Structure of the Mammalian Ribosome–Channel Complex at 17 Å Resolution

doi:10.1016/S0022-2836(02)01111-7 available online at http://www.idealibrary.com on

w B

J. Mol. Biol. (2002) 324, 871–886

Structure of the Mammalian Ribosome –Channel ˚ Resolution Complex at 17 A David Gene Morgan1,2,3†, Jean-Franc¸ois Me´ne´tret1†, Andrea Neuhof2, Tom A. Rapoport2 and Christopher W. Akey1* 1

Department of Physiology and Biophysics, Boston University School of Medicine, 700 Albany St., Boston, MA 02118-2526 USA 2

Howard Hughes Medical Institute and Department of Cell Biology, Harvard Medical School, 240 Longwood Avenue Boston, MA 02115, USA 3 Department of Molecular Cellular and Developmental Biology, University of Colorado Boulder, CO 80309, USA

The co-translational translocation of proteins into the endoplasmic reticulum (ER) lumen and the biogenesis of membrane proteins require ribosome binding to a membrane channel formed by the Sec61p complex. We ˚ structure of a mammalian ribosome – channel now report the 17 A complex derived from ER membranes. Atomic models of the ribosomal subunits were aligned to the programmed ribosome from Thermus thermophilus, to provide a common reference frame. The T. thermophilus ribosome, and by extension all known high resolution subunit models, were then docked within our map of the ribosome – channel complex. The structure shows that the ribosome contains a putative tRNA in the exit site, and a comparison with a non-programmed, yeast ribosome suggests that the L1 stalk may function as a gate in the tRNA exit path. We have localized six major expansion segments in the large subunit of the vertebrate ribosome including ES27, and suggest a function for ES30. The large ribosomal subunit is linked to the channel by four connections. We identified regions in the large subunit rRNA and four proteins that may help form the connections. These regions of the ribosome probably serve as a template to guide the assembly of the asymmetric translocation channel. Three of the connections form a picket fence that separates the putative translocation pore from the attachment site of an additional membrane component. The ribosome –channel connections also create an open junction that would allow egress of a nascent chain into the cytosol. At a threshold that is appropriate for the entire complex, the channel is rather solid and the lumenal half of the putative translocation pore is closed. These data suggest that the flow of small molecules across the membrane may be impeded by the channel itself, rather than the ribosome – channel junction. q 2002 Elsevier Science Ltd. All rights reserved

*Corresponding author

Keywords: ER channel; ribosome; Sec61p complex; co-translational translocation

Introduction During co-translational translocation many proteins are transported into the lumen of the endoplasmic reticulum (ER), while others are integrated into the membrane. This process requires a † These two authors contributed equally to this work. Abbreviations used: EM, electron microscopy; ER, endoplasmic reticulum; ES, expansion segment; TM, trans-membrane; TRAP, translocon-associated protein; OST, oligosaccharyl transferase; FSC, Fourier shell correlation; SSNR, spectral signal to noise ratio; PTC, peptidyl transferase center; TE, tunnel exit. E-mail address of the corresponding author: [email protected]

direct interaction between the translating ribosome and the translocation channel, thereby allowing the nascent polypeptide chain to be transferred from a tunnel in the ribosome into a pore within the channel.1,2 A complete understanding of this process will require detailed structures of the eukaryotic ribosome, the channel, and the ribosome – channel complex with and without a nascent chain. A large body of data is now available on the and function of prokaryotic structure3 – 6 ribosomes.7,8 In addition, a model of the yeast ribosome has been generated, based on atomic structures of the archaebacterial subunits and a ˚ map obtained by electron cryo-microscopy.9 15 A

0022-2836/02/$ - see front matter q 2002 Elsevier Science Ltd. All rights reserved

872

Structure of the Native Ribosome –Channel Complex

Figure 1. Fourier shell curves for the mammalian ribosome– channel complex. The curve with open circles represents a comparison between half 3D volumes calculated with the conjugate gradient backprojection method. The resolution ˚ , based on is estimated to be 17.5 A the FSC 0.5 cutoff, while the 5s value (continuous line) suggests a ˚ . The curve with value of , 13 A filled circles was calculated using the 3D SSNR method and gives a ˚. resolution of ,16.3 A

This work revealed a P-site tRNA within the decoding and peptidyl transferase centers. Much less is known about the mammalian ribosome, which is considerably larger than the yeast ribosome. This increase is due to the size of six rRNA expansion segments (ES) located on the surface of the large subunit.10 – 12 The function of these segments is not known but the deletion of ES27 is lethal.13 Intriguingly, ES27 in the yeast ribosome moves away from the tunnel exit (TE) domain of the large subunit when the channel is present.14 The ER channel is a ring-like oligomer consisting of three to four copies of the Sec61p complex.15 – 17 The channel is formed from the heterotrimeric Sec61p complex, which consists of an a-subunit with ten trans-membrane (TM) domains and two smaller subunits (b and g), each with a single TM domain.18 This complex is conserved in evolution and the bacterial homolog (SecYEG) shares a similar ring-like morphology.19 In addition, both the Sec61p and SecYEG complexes bind to ribosomes directly.20 – 23 The Sec61p complex may exist as individual heterotrimers in the plane of the membrane, and channels are assembled upon addition of ribosomes, as demonstrated by freeze-fracture electron microscopy (EM).15 However, channels persist in ER membranes after ribosomes are removed with high salt and puromycin, which suggests that the channels may be stabilized by interactions with other membrane components.15 In a current model, the ribosome interacts with the Sec61p channel to form a junction that is sealed towards the cytosol. This in turn, creates a continuous path for the nascent chain from the ribosome to the ER lumen for secretory proteins or the lumenal segments of membrane proteins.24,25 When the ribosome synthesizes a TM domain, the junction must open to allow the following polypeptide segment to emerge into the cytosol. This model is supported by experiments that demonstrate changes in the accessibility of fluorescent probes incorporated into the nascent chain when quenching reagents are added to the cytosol.26 These experiments suggest that the ribosome – channel junction normally provides a seal for ions

and other small molecules, preventing their passage through the membrane. Other experiments with fluorescence quenchers of different sizes suggest that the pore diameter changes significantly in the presence of a nascent chain.27 Recent EM structures have resulted in a different model for the ribosome –channel junction.14,16,17 In these studies, only a few connections between the ribosome and the channel were seen, regardless of whether a translocating nascent chain was present. In addition, candidate regions of the large subunit that may contribute to the connections have been described for a translating yeast ribosome – channel complex.14 Thus, the EM structures do not support the concept of a tight seal between the ribosome and the channel. Rather, it was suggested that a seal to small molecules might exist within the channel itself. The gap between the ribosome and the channel may be important for membrane protein integration, as it would provide a lateral path into the cytosol, when a TM domain stops the transfer of a polypeptide chain through the channel. This would allow the cytosolic domain of a membrane protein to emerge from the ribosome – channel junction without the necessity for a large conformational change. Reconstitution experiments have demonstrated that the Sec61p complex is sufficient to form a membrane channel. However, it is clear that other components associate with the channel. In particular, the translocon-associated protein (TRAP) and the oligosaccharyl transferase (OST) remain bound to the ribosome –Sec61p complex after solubilization of mammalian ER membranes.17,28 While the function of TRAP is unknown, the primary role of OST involves the N-glycosylation of nascent chains.29 Although neither protein is essential for translocation, one of these multi-subunit complexes is likely responsible for the lumenal protrusion in channels derived from ER membranes.17 Another protein, RAMP4, associates with ribosomes and may be important for the translocation of certain proteins.30 Our previous structures of the mammalian, ribosome – channel complex were limited by their

Structure of the Native Ribosome –Channel Complex

873

Figure 2. Surface views of the mammalian ribosome– channel complex. (a) A front view of the ribosome– channel complex reveals the presence of a tRNA (red) in the E-site, located between the small (S) and large subunits (L). A bridge connects the small and large subunit (B). The native channel (gold) is connected to the large subunit by three prominent connections (C). An extension of rRNA from ES27 (Ext) encroaches on the ER membrane. (b) A back view of the complex with the small subunit on the right reveals the three largest connections (C) and inter-subunit bridges, which are marked by a B and a white dot (EB8). (c) A top view reveals two bridges (labeled EB1a and EB1b/c) located between the head of the small subunit and the central protuberance of the large subunit. These bridges correspond to the feature labeled B in (a) and (b). (d) A bottom view of the ribosome –channel complex reveals the elliptical shape of the native channel and a prominent lumenal domain (LD), which points towards a dimple that likely forms the exit of the translocation pore (P). Bridge EB11 is indicated by a white dot. The scale bar ˚. represents 100 A

˚ ) and by difficulties in choosing resolution (, 27 A an appropriate threshold for the channel and the ribosome.17 We now report the 3D structure of a mammalian, ribosome – channel complex derived ˚ resolution, this map from ER membranes. At 17 A provides insight into the biological function of the mammalian ribosome, the ribosome – channel junction, and the channel.

Results Overview of the ribosome – channel complex The preparation of ribosomes in a defined state is an important factor in obtaining higher resolution with single-particle methods, because the small and large subunits may ratchet past each other during translation.31 Thus, we assembled a non-translocating, native ribosome –channel complex as follows. Puromycin-treated ribosomes and ribosome-stripped membranes were prepared from rough microsomes. Purified ribosomes were then added back to stripped ER membranes and the resulting membranes separated from unbound ribosomes by floatation on a sucrose gradient. These membranes were solubilized and the ribosome – channel complexes were concentrated by centrifugation and frozen rapidly on carbon-coated grids.17 In total, , 26,000 particles were used to calculate a 3D map that was corrected for phase reversals of the contrast-transfer function. The final data were

divided into halves and compared using Fourier shell correlation (FSC; Figure 1). The resolution is ˚ at the FSC0.5 cutoff, while estimated to be , 17.5 A ˚. the 5s cutoff (5 £ noise) gives a value of , 13 A We used the 3D spectral signal to noise ratio (SSNR) to estimate the resolution,32 and this ˚ . Thus, a nominal method gave a value of , 16.3 A ˚ resolution of 17 A is used to describe the structure. Surface views of the native, ribosome – channel complex are shown in Figure 2. The frontal view (Figure 2(a)) corresponds to a view along the plane of the ER membrane with the small ribosomal subunit (S) on the left and the large subunit (L) on the right. Many new features are observed within the ribosome, including an L-shaped density located between the small and large subunits (shown in red). This feature corresponds to a putative tRNA located within the E-site and is discussed later. A second feature forms a bridge between the head of the small subunit and the central protuberance of the large subunit (B in Figure 2(a)). This feature is seen also in a view from the back of the ribosome – channel complex (Figure 2(b)), while from the top it is resolved into a pair of bridges (white dots in Figure 2(c)). These bridges correspond to EB1a and EB1b/c in the yeast ribosome.9 Thus, in total there are seven strong bridges between the mammalian ribosomal subunits, which include EB2, EB3, EB7, EB8 and EB11 (note: EB612 has now been renamed EB11, to conform to the nomenclature used by Spahn et al.;9 and see Figure 2(d)). These bridges all have

874

Structure of the Native Ribosome –Channel Complex

Figure 3. Docking atomic structures of prokaryotic subunits into a map of the mammalian ribosome. (a) The overall fit of the atomic structure of the T. thermophilus small subunit is shown within the corresponding region of the EM map. Proteins are colored in turquoise and rRNA is shown in gold. (b) The head of the small subunit is shown from the interface side, with a transparent surface from the EM map superimposed on the atomic structure. The positions of H18, S7 and S11 are indicated. The head of the small subunit rests on H18 (white dot), which helps form the mRNA entrance tunnel. The head and the back of the platform are connected by S7 and S11, which may form an exit tunnel for the mRNA (marked by a red dot). A region of H44 indicated by a white bracket is outside of the EM envelope. (c) The fit of the T. thermophilus large subunit is shown within the corresponding region of the EM map of the ribosome– channel complex. Landmarks of the crown view are labeled, including the L1 and L7/L12 stalks, and the central protuberance (CP). (d) A close-up view of the interface canyon and central protuberance of the large subunit is shown, with a transparent surface from the EM map superimposed on the atomic structure. The positions of rRNA helices H38, H76 and H69 (white dot) are indicated. H38 forms part of the A-site finger, while H69 forms part of EB2. The L1 stalk is supported upon a pillar formed by H76. (e) The atomic structure of the H. marismortui large subunit was docked into the EM map. The fit of the rRNA helices is shown in this cutaway view, with the channel located at the bottom and the small subunit (S) on the left. The nascent chain exit tunnel is constricted near the peptidyl transferase center (PTC). This threshold also reveals connections between the ribosome and the channel (C2 and C3). (f) A second view of the EM map is shown with the docked H. marismortui large subunit. The small subunit (S) is located on the right. The interface canyon (IC) is shown in cross-section, along with connection C1. Protein L22 at the channel-docking surface is highlighted (purple) and its fit supports the choice of the threshold.

counterparts in the prokaryotic ribosome, with the exception of EB8 (Figure 2(b)), which is located at the back of the platform on the small subunit.12 At the bottom of the ribosome, the channel from native ER membranes (shown in gold; Figure 2(a), (b) and (d)) is connected with the large subunit through three major connections (labeled C). A fourth connection is not visible at this threshold. A prominent domain (LD) is attached to the channel by a stalk and extends into the ER lumen. In a bottom view of the complex, the tip of the lumenal domain is located adjacent to a dimple, which marks the probable exit point of the translocation pore (P in Figure 2(d)).

A comparison with prokaryotic ribosomes Recent work has established a model of the yeast ribosome because the eukaryotic ribosome shares a conserved framework with its prokaryotic counterparts.9 Vertebrate ribosomes are considerably larger than those from yeast, but are expected to share conserved features with prokaryotic and yeast ribosomes.11,12,33,34 We aligned the highresolution structures of small subunits from Thermus thermophilus4,5 and large subunits from Haloarcula marismortui and Deinococcus radiodurans3,35 with the intact T. thermophilus ribosome6 to provide a common reference frame. A molecular envelope from the T. thermophilus

Structure of the Native Ribosome –Channel Complex

875

Figure 4. A putative E-site tRNA in the mammalian ribosome– channel complex and the L1 stalk. (a) A top view of the interface between the small (S) and large (L) subunits is shown in a thick slab of density, with the head of the small subunit and the top of the central protuberance on the large subunit removed. The E-site tRNA (shown ˚ X-ray strucin pink) from the 5.5 A ture of the T. thermophilus ribosome6 is docked in density between the subunits, adjacent to the L1 stalk. Expansion segment 30 (ES30) and helix 78 (H78) are indicated. Together with the L1 protein (L1p) and H77 (not labeled), they form the top of the L1 stalk. In addition, a major bridge (EB2) connects the interface canyon on the large subunit to the small subunit. (b) The view in (a) has been rotated ,708 back into the plane of the page to reveal contacts between the E-site tRNA and the underside of the head of the small subunit. Both bridges EB2 and EB3 are visible in this orientation. (c) A domain of the L1 protein (L1p) was docked into the top of the L1 stalk, as described in the text. (d) Helix 78 (H78) from the progammed T. thermophilus ribosome was modeled into the mammalian L1 stalk. We suggest that H78 reorients towards the lateral surface of L1 in mammals, to accommodate the additional nucleotides of ES30. (e) The L1 stalk of the nonprogrammed yeast ribosome12 is shown in yellow mesh, superimposed on the large subunit from the map of the mammalian ribosome – channel complex. Helix 76 from the atomic structure of the D. radiodurans large subunit35 is shown in green and fits well within the yeast ribosome map. The position of H76 from the programmed T. thermophilus ribosome is indicated by white dots. Thus, the eukaryotic L1 may pivot about the base of H76 (red dot) to function as a lateral gate for the E-site tRNA.

ribosome was then used to manually dock the X-ray structure into the EM map using O.36 We next separated the canine ribosomal subunits in the map by cutting through the bridges between them.12,37 This allowed us to evaluate the overall fit of the atomic subunit models. The small subunit is shown in Figure 3(a), as viewed from the solvent side. This docking used distinctive features of the head, shoulder and the mRNA entrance tunnel. A close-up of the head and neck region is shown in Figure 3(b), as viewed from the interface side. In this view, the mRNA entrance tunnel formed by the head and H18 is well resolved.6 In addition, a completely enclosed exit tunnel for mRNA (see the red dot in Figure 3(b)) appears to be formed by vertebrate homologs of proteins S7 (RS5) and S11 (RS14). Extra density within the tunnel wall may be due to the larger size of the mammalian proteins, which together account for an additional 69 residues. A similar tunnel was observed in the small subunit of the yeast ribosome.9 Thus, a complete path for the transit of mRNA is delineated by a pair of tunnels that are located above the platform of the small subunit. Intriguingly, helix 44 appears to be less well ordered in our map, as there is no density for nucleotides , 1400 –1406 or 1451– 1463, in a region below the platform (white bracket in Figure 3(b)). The local disorder of H44 may reflect the nontranslating state of these ribosomes. The vertebrate large subunit is shown in Figure 3(c), as viewed from the subunit interface. The

canyon, the central protuberance, the L1 stalk, and the nascent chain exit tunnel were used to guide the docking of the atomic model from the T. thermophilus ribosome.6 A close-up view of the interface canyon allows an evaluation of the docking (Figure 3(d)). Overall, we found a good fit for the interface canyon and H69 (see the white dot in Figure 3(d)), which forms part of bridge EB2. How˚ within the A-site finger ever, we moved H38 , 8 A to improve its fit to the density. The L1 stalk region contains the putative E-site tRNA and was modified by reorienting H78 in a direction away from the subunit interface (Figure 3(c)). The known atomic structures from H. marismortui, T. thermophilus and D. radiodurans3,6,35 could each be fit within the central core of the mammalian large subunit because the models are similar in this region. The general quality of the fit can be judged from cross-sections of the large subunit into which the atomic structure of the H. marismortui large subunit is docked (Figure 3(e) and (f)). Large subunit proteins were omitted from these cross-sections to allow a clear view of the rRNA within the density. Most proteins also fit well, as shown for L22 (Figure 3(f)), which is located at the channel-docking surface and extends along the tunnel.38 In the map, the nascent chain exit tunnel is constricted at the entrance, near the peptidyl transferase center (PTC; Figures 3(e), and 7(c) and (d)). We observed an extensive tunnel system within the large subunit (Figure 7(c) and (d)), as noted previously for the Escherichia coli ribosome.39 When taken together, the quality of

876

Structure of the Native Ribosome –Channel Complex

Figure 5. Expansion segments in the large subunit of the mammalian ribosome. (a) A view of the ribosome – channel complex is shown from the back of the small subunit (blue), along the plane of the membrane. A comparison with a map of the yeast ribosome12 and with the known atomic structures of the prokaryotic large subunits, allowed the major expansion segments to be identified (shown in red). The positions of ES27, ES30 and fingers 3 and 4 are indicated. A portion of ES27, identified previously as the yeast spine, is colored in green and an extension (Ext) points down towards the membrane. (b) A view of the ribosome– channel complex is shown from the back of the large subunit. Two major spines (labeled 1 and 2, in red) are interconnected to form a V-shaped feature that spans the back of the large subunit. Expansion segments 7, 15 and 39 originate from the prokaryotic large subunit ˚. in the region outlined by the ellipse. The scale bar represents 100 A

the overall fit indicates that helical segments of rRNA and regions of ribosomal proteins can be identified in our map using the known atomic structures and their homology with eukaryotic components. The exit site for tRNA Our 3D map reveals an L-shaped rod of density located between the subunits, adjacent to the L1 stalk (Figures 2(a) and (b), and 4(a) and (b)). The density of this feature is comparable to known rRNA components in the map. In addition, the shape and position of this peak are nearly identical with an E-site tRNA observed in the programmed T. thermophilus ribosome.6 This suggests that a tRNA molecule is retained in the E-site after programmed mammalian ribosomes are treated with puromycin. Intriguingly, the E-site tRNA in the T. thermophilus ribosome originated from endogenous tRNAs that survived extensive purification.6 The putative E-site tRNA in our map proved to be docked robustly, as it survived treatment with digitonin and high salt. The putative tRNA makes distinct contacts to the ribosome that involve the head of the small subunit, the canyon rim of the large subunit and H77 of the L1 stalk. Importantly, this is the first time that a putative E-site tRNA has been observed in a eukaryotic ribosome. The L1 stalk forms part of the binding site for the E-site tRNA, and was rather disordered in maps of non-programmed mammalian ribosomes.12,17 However, in this map the L1 stalk is well ordered and its fit to the corresponding region of the T. thermophilus ribosome is good, but not perfect (Figure 4(c) and (d)). In particular, H76 is bent outwards away from the subunit interface, which may reflect a conformational change of the L1 stalk (Figures 3(d) and 4(d)). An expansion segment (ES30) is located at the tip of H78 and its size (, 55 nt) may have induced H78 to reorient later-

ally, thereby minimizing a possible steric clash with the small subunit (Figure 4(d)). The reorientation of the vertebrate L1 stalk may necessitate local changes in H77 and movements of the two domains within L1p (RL10A in mammals). The L1 protein from H. marismortui is similar in size to its mammalian counterpart. Thus, we moved L1p ˚ , to optimize the fit of the top upwards by , 10 A domain in the density (Figure 4(c); amino acid residues 67 – 160). The lower domain of L1p must also be moved towards the back of the large subunit, to accommodate changes in H77 and H78 (not shown). Overall, the data suggest that components of the vertebrate L1 stalk have undergone a concerted conformational change, relative to their T. thermophilus counterparts. A higherresolution EM map will be required to build a pseudo-atomic model of the mammalian L1 stalk. We also noticed that the L1 stalk is bent away from the E-site in a previous map of the nonprogrammed yeast ribosome (Figure 4(e)).12 Intriguingly, the yeast L1 stalk is ordered enough to allow an alternate conformation of H76 from the D. radiodurans large subunit35 to be docked within the map. A comparison between the canine and yeast maps suggests that the E-site tRNA may lock the flexible L1 stalk into a defined conformation in which H76 moves closer to the large subunit (Figure 4(e))6,35 Given this flexibility, it is possible that the L1 stalk may act like a gate to release the E-site tRNA by pivoting about the base of H76.35,40 Expansion segments in the large subunit The mammalian ribosome is , 0.4 MDa larger than its yeast counterpart, due primarily to the size of six expansion segments in the large subunit.10,41 In the current study, we have used the known atomic structures of prokaryotic large subunits, along with EM maps of the yeast and canine

Structure of the Native Ribosome –Channel Complex

877

Figure 6. Candidate regions of the large subunit that may form connections within the ribosome– channel junction. (a) A bottom view of the channel-docking surface on the large subunit is shown as a wire mesh. Six proteins of the H. marismortui large subunit are color-coded38 and selected regions of the rRNA are highlighted as yellow ribbons, while the remaining rRNA is shown as a yellow strand. The nascent chain exit tunnel is indicated by an asterisk (p ). (b) The EM map has been removed to allow a clear view of the channeldocking surface. The positions of four connections with the ER channel are indicated by white circles. Conserved components of the H. marismortui large subunit that may contribute to the connections are indicated. (c) A crosssection of the EM map reveals a picket fence between the large subunit and the channel, comprised of connections C1, C2 and C4. Components of the H. marismortui large subunit that may contribute to their formation are shown. (d) A view of the largest connection (C2) is revealed in a thick slab of the EM map, rotated ,908 from the view in (c). L23 is closest to the centrally located C2 connection. (e) Connection C3 is located adjacent to the nascent chain exit tunnel and may be formed by a contact with H24 and possibly with L24.

ribosome, to evaluate the vertebrate expansion segments. As noted previously,12 there are spines and finger-like features on the canine large subunit that extend beyond the boundary of the docked yeast large subunit (red features in Figure 5). However, there is no additional RNA located directly between the subunits. Although the positions of the expansion segments are conserved, their sizes vary and they have no known functions. However, specific roles are emerging for ES30 and ES27. Expansion segment 30 is found as an additional density that extends from H78 on the solvent side of the L1 stalk (Figures 4(c), and 5(a) and (b)). We estimate that roughly half of the 55 nucleotides in ES30 may be disordered. Given the possible role of the L1 stalk as a gate in the tRNA exit path,35,40 it is intriguing that an expansion segment would be incorporated into this feature. As noted in the previous section, the increased size of the vertebrate ES30 may have caused significant conformational changes in the L1 stalk. Thus, ES30 may influence the dynamic properties of the putative lateral gate for tRNAs in higher eukaryotes. Expansion segment 27 is an extension of H63 and is located at the base of the large subunit, below the subunit interface.9 In yeast ribosomes, this region contains an elongated spine,9,12 while in rabbit ribosomes only two finger-like projections were found (Figure 5(a)).11,12 In the canine ribosome – channel complex, there are two fingers and an elongated spine in this region (Figure 5(a)). In addition, an extension from ES27 points down towards the membrane surface (Figure 2(a) and

(b), and 5(a)) and could mediate additional interactions between the large subunit and the ER membrane. This extension from ES27 is not present in the yeast large subunit. The measured contour length of these features suggests that most of the additional nucleotides in vertebrate ES27 (, 480) may extend into solution, where they are disordered. The additional rRNA in the vertebrate ES27 may cause further flexibility in this feature, as the spine was not observed in earlier maps of the vertebrate ribosome.11,12 In hindsight, the ES27 spine is present in most previous maps of ribosome – channel complexes when they are contoured at a lower threshold,17 although it is generally more prominent in specimens derived from native membranes. In yeast, the ES27 spine was found next to the TE domain of the large subunit, while it relocates to a position below the small subunit when the channel is present.14 The rearrangement of ES27 may be required for the channel to bind to the ribosome. In addition, ES27 may play a role in events at the nascent chain TE, prior to arrival of the channel.14 ES7, ES15 and ES39 are located on the solventexposed back of the bacterial large subunit, and originate within the region indicated by the ellipse on the canine ribosome (Figure 5(b)). These expansion segments contain , 665 nt and must give rise to the two large, interconnected spines (denoted 1 and 2) and the two fingers located below L7/L12 (marked 3 and 4). By modeling, we found that only a single helix would fit within each rod-like density. This suggests that the expansion segments are formed by extended, helical protrusions. We

878

Structure of the Native Ribosome –Channel Complex

Figure 7. The native protein translocation channel. (a) A top view of the ER channel is shown, as viewed from the cytoplasm. The picket fence formed by connections C1, C2 and C4 is indicated, and the position of C3 is marked with a white circle. In addition, there are two cytoplasmic entrances into the pore (labeled CE and CE2). (b) A central slice of the ER channel reveals the acentric pore lumen and the base of the stalk, which connects the channel to the lumenal domain. (c) A cross-section of the EM map reveals the topology of the nascent chain exit tunnel, the pore lumen and exit. The nascent chain (white dots) may exit from the tunnel in the large subunit, and enter the channel at the pore entrance located in front of the picket fence (CE in (a)). The position of the pore exit suggests that the nascent chain may have to bend somewhat as it threads across the channel. (d) A view of the EM map, rotated , 908 relative to (c), reveals the cup-like shape of the pore in this orientation.

measured the total contour length of these features. When combined, these data suggest that only , 60% of the rRNA in ES7, ES15 and ES39 is well ordered. Overall, the three expansion segments span the entire back and L7/L12 side of the large subunit. They make contacts with domains I, II and III in the 28 S rRNA and with the 5 S rRNA. Thus, the ordered regions of ES7, ES15 and ES39 appear to form braces on the back and sides of the large subunit that may interconnect different rRNA domains, while the disordered portions could mediate interactions with cellular components during the biogenesis and intracellular targeting of ribosomes. Finally, ES9 is located on the L1 side of the central protuberance. Size differences between the yeast and canine maps in this region are consistent with the expected 30– 40 nt difference between yeast and mammals (Figure 5(b)).9 Connections between the ribosome and the channel We detected a gap between the ribosome and the channel in previous structures of mammalian and yeast ribosome –channel complexes.17 Inspection of the new map reveals a similar ribosome –channel junction (Figure 2(a) and (b)) and, importantly, the gap is present at all reasonable thresholds. The open structure of the junction is shown in more ˚ wide gap is detail in Figure 6(c)– (e). The 15– 20 A

spanned by four stalk-like connections. Three of the connections are relatively strong (denoted C1, C2 and C4) and form a picket fence across the top of the channel (Figure 6(b) and (c)). This linear arrangement is illustrated further by Figure 7(a), which shows the connections as viewed from the ribosome. An additional connection (C3) is located next to the TE and is weaker (Figures 6(e) and 7(d)). The C2 connection extends towards the cen˚ ter of the channel and has a diameter of , 20 A (Figures 6(c), and 7(a), (c) and (d)). A similar pattern of connections was observed by superimposing ten different 3D maps17 and, more recently, in the yeast ribosome – channel complex.14 Thus, the junction of the ribosome – channel complex appears to be universally conserved in eukaryotes, which suggests that the gap must play an important role in protein translocation. We next identified regions of the vertebrate ribosome that may participate in forming the connections. This is possible because a single threshold can be used to describe both the ribosome and the channel. After docking the H. marismortui large subunit into our map, the channel was removed to reveal the docking surface (Figure 6(a)). A simplified view of this region is shown in Figure 6(b), along with the positions of the four connections (white circles). Regions of the bacterial 23 S rRNA in close proximity to the connections are shown as yellow ribbons and six ribosomal proteins located near the TE are highlighted (Figure 6(a) and (b)).

879

Structure of the Native Ribosome –Channel Complex

Table 1. Ribosomal components that may participate in forming connections between the large subunit and the ER channel Connections

H. marismortui

Yeasta

Human

C1

L19e (148)c

C2

L23 (84)

C3

L24 (119)

C4

L29 (70)

RL19 (188) RL25 (141) RL26 (126) RL35 (120)

RL19 (196) RL2B (156) RL26 (145) RL35 (122)

23 S rRNAb H59 H50/H53 H24 H7

a

The large subunit proteins and rRNA regions involved in forming the yeast ribosome–channel junction were identified by Beckmann et al.14 b From H. marismortui. c The number of amino acid residues in the protein are indicated in parentheses.

This analysis suggests that both rRNA and proteins may help to form the connections. For example, connection C1 may be formed by protein L19e and/or rRNA centered on nucleotide 1631 in H59 (Figure 6(b) and (c)). Connection C2 may contain protein L23, which is the closest component (Figure 6(c) and (d)). However, the shape and length of the connection suggests that it may be formed, in part, from rRNA. In this case, the best candidate is a region of H50 comprising nucleotides 1415 –1443. However, this helix would have to undergo a significant conformational change (Figure 6(d)). The C2 connection may be formed by contributions from H53 (Figure 6(d)). The C3 connection may be formed by protein L24 and/or rRNA centered on nucleotide 490 in H24 (Figure 6(e)). Finally, the data suggest that connection C4 may be formed from a loop comprising nucleotides 93– 95 of H7, with possible contributions from protein L29 (Figure 6(c)). Because of its loop-like character, H7 could flip downwards to contact the top of the channel. Each of the rRNA regions that have been implicated in forming the connections is well conserved from bacteria to mammals, with the exception of H59, which is 12 nucleotides larger in mouse. The extension of H59 may help form the C3 connection, as proposed for the yeast ribosome – channel complex.14 In addition, candidate proteins of the H. marismortui large subunit are generally larger in yeast and mammals (Table 1) and could make significant contributions. Importantly, the pattern of connections in the yeast and mammalian channels is comparable14,17 (and this work) and utilizes similar helical elements located within the 50 half of the large subunit rRNA (Table 1). The native channel from mammalian ER contains an additional membrane protein component.17 When taken together, these data suggest that the presence of a heterologous membrane protein does not alter the pattern of connections between the channel and the ribosome. The picket fence alignment of the

three strongest connections, and the placement of the C2 connection in the channel center (Figure 6(d)) may place constraints on conformational changes associated with protein translocation. In addition, the conserved docking surface around the TE12,14 (and this work) suggests that the SecYEG channel may dock to a prokaryotic ribosome in a similar manner. Structure of the native channel The channel derived from native membranes is much larger than the channel comprised of purified Sec61p and its elliptical shape (Figures 2(d) and 7(a)) is due to the presence of an additional membrane protein.17 The expanded region lies at one end of the channel and is connected to a lume˚ from the channel nal domain, whose tip is , 35 A (Figures 2(a), and 7(c) and (d)). The additional component in native channels may correspond to either TRAP or OST,17 although its size is more consistent with TRAP. This heterologous membrane protein may either associate with the lateral surface of the channel or be intercalated between subunits. In the latter case, the degree of intercalation must be limited, due to the conserved topology of the connections. Importantly, this association occurs at a precise site with respect to the ribosome, even though the additional component does not make direct contact with the ribosome. Thus, the connections may guide the assembly of an asymmetric channel, which then provides a unique binding site for the additional membrane component. At this threshold, the native channel is less hollow than observed previously. However, an internal chamber is positioned adjacent to the C2 connection (Figure 7(a) and (b)). There are two entrances to the chamber from the cytoplasm. One entrance (marked CE in Figure 7(a)) is nearly aligned with the nascent chain exit tunnel (Figure 7(c)). Remarkably, the pore does not form a continuous conduit across the entire channel. However, at higher thresholds the inner chamber is enlarged and a dimple on the lumenal surface of the channel begins to form a small pore (not shown). Intriguingly, the dimple seen in Figure 2(d) is in line with the tunnel of the large ribosomal subunit (Figure 7(c)). Thus, the nascent chain would have to undergo only a slight bend as it traverses the channel (see the dots in Figure 7(c)). The pore within the channel has a cup-like crosssection (Figure 7(d)),16,17 but this profile occurs only on the long axis of the channel. In fact, the channel has a cytoplasmic cover, formed, in part, by the strongest connection (C2). At the threshold used in previously published maps, the C2 connection was left hanging and disconnected above the channel, which clearly is not reasonable.17 Hence, we suggest that the current threshold may portray the channel and the connections more accurately. This is supported by the fact that the current threshold provides a good fit of the atomic models

880

of the ribosomal subunits within the map. Thus, the lack of a continuous pore that spans the channel may be due to the non-translocating state of the complex or reflect an inability to visualize the pore at the current resolution. Recent structures of the yeast channel also revealed a solid channel without a continuous pore.14 Our structure suggests that the narrowest region of the trans˚ in location pore may be smaller than , 10– 12 A ˚ on the diameter, and span a distance of , 15 A lumenal side of the channel (Figure 7(d)).

Discussion We have determined the structure of a native, ˚ mammalian ribosome – channel complex at 17 A resolution. The improved resolution allows a fit with atomic models derived from prokaryotic ribosomes. In addition, the new structure gives a better view of the ribosome – channel junction and allows ˚ structure of the yeast a comparison with a 15 A ribosome – channel complex, which contains purified Sec61p.14 These results have direct implications for the function of mammalian ribosomes and for models of protein translocation through the ER channel. The mammalian ribosome The mammalian ribosome is much larger than its prokaryotic counterparts. Even so, we found an excellent fit between the mammalian ribosome and X-ray structures of the prokaryotic ribosomal subunits for regions involved in translation, subunit –subunit interactions and channel docking. This, in turn, allowed us to identify candidate regions on the large subunit that may participate in forming connections to the ER channel. We suggest that the L1 stalk may function as a gate to release the E-site tRNA from the ribosome. Finally, six expansion segments were localized on the large subunit and, when combined with previous work,11,12,14,34 our structure provides insight into possible roles for these rRNA components. We find that treatment of naturally programmed ribosomes with puromycin results in a putative tRNA being retained in the E-site, adjacent to the L1 stalk. A comparison of our EM map with the ˚ structure of the T. thermophilus ribosome6 5.5 A suggests that exit sites in prokaryotic and eukaryotic ribosomes are similar. However, the L1 stalk in a non-programmed yeast ribosome adopts a different conformation in which the stalk is bent away from the E-site. Thus, the L1 stalk may function as an exit gate for tRNAs in eukaryotes by pivoting about a point at the base of H76.35,40 A previous 3D map of naturally programmed, ribosome – channel complexes did not reveal a tRNA in either the P- or E-site.17 However, a mixed population of nascent chains should have been present in these complexes. In contrast, a P-site tRNA was well preserved in the structure of

Structure of the Native Ribosome –Channel Complex

a programmed, yeast ribosome – channel complex.14 The experimental factors that govern the stability of P- and E-site tRNAs within the ribosome are not well understood. However, the P- and E-site tRNAs each make a number of defined contacts within the eukaryotic ribosome14 (and this work;) and the nascent chain may help stabilize the P-site tRNA. Importantly, our data suggest that the puromycin reaction may lock the ribosome into a defined conformation with a tRNA in the E-site. This should prove useful for higherresolution studies of the ribosome–channel complex. Although the conformation of the ribosome with an E-site tRNA may not be strictly physiological, it appears to have many hallmarks of the active ribosome. For example, we find six major bridges that are conserved between prokaryotic and eukaryotic ribosomes, including B1a and B1b/c, which are located between the head of the small subunit and the central protuberance.9,12 In addition, the path for mRNA across the neck of the small subunit is flanked on either side by tunnels, which would guide it through the decoding center. As found in prokaryotic ribosomes, the mRNA entrance tunnel is formed by the head of the small subunit, which rests on H18,6,42 while the putative mRNA exit tunnel appears to be formed by homologs of proteins S7 (RS5) and S11 (RS14).9,42 A longstanding mystery concerns the function of the expansion segments, which are responsible for the large increase in mass of the mammalian ribosome (, 0.4 MDa), relative to its yeast counterpart. There are six expansion segments in the large subunit that account for most of this increase. These segments form helical protrusions on the surface of the ribosome, and may have different roles, depending upon their location. Moreover, at least half of the rRNA in the expansion segments appears to be disordered. The expansion segments appear to be structurally conserved, as they are similar in all vertebrate ribosomes that have been studied,11,12,34 implying that they play important roles. For example, ES30 may cause the L1 stalk to reorient laterally, away from the interface with the small subunit of the mammalian ribosome. This could alter the functional dynamics of the L1 stalk and affect its role as a lateral gate for E-site tRNAs. Intriguingly, ES30 in yeast is much smaller than in mammals (6 nt versus , 55 nt). In addition, an elongation factor (EF3) that functions as an NTPase and an exit site factor has been described in yeast,43 but no homolog has been identified in higher eukaryotes. Perhaps there is a relationship between the size of ES30 and the necessity for a separate EF3, as both may influence the function of the L1 stalk as an exit gate. We found that the additional nucleotides in vertebrate ES27 (relative to yeast) are rather mobile, even though the proximal portion of ES27 adopts a defined orientation when a channel is bound to the ribosome (this work).14 In previous work, the ES27 spine was oriented such that its path crosses the TE domain on the isolated ribosome.14 Hence,

881

Structure of the Native Ribosome –Channel Complex

it was suggested that ES27 may reorient during channel docking to form the ribosome – channel complex. This is intriguing, because SRP with a bound signal sequence binds at the TE. SRP then undergoes a rearrangement when it contacts the SRP receptor.44 Thus, channel docking may require a number of changes at the TE domain that serve to minimize steric clashes, while allowing the signal sequence to be inserted within the channel.14,44 Expansion segments 7, 15 and 39 are located on the back of the large subunit and together account for the two fingers located below L7/L12 and two large spines. The ordered helical regions of these expansion segments nearly enclose the back of the subunit, where they may act as braces that interconnect different domains of the 28 S rRNA.9 In addition, the large disordered regions of the expansion segments may play as yet undetermined roles in ribosome biogenesis or targeting in the mammalian cell. The ribosome– channel junction Available structural data suggests that the ribosome – channel junction is highly conserved in all eukaryotes.14,16,17 The conservation of the ribosome – channel junction is consistent with recent data which show that prokaryotic ribosomes bind to the mammalian channel and conversely, that mammalian ribosomes bind to the prokaryotic channel.22 Indeed, the similarity of the channeldocking surface on the large subunit suggests that the mechanism of co-translational translocation may be similar in all organisms.12,16 Thus, the wealth of biochemical and structural information on the prokaryotic ribosome can now be used to illuminate this process. In previous studies, the ribosome – channel junction was found to be rather open, and this feature is maintained at higher resolution (this work).16 ˚ wide in regions Thus, the gap is roughly 15 –20 A outside the connections. This would be sufficient to allow a nascent chain to exit laterally from within the ribosome –channel junction into the cytosol. Moreover, the most likely path for the egress of the nascent chain is in a direction away from the picket fence formed by the connections C1, C2 and C4. The gap between the ribosome and channel would allow the cytoplasmic domain of a membrane protein to emerge from the junction, without the necessity of a large conformational change. Major interaction sites between the Sec61p complex and the ribosome are provided by loops 6 and 8 of Sec61a.23 On the basis of a comparison with the X-ray structure of the H. marismortui large ribosomal subunit, we identified components in the ribosome that may participate in forming the connections. These include rRNA (H59, H50/H53, H24 and H7) and vertebrate homologs of proteins L19e, L23, L24 and L29. The similarity of the yeast and vertebrate ribosome – channel complexes is underscored by the identification of similar candidate regions that may help

form the connections.16 Previously, it was suggested that rRNA segments would play a decisive role in forming the connections, since the binding of non-translating ribosomes was not affected by treatment with protease, and ribosomal RNA alone possessed binding characteristics similar to those of intact ribosomes.22 However, the current data suggest that proteins of the large subunit may play a role in channel binding (this work).16 The arrangement of the three major connections is striking. They form a picket fence that essentially divides the native channel into two parts. One part contains the acentrically placed translocation conduit, while the other interacts with an additional membrane protein containing a large lumenal domain. The connections are present at the same position in channels formed by the purified Sec61p complex,17 which indicates that they do not involve any other membrane proteins. A recent 3D structure of the SecYEG complex45 indicates ˚ from the that the cytoplasmic loops extend , 15 A membrane surface. Thus, these loops may make a substantial contribution to the formation of the connections. The distance between connections C1 and C4, as well as the topology of the translocation pore, suggests that two different copies of Sec61a are involved in forming these interactions. In addition, C3 located near the TE may connect the ribosome to a third Sec61a molecule and help stabilize the junction. Thus, the number of connections is consistent with the idea that three or possibly four Sec61p complexes may form the channel.15 – 17 The data suggest that each Sec61p complex may make one connection, or in one case perhaps two connections, with the ribosome (e.g. C1 and C2). Hence, each Sec61p complex must reside in a distinct environment, dictated by its association with the large ribosomal subunit. Importantly, a similar pattern of connections is observed, irrespective of whether a nascent chain is present (this work).16 This suggests that conformational changes associated with protein translocation may not alter greatly the connections of the channel with the ribosome.

The native channel The structure of the native, mammalian channel comprises the channel proper, formed by the Sec61p complex and an extension that is formed by either the TRAP or OST complexes.17 Subunit boundaries between heterotrimeric Sec61p complexes are not revealed at the current resolution, but the size of the native, mammalian channel is consistent with there being three to four copies of the Sec61p complex present. This estimate is based, in part, on the size of the homologous SecYEG channel, whose structure was determined by electron crystallography.45 Molecular modeling with rhodopsin suggests that there are three copies of Sec61p in the yeast ribosome –channel complex.16 However, our current data do not

882

allow us to differentiate between three or four Sec61p complexes per channel. The native ER channel is asymmetric, unlike many other channels that possess perfect or nearly perfect rotational symmetry. Given that the Sec61p complexes are in different environments, we propose that ribosome binding may trigger a stepwise assembly of the channel. In this model, the ribosome would first bind to one or two Sec61p heterotrimers, to form the major connections. This in turn, would create specific sites for the association of additional Sec61p molecules and a membrane protein such as TRAP or OST. Other membrane proteins, such as TRAM, signal sequence peptidase, and RAMP4 may then associate at the periphery of the Sec61p channel.18 An additional result from our structure concerns the apparent lack of a continuous translocation pore. We believe that the threshold used in this study represents the ribosome, the channel and the junction between them accurately. This is important, as the lower threshold plays a key role in delineating the channel with a more solid morphology. In previous work, the channel contained a cup-shaped pore, due in part to the choice of threshold, partial channel occupancy and possible effects of TX-100.14,17 More recently, it was shown that the purified yeast Sec61p channel is solid with no apparent pore, at resolutions of , 15 and ˚ for translocating and non-translocating com19 A plexes, respectively.16 In contrast, we find that the ˚ between the pore is closed for a distance of , 15 A bottom of an internal chamber and a dimple on the lumenal surface of the channel. At the current resolution, we cannot exclude the possibility that the pore is in fact open, but it must have a diameter ˚ . This raises the possibility of less than , 10 –12 A that only a subset of the Sec61p complexes within the channel may form the narrowest region of the translocation pore. Indeed, a dimeric complex of SecYEG can be isolated with a nascent chain,46 yet the native ER channel likely accommodates three to four Sec61p complexes. The internal chamber within the channel has two cytoplasmic entrances (Figure 7(a)), with one located underneath the nascent chain exit tunnel (CE, Figure 7(a) and (c)). This unusual topology results, in part, from the interaction of connection C2 with the center of the channel. The role of the chamber remains to be elucidated, but it may be well defined in our map due to the additional membrane protein that is present in the native mammalian channel. Accumulating evidence suggests that the pore diameter may not change much when a nascent chain is present. First, previous structures of ribosome – channel complexes suggest that a similar pore size is present with or without a nascent chain.17 Second, the structure of a yeast ribosome – channel complex with a nascent chain revealed a rather solid channel with no discernible translocation pore. This property was similar in the control channel without a nascent chain.16 A pore size of ˚ was postulated for translocating channels 40 –60 A

Structure of the Native Ribosome –Channel Complex

on the basis of fluorescence quenching experiments.27 This now appears unlikely, because the required conformational changes would probably affect the connections between the channel and the ribosome. The small size of the pore and the absence of a large internal chamber suggest that polypeptide chains may not form significant tertiary structure within the channel. This is consistent with experiments in which small domains of translocating polypeptide chains fold only after they emerge into the ER lumen.47 The position of the proposed translocation conduit may contribute to its role in membrane protein integration, as TMs could slip between adjacent Sec61p complexes into the lipid phase. The geometry of the channel may facilitate sampling of the lipid environment by the nascent chain, because the pore would retain only small amounts of water. Finally, the position of the pore may allow nascent chains to interact with TRAM48,49 and provide access for other translocon components such as signal peptidase. General considerations for cotranslational translocation All the structural data point to the existence of a highly conserved ribosome – channel junction.14,16,17 Although the highest-resolution structures have been determined from soluble complexes, a similar junction is observed when ribosome –channel complexes remain associated with membranes.17 In addition, this work shows that the ribosome – channel junction is rather open at all reasonable thresholds and should allow the passage of small ˚ .14,16,17 In premolecules up to a diameter of , 15 A vious work, it was proposed that a seal to small molecules is provided by the ribosome – Sec61p junction at the cytoplasmic side of the channel, and by the binding of BiP at its lumenal side.24,25,27,50 Our data suggest an alternate possibility in which the passage of small molecules would be restricted within the channel itself,16 rather than at the junction. This is based on the observation that the pore is narrow or closed in our structure without a nascent chain. When a nascent chain is present, the pore may open or become wider, but the chain would occupy much of the internal space. In addition, a narrow pore would help to maintain the different environments on the two sides of the ER membrane. Hence, there would be no need to invoke a tight seal at the ribosome – membrane junction formed by other translocon components. The open topology of the ribosome – channel junction observed in these structures is consistent with experiments that show that a nascent chain with 63 residues may be cleaved by protease.51 The new structure suggests how a nascent polypeptide chain may travel from the peptidyltransferase site to the ER lumen. Inside the ribosome, the nascent chain would move down the tunnel, ignoring the smaller lateral branches. Once it has

883

Structure of the Native Ribosome –Channel Complex

reached the TE site, the chain would emerge into the cytosol and, if a hydrophobic signal sequence were present, it would be sequestered by SRP. During SRP-SR-mediated docking of the ribosome to the ER channel, the signal sequence and part of the nascent chain would be bound as a loop at a specific site in the wall of the narrow pore.52 The portion of the nascent chain following the signal sequence would then span the gap in the ribosome – membrane junction. Further chain elongation would allow the polypeptide to emerge into the ER lumen. The conservation of connections in both translocating and non-translocating complexes further suggests that the nascent chain may not induce a major conformational change in the channel (this work).16 Thus, the polypeptide would be located inside a rather narrow tunnel or pore throughout most of its journey. The nascent chain within the tunnel could adopt an extended or a-helical structure, but folding would be prohibited until it emerges into the lumenal space.47,53 Membrane proteins would be synthesized in a similar manner, until a TM domain emerges from the ribosome. The structure of the ribosome –channel junction suggests that the TM domain may exit laterally between Sec61p heterotrimers into the lipid phase, at a site pointing away from the three major connections. Once the TM domain has moved into the lipid phase, the polypeptide segment following the TM domain would be dragged through the gap between the ribosome and the channel, and further chain elongation would allow the nascent chain to proceed into the cytosol.

Experimental Procedures Preparation of native ribosome –channel complexes and electron cryo-microscopy Canine ribosomes and stripped rough microsomal membranes were prepared by incubating rough microsomes with puromycin and high salt. In brief, the microsomes were incubated in 50 mM Hepes –KOH (pH 7.5), 100 mM sucrose, 150 mM potassium acetate, 5 mM magnesium acetate, 1.5 mM DTT, and protease inhibitors with 0.1 mM GTP and 1.5 mM puromycin for one hour at 0 8C, followed by ten minutes at 37 8C. The ribosomestripped membranes were floated in a cesium chloride discontinuous gradient, and the ribosomes were collected from the pellet fraction, resuspended in 50 mM Hepes– KOH (pH 7.5), 250 mM sucrose, 2 mM DTT and frozen. Fifty equivalents of ribosome-stripped membranes (PK-RMs) were incubated with , 125 mg of purified canine ribosomes on ice for ten minutes followed by 15 minutes at 27 8C. The membranes were floated in a 2 M sucrose gradient containing 150 mM potassium acetate, and then solubilized in 30 mM Hepes– KOH (pH 7.5), 500 mM potassium acetate, 10 mM magnesium acetate, 1.5% (w/v) digitonin for 30 minutes on ice. After removal of aggregates by centrifugation for five minutes in a microfuge, ribosomes and the associated channel complex were sedimented by centrifugation for 20 minutes at 100,000 rpm in a TLA100 Beckman tabletop ultracentrifuge. The pellet was resuspended in 30 mM Hepes– KOH (pH 7.5), 50 mM potassium acetate,

10 mM magnesium acetate, 1.5% digitonin. To prepare samples for electron cryo-microscopy, freshly prepared specimens were frozen on Cu support grids with a continuous carbon film.17 Specimens were loaded onto a liquid nitrogen-cooled, Oxford cryo-holder, and inserted into an FEI-Philips Tecnai F20 electron microscope. Images were recorded at 200 kV and a magnification of 50,000 £ on Kodak SO163, and were developed for 12 minutes in concentrated D19 to give a final absorbance of ,0.8 – 1 at 540 nm.

Image processing Micrographs were screened using optical diffraction and those exhibiting astigmatism, drift or defocus gradients were eliminated. The remaining micrographs were scanned with a 7 mm raster using an eight bit Zeiss ˚ ) and converted SCAI scanner, binned 3 £ 3 (pixel size 4.2 A to SPIDER format.12 Individual ribosome–channel complexes were windowed into 128 £ 128 boxes using a crosscorrelation procedure12 and bad particles removed. A total of 128 micrographs yielded ,26,000 particles suitable for processing. Defocus parameters were determined from the summed power spectra of 180 non-overlapping areas from each scanned micrograph. Individual micrographs were segregated into nine defocus groups (from 1.1 mm to 3.4 mm underfocus) and aligned using Radon methods12,54 with our previous rabbit ribosome structure as the first reference. Three alignment cycles were performed over all Euler space using an updated reference, which was generated using SPIDER’s Fourier inversion method.55 A final reconstruction from each of the nine groups was then generated using the conjugate gradient method in SPIDER. The individual reconstructions were corrected for the global defocus effects, weighted differentially to reflect their contribution to the complete data set and combined. The particles were then re-segregated into 128 groups on the basis of the original micrographs and aligned to the CTF-corrected 3D reconstruction using Radon methods. The CTF-corrected, 3D reference volume was modulated to simulate the effect of the defocus present in each group during subsequent alignment. The alignment parameters were used to generate individual group reconstructions using an improved 3D Radon inversion method.56 After processing all groups, a new global 3D reconstruction was generated by correcting each group 3D reconstruction for defocus effects, weighting each group differentially on the basis of the number of particles, and then combining the group 3D reconstructions. We repeated this cycle of alignment and calculation of a global, CTF-corrected 3D reconstruction four times. The resolution range in both the 3D reference and the individual 2D images were increased after each iteration, and the size of the angular search step was decreased. For example, during the final Radon alignment process, the 3D reference was Fermi low-pass fil˚ resolution, the 2D images themselves tered to 1/12 A ˚ resolution, and the were filtered similarly to 1/10 A angular search was performed at 0.58 intervals. Reconstructions were generated for each of the 128 groups using a number of different algorithms. General weighted back-projection54 is practical given the small size of the individual image sets, and is faster in practice than iterative (BP RP) and conjugate gradient (BP CG) methods implemented in SPIDER. We also used the extremely fast 3D Radon inversion procedure. FSC curves between the various reconstructions indicated that the

884

Structure of the Native Ribosome –Channel Complex

FSC0.5 value for half data sets from each reconstruction ˚ and 1/17.5 A ˚ , with the conjugate varies between 1/22 A gradient method giving the highest resolution. If the 5 £ noise level is chosen as the resolution criterion, each of the reconstruction methods yields 3D maps that are internally ˚ and 1/12 A ˚ resolution. self-consistent to between 1/17 A We did not compensate the final 3D maps for effects of the envelope function of the electron microscope. As a further means to evaluate the different reconstruction methods and the resolution of this data set, the entire collection of 2D images were individually CTF-corrected and subjected to iterative cycles of Radon alignment against a CTF-corrected 3D volume. The large number of images in this single group precluded the use of general weighted back-projection, but 3D reconstructions were generated using the other methods mentioned above. In these cases, the FSCs were similar to those obtained using alignments that did not use CTF-corrected single particle images. The 3D SSNR method32 gave an estimate of ˚ for the resolution of this data set. ,16.3 A The final threshold for the 3D map was chosen on the basis of two criteria. First, the fit between the atomic structures of the docked ribosomal subunits must be reasonable. Second, the channel must be connected to the large ribosomal subunit and should not be fragmented. Using these criteria, we selected a threshold that gives a ribosomal volume that is , 1.3 to 1.4 times larger than that expected, based on a simple calculation using the molecular mass and the partial specific volumes of rRNA and protein. The Figures were made using WEB,55 O,36 GIMP† and Adobe Photoshop.

functionally activated small ribosomal subunit at ˚ resolution. Cell, 102, 615– 623. 3.3 A Yusupov, M. M., Yusupova, G. Zh., Baucom, A., Lieberman, K., Earnest, T. N., Cate, J. H. D. & Noller, H. F. (2001). Crystal structure of the ribosome at ˚ resolution. Science, 292, 883– 896. 5.5 A Green, R. & Noller, H. F. (1997). Ribosomes and translation. Annu. Rev. Biochem. 66, 679– 716. Ramakrishnan, V. (2002). Ribosome structure and the mechanism of translation. Cell, 108, 557– 572. Spahn, C. M., Beckmann, R., Eswar, N., Penczek, P. A., Sali, A., Blobel, G. & Frank, J. (2001). Structure of the 80 S ribosome from Saccharomyces cerevisiaetRNA-ribosome and subunit – subunit interactions. Cell, 107, 373– 386. Gerbi, S. A. (1996). Expansion segments: regions of variable size that interrupt the universal core secondary structure of ribosomal RNA. In Ribosomal RNA, Structure, Evolution, Processing, and Function in Protein Biosynthesis (Zimmermann, R. A. & Dahlberg, A. E., eds), pp. 71 – 87, CRC Press, New York. Dube, P., Bacher, G., Stark, H., Mueller, F., Zemlin, F., van Heel, M. & Brimacombe, R. (1998). Correlation of the expansion segments in mammalian rRNA with the fine structure of the 80 S ribosome: a cryoelectron microscopic reconstruction of the rabbit ˚ resolution. J. Mol. Biol. reticulocyte ribosome at 21 A 279, 403–421. Morgan, D. G., Me´ne´tret, J. F., Rademacher, M., Neuhof, A., Akey, I. V., Rapoport, T. A. & Akey, C. W. (2000). A comparison of the yeast and the rabbit 80 S ribosome reveals the topology of the nascent chain exit tunnel, inter-subunit bridges and mammalian rRNA expansion segments. J. Mol. Biol. 301, 301– 321. Sweeney, R., Chen, L. & Yao, M. C. (1994). An rRNA variable region has an evolutionarily conserved essential role despite sequence divergence. Mol. Cell. Biol. 14, 4203– 4215. Beckmann, R., Spahn, C. M., Eswar, N., Helmers, J., Penczek, P. A., Sali, A. et al. (2001). Architecture of the protein-conducting channel associated with the translating 80 S ribosome. Cell, 107, 361– 372. Hanein, D., Matlack, K. E. S., Jungnickel, B., Plath, K., Kalies, K.-U., Miller, K. R. et al. (1996). Oligomeric rings of the Sec61p complex induced by ligands required for protein translocation. Cell, 87, 721– 732. Beckmann, R., Bubeck, D., Grassucci, R., Penczek, P., Verschoor, A., Blobel, G. & Frank, J. (1997). Alignment of conduits for the nascent polypeptide chain in the ribosome– Sec61 complex. Science, 19, 2123– 2126. Me´ne´tret, J. F., Neuhof, A., Morgan, D. G., Plath, K., Radermacher, M., Rapoport, T. A. & Akey, C. W. (2000). The structure of ribosome – channel complexes engaged in protein translocation. Mol. Cell, 6, 1219– 1232. Rapoport, T. A., Jungnickel, B. & Kutay, U. (1996). Protein transport across the eukaryotic endoplasmic reticulumn and bacterial inner membranes. Annu. Rev. Biochem. 65, 271– 303. Meyer, T. H., Me´ne´tret, J. F., Breitling, R., Miller, K. R., Akey, C. W. & 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. Go¨rlich, D., Prehn, S., Hartmann, E., Kalies, K. U. & Rapoport, T. A. (1992). A mammalian homolog of Sec61p and SecYp is associated with ribosomes and

6.

7. 8. 9.

10.

11.

12.

13.

Acknowledgements We thank T. Walz for the use of his Tecnai F20 electron microscope and Y. Cheng for technical help. D.G.M. was supported by the HHMI. Both the T.A.R. and C.W.A. laboratories are supported by NIH grants, and T.A.R. is an investigator of the Howard Hughes Medical Institute.

14.

15.

16.

References 1. Matlack, K. E. S., Mothes, W. & Rapoport, T. A. (1998). Protein translocation-tunnel vision. Cell, 92, 381– 390. 2. Johnson, A. E. & Waes, M. A. (1999). The translocon: a dynamic gateway at the ER membrane. Annu. Rev. Cell Dev. Biol. 15, 799– 842. 3. Ban, N., Nissen, P., Hansen, J., Moore, P. B. & Steitz, T. A. (2000). The complete atomic structure of the ˚ resolution. Science, large ribosomal subunit at 2.4 A 289, 905– 920. 4. Wimberly, B., Brodersen, D. E., Clemons, W. M., Jr, Morgan-Warren, R. J., Carter, A. P., von Rhein, C. et al. (2000). Structure of the 30 S ribosomal subunit. Nature, 407, 327– 339. 5. Schluenzen, F., Tocilj, A., Zarivach, R., Harms, J., Gluhmann, M., Janeli, D. et al. (2000). Structure of † www.gimp.org

17.

18.

19.

20.

Structure of the Native Ribosome –Channel Complex

21.

22.

23.

24.

25.

26.

27.

28.

29. 30.

31. 32. 33. 34.

35.

36.

nascent polypeptides during translocation. Cell, 71, 489–503. Kalies, K. U., Go¨rlich, D. & Rapoport, T. A. (1994). Binding of ribosomes to the rough endoplasmic reticulum mediated by the Sec61p-complex. J. Cell Biol. 126, 925– 934. Prinz, A., Behrens, C., Rapoport, T. A., Hartmann, E. & Kalies, K.-U. (2000). Evolutionarily conserved binding of ribosomes to the translocation channel via the large ribosomal RNA. EMBO J. 19, 1900– 1906. Raden, D., Song, W. & Gilmore, R. (2000). Role of the cytoplasmic segments of Sec61alpha in the ribosomebinding and translocation-promoting activities of the Sec61 complex. J. Cell Biol. 150, 53 – 64. Crowley, K. S., Reinhart, G. D. & Johnson, A. E. (1993). The signal sequence moves through a ribosomal tunnel into a noncytoplasmic aqueous environment at the ER membrane early in translocation. Cell, 73, 1101– 1115. Haigh, N. G. & Johnson, A. E. (2002). A new role for BiP: closing the aqueous translocon pore during protein integration into the ER membrane. J. Cell Biol. 156, 261– 270. Liao, S., Lin, J., Do, H. & Johnson, A. E. (1997). Both lumenal and cytosolic gating of the aqueous ER translocon pore are regulated from inside the ribosome during membrane protein integration. Cell, 90, 31 – 41. Hamman, B. D., Chen, J. C., Johnson, E. E. & Johnson, A. E. (1997). The aqueous pore through the ˚ during translocon has a diameter of 40 – 60 A cotranslational protein translocation at the ER membrane. Cell, 89, 535– 544. Go¨rlich, D. & Rapoport, T. A. (1993). Protein translocation into proteoliposomes reconstituted from purified components of the endoplasmic reticulum membrane. Cell, 75, 615– 630. Knauer, R. & Lehle, L. (1999). The oligosaccharyltransferase complex from yeast. Biochim. Biophys. Acta, 1426, 259– 273. Schro¨der, K., Martoglio, B., Hofmann, M., Holscher, C., Hartmann, E., Prehn, S. et al. (1999). Control of glycosylation of MHC class II-associated invariant chain by translocon-associated RAMP4. EMBO J. 18, 4804–4815. Frank, J. & Agrawal, R. K. (2000). A rachet-like intersubunit reorganization of the ribosome during translocation. Nature, 406, 318– 322. Penczek, P. (2002). Three-dimensional spectral signal-to-noise ratio for a class of reconstruction algorithms. J. Struct. Biol. 138, 1 – 34. Verschoor, A., Warner, J. R., Srivastava, S., Grassucci, R. A. & Frank, J. (1998). Three-dimensional structure of the yeast ribosome. Nucl. Acids Res. 26, 655– 661. Dube, P., Wieske, M., Star, H., Schatz, M., Stahl, J., Zemlin, F. et al. (1998). The 80 S rat liver ribosome at ˚ resolution by electron cryomicroscopy and 25 A angular reconstitution. Structure, 6, 389– 399. Harms, J., Schluenzen, F., Zarivach, R., Bashan, A., Gat, S., Agmon, I. et al. (2001). High resolution structure of the large ribosomal subunit from a mesophilic eubacterium. Cell, 107, 679–688. Jones, T. A., Zou, J.-Y., Cowan, S. W. & Kjeldgaard, M. (1991). Improved methods for building protein models in electron density maps and the location of errors in these models. Acta Crystallog. sect. A, 47, 110 – 119.

885

37. Owen, C. H., Morgan, D. G. & DeRosier, D. J. (1996). Image analysis of helical objects: the Brandeis helical package. J. Struct. Biol. 116, 167– 175. 38. Nissen, P., Hansen, J., Ban, N., Moore, P. B. & Steitz, T. A. (2000). The structural basis of ribosome activity in peptide bond synthesis. Science, 289, 920– 930. 39. Gabashvili, I. S., Gregory, S. T., Valle, M., Grassucci, R., Worbs, M., Wahl, M. C. et al. (2001). The polypeptide tunnel system in the ribosome and its gating in erythromycin resistance mutants of L4 and L22. Mol. Cell, 8, 181– 188. 40. Gomez-Lorenzo, M. G., Spahn, C. M. T., Agrawal, R. K., Grassucci, R. A., Penczek, P., Chakraburtty, K. et al. (2000). Three-dimensional cryoelectron microscopy localization of EF2 in the Saccharomyces ˚ resolution. EMBO cerevisiae 80 S ribosome at 17.5 A J. 19, 2710– 2718. 41. Schnare, M. N., Damberger, S. H., Gray, M. W. & Gutell, R. R. (1996). Comprehensive comparison of structural characteristics in eukaryotic cytoplasmic large subunit (23 S-like) ribosomal RNA. J. Mol. Biol. 256, 701– 719. 42. Yusupova, G. Z., Yusupov, M. M., Cate, J. H. & Noller, H. F. (2001). The path of messenger RNA through the ribosome. Cell, 106, 233– 241. 43. Triana-Alonso, F. J., Chakraburtty, K. & Nierhaus, K. H. (1995). The elongation factor 3 unique in higher fungi and essential for protein biosynthesis is an E site factor. J. Biol. Chem. 270, 20473 – 20478. 44. Pool, M. R., Stumm, J., Fulga, T. A., Sinning, I. & Dobberstein, B. (2002). Distinct modes of signal recognition particle interaction with the ribosome. Science, 297, 1345– 1348. 45. Breyton, C., Haase, W., Rapoport, T. A., Kuhlbrandt, W. & Collinson, I. (2002). Three-dimensional structure of SecYEG, the bacterial protein-translocation complex SecYEG. Nature, 418, 662– 664. 46. Bessonneau, P., Besson, V., Collinson, I. & Duong, F. (2002). The SecYEG preprotein translocation channel is a conformationally dynamic and dimeric structure. EMBO J. 21, 995–1003. 47. Kowarik, M., Martoglio, B., Kueng, S. & Helenius, A. (2002). Protein folding during co-translational translocation in the ER. Mol. Cell, 10, 769– 778. 48. Do, H., Falcone, D., Lin, J., Andrews, D. W. & Johnson, A. E. (1996). The cotranslational integration of membrane proteins into the phospholipid bilayer is a multistep process. Cell, 85, 369– 378. 49. Voigt, S., Jungnickel, B., Hartmann, E. & Rapoport, T. A. (1996). Signal-sequence dependent function of the TRAM protein during early phases of protein transport across the ER membrane. J. Cell Biol. 134, 25 – 35. 50. Hamman, B. D., Hendershot, L. M. & Johnson, A. E. (1998). BiP maintains the permeability barrier of the ER membrane by sealing the lumenal end of the translocon pore before and early in translocation. Cell, 92, 747– 758. 51. Jungnickel, B. & Rapoport, T. A. (1995). A posttargeting signal-sequence recognition event in the endoplasmic reticulum membrane. Cell, 82, 261– 270. 52. Plath, K., Mothes, W., Wilkinson, B. M., Stirling, C. J. & Rapoport, T. A. (1998). Signal sequence recognition in posttranslational protein transport across the yeast ER membrane. Cell, 94, 795– 807. 53. Mingarro, I. I., Nilsson, I. I., Whitley, P. & von Heijne, G. (2000). Different conformations of nascent polypeptides during translocatioin across the ER membrane. BMC Cell Biol. 1, 1 – 3.

886

Structure of the Native Ribosome –Channel Complex

54. Radermacher, M. (1994). Three-dimensional reconstitution from Radon projections: orientational alignment via Radon transforms. Ultramicroscopy, 53, 121– 136. 55. Frank, J., Radermacher, M., Penczek, P., Zhu, J., Li, Y., Ladjadj, M. & Leith, A. (1996). SPIDER and WEB: processing and visualization of images in 3D electron

microscopy and related fields. J. Struct. Biol. 116, 190– 199. 56. Lanzavecchia, S., Bellon, P. L. & Radermacher, M. (1999). Fast and accurate three-dimensional reconstruction from projections with random orientations via Radon transforms. J. Struct. Biol. 128, 152– 164.

Edited by W. Baumeister (Received 7 June 2002; received in revised form 25 September 2002; accepted 1 October 2002)