Cell. Vol. 69, 725-735, May 29, 1992, Copyright 0 1992 by Cell Press

The N-End Rule

Alexander Varshavsky Division of Biology California institute of Technology Pasadena, California 91125

The N-end rule relates the in vivo half-life of a protein to the identity of its N-terminal residue. Since its discovery in 1986, distinct versions of the N-end rule have been shown to operate in all organisms examined, from animals and plants to yeast and bacteria. This review considers intracellular protein degradation with an emphasis on the N-end rule. The in vivo half-lives of proteins range from less than a minute to many days. Among the functions of intracellular proteolysis are the elimination of abnormal proteins, the maintenance of amino acid pools in cells affected by stresses such as starvation, and the generation of protein fragments that act as hormones, antigens, or other effectors. Yet another role of proteolytic pathways is to confer short half-lives on proteins whose concentrations must vary with time and alterations in the state of a cell. Metabolic instability is therefore a property of many regulatory proteins. Short half-lives of these proteins provide a way to generate their spatial gradients and allow for rapid adjustment of the proteins’ concentrations through changes in the rates of their synthesis or degradation. Many other proteins, while long-lived as components of larger complexes such as ribosomes and oligomeric proteins, are metabolically unstable as free subunits (reviewed by Finley and Chau, 1991; Hershko, 1991; Rechsteiner, 1991; Jentsch et al., 1991; Jones, 1991; Klausner and Sitia, 1990; Rivett, 1990; Stadtman, 1990; Gottesman, 1989; Olson and Dice, 1989; Bond and Beynon, 1987). Examples of eukaryotic proteins that are either conditionally or constitutively short-lived in vivo include cell cycle regulators such as cyclins (Hunt, 1990; Nurse, 1990; Glotzer et al., 1991) regulators of transcription and cell differentiation, such as the MATa repressor of the yeast Saccharomyces cerevisiae (Hochstrasser and Varshavsky, 1990) the Drosophila ftz protein (Kellerman et al., 1990), the mammalian c-Myc and MyoD proteins (Ltischer and Eisenman, 1990; Weintraub et al., 1991) the plant protein phytochrome (Shanklin et al., 1989) and enzymes at key steps in metabolic pathways, such as mammalian HMG-CoA reductase and ornithine decarboxylase (Chun et al., 1990; Ghoda et al., 1989). Among the many shortlived proteins in a bacterium such as Escherichia coli are the growth regulator SulA (Mizusawa and Gottesman, 1983) transposases of certain transposons (Derbyshire et al., 1990) IJ~~,astress-specificsubunit of RNApolymerase (Grossman et al., 1987; Tilly et al., 1989), and short-lived proteins of bacteriophages that infect E. coli, such as the h phage protein cll, whose half-life in an infected cell is a major determinant of the lysis-lysogeny decision by h (Banuett et al., 1986; Ho et al., 1988). Many (if not all) of

Review

these proteins appear to be degraded by pathways that are independent of the N-end rule. Amino acid sequences, conformational determinants, or chemically modified protein structures that confer metabolic instability are called degradation signals, or degrons (Varshavsky, 1991). The N-end rule(Bachmairet al., 1986) is the manifestation of a degradation signal called the N-degron, the first intracellular degron identified. I discuss the hierarchical structure of the N-end rule, the modular organization of the N-degron, the function of ubiquitin, cistrans recognition and subunit-specific degradation of N-end rulesubstrates, mechanisticdissection of the N-end rule pathway, and the current understanding of its functions. Brief History That most protein molecules do not last as long as the cell in which they reside became clear more than 50 years ago. In eukaryotes, one site of intracellular protein degradation was identified as a cytoplasmic organelle, the lysosome (de Duve, 1969). Like their extracellular cousins, lysosomal proteases are ATP-independent enzymes. The finding that a large fraction of intracellular proteolysis requires ATP (Simpson, 1953) was among the first to suggest that proteins are also degraded outside of lysosomes. Mechanistic studies of ATP-dependent proteolysis were initiated by the finding that an extract from rabbit reticulocytes is able to carry out ATP-dependent degradation of certain proteins (Etlinger and Goldberg, 1977). This degradation was shown to require a proteolytic cofactor that exists either free or covalently conjugated to short-lived proteins prior to their destruction in reticulocyte extract (Ciechanover et al., 1978; Hershkoet al., 1980). Thecofactor was identified as ubiquitin (Ub) (Wilkinson et al., 1980), a 76 residue protein originally described in studies unrelated to protein degradation (Goldstein et al., 1975). The first in vivo Ub conjugate was found to be an abundant mammalian chromosomal protein that is a branched conjugate of Ub and histone H2A; in this conjugate, an isopeptide bond links the C-terminal Gly-76 of Ub to the s-amino group of Lys-119 in H2A (reviewed by Busch, 1984). The coupling of Ub to acceptor proteins was subsequently found to involve a preliminary ATP-dependent step in which Gly-76 of Ub is joined, via a thioester bond, to a Cys residue in the Ub-activating (El) enzyme. The activated Ub is then transferred to a Cys residue in one of several Ubconjugating (E2) enzymes (reviewed by Pickart, 1988; Jentsch et al., lQQl), which couple Ub to its ultimate acceptors, yielding isopeptide bond-linked Ub-protein conjugates (Figure 1). The mouse cell line ts85 was shown to be temperature sensitive for growth because of a thermolability of its mutant Ub-activating enzyme; nonlysosomal protein degradation is inhibited at nonpermissive temperature in ts85 cells but not in their wild-type counterparts (Finley et al., 1984). These findings provided genetic argument for the

Cell 726

Ub,;

Ub;

Protdn

Protein

ws-Ub

?

Is-Ub mts-Ub ms-“b ms-“b ms-“b ms-“b Is-“b ms-Ub ms-Ub

Figure 1. The Ub System (S. cerevisiae) The yeast Ub genes (two of which contain introns) encode fusions of Ub to itself (MM) or specific ribosomal proteins (U&7-W/3). These fusions are processed by “b-specific, functionally overlapping proteases encoded by YUH7, UBPI-UBP3, and at least one other gene, yielding mature Ub (Tobias and Varshavsky, 1991 and references therein). Ub conjugated through high energy thioester bonds (*) is shown to the right of an acceptor protein; for low energy (isopeptide bond) conjugates, Ub is at the left, followed by a subscript denoting the number of Ub moieties joined to the acceptor protein. The posttranslational coupling of Ub to other proteins involves a preliminary ATP-dependent step, in which the last residue of Ub is joined, via a thioester bond, to a Cys residue in the “b-activating (El) enzyme encoded by MA7 (McGrath et al., 1991). The activated Ub is transferred to a Cys residue in one of at least ten distinct Ub-conjugating (E2) enzymes. E2 enzymes couple the activated Ub to its ultimate acceptor proteins, yielding isopeptide bond-linked Ub-protein conjugates. Ubc2 and Ubd are also known as Rad6 and Cdc34, respectively (Finley and Chau, 1991). Many if not all of the “b-conjugating enzymes may be guided to their in vivo substrates by EP-bound recognition proteins called recognins or E3. The only yeast recognin identified to date is N-recognin; it is associated with the Ubd Ub-conjugating enzyme (see Figure 4). A targeted, ubiquitinated substrate is degraded by the 26s proteasome, an ATP-dependent multicatalytic protease (see main text).

hypothesis (Hershko et al., 1980) that ubiquitination of a short-lived eukaryotic protein is an obligatory predegradation step. Subsequent studies have revealed many functional and mechanistic complexities of the Ub system, its extraordinary conservation among yeast, animals, and plants, and its absence from bacteria (Finley and Chau, 1991; Tobias et al., 1991). Figure 1 summarizes a part of what is known about the yeast Ub system.

The N-End Rule The understanding that the degradation of a short-lived intracellular protein may require a preliminary step of Ub conjugation leaves unsolved the problem of targeting: how are proteins initially recognized as substrates for degradation? One answer to this question came from the study of Ub genes, which encode fusions of Ub to itself or other proteins (tjzkaynak et al., 1987; Finley et al., 1987, 1989). When an engineered Ub-S-galactosidase fusion was expressed in S. cerevisiae, it was efficiently deubiquitinated by Ub-specific processing proteases (Bachmair et al., 1986). Moreover, the deubiquitination of Ub-X-pgal occurred irrespective of the identity of the residue X at the Ub-bgal junction, proline being the single exception. By allowing a bypass of the “normal” N-terminal processing of a newly formed protein, this result yielded an in vivo method for generating different residues at the N-termini of otherwise identical test proteins (Figure 2A). Remarkably, depending on the identity of their N-terminal residues, X-Pgal proteins were either long lived or metabolically unstable (Bachmair et al., 1986). The resulting code was named the N-end rule (Table 1). The N-end rule is now known to operate in S. cerevisiae (Bachmair and Varshavsky, 1989), in an extract from rabbit reticulocytes (Varshavsky et al., 1988; Gonda et al., 1989), and in E. coli (Tobias et al., 1991) (see Figure 3, Figure 4, and Figure 5; Table 1). An N-end rule was also demonstrated (but not yet determined systematically) in a mouse cell line (Townsend et al., 1988) and in tobacco (A. Bachmair, personal communication). The N-end rulesof E. coli, S. cerevisiae, and rabbit reticulocyte form nested sets of destabilizing residues, the largest set being that of reticulocyte (see Figure 3). Discovery of the N-end rule has accounted for an earlier observation by Hershko et al. (1984) that a chemical modification of the a-amino group at the N-terminus of a short-lived protein inhibits its degradation in reticulocyte extract. Moreover, it became clear that many of the model substrates used in earlier studies of the ATP-dependent protein degradation in reticulocyte extract bear destabilizing N-terminal residues and are in fact N-end rule substrates (Bachmair et al., 1986; Gonda et al., 1989). Hierarchical Structure of the N-End Rule Bacteria and eukaryotes have long been known to contain aminoacyl tRNA-protein transferases (aa-transferases), which conjugate specific amino acids to the N-termini of acceptor proteins (reviewed by Soffer, 1980). Also, in studies initially unrelated to amino acid-protein conjugation, the Ub-dependent degradation of some proteins was found to require tRNA (Ciechanover et al., 1985). The N-end rule provided an explanation for these observations by suggesting a function for aa-transferases. Specifically, we proposed that there exist secondary destabilizing N-terminal residues that function through their conjugation, by an aa-transferase, lo other, primary destabilizing residues (Bachmair et al., 1986). Ferber and Ciechanover (1987) confirmed this hypothesis by showing that the Ubdependent degradation of certain N-end rule substrates

Review: The N-End Rule 727

processing protease

II

so

111

da

C

Ka

1

rr-r;h~;;~~~~~lrnration

1

rr-r;h~;~~~~~i,n”i,,

D &recognition

Figure 2. The Ub Fusion Technique

and the N-Degron

(A) Linear fusions of Ub to other proteins are cleaved after the last residue of Ub in vivo and in cell extracts, making it possible to generate different residues at the N-termini of otherwise identical proteins. (B) Two-determinant organization of the eukaryotic Ndegron. d and s, destabilizing and stabilizing N-terminal residues, respectively; black ovals, multi-Ub chain. In construct Ill. the second-determinant lysine (K) is absent. (C) Cis recognition of the N-degron located in one subunit of a dimeric protein. The other subunit lacks the first determinant of the N-degron. (D) Trans recognition, in which the first (d) and second (K) determinants of the N-degron reside in different subunits of a dimeric protein. (E) A model for substrate recognition in the eukaryotic N-end rule pathway. In step I, N-recognin binds to adestabilizing N-terminal residue(d). followed by capture of the second-determinant lysine (K) by a Ub-conjugating (E2) enzyme that is associated with N-recognin (step II). Multiubiquitination by the E2 enzyme (step Ill) commences when both E2 and N-recognin have bound their respective determinants. It is unknown whether the lysine-binding site is located in the E2 enzyme or the ES-bound N-recognin. It is also unknown whether the binding of determinants of the N-degron is temporally ordered as shown, i.e., whether the lysine-binding site is unavailable until the d-binding site of N-recognin is filled.

requires the conjugation of Arg, a primary destabilizing residue, to their N-termini. Subsequent studies (Gonda et al., 1989; Balzi et al., 1990; Baker and Varshavsky, 1991) havedetermined that thesecondarydestabilizing residues in yeast are Asp and Glu, while in mammalian reticulocytes they are Asp, Glu, and Cys (see Figure 3 and Figure 4). Yet another level in the hierarchical organization of the N-end rule was revealed by the finding (Gonda et al., 1989) that in both yeast and mammals, N-terminal Asn and Gln are tertiary destabilizing residues-they function through their conversion, by a specific deamidase, into the secondary destabilizing residues Asp and Glu (see Figure 3 and Figure 4). Although both tertiary and secondary destabilizing residues of the eukaryotic N-end rules are stabilizing in E. coli, the bacterial N-end rule does contain two secondary destabilizing residues, Arg and Lys, which are primary destabilizing residues in eukaryotes (see Figure 3, Figure 4, and Figure 5). In E. coli, N-terminal Arg and Lys were

shown to be destablizing through their conjugation to primary destabilizing residues Leu or Phe by the Leu, Phe tRNA-protein transferase (L/F-transferase), an enzyme of previously unknown function (see Figure 5) (Tobias et al., 1991). This difference between the eukaryotic and bacterial N-end rules is consistent with the observed distribution of Arg-tRNA-protein transferase (R-transferase), which is confined to eukaryotes, and L/F-transferase, which is present at least in gram-negative bacteria but is absent from eukaryotes (Soffer, 1980). This difference also suggests that secondary and tertiary destabilizing residues were recruited relatively late in the evolution of the N-end rule, after the divergence of bacterial and eukaryotic lineages. The N-Degron The N-end rule is the manifestation of the N-degron (Varshavsky, 1991). In eukaryotes, the N-degron comprises

Cell 720

Table 1. The N-End Rule Half-Life of X-6gal Residue X

E. coli

S. cerevisiae

Aw

2 min 2 min 2 min 2 min 2 min 2 min >lO hr >lO hr >10 hr >lO hr >lO hr >I0 hr >lO hr >lO hr >I0 hr >lO hr >lO hr >lO hr

2 min 3 min 3 min 3 min 3 min 10 min 3 min 30 min 3 min 30 min 3 min 10 min >20 hr >20 hr >20 hr >20 hr >20 hr >20 hr >20 hr >20 hr

LYS Phe Leu Trp W His Ile Asp Glu Asn Gln CYS Ala Ser Thr GIY Val Pro Met

>lO hr

Approximate in vivo half-lives of X-6gal proteins in E. coli at 36% (Tobiasetal., 1991)and in S.cerevisiaeat30°C(Bachmairetal., 1966; Bachmair and Varshavsky, 1969). A question mark at Pro indicates its unknown status in the E. coli N-end rule (Tobias et al., 1991).

two distinct determinants: a destabilizing N-terminal residue and a specific internal Lys residue(s) (Figure 28) (Bachmair and Varshavsky, 1989; Chau et al., 1989; Johnson et al., 1990). The Lys residue is the site of attachment of a multi-Ub chain, whose formation is required for the degradation of at least some N-end rule substrates (Chau et al., 1989). In a multi-Ub chain, several Ub moieties are attached sequentially to the initial acceptor protein, forming a chain of Ub-Ub conjugates in which the C-terminal Gly-78 of one Ub is joined to Lys-48 of the adjacent Ub (Chau et al., 1989; Johnson et al., 1992). What causes a certain lysine in an N-end rule substrate to be selected as the multiubiquitination site? The available evidence suggests that this lysine is in spatial (but not necessarily “linear”) proximity to the destabilizing N-terminal residue and that the region containing either the lysine or the N-terminal region is mobile (Bachmair and Varshavsky, 1989; Johnson et al., 1990). In a stochastic view of the N-degron, each lysine of an N-end rule substrate can be assigned a probability of being utilized as a multiubiquitination site, depending on time-averaged spatial location, orientation, and mobility of the lysine. For some, and often for most of the lysines in an N-end rule substrate, the probability of serving as a multiubiquitination site would be infinitesimal because of the lysine’s lack of mobility and/ or its distance from the destabilizing N-terminal residue (Bachmair and Varshavsky, 1989). This “stochastic capture” mechanism (Figure 2E and below) is consistent with data about X+gal, related test proteins, and other model substrates of the N-end rule pathway (Dunten et al., 1991; Sokolik and Cohen, 1991; Dunten and Cohen, 1989). Another possibility is that the binding of a primary destabilizing residue of an N-end rule substrate is followed by

a directional scanning of the substrate’s polypeptide chain by the proteolytic machine (the recognition-ubiquitination-degradation complex). Scanning starts at the N-terminus of the substrate and stops when N-terminusproximal lysine is encountered. A lysine-attached multi-Ub chain is then formed, yielding a predegradation intermediate. In both the stochastic capture and scanning models, the efficiency of a search for a multiubiquitination site is strongly influenced by structural and kinetic aspects of the substrate’s conformation, with a folded conformation either slowing or precluding the search. In contrast to the eukaryotic N-end rule pathways, no Ub-like covalent modification of N-end rule substrates has been detected in E. coli. Moreover, while the conversion of multiubiquitination-site lysines into arginines (which cannot be ubiquitinated) renders an N-end rule substrate long-lived in eukaryotes, this modification does not impair the N-end rule-mediated degradation of the same substrate in E. coli (Tobias et al., 1991). Thus, bacteria lack a homolog of eukaryotic Ub. Furthermore, their N-end rule pathway lacks the requirement for a lysine-specific modification. The bacterial N-degron may be a one-determinant signal, comprising only the exposed, destabilizing N-terminal residue (see Figure 3 and Figure 5). On the Function of Ub What might be the function of a multi-Ub chain attached to a targeted short-lived protein? One possibility is that the chain’s formation on a targeted substrate produces an additional binding site (or sites) for components of the proteolytic machine. The resulting decrease in the rate of dissociation of the machine-substrate complex could be used to facilitate subsequent proteolytic steps. For example, if the rate-limiting step for the first proteolytic cleavage is a spontaneous unfolding (driven by thermal fluctuations) of a relevant region of the substrate, proteolysis of that region would be facilitated by stabilization of the machinesubstrate complex, because the longer the allowed “waiting” time, the greater the probability of a required unfolding event. One prediction of this model is that the N-end rule-mediated degradation of a substrate whose conformation presents less of a kinetic impediment to the proteolytic machine should be less dependent on Ub and ubiquitination than that of a more stably folded substrate. The discovery of subunit selectivity in Ub-dependent protein degradation (see below) suggests additional roles for substrateattached Ub moieties. For example, after binding to a multi-Ub chain linked to a subunit of an oligomeric protein, the proteolytic machine could use a mechanochemical process analogous to those that underlie the action of chaperonins (Rothman, 1989; Creighton, 1991) todissociate the targeted subunit from the rest of the protein before or during the subunit’s degradation (Johnson et al., 1990). These ideas could also apply to nonproteolytic functions of Ub. For instance, the emerging role of Ub in the translocation of some proteins across membranes (Finley and Chau, 1991) may be mediated by the binding of Ub moieties on a translocation substrate to a component of the translocation machine. By increasing the stability of a ma-

Review: The N-End Rule 729

chine-substrate complex, this binding may increase the probability of a translocation substrate undergoing a local unfolding that could be exploited by the translocation machine. Ubiquitination of a nascent translocation substrate could result in a kinetic competition between Ub-dependent proteolysis of the substrate in the cytosol and translocation-mediated escape of the substrate into a compartment such as the endoplasmic reticulum, whose lumen lacks the Ub system (N. Johnsson and A. V., unpublished data). The outcome of such a competition is likely to be influenced by the rate of folding of a substrate and by the strength of its translocation signal. The potential role of Ub and multi-Ub chains as conformation-perturbing devices (Varshavsky et al., 1988) is not necessarily in conflict with the notion of Ub as an affinity tag. This latter idea may also apply to Ub that is conjugated either to cytoskeletal proteins or to proteins at the cell surface, the inner face of the plasma membrane, or the cytosolic surfaces of vesicles (Siegelman and Weismann, 1988; Meyer et al., 1986; Yarden et al., 1986; Murti et al., 1988; Ball et al., 1987). Some of these proteinconjugated Ub moieties might serve as binding sites for kinesin or kinesin-like motor proteins. Similarly, Ub conjugated to histones in nucleosomes (Levinger and Varshavsky, 1982) might mediate the interactions of chromosomes with nucleoskeletal proteins. Among the advantages of Ub as an affinity tag are orientational flexibility (due to its extended C-terminus), reversibility (Ub can be removed by Ubspecific isopeptidases), and the possibility of a target-specific regulation of ubiquitination and deubiquitination. A previously established function of Ub that has not been encompassed by the above discussion is its role as acotranslational chaperone: in natural Ub fusions with ribosomal proteins, the transient association between Ub and these proteins promotes their incorporation into ribosomes and is required for efficient ribosome biogenesis (Finley et al., 1989).

Cis and Trans Recognition of the N-Degron In a monomeric substrate of the eukaryotic N-end rule pathway, both determinants of the N-degron are recognized in cis (Figures 28 and 2C). However, in a multisubunit protein, these determinants could reside in different subunits and might be recognized in trans (Figures 2C and 2D). By following the metabolic fates of individual subunits in an X+gal tetramer, the N-degron of X-8gal was shown to function either in cis or in trans (Johnson et al., 1990). These results are consistent with a model (Figure 2E) in which the recognition complex has a binding site for a substrate’s destabilizing N-terminal residue, and a lysinebinding site. Occupation of both sites is postulated to be required for multiubiquitination of the bound lysine of the substrate (Bachmair and Varshavsky, 1989). Since the N-terminal region of X+gal is likely to be segmentally mobile (Bachmair and Varshavsky, 1989; Johnson et al., 1990), at least one of its Lys residues could be transiently in spatial proximity to the destabilizing N-terminal residue of either its own or a different subunit in the same X-8gal tetramer, the latter allowing trans recognition.

Trans recognition may also be relevant to other targeting signals, including other degrons. Features of a signal that could be recognized either in cis or in trans are the presence of more than one distinct determinant and an absence of a strict constraint on the “linear” distance between determinants. These properties are characteristic of signals in DNA that mediate transcriptional regulation and depend on the bending and loop-forming properties of the DNA (Ptashne and Gann, 1990). Indeed, the promoter and the enhancer need not be located within the same DNA molecule for the enhancer-mediated activation of at least some promoters (Dunaway and Droge, 1989). The E6 protein of a human papillomavirus binds to the human ~53 protein and promotes its Ub-dependent degradation in reticulocyte extract (Scheffner et al., 1990; Crook et al., 1991) which is reminiscent of trans targeting in the N-end rule pathway (Figure 2D). Wild-type ~53 is a tumor suppressor and a target for several oncoproteins encoded by DNA tumor viruses. It remains to be seen whether the E6-induced degradation of p53 is mediated by an N-degron, or whether a different Ub-dependent degron is involved. Subunit-Specific Degradation of N-End Rule Substrates The trans recognition experiments (Figures 2C and 2D) also demonstrated that only those subunits of a short-lived X+gal tetramer that contain the multiubiquitination site (but not necessarily the destabilizing N-terminus) are actually degraded (Johnson et al., 1990). Destruction of an oligomeric N-end rule substrate is thus confined to those subunits that can be ubiquitinated. A mechanism that may be responsible for this feat of selectivity was outlined above. Subunit selectivity is likely to emerge as a general feature of intracellular protein degradation. It may underlie the alteration of transcriptional repressors and activators via combinatorial variation of their subunit composition, a common theme in gene regulation (Johnson and McKnight, 1989). Another example may be thedestruction of cyclins, the short-lived subunits of multiprotein complexes that regulate cell cycle progression in eukaryotes (Hunt, 1990; Nurse, 1990). Subunit selectivity is also characteristic of at least one of the two Ub-dependent degrons in the naturally short-lived transcriptional repressor MATa of S. cerevisiae. Neither of the degrons in MATa is an N-degron (Hochstrasser and Varshavsky, 1990; Hochstrasser et al., 1991). Trans recognition and subunit-specific protein degradation (Johnson et al., 1990) may allow the construction of a new class of dominant negative mutants in which a longlived protein could be destabilized by targeting it for degradation in trans. By using the Ub fusion technique (Figure 2A), the first determinant of the N-degron (Figure 28) could be generated in a specially designed targeting protein that is an in vivo ligand of the protein of interest and has the additional (if necessary, engineered) property of lacking an efficient multiubiquitination site of the N-degron. In this design, the multiubiquitination-site lysine is provided by the bound protein of interest, which would be targeted for

Cell 730

~______ i--

AminoacIds FLWYRKH

Figure 3. Comparison

of Eukaryotic

~___~

~~ ._

I NQDECASTGVPM

and Bactenal N-End Rules

Open circles, stabilizing residues; filled circles, triangles, and crosses denote, respectively, primary, secondary, and tertiary destabilizing residues in the N-end rulesof E. coli, S. cerevisiae, and rabbit reticulocytes. An asterisk at Ile indicates that it is a weakly destabilizing residue in yeast and a borderline destabilizing residue in reticulocytes. A question mark at Pro indicates its unknown status in the E. coli N-end rule. Modified from Tobias et al. (1991).

degradation in trans (Figure 2D). Lacking the multiubiquitination site, the targeting protein would not be destroyed and therefore should act catalytically rather than stoichiometrically. A short peptide, a peptide mimetic, or even an unrelated compound such as an oligonucleotide or a low molecular weight enzyme inhibitor that bears a destabilizing amino acid residue in a stereochemically appropriate context might also function as a targeting ligand in the trans degradation technique. While the feasibility of these applications remains to be ascertained, the strategy of trans targeting may have already been exploited by a papillomavirus (Scheffner et al., 1990; see above) to subvert the growth control of a host cell. The N-End Rule Pathways

The consistent absence of the twelve destabilizing residues of the yeast N-end rule (Figure 3) from the mature N-termini of long-lived, noncompartmentalized proteins in both bacteriaandeukaryotes(Bachmairet al., 1986) isdue to the properties of Met aminopeptidase, which removes N-terminal Met if and only if the second residue is stabilizing in the yeast N-end rule (Arfin and Bradshaw, 1988; Hire1 et al., 1989; Sherman et al., 1985). In particular, Ala, Ser, and Thr, which are stabilizing in E. coli and yeast (Figure 3) do not prevent the removal of N-terminal Met by Met aminopeptidase and therefore can be exposed at the N-termini of proteins that initially bear these residues at the second position. However, since Ala, Ser, and Thr are destabilizing in the reticulocyte N-end rule (Figure 3) Met aminopeptidase may be involved in generating a subset of potential N-end rule substrates in mammalian reticulocytes. With the exception of this possibility and a physiological substrate of the N-end rule pathway produced by the processing protease of Sindbis virus (see below), nothing is known about such substrates or proteases that generate them. Genetic dissection of the yeast N-end rule pathway has resulted in the isolation of genes encoding deamidase (LEA 7) (R. Baker and A. V., unpublished data), R-transferase (ATEI) (Balzi et al., 1990), and N-recognin (UBR7) (Bartel et al., 1990) (Figure 4). N-recognin (also known as E3) is the 225 kd recognition complonent of the pathway

that selects potential proteolytic substrates by binding to their primary destabilizing N-terminal residues (Bartel et al., 1990). N-recognin has two distinct substrate-binding sites. The type 1 site is specific for the basic N-terminal residues Arg, Lys, and His. The type 2 site is specific for the bulky hydrophobic N-terminal residues Phe, Leu, Trp, Tyr, and Ile (Figure 4) (Baker and Varshavsky, 1991; Gonda et al., 1989; Reiss et al., 1988). Yeast N-recognin is physically associated with the Ubconjugating (E2) enzyme encoded by the UBCP (RADG) gene (Dohmen et al., 1991). Ubc2 is one of at least ten distinct E2 enzymes in S. cerevisiae (see Figures 1 and 4). The N-end rule pathway is inactive in either a ubc2 or a ubrl null mutant (Bartel et al., 1990; Dohmen et al., 1991). The known functions of UBCP include DNA repair, induced mutagenesis, sporulation, and regulation of retrotransposition (Jentsch et al., 1987, 1991; Madura et al., 1990 and references therein). Further analysis should allow precise distinctions between those functions of the Ubc2 enzyme that are dependent on its interaction with N-recognin and those mediated by other, still uncharacterized Ubc2associated recognins (Sharon et al., 1991) specific for signals distinct from the N-degron. For instance, both UBR7 and UBCP are expressed during sporulation (Madura et al., 1990); however, while the absence of the N-end rule pathway (in ubrl mutants) results in a mild sporulation defect (increased frequency of asci containing fewer than four spores) (Bartel et al., 1990) the absence of both the N-end rule pathway and other UBCP-dependent pathways (in ubc2 mutants) precludes or severely perturbs sporulation (Dohmen et al., 1991 and references therein). In addition to binding the Ubc2 enzyme, N-recognin apparently binds to R-transferase (Atel) and deamidase (Deal), with the latter enzyme possibly bound indirectly, through its interaction with R-transferase (R. Baker and A. V., unpublished data). These affinities might be expected, given the hierarchical structure of the N-end rule (Figure 4). Thus, the action of the Deal deamidase on a substrate bearing a tertiary destabilizing N-terminal residue, Asn or Gln, could be followed by a channeling of the deamidated substrate to an active site of the deamidasebound R-transferase, and from there, after arginylation of N-terminal Asp or Glu, to the type 1 binding site of N-recognin (Figure 4). Following the UbcBmediated multiubiquitination of a targeted N-end rule substrate (see Figure 2E and discussion above), it is degraded by the 26s proteasome, a large (-1,500 kd), ATP-dependent multicatalytic protease that contains more than twenty distinct subunits. While the 26s proteasome is relatively unstable in vitro, its proteolytically active component, the 20s proteasome, is stable and was described long before it was realized that this cylindershaped particle has protease activity (reviewed by Goldberg, 1992; Finley and Chau, 1991; Orlowski, 1990). Mutations in specific subunits of the yeast 20s proteasome that impair its function result in the metabolic stabilization of both N-end rule substrates and other short-lived proteins that are degraded by Ub-dependent pathways in wild-type cells (Heinemeyer et al., 1991; D. Wolf and S. Jentsch, personal communication).

Review: The N-End Rule 731

tertiary destabilizing residues

t 1 deamidase

J( secondary

Deal

D E-

destabilizing residues

I/

R-~RNAR

R-transferase

Ate1

I

a

k

I

TYPO t ’

primary destabilizing residues

or

Ubci- a

R

K H

Ubc2 ATP *

degradation

Ubrl

Figure 4. The S. cerevisiae N-End Rule Pathway Conversion of N-terminal N and Cl to D and E is carried out by a deamidase encoded by the DEA7 gene. Conjugation of the primary destabilizing residue R to the secondary destabilizing N-terminal residues D and E is catalyzed by Arg tRNA-protein transferase (R-transferase). Type 1 and type 2 primary destabilizing N-terminal residuesare bound bytwo distinct sitesof N-recognin. N-recognin formsacomplex with the Ubc2 Ub-conjugating enzyme, one of at least ten E2 enzymes in S. cerevisiae. If a substrate bears both determinants of the Ndegron. UbcP uses activated Ub to catalyze formation of a substrate-linked multi-Ub chain. The latter is required for the degradation of at least some N-end rule substrates by the ATP-dependent 26s proteasome. Modified from Dohmen et al. (1991).

The only characterized mammalian N-end rule pathway is that of rabbit reticulocytes (Gonda et al., 1989), which is similar but not identical to that of yeast (see Figure 3). In particular, reticuiocytes contain a second N-recognin specific for N-terminal Ala, Ser, and Thr, which are stabilizing residues in yeast (Gonda et al., 1989; Helier and Hershko, 1990). The E. coli N-end rule pathway differs from its eukaryotic counterparts in its independence from Ub, the absence of deamidase and R-transferase, and the presence of L/F-transferase, whose function was discussed earlier (Figure 5). The Clp(Ti) protease (Maurizi et al., 1990; Goldberg, 1992), one of the two known ATP-dependent proteases in E. coii (the other being Lon), is required for the in vivo degradation of N-end rule substrates (Tobias et al., 1991). Clp is a mu750 kd protein containing two types of subunits, CipA (81 kd) and ClpP (21 kd) (Maurizi et al., 1990). While much simpler in subunit composition than the 20s proteasome of eukaryotes, the E. coii Clp is comparably large and also ATP dependent (Gottesman, 1989).

Functions Most

of the N-End Rule

noncompartmentalized

proteins

lack destabilizing

residues at their N-termini (Bachmair et al., 1988). While this bias is largely due to the properties of Met aminopeptidase (see above), it may also stem in part from underrepre-

sentation of short-lived proteins in the presently known set of N-terminal sequences. Thus far, the only physiological substrate of the N-end rule pathway identified in either bacteria or eukaryotes is RNApolymeraseof the Sindbisvirus (a plus-stranded RNA virus). This RNA polymerase is produced by an endoproteolytic cleavage of the viral precursor polyprotein, bears an N-terminal Tyr (a destabilizing residue; see Figure 3), and is degraded by the N-end rule pathway (deGroot et al., 1991). Moreover, Tyr is also an N-terminal residue of RNA polymerases of other alphaviruses (Strauss et al., 1988), suggesting that these homologsof the Sindbis polymerase are also degraded by the N-end rule pathway. Although C/PA- mutants of E. coli lack the N-end rule pathway (Tobias et al., 1991), they grow at wild-type rates, appear phenotypicaiiy normal, and do not stabilize several short-lived E. coli proteins (Gottesman, 1989). The absence of the N-end rule pathway in ubrl mutants of S. cerevisiae results in a slight retardation of growth and a subtle sporulation defect (Bartel et al., 1990). However,

Cdl 732

zation) of proteins during differentiation and other changes in the state of a cell. Whether the N-end rule is indeed a soft-wired code, and whether its variations are physiologically relevant remain to be determined.

secondary destabilizing residues L-IRNA’ L/F-transferase

Aat

?

6”‘ l

-

d+ *

&

degradation ClP protease

ClpA, ClpP

Figure 5. The E. coli N-End Rule Pathway Conjugation of the primary destabilizing N-terminal residues L or F to the secondary destabilizing residues R and K is catalyzed by Leu. Phe-tRNA-protein transferase (L/F-transferase). The degradation of a substrate bearing a primary destabilizing N-terminal residue is mediated by the ATP-dependent protease Clp encoded by the genes c/p/l and cl@. A question mark indicates a hypothetical recognition step, invoked by analogy to the eukaryotic N-end rule pathway. Modified from Tobias et al. (1991).

the viability of null ubrl mutants and their wild-type sensitivity to a variety of metabolic and physical stresses (Bartel et al., 1990) indicate that this pathway is nonessential in yeast as well. Thus, cell viability appears not to depend on the degradation of N-end rule substrates. To identify pathways that complement the hypothetical essential functions of the N-end rule, a “synthetic lethal” screen was used to isolate a yeast mutant whose viability requires the presence of the N-end rule pathway (Ota and Varshavsky, 1992). An extragenic suppressor of this mutation encodes a putative phosphotyrosine phosphatase, suggesting a connection between the N-end rule, dephosphorylation of phosphotyrosine in proteins, and functions essential for cell viability (Ota and Varshavsky, 1992). The divergence between the yeast and reticulocyte N-end rules with respect to Ala, Ser, and Thr (see Figure 3) appears particularly significant because these residues are often present at the mature N-termini of long-lived, noncompartmentalized proteins(Gondaet al., 1989). This, and the presence in reticulocytes of a distinct N-recognin specific for N-terminal Ala, Ser, and Thr (Gonda et al., 1989; Reiss et al., 1988), raise the possibility that the N-end rule of reticulocytes may differ from the N-end rule of a ieticulocyte precursor, and that Ala, Ser, and Thr may be stabilizing residues in such a precursor. While the selective destruction of previously long-lived proteins is unusually extensive during the conversion of reticulocytes into erythrocytes, it occurs in other cell lineages as well. Controlled alterations of the N-end rule may provide a mechanism for selective degradation (or selective stabili-

Regulatable Degron: A New Method for Producing Conditional Mutants A frequent problem with conditional mutants is their leakiness, i.e., unacceptably high residual activity of either a temperature-sensitive protein at nonpermissive temperature or a gene of interest in the “off” state of its promoter. Another problem stems from the “phenotypic” lag that often occurs between the imposition of nonpermissive conditions and the emergence of a relevant null phenotype. Phenotypic lag tends to be longer with proteins that are required in catalytic rather than stoichiometric amounts. It is also characteristic of those temperature-sensitive mutants in which nonpermissive temperature inactivates a newly formed protein but not its mature counterpart (Pringle, 1975). Some of these difficulties can be reduced or bypassed by connecting a transplantable N-degron (see Figure 28) to a long-lived protein of interest in a setting that allows regulatable activation of the attached N-degron. Specifically, the protein is expressed in yeast cells in which the Uf3Rl gene, encoding N-recognin (see Figure 4), had been placed under the control of an inducible promoter. The metabolic stability and hence the steady-state levels of the protein bearing the N-degron are either normal or extremely low depending on whether the lJBR7 gene is repressed or induced (J. Dohmen, K. Madura, 8. Bartel, and A. V., unpublished data). Such ns (N-end rule-sensitive) mutants can be constructed with any protein whose function is not perturbed by an N-terminal extension. The rapid disappearance of ns proteins under nonpermissive conditions (induced UBR7) should decrease the phenotypic lag. Attachment of the N-degron to either a temperaturesensitive protein or a protein expressed from an inducible promoter (Park et al., 1992) could reduce the leakiness of existing conditional mutants: under nonpermissive conditions, the destruction of a target protein by the activated N-end rule pathway should be synergistic with the denaturing effect of nonpermissive temperature or the “off” state of an inducible promoter. Concluding Remarks The N-degron remains the only Ub-dependent degradation signal that is understood in some detail. Its properties of modular organization, cisltrans recognition, and subunit selectivity are likely to recur in other, presently unknown or less well understood degrons. The eukaryotic N-end rule pathway includes a step of Ub conjugation to a specific lysine(s) of a targeted substrate. Ubiquitination is required for the degradation of at least some N-end rule substrates in eukaryotes but not in bacteria, which have the N-end rule pathway but lack the Ub system. This striking evolutionary dichotomy remains to be explored. Dissection of the manner in which the Ubdependent steps are bypassed in the bacterial N-end rule pathway may help clarify aspects of the Ub function that

Review: The N-End Rule 733

have kept the Ub system nearly unchanged throughout the evolution of eukaryotes but were apparently dispensable in the course of bacterial evolution. One possibility is that the substrate-attached multi-Ub chain may decrease the rate of dissociation of the proteolytic machine-substrate complex and thereby facilitate the degradation of substrates whose relatively stable conformations pose a kinetic impediment for the proteolytic machine. If so, bacteria may contain proteins whose function in the N-end rule pathway is Ub-like but involves a noncovalent, lysine-independent binding to a targeted N-end rule substrate. The demonstrated contribution of multi-Ub chains to subunit selectivity of protein degradation in eukaryotes leads to another testable prediction: if bacteria lack a functional equivalent of the multi-Ub chain, they may also lack the ability of Ub-dependent (eukaryotic) pathways to target an individual subunit within a multisubunit protein. At present, mechanistic aspects of the N-end rule are understood much better than its functions. Only one physiological substrate of the N-end rule pathway (or rather, a class of such substrates) has been identified thus far. The conjecture that the N-end rule can be altered by physiologically relevant changes in the state of a cell remains to be tested. Six years after its discovery, there is still much to learn about the N-end rule.

Chau, V., Tobias, J. W., Bachmair, A., Marriott, D., Ecker, D., Gonda, D. K., and Varshavsky, A. (1989). A multiubiquitin chain is confined to specific lysine in a targeted short-lived protein. Science 243, 1576 1583. Chun, R. T., Bar-nun, S., and Simoni, J. (1990). The regulation of degradation of 3-hydroxy3-methylglutaryl-CoA reductase requires a short-lived protein and occurs in the endoplasmic reticulum. J. Biol. Chem. 265, 22004-22010. Ciechanover, A., Hod, Y., and Hershko, A. (1978). A heat-stable polypeptide component of an ATP-dependent proteolytic system from reticulocytes. Biochem. Biophys. Res. Commun. 81, 1100-1105. Ciechanover, A., Wolin, S. L., Steitz, J. A., and Lodish, H. F. (1985). Transfer RNA is an essential component of the ubiquitin- and ATPdependent proteolytic system. Proc. Natl. Acad. Sci. USA 82, 13411345. Creighton. T. E. (1991). Unfolding protein folding. Nature 352, 17-18. Crook, T., Tidy, J. A., and Vousden, K. H. (1991). Degradation of ~53 can be targeted by HPV E6 sequences distinct from those required for p53 binding and trans-activation. Cell 67, 547-556. de Duve, C. (1969). The lysosome in retrospect. In Lysosomes in Biology and Pathology, Volume 1, J. T. Dingle and H. B. Fells, eds. (Amsterdam: North-Holland), pp. 3-40. deGroot, R. J., Riimenapf, T., Kuhn, R. J., Strauss, E.G., and Strauss, J. H. (1991). Sindbis virus RNA polymerase is degraded by the N-end rule pathway. Proc. Natl. Acad. Sci. USA 88. 8967-8971, Derbyshire, K. M., Kramer, M., and Grindley, N. D. F. (1990). Role of instability in the cis action of the insertion sequence IS903 transposase. Proc. Natl. Acad. Sci. USA 87, 4048-4052. Dohmen, R. J., Madura, K., Bartel, B., and Varshavsky, A. (1991). The N-end rule is mediated by the UbcP (Rad6) ubiquitin-conjugating enzyme. Proc. Natl. Acad. Sci. USA 88, 7351-7355.

Acknowledgments

Dunaway, M., and Droge, P. (1969). Transactivation of the Xenopus rRNA gene promoter by its enhancer. Nature 341, 657-659.

I am greatly indebted to the past and present membersof the laboratory for helpful discussions and advice. I thank B. Bartel, Ft. Baker, J. Dohmen, M. Ellison, D. Finley, M. Hochstrasser, E. Johnson, N. Johnsson, K. Madura, J. McGrath, I. Ota, J. Tobias, and P. Waker for comments on the manuscript and B. Doran for secretarial assistance. Work in the author’s laboratory is supported by grants from the National Institutes of Health.

Dunten, R. L., and Cohen, R. E. (1989). Recognition of modified forms of ribonuclease A by the ubiquitin system. J. Biol. Chem. 264,1673916747. Dunten, R. L., Cohen, R. E., Gregori, L., and Chau, V. (1991). Specific disulfide cleavage is required for ubiquitin conjugation and degradation of lysozyme. J. Biol. Chem. 266, 3260-3267.

References

Etlinger, J. D., and Goldberg, A. L. (1977). A soluble ATP-dependent proteolytic system responsible for the degradation of abnormal proteins in reticulocytes. Proc. Natl. Acad. Sci. USA 74, 54-58.

Arfin, S. M., and Bradshaw, R. A. (1988). Cotranslational processing and protein turnover in eukaryotic cells. Biochemistry 27, 7979-7984.

Ferber, S., and Ciechanover, A. (1987). Role of arginine-tRNA in protein degradation by the ubiquitin pathway. Nature 326, 808-811.

Bachmair, A., and Varshavsky. A. (1989). The degradation short-lived protein. Cell 56, 1019-1032.

Finley, D., and Chau. V. (1991). Ubiquitination. 7. 25-70.

Bachmair, A., Finley, D., and Varshavsky, a protein is a function of its amino-terminal 186.

signal in a

A. (1986). In vivo half-life of residue. Science 234, 179-

Baker, R. T., and Varshavsky, A. (1991). Inhibition of the N-end rule pathway in living cells. Proc. Natl. Acad. Sci. USA 88, 1090-1094. Ball, E., Karlik, C. C., Beall, C. J., Saville, D. L., Sparrow, J. C., Bullard, B., and Fyrberg, E. A. (1987). Arthrin. a myofibrillar protein of insect flight muscle, is an actin-ubiquitin conjugate. Cell 51, 221-228. Balzi, E., Choder, M., Chen, W.. Varshavsky, A., and Goffeau, A. (1990). Cloning and functional analysis of the arginyl-tRNA-protein transferase gene ATE7 of Saccharomyces cerevisiae. J. Biol. Chem. 265, 7464-7471. Banuett. F., Hoyt, M. A., MacFarlane, L., Echols, H., and Herskowitz, I. (1986). hf/B, a new Escherichia coli locus regulating lysogeny and the level of bacteriophage lambda cll protein. J. Mol. Biol. 787, 213224. Bartel, B., Wunning, I., and Varshavsky, A. (1990). The recognition component of the N-end rule pathway. EMBO J. 9. 3179-3189. Bond, J. S., and Beynon, R. J. (1987). Proteolysis regulation. Mol. Aspects Med. 9, 173-287. Busch, H. (1984). Ubiquitination 262.

and physiological

of proteins. Meth. Enzymol. 706,238-

Annu. Rev. Cell Biol.

Finley, D., Ciechanover, A., and Varshavsky, A. (1984). Thermolability of ubiquitin-activating enzyme from the mammalian cell cycle mutant ts65. Cell 37, 43-55. Finley, D.,C)zkaynak, E.,andVarshavsky, A.(1987).Theyeastpolyubiquitin gene is essential for resistance to high temperatures, starvation, and other stresses. Cell 48, 1035-1046. Finley, D., Bartel, B., and Varshavsky, A. (1989). The tails of ubiquitin precursors are ribosomal proteins whose fusion to ubiquitin facilitates ribosome biogenesis. Nature 338, 394-401. Ghoda, L., Wetters, T. D., Macrae, M., Ascherman, D., and Coffino, P. (1989). Prevention of rapid intracellular degradation of ODC by a carboxyl-terminal truncation. Science 243, 1493-1495. Glotzer, M., Murray, A. W., and Kirschner, M. W. (1991). Cyclin is degraded by the ubiquitin pathway. Nature 349, 132-138. Goldberg, A. (1992). The mechanism and functions of ATP-dependent proteases in bacterial and animal cells. Eur. J. Biochem. 203. 9-23. Goldstein, G., Scheid, M., Hammerling. U., Boyse, E. A., Schlesinger, D. H., and Niall, H. D. (1975). Isolation of a polypeptide that has lymphocyte-differentiating properties and is probably represented universally in living cells. Proc. Natl. Acad. Sci. USA 72, 11-15. Gonda, D. K., Bachmair,

A., Wcnning,

I.. Tobias, J. W., Lane, W. S.,

Cell 734

and Varshavsky, A. (1989). Universality rule. J. Biol. Chem. 264, 18700-16712. Gottesman, S. (1989). Genetics Annu. Rev. Genet. 23, 163-168.

and structure

of proteolysis

of the N-end

in Escherichia

coli.

Grossman, A. D.. Strauss, D. G., Walter, W. A., and Gross, C. A. (1987). 03* synthesis can regulate the synthesis of heat shock proteins in Escherichia coli. Genes Dev. 7, 179-184. Heinemeyer. W., Kleinschmidt. J. A.. Saidowsky, J., Escher, C., and Wolf, D. H. (1991). Proteinase ycsE, the yeast proteasomelmulticatatytic proteinase: mutants unravel its function in stress-induced proteolysis and uncover its necessity for cell survival. EMBO J. 10, 555-562. Heller, H., and Hershko, A. (1990). A ubiquitin-protein ligase specific for type III protein substrates. J. Biol. Chem. 265, 6532-6535. Hershko, A. (1991). The ubiquitin pathway for protein degradation. Trends Biochem. Sci. 16. 265-268 Hershko, A., Ciechanover, A., (1980). Proposed role of ATP proteins with multiple chains proteolysis. Proc. Natl. Acad.

Heller, H., Haas, A. L., and Rose, I. A. in protein breakdown: conjugation of of the polypeptide of ATPdependent Sci. 77, 1783-1786.

Hershko, A., Heller, H., Eytan, E., Kaklij, G., and Rose, I. A. (1984). Role of the a-amino group of protein in ubiquitin-mediated protein breakdown. Proc. Natl. Acad. Sci. USA 87, 7021-7025. Hirel, P. H., Schmitter, J. M., Dessert, P.. Fayat, G., and Blanquet, S. (1989). Extent of N-terminal methionine excision from Escherichia coli proteins is governed by the side-chain length of the penultimate amino acid. Proc. Natl. Acad. Sci. USA 86, 8247-8251. Ho, Y. S., Mahoney, M. E., Wulff. D. L., and Rosenberg, M. (1988). Identification of the DNA binding domain of the phage lambda cll transcriptional activator and the direct correlation of cll protein stability with its oligomeric forms. Genes Dev. 2, 184-195. Hochstrasser. transcriptional

M., and Varshavsky. A. (1990). In vivo degradation of a regulator: the yeast a2 repressor. Cell 67, 697-708.

Hochstrasser, M., Ellison, M., Chau, V., and Varshavsky, A. (1991). The short-lived MATa transcriptional regulator is ubiquitinated in viva. Proc. Natl. Acad. Sci. USA 88, 4606-4610. Hunt, T. (1990). Destruction

is our delight. Nature 349, 100-101.

Jentsch, S., McGrath, J. P., and Varshavsky. A. (1987). The yeast DNA repair gene RADG encodes a ubiquitin-conjugating enzyme. Nature 329, 131-134. Jentsch, S., Seufert, W., and Hauser, H.-P. (1991). Genetic analysis of the ubiquitin system. Biochim. Biophys. Acta 7089, 127-139. Johnson, E. S., Gonda, D. K., and Varshavsky. A. (1990). Cis-bans recognition and subunit-specific degradation of short-lived proteins. Nature 348, 287-291. Johnson, E. S., Bartel, B, Seufert, W., and Varshavsky, Ubiquitin as a degradation signal. EMBO J. 7 7, 497-505.

A. (1992).

Johnson, P. F.. and McKnight, S. L. (1989). Eukaryotic transcriptional regulatory proteins. Annu. Rev. Biochem. 58. 799-839. Jones, E. W. (1991). Three proteolytic systems of the yeast Saccharomyces cerevisiae. J. Biol. Chem. 266, 7963-7966. Kellerman, K. A., Mattson, D. M., and Duncan, I. (1990). Mutations affecting the stability of the fushi farazu protein of Drosophila. Genes Dev. 4, 1936-1950.

ClpP, the proteolyttc component of the ATP-dependent of Escherichia coli. J. Biol. Chem. 265, 12536-12545.

Clp protease

McGrath. J. P.. Jentsch, S., and Varshavsky, A. (1991). UBAI. an essential yeast gene encodmg ubiquitin-activating enzyme. EMBO J. 70, 227-237. Meyer, E. M., West, C. M., and Chau, V (1986). Antibodies directed against ubiquitin inhibit high affinity 3H-choline uptake in rat cerebral cortical synaptosomes. J. Biol. Chem. 267, 14365-14368. Mizusawa, S., and Gottesman, S. (1983). Protein degradation in Escherichia coli: the /on gene controls the stability of the SulA protein. Proc. Natl. Acad. Sci. USA 80. 358-362. Mum, K. G.. Smith, H. T., and Fried, V. A. (1988). Ubiquitin is a component of the microtubular network. Proc. Natl. Acad. Sci. USA 85,30193023. Nurse, P. (1990). Universal control mechamsm M-phase. Nature 344, 503-508.

regulatmg

onset of

Olson, T. S.. and Dice, J. F. (1989). Regulation of protein degradation rates in eukaryotes. Curr. Opin. Cell Biol. 7. 1194-1200. Orlowski, M. (1990). The multicatalytic proteinase complex, a major extralysosomal proteolytic system. Biochemistry 29, 10289-10297. Ota, I., and Varshavsky, A. (1992). A gene encoding a putative tyrosme phosphatase suppresses lethality of an N-end rule-dependent mutant. Proc. Natl. Acad. Sci. USA 89, 2355-2359. Czkaynak, E., Finley, D.. Solomon, M. J., and Varshavsky, A. (1987). The yeast ubiquitin genes: a family of natural gene fusions. EMBO J. 6, 1429-1440. Park, E. C., Finley, D.. and Szostak, J. W. (1992). Astrategyforgeneration of conditional mutants by protein destabilization. Proc. Natl. Acad. Sci. USA 89, 1249-1252. Pickart, C. (1988). Ubiquitin activation and ligation. In Ubiquitin, Rechsteiner, ed. (New York; Plenum Press), pp. 77-99.

M.

Pringle, J F. (1975). Induction, selection and experimental uses of temperature-sensitive and other conditional mutants of yeast. In Methods in Cell Biology, Vol. 12, D. M. Prescott, ed. (New York: Academic Press), pp. 233-272. Ptashne, M.. and Gann, A F. (1990). Activators 346, 329-331.

and targets. Nature

Rechsteiner, M. (1991). Natural substrates of the ubiquitin proteolytic pathway. Cell 66, 615-618. Reiss, Y., Karm. D., and Hershko, A. (1988). Specificity of binding of NH&erminal restdue of proteins to ubiquitin-protein ligase. Use of amino acid derivatives to characterize specific binding sites. J. Biol. Chem. 263, 2693-2698. Rivett, A. J. (1990). Eukaryotic Biol. 2. 1143-1149.

protein degradation.

Curr. Opin. Cell

Rothman, J. E. (1989). Polypeptide chatn binding proteins: catalysts of protein folding and related processes in cells. Cell 59, 591-601. Scheffner. M., Werness. B. A., Huibregtse, J. M., Levine, A. J., and Howley, P. M. (1990). The E6 oncoprotein encoded by human papillomavirus types 16 and 18 promotes the degradation of ~53. Cell 63. 1129-1136.

in the endo-

Shanklin. J., Jabben. M., and Vierstra, R. D. (1989). Partial purification and peptide mapping of ubiquitin-phytochrome conjugates from oat. Biochemistry 28, 6028-6034.

Levinger. L., and Varshavsky, A. (1982). Selective arrangement of ubiquitinated and Dl protein-containing nucleosomes within the Drosophila genome. Cell 28, 375-385.

Sharon, G., Raboy, B., Parag, H. A., Drmitrovksy, D., and Kulka, R. G. (1991). RADG gene product of Saccharomyces cerevisiae requires a putative ubiquitin-protein ligase (E3) for the ubiquitination of certain proteins. J. Biol. Chem. 266, 15890-15894.

Wscher, B., and Eisenman, R. N. (1990). New light on Myc and Myb. Part I. Myc. Genes Dev. 4, 2025-2035.

Sherman, F., Stewart, J. W., andTsunasawa. S. (1985). Methionineor not methionme at the beginning of a protein? BioEssays 3, 27-31.

Madura, K., Prakash, S., and Prakash, L. (1990). Expression of the Saccharomyces cerevisiae DNA repair gene RADG that encodes a ubiquitin-conjugating enzyme increases in response to DNA damage and in meiosis but remains constant during the mitotic cell cycle. Nucl. Acids Res. 78, 771-778.

Siegelman. M., and Weissman. I. L. (1988). Lymphocyte homing receptors, ubiquitin, and cell surface proteins. In Ubiquitin, M. Rechsteiner, ed. (New York: Plenum Press), pp. 239-270.

Maurizi, M. R., Clark, W. P., Katayama, Y., Rudikoff, S., Pumphrey, J., Bowers, B., and Gottesman, S. (1990). Sequence and structure of

Soffer. R. L. (1980). Biochemistry and biology of aminoacyl-tRNAprotein transferases. In Transfer RNA: Biological Aspects, D. Sdll, J.

Klausner, R. D., and Sitia, R. (1990). Protein degradation plasmic reticulum. Cell 62, 611-614.

Simpson, M. V. (1953). The release of labeled amino acids from the proteins of rat liver slices. J. Biol. Chem. 207, 143-145.

Review: The N-End Rule 735

Abelson, and P. Ft. Schimmel, eds. (Cold Spring Harbor, New York: Cold Spring Harbor Laboratory), pp. 493-505. Sokolik, C. W., and Cohen, Ft. E. (1991). The structures of ubiquitin conjugates of yeast iso-2cytochrome c. J. Biol. Chem. 266, QIOO9107. Stadtman, E. R. (1990). Covalent modification reactions are marking steps in protein turnover. Biochemistry 29, 6323-6331. Strauss, E. G., Levinson, R., Rice, C. M., Dalrymple. J., and Strauss, J. H. (1966). Nonstructural proteins nsp3 and nsp4 of Ross River and CSNyong-nyong viruses: sequence and comparison with those of other alphaviruses. Virology 164, 265-274. Tilly, K., Spence, J., and Georgopoulos, C. (1989). Modulation of stability of the Escherichia coli heat shock regulatory factor L?, J. Eacteriol. 171, 1585-I 569. Tobias, J. W., and Varshavsky, A. (1991). Cloning and functional analysis of the ubiquitin-specific protease gene UBP7 of Saccharomyces cerevisiae. J. Biol. Chem. 266, 12021-12028. Tobias, J. W., Shrader, T. E., Recap, G., and Varshavsky, The N-end rule in bacteria. Science 254, 1374-1377.

A. (IQQi).

Townsend, A., Bastin, J., Gould, K., Brownlee, G.. Andrew, M.. Coupar, B., Boyle, D., Chan, S., and Smith, G. (1988). Defective presentation to class l-restricted cytotoxic T lymphocytes in vacciniainfected cells is overcome by enhanced degradation of antigen. J. Exp. Med. 768, 1211-1224. Varshavsky,

A. (1991). Naming a targeting signal. Cell 64, 13-15.

Varshavsky, A., Bachmair, A., Finley, D., Gonda, D., and Wunning, I. (1988). The N-end rule of protein turnover. In Ubiquitin, M. RechSteiner, ed. (New York: Plenum Press), pp. 287-324. Weintraub, H., Davis, R., Tapscott, S., Thayer, M., Krause, M., Benezra. R., Blackwell., T. K., Turner, D., Rupp, R., Hollenberg, S.,Zhuang, Y.. and Lassar. A. (1991). The myoD gene family: nodal point during specification of the muscle cell lineage. Science 257, 761-766. Wilkinson, K. D., Urban, M. K., and Haas, A. L. (1980). Ubiquitin is the ATP-dependent proteolytic factor I of rabbit reticulocytes. J. Biol. Chem. 255, 7529-7532. Yarden, Y., Escabedo, J. A., Kuang, W. J., Yang-Feng, T. L., Daniel, T. O., Tremble, P. M., Chen, E. Y., Ando, M. E., Harkins. R. N., Francke, U., Fried, V. A., Ullrich, A., and Williams, L. T. (1986). Structure of the receptor for platelet-derived growth factor helps define a family of closely related growth factor receptors. Nature 323,226-232.