Molecular Cell, Vol. 6, 1219–1232, November, 2000, Copyright 2000 by Cell Press
The Structure of Ribosome-Channel Complexes
Engaged in Protein Translocation
complex, which consists of an ? subunit with ten trans-
membrane (TM) domains as well as smaller ? and ?
subunits, each with a single TM domain (reviewed in
Matlack et al., 1998). The Sec61p complex is conserved
among all eukaryotes and has a bacterial homolog, the
SecYEG complex. Initial evidence that the Sec61p com-
plex forms a channel came from crosslinking experi-
ments that demonstrated its proximity to translocating
polypeptide chains (Mothes et al., 1994). Subsequently,
and prokaryotic complexes form similar ring-like struc-
tures both in detergent solution and in membranes (Ha-
nein et al., 1996; Beckmann et al., 1997; Meyer et al.,
1999). The size of these rings suggests that they are
formed from multiple copies of the Sec61p or SecYEG
Although reconstitution experiments demonstrated
that the Sec61p complex is sufficient to form a mem-
brane channel (Go ¨rlich and Rapoport, 1993), other com-
ponents may be associated with the complex. Candi-
dates include the translocon-associated protein (TRAP)
complex and the oligosaccharyl transferase (OST), both
of which are tightly associated with native ribosome-
channel complexes (Go ¨rlich et al., 1992). In the mamma-
lian ER membrane, TRAP is abundant and consists of
four subunits (Hartmann et al., 1993). However, its func-
tion is not known. The OST is a ubiquitous multisubunit
complex whose primary role may be the glycosylation
of nascent proteins (Silberstein and Gilmore, 1996).
The junction between the ribosome and the channel
must play a critical role in the cotranslational transport
of proteins across the ER membrane. According to the
prevalent view, a continuous seal around the junction
into the membrane channel. The existence of a tight
by fluorescence quenching experiments (Crowley et al.,
1994; Hamman et al., 1997). On the other hand, a three-
nontranslating yeast ribosomes and the purified yeast
Sec61p complex indicated only one connection be-
tween the ribosome and the Sec61p complex, with a
significant gap between them (Beckmann et al., 1997).
However, the reconstitution of ribosome-Sec61p com-
plexes in detergent and, in particular, the absence of a
nascent chain make it difficult to assess the physiologi-
cal significance of this observation.
is known to be dependent on the nascent polypeptide
chain. Nontranslating ribosomes or ribosomes with
short nascent chains bind weakly to the Sec61p com-
plex and can be removed by high salt concentrations
(Jungnickel and Rapoport, 1995). Ribosomes carrying
long, translocating chains bind in a salt-resistant man-
ner. Also, the ribosome-channel junction does not pro-
tect short nascent chains against proteolysis but does
shield longer ones. The transition to a strong interaction
occurs at a critical chain length (approximately 70 resi-
dues forpreprolactin) and involves signalsequence rec-
ognition by the Sec61p complex and opening of the
Jean-Franc ¸ois Me ´ne ´tret,*# Andrea Neuhof,†#
David Gene Morgan,*†# Kathrin Plath,†
Michael Radermacher,‡Tom A. Rapoport,†
and Christopher W. Akey*§
*Department of Physiology and Biophysics
Boston University School of Medicine
Boston, Massachusetts 02118
†Howard Hughes Medical Institute and
Department of Cell Biology
Harvard Medical School
Boston, Massachusetts 02115
‡Max Planck Institut fur Biophysik
Cotranslational translocation of proteins requires ri-
bosome binding to the Sec61p channel at the endo-
tron cryomicroscopy to determine the structures of
ribosome-channel complexes in the absence or pres-
the structures are similarand contain 3–4 connections
between the ribosome and channel that leave a lateral
opening into the cytosol. Therefore, the ribosome-
channel junction may allow the direct transfer of poly-
peptides into the channel and provide a path for the
egress of some nascent chains into the cytosol. More-
over, complexes solubilized from mammalian ER
membranes contain an additional membrane protein
thathas alarge,lumenalprotrusion andisintercalated
into the wall of the Sec61p channel. Thus, the native
channel contains a component that is not essential
Many secretory and membrane proteins are cotransla-
tionally transported across or integrated into the endo-
plasmic reticulum (ER) membrane. Secretory proteins
pass through the membrane completely, while mem-
in the cytoplasm (reviewed in Hedge and Lingappa,
1997; Matlack et al., 1998). Both types of proteins are
synthesized on membrane bound ribosomes and are
ER membrane. How the translating ribosome binds to
the channel is unknown.
A channel in the ER membrane was detected by con-
ductance and fluorescence quenching experiments (Si-
mon and Blobel, 1991; Crowley et al., 1994). The major
component of the channel is the heterotrimeric Sec61p
§To whom correspondence should be addressed (e-mail: akey@
#These authors contributed equally to this paper.
1995; Crowley et al., 1994). It is not clear whether the
nascent chain causes a conformational change that in-
creases the strength of the ribosome-channel interac-
junction would permit polypeptides, such as secretory
proteins, to pass from the ribosome tunnel through the
channel and into the ER lumen, it would present a prob-
lem in certain cases. For example, the transport of na-
scent membrane proteins must stop when a TM domain
tide segment to form a cytosolic domain. In this case
the continuous connection between the ribosome and
the channel would have to be broken to allow this seg-
ment to emerge into the cytosol.
To better understand the behavior of the ribosome-
channel junction, we have determined the 3D structures
of ribosome-channel complexes, both in the absence
and presence of translocating nascent chains.
20 A˚) is present between them (Figure 2A). We conclude
that the structural features of ribosome-channel com-
plexes are essentially the same prior to solubilization,
whether the ribosomes are directly bound to channels
in detergent or bound to channels in membranes.
Next, we determined the structure of ribosome-chan-
nel complexes in which the ribosome carries a translo-
cating nascent chain. We used rabbit reticulocyte ribo-
somes carrying the first 86 amino acids of preprolactin
(86mer) as previous results have demonstrated that the
86mer induces a salt-resistant interaction of the ribo-
some with the channel and is fully inserted into the
Sec61p channel (Jungnickel and Rapoport, 1995).
We first tested whether complexes of translating
ribosomes and the Sec61p channel could be isolated.
The 86mer was synthesized in a reticulocyte lysate in
the presence of35S-methionine. Proteoliposomes were
added that contained purified canine Sec61p complex
and the SRP receptor. The latter is required for mem-
brane targeting of ribosome–nascent chain complexes.
After translation, the vesicles were floated in a sucrose
gradient at low (100 mM) or high (500 mM) salt concen-
tration. In both cases, significant amounts of the 86mer
were found in the floated fractions (Figure 1B, lanes 3–6)
and remained bound to ribosomes as peptidyl tRNA, as
demonstrated by precipitation with cetyltrimethylam-
monium bromide (CTABr; Figure 1B, lanes 4 versus 3
and 6 versus 5). Very little 86mer was found in floated
vesicles that lacked the Sec61p complex and the SRP
receptor (Figure 1C). When floated complexes con-
taining the 86mer were solubilized in digitonin at high
salt concentration and subjected to sedimentation, a
portion of the 86mer was recovered in the pellet fraction
(Figure 1B, lanes 7–10), and most of the material re-
mained CTABr-precipitable (lanes 8 versus 7 and 10
versus 9).The complex ofribosome, 86mer,and Sec61p
was very stable, even when it was generated with native
membranes instead of proteoliposomes (Figure 1D).
Most of the 86mer remained CTABr-precipitable for at
least 5 hr (lanes 14 versus 13), and these results indi-
cated that the nascent chain stayed associated with
the ribosome as peptidyl tRNA. In addition, the nascent
chain remained protease resistant, and this was indica-
tive of its location inside the ribosome-channel complex
(lanes 15 versus 13). Together, these data demonstrate
that ribosome-channel complexes can be produced
with a defined translocating nascent chain. Samples
for electron cryomicroscopy were prepared similarly,
except that the 86mer was synthesized in the absence
some-Sec61p complex, we added proteoliposomes to
a nonprogrammed reticulocyte lysate. We followed this
with flotation at low salt concentration and solubiliza-
tion. Immunoblotting with an antibody to the ribosomal
protein S26 showed that ribosomes were present in the
final pellet fraction (Figure 1E, lane 2). No ribosomes
were seen in the pellet if the flotation was performed at
high salt concentration (lane 3), indicating that non-
translating ribosomes were effectively removed. It should
be noted that solubilization of the complexes of non-
translating ribosomes and channels was carried out at
the interaction becomes salt resistant (Go ¨rlich et al.,
Ribosomes Associated with Purified
In a previous study, electron cryomicroscopy was used
to determine the 3D structure of detergent-solubilized
yeast Sec61p complex bound to yeast ribosomes lack-
the structural analysis with complexes formed by bind-
ing ribosomes to the Sec61p complex in membranes,
which created a more natural interaction. Isolated yeast
ribosomes (Figure 1A, lane 1) were mixed with proteoli-
posomes containing the purified yeast Sec61p complex
ing the vesicles in a sucrose gradient. Ribosomes were
the Sec61p complex (lane 8 versus lane 5). The floated
vesicles were then solubilized in digitonin, and the ribo-
plex (lane 9). In the absence of added ribosomes, no
Sec61p complex was found in the pellet (not shown).
After resuspending the pellet, the ribosome-channel
cryomicroscopy. We used a yeast ribosome map (Mor-
gan et al., 2000) as a first reference to determine the 3D
structure of yeast ribosome-channel complexes. As a
starting point, a threshold level was chosen for each
structure so that the ribosome volume calculated from
the density above the threshold would be consistent
with its calculated volume (100% level).
Overall, the yeast ribosome-channel complex at ap-
proximately 25 A˚resolution is similar to that reported
by Beckmann et al. (1997) (Figure 2A shows a frontal
view, and Figure 2B shows a view from the ER lumen;
see Table 1 for statistics). The channel has an outer
diameter of approximately 91 A˚, and the nascent-chain
exit site on the large ribosomal subunit is aligned with
the pore (Figure 2C; the tunnel exit site is encircled by
the gold channel outline). At this threshold, there is no
visible connection between the ribosome and the
Sec61p channel; hence, a sizable gap (approximately
Three-Dimensional Structure of the Ribosome-Channel Complex
Figure 1. Isolation
(A) Yeast ribosomes were mixed with proteo-
liposomes containing purified yeast Sec61p
complex (Sec61p) or with liposomes lacking
protein. The membranes were floated in a su-
crose gradient and solubilized in digitonin,
and the ribosomes were sedimented. Equiva-
lent aliquots of the individual components, of
the original mixtures (T, for total), and of the
floated (F) and pellet (P) fractions were sub-
with Coomassie blue or by immunoblotting
cate the positions of the three subunits of the
(B) Radiolabeled 86mer ofpreprolactin(pPL86)
was synthesized in a reticulocyte lysate in
the presence of proteoliposomes containing
canine Sec61p complex and SRP receptor.
The membranes were floated in a sucrose
gradient at either 100 or 500 mM salt and
solubilized in digitonin at the appropriate salt
concentration, and the ribosomes were sedi-
mented. Equivalent aliquots of the original
and F500) or after pelleting of the ribosomes
(P100and P500) were analyzed by SDS-PAGE
followed by autoradiography. Each of the
with CTABr (CTABr-ppt). The percentage of
CTABr-precipitable material is given below
(C) As in (B), except that liposomes without
protein were used.
(D) As in (B), except that ribosome-stripped
native membranes (PK-RM) were used and
flotation andsolubilization wereperformed at
500 mM salt. All samples were analyzed by
proteinase K (ProtK). The final pellet fraction
was also incubated on ice for either 1 or 5 hr
to test the stability of the ribosome-channel
in F and P samples than in T samples.
(E) As in (B), except that a nonprogrammed
translation mix was used and the ribosomes
in the original sample (T; 10% loaded) and in
the final pellet fraction (P) were detected by
immunoblotting with an antibody to the ribo-
somal protein S26.
1992), in a manner similar to that observed with translat-
ing ribosomes on intact membranes.
then used to determine the 3D structures of mammalian
ribosome-channel complexes both without and with the
preprolactin 86mer (Figures 2D–2F and 2G–2I; see Ta-
ble 1). The structures were similar to those from yeast,
except that the channel had a diameter of ?100 A˚(com-
pare Figures 2A and 2B with 2D and 2E). Surprisingly,
there was little difference in the structures with and
without a nascent polypeptide chain (compare Figures
2D and 2E with 2G and 2H). At the 100% threshold
level, there was again no visible connection between
a sizablegap (approximately20 A˚)was presentbetween
them (Figures 2D and 2G). When the threshold was low-
ered to enclose approximately 150% of the calculated
ribosome volume, several connections were observed
and the gap became narrower (Figures 2F and 2I). One
of the connections appears to be in a position similar
to that observed for the yeast complex (Beckmann et
Ribosome-Sec61p Complexes Derived
from Native Membranes
To test whether the structure of the ribosome-channel
complex is altered in the presence of other ER com-
ponents, we repeated the analysis with complexes
produced from intact ER membranes. Ribosomes with
and without the preprolactin 86mer were bound to ca-
nine microsomes stripped of ribosomes by a previous
treatment with puromycin and high salt (PK-RM). Ribo-
some-channel complexes were prepared for electron
microscopy, and their 3D structures were determined.
Figure 2. Ribosomes Associated with Purified Sec61p Complexes
(A) The complex of yeast ribosome (green) and yeast channel (gold) lacking a nascent chain is viewed along the plane of the ER membrane
(frontal view). The threshold level was chosen to encompass 100% of the ribosomal volume. The small (S) and large (L) ribosomal subunits
(B) The yeast complex is viewed from the ER lumen (bottom view). This view is generated by a 90? rotation about the horizontal axis, followed
by a 90? rotation about the vertical axis in the plane.
(C) A similar view as in (B), with the circumference of the yeast channel outlined in gold to reveal the nascent chain tunnel exit (TE).
(D) The complex of rabbit ribosome (purple) and canine channel (gold), lacking a nascent chain, is shown in frontal view.
(E) The mammalian complex lacking a nascent chain shown in bottom view.
(F) The mammalian complex, lacking a nascent chain, is shown in frontal view with a threshold that encompasses approximately 150% of the
expected ribosomal volume.
(G) The complex of rabbit ribosome (purple) and canine channel (red), with the 86mer of preprolactin present, is shown in frontal view.
(H) The mammalian complex with a nascent chain shown in bottom view.
(I) The mammalian complex with a nascent chain is shown in frontal view with a threshold that encompasses approximately 150% of the
expected ribosomal volume.
Connections (c) between the channel and large subunit are indicated by arrows. Scale bar ? 100 A˚. The color code of the ribosomes and of
the channels plus and minus nascent chain (NC) is shown as a vertical bar on the lower right.
The resulting structures demonstrate features that are
similar to those seen previously (Figures 3A–3F). At a
tions were observed between the ribosome and the
channel (Figures 3C and 3F) at approximately the same
positions (Figures 2F and 2I and Beckmann et al., 1997).
However, the channels were noticeably larger and pos-
sessed an additional domainemerging from the lumenal
side (to be discussed later). Although this feature was
less pronounced when the 86mer was present, overall
We wished to exclude the possibility that our results
were biased by the choice of the preprolactin nascent
chain or by the de novo assembly of ribosome-channel
complexes in vitro. We therefore determined the 3D
structure of native ribosome-channel complexes as-
sembled in vivo in which the ribosomes carry a mixed
population of endogenous nascent chains. Rough ca-
nine microsomes were washed with high salt to remove
all ribosomes that do not carry translocating nascent
Three-Dimensional Structure of the Ribosome-Channel Complex
Table 1. Summary of Ribosome-ER Channel 3D Data Sets
RibosomesChannels MembranesNascent Chains ParticlesResolutiona(A˚)
aResolution was determined with the FSC0.5criterion. For each data set, pairs of 3D volumes were calculated with increasing numbers of
particles up to the total number divided by two. The appropriate volumes were then compared, and their resolutions were plotted as a function
of increasing particle number. This allowed us to estimate the resolution of the final complete 3D data sets by extrapolation.
The “nd” stands for “not determined.”
chains. The membranes (K-RMs) were then solubilized
Sec61p complexes were pelleted and analyzed by elec-
tron cryomicroscopy. Again, this 3D structure showed
all the features seen previously (Figures 3G–I). As ex-
pected, thestructure of thecanine ribosomewas similar
to that of the rabbit ribosome, and the exit site of the
nascent chain was aligned with the pore of the mem-
brane channel (Figure 3I). Connections between the ri-
bosome and the channel were only visible at lower
threshold levels (compare the 100% level in Figure 3G
In conclusion, the ribosome-channel junction does
not change significantly, even when ribosomes carry a
heterogeneous mixture of nascent chains at different
stages of translocation. It should be noted that com-
plexes containing a nascent chain did not show a sig-
nificant density for peptidyl tRNA in the resulting
structures, even though CTABr precipitation, protease-
protection, and high-salt resistance all indicated the
presence of a nascent chain. We believe that sample
preparation and the nonphysiological mechanism used
ble binding sites on the ribosome (Agrawal et al., 2000).
Hence, the tRNA density would be lost in the averaged
on the chosen threshold level and must be a genuine
feature of the ribosome-channel junction.
whose density was similar to that outside the ribosome,
at a confidence level greater than 99% (blue contour in
Figure 4D). These results illustrate that the gap between
conducting tunnel in the ribosome. A similar analysis
performed with all other structures confirmed that the
gap is a general feature of the ribosome-channel
Interaction sites between the ribosome and the chan-
mined structures, to the point where individual connec-
tions could be seen. Bottom views of the ribosomes
were then superimposed, and connections to the chan-
as gray regions within the native channel, and the num-
bers indicate how many structures contributed to each.
Three connections were located on one side of the tun-
nel exit in nine out of ten maps and are thus very likely
real features. A fourth connection was present in only
four maps. Together, they appear to form a horseshoe-
like collar that provides a major lateral opening toward
the cytosol (see arrow in Figure 4E). This opening is
oriented approximately in the direction of the interface
between the small and large ribosomal subunits. In Fig-
ure 4F the connections were overlayed onto a map of
the native channel complex, which is contoured at a
higher threshold level to indicate dense features that
may correspond to ribosomal RNA. Interestingly, each
of the connections to the channel was located over a
high-density feature (Figure 4F). Together with a recent
cient to mediate interactions with the membrane chan-
nel (Prinz et al., 2000), these results suggest that several
distinct rRNA regions may be involved.
To exclude the possibility that solubilization in deter-
gent artificially generated a gap, we visualized ribo-
somes bound to intact membranes. Yeast ribosomes
were incubated with proteoliposomes containing puri-
fied yeast Sec61p complex, and the vesicles were puri-
The Ribosome-Channel Junction
plexes containing endogenous nascent chains derived
from native membranes. When the threshold was low-
ered to include 200% of the ribosome volume, the back-
4B), and the pore was filled in (white contour in Figure
4C). Although this threshold level is clearly too low, a
gap in the ribosome-channel junction is still visible as
a pronounced lateral passage that leads from the ribo-
somal exit site to the cytosol (Figure 4C, dotted line).
Therefore, we conclude that the gap is not dependent
Figure 3. Ribosomes Associated with the Sec61p Complex Derived from Native Membranes
(A) The complex of nontranslating rabbit ribosomes (purple) and canine channels (gold), which was derived from ribosome-stripped membranes
(PK-RM), is viewed along the plane of the ER membrane (frontal view). The threshold level was chosen to encompass 100% of the ribosomal
volume. The small (S) and large (L) ribosomal subunits are indicated.
(B) The mammalian complex, derived from PK-RM, without a nascent chain is viewed from the ER lumen (bottom view). The threshold is at
100% ribosome volume.
(C) As in (A), with a threshold that encompasses approximately 150% of the expected ribosomal volume. Connections (c) between the channel
and large subunit are indicated by arrows.
(D) The complex of rabbit ribosome (purple) and canine channel derived from PK-RM (red), with the 86mer of preprolactin present, is shown
in frontal view.
(E) The mammalian complex, derived from PK-RM, with a nascent chain present is shown in bottom view. The threshold is at 100% ribosome
(F) As in (D), with a threshold that encompasses approximately 150% of the expected ribosomal volume.
(G) The complex of canine ribosomes (blue) and canine channels (red), derived from native high salt–washed ER membranes (K-RM), with
endogenous nascent chains present, is shown in a frontal view.
(H) The mammalian complex with endogenous nascent chains is shown in bottom view.
(I) A similar view as in (H), with the circumference of the canine channel outlined in red to reveal the nascent-chain tunnel exit. The smaller
hole is only a dimple. Scale bar ? 100 A˚. The color code of the ribosomes and of the channels plus and minus nascent chain (NC) is shown
as a vertical bar on the lower right.
fied by flotation. We collected images in which the ribo-
somes had the correct orientation relative to the plane
of the membrane (Figure 5A and see Experimental Pro-
cedures). A 3D structure was calculated at approxi-
edge demonstrates that there is a pronounced gap be-
tween the ribosome and the membrane (Figure 5A, left).
brane is visible, shows that the gap is similar to that
observed in solubilized complexes (Figures 5C versus
5E). This was verified by measuring the distance be-
unit and the middle of the membrane or channel (arrows
in Figures 5C and 5E). Thus, the ribosome-channel junc-
tion does not change during solubilization. Although the
channel is difficult to see because its density is similar
to that of the membrane, a dense feature is observed
at the expected position when the threshold level is
raised (Figure 5D).
A similar analysis was performed with native mem-
Three-Dimensional Structure of the Ribosome-Channel Complex
Figure 4. A Gap between the Ribosome and
(A) 3D structure of canine ribosome-channel
complexes derived from native membranes.
ent. Shown is a frontal view at a threshold
level corresponding to 150% of the ribosome
volume. Arrows point to connections. Grey
features indicate background noise. Scale
bar ? 100 A˚.
(B) Similar view as in (A), except that the
ume of approximately 200%.
(C) A thinsection is shown fromthe map, with
the channel oriented at the bottom and the
approximate position of the membrane (M)
indicated. The map is contoured at three
thresholds, corresponding to 100% (green),
150% (yellow), and approximately 200%
(white dots) form a continuous path from the
peptidyltransferase center (PTC) to the cyto-
sol with a diameter of approximately 20 A˚,
even at a 200% threshold. Maps were dis-
played in “O” (Jones et al., 1991).
(D) A view similar to that in (C), with two
thresholds for the ribosome-channel com-
plex, 100% (green) and approximately 200%
tical map (in blue), which defines regions that
are lacking positive electron density (?99%
confidence interval) and may correspond to
solvent. The tunnel is labeled “T.”
(E) The bottom views of ribosomes from ten
structures, including the first 6 listed in Table
1 and four other unpublished maps, were su-
perimposed, and connections to the channel
were identified. These connections are shown
as gray regions within the outline of a semi-
cate how many structures contributed to a
connection). The tunnel exit (TE) is indicated,
chain is marked with an arrow.
(F) A view similar to that in (E), with the con-
nections overlayed in red onto high-density
features (blue mesh) that surround the tunnel
exit. The lines indicate the boundaries be-
tween the small (S) and large (L) ribosomal
brane bound ribosomes in salt-washed canine micro-
somes. Because of technical problems, we were able
to analyze only approximately 200 particles to generate
a low-resolution 3D structure. A projection of this struc-
ture (Figure 5B) also demonstrated that the spacing of
the ribosome-membrane junction was similar to that
observed in the yeast ribosome Sec61p vesicles. There-
fore, we conclude that a gap is an intrinsic feature of
the native ribosome-channel junction.
the tunnel exit (Figure 4E). We estimate that the lumenal
son with the channel itself. In Figures 6A–6D, the chan-
nels from native membranes (K-RM) are compared with
those formed by the purified Sec61p complex and are
shown in isolation without the ribosome, either from
the lumen of the ER (Figures 6A and 6B) or along the
membrane plane (Figures 6C and 6D). The channels are
corresponding to a range that likely includes the true
channel volume. When compared with purified Sec61p
complexes (approximately 100 A˚ diameter), channels
mately 125 A˚in the longest dimension). The increased
channel size is consistent with previous results using
The channel formed by purified Sec61p complexes
Structure of the Translocation Channel
While the ribosome-channel junction is similar for struc-
tures derived from purified Sec61p complexes or from
native membranes, the channels themselves differ sig-
nificantly. Native channels contain a prominent lumenal
domain (compare Figures 2 and 3) that is precisely ori-
ented with respect to the ribosome and points toward
Figure 5. Interaction of Ribosomes with Membranes
(A) Three images of yeast ribosomes associated with channels in vesicles are shown on the right (protein and lipid are white). A projection
map of the final 3D reconstruction is shown on the left. The small (S) and large (L) ribosomal subunits are labeled, and the membrane is
(B) Three images of canine ribosomes associated with channel complexes in native membranes (K-RM) are shown on the right. A projection
map of the final 3D reconstruction is shown on the left.
(C) A 3D surface for the yeast ribosome-vesicle map is shown in the frontal view. A gap is seen between the ribosome (green) and the
membrane (gold). The double arrow gives the distance between a recognizable feature on the ribosome to the center of the membrane.
(D) A bottom view of the yeast vesicle 3D map is shown at a higher threshold to reveal significant channel density over the nascent-chain
(E) For comparison, a frontal view of the solubilized yeast ribosome-channel structure is shown. Note that the length of the double arrow is
similar to that in (C). Scale bar ? 100 A˚.
has a cup shape with a large vestibule located toward
the ribosome (approximately 50 A˚diameter) that tapers
to an approximately 20 A˚pore on the lumenal side (Fig-
ure 6E). The complex from which the channel was taken
did not have a nascent chain; however, similar channel
structures were obtained with the chain present (see
Figures 2D–2I). The channel derived from native mem-
branes looks similar to the purified channel when sec-
tioned along the short axis (Figure 6F, middle panel). In
a section cut approximately 45? away, the pore is both
larger and maintains a constant diameter (Figure 6F,
right panel). The central pore of the native channel ex-
tends from a point opposite the ribosomal tunnel exit
to the site where the lumenal protrusion emerges (the
old level). When purified channels are overlayed with
those derived from native membranes, their similarity in
the region over the ribosome tunnel exit is apparent
(Figures 6G and 6H, center panels), while major differ-
the channel (Figure 6H, central panel). Importantly,
large-scale differences were not detected between
channels that differ by the presence of a nascent chain.
Specifically, the pore size remained the same, unlike
what was seen in fluorescence quenching experiments
using native membranes (Hamman et al., 1997, 1998).
Native Channels Contain an Additional
We first tested whether soluble, lumenal proteins con-
tribute to the structural difference between native and
purified channels. Native membranes (K-RM) were ex-
tracted with digitonin at low salt concentrations to re-
move most lumenal proteins (Figure 7A, Sotot) and then
termined structure was identical to that obtained with
nonextracted membranes and showed the lumenal do-
main (compare Figures 3G and 7B). This result was con-
a crude detergent extract of ribosome-stripped micro-
somes. The vesicleswere incubated with nontranslating
ribosomesand subjectedtotheusual proceduretogen-
erate ribosome-channel complexes. The structure of
lar to those obtained from native membranes, although
they appear to have a lower channel occupancy as evi-
denced from the disconnected lumenal domain and
smaller channel size.
Three-Dimensional Structure of the Ribosome-Channel Complex
Figure 6. Structure of the Translocation Channel
(A) The purified canine Sec61p channel in the presence of the 86mer of preprolactin is viewed from the ER lumen at thresholds equivalent to
120%, 110%, and 100% of the associated ribosome volume.
(B) The canine channel derived from native membranes in the presence of a mixed population of nascent chains is viewed from the ER lumen
at a threshold series as in (A).
(C) As in (A), but viewed from the side along the membrane plane.
(D) As in (B), but viewed from the side along the membrane plane. The lumenal domain (LD) is indicated.
(E) The purified canine Sec61p channel in the absence of a nascent chain is shown in gold as a wiremesh surface viewed from the ribosome
(left panel). The center and right panels show slices through the channel. These slices correspond to a view along the shortest axis and a
view offset by approximately 45?, respectively.
(F) The canine channel in the presence of a mixed population of nascent chains is shown in orientations similar to those in (E).
(G) An overlap view (center panel) of the two maps from the left panels in (E) and (F) is shown. The native channel is semitransparent, and
the lumenal domain has been removed. The original maps are shown on either side.
(H) An overlap view (center panel) of the two maps from the right-most panels in (E) and (F) are shown. The native channel is semitransparent.
The original maps are shown on either side. Scale bar ? 100 A˚.
Next we determined which membrane proteins are
present in the complexes used for structural analysis
of native channels. The ribosome-channel pellet, which
would normally be analyzed in the microscope, was
resuspended and treated with puromycin and high salt
to dissociate the ribosomes and release associated
membrane proteins (Go ¨rlich et al., 1992). The ribosomal
subunits were sedimented, and the supernatant was
analyzed either directly by SDS-PAGE (Figure 7A, Sptot)
or after extraction with Triton X-114 to enrich for integral
membrane proteins (SpTX-114). As expected, this frac-
tion contained the Sec61p complex. In addition, it con-
tained the TRAP and OST complexes (indicated with
symbols in the figure). Immunoblot analysis demon-
strated that the majority of these proteins were re-
leased from the ribosomes by puromycin treatment (not
shown). A number of other proteins, including signal
were not present in the ribosome-channel complex.
Therefore, we conclude that either the TRAP or OST
complex is responsible for the differences between na-
tive and purified channels.
Two major conclusions can be deduced from our struc-
tural analysis. First, the junction between the ribosome
and the channel is not continuous, and a lateral opening
to the cytosol is always present. Second, channels de-
rived from native membranes differ significantly from
those formed by the purified Sec61p complex. They
contain either the TRAP or OST complex, which contrib-
ute to the formation of a lumenal protrusion.
A Gap between the Ribosome and the Channel
Beckmann et al. (1997) demonstrated the existence of
a sizable gap between the ribosomal large subunit and
the ER channel. This gap was believed to be due to the
lack of a nascent chain. We now provide evidence that
the ribosome-channel interaction does not become
more extensive when a translocating nascent chain is
present. We consider it unlikely that the gap between
the ribosome and the membrane channel is an artifact
for several reasons. First, it has now been seen in 15
3D structures by using different methods of complex
Figure 7. A Lumenal Protrusion in Native
Channels Formed Either by TRAP or OST
(A) Pancreatic microsomes washed with high
salt (K-RM) were extracted in digitonin with-
out salt. This resulted in a supernatant (tot;
10 eq loaded) and a pellet fraction. The pellet
was extracted first with a buffer containing
digitonin and 0.5 M salt, which resulted in a
supernatant (S0.5tot, 10 eq) and a pellet con-
taining ribosomes and associated membrane
proteins. To release the membrane proteins,
the pellet was treated with puromycin in dig-
itonin at 1.2 M salt. After centrifugation, a
supernatant (Sptot, 50 eq) and a pellet (Pp, 5
pmol ribosomes loaded) were obtained. To
enrich for membrane proteins, a portion of
all supernatant fractions was extracted with
Triton X-114, and the detergent phase was
analyzed (S0TX-114 (20 eq), S0.5TX-114 (20
eq), SpTX-114 (100 eq). All samples were sub-
with Coomassie blue. Lanes 9–12 show puri-
fied canine ribosomes, Sec61p, OST, and
TRAP complexes. Dots, triangles, and dia-
monds indicate subunits of the OST, Sec61p,
and TRAP complexes, respectively.
(B) Native membranes (K-RM) were extracted
with digitonin at low salt concentrations to re-
move most lumenal proteins. The membranes
were subjected to the usual procedure to
generate ribosome-channel complexes, and
their structure was determined.
(C) Ribosome-stripped microsomes (PK-RM)
were solubilized in deoxyBigCHAP at high
salt concentration, and the extract was
treated with hydrophobic beads to remove
the detergent. The resulting proteoliposomes
were incubated with nontranslating ribo-
somes and solubilized, and a structure was
formation, sourcesof material,and methodsof analysis.
Second, the presence of a gap is not dependent on
the chosen threshold level. Even at unrealistically low
thresholds, there is at least one lateral passage with a
diameter of approximately 20 A˚leading from the tunnel
and the channel appears likely, there is still the possibil-
ity that it contains some material that is not visible at
of ribosomal RNA or proteins or of the Sec61p complex
may be present. In addition, we cannot exclude confor-
mational changes caused by the nascent chain that are
below the resolution limit. As the three or four connec-
tions between the ribosome and the channel were seen
only at low thresholds, we suggest that they are small
or flexible. However, by acting together they may stably
tether the channel to the ribosome.
The existence of a gap between the ribosome and the
channel does not contradict previous data that demon-
strated that polypeptides passing into the lumen of the
ER are protected against externally added proteases
(Connolly et al., 1989); the gap is too narrow to allow
access of a protease to the translocating polypeptide
chain. The data are also consistent with electrophysio-
logical experiments showing that a nascent chain pre-
and Blobel, 1991) if one assumes that the block occurs
within the membrane channel. However, our results are
more difficult to reconcile with fluorescence quenching
data that show the existence of a seal for ions between
the ribosome and the channel. For example, fluorescent
probes in short preprolactin chains of 56 or 64 amino
acids could not be quenched by iodide ions added to
the cytosolic compartment (Crowley et al., 1994). Per-
haps the diffusion of iodide ions is inhibited by charges
on the ribosome and/or the channel surfaces lining the
pathway or by flexible material. It should also be noted
that the same short nascent chains can be cleaved by
protease (Jungnickel and Rapoport, 1995). This finding
supports the existence of a lateral opening between the
ribosome and channel.
If the ribosome-membrane junction does not form a
seal, then how is the permeability barrier maintained
between the ER lumen and the cytosol? One possibility
is that the nascent chain in the channel provides suffi-
cient hindrance to the flow of small molecules and that
the channel is disassembled when it is not in use. A
more provocative idea would be that the active channel
provides only a minimal hindrance to the flow of small
molecules. A leaky ER membrane channel may not
cause problems in maintaining gradients if pumps such
as the Ca2?ATPase are active enough.
Three-Dimensional Structure of the Ribosome-Channel Complex
Figure 8. Possible Pathways for the Nascent Chains of Secretory and Membrane Proteins
(A) A thin section was generated from the 3D structure of the ribosome-channel complex derived from native canine membranes. Internal
solid surfaces revealed by the front cutting plane are filled with blue. The channel is colored in red. The predicted position of the preprolactin
86mer that extends from the peptidyltransferase center (PTC) to the channel is indicated by a dotted line. The chain likely adopts a loop
structure and is representative of early stages of translocation. The small (S) and large (L) ribosomal subunits are labeled, and the boundary
between them is marked by a dashed line.
(B) The possible path for secretory nascent chains from the PTC to the ER lumen is indicated.
(C) The possible path of a nascent membrane protein is shown when a cytosolic domain is being synthesized following the integration of a
TM segment into the membrane (M).
A gap between the ribosome exit site and the mem-
brane channel raises the question of how the polypep-
tide chain is translocated. Early during translocation,
when a nascent secretory protein contains approxi-
mately 60 residues, the signal sequence will have
emerged from the ribosomal tunnel, but the chain would
be too short to insert into the Sec61p channel as a loop.
The flexible N termini of these chains may sometimes
extend through the gap into the cytosol, and this would
explain why they are accessible to proteolysis (Jung-
nickel and Rapoport, 1995). With nascent chains longer
than approximately 70 residues the signal sequence
would be inserted into the pore as a result of its affinity
for a specific binding site in the channel wall (Plath et
as a loop would guide the following part of the polypep-
tide chain into the channel pore so that no part would
be exposed to the cytosol. This would explain why such
chains are protease resistant. As the chain is elongated
during translation, polypeptide segments would pass
from the ribosomal tunnel, cross the gap, and emerge
through the pore in the Sec61p channel into the ER
lumen (Figure 8B). The polypeptide chain would nor-
mally pass in the forward direction during translation as
the gap may be narrow enough to prevent polypeptide
segments from looping out into the cytosol.
The existence of a gap simplifies models of how cyto-
solic segments of membrane proteins may escape
through the ribosome-channel junction. We propose
that the ribosome-channel junction remains unchanged
or undergoes only small conformational changes when
a hydrophobic membrane anchor stops the transfer of
the polypeptide chain through the channel. The mem-
brane anchor may then exit the channel laterally and
drag the nascent chain with it through the gap in the
ribosome-channel junction. It would thus allow the fol-
lowing segment to emerge into the cytosol (Figure 8C).
Even if the gap contained flexible material, it would not
be expected to totally block the egress of the nascent
emerge from the membrane bound ribosome close to
the cytosolic end of the Sec61p channel, and its hydro-
phobicity could thus be continuously probed. If suffi-
ciently hydrophobic, the segment would insert into the
channel as a loop and reinitiate translocation of a lume-
A similar model would apply to translocational paus-
ing during the synthesis of certain secretory proteins,
would stop polypeptide movement through the channel
and lead to the transient looping of a polypeptide seg-
ment into the cytosol. A lateral opening between the
polypeptides can be transported backward from the
ER lumen into the cytosol, despite the ribosome being
bound to the channel (Ooi and Weiss, 1992). Finally,
should ribosomes synthesizing cytosolic proteins bind
to the channel, the existence of a passageway into the
cytosol would allow them to continue translation.
The Native Protein Translocation Channel
Our results reveal that native channels contain an inte-
gral membrane protein that is distinct from the Sec61p
complex and that may correspond to either the TRAP
or OST complex. However, a structural comparison of
the native and purified channels suggests that the size
differencemayarise fromonlyoneof thetwomembrane
complexes. Neither of these proteins is essential for
Rapoport, 1993). Thus, our data indicate that nonessen-
tial membrane proteins are recruited to the native trans-
Interestingly, the lumenal protrusion formed by the
additional component is precisely oriented within the
ribosome-channel complex even though it does not
PC) in ethanol was added to a final concentration of 1 mol% of total
phospholipid to follow the fractionation of membranes under UV
light. The translation mixture was adjusted to a final volume of 150
posomes), both in 30 mM HEPES/KOH (pH 7.8) and 10 mM magne-
sium acetate buffer. For ribosomes with or without a nascent chain,
the buffer also contained 500 mM or 100 mM potassium acetate,
respectively. The samples were transferred to a 7 ? 20 mm polycar-
bonate tube (Beckman), overlayed with 30 ?l of the same buffer
without sucrose, and spun for 1 hr at 100,000 rpm at 2?C in a Beck-
man TLA100 rotor. A UV transilluminator was used to visualize the
floatedmembranes. Theywerediluted about1:3in buffercontaining
30 mM HEPES/KOH (pH 7.8); 10 mM magnesium acetate; 1.5%
digitonin (final concentration); and 100 mM potassium acetate for
proteoliposomes with ribosomes lacking nascent chains or 500 mM
potassium acetate for samples containing nascent chains and for
all PK-RM samples. After incubation for 15 min at 4?C, the samples
were centrifuged for 20 min at 100,000 rpm at 2?C in a TLA100 rotor.
The pellet was resuspended in 30 mM HEPES/KOH (pH 7.8), 1.5%
digitonin, 100 mM potassium acetate, and 10 mM magnesium ace-
tate. Aliquots were analyzed by SDS-PAGE and immunoblotting
against the ? and ? subunits of the Sec61p complex. CTABr precipi-
tation and treatment with proteinase K were done as described
(Mothes et al., 1997). All samples were kept at 4?C and frozen for
electron cryomicrosocopy within approximately 4 hr of preparation.
Preparation of yeast ribosomes with bound Sec61p was similar
to that of the mammalian complexes lacking nascent chains except
stituted proteoliposomes containing the yeast Sec61p complex
were isolated by flotation and directly analyzed by electron mi-
make direct contact with the ribosome. Because the
the additional component may become associated at
a specific site. The additional membrane protein also
causes drastic changes in the channel. This component
seems to be intercalated into the walls of the Sec61p
channel and to increase the size of the membrane-
embedded portion. In addition, it changes the shape
and size of the pore. In the purified channel the central
pore is cylindrical, with a diameter of about 20 A˚on its
lumenal side, whereas in the native channel the pore is
elliptical, with a size of approximately 20 ? 50 A˚. The
purified Sec61p channel has a cup-like cross section
with a rather narrow constriction toward the lumen. This
shape is similar to the funnel-like shape observed by
Beckmann et al. (1997). The shape would be ideal to
collect the nascent chain when it begins to emerge from
the ribosome and could restrict its lateral movement.
of other molecules through the membrane when the
nascent chain is present. The channel derived from na-
tive membranes has a uniform cross section along its
cation. During the initiation of translocation, perhaps
one state of the channel is converted into the other by
the recruitment of the membrane component that forms
the lumenal protrusion.
If the additional component is OST, its function would
be obvious; the tip of the lumenal protrusion may recog-
nize glycosylation sites in the nascent chain and attach
a carbohydrate chain. On the other hand, TRAP appears
to be more abundant in the ribosome-channel complex,
and its size is more consistent with the changes in the
membrane domain and the size of the lumenal protru-
sion. There are a total of seven TM domains, and three
of the four TRAP subunits have lumenal domains. In
addition, TRAP? can be crosslinked to nascent chains
in the ER lumen, whereas no crosslinks have been de-
tected with OST (Mothes et al., 1994). The function of
TRAP is currently unknown. However, it may enhance
the translocation efficiency of some proteins, be in-
volved in interactions with lumenal chaperones or modi-
fication enzymes, or regulate the opening of the pore.
Enrichment of Membrane Proteins Associated with Ribosomes
K-RM (180 equivalents [eq]) were extracted at 0.5 eq/?l with buffer
A (50 mM HEPES [pH 7.8], 1.5% digitonin, 10 mM magnesium ace-
tate, 2 mM dithiothreitol, 1:1000 protease inhibitors). The sample
was centrifuged for 40 min at 70,000 rpm in a Beckman TLA100.3
rotor at 4?C in microfuge tubes. The pellet was resuspended at 0.5
eq/?l in buffer A containing 0.5 M potassium acetate. After the
removal of aggregates in amicrofuge, the samples were centrifuged
for 40 min at 70,000 rpm in a TLA100.3 rotor. The pellet was resus-
pended at 0.5 eq/?l in buffer A containing 1.2 M potassium acetate.
Puromycin (2 mM) and GTP (1 mM) were added, and the sample
was incubated for 30 min on ice and 15 min at 37?C. It was subse-
quently spun for 40 min at 70,000 rpm in a TLA100.3 rotor. The pellet
was resuspended at 3 eq/?l in 50 mM HEPES (pH 7.8), 250 mM
wereloaded forSDS-PAGE. Asample containing50 eqof thesuper-
natants was used to analyze the protein composition. Proteins were
precipitated with 20% PEG and the pellet was washed in methanol
and subjected to SDS-PAGE. The supernatants (100 eq) were sub-
jected to Triton X-114 extraction as described by Go ¨rlich et al.
Preparation of Ribosomes and Membrane Vesicles
Yeast and rabbit reticulocyte ribosomes were purified as described
(Morgan et al., 2000). The purification of canine SRP receptor and
canine and yeast Sec61p complexes as well as their reconstitution
into proteoliposomes were carried out as described (Go ¨rlich et al.,
1993; Jungnickel and Rapoport, 1995; Hanein et al., 1996). Canine
microsomes were stripped of ribosomes with puromycin and high
salt (PK-RMs) (Neuhofet al., 1998). Salt-washedcanine microsomes
Electron Cryomicroscopy of Ribosome-Channel Complexes
Suspensions wereloaded onto300 meshgrids, withthin continuous
carbon film supported by a holey carbon mesh, that had been pre-
viously glow-discharged in air. The specimens were blotted and
ment at 4?C (?85% relative humidity). A Gatan cryotransfer system
and cryoholder (model 626-DH) were used to transfer grids into a
Philips CM12 transmission electron microscope equipped with a
cryoblade–type anticontaminator and specimen relocation system.
All electron micrographs were recorded at 100 kV under minimal
dose conditions with a LaB6filament, and a defocus range of ?1
cation on KODAK SO163 film and developed for 12 min in full-
strength D19 developer (KODAK). In some cases, images were re-
corded with the specimen tilted at 30? by using a dynamic defocus
spot scan package developed by Dr. I. Tews.
Preparation of Ribosome Channel Complexes
Transcripts coding for preprolactin 86mer were produced by in vitro
transcription with the Ribomax SP6 kit (Promega) (Jungnickel and
Rapoport, 1995). The mRNA was translated in rabbit reticulocyte
lysate in the presence of either PK-RMs or proteoliposomes con-
taining purified SRP receptor and Sec61p complex for 20 min at
27?C. To generate complexes lacking a nascent chain, the mRNA
was omitted. After translation, 1-acyl-2-(6-[7-nitro-2,1,3-benzoxadi-
azole-4-yl amino]-caproyl)-sn-glycero-3-phosphocholine (C6-NBD-
Three-Dimensional Image Processing and Analyses
Micrographs were inspected by optical diffraction, and those dis-
playing minimal astigmatism and drift were chosen for processing.
Three-Dimensional Structure of the Ribosome-Channel Complex
Entire negatives were digitized with a Zeiss SCAI scanner by using
a 7 ?m raster, binned to a pixel size of 14 ?m (corresponding to
5 A˚/pixel), and converted to SPIDER format (approximately 5000 ?
6000 pixels). Image processing was done with the SPIDER software
package (Frank et al., 1996). In most cases, particle picking was
semiautomated. The image was first cross-correlated against a ref-
erence calculated by rotationally averaging the frontal view of the
yeast ribosome.Image featureswith across-correlation peakhigher
than 0.5 were windowed (128 ? 128 pixels) from the original micro-
cles. In difficult cases, particles were picked interactively from large
sections of the original micrograph in WEB. Each precentered, 2D
dataset of ribosome-channel complexes was first aligned against
the corresponding ribosome model truncated to approximately 50 A˚
resolutionby usingRadon alignmentmethods (Radermacher,1994).
Multiple alignment cycles were done as described previously (Mor-
gan et al., 2000), and final 3D volumes were obtained by R-weighted
back projection. We restricted the resolution of the final maps so
that the CTF does not play a critical role in forming the gap (Morgan
et al., 2000).
The threshold representing 100% of the ribosomal volume was
chosen on the basis of calculated and experimentally measured
partial specific volumes and the known mass of ribosomal protein
and RNA. The 100% ribosomal volumes used in this work were
3.75 ? 106A˚3(yeast) and 4 ? 106A˚3(mammals).
breaking the data set into subsets containing 500 particles and
(Trachtenberg and DeRosier, 1987).
Beckmann, R., Bubeck, D., Grassucci, R., Penczek, P., Verschoor,
A., Blobel, G., and Frank, J. (1997). Alignment of conduits for the
K to partially translocated nascent polypeptides in intact and deter-
gent-solubilized membranes. J. Cell Biol. 108, 299–307.
Crowley, K.S., Liao, S.R., Worrell, V.E., Reinhart, G.D., and Johnson,
A.E. (1994). Secretory proteins move through the endoplasmic retic-
ulum membrane via an aqueous, gated pore. Cell 78, 461–471.
Dubochet, J., Adrian, M., Chang, J.-J., Homo, J.-C., Lepault, J.,
McDowall, A.W., and Schultz, P. (1988). Cryo-electron microscopy
of vitrified specimens. Quarterly Review of Biophysics 21, 129–228.
Frank, J., Radermacher, M., Penczek, P., Zhu, J., Li, Y., Ladjadj, M.,
and Leith, A. (1996). SPIDER and WEB: processing and visualization
of images in 3D electron microscopy and related fields. J. Struct.
Biol. 116, 190–199.
Go ¨rlich, D., Prehn, S., Hartmann, E., Kalies, K.U., and Rapoport, T.A.
(1992). A mammalian homolog of Sec61p and SecYp is associated
with ribosomes and nascent polypeptides during translocation. Cell
Go ¨rlich, D., and Rapoport, T.A. (1993). Protein translocation into
proteoliposomes reconstituted from purified components of the en-
doplasmic reticulum membrane. Cell 75, 615–630.
Hamman, B.D., Chen, J.C., Johnson, E.E., and Johnson, A.E. (1997).
The aqueous pore through the translocon has a diameter of 40–
60 A˚during cotranslational protein translocation at the ER mem-
brane. Cell 89, 535–544.
Hamman, B.D., Hendershot, L.M., and Johnson, A.E. (1998). BiP
maintains the permeability barrier of the ER membrane by sealing
thelumenalend ofthetransloconporebeforeand earlyintransloca-
tion. Cell 92, 747–758.
Hanein, D., Matlack, K.E.S., Jungnickel, B., Plath, K., Kalies, K.-U.,
Miller, K.R., Rapoport, T.A., and Akey, C.W. (1996). Oligomeric rings
of the Sec61p complex induced by ligands required for protein
translocation. Cell 87, 721–732.
Hartmann, E., Go ¨rlich, D., Kostka, S., Otto, A., Kraft, R., Knespel,
S., Burger, E., Rapoport, T.A., and Prehn, S. (1993). A tetrameric
complex of membrane proteins in the endoplasmic reticulum. Eur.
J. Biochem. 214, 375–381.
Hedge, R.S., and Lingappa, V.R. (1996). Sequence-specific alter-
ation of the ribosome-membrane junction exposes nascent secre-
tory proteins to the cytosol. Cell 85, 217–228.
Hedge, R.S., and Lingappa, V.R. (1997). Membrane protein biogene-
sis: regulated complexity at the endoplasmic reticulum. Cell 91,
Jones, T.A., Zou, J.-Y., and Cowan, S.W. (1991). Improved methods
of errors in these models. Acta Crystallogr. A 47, 110–119.
Jungnickel, B., and Rapoport, T.A. (1995). A posttargeting signal
sequence recognition event in the endoplasmic reticulum mem-
brane. Cell 82, 261–270.
Matlack, K.E.S., Mothes, W., and Rapoport, T.A. (1998). Protein
translocation: tunnel vision. Cell 92, 381–390.
Meyer,T.H.,Me ´ne ´tret,J.F.,Breitling,R.,Miller,K.R.,Akey,C.W.,and
Rapoport, T.A. (1999). The bacterial SecY/E translocation complex
forms channel-like structures similar to those of the eukaryotic
Sec61p complex. J. Mol. Biol. 285, 1789–1800.
Migliaccio, G., Nicchitta, C.V., and Blobel, G. (1992). The signal
sequence receptor, unlike the signal recognition particle receptor,
is not essential for protein translocation. J. Cell Biol. 117, 15–25.
Morgan, D.G., Me ´ne ´tret, J.F., Rademacher, M., Neuhof, A., Akey,
I.V., Rapoport, T.A., and Akey, C.W. (2000). A comparison of the
yeast and the rabbit 80S ribosome reveals the topology of the na-
scent chain exit tunnel, inter-subunit bridges and mammalian rRNA
expansion segments. J. Mol. Biol. 301, 301–321.
Mothes, W., Prehn, S., and Rapoport, T.A. (1994). Systematic prob-
ing of the environment of a translocating secretory protein during
translocation through the ER membrane. EMBO J. 13, 3937–3982.
Processing of Ribosomes on Intact Membranes
eter) were generally seen in edge views. To select ribosomes bound
to channels, we took advantage of the fact that they should freely
rotate about an axis perpendicular to the membrane plane; nonspe-
cifically bound ribosomes should adopt all possible angular orienta-
tions. An initial 2D alignment was used to place the membranes in
a similar orientation. The soluble yeast ribosome-Sec61p complex
was used as a reference after the rotation of this map so that 3D
rotation around the axis normal to the membrane could be readily
identified. After iterative cycles of alignment, images whose align-
ment angles differed significantly from those expected were elimi-
nated, andthe updatedreference wasobtained byR-weighted back
All particles were evaluated with a second procedure to test for
genuinely bound ribosomes. A model 3D volume was constructed
with an idealized vesicle surface passing through the Sec61p chan-
nel within the 3D map of the yeast ribosome-channel complex. This
model was then projected at the 3D alignment angles determined
and actual positions of the membrane. When these two images
agreed, the ribosome was considered to be associated with an
underlying channel. Overall, there was good agreement in the parti-
cles selected with either criteria.
We thank I. V. Akey for yeast ribosomes, J. Stahl for antibodies to
S26, and K. Matlack for help with experiments, suggestions, and
critical reading of the manuscript. J. F. M. was supported by NIH
training grants, and D. G. M. was supported by the HHMI. Both the
C. W. A. and T. A. R. laboratories are supported by NIH grants, and
T. A. R. is a Howard Hughes Medical Institute investigator.
Received August 16, 1999; revised September 29, 2000.
Agrawal, R.K., Spahn, C.M., Penczek, P., Grassucci, R.A., Nierhaus,
K.H., and Frank, J. (2000). Visualization of tRNA movements on the
Escherichia coli 70S ribosome during the elongation cycle. J. Cell
Biol. 150, 447–460.
Mothes, W., Heinrich, S.U., Graf, R., Nilsson, I., von Heijne, G., Brun-
ner, J., and Rapoport, T.A. (1997). Molecular mechanism of mem-
brane protein integration into the endoplasmic reticulum. Cell 89,
Neuhof, A., Rolls, M.M., Jungnickel, B., Kalies, K.-U., and Rapoport,
T.A. (1998). Binding of signal recognition particle gives ribosome/
nascent chain complexes a competitive advantage in endoplasmic
reticulum membrane interaction. Mol. Biol. Cell 9, 103–115.
Ooi, C.E., and Weiss, J. (1992). Bidirectional movement of a nascent
polypeptide across microsomal membranes reveals requirements
for vectorial translocation of proteins. Cell 71, 87–96.
Plath, K., Mothes, W., Wilkinson, B.M., Stirling, C.J., and Rapoport,
T.A. (1998). Signal sequence recognition in posttranslational protein
transport across the yeast ER membrane. Cell 94, 795–807.
Prinz, A., Behrens, C., Rapoport, T.A., Hartmann, E., and Kalies,
K.-U. (2000). Evolutionarily conserved binding of ribosomes to the
translocation channel via the large ribosomal RNA. EMBO J. 19,
Radermacher, M. (1994). Three-dimensional reconstitution from Ra-
don projections: orientational alignment via Radon transforms. Ul-
tramicroscopy 53, 121–136.
Silberstein, S., and Gilmore, R. (1996). Biochemistry, molecular biol-
ogy, and genetics of the oligosaccharyl transferase. FASEB J. 10,
Simon, S.M., and Blobel, G. (1991). A protein-conducting channel
in the endoplasmic reticulum. Cell 65, 371–380.
Trachtenberg, S., and DeRosier, D.J. (1987). Three dimensional
structure of the frozen-hydrated flagellar filament: the left handed
filament of Salmonella typhimurium. J. Mol. Biol. 195, 581–601.