|Home | About | Journals | Submit | Contact Us | Français|
Pre-ribosomal particles evolve in the nucleus through transient interaction with biogenesis factors, before export to the cytoplasm. Here, we report the architecture of the late pre-60S particle purified from Saccharomyces cerevisiae through Arx1, a nuclear export factor with structural homology to methionine aminopeptidases, or its binding partner Alb1. Cryo-electron microscopy reconstruction of the Arx1-particle at 11.9 Å resolution reveals regions of extra densities on the pre-60S particle attributed to associated biogenesis factors, confirming the immature state of the nascent subunit. One of these densities could be unambiguously assigned to Arx1. Immuno-electron microscopy and UV cross-linking localize Arx1 close to the ribosomal exit tunnel in direct contact with ES27, a highly dynamic eukaryotic rRNA expansion segment. The binding of Arx1 at the exit tunnel may position this export factor to prevent premature recruitment of ribosome-associated factors active during translation.
Ribosomes are key components of the cellular infrastructure that catalyze protein synthesis. They are evolutionary conserved throughout all kingdoms of life - bacteria, archaea and eukaryotes. However, eukaryotic organisms have evolved a sophisticated synthesis and assembly pathway for their 60S and 40S ribosomal subunits, which consist of four ribosomal RNA species (25S, 5.8S, 5S, and 18S rRNA, respectively) and about 80 ribosomal proteins (r-proteins). Ribosome assembly involves a myriad of eukaryote-specific biogenesis factors (between 150-200). In addition, nucleocytoplasmic transport through the nuclear pore complexes (NPCs) is necessary because the nuclear envelope separates the nucleoplasm, where most of the ribosomal assembly takes place, from the cytoplasm, where the mature ribosomes function in translation1,2. The active transport channel of the NPC is formed by a meshwork of hydrophobic phenlyalanine-glycine (FG) rich repeat domains of a subgroup of nucleoporins. Transient interaction of transport receptors with these FG motifs allows passage of attached cargo through the transport channel3. Maturing pre-60S subunits are huge, hydrophilic particles of >2.5 MDa and their efficient translocation may therefore pose particular problems and require a special effort.
Although the nuclear export of ribosomal subunits has been analyzed in detail4, little is known about the architecture and composition of the exported pre-60S particles compared to mature 60S subunits. Three-dimensional structures of mature eukaryotic ribosomes were revealed during the last decade by modeling of structures into cryo-electron microscopy (cryo-EM) reconstructions and, recently, by solving crystal structures of complete 80S ribosomes and separate 40S or 60S subunits5-9. Various pre-60S particles have been isolated via associated biogenesis factors as bait proteins and their composition has been determined10. Few ribosomal precursor particles have been analyzed by EM and factors on the surface have been mapped. Examples include analyses of pre-40S particles and localization of associated factors by cryo-EM11,12, and analyses of pre-60S intermediates by negative stain EM with localization of several ribosomal and non-ribosomal proteins by immuno-EM13,14.
Here, we focus on a late pre-60S particle that is associated with the biogenesis factor Arx1, which is located both in the nucleoplasm and cytoplasm. This particle is thought to represent an export intermediate, since it carries several nuclear export factors. One of them is the essential export adaptor Nmd3 that contains a nuclear export signal (NES) recognized by Crm1 (also known as exportin1 or Xpo1), the general nuclear export receptor for NES carrying cargo, in concert with RanGTP15,16. When re-bound to the isolated mature 60S subunit, Nmd3 was found to be located at the interface of the 60S subunit close to the r-protein Rpl10 by cryo-EM17. Another export receptor of the 60S subunit is the Mex67-Mtr2 heterodimer, which has been suggested to be recruited to double-stranded RNA, possibly to the 5S rRNA18,19. In addition, Arx1 has been identified as a third export factor with unusual properties; it has a methionine aminopeptidase fold and can bind FG repeat nucleoporins, thereby mediating translocation through the FG repeat channel of the NPC20,21. Furthermore, Ecm1 was suggested to act in 60S subunit export, since it too can interact with FG repeats of the channel nucleoporins22. Several other factors have also been implicated in nuclear export of the 60S subunit including the HEAT-repeat protein Rrp1223, and Npl324.
We sought to further characterize the Arx1-purified pre-60S particle to gain insight into the process of how the huge pre-60S subunit might translocate through the active transport channel of the NPC. Cryo-EM of the isolated pre-60S particle uncovered structural differences from the mature 60S subunit. Biochemical analysis, negative stain EM combined with immuno-labeling, and RNA-protein cross-linking revealed the position of Arx1 and other pre-60S factors on the nascent 60S subunit. The binding site of Arx1 was further determined on an earlier pre-60S particle (Rix1-particle) that is associated with Rea1 and the Rix1-complex, allowing the comparison of two successive nascent 60S subunits. Altogether, our data indicate that the pre-60S particle that exits the nucleus has not yet adopted its mature topology. A scattered distribution of these nuclear export factors on the surface of this pre-60S subunit could facilitate transport through the NPC.
To gain insight into the architecture of pre-60S particles carrying nuclear export factors, we focused on Arx1. This factor forms a salt-resistant complex with Alb1 (Arx1 little brother)20,25 and is stably associated with late pre-60S particles in the nucleoplasm and cytoplasm26. To determine whether Arx1 and Alb1 are indeed associated with the same pre-60S particles, we affinity-purified various nascent 60S subunits via different TAP-tagged non-ribosomal bait proteins from yeast expressing HA-tagged Alb1. These particles, which range from early nucleolar, intermediate nucleoplasmic to late cytoplasmic stages, contained the expected pre-ribosomal factors as revealed by SDS-PAGE and Coomassie staining (Fig. 1a). Western analysis indicated the presence of the known nuclear export factors Nmd3 and Mex67-Mtr2 in the Arx1-TAP purification. When we analyzed Alb1-HA, we detected it in the same pre-60S particles that also contained Arx1, consistent with the model that Arx1 directly binds to the pre-60S subunit, and that Alb1 is recruited to the particle by Arx120.
These data also suggested that Arx1 and Alb1 are bound to intermediate nucleoplasmic particles associated with Rix1 and Nug1, but not to earlier nucleolar pre-60S particles that carry Ssf1 or Nsa1. In contrast, the export factors Nmd3 and Mex67-Mtr2 are absent from the Rix1-associated particles, but strongly enriched in the Arx1-associated particles (Fig. 1a). To confirm that Arx1 is a bona fide component of the Rix1-associated particle, we generated yeast strains expressing either integrated Rix1-TAP, Sda1-TAP or Lsg1-TAP, together with Arx1-Flag, and performed split-tag affinity purifications. Via the ProtA-tag we affinity-purified the Rix1-TAP, Sda1-TAP and Lsg1-TAP proteins from cell lysates. We then passed the TEV-eluates over an anti-Flag peptide column to select for Arx1-Flag, followed by elution with Flag peptides. The different pre-60S particles associated with Rix1-TAP (nucleoplasmic), Sda1-TAP (nucleoplasmic), and Lsg1-TAP (cytoplasmic) each contained Arx1-Flag with similar stoichiometry to other co-purified biogenesis factors (Fig. 1b). We obtained comparable results when we purified Rix1-TAP, Sda1-TAP, and Lsg1-TAP, respectively, from yeast strains expressing Alb1-Flag (data not shown).
These data support the model that the Arx1-Alb1 heterodimer is recruited to a late stage of the maturing nucleoplasmic pre-60S subunit, but prior to the recruitment of Nmd3 and Mex67-Mtr2. Notably, Arx1-Alb1 are associated with nascent 60S subunits before (Rix1) and after (Lsg1) nuclear export, strongly indicating that they are exported in association with the pre-60S particles. In contrast, other biogenesis factors such as Rix1, Rea1, Rsa4, or Lsg1, which all co-precipitate Arx1, are restricted to more distinct biogenesis intermediates, which are considered either exclusively nucleoplasmic (Rea1, Rsa4) or cytoplasmic (Lsg1).
To visualize the structure of the Arx1-particle, we performed cryo-EM single-particle analysis, using Alb1-TAP affinity-purified pre-60S subunits. Comparison of the cryo-EM reconstruction of the Arx1-particle at 11.9 Å resolution (see Supplementary Fig. 1a for Fourier shell correlation plot) with the reconstruction of a mature 60S subunit revealed substantial additional mass in several areas of the Arx1-particle (Fig. 2a, Supplementary Fig. 1b). A globular density (red) is situated directly in front of the ribosomal exit tunnel and a large bulky density (orange) is located at the common helix of the 3′-end of the 5.8S rRNA and the 5′-end of the 25S rRNA. Also the central protuberance region is surrounded by additional mass forming an elongated shape (yellow) and a triangular density is visible in the center of the intersubunit surface (green). An elongated density (cyan) extends from the stalk base to the E-site, blocking access to all of the tRNA binding sites and the peptidyl transferase center (PTC). Furthermore, a globular shape (blue) is observed at the translation factor binding site and a chain-like density (purple) extends from this position along the surface of the pre-mature subunit towards the exit tunnel.
In negative stain EM combined with single particle analysis, Arx1-TAP and Alb1-TAP affinity-purified particles each classify into three comparable main orientations with characteristic structures, the foot, knob and nose (Supplementary Fig. 2; for composition, see also Fig. 1a). Characterized by these common features, we will refer to these particles collectively as the “Arx1-particle”. The cryo-EM structure of the Arx1-particle was readily reconciled with the negative stain class averages, suggesting that the foot, knob, and part of the nose structure are possibly formed by pre-ribosomal factors (Fig. 2b) or may contain rRNA not finally processed or not yet in its final conformation (see Discussion).
Overall, the structures of the mature 60S subunit that were previously determined by cryo-EM6 and x-ray crystallography27 nicely fit into the electron density observed for the pre-mature 60S subunit in our reconstruction of the Arx1-particle (Fig. 2c and data not shown), indicating that maturation is rather complete for most parts of the subunit. However, several features of the pre-60S particle still differ from the mature subunit (Supplementary Fig. 1b). The most obvious rearrangement is observed in the region of the central protuberance. The 5S rRNA and helix 38 (A-site finger) of the 25S rRNA cannot be clearly recognized at their final positions in the Arx1-particle (Fig. 2c). Also, the late-joining r-protein Rpl10 that is located between the central protuberance and the stalk base in the mature subunit seems not to be present in the pre-60S structure. The density found at the mature location of Rpl10 does not fit the shape of the r-protein. Moreover, the electron density observed in this position appears to represent RNA rather than protein, because it displays typical features of RNA density: good visibility at high contour level and helical twist (Fig. 2c). It is possible that the A-site finger may be rearranged to occupy this space. Another rearranged structural feature of the 60S subunit is helix 69 of the 25S rRNA, located in the center of the intersubunit surface where it is involved in the formation of two intersubunit bridges27. While it points towards the A-site in the mature subunit, it is turned by almost 180° in our reconstruction, now pointing towards the E-site (Fig. 2a, top row). Furthermore, we did not observe any density for the P-stalk in our reconstruction and we found that the stalk base was different from the mature subunit in the region of P0 (Supplementary Fig. 1b). In addition to P1 and P2 that were clearly absent, Rpl12, which forms the base of the stalk together with P0, appeared also to be absent from the nascent 60S subunit.
Taken together, our reconstruction of the Arx1 particle provides the first cryo-EM map of a pre-60S subunit. This structure underscores the immature stage of the nascent subunit and reveals a number of extra electron densities that could correspond to bound ribosome biogenesis factors (see below).
To assign pre-ribosomal proteins to the extra densities observed in our cryo-EM reconstruction of the Arx1-particle, we determined the location of selected HA-tagged r-proteins and ribosome biogenesis factors on Alb1-TAP purified particles. Complementation assays showed the functionality of the HA-tagged proteins (Supplementary Fig. 3a). We used anti-HA antibodies to generate an additional mass in close proximity to the respective HA-tagged proteins, allowing visualization by negative stain EM. In parallel, we analyzed the efficiency of antibody labeling by SDS-PAGE and Coomassie staining (Supplementary Figure 3b). We labeled Rpl3-HA and Rpl5-HA, which were previously localized on the Rix1-particle14, with the anti-HA antibody on the Arx1-particle (Fig. 3a). Rpl5 localized to the region of the nose structure, supporting our conclusion that the nose might represent part of the central protuberance. Immuno-labeling of Rpl8 and Rpl26 on the Arx1-particle allowed comparison of the Arx1-particle to the atomic structure of the mature 60S subunit (Fig. 3a, Supplementary Fig. 3c, d).
We observed the antibody signal for immuno-labeled Arx1-HA in the immediate vicinity of the knob structure (Fig. 3a). We identified the knob structure as the additional density observed directly at the ribosomal exit tunnel in our cryo-EM reconstruction of the Arx1-particle (red density, Fig. 2a, b). Consistently, the crystal structure of the human Arx1 homolog Ebp128,29 fits well into this part of the cryo-EM map (Fig. 3b). The Ebp1 structure leaves parts of the density empty, however, Arx1 is larger than Ebp1 (64 kDa vs. 44 kDa due to loop insertions) and some of the density may also belong to Arx1’s much smaller binding partner Alb1 (19 kDa). Notably, the observed density appears smaller than expected for the complete Arx1-Alb1 heterodimer (83 kDa). Missing density can be explained, however, by possible flexibility of Alb1 or the loop insertions in Arx1 of yeast. These results indicate that the density observed in front of the ribosomal exit tunnel represents Arx1, possibly together with Alb1.
The ribosome biogenesis factor Tif6 prevents premature subunit joining and was found to bind to the intersubunit surface of mature 60S subunits, contacting Rpl23, Rpl24, and the sarcin-ricin rRNA loop9,30. The non-ribosomal density observed at exactly this position in our cryo-EM reconstruction (blue density, Fig. 2a, b) matches Tif6 in size and shape and the crystal structure of Tif631 fits well into the density (Fig. 3c). Consistently, we found the antibody signal for Tif6-HA at the top of the particle (Fig. 3a). Taken together, these data indicate that the density labeled in blue represents Tif6, and confirm its location on the pre-mature 60S subunit.
The immuno-EM signal for Nsa2-HA localized to the top of the Arx1-particle, close to the signal for Rpl5 (Fig. 3a, Supplementary Fig. 3c, d). Nsa2 is a biogenesis factor that is required for processing of internal transcribed spacer 2 (ITS2) and, like Tif6, is already present in earlier particles14,32. Based on our immuno-labeling data, the cyan density stretching from the E-site to the stalk base in the Arx1-particle reconstruction might represent – or at least contain – Nsa2 (Fig. 2a, b and and3a).3a). We also immuno-labeled HA-tagged Nmd3, Mex67, Mtr2, and Ecm1 on Alb1-TAP, but none of these proteins were efficiently labeled by the antibody as evaluated by SDS-PAGE and Coomassie staining (see also Supplementary Figure 3b) and thus were not further analyzed by EM (data not shown).
Taken together, immuno-labeling combined with negative stain EM revealed the relative positions of several biogenesis factors on the surface of the Arx1-particle and – together with the positions of marker r-proteins – allowed their assignment to additional densities observed on the Arx1 pre-60S particle reconstructed by cryo-EM.
To experimentally determine where Arx1 contacts the surface of the pre-60S subunit, we applied the CRAC (UV cross-linking and analysis of cDNA) methodology previously used to successfully identify RNA-protein interactions in ribosomal subunits33-35. We found that Arx1 directly contacts several rRNA elements that cluster on the 60S subunit surface near the exit tunnel. In particular, Arx1 was efficiently cross-linked to helix 59 (expansion segment ES24) of the 25S rRNA as well as to ES27, an expansion of helix 63 of the 25S rRNA (Fig. 4a–c). Helix 59 was reported to contact the Sec61-complex within the ER membrane36, whereas ES27 is a highly dynamic rRNA element with two major conformations, one pointing towards the L1-stalk (in), and the other reaching the peptide exit tunnel (out). ES27 is essential for ribosome biogenesis, possibly involved in pre-rRNA processing or in stabilization of mature rRNA37. In the mature ribosome, ES27 was suggested to coordinate recruitment of non-ribosomal factors (e.g. chaperons) to the exit tunnel, from which the nascent polypeptide chain emerges6,36. Arx1 is likely to contact ES27 in the out position close to the exit tunnel, because a simultaneous interaction with both, helix 59 and ES27, is possible only in this conformation (Fig. 4c). Consistently, analysis of the Arx1 binding site in our cryo-EM reconstruction using the high-resolution crystal structure of the mature 60S subunit showed that Arx1 contacts rRNA helix 59 (Fig. 4d), as well as ES27 in the out position (visualized only at lower contour levels; Fig. 4e). Arx1 also contacts Rpl25 and Rpl35 and binds close to Rpl19 (Fig. 4d). Notably, due to low cross-linking efficiency, we could not determine reproducible cross-linking sites for Alb1. Similarly, we could not detectably cross-link Tif6 (Granneman, unpublished data). In conclusion, the CRAC data for Arx1 are in excellent agreement with the immuno-labeling results and its binding site observed in cryo-EM, supporting the model that Arx1 binds in proximity to the exit tunnel and hence could affect this hallmark structure of the ribosome during biogenesis (see Discussion).
Previous studies analyzed the Rix1-associated particle by negative stain EM and localized several non-ribosomal biogenesis factors (e.g. Rea1, Rix1, Rsa4) as well as r-proteins (e.g. Rpl5, Rpl3) on this nascent 60S subunit by immuno-labeling13,14. Since Arx1 is also present on the Rix1-particle (see above), we wanted to find out where it localizes on this intermediate with respect to the other factors. Hence, we affinity-purified Rix1-TAP from yeast expressing chromosomal Arx1-HA, followed by immuno-labeling using anti-HA antibodies (Supplementary Fig. 4a). We localized Arx1 to the knob-like structure on the top of the Rix1-particle, opposite the long protruding tail that corresponds to the huge AAA+ ATPase Rea1 (Fig. 5a, b; Supplementary Fig. 4b, c). Thus, Arx1, likely together with Alb1, forms the knob-like protrusion that is a distinct structural landmark on the top of the Rix1-particle. This position is relatively close to r-protein Rpl3 and more distant from Rpl5 (Fig. 5b). Together, these data suggest that the knob structure represents the Arx1-Alb1 heterodimer in both the Rix1- and Arx1-particles.
The Arx1-TAP and Alb1-TAP particle preparations should include a subset of Rix1-associated particles, with a typical Rea1 tail (see above). We therefore performed a classification of the Alb1-TAP affinity-purified particles visualized by EM, and analyzed the distribution of the particles among the different classes. This quantitative analysis indicated that ~45% of the Alb1-TAP purified particles belong to the three main orientations (views 1–3) referred to as the “Arx1-particle”, whereas only 12% were classified as “Rix1-particle“ exhibiting the typical Rea1 tail (Supplementary Fig. 7). Interestingly, the overall morphology of “view 2“ Arx1-particles resembles a class of tail-less Rix1-TAP purified particles that was previously generated by ATP-treatment of purified Rix1-particles to release Rea1 and the Rix1-complex (Fig. 5c; see ref. 14). Thus, the Rix1-particle carrying Rea1, Rix1-complex and Rsa4 appears to be the precursor of the Arx1-particle, which lost these latter factors due to the action of Rea114 while recruiting new factors, which include the nuclear export factors Nmd3 and Mex67-Mtr2 (see Discussion).
Here we have analyzed the EM structure of pre-60S particles that contain the Arx1-Alb1 heterodimer. These are relatively late particles of the 60S biogenesis pathway that are eventually exported from the nucleus into the cytoplasm. On the one hand, Arx1-Alb1 is associated with a well-studied nucleoplasmic pre-ribosomal particle that carries the dynein-related AAA+ ATPase Rea1 and the Rix1-complex (Rix1-particle). However, the major pool of Arx1-Alb1 associated pre-60S particles, termed the Arx1-particle, lack Rea1 and Rix1, and represent a biogenesis intermediate that follows the Rix1-particle. The Arx1-particle exhibits characteristic structures such as the foot, knob, and nose, which are prominent features seen both by negative stain and cryo-EM. We show for a few of these structures that they corresponded to additional non-ribosomal densities on the pre-60S particle absent from the mature 60S subunit. Employing immuno-EM, the positions of several ribosomal and non-ribosomal proteins could be mapped. To our knowledge, the Arx1-particle is the first immature 60S subunit precursor for which a cryo-EM structure has been determined.
In our reconstruction, the core of the Arx1-particle is similar to the mature 60S subunit, indicating a late assembly stage. However, several important functional sites of the mature 60S subunit are not yet fully developed on the Arx1-particle. Apparently, the flexible P-stalk is absent from the Arx1-particle. This structure is composed of ribosomal proteins P0 and acidic P1 and P2 on the mature 60S subunit. The P-stalk functions in recruiting translation factors to the GTPase center38. However, the stalk base seems to be under reconstruction in the Arx1-particle. Mrt4, a P0-paralog that requires the Yvh1 factor for its release from the 60S pre-ribosome39,40, as well as Yvh1 and P0 are detected in the Arx1-particle by mass-spectrometry. On the other hand, P1 and P2 are known to be late joining r-proteins41 that likely associate with cytoplasmic 60S subunits following release of Arx1-Alb1.
Moreover, the Arx1-purified pre-60S particle differs from the mature subunit in the region of the central protuberance. Together with the observed extra densities, the immature central protuberance gives rise to the characteristic nose structure of the Arx1-particle, suggesting that final conformational rearrangements of the central protuberance and/or release of ribosome biosynthesis factors from this region may not have yet occurred. The yellow extra density at the central protuberance (Fig. 2a, b) may contain part of the structures that form the central protuberance in the mature ribosome, e.g. 5S rRNA, Rpl5, and Rpl11. Possible ribosome biogenesis factors that could contribute to the yellow density are the Mex67-Mtr2 complex or the GTPase Nug1, both of which were shown to interact with 5S rRNA in vitro18,42. The pre-ribosomal factors responsible for the additional density in the pre-ribosomal structure, other than Tif6 and Arx1-Alb1, have not been unambiguously identified. The localization of the green shape (Fig. 2a, b) – albeit smaller – is similar to that found for MPB-tagged Nmd3 on the intersubunit surface of mature 60S in a previous cryo-EM study17. Interestingly, the purple density stretches far across the pre-ribosomal surface and contacts both Arx1 and the Tif6 density, thus possibly providing a means of direct crosstalk between the distant areas around the stalk base and the exit tunnel. This might enable communication of the state of maturation at the stalk base and the release of Arx1 at the exit tunnel. Likewise, the cyan density might coordinate and communicate the progression of maturation around the stalk base and the tRNA binding sites. Another biogenesis factor Rei1, implicated in release of Arx1-Alb1 from the subunit25,43, might be represented by the purple density that also contacts Arx1. Other candidate proteins responsible for the extra densities are Ecm1 and several GTPases including Nog1, Nug1, Nog2, and Lsg1 (involved in the release of Nmd344).
Notably, all these additional densities on the Arx1-particle were observed at functionally relevant sites of the 60S subunit, blocking access to the ribosomal peptidyl transferase center (PTC), the tRNA binding sites, the stalk base, the intersubunit surface and the exit tunnel. A recent study showed that biogenesis factors present on a late cytoplasmic pre-40S particle block all sites important for translation initiation12. Similarly, the positioning of biogenesis factors, as well as the immature state of important structures on the Arx1-particle are hindering untimely onset of translation. Furthermore, it was reported that final cytoplasmic pre-40S maturation involves a translation-like cycle45. It is possible that also pre-60S subunits engage in such a quality control checkpoint to assure functionality and final rRNA maturation and factor release. Together, these observations illustrate the multitude of control mechanisms that keep the nascent subunit from pre-mature engagement in translation during this advanced phase of maturation.
The prominent foot structure of the Arx1-particle is located above the region where the 3′-end of the 5.8S rRNA and the 5′-end of the 25S rRNA end in a common helix in the mature subunit. Arx1-associated particles mainly contain mature 25S and 5.8S rRNA. However, a small amount of 27SB pre-rRNA (in which the 3′-end of 5.8S and 5′-end of 25S rRNA are linked by ITS2) and some 7S pre-rRNA (in which the 3′-end of 5.8S rRNA is extended) are also recovered. This indicates that C2 cleavage, which generates the 3′-end of the 7S pre-rRNA, has occurred in most Arx1-associated particles, but processing of 7S to the 5.8S rRNA has not yet taken place in all particles26 (data not shown). This is in agreement with the reported cytoplasmic location of the final step in 5.8S rRNA processing46. The foot structure may therefore represent retained factors involved in processing and/or folding of ITS2 or it might protect this region and/or offer a binding platform for processing factors until nuclear export and/or maturation of this area are completed.
The position of the unconventional export receptor Arx1 could be unambiguously assigned to the characteristic knob structure observed on either the Rix1- or Arx1-purified 60S subunit particle. The knob structure is located directly in front of the exit tunnel and the observed binding site of Arx1 is in good agreement with an earlier suggestion that it may bind in close proximity to Rpl2543. The flexible rRNA expansion segment ES27 was suggested to be involved in regulating the access of non-ribosomal factors to the exit tunnel36. Binding of Arx1 to the pre-ribosome at the exit tunnel and interaction with ES27 might favor the ES27out conformation, which potentially facilitates nuclear export of the subunit. Arx1 is structurally related to methionine aminopeptidases (Met-APs), which remove N-terminal methionine from the nascent polypeptide chain; however, Arx1 lacks this enzymatic activity20. It is therefore plausible that Arx1 binds the pre-60S subunit in a similar manner and position to Met-AP on the mature 60S subunit. We speculate that Arx1 could function as a placeholder for Met-AP and/or other cytoplasmic translation-associated factors that bind to the exit tunnel. In this way, Arx1 could restrict the access of such factors to the exit tunnel region of the immature 60S subunit. This could avoid steric hindrance during export of the 60S subunit (note that Arx1 also acts as export factor; see below) or premature association of translation-competence. Last but not least, Arx1 may act as quality control factor to ensure correct assembly of this key site on the 60S subunit before the particle is exported to the cytoplasm.
Finally, the localization of the export factor Arx1 supports the model that transport receptors are distributed over different regions of the pre-60S subunit. The export adapter Nmd3 was found to interact with the intersubunit surface of the mature 60S subunit17. Mex67-Mtr2 was shown to interact with 5S rRNA in vitro18 and may thus bind to the central protuberance at the “top” of the subunit. Arx1 binds close to the exit tunnel on the opposite side of the pre-60S subunit, a location that is very distant from the predicted interaction sites of the other two export factors. It might have been envisaged that nuclear export factors would be needed to initiate subunit export and then “tow” the subunit through the nuclear pore complex. However, the dispersed distribution of export factors on the pre-60S surface supports a view that no single region of the 60S subunit is sufficient for efficient passage through the nuclear pore. Rather, several export factors scattered over the complete surface of the pre-60S subunit are required to shield and efficiently export this huge cargo through the hydrophobic FG repeat phase of the nuclear pore into the cytoplasm.
Yeast strains used in this study are listed in Supplementary Table 1. The genotype of the DS1-2b strain is MATalpha his3-Δ00 leu2-Δ1 trp1-Δ63 ura3-Δ52 derived from FY23xFY8626. TAP-tag or 3HA-tag were genomically integrated at the C-terminus as described49,50. For genomic integration of the Flag-tag the sequence of the Flag-epitope was integrated into the F3 primer50. For genomic integration of C-terminal His6-TEV-ProtA-tag the integration cassette plasmid pFA6a-HTpA-HIS3MX4 (kindly provided by D. Kressler) was used.
The CRAC method and bioinformatics analysis were performed as described35 using the Arx1-HTP (His6-TEV-ProtA) strain for CRAC analysis of Arx1 and the parental strain DS1-2b as a negative control. Two independent experiments were performed and the results merged.
Particles were essentially purified from 2-4 liters YPD liquid cultures grown to log phase at 30°C via TAP (tandem affinity purification) in standard buffer (50 mM Tris-HCl [pH 7.5], 100 mM NaCl, 1.5 mM or 5 mM MgCl2 [where indicated], 0.075 % (v/v) NP-40) as described26 with the following modifications. NP-40 was present only during cell lysis. Final eluates were in standard buffer without NP-40, with 0.5 mM DTT and 3.5 mM EGTA.
For split-tag purifications the TEV-protease eluate from the first affinity purification step of a TAP purification exploiting the ProtA-tag present on the bait protein was applied to anti-Flag M2-agarose from mouse (Sigma) for 1 hour at 4°C, followed by elution for 45 minutes at 4°C with 3xFlag peptide (Sigma) in standard buffer without NP-40 with 0.5 mM DTT.
Anti-HA antibody (monoclonal HA.11, Covance) was present in 1:50 (used for negative stain EM) or 1:200 dilution during the incubation on Calmodulin beads. Binding time was extended to 1.5 hours at 4°C. The tandem affinity-purified, split-tag purified, or TAP-purified and immuno-labeled particles were negatively stained with 2% uranyl acetate using the sandwich technique, imaged under low dose at 200 kV at 27,500 fold magnification (5.2 Å/pixel) with a Philips CM200 FEG transmission electron microscope supplied with a 2k x 2k CCD camera (TVIPS-GmbH) or at 100 kV at 33,000 or 53,000 fold magnification (3.6 Å/pixel or 2.2 Å/pixel) with a Philips CM120 BioTWIN LaB6 cathode transmission electron microscope supplied with a 4k x 4k CCD camera (FEI slow scan; used for negative stain of Fig. 4), all essentially as described14. Boxing of particles using “Boxer”51 and image processing using IMAGIC-552 for single particle analysis were also essentially done as described14.
For random conical tilt53, micrographs were taken at 0° and −55° tilt using the CM200 FEG microscope. A map was calculated from the tilted particles belonging to one class of untilted particles using IMAGIC-553 and SPIDER54 software and displayed using the UCSF Chimera55 software.
Image and particle numbers analyzed were 176/4201 for Arx1-TAP, 250/8449 for Alb1-TAP Arx1-HA without anti-HA, 201/5236 for Alb1-TAP Rpl3-HA, 213/3310 for Alb1-TAP Rpl5-HA, 259/5933 for Alb1-TAP Rpl8-HA, 224/4331 for Alb1-TAP Rpl26-HA, 200/3835 for Alb1-TAP Arx1-HA, 162/4983 for Alb1-TAP Nsa2-HA, 116/3216 for Alb1-TAP Tif6-HA, 200/1476 for Rix1-TAP Arx1-HA, 170/2141 for Rix1-TAP, all with anti-HA, and 55/7219 for Alb1-TAP, all at 1.5 mM MgCl2, and 60/11,922 for Alb1-TAP at 5 mM MgCl2.
As previously described56, samples were applied to carbon-coated holey grids, and micrographs were recorded under low-dose conditions (20 e−/Å2) on a Titan Krios TEM (FEI Company) microscope at 200 kV in a defocus range of 1.0–3.5 μm with a TemCam F416 camera (4,096 × 4,096 pixel, TVIPS GmbH) resulting in a pixel size of 1.049 Å on the object scale. The data were processed with the SPIDER software package54. Images were manually inspected for good areas and power spectra. Particles were automatically picked from 6,359 micrographs using projections of the crystal structure of the S. cerevisiae 60S ribosomal subunit (PDB: 3O587) as a template resulting in 222,229 particles. After initial alignment to the 60S ribosomal subunit, remaining non-ribosomal particles and 80S contaminations were removed by semi-supervised classification using iterative multi-reference 3D projection alignment. Templates were provided only for the initial classification step. Subsequently, the respective output maps of the previous classification rounds were used as new templates. Non-ribosomal particles were identified by their higher cross-correlation to an essentially featureless high-contrast density than to ribosomal references. An 80S-containing population of particles accumulated when the prominent orange-colored ligand was removed from the density and the resulting map was supplied as alternative reference during classification. 54,196 pre-ribosomal particles were then classified using maps of consecutive refinement rounds as templates. Using these quasi identical templates, the dataset was classified according to the most relevant intrinsic heterogeneity (e.g. density in the intersubunit space) minimizing the risk of introducing bias by artificial templates. Subsequently, the data were analyzed by focused sorting according to the density below the exit tunnel using a cylindrical binary mask (removing 6,795 particles)47,57. 8,322 particles were used for an initial reconstruction of the Arx1 pre-60S ribosomal particle at 14.3 Å resolution. In an effort to improve the reconstruction the entire data set was then re-processed. Considering the substantial additional mass identified on the pre-60S subunit in the previous reconstruction, this reconstruction was now used as template for the initial alignment. This increased the number of identified pre-ribosomal particles to 112,632. These particles were again classified using maps of consecutive refinement rounds as templates, resulting in a stable set of 63,943 particles that was used for reconstruction of the Arx1-particle. The final contrast transfer function corrected reconstruction has a resolution of 11.9 Å, based on the Fourier Shell Correlation with a cutoff value of 0.5. The density for the conserved 60S core and additional densities were isolated manually using the UCSF Chimera55 software.
Antibodies used for Western analysis in the following dilutions were anti-HA 1:3,000 (HA.11 mouse monoclonal antibody, clone 16B12, Cat.-No. MMS-101R, Covance, Berkeley, California, USA), anti-Arx132 1:2,000, anti-Nmd358 1:5,000, polyclonal rabbit anti-Mex67 1:5,000 (gift from Catherine Dargemont), anti-Mtr259 1:500, anti-Rpl360 1:4,000, anti-Rei125 1:5,000, anti-CBP 1:2,000 (Cat.-No. CAB1001, Thermo Scientific Open Biosystems, Rockford, Illinois, USA), goat-anti-mouse 1:6,000 (Cat.-No. 170-6516) and mouse-anti-rabbit horse radish peroxidase conjugated antibodies 1:6,000 (Cat.-No. 170-6515, both BIORAD, Munich, Germany). Page Ruler Unstained Protein Ladder (Thermo Scientific, Rockford, Illinois, USA) was used as a protein marker, Brillant Blue G-Colloidal Concentrate Electrophoresis Reagent (Sigma-Aldrich, Munich, Germany) was used for Coomassie stain, and 4-12% NuPAGE Bis-Tris Gels (Novex, Darmstadt, Germany) together with NuPAGE MOPS SDS Running Buffer (Invitrogen, Darmstadt, Germany) were used for SDS-PAGE.
We are very thankful to the EMBL-Heidelberg for providing the EM facility and computational infrastructure without which this work would not have been possible, and especially M. Diepholz, C. Blachiere-Batisse, J. Briggs, F. Thommen, and M. Wahlers for advice and technical support. The plasmid pFA6a-HTpA-HIS3MX4 was a kind gift of D. Kressler (Unit of Biochemistry, Department of Biology, University of Fribourg, Fribourg, Switzerland). We thank C. Dargemont (Institut Jacques Monod, Universités Paris VI and VII, Centre National de la Recherche Scientifique, Paris, France), A. W. Johnson (Molecular Genetics & Microbiology, The University of Texas at Austin, Austin, Texas, USA), A. Lebreton (Unité des Interactions Bactéries-Cellules, Institut Pasteur, Paris, France), and J. R. Warner (Department of Cell Biology, Albert Einstein College of Medicine of Yeshiva University, Bronx, New York, USA) for providing antibodies. This work was supported by grants from the Wellcome Trust (D.T.) and the Deutsche Forschungsgemeinschaft (DFG Hu363/10-4 to E.H.).
ACCESSION CODES The cryo-EM reconstruction density map has been deposited in the Electron Microscopy Data Bank (EMDB) under accession code EMD-5513, atomic coordinates of the crystal structures modeled into the reconstruction can be found at the Protein Data Bank entry 3J2I.
COMPETING FINANCIAL INTERESTS The authors declare no competing financial interests.