|Home | About | Journals | Submit | Contact Us | Français|
Ribosome assembly involves rRNA transcription, modification, folding and cleavage from precursor transcripts, and association of ribosomal proteins (Rps). In bacteria, this complex process requires only a handful of proteins in addition to those needed for rRNA transcription, modification and cleavage, while in eukaryotes a large machinery comprising ~ 200 proteins in the yeast S. cerevisiae has been identified. Furthermore, while the bacterial assembly factors generally produce only cold-sensitive phenotypes upon deletion, most of the eukaryotic assembly factors are essential, comprising ~ 20% of essential yeast proteins. This review explores recent rapid progress in the structural and functional dissection of the 40S assembly machinery.
Over the last 10 years, the ribosome assembly field has undergone a dramatic transformation. Once essentially considered a solved question (after all, ribosomes could be reconstituted in vitro) the field was pushed wide open by ground-breaking proteomic studies revealing the large complexity of the eukaryotic ribosome assembly machinery [1-3]. Once the dizzying number of ribosome assembly factors became evident, with almost no functional data available, making sense of ribosome assembly seemed almost impossible. Excitingly, the dust is now settling and we can see the possibility of understanding the vast ribosome assembly machinery. This review focuses on structural insights into small subunit assembly that have been recently garnered. Importantly, we believe that some general features deduced here are relevant for large subunit assembly as well, and some references to that are made. Interested readers are also referred to other recent reviews with more functional focus [3-5].
Box C/D snoRNAs bind co-transcriptionally to promote pre-rRNA methylation at certain 2′-OH residues [6*], as suggested by their binding to components of the PolI transcription machinery . Deletion of individual snoRNAs has, with the exception of U3, U14 and snR30 [7-9], no effect on yeast growth or ribosome assembly. However the concurrent deletion of several snoRNAs often produces synthetic phenotypes [10,11]. Mapping of the box C/D snoRNA binding sites within ribosomes indicates that their binding prevents the formation of secondary and sometimes tertiary structure elements (Figure 1), perhaps to keep the rRNA structure more fluid and facilitate the insertion of Rps, many of which have unstructured extensions that insert into the rRNA core [12, *13, 14]. snR30 prevents the premature formation of a large eukaryote-specific structure, which could lock in the tertiary structure of the body [15*]. Furthermore, several rRNA duplexes whose formation is inhibited by snoRNA binding interact with later-binding Rps (Figure 1), indicating that snoRNA binding might delay the incorporation of these proteins.1
Upon methylation snoRNPs are removed from pre-rRNAs by Prp43 [17-20]. It is not known if all snoRNAs dissociate prior to early pre-rRNA cleavage steps, or if some remain bound longer. Quantitative studies in vertebrates have shown that the snoRNA-dependent 2′-methylations at G1448 in 18S and A394 in 28S are not complete during transcription, while essentially all 20S (pre-18S) and 32S (pre-28S) rRNA molecules are methylated at these sites . Furthermore, Prp43 genetically interacts with proteins required for late cytoplasmic 40S assembly steps , indicating that perhaps some snoRNAs are not removed until later in assembly. The finding that all pre-18S and pre-28S rRNAs are methylated at the two investigated residues, if confirmed for additional methylation sites, also indicates an impressive efficiency of the snoRNP system. Perhaps the transcription machinery includes factors that facilitate the annealing of snoRNAs. Alternatively, it is possible that transcripts not bound to snoRNAs could be efficiently degraded, consistent with recent observations [*6].
Early work with bacterial ribosomes uncovered the order in which Rps bind, and how binding of a first set of proteins (primary binders) promotes rRNA structural changes that facilitate binding of a secondary and then tertiary set of proteins [23,24]. More recently, mechanistic studies have shown that Rps binding to rRNA is not a simple two-state process, in which the individual interactions of a single protein are made all at once; instead, even the interactions of a single Rps occur in multiple steps [**25].
More globally, a recent EM study of in vitro reconstituted assembly intermediates has shown that binding of proteins to the body precedes binding to the platform and head, with the formation of stable assembly intermediates that lack the head and sometimes even the platform, or are missing individual protein components from these regions [**26].
Excitingly, these in vitro results with bacterial 30S subunits are highly similar to results from yeast in vivo studies. Most Rps in the body are required for early co-transcriptional processing steps, while only a subset of platform binding proteins is required (Figure 1). Head-binding proteins are not required until the later export and cytoplasmic cleavage steps [13,*27]. Even more strikingly, intermediates in the bacterial assembly reaction lack the homologs of the head binders Rps3 and Rps20, as well as the platform binders Rps14 and Rps26, but are otherwise completely assembled [**26]. Similarly, we have recently shown that in a late yeast assembly intermediate Rps26 is absent, and Rps14 is not at its final position. Additionally, Rps3 is present but not correctly positioned, Rps20 and Rps29 are only bound to a fraction of molecules and Rps10, which also binds to the head but has no homolog in bacteria, is absent. The absence of Rps10 and Rps26 is explained by the overlap of their binding sites with assembly factors suggesting that their incorporation is delayed until assembly factors dissociate, perhaps even after rRNA cleavage is complete [**28].
The similarity in the assembly of Rps in vitro and in vivo and from different kingdoms of life, suggests that the main factors governing protein binding are conserved and must be intrinsic to rRNA and proteins. Furthermore, assembly factors are not producing these patterns although they might modulate them. For example, by delaying Rps incorporation until snoRNAs are dissociated or 18S maturation is complete, as described above.
Interestingly, deletions of a single copy of most Rps are synthetically lethal or slow growing with deletion of the cytoplasmic Dom34, required for degradation of stalled 80S ribosomes, including those containing 18S precursors [*29,**30]. This finding suggests that ribosome assembly can continue in the absence of an early binding Rps. However, because these ribosomes cannot function, they stall at the 80S stage, and are then degraded.
In addition to Rps from the body and platform, a large set of assembly factors is required for the early, co-transcriptional rRNA cleavage steps. These proteins, as well as the essential snoRNA U3, are collectively referred to as the small subunit (SSU) processosome, and comprise several subcomplexes: UtpA, UtpB, UtpC and Mpp10/U3 [31-35]. Data in the literature indicate that there is at least one additional complex, consisting of Utp7, Utp14 and Sof1. Interactions within the UtpA and UtpB subcomplexes, as well as among complexes have been mapped by yeast-2-hybrid screens and the conceptually related protein complementation assay (PCA) [36,37], and are summarized in Figure 2 (see also: ). This and other work suggests the following assembly steps for the SSU processosome: The UtpA subcomplex binds first to nascent pre-rRNA transcripts [*38], somehow changing its structure to facilitate subsequent processing steps. This function is required under normal conditions but becomes dispensible when quality control mechanisms that degrade slowly-assembling particles are inactivated . Base pairing between U3 and two regions in the pre-rRNA 5′ to the 18S rRNA (5′-ETS, Figure 3A) is required for incorporation of the U3/Mpp10 complex, which is bound to UtpB [40,41]. At this later stage, the base pairs between U3 and 18S rRNA sequences are formed [*42], stabilized by Imp4 [*43]. These base pairs are mutually exclusive with the formation of the central pseudoknot. Incorporation of the UtpC complex is independent of these events but requires Rrp5 , which binds near the 3′-end of 18S rRNA and within the ITS1 spacer sequence between 18S and 5.8S rRNA .
These interactions, as well as existing structural and biochemical data, provide insight into the organization of early pre-40S ribosomes despite the lack of any direct structural data (Figure 3): Early SSU-dependent cleavages occur in a body-like particle, with a partially assembled platform and head, but with formation of the central pseudoknot inhibited by U3 snoRNA binding. The 5′-end of 18S rRNA is located on the solvent side, suggesting that the hairpin with cleavage sites A0 and A1 extends in that direction. On the subunit interface of the mature particle, the central pseudoknot is rendered inaccessible by the top of the decoding site helix, H44 (Figure 3B). However, in early pre-40S particles, the top of H44 is not yet formed, leaving the central pseudoknot accessible to U3 binding [*45]. Furthermore, in the early pre-40S structure cleavage site A2 in ITS1 is located near the top of H44. Cleavage site A2 is recognized by the nuclease Rcl1 (D.M. Horn, S.L. Mason & K.K., in revision), which binds the GTPase Bms1. Bms1 interacts with U3 snoRNA  and H44 (K.K., unpublished), and based on the substantial sequence similarity with Tsr1 , is likely positioned to the left of H44, where Tsr1 is bound in later pre-40S particles [**28]. Together these data suggest that the U3 snoRNP is located on the subunit interface, where it interacts with the Bms1/Rcl1 complex. As a result, the 5′-ETS sequence between the U3 binding sites and the A0/A1 cleavage sites must wrap around from the solvent side to the subunit interface. As ITS1 binds near the platform [**28], it is likely that the 5′-ETS binds near the assembling beak, which could explain the interactions between the UtpB component Utp18, and the Ltv1/Enp1 subcomplex located at the beak. It will be exiting to obtain structures of such particles in the next several years.
While the early assembly events described above occur co-transcriptionally in the nucleolus, a post-transcriptional stage in 40S ribosome assembly takes place in the cytoplasm. These two stages are not only separated physically, their hierarchy is imposed by a structural switch in pre-40S ribosomes, and the almost complete exchange of the bound assembly factors: Early in 40S assembly, base pairing between the 3′-strand of H44 and ITS1 prevents premature cleavage at the 3′-end, formation of the decoding site, while allowing access of U3 snoRNA to the central pseudoknot [*45]. Additionally, while early assembly events require a large number of SSU components , the assembly intermediate that accumulates in the cytoplasm contains only seven assembly factors [*48, **28]; of these, only Dim1 and Pno1 bind early assembly intermediates. This suggests that at some point between cleavage at site A1 (to generate the mature 5′-end) and before nuclear export, the SSU components must dissociate to allow for binding of the late 40S assembly factors. What triggers this remodeling (perhaps cleavage at sites A1 or A2), or whether proteins catalyze SSU processome removal, akin to the AAA+-ATPases, which perform similar roles during 60S assembly [49, **50, 51], is unknown. Similarly, the inhibitory H44-ITS1 duplexes must be unwound to allow for the conformational switch in H44, required for cleavage at the 3′-end of 18S rRNA [*45].
Cytoplasmic 40S assembly requires seven stably bound assembly factors: the methylase Dim1, the nuclease Nob1 and its regulator Pno1, the export adaptor Ltv1 and its binding partner Enp1, the kinase Rio2 and the GTPase-like protein Tsr1. In addition, the kinase Rio1 and the ATPase Fap7 also transiently interact with this particle to produce mature 40S subunits. While the binding sites for these factors have been recently mapped [52,*53], only a few have known functions. Dim1 methylates two conserved adenosines near the 3′-end in a universally conserved reaction . Pno1 modulates Nob1′s RNA binding activity and is required for efficient Nob1-dependent formation of the 3′-end of 18S rRNA .
Cytoplasmic pre-40S subunits are inundated with large concentrations of mRNAs, translation initiation factors and 60S subunits, but premature formation of 80S subunits leads to efficient degradation of the pre-18S rRNA [**30]. To chaperone the pre-40S particle all seven cytoplasmic assembly factors cooperate to inhibit each step in the translation initiation pathway [**28]. Assembly factors bound to the subunit interface (Tsr1, Rio2 and Dim1) and the platform (Pno1 and Nob1) prevent binding of translation initiation factors eIF1, eIF1A, and eIF3, while Enp1 and Ltv1 prevent opening of the mRNA channel. Tsr1 blocks joining of 60S subunits and, finally, the decoding site is deformed. Interestingly, with the exception of Dim1, there are no bacterial homologs of these proteins. Nevertheless there are non-homologous assembly factors, which could have analogous functions. E.g., mRNA recruitment in bacteria requires the Shine-Dalgarno sequence, which is bound and possibly blocked by the GTPase Era . Similarly, the bacterial KH-domain containing protein RbfA deforms the decoding site .
If Tsr1 blocks premature joining of 60S subunits with assembling 40S subunits, then how can pre-18S rRNA-containing 40S subunits enter polysomes as described [**30]? This apparent paradox indicates that there is more than one cytoplasmic 40S assembly intermediate: one or more that contain Tsr1 and at least one later particle, which enters polysomes and does not contain Tsr1. Consistent with this hypothesis, it has long been known that cytoplasmic 40S ribosome maturation involves two chemical changes to the 18S rRNA. These include cleavage at the 3′-end to produce the mature 18S rRNA [22,*45] and the universally conserved methylation of two neighboring adenosines in H45, the terminal stemloop . In addition, Rps10, Rps26, and Rps3 must be incorporated, the latter from a salt-labile conformation [*48, **28]. Furthermore, two conformational rearrangements, the formation of the decoding site [**28] as well as the formation of contacts between the very C-terminus of Rps14 and 18S rRNA  take place. Finally, the seven stably bound assembly factors dissociate. This extensive (and likely incomplete) list of distinct steps in cytoplasmic 40S assembly suggests that this stage is substantially more complex than previously thought. What is the order of these steps and how are they regulated? The answer to these questions awaits further study but promises to be as intricate as the recently unraveled 60S maturation pathway [*59].
Towards a better understanding of 40S assembly, we must now provide richer functional as well as structural data. Drawing on the identification and mapping of subcomplexes, this will likely require generating of point mutations that more subtly interfere with protein function, perhaps by disrupting a protein-protein or protein-RNA interface [36,46,50,55]. This will be useful because depletion of one assembly factor often leads to co-depletion of many others, thereby obscuring the role of any single protein. Furthermore, novel intermediates need to be identified, as assembly is clearly much more complex than Northern blots reveal. This will require the development of new techniques such as CRAC crosslinking [*53, 60] or the revisiting of old ones, such as RNA structure probing [*42,*45]. High-resolution structures of individual assembly factors, or smaller complexes, as well as medium-resolution electron microscopy-derived structures of pre-ribosomal complexes will also have a tremendous impact on better understanding the assembly process.
As the ribosome assembly field matures its interface with other fields will also be explored: How do ribosomes transition from assembly to translation? Does the first round of translation initiation utilize the same mechanism and translation initiation factors as any subsequent one? If there are specialized ribosomes, as suggested , are there specialized assembly factors to generate certain subpopulations? And finally, how exactly is the ribosome assembly machinery integrated with the cell cycle?
I thank the members of my laboratory for many helpful discussions and comments on the manuscript. Work on 40S ribosome assembly in my lab is funded by the NIH (1R01GM086451), NSF (MCB0845156), and a Beginning Investigator grant from the AHA.
1The same conclusion can be drawn for box H/ACA snoRNAs, which introduce pseudouridylation into rRNA , but which are not discussed for space reasons. Interestingly, snoRNA binding sites can overlap, indicating either that they function one after the other, or that they occur alternatively, thereby combinatorially increasing the diversity in the ribosome pool.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.