Structural impact of K63 ubiquitin on yeast translocating ribosomes under oxidative stress

Structural Analysis of K63 Ubiquitinated Ribosomes.

To determine the structural impact of K63 ubiquitination of ribosomes, we isolated K63 ubiquitinated ribosomes under oxidative stress and imaged them using single-particle cryo-EM. Incubation of yeast cells with 0.6 mM of H₂O₂ for 30 min is sufficient to induce large accumulation of K63 ubiquitin on free 40S and 60S subunits, 80S ribosomes, and polysomes (Fig. 1A). Moreover, untreated cells have nearly undetectable levels of K63 ubiquitin (10), similar to what is observed for the K63R ubiquitin mutant (Fig. 1A), which contains a mutation that prevents the elongation of K63 ubiquitin chains. After cell disruption, K63 ubiquitinated ribosomes were isolated through pulldown using a K63 ubiquitin trap (10, 11, 15), which consists of a series of ubiquitin binding domains with high affinity for K63 ubiquitin chains (SI Appendix, Fig. S1A). Our 3D classification revealed that free 40S and 60S subunits as well as fully assembled 80S ribosomes were present among the K63 ubiquitinated particles (Fig. 1B and SI Appendix, Figs. S1B and S2). This finding agrees with our polysome profile analysis that shows widespread distribution of K63 ubiquitin on a variety of ribosomal species (Fig. 1A). Free 40S particles were abundant in the sample but did not converge into high-resolution structures likely due to their intrinsic flexibility and the presence of mixed noise particles. In contrast, the fraction of particles identified as 60S (32%) and 80S (40%) were structurally more homogeneous than the 40S, and refinement of the 80S subpopulation resulted in high-resolution reconstructions (Fig. 1B and SI Appendix, Fig. S2). Although ubiquitination of free subunits might be functionally relevant, here, we focused on the intact 80S particles, which are translating ribosomes and constitute the most abundant ribosomal class targeted by K63 ubiquitin under stress.

Our 3D classification analysis generated two main classes containing intact 80S ribosomes (Fig. 1C and SI Appendix, Fig. S2). Inspection of two-dimensional (2D) class averages shows a class of isolated monosomes (80S) and a class of ribosomes that contains additional densities adjacent to the 40S entry and exit mRNA sites (80S+) (Fig. 1C and SI Appendix, Fig. S3A). These additional densities in the 80+ class are consistent with the presence of neighboring ribosomes localized proximally to the main 80S particle in a polysome arrangement. Our 3D reconstruction of this 80S+ conformation showed that the adjacent densities are composed of two distinct parts resembling a 40S subunit associated with a 60S (Fig. 1C and SI Appendix, Fig. S3 A and B). The 80S+ model revealed an arrangement where the 40S subunits face each other, and the 60S subunits localize toward the outside of the high-order structure (Fig. 1C and SI Appendix, Fig. S3B). To confirm that these additional densities represent a polysomal conformation, we isolated yeast polysomes using profiling separation and determined their structures using cryo-ET and subtomogram averaging (SI Appendix, Fig. S3 C–E). The 3D nature of image acquisition by cryo-ET revealed that polysomes can be arranged in the same conformation obtained by the single-particle cryo-EM analysis of K63 ubiquitinated 80S ribosomes. Our cryo-ET analysis of polysomes showed a clear prevalence for the small subunits facing each other, producing a pseudohelical organization (Fig. 1D) that we expected to be observed in the 80S+ class. These findings support the fact that the 80S+ class is, indeed, composed of polysomes. Therefore, the combined cryo-EM/ET studies corroborate that monosomes and polysomes are present in the K63 ubiquitinated pool of ribosomes, confirming our previous analysis that assembled ribosomes were targeted by K63 ubiquitin under oxidative stress (10, 11).

We next showed that 80S particles from the polysome class converged into a 3.5-Å resolution structure of elongating ribosomes (Fig. 2A). The maps we obtained showed high-resolution features, such as density for individual side chains of amino acids and π–π stacking of ribosomal RNA (rRNA) bases. Although it is expected that protein synthesis would be globally repressed in response to stress, translation must still occur to produce stress-related proteins (3, 4). In this context, cells would exhibit different 80S conformation states and intermediates as thousands of ribosomes still progress through the translation cycle (13). To our surprise, K63 ubiquitinated monosomes and polysomes were highly homogenous (SI Appendix, Fig. S4A), sharing several structural features of elongating ribosomes. Among these features, we observed the ratchet 40S rotated conformation and densities for the L1 stalk and for two tRNA molecules in the decoding center (Fig. 2B). When we aligned K63 ubiquitinated ribosomes with previously published rotated and nonrotated structures (16), we found that K63 ubiquitinated molecules are present in the pretranslocation stage of elongation (SI Appendix, Fig. S4 B and C), suggesting that K63 ribosomes occupy a single and well-defined stage of translation. Translocation is a step of the elongation process characterized by a ratchet movement, a 40S–60S rotation that moves the transcript within the decoding center and the nascent polypeptide chain from the A to the P site, allowing ribosomes to accommodate another charged tRNA (13, 17). Our analysis shows that K63 ubiquitinated ribosomes were captured specifically at the pretranslocation stage 2 of translation where ribosomes are rotated and the tRNAs are present in a hybrid position with the first molecule in transition from the A to the P site and the second molecule from the P to the E site (Fig. 2B). This intermediate is formed prior to the binding of eEF2, a eukaryotic elongation factor required for translocation to occur.

To better understand the structural properties of K63 ubiquitinated molecules, we further classified the K63 ubiquitinated polysome particles into three subclasses. Although one of the subclasses (subclass 3) did not achieve high resolution (SI Appendix, Fig. S2A), the quality of the remaining two reconstructions allowed us to detect several structural differences between them. Polysomal subclasses 1 and 2 aligned well with translocating ribosomes (Fig. 2C), were present in the ratcheted state, and had tRNAs in the hybrid position (SI Appendix, Fig. S4D). When aligned to translocating ribosomes (14), the K63 polysome subclass 1 (58%) showed a 4° rotation in the 40S subunit that is not present in subclass 2 (21%) (Fig. 2C), a swivel-like motion that is associated with the translocation step of elongation (18). At low resolution, the three subclasses also showed differences in the position of the L1 stalk with subclass 1 presenting the L1 stalk in its closed position, while the stalk adopted the open conformation in the other two subclasses (Fig. 2D). The L1 stalk is a 60S structural component that interacts with tRNA during translocation and assists the release of tRNA from the E site (19). Furthermore, by comparing the high-resolution polysome subclasses 1 and 2, we observed a minor shift in the position of the hybrid tRNA in the P/A position with the major subclass being slightly more advanced in the translocation process than the minor class (SI Appendix, Fig. S4 D and E). These results corroborate that K63 ubiquitinated polysomes occupy a restricted conformational space within the pretranslocating stage of elongation. Our previous work showed that K63 ubiquitination of ribosomes under stress impairs their capacity to produce proteins and acts postinitiation (11). Building on these previous findings, our new data suggest that K63 ubiquitin might, indeed, act postinitiation by stalling translation of both 80S monosomes and polysomes at the pretranslocation stage 2 of elongation.

K63 Ubiquitin Is Required for Conformational Changes at the Ribosomal Binding Sites for eEF2.

Our published proteomics data showed that K63 ubiquitin modifies several lysine residues proximally to the ribosome decoding center (11). Here, we used the structural location of these modifications to understand how K63 ubiquitinated ribosomes are captured at pretranslocation. The 40S subunit, which is the target of multiple K63 ubiquitin sites (Fig. 3A), is also a highly dynamic subunit, and provides a challenging surface for molecular investigation. Although EM density for K63 ubiquitin chains was not visible in the cryo-EM maps due to expected high flexibility and likely stochasticity of ubiquitination (20), we observed that K63 ubiquitin impacts several ribosomal proteins that interact with eEF2 (Fig. 3A). A suboptimal interaction of K63 ubiquitinated ribosomes and eEF2 could determine ubiquitin’s role in translocation.

To further investigate the role of K63 ubiquitin on eEF2 interacting proteins, we compared our K63 ubiquitinated model to unmodified translocating ribosomes (PDB ID 6GQ1) and verified that K63 ubiquitinated ribosomes show several conformational changes for proteins present at the interface of the 40S and the 60S subunits (Fig. 3B). Compared to unmodified ribosomes, densities for ribosomal proteins uL10 (Rpp0) and uL11 (Rpl12) were only partially detected in our cryo-EM maps, likely due to increased structural flexibility. uL10 is highly ubiquitinated under stress (11) and the presence of flexible ubiquitin chains could explain the dynamics of this protein. uL10 is an acidic protein essential for the formation of the P stalk, a pentameric structure involved in the binding of the elongation factor eEF2 and in the formation of the GTPase associated center (21–23). The C terminus of uL10 interacts with a heterotetramer of P1 and P2 to form the P stalk, while conserved rRNA loops connect uL10 to uL11 (24, 25). The map suggests that the P stalk occupies an open conformation in the presence of K63 ubiquitin (Fig. 3B), which is rarely observed in actively translating ribosomes (SI Appendix, Fig. S4F). Together with uL10, uL11 forms the base that stabilizes the P stalk (26), and our results show that both uL10 and uL11 are highly flexible in the K63 ubiquitinated structure. P1 and P2, the other ribosomal proteins involved in the formation of the P stalk, are intrinsically flexible containing dynamic hinges associated with their C terminus (21, 27), thus, explaining the lack of density for the [P1-P2]₂ structure.

Finally, we also observed higher flexibility for eL40, a 60S protein that makes a distal contact with uL10 and eEF2 (Fig. 3 A and B). As the elongation factor eEF2 binds to the 60S–40S interface, contacts with 40S proteins are also expected in addition to contacts with 60S ribosomal proteins in the P stalk (23). Two proteins in the beak of the 40S subunit, eS12 (Rps12) and eS31 (Rps31), interact with domains II and IV of eEF2 (23), and inspection of these regions in the K63 ubiquitinated ribosomes also revealed increased flexibility in comparison to unmodified translocating ribosomes (Fig. 3B). Similar to uL10, eS12 has been shown to be highly K63 ubiquitinated in response to stress (11), suggesting that this modification impacts the structural disposition of the region. Collectively, our cryo-EM maps for K63 ubiquitinated ribosomes show high destabilization of several eEF2 binding regions, indicating that flexibility of the 40S–60S interface could result in an ineffective binding or prevention of the GTPase activity of this elongation factor. Our structural analysis provides mechanistic insight on how K63 ubiquitin could trap polysomes at elongation under stress by preventing eEF2-mediated translocation.

Although K63 ubiquitinated polysomes were captured at the pretranslocation stage of translation and showed specific structural changes related to the elongation factor eEF2, our data did not allow us to determine whether these modifications were mediated by oxidative stress, or whether K63 ubiquitin induces these changes and prevents the progression of translation past translocation. Thus, to determine the role of K63 ubiquitin in ribosomal conformational changes, we resolved the cryo-EM structures of ribosomes isolated from yeast cells containing the K63R ubiquitin mutation treated with H₂O₂. This lysine to arginine mutation allows cells to ubiquitinate protein targets but prevents the extension of K63 ubiquitin chains (28). In this context, H₂O₂ treatment would induce redox changes without the effect of K63 ubiquitin chains, thus, helping to define the impact of K63 ubiquitin in the ribosome structure. We collected and analyzed cryo-EM images from ribosomes lacking K63 ubiquitin under stress (SI Appendix, Figs. S5 and S6) and obtained a consensus reconstruction resolved at 2.7 Å where several high-resolution features could be visualized for ribosomal proteins and rRNA (SI Appendix, Fig. S6D). Subsequent 3D classification analysis produced one 60S class (28%) and four distinct 80S classes with no distinguishable polysomal population (Fig. 4A and SI Appendix, Fig. S5). Polysomes from the K63R mutant are known to be biologically less stable than the WT (10, 29), which could account for the lack of polysomes in the mutant sample. Among the four 80S classes of the K63R mutant, we observed: (i) ribosomes in the classic state with one tRNA in the P position (36%); (ii) ribosomes in the classic state with tRNAs in the A and P positions (14%); (iii) ribosomes in the rotated state with tRNAs in the A and P positions (11%); and (iv) ribosomes in the pretranslocation 2 stage with tRNAs in the hybrid A/P and P/E positions (11%). The percentages are based on the total number of 80S ribosomes after exclusion of empty 80S and 40S particle populations. Moreover, we confirmed that subdivision of the K63R dataset into structural classes is robust as reclassification of the images resulted in consistent particle assignments and produced the same four conformational states. The analysis also consistently revealed the presence of defined translation states, which were determined by their rotated state and the density of tRNAs within the decoding center. All these classes converged to maps in the ∼4 Å resolution range where several high-order features of 80S ribosomes were also identified (Fig. 4A and SI Appendix, Figs. S5 and S6). The fact that we observed multiple elongation intermediates suggests that K63 ubiquitin and not H₂O₂ treatment induces the accumulation of ribosome particles at the pretranslocation stage. Moreover, the pretranslocation stage 2 has lower representation in the K63R mutant population under stress (Fig. 4A), indicating that our analyses are not biased toward most abundant or stable ribosomal species. Our data, thus, suggest that K63 ubiquitin promotes the accumulation of assembled ribosomes and polysomes in pretranslocation 2, a well-defined stage of translation elongation.

In comparison, we observed ribosomes from the K63R ubiquitin mutant strain in many stages of translation during oxidative stress, which included a minor pretranslocating class (Fig. 4A). However, it remained unclear whether the changes to the eEF2 binding sites (Fig. 3B) required K63 ubiquitination and could, therefore, prevent proper translocation. We, then, aligned the map for the K63 ubiquitinated ribosomes with the K63R mutant class at pretranslocation (Fig. 4 B–E and SI Appendix, Fig. S7). In agreement with the analysis above (Fig. 3B), pretranslocating ribosomes from yeast bearing the K63R mutation favors the closed conformation of the L1 stalk, which is largely open in the monosomal fraction of the K63 ubiquitinated sample (Fig. 4B). In addition, the structure of the K63R mutant shows a closed conformation of the P stalk (Fig. 4 B–E), corroborating that K63 ubiquitin is required to destabilize the P stalk under oxidative stress. Interestingly, we also identified regions where the ubiquitinated particles would display additional density when compared to the mutant (SI Appendix, Fig. S7A). In one such region, the K63 ubiquitinated structure shows a mRNA bridge of density between the uS3 (Rps3) and the 18S rRNA h16 domains, suggesting that ubiquitin could help stabilize this connection. uS3 is a ribosomal protein targeted by ubiquitin (11), which localizes by the mRNA entry channel, enhances binding of the mRNA to the 40S subunit, and supports initiation accuracy (31). Finally, we also identified the ES6a rRNA segment in the K63 ubiquitinated ribosomes that display different conformations when compared to the K63R mutant (SI Appendix, Fig. S7A), suggesting that K63 ubiquitin could induce a variety of local changes on ribosomal proteins and rRNA under stress. Consistent with our early interpretations, we proposed that the structural changes observed in the K63 ubiquitinated ribosomes are driven by the presence of K63 ubiquitin rather than by indirect effects of oxidative stress. In agreement with our model, the K63R mutant and published translocating ribosomes (14) show the same disposition for the L1 stalk and P stalk (SI Appendix, Fig. S7 B and C), which differ from the K63 ubiquitinated particles and reinforces the understanding that these modifications are induced by the presence of K63 ubiquitin chains. Taken together, our structural analyses indicate that K63 ubiquitination of ribosomes impacts a particular stage of the translation cycle by halting translocation due to the induction of local flexibility at the eEF2 binding sites.

Isolation of K63 Ubiquitinated Ribosomes.

The isolation of K63 ubiquitinated ribosomes was conducted based on protocols used in our previous studies (10, 11) with minor modifications. Yeast Saccharomyces cerevisiae cells strain SUB280 (51) were grown to the midlogarithmic phase (optical density₆₀₀ [OD₆₀₀] ∼0.5) in 2.4 L of minimal medium containing 0.67% yeast nitrogen base, 2% dextrose, and required amino acids at 30 °C. Cells were treated with 0.6-mM H₂O₂ for 30 min to induce oxidative stress and accumulation of K63 ubiquitin chains. Yeast cells were disrupted by bead beating in lysis buffer containing 50-mM 2-amino-2-hydroxymethyl-1,3-propanediol (Tris)⋅HCl pH 7.5, 100-mM KCl, 20-mM MgCl₂, 10% sucrose, 20-mM iodoacetamide, a cOmplete mini (ethylenedinitrilo)tetraacetic acid (EDTA)-free protease inhibitor kit (Roche), and 50 nM of FLAG K63-tandem ubiquitin binding entities; LifeSensors, catalog no. UM604). Lysate was cleared by centrifugation and loaded into a column containing M2 anti-FLAG resin (Sigma-Aldrich). The sample was eluted with 100 µg/mL of 3×-FLAG peptide (Sigma-Aldrich) in buffer containing 50-mM Tris⋅HCl pH 7.5, 100-mM KCl, 5-mM MgCl₂, and 10% glycerol. Enrichment for large complexes and sample concentration prior to cryo-EM analysis was performed using Amicon centrifugal filter units with 100-kDa cutoff.

Polysome Analysis.

Polysome profiling was performed as described previously (11). Briefly, cell lysis was performed in 50-mM Tris-acetate pH 7.0 buffer containing 150-mM NaCl, 15-mM MgCl₂, 20-mM iodoacetamide, 0.15-mg/mL cycloheximide, 0.2-mg/mL heparin, and a cOmplete mini EDTA-free protease inhibitor kit (Roche). A total of 400 μg of RNA was sedimented by ultracentrifugation in a 7–47% sucrose gradient for 150 min at 38,000 rpm (Beckman Coulter SW41Ti rotor) at 4 °C. Gradients were fractionated from top to bottom using a Brandel density gradient fractionator system, and data were collected using Peak Chart software. Protein was precipitated with 10% tricarboxylic acid and washed with ice-cold 100% acetone prior to immunoblotting. Immunoblot was performed in poly(vinylidene difluoride) membranes using anti-K63 ubiquitin antibody at 1:6,000 dilution (EMD Millipore, catalog nos. 05–1308, clone apu3); anti-Rps10 at 1:6,000 (Sigma-Aldrich, catalog no. WH0004736M1); and anti-Rps6 at 1:4,000 (Abcam catalog no. ab40820).

Isolation of Yeast Polysomes.

W303 yeast cells were grown in YPD until OD₆₀₀ of ∼0.6 and treated with 0.1-mg/mL cycloheximide for 10 min on ice. Cells were pelleted and resuspended in polysome buffer (20-mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid, pH 7.4, 10-mM MgCl₂, 150-mM KCl, 1-mM dithiothreitol, and 0.1-mg/mL cycloheximide) for lysis. To enrich for polysomes, cleared lysates (13.2 OD₂₆₀ equivalent) were loaded onto 10–50% sucrose gradients prepared in polysome buffer and centrifuged for 90 min at 220,000× g (Beckman Coulter SW41Ti rotor). Gradients were fractionated from top to bottom using a Gradient Master system (Biocomp) while monitoring absorbance at 260 nm. Based on the acquired absorption profile (polysome profile), a gradient fraction expected to contain polysomes with, at least, five ribosomes was used for subsequent cryo-ET analysis.

Isolation of Ribosomes from the K63R Mutant Strain.

Yeast S. cerevisiae cells from the the SUB413 (K63R) (28) strain were grown to midlogarithmic phase (OD₆₀₀ ∼0.5) in 200 mL of minimal medium containing 0.67% yeast nitrogen base, 2% dextrose, and required amino acids at 30 °C. Cells were treated with 0.6-mM H₂O₂ for 30 min to induce oxidative stress. Yeast cells were disrupted by bead beating in lysis buffer containing 50-mM Tris-acetate pH 7.0, 50-mM NaCl, 15-mM MgCl₂, 20-mM iodoacetamide, and a complete EDTA-free protease inhibitor kit (Roche). Lysate was cleared by centrifugation and loaded onto buffered sucrose cushion containing 50% sucrose, 50-mM Tris-acetate pH 7.0, and 15-mM MgCl₂. The sample was pelleted by centrifugation at 70,000 rpm for 2 h at 4 °C in a Beckman Optima Max TL ultracentrifuge using a TLA110 rotor. Ribosome pellets were resuspended in buffer containing 50-mM Tris⋅HCl pH 7.5, 100-mM KCl, 10-mM MgCl₂, and 10% glycerol.

Cryo-EM Sample Preparation and Data Collection.

Single-Particle Cryo-EM Data Processing.

For the K63 ubiquitinated ribosomes dataset, movie alignment was performed in RELION (52), and the contrast transfer function was determined on the motion corrected sum of frames using CTFFIND4.1 (53). Ribosome particles were boxed out using template-free particle picking. Two consecutive rounds of 2D classification were preformed to clean the extracted particles. A total of 393,353 clean particles were used for further processing (see SI Appendix, Fig. S2 for details). An ab initio 3D reference was generated using cryoSPARC (54). The rest of the data processing was performed in RELION. The initial 3D classification was performed using 3× binned particles. The 3D classification into 10 classes identified five distinct conformations as follows: tRNA bound 80S polysome (80S+), tRNA bound mono-80S, and empty 80S, 60S, and 40S. Particle images corresponding to the 80S+ conformation were reextracted without binning and further refined to ∼4-Å resolution followed by contrast transfer function (CTF) refinement and particle polishing, which improved the resolution to 3.5 Å. The empty 80S subclasses were combined, and the corresponding particles reextracted using a binning factor of 2, followed by 3D refinement. Two additional rounds of 3D classification (without alignment) using a soft shape mask were conducted resulting in the identification of ∼53,000 additional particles corresponding to the tRNA-bound mono-80S conformation from the empty 80S class. These particles were, then, combined with the previously identified tRNA-bound mono-80S group (SI Appendix, Fig. S2) and further refined to 3.2-Å resolution after CTF refinement and particle polishing. Empty ribosomes were not included in the final quantification of occupancy percentages.

For the K63R mutant dataset, 232,124 clean particles were identified using the same protocol described above. A high-resolution consensus reconstruction (without 3D classification) was produced using automated particle sorting (55), per-particle drift correction, CTF correction, and data-driven dose weighting (56), resulting in a map at 2.7-Å resolution (SI Appendix, Fig. S6 D and E). Six classes were, then, assigned during 3D classification using 3× binned particles. No polysomal class was observed, and all of the 80S classes were combined and further refined. A 3D classification without alignment was conducted to generate four subclasses. Based on particle numbers and map quality, especially the 40S region, two subclasses were selected to perform further refinement using 2× binned particles. One more round of 3D classification was performed for the minor class, and two subclasses were refined to 3.6-Å (80S with tRNA in P site) and 4.3-Å resolutions (80S in the classic state with tRNAs in the P, A sites), respectively, using unbinned particles. For the major class, two more 3D classification rounds without alignment were performed to sort out the other two subclasses. These were, in turn, refined to 4.4-Å (80S in the rotated state with tRNAs in P, A sites) and 4.2-Å resolutions (80S with tRNAs in A/P, P/E sites), respectively. To confirm the final particle assignments, an independent run of 3D classification was executed which resulted in the same final four conformations with percentage particle overlaps of 95%, 97%, 72%, and 87%, respectively. A summary of all of the cryo-EM data and maps produced in this study are included in SI Appendix, Table S1.

Cryo-ET Sample Preparation and Subtomogram Averaging of Yeast Polysomes.

Polysomes were loaded onto lacey carbon molybdenum grids (Ted Pella), 10-nm colloidal gold fiducials were added, and the sample was vitrified using an FEI Vitrobot (Mark IV). Single-axis tilt series were acquired on a FEI Titan Krios transmission electron microscope (TEM) (Thermo Fisher Scientific) equipped with a K2 Summit direct electron detector (Gatan, Inc., Pleasanton, CA), operated in movie mode with five to seven frames per projection image. The TEM was operated at an acceleration voltage of 300 kV, a nominal defocus of −4 µm, and 53,000× magnification (2.62-Å pixel size). Some 47 tilt series were recorded from −60 to 60° (first half: 20 to −60°; second half: 22 to 60°) with an angular increment of 2° and a total electron dose of 80–100 e⁻/A² using the Serial EM acquisition software (57). Slit width of the Gatan imaging filter was 20 eV, zero loss. No Cs corrector was used. Frames from the K2 direct electron detector were aligned using in-house developed motion correction software based on the algorithm previously published (58). Correction of phase reversals due to the CTF was performed using MATLAB scripts and PyTom (59) on single projections (60). Tilt series were aligned by interactively located gold fiducials and reconstructed (voxel size: 2.1 nm) by weighted backprojection using the TOM-software package (61). Ribosomes were localized by template matching against a subtomogram average of the yeast 80S ribosome (62) and visual inspection of tomogram areas corresponding to peaks of the resulting cross-correlation function using PyTom and the AV3 software package (63). Subtomograms centered at the retained coordinates were individually reconstructed from the weighted aligned projections using AV3 (voxel size: 0.262 nm) and iteratively aligned using PyTom until convergence. A total of 7,125 subtomograms were used for the final polysome structure. For visualization of polysomal arrangements observed in the tomograms, the subtomogram average was replicated in a volume using the positions and orientations obtained by template matching. A model was generated after superposition of the subtomogram average on outer densities, replicating the preferred neighbor configurations five times and placing 80S into extrapolated coordinates and angles (64). The putative pathway of interconnecting polysomal mRNA was interpolated based on the positions of mRNA entry and exit sites of adjacent ribosomes.

Atomic Model Refinement.

Coordinates for the yeast 80S ribosome (PDB ID 6GQ1) (14) were rigidly fit into the density maps of K63 ubiquitinated ribosomes and the K63R ubiquitin mutant using UCSF Chimera X (65), subsequently, refined with coot (66) and subjected to real space refinement in PHENIX (67). Validation statistics for both models are included in SI Appendix, Table S1.