. 2007), and Rpn13 via a pleckstrin homology (PH) domain known as the Pleckstrin-like Receptor for Ubiquitin (PRU) domain (Husnjak et al. 2008). Rpn10 and Rpn13 are situated on opposite sides of the substrate entry port, with Rpn13 more distant from the port due its apical position (Figure 5C; Lander et al. 2012; Sakata et al. 2012). Although not proximal to one another (Figure 5B), Rpn10 and Rpn13 might simultaneously engage the same ubiquitin chain, given adequate chain length. The UIM element of Rpn10 appears to contact the coiled-coil domain shared by Rpt4 and Rpt5 (Figure 5B). Rpt5 has been hypothesized to be a ubiquitin receptor based on cross-linking studies (Lam et al. 2002), though never shown to bind ubiquitin directly; and the proximity of its coiled-coil element to the UIM of Rpn10 (Lander et al. 2012) provides a plausible explanation of the cross-linking result. The ability of Rpn10 to recognize ubiquitin chains is regulated by its ubiquitylation; ubiquitin covalently linked to Rpn10 can fold back to occupy the UIM site (Isasa et al. 2010).

The RP also recognizes ubiquitin conjugates through a family of UBL–UBA proteins that serve as shuttling receptors: Rad23, Dsk2, and Ddi1 (Table 4; Schauber et al. 1998; Chen and Madura 2002; Elsasser et al. 2002; Rao and Sastry 2002; Saeki et al. 2002a; Elsasser and Finley 2005; Finley 2009; Rosenzweig et al. 2012). The N-terminal UBL (ubiquitin-like) domain in each shuttling receptor serves as a docking site for the proteasome, and the UBA domain (or domains) binds ubiquitin chains. Rpn1 and Rpn13 have been identified as receptor sites for UBL–UBA proteins (Elsasser et al. 2002; Saeki et al. 2002b; Husnjak et al. 2008; Peth et al. 2010; Gomez et al. 2011; Rosenzweig et al. 2012). Of the five proteasomal ubiquitin receptors described above, none is essential, and there is some degree of functional redundancy in addition to distinct roles. The biochemical basis of their functional differentiation remains largely unknown.

The shuttling receptors have divergent properties. Ddi1, for example, contains an aspartyl protease domain that is likely to be functional based on its crystal structure and on the identification of a defined phenotype in an active-site substitution mutant (Sirkis et al. 2006; White et al. 2011). Thus, the protease activity of Ddi1 could possibly provide an alternative to the proteasome as a means to attack ubiquitylated proteins. Dsk2 is distinguished by the existence of an extraproteasomal pool that is largely complexed to a free pool of Rpn10 (van Nocker et al. 1996; Matiuhin et al. 2008; Zhang et al. 2009a). In this complex, the UBL domain of Dsk2 binds the UIM element of Rpn10, which is the ubiquitin-binding element of Rpn10 (Zhang et al. 2009a). Interestingly, the UBL–UIM interaction can be displaced by a substrate-bound ubiquitin chain to form a ternary complex, that, with an unoccupied Dsk2 UBL domain, is activated for proteasome binding. Despite this interaction, Dsk2 does not bind proteasomes via Rpn10 (Elsasser et al. 2002; Matiuhin et al. 2008). Interestingly, a mammalian homolog of Dsk2 has been implicated in amyotrophic lateral sclerosis (Deng et al. 2011). As described below, Rad23 participates in the nucleotide excision repair (NER) pathway of DNA repair. Finally, mutated variants of the shuttling receptors have been studied as unique model substrates of the proteasome, although their wild-type forms are relatively stable (Heessen et al. 2005; Fishbain et al. 2011; Heinen et al. 2011; Sekiguchi et al. 2011)

Deubiquitylation at the proteasome:

The lid is positioned for the most part laterally to the base, and its subunits extend like fingers to contact the base at many points (Figure 5C; Lander et al. 2012; Lasker et al. 2012). A key function of the lid is to deubiquitylate proteasome substrates, an activity mediated by its subunit Rpn11 (Maytal-Kivity et al. 2002; Verma et al. 2002; Yao and Cohen 2002). Rpn11, a metalloprotease, is thought to cleave at the substrate-proximal tip of the chain, thus removing the chain entirely. Rpn11 activity is typically dependent on ATP, though it is unlikely that Rpn11 is an ATPase. Rpn11 activity is likely coupled to ATP hydrolysis by Rpt proteins of the base, which is thought to translocate the substrate through the axial channel formed by the Rpt proteins. Presumably the substrate-attached chain is thereby moved toward the entry port of the channel, where it may encounter Rpn11. In agreement with this model, Rpn11 is found near the entry port of the substrate translocation channel (Figure 5B; Lander et al. 2012).

Remarkably, the lid is paralogous to two free complexes found in eukaryotic cells, eIF3 and the COP9 signalosome complex (Glickman et al. 1998). It appears that in the course of evolution the lid gave rise to the COP9 signalosome and eIF3. The COP9 signalosome is active in the removal of the ubiquitin-like protein Rub1 (see above) from covalent adducts to the cullin Cdc53 (Cope et al. 2002). Thus, the COP9 signalosome functions analogously to the lid, except that as it lost its association with the proteasome, its specificity was modified so that it cleaves a ubiquitin-like protein rather than ubiquitin.

Ubp6 is a second major proteasome-associated deubiquitylating enzyme (Verma et al. 2000; Leggett et al. 2002). ubp6 null mutants are ubiquitin deficient (Amerik et al. 2000b; Leggett et al. 2002), due to elevated rates of ubiquitin turnover by the proteasome (Chernova et al. 2003; Hanna et al. 2003). Thus Ubp6 serves, like Rpn11, to protect ubiquitin from degradation by the proteasome by removing ubiquitin before it is translocated into the CP. However, Ubp6 does so quite differently from Rpn11. First, the position of Ubp6 is distant from the substrate entry port (Figure 5B; Lander et al. 2012) Unlike Rpn11, Ubp6 disassembles ubiquitin chains in an ATP-independent manner. Ubp6 serves to inhibit protein degradation by the proteasome, using two distinct mechanisms. Its deubiquitylating activity can shorten a chain before the substrate is productively engaged by the proteasome, leading to release of intact substrate. This has been shown most clearly with Ubp6’s mammalian ortholog, Usp14 (Lee et al. 2010). Second, a catalytically inactive form of Usp14 can also inhibit protein degradation, through an unknown mechanism (Hanna et al. 2006). Finally, Ubp6 can influence gating of the substrate translocation channel (Peth et al. 2009). ubp6 mutants show an exceptional ability to tolerate aneuploidy (Torres et al. 2010), owing apparently to enhanced quality-control protein degradation, perhaps reflecting enhanced proteasome activity.

Substrate deubiquitylation by the proteasome is antagonized by Hul5, a proteasome-associated ubiquitin ligase (Crosas et al. 2006). Numerous proteins are stabilized or degraded nonprocessively in hul5 mutants, consistent with a generalized E4 activity of Hul5 (Crosas et al. 2006; Kohlmann et al. 2008; Aviram and Kornitzer 2010; Fang et al. 2011). The balance of Hul5 and Ubp6 activity can fine tune proteasome activity to cellular conditions (Hanna et al. 2007; Fang et al. 2011; Park et al. 2011). In particular, Hul5 has been shown to be the major ubiquitin ligase targeting misfolded cytosolic proteins upon heat stress (Fang et al. 2011).

Initiation sites:

Some proteins are resistant to protein degradation by the proteasome, even when modified by canonical ubiquitin chains. One potential explanation is that such proteins are inherently resistant to unfolding. However, this property does not correlate with the thermal melting profile of these proteins (Lee et al. 2001). Such proteins can be converted into favored substrates by appending short peptide segments to their N- or C termini (without perturbing their thermal melting profile). Unstructured peptide elements that are necessary for proteasome-mediated degradation (Prakash et al. 2004; Takeuchi et al. 2007; Schrader et al. 2009) are known as initiation sites. Such sites may be employed to dissociate specific subunits of a protein complex for selective degradation (Johnson et al. 1990; Verma et al. 2001; Prakash et al. 2009). Degradation is thought to proceed from an initiation site (Piwko and Jentsch 2006; Schrader et al. 2009), usually continuing to completion. In rare cases, degradation is interrupted and stable protein fragments escape from the proteasome, owing to the inability of the proteasome to effect complete substrate unfolding. As described below, this type of mechanism is used to activate certain transcription factors (Hoppe et al. 2000; Piwko and Jentsch 2006; Schrader et al. 2009).

Rpt ring:

Crystallographic studies on the homohexameric Proteasome-Activating Nucleotidase (PAN) complex of Archaea, which is orthologous to the Rpt ring, have identified major structural features (Zhang et al. 2009b). A channel is formed at the center of the ring of ATPase domains, and within this channel are two “pore loops” that are likely to contact substrates (Figure 5A). When ATP is hydrolyzed, conformational changes of the ATPase domains are thought to move the pore loops along the axis of the channel, providing the driving force for substrate unfolding and translocation (Sauer and Baker 2011; Erales et al. 2012). The pore loops are expected to interact first with the initiation sites of the substrate, and then to track along the polypeptide as substrate translocation into the CP proceeds.

The Rpt proteins also contain oligonucleotide/oligosaccharide-binding (OB) domains (Zhang et al. 2009b), positioned on the N-terminal sides of the ATPase domains (Figure 5A). In the case of PAN, the OB domain self-assembles into a homohexameric ring complex (also known as the N ring). This ring is coaxial with the ATPase domain ring (Figure 5A). Most likely the OB ring serves as the substrate entry port of the proteasome, and the substrate’s initiation site must thread through the central channel of the OB ring before coming into contact with the pore loops of the ATPase domain. Whether the OB ring engages substrates or provides a passive pore, one likely function of this ring is to impose a stringent criterion on the length of a functional initiation sequence. The presence of the OB domain may allow for eukaryotic proteins to have significant stretches of unstructured sequence without being readily degraded by the proteasome.

The OB domain of PAN forms a trimer of dimers. Each dimer is asymmetric in that the peptide bond at Pro91 is in the trans configuration in one subunit but in cis in its partner. Pro91 is positioned between the coiled-coil and OB domains, so this kink in the trajectory of the main chain allows for the α-helical elements emerging from partnered OB domains to coalesce into a coiled coil. This trimer of dimers arrangement is evidently replicated in the yeast proteasome (Zhang et al. 2009b), with “cis-Rpt’s” alternating around the ring (Tomko et al. 2010).

Interface between the RP and CP:

The Rpt proteins belong to the ATPases Associated with a variety of cellular Activities (AAA) family of ATPases. A distinguishing feature of the AAA family is the C domain, which is positioned at the perimeter of the ATPase domain. The C-terminal “tails” of the Rpt proteins are thought to be flexible, and some or all of the tails emerge from the C domains and insert into the α pockets of the CP. A motif at the end of the tail, the HbYX motif, is found on three of the six Rpt proteins, and these three Rpts—Rpt2, Rpt3, and Rpt5—are critical for CP gating (Smith et al. 2007). The Rpt tails have been mapped to the α pockets into which they insert by cross-linking (Tian et al. 2011). Surprisingly, the interface has an asymmetric character, with fixed contacts between the Rpt2, Rpt6, and Rpt3 tails and the α pockets into which they insert and on the other side of the ring, a less defined pocket specificity among the other tails (Tian et al. 2011). The CP–RP interface is stabilized not only by the insertion of Rpt tails into α pockets, but also presumably by an interaction between Rpn6 and α2 (Lander et al. 2012; Pathare et al. 2012).

Blm10 and ubiquitin-independent protein degradation by the proteasome

Not all proteasome substrates require modification by ubiquitin. One example is ornithine decarboxylase (ODC, encoded by SPE1), which catalyzes the committed step in polyamine biosynthesis, and is under intricate feedback control (Kurian et al. 2011). ODC is antagonized by ODC antizyme (Oaz1). When polyamine levels are high, Oaz1 is induced and binds to ODC. This exposes a peptide in ODC that can serve as an initiation site; ODC is then unfolded by the RP and degraded by the CP (Takeuchi et al. 2008).

In contrast to ODC, the proteasome’s ubiquitin-independent substrates may typically be degraded without the participation of the RP. Other factors can replace the RP on the cylinder end of the CP, open the CP channel, and promote protein degradation (Finley 2009). The most conserved of these “CP activators” is Blm10, a 246-kDa HEAT-repeat protein (Schmidt et al. 2005). Approxiately 20% of proteasomes in yeast are hybrid RP–CP–Blm10 complexes (Schmidt et al. 2005). Blm10 binds to the cylinder end of the CP in the form of a turban and inserts its C-terminal HbYX element into the α5/α6 pocket to open the CP gate (Sadre-Bazzaz et al. 2010). An aperture in Blm10, though small, could provide access to the CP channel for an unfolded protein. Perhaps in this way, Blm10 promotes degradation of Sfp1, a transcriptional activator of ribosomal protein genes (Dange et al. 2011; Lopez et al. 2011). Blm10 also participates in assembly of the CP (Fehlker et al. 2003; Marques et al. 2007).

Proteasome activators such as Blm10 seem to lack both the capacity to recognize ubiquitin and to hydrolyze ATP. Their ability to promote protein degradation relies on opening of the CP channel, to provide access to substrate. They may preferentially catalyze the degradation of proteins that can bypass an ATP-dependent unfolding step, either because the substrate spontaneously unfolds at a high frequency or is constitutively unfolded (Dange et al. 2011).

Regulation of proteasome activity

The transcription factor Rpn4 recognizes consensus binding elements upstream of all genes encoding major proteasome components (Mannhaupt et al. 1999; Leggett et al. 2002). The protein is extremely unstable, being a substrate for the Ubr2 ligase (Wang et al. 2004; Ju et al. 2008), and Rpn4 is also degraded by the proteasome in a ubiquitin-independent pathway (Ju and Xie 2006; Ha et al. 2012). Consequently, when proteasome function is compromised, Rpn4 levels rise, leading to homeostatic restoration of proteasome activity (Xie and Varshavsky 2001; Metzger and Michaelis 2009; Wang et al. 2010). Under conditions of “proteasome stress,” proteasomes also exhibit altered composition (Park et al. 2011). Chronic upregulation of proteasome activity by overexpression of Rpn4 leads to extended replicative lifespan in yeast (Kruegel et al. 2011; see also Chen et al. 2006).

Proteasome Assembly

CP assembly

An early step in CP assembly is formation of the seven-membered α ring. This ring is then used as a template for assembly of the β ring. The resulting structures, or “half-mers,” are subsequently joined through β ring–β ring interactions to form the mature α7β7β7α7 CP. The proteolytic sites of the CP are held in an inactive state until the α7β7β7α7 complex is fully assembled, so that the proteolytic sites are never active unless sequestered from the cytoplasm. This pathway is ordered through the action of five dedicated assembly chaperones (Table 5) (Ramos et al. 1998; Le Tallec et al. 2007; Li et al. 2007c; reviewed by Kusmierczyk and Hochstrasser 2008).

View this table:
Table 5  Assembly chaperones for the proteasome

The Pba1Pba2 heterodimer binds the outer, RP-binding surface of the α ring, and the Pba3Pba4 heterodimer the inner surface, which abuts the β ring in the mature particle. Interestingly, Pba1 and Pba2 contain HbYX motifs, suggesting that they may suppress premature Rpt tail insertion into nascent CP species (Kusmierczyk et al. 2011). The crystal structure of a Pba3Pba4–α5 ternary complex indicates that these chaperones occlude interaction surfaces between the α and β rings (Yashiroda et al. 2008). However, the Pba3Pba4 heterodimer has more complex effects on assembly, since its absence results in the substitution of α4 for α3 in a subset of proteasomes (Kusmierczyk et al. 2008).

The three catalytically active β subunits, as well as two of the catalytically inactive subunits, are synthesized with N-terminal propeptides. Propeptide removal follows upon the joining of two half-mers, reflecting that formation of the interface between β rings is required for the proteolytic sites to acquire catalytic activity (Arendt and Hochstrasser 1997). Interestingly, the propeptide of β5 is essential for this subunit’s incorporation into the CP (Chen and Hochstrasser 1996). The β5 propeptide also interacts physically with Ump1 (Heink et al. 2005), a chaperone that suppresses half-mer dimerization until the β ring is complete (Li et al. 2007c). The β ring is completed with the addition of the β7 subunit (Marques et al. 2007). This subunit has a C-terminal tail that reaches to the neighboring β ring and inserts into the interface between β1 and β2. As half-mers are joined, Ump1 is encapsulated in the nascent CP and degraded (Ramos et al. 1998).

RP assembly

The base and lid appear to have independent assembly pathways, and are joined to form the RP late in the pathway. Base assembly involves four dedicated and evolutionarily conserved chaperones, which are not found in mature proteasomes (Table 5) (Funakoshi et al. 2009; Kaneko et al. 2009; Le Tallec et al. 2009; Park et al. 2009; Roelofs et al. 2009; Saeki et al. 2009a). Each of these “RP chaperones” binds to the C domain of an Rpt protein, which constitutes a notable example of convergent evolution, because they have no sequence or structural homology.

The base is assembled from three precursor complexes or modules. Each module is defined by a pair of Rpt proteins, containing one cis and one trans subunit with respect to Pro91 (Rpt1Rpt2, Rpt3Rpt6, and Rpt4Rpt5) (Funakoshi et al. 2009; Kaneko et al. 2009; Saeki et al. 2009a; Tomko et al. 2010). Thus, the slow steps in Rpt ring assembly are those involving the presumably weak interdimeric interfaces. Each module contains at least one RP chaperone as well: the Rpt1Rpt2 module is found with Hsm3 bound to Rpt1, whereas the Rpt4Rpt5 module has Nas2 on Rpt5. The final module has both Nas6 and Rpn14 bound to Rpt3 and Rpt6, respectively.

The Rpt C-terminal tails appear to be critical for assembly; deletion of a single amino acid from the C terminus of either Rpt6 or Rpt4 leads to a dramatic defect in RP formation (Park et al. 2009). The Rpt proteins that have strong effects on gating—Rpt2, Rpt3, and Rpt5—have, by contrast, little effect on proteasome assembly. The assembly phenotypes of the rpt6 and rpt4 mutants suggest a role of the CP in RP assembly, and indeed, RP assembly is defective in several mutants whose primary defect is in CP assembly (Kusmierczyk et al. 2008; Park et al. 2011). Such CP mutants do not interfere nonspecifically with RP assembly but rather block an early step of Rpt ring formation, consistent with a templating model (Park et al. 2011)

How the chaperones promote proper base assembly remains to be solved. Interestingly, when bound to an Rpt C domain, some or all of the RP chaperones may project partly in the direction of the CP (Roelofs et al. 2009; Barrault et al. 2012). Chaperone binding is thus hypothesized to occlude contacts between the Rpt tail and its cognate α pocket, thus minimizing the formation of premature or incorrect RP–CP contacts (Roelofs et al. 2009). The RP chaperones Hsm3, Nas6, and Rpn14 do not bind the Rpt tail, only the proximal C domain (Roelofs et al. 2009; Takagi et al. 2012). Thus, they may occlude the tail by virtue of the proximity of the C domain to the tail. Consistent with this idea, Rpn14 remains stably associated with the proteasome when the C-terminal amino acid of Rpt6 is deleted (presumably leading to poor engagement of this tail), and Nas6 remains associated with proteasomes in the C-terminal mutant of Rpt3 (Park et al. 2009). Thus, negative regulation of the insertion of Rpt C termini into the α pockets of the CP may be a key mechanism of chaperone action. Nas2 appears to bind both the C domain and the tail itself, so it may well conform to the model (Lee et al. 2011). Some chaperones may positively regulate lateral interactions between ATPases as well, as indicated by recent work on Hsm3 (Barrault et al. 2012).

Whereas base assembly is guided by multipe chaperones, no chaperones have been identified for lid assembly. A landmark in lid assembly is the incorporation of Rpn12, the last subunit to join the complex (Fukunaga et al. 2010; Tomko and Hockstrasser 2011). The arrival of Rpn12 converts the nascent lid into a state competent to join with the base to form the RP. This property of Rpn12 is likely accounted for by direct contacts between this subunit and the base (Tomko and Hockstrasser 2011).

Cdc48 ATPase

An essential factor involved widely in ubiquitin-dependent processes is the chaperone Cdc48 (Meyer et al. 2012). This enzyme’s ortholog in mammalian cells is p97 or valosin containing protein (VCP). Cdc48 belongs to the AAA family of ATPases (Halawani and Latterich 2006; Ye 2006; Jentsch and Rumpf 2007). It comprises two AAA ATPase domains, D1 and D2, and a terminal N domain (Figure 6). The chaperone assembles into cylindrical homohexamers that undergo nucleotide-dependent conformational changes, predominantly between the N and D1 domains (Pye et al. 2006). Genetic defects in the chaperone give rise to VCP disease (Ju and Weihl 2010), a progressive autosomal disorder associated with inclusion body myopathy, Paget disease of the bone, and frontotemporal dementia, accompanied by a marked accumulation of polyubiquitylated proteins.

Figure 6 

Structure of p97/Cdc48. Left: Ribbon representations of full-length p97. Top and side views are shown. The N, D1, and D2 domains are indicated in different colors. Right: Ribbon representations of p97 N and D1 domains interacting with p47. Top and side views, as at left. These images were reproduced with permission from Dreveny et al. (2004).

The first link of Cdc48 to ubiquitin was found in a screen for stabilizing mutants in the N-end rule pathway (Ghislain et al. 1996). However, Cdc48 participates in diverse cellular processes such as cell cycle progression, homotypic membrane fusion, DNA repair, and transcription factor processing. To support this wide range of functions, ancillary proteins regulate Cdc48 activity toward individual substrates in a spatially and temporally controlled manner (Schuberth and Buchberger 2008). The majority of ancillary factors are ubiquitin receptors that deliver ubiquitylated proteins to Cdc48, suggesting that it acts downstream of ubiquitylation and upstream of the proteasome. At least some Cdc48-associated proteins bind mutually exclusively to Cdc48 and thus define functionally distinct subcomplexes (Schuberth and Buchberger 2008). Interestingly, Cdc48 action may also lead to proteolysis in the vacuole. Cdc48 acts on substrates modified with the ubiquitin-like molecule Atg8 in the course of macroautophagy (Krick et al. 2010) and it plays a role in ribophagy under starvation conditions (Ossareh-Nazari et al. 2010).

Proteins that associate with Cdc48 are classified according to their Cdc48-binding domains. For example, regulatory cofactors containing ubiquitin regulatory X (UBX) domains or suppressor of high-copy PP1 (SHP) boxes bind to distinct regions within the N domain (Schuberth and Buchberger 2008). In contrast, several proteins harboring a peptide:N-glycanase/UBA or UBX-containing (PUB) or PLAP, Ufd3, and Lub1 (PUL) domain bind near the C terminus of Cdc48 (Madsen et al. 2009).

Cdc48’s mechanism of action is not fully understood. Initial insight derived from analysis of the oleic acid (OLE) pathway, which controls the synthesis of unsaturated fatty acids in yeast, from the ERAD pathway, and from analysis of membrane fusion events. Transcription of OLE1 is driven by Spt23 and its homolog Mga2. Spt23 is synthesized as an inactive precursor (p120), which is anchored in the membrane of the ER. In the absence of unsaturated fatty acids, p120 is ubiquitylated by the ubiquitin ligase Rsp5 and cleaved by the proteasome. The resulting p90 fragment lacks a transmembrane anchor and can thus drive transcription of OLE1 (Hoppe et al. 2000). A complex of Cdc48 and the cofactors Ufd1 and Npl4 is involved in Spt23 activation. After activation, Spt23 exists in the ER membrane as a homodimer, only one subunit of which is processed. Cdc48Ufd1Npl4 binds and mobilizes the ubiquitylated and processed p90, separating it from the unprocessed Spt23. These data suggested that Cdc48 acts as a segregase to disassemble protein complexes (Rape et al. 2001; Jentsch and Rumpf 2007; Shcherbik and Haines 2007).

In support of a segregase function, Cdc48 is required to remove ubiquitylated Rpb1, the largest subunit of RNA Pol II, from chromatin. Rbp1 turnover is induced by UV (see below) and dependent on Cdc48, Ufd1, Npl4, Ubx4, and Ubx5 (Verma et al. 2011). In the absence of functional Cdc48, with Rpb1 degradation inhibited, ubiquitylated forms of Rbp1 are still delivered to the proteasome (Verma et al. 2011). These findings suggest that Cdc48 need not function strictly upstream of the proteasome but might act on proteasome-bound ubiquitin–protein conjugates. In some cases, however, the ATPases of the proteasome seem to be sufficient to extract a proteolytic target from a protein complex (see above). The mechanism whereby Cdc48 separates subunits of a protein complex from one another remains to be understood at the biochemical level. In particular, it is unresolved whether this activity involves threading of substrates through a central channel in Cdc48, in analogy to the mechanism of the proteasome.

In the turnover of ERAD substrates, Cdc48 acts subsequently to ubiquitylation but prior to the proteasome. The cofactors Ufd1 and Npl4 contain ubiquitin binding domains and participate in the export of polyubiquitylated proteins from the lumen and membrane of the ER (Bays et al. 2001b; Ye et al. 2001; Braun et al. 2002; Jarosch et al. 2002; Rabinovich et al. 2002). The integral membrane protein Ubx2 recruits Cdc48Ufd1Npl4 to the ER membrane and establishes its interaction with ubiquitin ligase complexes involved in ERAD (Neuber et al. 2005; Schuberth and Buchberger 2005). In membrane fusion processes, Cdc48 associates with the cofactor Shp1/Ubx1 (or, in mammals, with the homologous protein p47), to promote the homotypic fusion of membranes derived from the nucleus, the ER, and the Golgi apparatus (Hetzer et al. 2001).

Notably, some Cdc48 partner proteins modify the ubiquitin chains of bound substrates, thereby regulating the fate of these substrates. Thus, the antagonistic actions of the E4 enzyme Ufd2 and the DUB Otu1 determine the length of polyubiquitin chains on certain Cdc48 substrates, such as Spt23. According to a current model, Cdc48 accepts oligoubiquitylated substrates from ligases and then adjusts the length of a polyubiquitin chain prior to the substrate’s dissociation from the Cdc48Ufd1Npl4 complex. The handoff of ubiquitylated proteins from Cdc48 to the proteasome is facilitated by the interaction of Ufd2 with the UBL domains of Rad23 and Dsk2. Once dissociated from Ufd2, these UBL domains are free to deliver substrate to the proteasome (Richly et al. 2005; Rumpf and Jentsch 2006; Hänzelmann et al. 2010; see also Kim et al. 2004).

Substrate Recognition in the Ubiquitin Pathway

Quality-control protein degradation

One of the major functions of the ubiquitin–proteasome system is the disposal of misfolded and damaged proteins. Cells are highly sensitive to such proteins (Geiler-Samerotte et al. 2011) and possess several mechanisms, in addition to the ubiquitin–proteasome system, to neutralize them (Liu et al. 2011). Misfolded proteins are often localized in subcellular compartments that may either reduce the toxicity of these proteins or promote efficient quality-control protein turnover (Kaganovich et al. 2008). Several E3s involved in protein quality control have been discovered. They include the nuclear quality control ligase San1 (Gardner et al. 2005a) and the endoplasmic-reticulum–associated degradation E3s Hrd1 and Doa10 (see below). The principal quality control ligases in the cytoplasm appear to be Ubr1, Hul5, and the ribosome-bound ubiquitin ligase Rkr1/Ltn1 (Eisele and Wolf 2008; Bengtson and Joazeiro 2010; Heck et al. 2010; Nillegoda et al. 2010; Fang et al. 2011). In addition, Ubr2 may contribute to removal of misfolded cytopasmic proteins (Nillegoda et al. 2010).

How do these ubiquitin ligases recognize their substrates? San1 appears to bind directly to misfolded proteins through exposed hydrophobic patches, without the need for chaperones (Fredrickson et al. 2011; Rosenbaum et al. 2011). This interaction is mediated by multiple binding sites for substrates with different properties, embedded into intrinsically disordered regions in San1 (Rosenbaum and Gardner 2011; Rosenbaum et al. 2011). San1 ubiquitylates a range of mutant and thus presumably misfolded nuclear proteins, and deletion of the SAN1 gene induces a cellular stress response (Gardner et al. 2005a). Remarkably, even some misfolded cytoplasmic proteins are subject to San1-dependent degradation after Hsp70-dependent import into the nucleus (Prasad et al. 2010), suggesting an important function of the nucleus in ubiquitin/proteasome-mediated protein quality control. In contrast to San1, Ubr1 is dependent on molecular chaperones in its quality control functions (Heck et al. 2010; Nillegoda et al. 2010). Ubr1 might bind molecular chaperones and employ them for substrate recognition as described for the mammalian cytoplasmic quality control ligase Carboxyl terminus of Hsp70-Interacting Protein (CHIP) (McDonough and Patterson 2003). Alternatively, chaperones may promote substrate solubility or conformations that are directly recognized by Ubr1.

How Hul5 recognizes misfolded proteins is unknown. Hul5 is proposed to function as an E4 in the context of the proteasome (Crosas et al. 2006; see above), and it is thus possible that as yet unidentified E3s cooperate with Hul5 in substrate recognition and ubiquitylation. Consistent with this idea, hul5Δ mutants primarily affect poly- but not monoubiquitylation of misfolded proteins during heat stress (Fang et al. 2011).

The RING E3 Rkr1/Ltn1 is associated with ribosomes and ubiquitylates aberrant proteins arising from mRNAs that lack stop codons (Bengtson and Joazeiro 2010). Such nonstop mRNAs can result from errors in gene expression, and their poly(A) tails are translated into polylysine tracts. The positive charge of polylysine induces translational pausing due to strong electrostatic interaction with the negatively charged ribosome exit channel (Lu and Deutsch 2008). The resulting translationally paused or arrested nascent polypeptides seem to be targeted by Rkr1/Ltn1 (Bengtson and Joazeiro 2010). The precise mechanism is unknown, but translational pausing may transmit a conformational change to the surface of the ribosome that can be recognized by Rkr1/Ltn1.

Protein quality control in the endoplasmic reticulum

Proteins of the secretory pathway enter the ER through the Sec61 channel in an unfolded state and adopt their native conformation after clearing the channel. A protein quality-control system retains immature molecules in the ER until folding is completed. Terminally misfolded polypeptides are singled out by the ER protein quality-control system and routed to the cytoplasm for degradation by the ubiquitin–proteasome system. This highly conserved process, ERAD, promotes cellular homeostasis by preventing the accumulation and eventual aggregation of defective proteins within the secretory pathway (Hirsch et al. 2009; Buchberger et al. 2010; Smith et al. 2011b). Central to ERAD are membrane-bound ubiquitin ligases that are organized in multimeric protein complexes. They coordinate protein quality-control activities with cytoplasmic ubiquitylation, the action of the AAA–ATPase Cdc48, and the proteasome. The specificity of ERAD is primarily assured by substrate-recruitment factors that are integral components of ubiquitin ligase complexes. They selectively bind aberrant conformers and deliver them to downstream-acting factors (Meusser et al. 2005).

The ERAD pathway handles misfolded glycosylated and nonglycosylated proteins of the ER lumen, as well as membrane proteins, both single spanning and multispanning. The diversity of ERAD substrates is reflected in distinct pathways of ubiquitylation, as defined by individual E3 ligase complexes and their accessory factors: ERAD-C degrades proteins with defective cytosolic domains, ERAD-M targets lesions in transmembrane segments, and ERAD-L processes substrates with luminal defects (Vashist and Ng 2004). In yeast, two E3 ligase complexes, the HMG–CoA reductase degradation (HRD) ligase and Doa10, target this diverse pool of clients.

Ubiquitin ligase Doa10:

Doa10 (degradation of α2) was identified in a screen for factors required for the degradation of the soluble transcriptional repressor Matα2 (Swanson et al. 2001) and subsequently shown to act as well on ERAD-C substrates. Doa10 features an unusual N-terminal RING-finger domain and 14 transmembrane segments. This ligase functions with the E2 enzymes Ubc6 and Ubc7. Ubc6, a C-terminally anchored membrane protein (Sommer and Jentsch 1993), is also a target of Doa10-dependent turnover (Walter et al. 2001). Ubc7 is recruited to the ER membrane via Cue1 (Biederer et al. 1997). If Cue1 is missing, Ubc7 is mislocalized to the cytoplasm and targeted for proteasome-mediated degradation by Ufd4-dependent ubiquitylation. This turnover of Ubc7 is most likely signaled by a polyubiquitin chain synthesized on the E2 active site cysteine (Ravid and Hochstrasser 2007). Doa10 also associates with Ubx2, which stabilizes the interaction of Doa10 with Cdc48 (Neuber et al. 2005; Schuberth and Buchberger 2005). Doa10 is not only found in the ER membrane but also in the inner nuclear membrane, where it targets nuclear substrates for degradation (Deng and Hochstrasser 2006). As no substrate-recruitment factors for Doa10 have been identified to date, the selection of targets is perhaps accomplished by the ligase itself. Although distinct short-lived client proteins are processed by the HRD ligase and Doa10, some overlapping functions seem to exist: Double mutants of doa10 and hrd1 display enhanced cadmium sensitivity and show an activated unfolded protein response (Swanson et al. 2001; for further reading on the unfolded protein response (UPR), see Walter and Ron 2011).

HRD ubiquitin ligase:

ERAD-L and ERAD-M substrates are targeted by the HRD ligase. Key elements of this ligase complex have been identified in two genetic screens. In one of the screens, an ERAD-L substrate was used (Knop et al. 1996)—a mutant version of the vacuolar enzyme carboxypeptidase Y, CPY* (Finger et al. 1993). A mutation that stabilized CPY* was found in UBC7, which was the first indication that misfolded proteins of the ER lumen are degraded by cytoplasmic pathways (Hiller et al. 1996). Since UBC7 was also required for turnover of a mutant form of the translocation component Sec61 (sec61-2; Biederer et al. 1996) it became evident that ERAD-L and ERAD-M substrates can be degraded by the same cytoplasmic pathway. The other genetic screen was performed using an 3-hydroxy–3-methylglutaryl–CoA reductase (HMG-R) isozyme, Hmg2 (Hampton et al. 1996), which is not a bona fide misfolded protein. Instead, Hmg2 turnover is regulated through feedback control involving the mevalonate pathway. The two screens revealed overlapping genes (the “HRD” and the “DER” genes), indicating that Hmg2 is channeled into an ERAD pathway that also acts on misfolded proteins. Indeed, farnesol opens the conformation of Hmg2, which could make it accessible to the HRD ligase. Interestingly, the effect of farnesol requires an intact Hmg2 sterol-sensing domain (Shearer and Hampton 2005).

The HRD ligase complex is composed of at least six subunits and the requisite E2 enzymes (Figure 7A). The central component of this complex, Hrd1/Der3, comprises six transmembrane segments and ubiquitylates substrates at the cytoplasmic surface of the ER through its C-terminal RING-finger domain (Bordallo et al. 1998; Bays et al. 2001a). Similarly to Doa10, Hrd1 functions with Cue1-tethered Ubc7. Cue1 not only localizes with Ubc7 but also stimulates its enzymatic activity (Bazirgan and Hampton 2008). Ubx2, an additional component shared by Hrd1 and Doa10, connects Cdc48 to the HRD ligase pathway (Neuber et al. 2005; Schuberth and Buchberger 2005).

Figure 7 

HRD ubiquitin ligase. (A) HRD ubiquitin ligase consists of six core subunits: Hrd1 exposes a RING-finger domain on the cytoplasmic surface of the ER membrane and acts together with the E2 enzymes Ubc7/Cue1 and Ubc1 (both not depicted). Hrd3 together with Yos9 forms the ER luminal domain of the ligase complex. Usa1 bridges Hrd1 with Der1. Ubx2 binds Hrd1 and also, via a UBX domain, Cdc48. The transmembrane organization of the ligase complex suggests that it connects ER-luminal quality-control functions, dislocation, ubiquitylation, and the generation of pulling forces with proteolysis by the proteasome. (B) Hypothetical model of how the ER-luminal domain of the HRD ligase selects ERAD substrates. The glycans of misfolded proteins are processed by Htm1 to generate the glycan signal Man7GlcNAc2. Hrd3 first binds the misfolded protein in a “recruitment step” (left). Then Yos9 controls the identity of the glycan signal in a “commitment step” (center). Only when both interactions are productive is the client protein dislocated into the cytoplasm for proteasomal digestion.

Another key element of the HRD ligase is Usa1, comprising two transmembrane helices and a UBA domain. Usa1 mediates the interaction of Hrd1 with the small membrane protein Der1, which spans the membrane four times and is selectively required for the breakdown of ERAD-L substrates (Knop et al. 1996; Carvalho et al. 2006). Additionally, Usa1 acts as a scaffold that binds Hrd1 and promotes its dimerization. This function of Usa1 is generally required for proteolysis of ERAD-L and ERAD-M substrates (Horn et al. 2009; Carvalho et al. 2010).

In vivo cross-linking studies have suggested that Hrd1 may bind ERAD-L substrates (Carvalho et al. 2010). However, Hrd1 carries only small loops facing the lumen of the ER. Therefore, it is likely that the luminal substrate-binding module of the HRD ligase is formed primarily by Hrd3 and Yos9 (Denic et al. 2006; Gauss et al. 2006a,b). Hrd3 is a type I transmembrane protein that exposes an ∼80-kDa domain into the ER lumen. Yos9 interacts with the HRD ligase via Hrd3 and contains a mannose-6 phosphate receptor homology domain (MRH). The luminal Hrd3/Yos9 module links the ligase to the chaperone system of the ER by recruiting the Hsp70-type chaperone Kar2 to the E3 (Denic et al. 2006). Hrd3 binds misfolded CPY* irrespective of its glycan modifications and also in absence of Yos9. Therefore, Hrd3 was proposed to be the primary receptor for misfolded proteins at the ligase complex (Gauss et al. 2006b). Yos9 specifically binds terminal α1,6-bonded mannose moieties on misfolded glycoproteins (Quan et al. 2008). These are generated by Mns1 and Htm1, which convert Man9GlcNAc2 into Man7GlcNAc2 (Clerc et al. 2009; Gauss et al. 2011). The binding characteristics of Hrd3 and Yos9 reflect the key features of degradation signals in ERAD substrates, one being misfolding of the client, probably recognized by hydrophobic interactions, and the other, a specific glycan signal, a Man7GlcNAc2 modification (Figure 7B). Since glycoproteins that are not processed by Mns1 and Htm1 are protected from degradation, these two mannosidases act as a timer that allows newly synthesized proteins to be distinguished from those that have failed to fold correctly (Jakob et al. 1998). While these data apply to glycan-modified ERAD-L model substrates, targeting of ERAD-M client proteins may differ. A mutational analysis of the Hrd1p membrane anchors indicated that the transmembrane segments may play a crucial role in detecting misfolding of ERAD-M substrates (Sato et al. 2009).

Although the Hrd1 and Doa10 ligases exhibit similar activities in ERAD, their topological organization is different. While it is likely that the domains of Doa10 involved in substrate selection and ubiquitin conjugation both reside in the cytoplasm, these domains are separated by the ER membrane in the case of the HRD ligase. Thus, at least ERAD-L clients have to be exported from the ER prior to ubiquitylation. This process, termed dislocation or retrotranslocation, most likely involves a proteinaceous channel in the ER membrane. It has been speculated that such a channel may be formed by the components of the HRD ligase itself (Hampton et al. 1996; Swanson et al. 2001; Horn et al. 2009). However, a function of the translocon in dislocation has also been proposed, based on a physical interaction between Hrd3 and Sec61 (Schafer and Wolf 2009). Moreover, an apparent interaction of CPY* with Sec61 is maintained until the misfolded protein is ubiquitylated on the cytoplasmic surface (Schafer and Wolf 2009). These findings support previous genetic data pointing to a function of Sec61 in ERAD (Plemper et al. 1997).

Degradation signals

Mechanisms of substrate selection by ubiquitin ligases are diverse and rely on a variety of degradation signals (degrons) (reviewed in Ravid and Hochstrasser 2008). Generally we can distinguish between signal-specific degrons on regulatory proteins and degrons controlled by protein folding and assembly. We briefly discussed the latter in the previous section, as they are key to protein quality-control pathways.

The first systematically studied and perhaps most surprising degrons are determined by the N-terminal amino acid residue of the substrate protein (Bachmair et al. 1986; Varshavsky 2011). The N-end rule ubiquitin ligase Ubr1 (Bartel et al. 1990) binds proteins with different affinities depending on the side chain of the first amino acid and thereby relates the protein’s N terminus to protein stability (Table 6) (Choi et al. 2010). The specificity of Ubr1 is essentially complementary to that of methionine aminopeptidases, so that newly synthesized proteins will rarely present destabilizing residues; if the penultimate residue is destabilizing, methionine aminopeptidase will not remove the initiator methionine. Rather, destabilizing N-terminal residues are formed as a result of endoproteolytic cleavage by proteases such as separase (see below), or other post-translational events (Varshavsky 2011). For example, acidic N-terminal residues generated by endoproteases are not recognized by Ubr1, but are substrates for Ate1, an enzyme that ligates arginine to the substrate’s N terminus. This allows for subsequent Ubr1-mediated recognition, ubiquitylation, and degradation. N-terminal glutamine and asparagine residues are funneled into the N-end rule pathway by the action of Nta1, an N-terminal amidase, whose reaction products are in turn substrates for Ate1 (Baker and Varshavsky 1995). These pathways mediate major regulatory events in many eukaryotes, such as the sensing of oxygen and nitric oxide levels (Licausi et al. 2011; Varshavsky 2011).

View this table:
Table 6  N-end rule in Saccharomyces cerevisiae

Acetylated N termini, which are found in most proteins, are not recognized by Ubr1. Instead, they present separate degrons recognized by the Doa10 ligase (Hwang et al. 2010). Many proteins are metabolically stable despite the presence of these targeting elements, presumably due to poor exposure of their N termini, suggesting a potential involvement of this pathway in recognition of misfolded proteins and quality control degradation (Hwang et al. 2010).

Many other regulated degradation pathways also use post-translational modifications to activate degrons. For example, phosphorylation often generates a high-affinity interaction site for E3 recruitment. Such phosphodegrons are widely used by SCF ligases (Petroski and Deshaies 2005; Zimmerman et al. 2010; Duda et al. 2011). The cell cycle inhibitor Sic1 contains an array of phosphodegrons with relatively low affinities for SCFCdc4. This arrangement requires processive multiphosphorylation by the G1 and S-phase kinases Cln2/Cdc28 and Clb5/Cdc28 for efficient Sic1 degradation and transforms the graded kinase activity into a switch-like cell-cycle transition (Nash et al. 2001; Petroski and Deshaies 2003; Koivomagi et al. 2011).

Another interesting variety of degron formed by post-translational modification is recognized by the small ubiquitin-related modifier (SUMO)-targeted ubiquitin ligases (STUbLs) (Perry et al. 2008). These E3s contain SUMO interacting motifs that mediate binding to SUMOylated substrate proteins for ubiquitylation (see below).

Other degrons are less well defined and include short surface-exposed hydrophobic stretches such as the N-terminal degron in Matα2, which can be masked by heterodimerization with Mata1 (Johnson et al. 1998).

Sites in target proteins that recruit E3s and allow ubiquitin conjugation constitute the canonical form of a degradation signal in the ubiquitin–proteasome system. These sites are remarkably varied, consistent with the multiplicity of ubiquitin ligases and their diverse substrate recognition modes. However, additional features to support proteasome-mediated degradation of the substrate are also critical, particularly an unfolded segment to serve as an initation site for the proteasome, as discussed above (Prakash et al. 2004).

Ubiquitylation of Membrane Proteins

As discussed earlier, membrane-associated ubiquitin ligases play key roles in the protein quality-control pathway of the ER. However, other membrane systems of the cell are also sites of abundant ubiquitylation, where it acts to direct protein sorting. On the one hand, ubiquitylation drives transport from the trans-Golgi and the plasma membrane. On the other, it helps to concentrate proteins in the MVB compartment. The MVB sorting step leads ultimately to proteolysis of the cargo—not in the proteasome, however, but in the vacuole (Lauwers et al. 2010).

Ubiquitin function in endocytosis

The abundance of receptors and transporters at the plasma membrane is regulated by endocytosis, often in a signal-dependent manner. Internalized proteins are transported to the endocytic compartment. From there, they are either recycled to the plasma membrane or packaged into multivesicular bodies for delivery to lysosomes. Ubiquitin serves as an important internalization signal for endocytosis. In some cases, ubiquitin seems to act redundantly with other signals.

A function of ubiquitin in protein sorting at the plasma membrane was suggested by the observation that ubiquitylated Ste6, the yeast pheromone transporter, accumulated at the plasma membrane when endocytosis is blocked (Kölling and Hollenberg 1994). Moreover, ubiquitylation was found to be necessary and sufficient for ligand-induced endocytosis of the pheromone receptor Ste2 (Hicke and Riezman 1996). Accordingly, the nitrogen permease inactivator Npi1 was identified as the HECT domain ligase Rsp5 (Hein et al. 1995; Huibregtse et al. 1995). A number of other plasma membrane proteins were subsequently shown to undergo ubiquitylation, including the permeases Fur4 and Gap1.

After Rsp5-dependent selection and modification of the cargo, it is most likely recognized by endocytic adaptors (Shih et al. 2002). Yeast endocytic adaptors Ent1 and Ent2 bind ubiquitylated cargo via UIM domains and localize to the plasma membrane by interacting with phosphatidylinositol-(4,5)-bisphosphate and clathrin. An additional endocytic scaffold protein, Ede1, an EH domain protein, contains a UBA domain and it may also contribute to cargo interaction and concentration (Dores et al. 2010).

Several nutrient permeases are ubiquitylated on multiple lysines by short K63 ubiquitin chains, consistent with the linkage specificity of Rsp5 (Kim and Huibregtse 2009). Though a single ubiquitin molecule is sufficient to promote endocytosis, multiple monoubiquitylation and short K63 chains accelerate the rate of endocytosis, to an extent that may depend on the substrate (Galan and Haguenauer-Tsapis 1997; Springael et al. 1999). Notably, the Jen1 transporter shows a strict requirement for K63 chains (Paiva et al. 2009).

Rsp5 contains WW domains, named for the presence of two highly conserved tryptophan residues, which directly interact with PPx(Y/F) motifs in substrate proteins. However, many cargo molecules, including several permeases, do not carry such a motif. In these cases, interaction of the cargo with Rsp5 is mediated by a family of adaptor proteins, such as the arrestin-related trafficking adaptors (ARTs). Unlike mammals, yeast does not have canonical arrestins, in that the yeast ART proteins lack adaptin and clathrin-binding sequences. Instead they interact with Rsp5 through a PxY sequence (Lin et al. 2008; Leon and Haguenauer-Tsapis 2009; Nikko and Pelham 2009). ART proteins (Art1Art10) regulate the ubiquitylation of specific cargos at the plasma membrane in response to specific stimuli and may be part of a quality-control system that targets damaged and misfolded membrane proteins for degradation in the vacuole. This variety of adaptors explains how a single ubiquitin ligase can regulate endocytosis of many different proteins. Another level of regulation is provided by differential localization of the Rsp5 adaptors. For example, the adaptors Bul1 and Bul2 work both at the plasma membrane and the trans-Golgi (Nikko and Pelham 2009). The Art1Art10 proteins (Lin et al. 2008; Nikko and Pelham 2009; MacGurn et al. 2011) function mainly at the plasma membrane, while the Rsp5 adaptor proteins Ear1, Ssh4, Bsd2, Tre1, and Tre2 are located mainly at endosomes (Liu et al. 1997; Stimpson et al. 2006; Leon et al. 2008).

Nutrient uptake is controlled by a regulatory loop that adjusts the level of amino acid transporters at the plasma membrane through Art1. Npr1 (nitrogen permease reactivator 1 kinase) phosphorylates residues near the N terminus of Art1 to inhibit its transport to the plasma membrane. Endocytosis of amino acid transporters is thereby suppressed. Npr1 itself is negatively regulated by the TORC1 kinase, which thus acts on ubiquitin-dependent endocytosis to fine tune the activity and composition of proteins of the plasma membrane (MacGurn et al. 2011).

Function of ubiquitin in the MVB pathway

Proteins of the late endosome destined for proteolysis are sorted into multivesicular bodies, which deliver cargo to the yeast vacuole for degradation (Katzmann et al. 2001; Henne et al. 2011). Formation of these vesicles involves invagination of membranes into the endosomal compartment. Mature MVBs subsequently fuse with the lysosome and release their contents. Crucial players in the MVB pathway are specific multisubunit endosomal sorting complexes required for transport (ESCRT): ESCRT-0, ESCRT-I, ESCRT-II, ESCRT-III, and the AAA ATPase Vps4 (part of the fifth ESCRT complex). These act sequentially (Katzmann et al. 2001; Babst et al. 2002a,b) in early cargo recruitment and concentration (ESCRT-0, -I, and -II) and later in cargo deubiquitylation and membrane sculpting (ESCRT-III and Vps4).

A prerequisite for selective sorting and concentration of membrane proteins into endosomal microdomains and, eventually MVBs, is the ubiquitylation of cargo proteins (Katzmann et al. 2001; Lauwers et al. 2010; Henne et al. 2011). ESCRT complexes have several distinct ubiquitin-binding motifs for cargo recognition. Subunits of the ESCRT-0 complex bind ubiquitin in several ways: Vps27 contains a VHS (Vps27 Hrs STAM) and two UIM domains in tandem, while the Hse1 subunit contains both a UIM and a VHS domain (Bilodeau et al. 2002). Thus, ESCRT-0 contains five ubiquitin-binding domains. However, it remains unclear whether this allows binding of several cargoes simultaneously or binding with high affinity to poly- or multiubiquitylated cargo (Ren and Hurley 2010). In addition to its ubiquitin-binding activity, ESCRT-0 may bind cargo through interactions with the clathrin vesicle machinery, suggesting microdomains in which clathrin lattices, ESCRT-0, and ubiquitylated cargo meet. ESCRT-I also contains ubiquitin-binding domains: the subunit Vps23 contains a UEV domain (Pornillos et al. 2002), and Mvb12 harbors a novel ubiquitin-binding domain (Shields et al. 2009). So far only one ubiquitin-binding domain has been identified in ESCRT-II, an Npl4 Zinc Finger (NZF) motif in Vps36. Also participating in ubiquitin-dependent sorting are the ubiquitin-binding adaptor proteins Gga1 and Gga2, although their exact roles in the pathway are not understood (Lauwers et al. 2009, 2010).

Though it was initially assumed that monoubiquitylation is sufficient to direct targets into the MVB pathway, it is now accepted that K63 ubiquitin chains are needed. Whether these ubiquitin moieties are added at the endosome or persist from endocytosis-associated ubiquitylation events at the plasma membrane is unresolved. In any case, this modification must be removed prior to packaging of cargo into vesicles to avoid depletion of ubiqutin by its uptake into the vacuole. In yeast, this step involves the DUB Doa4 (Amerik et al. 2000a). Doa4 is recruited into the ESCRT-III complex by the adaptor protein Bro1, which also stimulates the deubiquitylating activity of Doa4 (Luhtala and Odorizzi 2004; Richter et al. 2007).

In addition to ubiquitin binding, ESCRT components also serve as targets for ubiquitylation. For instance, Vps27 can be monoubiquitylated (Polo et al. 2002; Stringer and Piper 2011). Although the physiological function of this modification remains unknown, Hrs, the human homolog of Vps27, is inhibited by monoubiquitylation because an intramolecular interaction between the UIM and the ubiquitin modification prevents the binding of ubiquitylated cargo. Similar observations have been made for the endocytic adaptor Eps15, raising the possibility of a general role for monoubiquitylation in downregulation of these pathways (Hoeller et al. 2006).

Ubiquitylation and protein import into peroxisomes

Peroxisomes are small cytoplasmic vesicles housing ∼50 enzymes that mediate β oxidation of fatty acids and other metabolic processes. Luminal proteins of the peroxisome are imported post-translationally by a complex, receptor-mediated process (Girzalsky et al. 2010). The import apparatus includes three RING ligases (Pex2, Pex10, and Pex12) that reside in the peroxisomal membrane, as well as as a cognate E2 (Ubc10/Pex4) (Williams et al. 2008; Platta et al. 2009). The principal role of ubiquitin in peroxisomes is apparently the monoubiquitylation of Pex5, a receptor for protein import into peroxisomes that delivers cargo to the peroxisome by cycling between the cytoplasm and the peroxisomal membrane (Carvalho et al. 2007; Platta et al. 2007; Grou et al. 2008). After cargo delivery, Pex5 must return to the cytoplasm for another round of import. This recycling step is dependent on Pex5 ubiquitylation. Other peroxisomal import receptors such as Pex18 and Pex20 are ubiquitylated and potentially follow a similar cycle. A heteromeric protein complex of the AAA family, composed of Pex1 and Pex6 monomers, mediates ATP-dependent extraction of Pex5 into the cytosol by an unknown mechanism (Platta et al. 2005). A hypothetical model accounting for these data are that the Pex1/Pex6 complex recognizes Pex5 via its ubiquitin modification, and functions analogously to Cdc48 in its extraction of membrane proteins from the ER in the ERAD-M pathway, as described above.

Nuclear Functions of the Ubiquitin System

Nuclear functions of ubiquitin fall into two general categories: on one hand, there is a nuclear form of the above-described protein quality-control pathways. On the other hand, as described below, ubiquitylation contributes, both in its proteolytic and its noncanonical mode, to virtually all aspects of DNA metabolism (Ulrich 2002), such as DNA replication and repair (Bergink and Jentsch 2009; Ulrich and Walden 2010), gene expression and chromatin structure (Muratani and Tansey 2003; Shilatifard 2006), and chromosome dynamics and segregation (Pines 2006).

Coupling cell cycle progression to DNA replication and chromosome segregation

DNA replication and chromosome segregation are intimately coupled to cell cycle progression and hence subject to regulation by the ubiquitin system, mostly by means of proteolytic destruction of important regulators. These processes have been discussed in detail in excellent reviews (Nasmyth 1996; Pines 2006; Diffley 2010).

Replication initiation:

At the G1-to-S transition, Sic1 degradation, initiated by SCFCdc4, allows the activation of the cyclin-dependent kinase (CDK) Cdc28 in complex with the S-phase–specific, B-type cyclins, Clb5 or Clb6, leading to phosphorylation and activation of the replication initiation factors Sld2 and Sld3 (Tanaka et al. 2007; Zegerman and Diffley 2007). In parallel, SIC1 transcription is terminated by ubiquitylation and destruction of a transcriptional activator, Swi5, also mediated by SCFCdc4 (Kishi et al. 2008).

There is also evidence for nondegradative contributions of ubiquitin to replication initiation. The cullin Rtt101 associates with early replication origins and ubiquitylates Spc16, a subunit of the FAcilitator of Chromatin Transactions (FACT) complex, predominantly via K63-linked chains (Han et al. 2010). FACT is a histone chaperone with functions in transcription, DNA replication, and repair (Winkler and Luger 2011). Deletion of RTT101 results in a weakening of the interactions between FACT and the replicative helicase (the Mcm2-7 hexamer), and a partial loss of both complexes from a subset of replication origins (Han et al. 2010). Whether or not ubiquitylation of Spc16 is responsible for this phenomenon has not been determined, but the activity of Rtt101 appears to impinge specifically on the replication-related functions of FACT.

Origin licensing:

The mechanism that limits replication to a single round per cell cycle is called origin licensing. This restricts the activation of replication origins to S phase and prevents renewed firing until the next cell cycle. All organisms use multiple strategies to achieve this goal, including ubiquitin-mediated proteolysis of key regulatory factors (Diffley 2010). In budding yeast, the primary target is Cdc6, a component of the prereplicative complex (pre-RC). Cdc6 is phosphorylated in late G1 and S phase, which causes ubiquitylation by SCFCdc4 and subsequent degradation (Drury et al. 1997). While in mammalian cells the pre-RC component Cdt1 is subject to proteolysis, its yeast homolog Tah11 is instead inactivated by export from the nucleus (Diffley 2010).

Chromosome segregation:

One of the most important features of cell division is the even distribution of replicated chromosomes to the daughter cells, a process controlled by ubiquitin-dependent proteolysis (Pines 2006). During this cell cycle stage, the dominant E3 is the APC/C, which couples mitosis to cytokinesis and ensures correct chromosome segregation (Harper et al. 2002). Important substrates include cyclins; components of the spindle checkpoint that monitor the correct assembly of the mitotic spindle; and the securin protein, Pds1, an inhibitor of separase (Esp1) that initiates anaphase by cleaving the cohesin subunit Scc1, thus allowing sister chromatid separation (Nasmyth et al. 2000). The C-terminal proteolytic fragment of Scc1 is subject to proteasomal degradation by the N-end rule pathway, initiated by ubiquitylation via the RING-finger E3 Ubr1 with the E2 Rad6, and interference with this process causes chromosome loss (Rao et al. 2001; see also Buonomo et al. 2000).

Responses to replication stress

Mechanisms of replication fork protection:

Whereas the coupling of replication initiation and origin licensing to the cell cycle mostly involves proteolytic functions of the ubiquitin system, the role of ubiquitylation in the course of replication appears more diverse. A number of ubiquitin ligases have been found to contribute to genome stability by protecting replication forks from stress, but their mechanisms of action are poorly understood.

The F-box protein Dia2 is a constitutive component of the replisome progression complex (RPC), tethered to the RPC components Mrc1 and Ctf4 by means of its N-terminal domain (Morohashi et al. 2009). Dia2 acts as a substrate adaptor for SCFDia2. Its association with replication forks appears to facilitate the replication of difficult templates and protects cells from DNA damage and replication stress (Mimura et al. 2009; Morohashi et al. 2009). Mrc1 and Ctf4 are ubiquitylated by SCFDia2 and seem to be degraded, but it is unclear whether these substrates are functionally critical (Mimura et al. 2009). Dia2 itself is an unstable protein, and it is stabilized by replication stress (Kile and Koepp 2010).

The cullin Rtt101, in addition to its role in replication initiation, also contributes to protecting replication forks from collapse when they encounter DNA lesions (Luke et al. 2006). Thus, rtt101Δ mutants are sensitive to DNA-damaging agents and unable to recover from damage-induced fork stalling. In response to DNA damage, Rtt101 forms a ubiquitin ligase with the RING finger protein Hrt1, the linker protein Mms1, and the putative substrate adaptor Mms22 (Zaidi et al. 2008). Interestingly, Rtt101’s function in replication fork protection appears unrelated to its action on the FACT complex, as the latter does not require the presence of Mms1 or Mms22 (Han et al. 2010). A second substrate adaptor, Crt10, is recruited to the Rtt101 cullin complex via Mms1 in a damage-independent manner and has been suggested to affect replication by regulating nucleotide levels (Fu and Xiao 2006; Zaidi et al. 2008).

An interesting example of cross-talk between ubiquitin and the small ubiquitin-related modifier SUMO has emerged from the identification of a class of E3s, called STUbLs, which recognize SUMO-modified proteins as targets for ubiquitylation (Perry et al. 2008). In budding yeast, the RING-finger proteins Hex3/Slx5 and Slx8 have been implicated in preventing the accumulation of DNA damage during replication (Zhang et al. 2006). They form a heterodimer that promotes the ubiquitylation and degradation of highly sumoylated cellular proteins (Uzunova et al. 2007; Xie et al. 2007). The SUMO moieties are recognized by SUMO-interacting motifs within Hex3. It remains to be resolved whether the contribution of SUMO-dependent ubiquitylation to replication fork protection is attributable to the removal of specific sumoylated proteins for regulatory purposes or to preventing bulk accumulation of potentially toxic poly-SUMO conjugates.

Control of DNA damage bypass:

An independent pathway for lesion processing during replication is called postreplication repair, DNA damage bypass, or DNA damage tolerance (Lawrence 1994). The process provides resistance to DNA-damaging agents, but is capable of generating genomic instability through damage-induced mutagenesis. It is initiated by ubiquitylation of the proliferating cell nuclear antigen (PCNA) Pol30 (Hoege et al. 2002), a homotrimeric sliding clamp that ensures processivity of the replicative polymerases and also acts as an interaction platform for a multitude of proteins involved in various aspects of DNA metabolism (Moldovan et al. 2007). The PCNA modification system (Figure 8) provides an example where mono- and polyubiquitylation at a single site elicit distinct cellular responses (Ulrich 2009), mediated by a range of ubiquitin receptors (Table 7).

Figure 8 

Modifications of the replication factor PCNA. During undisturbed replication, PCNA (blue ring shape) promotes processive DNA synthesis by replicative polymerases δ and ε (Pol δ/ε), and is modified by SUMO (red lollipop shape). The modification prevents binding of Eco1, but causes the recruitment of Elg1 and Srs2. Srs2 prevents the formation of the recombinogenic Rad51 filament (51), inhibiting unscheduled recombination at replication forks. Upon damage-induced replication fork stalling, PCNA is modified by mono- and polyubiquitin (black lollipop shapes) at postreplicative daughter-strand gaps. Monoubiquitylation recruits damage-tolerant DNA polymerases (TLS) for translesion synthesis, while K63 polyubiquitylation causes recruitment of Mgs1 and initiates damage bypass by template switching in an unknown manner. Conjugating enzymes, ligases, and DUBs are highlighted in shades of purple, green, and pink, respectively.

View this table:
Table 7  Ubiquitin receptors in the DNA damage response

Monoubiquitylation at a single conserved lysine, K164, mediated by the E2–E3 complex Rad6Rad18 (Hoege et al. 2002), is a prerequisite for a process called translesion synthesis (Stelter and Ulrich 2003). This reaction involves a series of specialized DNA polymerases capable of using damaged DNA as a template for DNA synthesis (Waters et al. 2009). Although there is evidence for ubiquitin-independent translesion synthesis in vertebrates, the principle by which ubiquitylated PCNA activates damage-tolerant polymerases appears to be conserved: a series of ubiquitin-binding domains of the Ubiquitin-Binding Zinc Finger (UBZ) or Ubiquitin-Binding Motif (UBM) type, present in a subset of the polymerases (Table 7), affords enhanced affinity for the monoubiquitylated form of PCNA and thereby allows their recruitment and activation in response to DNA damage (Bienko et al. 2005). In budding yeast, this applies to polymerase η (encoded by RAD30), which mediates error-free translesion synthesis over UV-induced lesions, and Rev1, which in cooperation with polymerase ζ (encoded by REV3 and REV7) is responsible for a large part of damage-induced mutagenesis (Garg and Burgers 2005; Guo et al. 2006; Parker et al. 2007).

The consequences of PCNA polyubiquitylation are less well defined. The modification is a prerequisite for an error-free pathway of template switching (Hoege et al. 2002), which mediates damage bypass by avoiding the use of damaged DNA as a replication template. PCNA polyubiquitylation involves the synthesis of a K63-linked chain by the E3 Rad5 in cooperation with the heterodimeric E2 Ubc13Mms2 (Hoege et al. 2002; Parker and Ulrich 2009). The modification is likely to serve a nondegradative function (Zhao and Ulrich 2010) and enhances the affinity of an ATPase, Mgs1, for PCNA (Hishida et al. 2006; Saugar et al. 2012). However, the recruitment of this protein to sites of replication problems by means of its UBZ domain (Table 7) cannot fully explain the function of PCNA polyubiquitylation in damage bypass.

PCNA ubiquitylation is induced by DNA damage and replication stress, which involves the recruitment of Rad18 to stretches of single-stranded DNA covered by replication protein A (Davies et al. 2008). Yet, unlike the Dia2- and Rtt101-dependent mechanisms discussed above, damage bypass can be separated from bulk genome replication (Daigaku et al. 2010; Karras and Jentsch 2010), indicating that it operates on postreplicative daughter-strand gaps rather than directly at the fork.

The function of PCNA is further diversified by additional modifications: Attachment of the small ubiquitin-like modifier SUMO (Smt3 in yeast) occurs constitutively during replication and involves predominantly the same site that is targeted for damage-induced ubiquitylation, K164 (Hoege et al. 2002; Parker et al. 2008). This SUMOylation event prevents unscheduled recombination by recruiting an antirecombinogenic helicase, Srs2 (Papouli et al. 2005; Pfander et al. 2005). As with the recruitment of damage-tolerant polymerases by monoubiquitylated PCNA, Srs2 is targeted to SUMO-modified PCNA by means of tandem receptor motifs that independently recognize SUMO and PCNA (Armstrong et al. 2012). Under conditions of replication stress, Srs2 thus allows damage processing by ubiquitin-dependent bypass. At the same time, PCNA sumoylation enhances the affinity of an alternative clamp loader, Elg1, for PCNA (Parnas et al. 2010) and prevents the interaction of PCNA with an acetyltransferase important for the establishment of sister chromatid cohesion, Eco1 (Moldovan et al. 2006). In contrast to ubiquitylation, the functions of PCNA sumoylation are unlikely to be fully conserved in vertebrates, although an Srs2-related protein, bearing SUMO- and PCNA-interacting motifs, was recently identified in humans and shown to restrict unscheduled homologous recombination (Moldovan et al. 2012).

DNA repair

Among the DNA repair pathways, NER specializes in removing bulky lesions from double-stranded DNA by means of excising the damaged stretch and filling the resulting gap by DNA synthesis (Hoeijmakers 2001). The pathway operates in two distinct modes, depending on the way in which lesions are initially recognized. In global genome repair (GGR), Rad4 serves as the principal lesion recognition factor, in complex with its binding partner Rad23 (Figure 9A). In transcription-coupled repair (TCR), an RNA polymerase II stalled at a lesion initiates the events that lead to preferential repair of actively transcribed genes (Figure 9, B and C). Both branches are influenced by the ubiquitin–proteasome system.

Figure 9 

Ubiquitylation during nucleotide excision repair. (A) For global genome repair, lesions are recognized by Rad4 in complex with Rad23. Ubiquitylation of Rad4 is important for subsequent steps of repair. Ubiquitylated Rad4 is degraded by the proteasome. (B) Lesions on the transcribed strand of actively expressed genes are repaired by transcription-coupled repair, where RNA polymerase II (RNA Pol II) contributes to lesion recognition. Following removal of the enzyme by the action of Rad26, strand unwinding, excision of the lesion and resynthesis proceed as in global genome repair. (C) An irreversibly stalled RNA polymerase II is targeted for ubiquitylation and proteasomal degradation in a Def1-dependent manner. This frees the lesion and allows global genome repair. Conjugating enzymes, ligases, and DUBs are highlighted in shades of purple, green, and pink, respectively. Distinct polyubiquitin chain linkages are indicated as K48 or K63.

Global genome repair:

GGR is affected by the ubiquitin system in several ways (Reed and Gillette 2007; Dantuma et al. 2009). A cullin-based E3, containing the Cul3, Elc1, the SOCS box protein Rad7, and the RING-finger protein Rad16, mediates ubiquitylation of Rad4 (Gillette et al. 2006) (Figure 9). Although the ubiquitylated protein is degraded by the proteasome, its modification rather than its degradation was found to be important for repair (Gillette et al. 2006). Yet, the overall efficiency of GGR is highly dependent on Rad4 levels, which are controlled by Rad23 (Lommel et al. 2002; Ortolan et al. 2004). As a consequence, Rad4 is strongly depleted in rad23Δ mutants, and the resulting damage sensitivity can in part be compensated by boosting Rad4 abundance. Thus, contrary to its role as a ubiquitin receptor in proteasomal targeting, Rad23 appears to stabilize Rad4 rather than induce its degradation. This may be mediated simply by binding to Rad4 and thereby preventing its misfolding (Dantuma et al. 2009) or alternatively, by means of shielding its ubiquitylated form from proteasomal access (Ortolan et al. 2000). It has even been reported that de novo protein synthesis is required for the stabilizing effect of Rad23 on Rad4, suggesting a regulation via damage-induced transcription (Gillette et al. 2006). In addition, the N-terminal ubiquitin-like domain of Rad23, known for its function as a proteasome docking site (Elsasser et al. 2002), contributes to GGR, possibly by providing a link between the NER machinery and the ATPase activities of the 19S cap (Watkins et al. 1993; Schauber et al. 1998; Russell et al. 1999b). Rad23’s UBA domains (Table 7), which as described above are critical for ubiquitin chain recognition by the proteasome, are not specifically required for NER (Bertolaet et al. 2001).

Transcription-coupled repair:

The influence of ubiquitylation on TCR may be viewed as a solution to the problem of an irreversibly stalled RNA polymerase II (Svejstrup 2010) (Figure 9C). In this situation, the enzyme’s large subunit, Rpb1, is ubiquitylated and degraded by the proteasome, which clears the transcription machinery from the site of damage and allows subsequent repair via GGR (Beaudenon et al. 1999; Woudstra et al. 2002). Degradation is dependent on the Coupling of Ubiquitin conjugation to ER degradation (CUE)-domain protein Def1 (Table 7), but whether this domain actually binds ubiquitin has not been determined. The E3 Rsp5 attaches either a single ubiquitin or a short K63-linked polyubiquitin chain to Rpb1, the latter of which may be trimmed by the DUB enzyme Ubp2 (Beaudenon et al. 1999; Harreman et al. 2009). A second E3, containing Cul3 in complex with Elc1, Ela1, and Hrt1, but not Rad7 and Rad16, has been implicated in Rpb1 modification as well (Ribar et al. 2006, 2007). This complex uses monoubiquitylated Rpb1 as a substrate for polyubiquitylation with a K48-linked chain (Harreman et al. 2009). Hence, Rpb1 polyubiquitylation, like PCNA modification, is characterized by the successive action of two E3s. As discussed above, degradation of polyubiquitylated Rpb1 requires the Cdc48Ufd1Npl4 complex (Verma et al. 2011), and the process might be balanced by Ubp3-mediated deconjugation (Kvint et al. 2008).

Regulation of gene expression and chromatin structure

Modulation of the transcription machinery:

Tight control over gene expression is essential for adaptation to changes in a cell’s environment. The ubiquitin–proteasome system contributes to this activity in many aspects (Muratani and Tansey 2003; Shilatifard 2006; Ouni et al. 2011; Geng et al. 2012). Among the most direct modes of influence is control over the levels of transcriptional regulators. Hence, many transcription factors are short-lived proteins, such as Gcn4, a target of SCFCdc4 (Meimoun et al. 2000), and Gal4, whose abundance is limited by SCFGrr1 (Muratani et al. 2005). In many cases, including Gcn4, the transcriptional activation domain was found to overlap with the degradation signal (Salghetti et al. 2000). Ubiquitin-dependent proteolysis was thus found to be intimately coupled to the activity of natural and engineered transcription factors (Salghetti et al. 2001; Lipford et al. 2005; Wang et al. 2010). These observations led to the hypothesis that periodic promoter clearance is important for maximal activity. In the case of Gal4, two parallel degradation pathways have been described: the SCFGrr1-dependent mode, which is independent of Gal4 activity and downregulates the protein in the absence of galactose, and a pathway mediated by the F-box protein Dsg1, which applies to activated Gal4 (Muratani et al. 2005).

Ubiquitin-dependent activation of transcription factors does not need to involve complete degradation, but can also proceed by proteolytic processing. This was observed for the transcriptional activators Spt23 and Mga2, whose membrane-bound precursors are ubiquitylated by Rsp5, leading to their processing and relocalization into the nucleus (Hoppe et al. 2000; see above).

Finally, as discussed above, RNA polymerase II itself is subject to ubiquitylation and degradation upon transcription stalling. In addition, ubiquitylation of Rpb1 and Rpb2 by the RING-finger E3 Asr1 apparently causes an eviction of the polymerase subunits Rpb4 and Rpb7, leading to the enzyme’s inactivation (Daulny et al. 2008). It remains to be seen, however, to what extent this strategy is used as a regulatory measure.

Several observations suggest possible nonproteolytic contributions of the proteasome to transcription. The RP, but not the CP, has been implicated in transcription elongation (Ferdous et al. 2001), and its ATPase subunits were found in association with active promoters (Gonzalez et al. 2002; Sulahian et al. 2006). Also, chromatin immunoprecipitation analyses have suggested limited overlap between RP and CP components on chromatin and have localized proteasome subunits to internal and 3′ regions of transcribed genes as well (Auld et al. 2006; Sikder et al. 2006). These data have resulted in a model postulating a nonproteolytic and chaperone-like activity of the proteasomal ATPases in transcription. In the context of transcription initiation, this has been linked to the recruitment and stimulation of the Spt-Ada-Gcn5 Acetyltransferase (SAGA) complex, a histone acetyltransferase (Lee et al. 2005). However, the physiological relevance of these findings is still debated (see for example Collins et al. 2009), and the mechanism by which the proteasome might act here remains to be elucidated.

Regulation of chromatin structure:

The higher organization of genes into chromatin and their accessibility by the transcription machinery are crucial determinants of gene expression (Osley 2006; Shilatifard 2006). The first evidence for a contribution of the ubiquitin system to chromatin structure came from the identification of monoubiquitylated histone H2B (Robzyk et al. 2000). The modification is attached to K123 within the C-terminal tail of H2B by Rad6 in complex with the RING–E3 Bre1 (Wood et al. 2003), and is a prerequisite for the subsequent di- and trimethylation of histone H3 on K4 and K79 by the methyltransferases Set1 and Dot1, respectively (Dover et al. 2002; Ng et al. 2002; Sun and Allis 2002). The relationships between the levels of ubiquitylated H2B and methylated H3 are complex, and the mechanism by which one modification induces the other has not been fully explained; but the reaction appears to occur cotranscriptionally and is important for telomeric gene silencing. H2B ubiquitylation is also observed on the body of transcribed genes and has been associated with transcriptional initiation and elongation, but also repression (Henry et al. 2003; Kao et al. 2004; Xiao et al. 2005; Osley 2006). In a reconstituted system derived from mammalian cells, the effect of H2B monoubiquitylation on elongation by RNA polymerase II is due to a stimulation of the FACT complex, and a similar situation may apply in yeast (Pavri et al. 2006).

Interestingly, optimal transcription in vivo requires both ubiquitylation and subsequent deubiquitylation of H2B. Deconjugation is mediated by the two DUBs, Ubp8 and Ubp10 (Henry et al. 2003; Emre et al. 2005; Gardner et al. 2005b). Ubp8 acts as an integral component of the SAGA complex, and during the early steps of transcription. In contrast, Ubp10 works independently and influences mainly telomeric silencing, indicating nonredundant roles (Emre et al. 2005; Gardner et al. 2005b). A recent genome-wide analysis of ubiquitin and methylation marks on H2B revealed that the two DUBs affect different pools of cellular H2B (Schulze et al. 2011).

The cross-talk of histone ubiquitylation and methylation affects not only gene expression, but also genome maintenance, via an influence of H2B monoubiquitylation and subsequent H3 K79 methylation on the DNA damage checkpoint (Game and Chernikova 2009). In this context, the checkpoint mediator protein Rad9 recognizes K79-dimethylated H3 via Rad9’s tudor domain and facilitates DNA repair by homologous recombination. As a consequence, bre1 and dot1 mutants are equally sensitive to agents that cause double-strand breaks (Game et al. 2006).

Processing of mRNAs:

Gene expression can be regulated post-transcriptionally by modulating the maturation, export, or stability of mRNA, all of which are affected by the ubiquitin system. Maturation of pre-mRNAs is controlled by the splicing factor Prp19, an E3 of the U-box type, whose activity is essential for spliceosome function (Ohi et al. 2003). A possible substrate of Prp19 is the splicing factor Prp8, which is ubiquitylated in vivo (Bellare et al. 2008), but also contains a ubiquitin-binding domain of the Jab1/Mpr1, Pad1, N-terminal (MPN) class that is essential for splicing (Bellare et al. 2006). Following splicing, mRNA is exported from the nucleus, and this process is guided by two E3s of the HECT family, Rsp5 and Tom1 (Duncan et al. 2000; Rodriguez et al. 2003). Rsp5 together with Ubc4 ubiquitylates the mRNA export factor Hpr1 in a transcription-dependent manner (Gwizdek et al. 2005). Ubiquitylated Hpr1 is targeted to the proteasome, but at the same time, the modification enhances interaction with the mRNA export receptor Mex67. This interaction is in part mediated by a UBA domain within Mex67, which was demonstrated to bind to polyubiquitin chains but also to interact with Hpr1 directly (Gwizdek et al. 2006). As a consequence, Mex67 stabilizes ubiquitylated Hpr1 by protecting it from proteasomal degradation. At the same time, it contributes to the recruitment of Hpr1 to actively transcribed genes, thus coordinating mRNA export with transcription (Gwizdek et al. 2006).

A relevant target of Tom1 appears to be Yra1, an adaptor protein linking mRNA to Mex67 (Iglesias et al. 2010). Monoubiquitylation and K48-diubiquitylation of Yra1 do not induce proteolysis, but promote dissociation from the Mex67–mRNP complex, which facilitates mRNA export. Finally, nonsense-mediated decay of mRNAs containing premature termination codons requires an RNA-dependent ATPase, Upf1, which also harbors a RING-related domain and displays E3 activity that is necessary for its function (Takahashi et al. 2008). However, the substrates and mechanism of this cytoplasmic pathway have not been elucidated.

Ccr4–Not complex:

The multisubunit Ccr4–Not complex impinges on chromatin modification and transcription elongation, but also on RNA processing, export, translation, and stability (Collart and Panasenko 2011). Its Not4 subunit, a RING-finger protein, displays E3 activity (Albert et al. 2002), but it is unclear how many of the functions ascribed to the Ccr4–Not complex actually require this activity. Two targets have been identified: Jhd2, a histone H3K4 demethylase involved in regulating gene expression (Mersman et al. 2009), and the so-called “nascent associated polypeptide complex,” a chaperone for nascent peptides at the ribosome that may be involved in protein quality control (Panasenko et al. 2006). Although Not4 cooperates with Ubc4 and/or Ubc5 in both cases, Jhd2 is polyubiquitylated and degraded, while the latter substrate undergoes monoubiquitylation. In addition, Not4 directly interacts with the proteasome and appears to contribute to its structural integrity (Panasenko and Collart 2011). It has been suggested that the complex acts as a general chaperone platform by means of associating with multiple interaction partners, but further research is clearly needed to uncover the basis of its multifunctionality.

Perspectives

For more than 25 years, the study of ubiquitylation in yeast has been a major driving force in the ubiquitin field, with countless original insights that have proven to be general across eukaryotes. Discoveries in this area have also fertilized many other aspects of cell biology, such as DNA repair and protein trafficking. Our understanding of ubiquitylation in yeast is more advanced than in other species but nonetheless far from mature.

In the coming years, this vast system will no doubt be charted more effectively through large-scale, mass-spectrometry–based proteomics efforts. The major goals of such studies will be to identify the set of all yeast proteins that undergo ubiquitylation; to identify the sites of ubiquitylation and the topologies of the ubiquitin chains at these sites, if any; to determine the set of yeast proteins that are substrates for the proteasome; and to match all substrates of the pathway to ubiquitin ligases, DUBs, and ubiquitin receptors that act on them. Such work should provide many fresh insights into the basic biology of yeast.

However, the type of pathway map that may emerge from studies of this kind will be limited. We will additionally need a better understanding of how the signaling information captured in the topology of ubiquitin chains is interpreted by ubiquitin receptors, and more generally deeper inroads must be made into the specificity, mechanisms, regulation, dynamics, and cell biology of the pathway. Vast networks of ubiquitin receptors, such as in the MVB and proteasome pathways, need to be deciphered. Ubiquitin chains are now studied as static entities, but they are likely to be very dynamic. It will be important, although challenging, to follow such key dynamics in the cell without perturbing pathway function.

Acknowledgments

We thank Geng Tian, Suzanne Elsasser, Soyeon Park, An Tyrrell, Karin Flick, Ingfei Chen, and especially Marion Schmidt for critical input and assistance with the figures. Work in our laboratories is supported by National Institutes of Health grants GM43601 and GM095526 (to D.F.), GM66164 and CA112560 (to P.K.); Deutsche Forschungsgemeinschaft grants SFB740, DIP, SPP1365 (to T.S.); and Cancer Research UK (H.D.U.)

Footnotes

  • Communicating editor: T. Davis

  • Received March 13, 2012.
  • Accepted July 28, 2012.

Literature Cited