Content uploaded by Hidde L Ploegh
Author content
All content in this area was uploaded by Hidde L Ploegh on Dec 08, 2014
Content may be subject to copyright.
A membrane protein required for
dislocation of misfolded proteins from
the ER
Brendan N. Lilley & Hidde L. Ploegh
Department of Pathology, Harvard Medical School, Boston, Massachusetts, 02115, USA
...........................................................................................................................................................................................................................
After insertion into the endoplasmic reticulum (ER), proteins that fail to fold there are destroyed. Through a process termed
dislocation such misfolded proteins arrive in the cytosol, where ubiquitination, deglycosylation and finally proteasomal proteolysis
dispense with the unwanted polypeptides. The machinery involved in the extraction of misfolded proteins from the ER is poorly
defined. The human cytomegalovirus-encoded glycoproteins US2 and US11 catalys e the dislocation of class I major histocompat-
ibility complex (MHC) products, resulting in their rapid degradation. Here we show that US11 uses its transmembrane domain to
recruit class I MHC products to a human homologue of yeast Der1p, a protein essential for the degradation of a subset of misfolded
ER proteins. We show that this protein, Derlin-1, is essential for the degradation of class I MH C molecules catalysed by US11, but
not by US2. We conclude that Derlin-1 is an important factor for the extraction of certain aberrantly folded proteins from the
mammalian ER.
Polypeptide chains destined for the secretory pathway emerge in the
lumen of the ER, where the process of folding is carefully monitored
to ensure that misfits are either repaired or destroyed. How cells
make the distinction between polypeptides that have yet to attain
their proper conformation and those that have exhausted their
folding options is not entirely clear
1
, but most terminally misfolded
polypeptides are transferred from the ER to the cytosol, where they
are destroyed by the proteasome in a ubiquitin-dependent manner
2
.
Proteins are thought to leave the ER through the Sec61 complex
3–7
,
but the involvement of the latter has yet to be demonstrated
conclusively. Substrates exposed to the cytosol are acted upon by
ER-associated components of the ubiquitin conjugation machin-
ery
8–10
, extracted from the ER membrane by the AAA ATPase
Cdc48p and its associated cofactors, Ufd1p and Npl4p (refs 11–
14), and degraded by the 26S proteasome
9,15
.Althoughsome
cytosolic components required for degradation have been charac-
terized in mammalian cells
2,11
, ER membrane proteins directly
involved in the dislocation reaction remain to be identified by
functional criteria.
Human cytomegalovirus encodes five proteins that disarm class I
MHC restricted antigen presentation, thus avoiding detection by
the immune system
16
. Two human cytomegalovirus proteins, US2
and US11, catalyse dislocation, the rapid transfer of the class I MHC
heavy chain (HC) from the ER to the cytosol, where N-glycanase,
ubiquitin-conjugating enzymes and finally the proteasome act on
them
3,17–20
. This pat hway resembles the mechanism by which
mammalian cells dispose of misfolded proteins, with US2 and
US11 conferring selectivity for class I molecules and accelerating
the rate constant of the dislocation reaction.
The transmembrane domain (TMD) of US11 is essential for its
ability to dislocate the class I HC
21
. The presence of a single
glutamine residue in the TM segment (Gln 192) suggested its
involvement in interactions of US11 with host proteins in the
plane of the membrane, confirmed by the inability of the Gln192Leu
(US11
Q192L
) mutant to sustain dislocation of the class I HC
21
.We
exploit the availability of this US11 point mutant in an affinity
purification approach aimed at isolating proteins that interact with
wild-type US11 (US11
WT
) but not with mutant US11 (US11
Q192L
).
We identify a human homologue of yeast Der1p, a protein required
for the degradation of a misfolded ER luminal protein, CPY*
(ref. 22), as the partner essential for US11 to perform its function.
In an accompany ing paper, Ye et al. used a different approach and
have reached a similar conclusion
23
. Whereas US11
WT
recruits the
class I HC to the newly identified protein, which we call Derlin-1,
US11
Q192L
fails to do so. A dominant-negative version of Derlin-1
impedes class I HC dislocation mediated by US11 but not by US2.
Combined, our results identify, by functional criteria, a mammalian
integral membrane protein required for the dislocation of certain
misfolded proteins from the ER to the cytosol.
Identification of US11-associated proteins
We developed an affinity purification approach that allowed us to
examine cellular proteins associated with US11
WT
and US11
Q192L
.
The purification strategy employed an amino-terminal affinity tag
composed of a haemagglutinin (HA) epitope followed by a short
spacer, a tobacco etch virus (TEV) protease cleavage site and an
additional spacer attached to the luminal domain and remainder of
US11 (Fig. 1a). HA–US11
WT
or HA–US11
Q192L
behaved like their
untagged counterparts in their ability to cause class I HC dis-
location, and migrated on SDS–polyacrylamide-gel electrophoresis
(SDS–PAGE) at the expected molecular mass of 42 kDa (Fig. 1b).
We performed a large-scale purification from digitonin lysates of
control U373 cells, and from cells that expressed either HA–US11
Wt
or HA–US11
Q192L
. Both Coomassie blue staining and subsequent
analysis by tandem mass spectrometr y (MS/MS) showed that
a significant number of polypeptides associated equally with
HA–US11
WT
and HA–US11
Q192L
(Fig. 1c) and therefore that
associations with this group of proteins are unlikely to account
for the mutant phenotype of US11
Q192L
. These proteins include ER
chaperones such as immunoglobulin heavy-chain binding protein
(BiP), calnexin, the AAA ATPase p97 and three subunits of the
oligosaccharide transfera se complex (OST48, ribophorin I and
ribophorin II). The significance of these interactions will be
addressed elsewhere. Consistent with our previous findings
21
was
our observation that the class I HC co-purified with HA–US11
Q192L
.
One 22-kDa polypeptide associated specifically with HA–US11
WT
and not with HA–US11
Q192L
(Fig. 1c). MS/MS analysis of a tryptic
digest of the 22-kDa protein identified two peptides from a
articles
NATURE | VOL 429 | 24 JUNE 2004 | www.nature.com/nature834
© 2004
Nature
Publishing
Group
predicted open reading frame, MGC3067 (NCBI GeneID 79139),
which encodes a small, hydrophobic protein of 251 amino acids
predicted to span the lipid bilayer four times, with both its amino
and carboxy terminus in the cytosol (Fig. 1d). Immunoblots using
an antiserum raised against peptides from the predicted protein
sequence of MGC3067 show its presence at equivalent levels in
HA–US11
WT
and HA–US11
Q192L
cell lines (Fig. 1e, left panel). The
22 kDa protein associates with HA–US11
WT
but not with HA–
US11
Q192L
(Fig. 1e, right panel).
MGC3067 encodes an evolutionarily conserved protei n with
homologues in all eukaryotes examined. Although the functions
of MGC3067 homologues in multicellular organisms are unknown,
one of the yeast proteins that bears limited similarity (about 10%
identity) to MGC3067 is Der1p, a known factor in the degradation
of misfolded ER proteins
22,24
. Another yeast open reading frame,
YDR411c, encodes a protein more similar to MGC3067 (about 20%
identity); like Der1p, YDR411cp is an ER-resident protein induced
by ER stress
25,26
.
MGC3067 and its homologues contain a domain named for yeast
Der1p (Der1-like domain, Pfam accession no. PF04511; ref. 27).
This domain is about 200 amino acids in length and is predomi-
nantly hydrophobic, with pockets of highly conserved and invariant
residues, many of which flank the predicted transmembrane regions
(Fig. 2a). Because MGC3067 contains a Der1-like domain and
shows similarity to Der1p, we name the protein product of
MGC3067 Derlin-1 (Der1-like protein 1).
Mammals have two additional Der1-like domain-containing
proteins homologous to Derlin-1, which we call Derlin-2 (NCBI
GeneID 51009, also known as F-LANa; ref. 28) and Derlin-3 (NCBI
GeneID 91319, murine orthologue known as IZP6; ref. 29). Derlin-2
and Derlin-3 are about 70% identical (see alignment in Supplemen-
tary Fig. S1) and probably originated from a gene duplication event
in mammals. The evolutionary relationships between the Derlin
proteins and their putative orthologues are shown in Fig. 2b.
Derlin-1 is a w idely expressed protein, w ith strong signals
obtained by immunoblotting from liver, spleen, pancreas, lung,
thymus and ovary. Despite the presence of Derlin-1 sequences in
brain cDNA libraries (NCBI UniGene cluster Hs.241576), we did
not detect immunoreactivity in brain (Fig. 2c).
The inferred integral membrane disposition of Derlin-1 was
confirmed by fractionation of U373 microsomes. Neither alkaline
nor urea extraction affected the membrane association of Derlin-1
or the known ER membrane protein calnexin, but each removed the
soluble ER protein PDI (protein disulphide isomerase) from the
Figure 1 Identification of US11-associated proteins. a, Schematic representation of
constructs used for affinity purification. b, Stability of the class I HC in cells expressing
US11
WT
, US11
Q192L
or their epitope-tagged counterparts. IP, immunoprecipitation.
c, Identification of HA–US11
WT
and HA–US11
Q192L
-associated proteins from the cell lines
indicated (2, control U373 cells; WT, HA–US11
WT
; Q192L, HA–US11
Q192L
). Polypeptides
identified by MS/MS analysis are indicated. CNX, calnexin. d, Sequence of human
Derlin-1. Peptides identified by MS/MS analysis are boxed, predicted TMDs (using
TMpred) are underlined, and peptide sequences used for antibody production are in bold.
e, Use of antiserum against Derlin-1 confirms the interaction of Derlin-1 with HA–US11
WT
.
Equal amounts of lysate or TEV eluate from the specified cell lines (as in c) were analysed
by immunoblotting with anti-US11 or anti-Derlin-1 antibodies.
articles
NATURE | VOL 429 | 24 JUNE 2004 | www.nature.com/nature 835
© 2004
Nature
Publishing
Group
particulate fraction (Fig. 2d). Derlin-1 is confined to the ER,
because it co-localizes with the ER chaperone calnexin (Fig. 2e).
Mild proteolysis eliminated the C-terminal epitope of Derlin-1 in
intact and detergent-disrupted microsomes, but the luminal protein
GRP94 remained intact unless detergent was added. Intact micro-
somes treated with prot ease did not yield Derlin-1 digestion
intermediates (Fig. 2f). Thus, at least the C terminus of Derlin-1
resides in the cytosol.
Association of the class I HC with Derlin-1
The anti-Derlin-1 antiserum was used for immunoprecipitations
from radiolabelled digitonin lysates of US11
WT
and US11
Q192L
cell
Figure 2 Characterization of Derlin-1. a, Alignment of Derlin-1 with putative orthologues.
Sequences (GenBank accession numbers indicated) from Danio rerio (Dr, AAH45413),
Drosophila melanogaster (Dm, NP_608632), Caenorhabditis elegans (Ce, NP_492721),
Arabidopsis thaliana (At, AAD03446) and Saccharomyces cerevisiae (Sc, YDR411cp:
AAB64889, Der1p: P38307) were aligned with human (Hs, NP_077271) and mouse
(Mm, NP_077169) Derlin-1 by using ClustalW. Residues identical to those in human
Derlin-1 are shaded, and the consensus above the aligned sequences indicates residues
that are identical in six or more of the aligned sequences. b, Phylogenetic analysis of
Der1-domain-containing proteins generated by using ClustalW and the Megalign
program. The species of origin and GenBank accession numbers are shown or are as
follows: Hs Derlin-2, NP_057125; Mm Derlin-2, NP_291040; Hs Derlin-3, AAH57830;
Mm Derlin-3, BAB32788. c, Analysis of Derlin-1 expression in mouse tissues by
immunoblotting. d, Derlin-1 is an integral membrane protein. Immunoblot analysis of
microsomes treated with the indicated reagents (HB, homogenization buffer) performed
on total (T), particulate (P) and soluble (S) fractions by using antibodies against calnexin
(anti-CNX), PDI and Derlin-1. e, Derlin-1 is an ER-resident protein. U373 cells were
stained with antibodies against calnexin (CNX), Derlin-1 (anti-Derlin-1) or Derlin-1
antiserum depleted of specific antibodies (anti-Derlin-1 D). Merging the images shows
the co-localization of Derlin-1 with calnexin. f, Analysis of Derlin-1 topology by protease
protection and immunoblotting with anti-GRP94 and anti-Derlin-1 antibodies. The
removal of the C terminus of Derlin-1 in the absence of detergent (Nonidet P40) is
consistent with the topology proposed.
articles
NATURE | VOL 429 | 24 JUNE 2004 | www.nature.com/nature836
© 2004
Nature
Publishing
Group
lines. Labelled U S11
WT
immunoprecipitates with Derlin-1 but
US11
Q192L
does not (Fig. 3b, compare lanes 1 and 2 with 5 and
6), which is consistent with the results of the large-scale purification.
Re-immunoprecipitation from the material captured by the anti-
Derlin-1 antibodies with anti-US11 serum confirms the presence of
US11
WT
and the absence of US11
Q192L
(see Fig. 5e, lower panel).
Labelled US11
WT
associates with Derlin-1 very soon after synthesis
(Fig. 3b), indicating that this complex forms rapidly. The detection
of radiolabelled Derlin-1 requires prolonged exposure of the auto-
radiogram (data not show n), possibly reflecting a low level of
Derlin-1 synthesis relative to US11
WT
and the class I HC during
the 30 min of pulse labelling.
Does the class I HC substrate also occur in a complex that
contains Derlin-1? We first showed in a pulse–chase experiment
that the class I HC disappears rapidly from US11
WT
cells (Fig. 3a,
lanes 1 and 2). Inclusion of proteasome inhibitors in pulse–chase
experiments with US11 cells causes the transient accumulation of a
deglycosylated class I HC intermediate whose N-linked glycan has
been removed by a cytoplasmic N-glycanase (Fig. 3a, lanes 3 and 4;
ref. 17). The US11
Q192L
mutant is incapable of accelerating degra-
dation of the class I HC (Fig . 3a, lanes 5–8). In cells expressing
US11
WT
, immunoprecipitations for Derlin-1 also recovered the
class I HC (Fig. 3b, lane 1). The class I HC associates with Derlin-
1 at early times after synthesis, but later, coincident with the
degradation of the class I HC, this association is lost (Fig. 3b,
lanes 1 and 2). In cells that express US11
Q192L
, no class I HC is
associated with Derlin-1 (Fig. 3b, lanes 5 and 6). Immunoprecipita-
tions with either preimmune serum or immune serum that had
been depleted of specific anti-Derlin-1 antibodies did not recover
any detectable amount of class I HC or US11
WT
(data not shown).
US11 therefore uses its TMD to recruit the class I HC into a complex
with Derlin-1.
In US11
WT
cells treated with proteasome inhibitor, we find
increased amounts of glycosylated class I HC in association with
Derlin-1 relative to non-treated cells (Fig. 3b, compare lanes 3 and
1). Indeed, prolonged treatment with proteasome inhibitor delays
the dislocation of class I HC to the cytosol
30
. When Derlin-1 was
Figure 3 Association of US11
WT
and the class I HC with Derlin-1. a, Digitonin lysates from
the indicated cell lines were used for direct immunoprecipitation (IP) with anti-heavy chain
(anti-HC) antibodies. Treatment with proteasome inhibitors (ZL
3
VS) leads to the
accumulation of deglycosylated class I HC (HC 2 CHO) only in US11
WT
cells. The majority
of the class I HC is in the deglycosylated form after the pulse. b, Digitonin lysates from the
indicated cell lines were immunoprecipitated with anti-Derlin-1 antibodies. Half of the
recovered material was analysed directly and the remaining half was used for re-
immunoprecipitation with anti-heavy chain antibodies (shown in c). Only US11
WT
is
recovered with Derlin-1, and predominantly glycosylated (HC þ CHO) class I HC is
recovered with Derlin-1 only in US11
WT
cells. Additional radiolabelled polypeptides of
105 kDa (p105) and 21 kDa (p21) associate with Derlin-1 in US11
WT
cells but not in
US11
Q192L
cells. c, Re-immunoprecipitation of the class I HC (anti-HC) from Derlin-1
immunoprecipitations in b confirms the presence of the class I HC. The autoradiogram in a
was exposed for 1 day; those in b and c were exposed for 6 days.
Figure 4 Expression of a Derlin-1 dominant-negative construct impedes US11-mediated
class I HC degradation. a, The stability of the class I HC in US11
WT
cells is substantially
increased by the expression of Derlin-1
GFP
. The graph shows quantification of the
experiment. Immunoprecipitation (IP) for transferrin receptor (anti-TfR) shows that
glycoprotein maturation is not affected in cells that express Derlin-1
GFP
or Derlin-2
GFP
. b,
The complex of US11
WT
and the class I HC persists when Derlin-1
GFP
is expressed.
Digitonin lysates were immunoprecipitated for US11, followed by re-immunoprecipitation
with anti-HC and anti-US11 antibodies. c, The complex of the class I HC and Derlin-1
persists in the presence of Derlin-1
GFP
. Digitonin lysates were immunoprecipitated for
Derlin-1, followed by re-immunoprecipitation with anti-HC antibodies.
articles
NATURE | VOL 429 | 24 JUNE 2004 | www.nature.com/nature 837
© 2004
Nature
Publishing
Group
immunoprecipitated from cells treated with proteasome inhibitor,
predominantly glycosylated class I HC was immunoprecipitated
with it after the onset of the chase, despite the fact that more than
50% of the total class I HC population was in the deglycosylated
form (Fig. 3, lane 3, compare panels a and b). A small amount of
deglycosylated class I HC was also immunoprecipitated with
Derlin-1 (Fig. 3b, c, lanes 3 and 4). Therefore, although Derlin-1
associates preferentially with glycosylated class I HC, it can interact
with class I HC that has been exposed to the cytosol.
In cells that express US11
WT
, several other proteins occur in
association with Derlin-1 in addition to US11
WT
and the class I HC.
Proteins of about 105 and 21 kDa associated with Derlin-1 after the
onset of the chase in cells expressing US11
WT
but not in those
expressing US11
Q192L
(Fig. 3b, compare lanes 1–4 with lanes 5–8).
The 105-kDa species is not the AAA ATPase, p97 (ref. 11) as
demonstrated by re-immunoprecipitation (data not shown).
These labelled species were not recovered with Derlin-1 at later
time points. Although we do not yet know the identity of these
proteins, their association with Derlin-1 in US11
WT
cells indicates
that they, too, might have a function in dislocation of the class I HC
mediated by US11.
Derlin-1 is required for dislocation
To test the function of Derlin-1 in the US11-mediated dislocation of
class I HC, we created a dominant-negative construct of Derlin-1
and, as a specificity control, Derlin-2. On the basis of the structure
of Derlin-1 and related proteins, we proposed that the addition of a
folded domain to what is predicted to be a flexible cytoplasmic tail
would interfere with interactions on the cytoplasmic side of the ER
but would leave intact the intramembrane interaction with the
US11 TMD. We therefore added green fluorescent protein (GFP) to
the C terminus of Derlin-1 and Derlin-2 (Derlin-1
GFP
and Derlin-
2
GFP
) and expressed these constructs in US11
WT
cells. Whereas
Derlin-1
GFP
was expressed at a level roughly equivalent to that of
endogenous Derlin-1, Derlin-2
GFP
was significantly overexpressed
relative to endogenous Derlin-2, as assessed by immunoblotting
(Supplementary Fig. S2a). Derlin-1 and Derlin-2
GFP
localize to the
ER (Supplementary Fig. S2b), and their expression did not
obviously perturb ER structure
31
(data not shown) or trafficking
of the transferrin receptor as assessed by N-glycan maturation
(Fig. 4a, top panels).
In US11
WT
cells expressing Derlin-1
GFP
, the class I HC was more
stable than in US11
WT
cells expressing Derlin-2
GFP
(Fig. 4a, middle
panels). The half-life of the class I HC in US11
WT
cells is extended
from less than 5 min (ref. 17) to more than 30 min in the presence of
Derlin-1
GFP
(Fig. 4a).
Shortly after synthesis, but before dislocation, the class I HC
exists as part of a complex in the ER that includes US11
WT
(ref. 21)
and Derlin-1 (Fig. 3). As dislocation proceeds, the class I HC is lost
from this complex. In cells that express Derlin-1
GFP
we observe a
prolonged association of the class I HC with both US11
WT
(Fig. 4b)
and with Derlin-1 (Fig. 4c). The presence of Derlin-1
GFP
thus
delays the exposure of the class I HC to the cytosol. Expression of
Derlin-1
GFP
does not preclude binding of the class I HC substrate to
either US11
WT
or Derlin-1 itself.
Specificity of Derlin-1 action
Both US11 and US2 catalyse dislocation of the class I HC, resulting
in intermediates and a final outcome that are similar
3,17
. Having
established that Derlin-1
GFP
blocks dislocatio n of class I HC
molecules in US11
WT
cells, we examined its effects on US2-
dependent dislocation of class I molecules.
At levels of Derlin-1
GFP
expression that block US11-dependent
dislocation (Supplementary Fig. S2a, c), the class I HC molecules in
cells that express US2 (US2 cells) are not affected (Fig. 5a, middle
panels ). Thus, the inhibition of dislocation by Derl in-1
GFP
in
US11
WT
cells is specific and cannot be attributed to a systemic
inhibitory effect on ER function. In US2 cells, we do not observe an
association of the class I HC with Derlin-1 (Fig. 5e), which is
consistent with a lack of Derlin-1 involvement in US2-mediated
class I HC dislocation. When US2 cells are treated with proteasome
inhibitors, the deglycosylated class I HC is observed, as it is in
Figure 5 Expression of a Derlin-1 dominant-negative construct does not inhibit
US2-mediated class I HC degradation, but blocks degradation of the US2 protein itself.
a, Stability of the class I HC is not affected in US2 cells expressing Derlin-1
GFP
compared
with control US2 cells, but degradation of glycosylated US2 protein (US2þCHO) is blocked
in the presence of Derlin-1
GFP
. Degradation of the non-glycosylated US2 (US22CHO) is
not affected. Immunoprecipitation (IP) for transferrin receptor (anti-TfR) shows that
glycoprotein maturation is not affected in cells that express Derlin-1
GFP
or Derlin-2
GFP
.
b–d, Quantification of the results in a: b, class I HC; c, US2þCHO; d, US22CHO. e, The
US2 protein, but not the class I HC, associates with Derlin-1 in US2 cells. Digitonin lysates
from the indicated cell lines were immunoprecipitated with anti-Derlin-1 antibodies
followed by re-immunoprecipitation first with anti-HC (upper panel) antibodies followed by
anti-US11 or anti-US2 antibodies (lower panel).
articles
NATURE | VOL 429 | 24 JUNE 2004 | www.nature.com/nature838
© 2004
Nature
Publishing
Group
US11
WT
cells
3
. However, in US2 cells treated with proteasome
inhibitors, we did not observe deglycosylated class I HC in associ-
ation with Derlin-1 (data not shown), whereas we did observe a
small amount of this class I HC intermediate associated with Derlin-
1 in US11
WT
cells (Fig. 3b, lanes 3 and 4). This supports the idea
that, in US11
WT
cells, the interaction between Derlin-1 and the
deglycosylated class I HC is specific and not due to an interaction
after lysis. The differential inhibition of US11- versus US2-mediated
dislocation by Derlin-1
GFP
suggests that US2 and US11 exploit
different pathways
32
for dislocation of the same substrate proteins.
Analysis of US2 cells also allowed us to address the fate of US2
itself, a short-lived type I membrane protein that lacks a cleavable
signal sequence
33
. The US2 protein exists in two forms in the cell: an
ER-inserted, glycosylated species, and a non-glycosylated, cytosolic
version. The cytosolic, non-glycosylated form is derived from the
failure of a fraction of the newly synthesized US2 to insert into the
ER. The ER-resident form o f US2 has a short half-life, and
represents an additional ER degradation substrate
33
.
In US2 cells that express Derlin-1
GFP
, US2 is stabilized but the
degradation of class I HC molecules and non-glycosylated US2
continues (Fig. 5a–d). A small amount of glycosylated US2 associ-
ates with Derlin-1 (Fig. 5e). The degradation of two type I
membrane proteins, the class I HC in US11
WT
cells and the US2
protein, therefore requires Derlin-1. We have not yet identified a
functional role for Derlin-2 in the degradation of ER proteins, but
the Derlin-2
GFP
fusion protein in no way impairs dislocation of the
class I HC in either US11
WT
cells (Fig. 4) or US2 cells (Fig. 5).
Discussion
We identified Derlin-1 as an ER membrane protein essential for
US11-mediated dislocation of the class I HC. Our approach rests on
the assumption that the active form of US11 should interact with
partner proteins no longer available to the inactive US11
Q192L
.
Indeed, affinity purification identified Derlin-1 as a protein that
meets this criterion. The use of a dominant-negative version of
Derlin-1 shows its direct involvement in US11-mediated dislo-
cation. We emphasize the importance of a functional readout of
US11 activity, as exemplified both by the distinction betw een
US11
WT
and US11
Q192L
and by the ability of a dominant-negative
version of Derlin-1 to block US11-dependent dislocation. Although
mere interactions between proteins involved in dislocation might
suggest functional relevance, as observed for US11
WT
, class I HC
and Derlin-1, these observations are confirmed by analysis of
US11
Q192L
and the effects of the dominant-negative form of
Derlin-1 on class I HC dislocation.
Derlin-1 is weakly similar to yeast Der1p, whose mechanistic
contribution to ER degradation is unknown. The degradation of
only a subset of ER substrates requires the involvement of Der1p,
which is consistent with the notion that multiple pathways operate
to clear misfolded proteins from the ER
22,24,34–38
. We show that
Derlin-1 mediates the degradation of two different type I membrane
proteins, the class I HC and US2. At present we know of no
additional proteins that require Derlin-1 for their degradation,
but we consider it likely that Derlin-1 acts to degrade cellular ER
proteins independently of the US11 and US2 systems studied here.
The possible role(s) of additional Der1-like domain-containing
proteins in ER degradation have yet to be examined. These proteins
might serve a function similar to that of Der1p and Derlin-1 but
might operate on distinct groups of substrate proteins. In v iew of
the diverse nature of proteins that exist in the secretory pathway,
there might be factors that operate independently of Derlin-1 to
achieve quality control.
The Sec61 complex might be the conduit by which proteins exit
the ER
3–7
. However, the structure of the analogous SecY complex
from archaea suggests a strict upper limit to the diameter of the
pore that spans the lipid bilayer
39
. What is currently known about
the dislocation reaction is hard to reconcile with the observed
dimensions of the SecY complex. Class I HC fusion proteins
prepared with the tightly folded domain of GFP, or with dihydro-
folate reductase stabilized by methotrexate analogues, are still
dislocated with kinetics seen for the unmodified class I HC
40,41
,as
is the high-mannose-bearing class I HC
20
, further indicating a
dislocation pore of considerable size.
We observe a significant amount of the glycosylated class I HC in
association with Derlin-1 before dislocation to the cytosol. When
cells are treated with proteasome inhibitor, some deglycosylated
class I HC occurs in association with Derlin-1. Expression of a
Derlin-1 dominant-negative construct causes g lycosylated class I
HC to persist in a complex with US11
WT
and Derlin-1, which is
indicative of class I HC retention in the ER. Derlin-1 is therefore
positioned in the ER membrane in such a way that it can interact
with substrate proteins both before and immediately after their
dislocation to the cytosol. The presence in Derlin-1 of four trans-
membrane segments now suggests further possibilities for the
creation of a protein-conducting channel
—
the disloco n
—
that
might be gated through the oligomerization of Derlin-1 or through
the association of Derlin-1 with accessories yet to be identified. The
formation of this channel might indeed be independent of Sec61.
US11 is a small 215-residue glycoprotein that consists of at least
two modules. The luminal domain is sufficient for interaction with
the class I HC
21
, whereas the TMD is required for interaction with
Derlin-1. Even though the interaction between the US11 TMD and
Derlin-1 might be indirect, contacts formed by Gln 192 in US11 are
essential for the interaction between US11 and Derlin-1. The US11
protein is remarkable for the speed with which it targets the class I
HC molecules for degradation: within minutes of its synthesis, the
class I HC is destroyed. We propose that US11 does so by recruit-
ment of the class I HC to Derlin-1. For other ER degradation
substrates, they must first be recognized as misfolded or un-
assembled, and then must be targeted to the machinery that per-
forms the dislocation reaction. We propose that the recruitment of
such misfolded proteins to Der1-like domain-containing proteins
represents a scheme for clearing the ER of the requisite range of
substrate proteins. A
Methods
DNA constructs and cell lines
HA–US11
WT
and HA–US11
Q192L
were constructed with standard methods, and sequence-
verified constructs were transferred into a pLNCX-based vector
21
for retrovirus
production. The murine K
b
signal sequence was included to direct the proteins to the ER.
Human Derlin-1 and Derlin-2 were cloned from U373 astrocytoma cDNA by using
polymerase chain reaction with reverse transcription, and GFP fusions were made by the
insertion of Derlin-1 and Derlin-2 complementar y DNAs into the pEGFP-N1 vector
(Clontech). The Derlin-1
GFP
and Derlin-2
GFP
constructs were subcloned into the pLHCX
retroviral expression vector (Clontech). The US11
WT
and US11
Q192L
cell lines have been
described
21,42
. US2 cells were derived by transducing U373 cells with a retroviral construct
containing the full-length US2 gene, followed by antibiotic selection and cloning by
limiting dilution. Methods for the production of retroviruses, infection of U373 cells and
subsequent antibiotic selection have been reported
21
. All U373 astrocytoma cell lines used
were maintained in culture as described
42
. Cells transduced with LHCX-based vectors were
selected with 125
m
gml
21
hygromycin B (Roche).
Antibodies and immunoblotting
Most of the antibodies used in this study have been reported
21,43
. Antibodies against
transferrin receptor (H68.4) and p97 were purchased from Zymed and Research
Diagnostics, respectively. Alexa-488-conjugated goat anti-mouse and Alexa-568-
conjugated goat anti-rabbit were from Molecular Probes. Anti-Derlin-1 and anti-De rlin-2
antisera were generated by immunizi ng rabbits with peptides coupled to keyhole-limpet
haemocyanin through an added Cys residue. The Derlin-1 sequences used were
SDIGDWFRSIPAITR(C), (C)RNFLSTPQFLYRWLPSRR and (C)RHNWGQGFRLGDQ.
Derlin-2 sequences used were AYQSLRLEYLQIPPVSR(C), (C)KAIFDTPDEDPNYN and
(C)EERPGGFAWGEGQRLGG. Antibodies specific to the C terminus of Derlin-1
(RHNWGQGFRLGDQ) were affinity purified as described
44
. Affinity-purified anti-
Derlin-1-C antibodies were used for immunoprecipitations, examin ation of Derlin-1
topology, and immunohistochemical analysis. Crude anti-Derlin-1 serum was used for
immunoblotting. Anti-GFP and anti-GRP94 sera were generated by immunizing rabbits
with bacterially expressed GFP and GRP94, respectively. Immunoblotting experiments
were performed as described
21
.
articles
NATURE | VOL 429 | 24 JUNE 2004 | www.nature.com/nature 839
© 2004
Nature
Publishing
Group
Purification of HA–US11
WT
and HA–US11
Q192L
, and MS/MS analysis
Cells (10
8
) were lysed in 30 ml of ice-cold lysis buffer (1% digitonin in 25 mM Tris-HCl pH
7.4, 150 mM NaCl, 5 mM MgCl
2
, 1 mM CaCl
2
with 1 mM phenylmethylsulphonyl fluoride
and complete protease inhibitors (Roche)). The lysate was cleared of debris by
centrifugation at 20,000 g for 15 min and was added to 0.35 ml of 12CA5-coupled
Sepharose
21
. The affinity resin was washed with 35 ml of wash buffer (composition as lysis
buffer, except with 0.1% digitonin and without protease inhibitors). Bound material was
eluted by treatment of the resin for 1 h at 25 8C with 80 units of TEV protease (Invitrogen)
in 0.25 ml of wash buffer. The eluate was exchanged into 20 mM NH
4
CO
3
pH 8.0 with
0.1% SDS with the use of Sephadex G-25 resin and was then freeze-dried. Polypeptides
were separated by 10% reducing SDS–PAGE and were revealed by Coomassie blue
staining. Polypeptides of interest were excised and subjected to trypsinolysis as
described
45
. Digested samples were separated by liquid chromatography and analysed by
MS/MS as described
46
. After data acquisition and processing, MS/MS data were searched
against the NCBI non-redundant database using the MASCOT program (Matrix Science).
Pulse–chase experiments and SDS–PAGE
Methods for the treatment of cells with the proteasome inhibitor ZL
3
VS (ref. 19), pulse
labelling, cell lysis, immunoprecipitation and analysis of class I HC stability in US11 and
US2 cells have been described
21
. Preparation of digitonin lysates and re-
immunoprecipitation of specified polypeptides from material captured from digitonin
lysates was performed as described
21
. N-Ethylmaleimide was included during digitonin
lysate preparation at a concentration of 2.5 mM. Methods for SDS–PAGE and
fluorography have been described
47
. In experiments examining proteins associated with
Derlin-1, the pulse labelling period was extende d to 30 min, and the cells were chased for
60 min.
Analysis of Derlin-1
Microsomes from U373 cells were produced as described
32
and were treated with 0.1 M
Na
2
CO
3
, pH 11.6, 2.5 M urea or 1% SDS as reported
24
. Treatment of U373 microsomes
with protease K was performed as described
48
. For the analysis of Derlin-1 tissue
distribution, organs from a PBS-perfused C57/BL6 mouse were removed and
homogenized in 100 mM Tris-HCl pH 7.6 with 8 M urea and 1% SDS. Equal amounts of
protein were precipitated with trichloroacetic acid and a 25
m
g sample from each tissue
source was analysed by reducing 12% SDS–PAGE followed by immunoblotting. Another
gel was run in parallel and was analysed by Coomassie blue staining to ensure equivalent
protein loading. Immunohistochemical analysis of Derlin-1 and calnexin localization in
U373 cells was performed with established methods
21
. Anti-Derlin-1 serum that had been
depleted of specific antibodies by affinity purification was used as a control.
Received 8 March; accepted 21 April 2004; doi:10.1038/nature02592.
1. Ellgaard, L. & Helenius, A. Quality control in the endoplasmic reticulum. Nature Rev. Mol. Cell Biol. 4,
181–191 (2003).
2. Kostova, Z. & Wolf, D. H. For whom the bell tolls: protein quality control of the endoplasmic
reticulum and the ubiquitin–proteasome connection. EMBO J. 22, 2309–2317 (2003).
3. Wiertz, E. J. et al. Sec61-mediated transfer of a membrane protein from the endoplasmic reticulum to
the proteasome for destruction. Nature 384, 432–438 (1996).
4. Bebok, Z., Mazzochi, C., King, S. A., Hong, J. S. & Sorscher, E. J. The mechanism underlying cystic
fibrosis transmembrane conductance regulator transport from the endoplasmic reticulum to the
proteasome includes Sec61beta and a cytosolic, deglycosylated intermediary. J. Biol. Chem. 273,
29873–29878 (1998).
5. de Virgilio, M., Weninger, H. & Ivessa, N. E. Ubiquitination is required for the retro-translocation of a
short-lived luminal endoplasmic reticulum glycoprotein to the cytosol for degradation by the
proteasome. J. Biol. Chem. 273, 9734–9743 (1998).
6. Plemper, R. K., Bohmler, S., Bordallo, J., Sommer, T. & Wolf, D. H. Mutant analysis links the
translocon and BiP to retrograde protein transport for ER degradation. Nature 388, 891–895 (1997).
7. Plemper, R. K. et al. Genetic interactions of Hrd3p and Der3p/Hrd1p with Sec61p suggest a retro-
translocation complex mediating protein transport for ER degradation. J. Cell Sci. 112, 4123–4134
(1999).
8. Sommer, T. & Jentsch, S. A protein translocation defect linked to ubiquitin conjugation at the
endoplasmic reticulum. Nature 365, 176–179 (1993).
9. Hiller, M. M., Finger, A., Schweiger, M. & Wolf, D. H. ER degradation of a misfolded luminal protein
by the cytosolic ubiquitin–proteasome pathway. Science 273, 1725–1728 (1996).
10. Bays, N. W., Gardner, R. G., Seelig, L. P., Joazeiro, C. A. & Hampton, R. Y. Hrd1p/Der3p is a
membrane-anchored ubiquitin ligase required for ER-associated degradation. Nature Cell Biol. 3,
24–29 (2001).
11. Ye, Y., Meyer, H. H. & Rapoport, T. A. The AAA ATPase Cdc48/p97 and its partners transport proteins
from the ER into the cytosol. Nature 414, 652–656 (2001).
12. Hitchcock, A. L. et al. The conserved npl4 protein complex mediates proteasome-dependent
membrane-bound transcription factor activation. Mol. Biol. Cell 12, 3226–3241 (2001).
13. Jarosch, E. et al. Protein dislocation from the ER requires polyubiquitination and the AAA-ATPase
Cdc48. Nature Cell Biol. 4, 134–139 (2002).
14. Bays, N. W., Wilhovsky, S. K., Goradia, A., Hodgkiss-Harlow, K. & Hampton, R. Y. HRD4/NPL4 is
required for the proteasomal processing of ubiquitinated ER proteins. Mol. Biol. Cell 12, 4114–4128
(2001).
15. Hampton, R. Y., Gardner, R. G. & Rine, J. Role of 26S proteasome and HRD genes in the degradation
of 3-hydroxy-3-methylglutaryl-CoA reductase, an integral endoplasmic reticulum membrane
protein. Mol. Biol. Cell 7, 2029–2044 (1996).
16. Tortorella, D., Gewurz, B. E., Furman, M. H., Schust, D. J. & Ploegh, H. L. Viral subversion of the
immune system. Annu. Rev. Immunol. 18, 861–926 (2000).
17. Wiertz, E. J. et al. The human cytomegalovirus US11 gene product dislocates MHC class I heavy chains
from the endoplasmic reticulum to the cytosol. Cell 84, 769–779 (1996).
18. Shamu, C. E., Flierman, D., Ploegh, H. L., Rapoport, T. A. & Chau, V. Polyubiquitination is required
for US11-dependent movement of MHC class I heavy chain from endoplasmic reticulum into cytosol.
Mol. Biol. Cell 12, 2546–2555 (2001).
19. Shamu, C. E., Story, C. M., Rapoport, T. A. & Ploegh, H. L. The pathway of US11-dependent
degradation of MHC class I heavy chains involves a ubiquitin-conjugated intermediate. J. Cell Biol.
147, 45–58 (1999).
20. Blom, D., Hirsch, C., Stern, P., Tortorella, D. & Ploegh, H. L. A glycosylated type I membrane protein
becomes cytosolic when peptide:N-glycanase is compromised. EMBO J. 23, 650–658 (2004).
21. Lilley, B. N., Tortorella, D. & Ploegh, H. L. Dislocation of a type I membrane protein requires
interactions between membrane-spanning segments within the lipid bilayer. Mol. Biol. Cell 14,
3690–3698 (2003).
22. Knop, M., Finger, A., Braun, T., Hellmuth, K. & Der Wolf, D. H. Der1, a novel protein specifically
required for endoplasmic reticulum degradation in yeast. EMBO J. 15, 753–763 (1996).
23. Ye, Y. et al A membrane protein complex mediates retro-translocation from the ER lumen into the
cytosol. Nature (this issue).
24. Taxis, C. et al. Use of modular substrates demonstrates mechanistic diversity and reveals differences in
chaperone requirement of ERAD. J. Biol. Chem. 278, 35903–35913 (2003).
25. Huh, W. K. et al. Global analysis of protein localization in budding yeast. Nature 425, 686–691 (2003).
26. Travers, K. J. et al. Functional and genomic analyses reveal an essential coordination between the
unfolded protein response and ER-associated degradation. Cell 101, 249–258 (2000).
27. Bateman, A. et al. The Pfam protein families database. Nucleic Acids Res. 32 (Database issue),
D138–D141 (2004).
28. Ying, H., Yu, Y. & Xu, Y. Cloning and characterization of F-LANa, upregulated in human liver cancer.
Biochem. Biophys. Res. Commun. 286, 394–400 (2001).
29. Tsukahara, M., Ji, Z. S., Noguchi, S. & Tsunoo, H. A novel putative transmembrane protein, IZP6, is
expressed in neural cells during embryogenesis. Dev. Growth Differ. 43, 285–293 (2001).
30. Story, C. M., Furman, M. H. & Ploegh, H. L. The cytosolic tail of class I MHC heavy chain is required
for its dislocation by the human cytomegalovirus US2 and US11 gene products. Proc. Natl Acad. Sci.
USA 96, 8516–8521 (1999).
31. Snapp, E. L. et al. Formation of stacked ER cisternae by low affinity protein interactions. J. Cell Biol.
163, 257–269 (2003).
32. Furman, M. H., Ploegh, H. L. & Tortorella, D. Membrane-specific, host-derived factors are required
for US2- and US11-mediated degradation of major histocompatibility complex class I molecules.
J. Biol. Chem. 277, 3258–3267 (2002).
33. Gewurz, B. E., Ploegh, H. L. & Tortorella, D. US2, a human cytomegalovirus-encoded type I
membrane protein, contains a non-cleavable amino-terminal signal peptide. J. Biol. Chem. 277,
11306–11313 (2002).
34. Vashist, S. et al. Distinct retrieval and retention mechanisms are required for the quality control of
endoplasmic reticulum protein folding. J. Cell Biol. 155, 355–368 (2001).
35. Plemper, R. K., Egner, R., Kuchler, K. & Wolf, D. H. Endoplasmic reticulum degradation of a mutated
ATP-binding cassette transporter Pdr5 proceeds in a concerted action of Sec61 and the proteasome.
J. Biol. Chem. 273, 32848–32856 (1998).
36. Walter, J., Urban, J., Volkwein, C. & Sommer, T. Sec61p-independent degradation of the tail-anchored
ER membrane protein Ubc6p. EMBO J. 20, 3124–3131 (2001).
37. Hill, K. & Cooper, A. A. Degradation of unassembled Vph1p reveals novel aspects of the yeast ER
quality control system. EMBO J. 19, 550–561 (2000).
38. Vashist, S. & Ng, D. T. W. Misfolded proteins are sorted by a sequential checkpoint mechanism of ER
quality control. J. Cell Biol. 165, 41–52 (2004).
39. Van den Berg, B. et al. X-ray structure of a protein-conducting channel. Nature 427, 36–44 (2004).
40. Fiebiger, E., Story, C., Ploegh, H. L. & Tortorella, D. Visualization of the ER-to-cytosol dislocation
reaction of a type I membrane protein. EMBO J. 21, 1041–1053 (2002).
41. Tirosh, B., Furman, M. H., Tortorella, D. & Ploegh, H. L. Protein unfolding is not a prerequisite for
endoplasmic reticulum-to-cytosol dislocation. J. Biol. Chem. 278, 6664–6672 (2003).
42. Rehm, A., Stern, P., Ploegh, H. L. & Tortorella, D. Signal peptide cleavage of a type I membrane
protein, HCMV US11, is dependent on its membrane anchor. EMBO J. 20, 1573–1582 (2001).
43. Tortorella, D. et al. Dislocation of type I membrane proteins from the ER to the cytosol is sensitive to
changes in redox potential. J. Cell Biol. 142, 365–376 (1998).
44. Sawin, K. E., Mitchison, T. J. & Wordeman, L. G. Evidence for kinesin-related proteins in the mitotic
apparatus using peptide antibodies. J. Cell Sci. 101, 303–313 (1992).
45. Kinter, M. & Sherman, N. E. Protein Sequencing and Identification using Tandem Mass Spectrometry
(Wiley, New York, 2000).
46. Borodovsky, A. et al. Chemistry-based functional proteomics reveals novel members of the
deubiquitinating enzyme family. Chem. Biol. 9, 1149–1159 (2002).
47. Ploegh, H. L. in Current Protocols in Protein Science (eds Coligan, J. E., Dunn, B. M., Ploegh, H. L.,
Speicher, D. W. & Wingfield, P. T.) 10.2.1–10.2.8 (Wiley, New York, 1995).
48. Green, N., Fang, H., Kalies, K.-U. & Canfield, V. in Current Protocols in Cell Biology (eds Bonifacino,
J. S., Dasso, M., Harford, J. B., Lippincott-Schwartz, J. & Yamada, K. M.) 5.2.1–5.2.27 (Wiley, New
York, 1998).
Supplementary Information accompanies the paper on www.nature.com/nature.
Acknowledgements We thank B. Kessler and E. Spooner for the preparation of samples and
assistance in analysis of mass spectrometry data; R. Tirabassi and K. Ryan for the production of
anti-GFP and anti-GRP94 antisera; D. Tortorella for cell lines; and members of the Ploegh
laboratory for helpful comments on the manuscript. B.N.L. is a Howard Hughes Medical Institute
Predoctoral Fellow. This work was supported by the National Institutes of Health.
Competing interests statement The authors declare that they have no competing financial
interests.
Correspondence and requests for materials should be addressed to H.L.P.
(ploegh@hms.harvard.edu).
articles
NATURE | VOL 429 | 24 JUNE 2004 | www.nature.com/nature840
© 2004
Nature
Publishing
Group
