Functional analysis of the proteasome regulatory particle

Abstract

We have developed S.cerevisiae as a model system for mechanistic studies of the 26S proteasome. The subunits of the yeast 19S complex, or regulatory particle (RP), have been defined, and are closely related to those of mammalian proteasomes. The multiubiquitin chain binding subunit (S5a/Mcb1/Rpn10) was found, surprisingly, to be nonessential for the degradation of a variety of ubiquitin-protein conjugates in vivo. Biochemical studies of proteasomes from rpn10 mutants revealed the existence of two structural subassemblies within the RP, the lid and the base. The lid and the base are both composed of 8subunits. By electron microscopy, the base and the lid correspond to the proximal and distal masses of the RP, respectively. The base is sufficient to activate the 20S core particle for degradation of peptides, but the lid is required for ubiquitin-dependent degradation. The lid subunits share sequence motifs with components of the COP9/signalosome complex, suggesting that these functionally diverse particles have a common evolutionary ancestry. Analysis of equivalent point mutations in the six ATPases of the base indicate that they have well-differentiated functions. In particular, mutations in one ATPase gene, RPT2, result in an unexpected defect in peptide hydrolysis by the core particle. One interpretation of this result is that Rpt2 participates in gating of the channel through which substrates enter the core particle.

Full text

Molecular Biology Reports 26: 21–28, 1999.
© 1999 Kluwer Academic Publishers. Printed in the Netherlands. 21
Functional analysis of the proteasome regulatory particle
Michael H. Glickman1,2, David M. Rubin1, Hongyong Fu3, Christopher N. Larsen1,
Olivier Coux4, Inge Wefes1, Günter Pfeifer5, Zdenka Cjeka5, Richard Vierstra3,Wolf-
gang Baumeister5, Victor Fried6& Daniel Finley1,∗
1Dept. of Cell Biology, Harvard Medical School, 240 Longwood Ave, Boston, MA 02115, USA; 2Present address:
Dept. of Chemistry, Technion-Israel Institue of Technology, Haifa, Israel; 3Dept. of Horticulture, University of
Wisconsin, Madison, WI 53706, USA; 4Present address: CRBM-CNRS, B.P. 5051, Route de Mende, Montpellier,
France; 5Max-Planck-Institut für Biochemie, D-82152 Martinsried, Germany 6Dept. of Cell Biology, New York
Medical College, Valhalla, NY 10595, USA; ∗Author for correspondence (Phone: (617) 432-3492; Fax: (617)
432-1144; E-mail: daniel−finley@hms.harvard.edu)
Key words: ATPase, proteasome, S. cerevisiae, ubiquitin
Abstract
We have developed S. cerevisiae as a model system for mechanistic studies of the 26S proteasome. The subunits
of the yeast 19S complex, or regulatory particle (RP), have been defined, and are closely related to those of
mammalian proteasomes. The multiubiquitin chain binding subunit (S5a/Mcb1/Rpn10) was found, surprisingly,
to be nonessential for the degradation of a variety of ubiquitin-protein conjugates in vivo. Biochemical studies of
proteasomes from 1rpn10 mutants revealed the existence of two structural subassemblies within the RP, the lid
and the base. The lid and the base are both composed of 8 subunits. By electron microscopy, the base and the lid
correspond to the proximal anddistal masses of the RP, respectively.The base is sufficient to activate the 20S core
particle for degradation of peptides, but the lid is required for ubiquitin-dependent degradation. The lid subunits
share sequence motifs with components of the COP9/signalosome complex, suggesting that these functionally
diverse particles have a common evolutionary ancestry. Analysis of equivalent point mutations in the six ATPases
of the base indicate that they have well-differentiatedfunctions. In particular, mutations in one ATPase gene, RPT2,
result in an unexpected defect in peptide hydrolysis by the core particle. One interpretation of this result is that
Rpt2 participates in gating of the channel throughwhich substrates enter the core particle.
Introduction
The ubiquitin-proteasome pathway is a major media-
tor of post-translational control in eukaryotes, which
functions in the control of cell proliferation, the cell
cycle, and other processes. Conjugation of ubiqui-
tin to substrates such as cyclins and p53 target them
for degradation by the proteasome. The proteasome
holoenzyme is a labile structure which readily disso-
ciates into multisubunit complexes known as the core
particle (CP) and the regulatory particle (RP; the RP
is also referred to as the 19S complex or PA700 in
mammals, and the µparticle in D. melanogaster).
The proteolytic active sites of the proteasome are
found in the CP, and are sequestered within the lumen
of this cylindrical complex [1,2]. Proteins apparently
enter the lumen of the CP through channels located
at each end of the cylinder. The binding of the RP
to the outer port of the CP channel implies that the
RP initiates substrate translocation into the CP [2,3].
Because the channel leading into the CP is narrow,
translocation is thought to require prior unfolding of
the substrate, perhaps by the RP itself. The selection
of ubiquitinated proteins for degradation is mediated
by the RP [3]. While the CP can hydrolyzesmall pep-
tides, its specific activity for such substrates is less
than that of the proteasome holoenzyme [4–6]. This
observation probably reflects that, in the free form

22
Figure 1. Comparison of proteasome purified from wild-type and
1rpn10 cells. (A) Wild-type and 1rpn10 lysates were fractionated
on columns containing DEAE-Affigel Blue, Resource Q, and S400
resin. Fractions from the DEAE-Affigeland S400 columns, contain-
ing the peak of peptidase activity, were visualized by nondenaturing
PAGE and fluorogenic peptide overlay. The observed species con-
tain either one regulatory particle (RP1CP) or two (RP2CP). A faint
species corresponds to free core particle (CP). Purified 1rpn10 frac-
tions from the S400 column contain two faster-migrating species
with peptidase activity. (B) Purified proteasomes (30 mg) from
both strains were tested for the ability to hydrolyze the fluorogenic
peptide suc-LLVY-AMC. Peptidase activity is given in arbitrary flu-
orescence units. Closed circles, wild-type proteasome holoenzyme
assayed in the presence of ATP.Open circles, wild-type proteasomes
preincubated (and assayed) in the absence of ATP to dissociate the
RP from the CP. Closed triangles, 1rpn10 in the presence of ATP;
Open triangles, dissociated 1rpn10 proteasomes assayed in the ab-
sence of ATP. (C, D) Purified proteasomes from both strains were
tested for the ability to hydrolyze 14 C-labeled casein or multiubiq-
uitinated 125I-labeled lys ozyme in the presence of ATP. Degradation
is measured as the production of TCA-soluble CPM at a given time
point. Background radioactivity was subtracted from all readings.
For details, see [21].
of the yeast CP, its channel exists predominantly in a
closed state [7].
Many studies of the proteasome have focused on
the CP, culminating in the solution of its crystal struc-
ture [2]. However, the dependence of the proteasome
on ubiquitin and ATP is conferred by the RP. Thus,
studies of the RP are likely to increase our under-
standing of substrate selection and other important
early steps in protein breakdown by the proteasome.
Here we discuss our recent studies of the RP from
S. cerevisiae.
Subunits of the yeast regulatory particle
To date, 18 subunits of the yeast RP have been iden-
tified (see Table 1 and[4,8]). Six of the RP subunits
are ATPases of the AAA family [9,10], and are desig-
nated [11] as Rpt1-6 (for Regulatory Particle Triple-A
protein). The remaining subunits of the RP are des-
ignated Rpn1-12 (Regulatory Particle Non-ATPase).
All of the known RP components are tightly associ-
ated with the particle, with the apparent exception of
Rpn4/Son1/Ufd5 [8,4]. Most of the known subunits
of the RP from yeast have mammalian homologs that
have been identified as RP subunits. Thus, the overall
conservation of the proteasome in eukaryotes is ex-
traordinary. The Rpt subunits are 66–76% identical
between yeast and humans, whereas the non-ATPase
subunits show a lower yet significant amount of se-
quence identity, typically in the range of 33–47% (Ta-

23
Table 1. Subunit composition of the S. cerevisiae RP
Subunit Yeast subunits Homologs % Identity1
Previous MW PI Human Bovine/ Other (Yeast-Human)
nomenclature Human
Rpn1 Hrd2/Nas1 109.4 4.32 S2/Trap-2 p97 mts4 41
Rpn2 Sen3 104.3 5.92 S1 p112 41
Rpn3 Sun2 60.4 5.37 S3 p58 33
Rpn4 Son1/Ufd5 60.1 5.20
Rpt1 Cim5/Yta3 52.0 5.32 S7/Mss1 76
Rpt2 Yta5 48.8 5.78 S4 p56 mts2 71
Rpt3 Yta2/Ynt1 48.0 5.38 S6/Tbp7 p48 MS73 66
Rpt4 Crl13/Sug2/Pcs1 49.4 5.53 S10b p42 CADp44 67
Rpt5 Yta1 48.2 4.93 S60/Tbp1 68
Rpn5 Nas5 51.8 5.79 p55 40
Rpn6 Nas6 49.8 5.90 S9 p44.5 41
Rpn7 49.0 5.16 S10 p44 36
Rpt6 Sug1/Cim3/Crl3 45.2 9.09 S8/Trip1 p45 m56 74
Rpn8 38.3 5.43 S12 p40 Mov-34 47
Rpn9 Nas7 45.9 5.51 p40.5 31
Rpn10 Mcb1/Sun1 29.7 4.73 S5a Mbp1/p54 34
Rpn11 Mpr1 34.4 5.66 Poh1 pad1 65.5
Rpn12 Nin1 31.9 4.8 S14 p31 mts3 32
1By Jotun-Hein method using MegAlign (gap penalty 11, gap length penalty 3).
ble 1). Alone among the Rpn subunits, Rpn11 is 65%
identical to its human counterpart. Of the known mam-
malian RP subunits, only S5b/p50.5[12,13] appears to
have no homologin yeast.
Role of Rpn10 in substrate recognition by the
proteasome
Substrate selection by the proteasome is thought to
be mediated by the interaction of RP subunits with
multiubiquitinated proteins. A human RP subunit,
designated S5a, has been found to bind both ubiquitin-
lysozyme conjugates and free ubiquitin chains and has
consequently been proposed to be the multiubiquitin
chain receptor of the proteasome [12,13]. The gene for
this subunit has been cloned from various eukaryotes
[12–19]. In yeast this gene was originally designated
MCB1, and is now knownas RPN10 (Table 1).
To test whether Rpn10 functions in the targeting
of ubiquitinated proteins to the proteasome, we con-
structed 1rpn10 knockout mutants in S. cerevisiae.
Unlike mutants in many other components of the
ubiquitin pathway, 1rpn10 knockout mutants are vi-
able and degrade many proteins normally [14]. Thus,
Rpn10 cannot be the only ubiquitin receptor for pro-
teasomal degradation. However, Rpn10 does play a
role in the degradation of some proteins, since 1rpn10
mutants are more sensitive to amino acid analogs than
wild-type yeast, and cannot degrade ubiquitin-Pro-
ßgalactosidase (Ub-Pro-ßgal), a known substrate of
the proteasome.
To further define the role of Rpn10 in conjugate
recognition and to identify the sites of multiubiquitin
chain binding, a series of deletions and point muta-
tions were constructed in the RPN10 gene [20]. A
stretch of conserved hydrophobic residues near the
C-terminus was found to be critical for recognizing
Lys48-linked multiubiquitin chains in vitro. However,
when these mutants were tested in vivo for their abil-
ity to complement 1rpn10 mutants, the multiubiquitin
chain recognition motif was found not to be required
for conferring resistance to amino acid analogs or for
restoring degradation of Ub-Pro-ßgal. Instead, a con-
served region near the N-terminus of Rpn10 was found
to be essential for these functions. Thus, a domain
near the N-terminus, and not the multiubiquitin chain-
binding site, appears to be most critical for Rpn10
function, at least in yeast.

24
Domain structure of the regulatory particle
The proteasome can be dissociated into two subcom-
plexes: the CP and the RP. The resolution of two
structural domains in the RP itself originated from
biochemical studies of the proteasome from 1rpn10
mutants [21]. After partial purification, wild-type and
1rpn10 proteasomes display comparable activity and
comparable electrophoretic mobility on nondenatur-
ing gels (Figure 1). Upon full purification, however,
1rpn10 proteasomes migrate faster during nondena-
turing gel electrophoresis than wild-type proteasomes.
This mobility shift results from the lack of 8 subunits
from the purified mutant proteasomes: Rpn3, Rpn5,
Rpn6, Rpn7, Rpn8, Rpn9, Rpn11, and Rpn12 [21].
When proteasomes are examined by electron mi-
croscopy, the wild-type regulatory particle appears
highly asymmetric and resembles an open wedge, the
proximal arm of which is bound to the core particle.
However, the mutant RP appears to be largely sym-
metric and hemispherical in form. In comparison to
wild-type, the key structural feature of 1rpn10 pro-
teasomes is that they are missing the distal arm of
the RP. This implies that the eight missing subunits
may constitute the distal mass of the RP. The proxi-
mal mass, including Rpn1, Rpn2, and Rpt1-Rpt6, will
be referred to as the base of the regulatory particle.
The mass density of the base in 1rpn10 proteasomes
predominantly overlaps with the proximal arm of the
wild-type RP.
Interestingly, the base alone is competent to ac-
tivate the peptide-hydrolyzing activity and casein-
hydrolyzing activity of the proteasome core particle
to levels almost equivalent to those of wild-type pro-
teasomes (Figure 1; [21]). However, degradation of
ubiquitin-protein conjugates (specifically, ubiquitin-
lysozyme conjugates) requires the intact RP. There-
fore, two basic activities of the regulatory particle can
be uncoupled; the proteasomal components that are
lost during purification are not required to activate the
core particle for peptide hydrolysis but are required for
ubiquitin-conjugate degradation.
1rpn10 proteasomes at an earlier stage of purifi-
cation are capable of degrading ubiquitin-protein con-
jugates with an activity close to that of wild-type [21].
Moreover, 1rpn10 mutants are competent to degrade
several ubiquitin-protein conjugates in vivo [14,20].
Thus, the ability of 1rpn10 proteasomes to degrade
ubiquitin-protein conjugate was apparently lost in par-
allel with the electrophoretic mobility shift during the
course of purification. Deletion of RPN10 decreases
the affinity of the lid for the base such that dissocia-
tion may occur as a result of supraphysiological salt
concentrations during chromatography.
The lid of the regulatoryparticle
The eight subunits that are released from the regu-
latory particle during purification of 1rpn10 protea-
somes copurify as a ∼400-kDa complex [21]. These
experiments, together with the electron microscopy
data mentioned above, indicate that the 400-kDa par-
ticle corresponds to the distal mass of the regulatory
particle. We refer to this stable subcomplex as the lid
of the proteasome. When lid particles are incubated
in the presence of complexes between the base and
the CP, intact proteasomes are efficiently reconstituted
[21].
Two structural motifs have recently been found
to be present in several proteasome subunits as well
as components of other protein complexes, such as
eIF3, and the COP9/signalosome complex (Table 2,
Figure 2, and [22,23]). The PINT/PCI domain is up
to 200 residues in length and is predicted to form an
α-helical structure [22,23]. These domains have been
found at the C-termini of Rpn3, Rpn5, Rpn6, Rpn7,
and Rpn9 (Figure 2; [22,23]. The MPN domain, found
in the N-termini of Rpn8 and Rpn11 (Figure 2), spans
approximately 140 residuesand is predicted to assume
aα/β structure [22]. Remarkably, all of the protea-
some subunits that possess these motifs are found in
the lid rather than the base of the RP or the CP. Fur-
thermore, all but one of the subunits of the lid contain
either an MPN or PINT/PCI motif. All 8 subunits of
the signalosome contain PINT/PCI or MPN motifs
as do 5 subunits of eIF3 [22,24]. Additionally, both
the signalosome and eIF3 contain one subunit with
specific similarity to Rpn8 and one with specific simi-
larity to Rpn11. The identification of a distinct domain
within the RP, the lid complex, provides an impor-
tant new basis for the interpretation of these sequence
relationships.
The sequence relationships described above further
validate consideration of the lid as a domain within the
RP. Indeed, identification of the lid domain helps to
define a new family of multisubunit assemblies, each
of which is broadly distributed among eukaryotes.
Seeger et al. [25] have proposed that the homologies
between proteasome and COP9/signalosome subunits
reflect common substrate binding sites, while Hof-
mann and Bucher [22] have suggested that the exis-

25
Table 2. A family of multisubunit complexes with common structural motifs
Complex Function kDa Total Subunits with Subunits with Refs.
subunits MPN motifs PINT/PCI motifs
RP lid protein degradation 400 8 2 5 21
COP9/Signalosome signal transduction 450 8 2 6 24,25
eIF3 initiation of translation 600 10 2 3 24,25,43, 44
tence of PINT/PCI and MPN domains in eIF3 and the
COP9/signalosome complex may indicate that these
complexes, like the lid, function in protein degrada-
tion. We propose that the relationship among these
particles reflects a common evolutionary ancestry, and
that a key step in the evolution of the modern protea-
some may have been the development of a binding in-
teraction between a precursor of the lid and a precursor
of the base. Simple homologs of the proteasome have
been described in archea, in which the RP consists
only of six identical ATPases [26,27]. The composi-
tion of the base particle is not markedly different from
this possible evolutionary precursor.
Models for functional cooperation between the
base and the lid
An important property of the base is that it is nearly
as efficient as the RP itself in stimulating the degra-
dation of peptides and the nonubiquitinated protein
substrate casein by the CP. Consistent with this obser-
vation, peptide hydrolysis by the CP can be inhibited
by mutations in the ATP-binding site of Rpt2, a sub-
unit of the base (see below and [28]). We suggest that
both results reflect a role of the base in opening of the
channel of the core particle. This model is in agree-
ment with structural data indicating that the channel
is closed in free core particles from yeast. Although
sufficient to stimulate peptide hydrolysis, the base fails
to promote the degradation of a ubiquitinated protein
substrate. A simple model accounting for these results
is that ubiquitin-protein conjugates might interact with
the lid through their ubiquitin moiety and with the base
through the substrate componentof the conjugate.
Two subunits of the base, Rpn1 and Rpn2, have
strong sequence similarity to the leucine-rich repeat
(LRR) domain, a common site for protein-protein in-
teraction [29]. The six ATPases of the base are also
likely to function through protein-protein interaction,
by analogy to the simple ATP-dependent proteases of
prokaryotes, which directly interact with substrates
[30, 31]. Thus, it is plausible that all eight components
of the base may engage in direct interactions with sub-
strates. This possibility is consistent with the structural
data which suggest that the substrate must translocate
through the center of the base to gain access to the CP.
It is clear that further experimentation will be required
to refine these simple models for substrate-proteasome
interactions. Given that lid and base complexescan be
purified in significant amounts, and can be used to re-
constitute the proteasome, it is now feasible to clarify
the functional distinctions between these domains of
the RP.
Proposed functions of ATP in the proteasome
ATP hydrolysis is strictly required for protein break-
down by the proteasome [5,6,32].The energy require-
ment has been suggested to reflect a role of the protea-
somal ATPases in substrate unfolding. In this model,
the role of ATP in the proteasome wouldbe analogous
to its role in the function of the ATPase ring complexes
known as chaperonins, which function to assist protein
folding. In the chaperonins, ATP hydrolysis is used to
drive large-scale structural transitions between states
of high- and low-substrate binding affinity, with an ap-
parently concerted motion of the subunits [33–36]. A
second possible function for the proteasomalATPases
was suggested by the crystal structure of the yeastCP.
Its cylindrical ends were found in a closed state [7],
suggesting that the proteasome channel is gated. The
ATPases, being localized to the base of the RP, are the
best candidates for mediating this gating.
Peptidase defect in rpt2RF proteasomes
The RPT genes are essential for vegetative growth
[7,37–39], but it is not clear whether this reflects a
functional requirement for each ATPase activity, or

26
Figure 2. Subunit organization of the proteasome regulatory particle. (A) Summary of identified domains in regulatory particle subunits.
Hatched box, PINT/PCI domain [22,23]. Striped box, MPN domain [22]. Open boxes, repeat motif with similarities to the LRR motif [29].
Rpn1 contains 9 such repeats, whereas Rpn2 contains 10. Checkered box, N-terminal conserved domain I of Rpn10 which is contained within
the N-terminal deletion used in this work [20]. Black box, the conserved domain III of Rpn10, containing the in vitro ubiquitin chain binding
site [20]. Stippled boxes represent AAA cassettes [an ATPase domain; [10]). The darker regions within the cassettes correspond to the highly
conserved Walker A and B motifs [10]. The sequence motifs that are most conserved among different RP proteins are the AAA cassettes. The
N-terminal portion of the PINT/PCI motif of Rpn9 is relatively divergent (K. Hofmann, pers. commun.). All domains are drawn to scale. (B)
A model for the regulatory particle. The proteasome is composed of two major subcomplexes: the CP and the RP. The RP itself contains the
∼600 kDa base and ∼360 kDa lid subcomplexes. Within the base are the six ATPases, or Rpt proteins; the two largest subunits, Rpn1 and
Rpn2; and Rpn10. The remaining eight Rpn subunits comprise the lid. Because the association of the lid and the base is relatively unstable in
the absence of Rpn10, this subunit is depicted at the interface of the two subcomplexes. The detailed arrangement of subunits within the lid and
base complexes is arbitrary. The CP consists of two types of heptameric rings of subunits (a and b). This figure taken from [21].
simply that in the complete absence of an Rpt subunit,
the proteasome fails to assemble. Active site mutants
provide an alternative way to study these genes. We
have constructed and characterized such mutants for
each of the RPT genes (see [28]). Here we will re-
strict discussion to one of these genes, RPT2.Ofthe
conservativesubstitutions of the invariant lysine of the
Walker-type P-loop, only rpt2K229R was lethal. To
obtain a viable rpt2R allele, an intragenic suppressor
mutation was obtained, in which Ser241 is replaced
by Phe. Ser241 is predicted to be outside of the active
site in the three-dimensional structure of the ATPase
domain [10]. The suppressor mutation alone (rpt2F)
had no detectable phenotypic effects in the absence
of the K229R substitution [28]. While the suppressor
mutation alleviates the lethal phenotype of the rpt2R
mutant, the rpt2RF strain remains strongly growth de-
fective, temperature sensitive, and unable to grow in
the presence of canavanine.
Based on structural studies, the channels of yeast
core particles are expected to exist predominantly in
a closed state [7]. Consistent with these data, the
peptidase activity of the yeast CP can be stimulated
approximately ten-fold by complex formation with the
RP [4]. Remarkably, this stimulation is not observed
in proteasomes from rpt2RF mutants tested. Indeed,
rpt2RF proteasomes have a lower peptidase activity on
a molar basis than do free CP [28]. Peptidase defects
have not been observed in any of the other rpt mu-
tants tested. Thus, Rpt2 apparently has a specialized
function among the proteasomal ATPases, involving
either the peptidase activity of the complex or access
of peptides to the lumen of the core particle.
The defect of rpt2RF proteasomes in hydrolyzing
small peptides that have no secondary structure indi-
cates that specific proteasomal ATPases havefunctions
other than, or in addition to, unfolding of protein
substrates. This is the first observation to link a pro-
teasomal ATPase to peptide hydrolysis. The RP was
previously known to stimulate the peptidase activity
of the CP [4,40], but the role of the ATPases in this
stimulation has been unclear. Moreover, the rpt2RF
effect is novel because it uncouples the stimulation of
peptidase activity from the role of ATP in holoenzyme
assembly.

27
Whether the rpt2RF mutation results in a defect
in allosteric control of the peptidases or a defect in
channel gating remains to be rigorously determined.
However, structural data provide significant support
for the existence of a gating mechanism, whereas there
is as yet little evidence for allosteric control of the
peptidase sites by the RP. Assuming that the rpt2RF
mutation causes a channel gating defect, the lack of
such an effect in four other rpt mutants suggests that
modulating the state of the proteasome channel is a
specialized function of Rpt2. However, it should be
noted that further characterization of the lethal rpt mu-
tants will be required to test whether this role of Rpt2
is unique. It is also uncertain whether the apparent
channel gating defect underlies the in vivo defects in
the turnover of ubiquitin-protein conjugates.
Concluding remarks
The proteasome holoenzyme was first identified in
mammalian cells in 1986 [41,42]. Since that time,
studies of the RP have focused mainly on the bio-
chemical and genetic identification of its subunits. As
described above, studies of the RP have recently be-
gun to turn towards mechanistic questions and the
functional characterization of subcomplexes from the
RP. Some of the more interesting issues are: How are
substrates recognized by the RP? How are substrates
unfolded? How is translocation of the substrate initi-
ated? What are the roles of ATP in the RP? How is the
proteasome channel gated? Because of the strengths
of yeast as a genetic system, it will provide unique
opportunities for addressing these issues.
References
1. Löwe J, Stock D, Jap B, Zwickl P, Baumeister W & Huber R
(1995) Science 268: 533–539
2. Baumeister W, Walz J, Zuhl F & Seemuller E (1998) Cell 92:
367–380
3. Pickart C (1997) FASEB J. 11: 1055–1066
4. Glickman MH, Rubin DM, Fried VA & Finley D (1998) Mol.
Cell. Biol. 18: 3149–3162
5. Hoffman L & Rechsteiner M (1994) J. Biol. Chem. 269:
16890–16895
6. DeMartino GN, Moomaw CR, Zagnitko OP, Proske RJ, Ma
CP, Afendis SJ, Swaffield JC & Slaughter CA (1994) J. Biol.
Chem. 269: 20878–20884
7. Groll M, Ditzel L, Löwe J, Stock D, Bochtler m, Bartunik HD
& Huber R (1997) Nature 386: 463–477
8. Fujimuro M, Tanaka K, Yokosawa H & Toh-eA (1998) FEBS
Lett. 423: 149–154
9. Patel S & Latterich m (1998) Trends Cell Biol. 8: 65–71
10. Beyer A (1997) Prot. Sci. 6: 2043–2058
11. Finley D, et al. (1998) Trends Biochem. Sci. 23: 244–245
12. Deveraux Q, Ustrell V, Pickart C & Rechsteiner M (1994) J.
Biol. Chem. 269: 7059–7061
13. Deveraux Q, Jensen C & Rechsteiner M (1995) J. Biol. Chem.
270: 23726–23729
14. van Nocker S, Sadis S, Rubin DM, Glickman MH, Fu H, Coux
O, Wefes I, Finley D & Vierstra RD (1996) Mol. Cell. Biol. 11:
6020–6028
15. van Nocker S, Deveraux Q, Rechsteiner M & Vierstra RD
(1996) Proc. Natl. Acad. Sci. 93: 856–860
16. Kominami K, Okura N, Kawamura M, DeMartino GN,
Slaughter CA, Shimbara N, Chung CH, Fujimura M, Yoko-
sawa H, Shimizu Y, Tanahashi N, Tanaka K & Toh-e A (1997)
Mol. Biol. Cell 8: 171–187
17. Haracska L & Udvardy A (1997) FEBS Lett. 412: 331–336
18. Haracska L & Udvardy A (1995) Eur. J. Biochem. 231: 720–
725
19. Young P, Deveraux Q, Beal RE, Pickart CM & Rechsteiner M
(1998) J. Biol. Chem. 273: 5461–5467
20. Fu H, Sadis S, Rubin DM, Glickman MH, van Nocker S,
Finley D & Vierstra RD (1998) J. Biol. Chem. 273: 1970–1989
21. Glickman MH, Rubin DM, Coux O, Wefes I, Pfeifer G, Cjeka
Z, Baumeister W, Fried VA & Finley D (1998) Cell 94: 615–
623
22. Hofmann K & Bucher P (1998) Trends Biol. Chem. 23: 204–
205
23. Aravind L & Ponting CP (1998) Prot. Sci. 7: 1250–1254
24. Wei N, Tsuge T, Serino G, Dohmae N, Takio K, Matsui M &
Deng, XW (1998) Curr. Biol. 8: 919–922
25. Seeger M, Kraft R, Ferrel K, Bech-Otschir D, Dumdey R,
Schade R, Gordon C, Naumann M & Dubiel W (1998) FASEB
J. 12: 469–478
26. Wolf S, Nagy I, Lupas A, Pfeifer G, Cejka Z, Müller SA, E ngel
A, De Mot R & Baumeister W (1998) J. Mol. Biol. 277: 13–25
27. Zwickl P, Woo KM, Klenk HP & Goldberg, AL (Submitted).
28. Rubin DM, Glickman MH, Larsen CN, Dhruvakumar S &
Finley D (1998) EMBO 17: 4909–4919
29. Lupas A & Baumeister W (1997) Trends Biochem. Sci. 22:
195–196
30. Gottesman S, Maurizi MR & Wickner S (1997) Cell 91: 435–
438
31. Gottesman S, Wickner S & Maurizi MR (1997) Genes Devel.
11: 815–823
32. Hershko A, Leshinsky E, Ganoth D & Heller H (1984) Proc.
Natl. Acad. Sci. USA 81: 1619–1623
33. Ditzel L, Lowe J, Stock D, Stetter KO, Huber H, Huber R &
Steinbacher S (1998) Cell 93: 125–138
34. Fenton WA & Horwich AL (1997) Protein Science 6: 743–760
35. Horovitz A (1998) Curr. Op. Struc. Biol. 8: 93–100
36. Kim S, Willison KR & Horwich AL (1994) Trends Biochem.
Sci. 19: 543–548
37. Ghislain M, Udvardy A & Mann C (1993) Nature 366: 358–
361
38. Schnall R, Mannhaupt G, Stuka R, Tauer R, Ehnle S, Schwar-
zlose C, Vetter I & Feldmann H (1994) Yeast 10: 1141–1155
39. Russell SJ, Sathyanarayana UG & Johnston SA (1996) J. Biol.
Chem. 271: 32810–32817

28
40. Chu-Ping M, Vu JH, Proske RJ, Slaughter CA & DeMartino
GN (1992) J. Biol. Chem. 269: 3539–3547
41. Hough R, Pratt G & Rechsteiner M (1986) J. Biol. Chem. 261:
2400–2408
42. Hough R, Pratt G & Rechsteiner M (1987) J. Biol. Chem. 262:
8303–8313
43. Asano K, Vornlocher HP, Richter-Cook NJ, Merrick WC, Hin-
nebusch AG & Hershey JWB (1997) J. Biol. Chem. 272:
27042–27052
44. Asano K, Kinzy TG, Merrick WC & Hershey JWB (1997) J.
Biol. Chem. 272: 1101–1109