The ribosomal stalk of the 60S subunit has been shown to play a crucial role in all steps of protein synthesis, but its structure and exact molecular function remain an unanswered question. In the present study, we show the low-resolution models of the solution structure of the yeast ribosomal stalk, composed of five proteins, P0–(P1–P2)2. The model of the pentameric stalk complex determined by small-angle X-ray scattering reveals an elongated shape with a maximum length of 13 nm. The model displays three distinct lobes, which may correspond to the individual P1–P2 heterodimers anchored to the C-terminal domain of the P0 protein.
- cryo-electron microscopy
- GTPase centre
- ribosomal P protein
- ribosomal stalk
- small-angle X-ray scattering (SAXS)
At all stages of protein synthesis the ribosome requires protein factors, known as tGTPases (translational GTPases), which act as molecular switches cycling between an active GTP-bound conformation and an inactive GDP-form, conferring unidirectionality and fidelity on the translational apparatus. The platform for the tGTPases, called the GAR (GTPase-associated region), is located on the large ribosomal subunit and is responsible for tGTPase activation . The major part of GAR is composed of a set of proteins which form the stalk The stalk is responsible for the recruitment of tGTPases and stimulation of factor-dependent GTP hydrolysis [2,3]. In bacteria this complex occurs in two configurations: pentameric [L10–(L12)4] in mesophiles and heptameric [L10–(L12)6] in thermophiles [2,4]. In archaeal/eukaryotic ribosomes the stalk is heptameric [L10–(L12)6] or pentameric [P0–(P1–P2)2] respectively [5–7]. The bacterial/eukaryotic L10/P0 proteins form the base of the stalk which interacts with rRNA. The rRNA-binding domain of these proteins is homologous and functionally interchangeable . In turn, the L12/P1–P2 proteins form the functional part of the stalk which is attached to the ribosome through the L10/P0 proteins, as (L12)2 or P1–P2 dimers [2,9–11]. An interesting case is found in lower eukaryotes, such as the yeast Saccharomyces cerevisiae, where the two P1/P2 proteins have evolved up to four different proteins, P1A, P1B, P2A and P2B [10,12]. In contrast with L10/P0, the L12/P1–P2 proteins are regarded as analogous . In spite of the wealth of ribosomal structures determined by crystallography or cryo-EM (electron microscopy), studies of the whole stalk remain poorly established. In cryo-EM analyses of bacterial/eukaryotic ribosomes, the stalk was only visible as a residual structure representing only a small part of the stalk [14,15]. For the bacterial ribosome, cryo-EM analysis only allowed a fragment of the L12 protein to be located: its C-terminal part was trapped close to the G′ domain of EF-G (elongation factor G) . Crystallography of ribosomes has also not provided any more information on the structure of the whole stalk. Fragments of L10 protein were visualized either on the ribosome as an rRNA-binding domain [2,17] or as fragments of individual L10 proteins [2,18]. Crystallographic analyses of isolated stalk complexes of L10–L12 from bacteria or archaea provided fragmentary information about the stalk architecture, showing the core structure excluding the functional CTDs (C-terminal domains) of L12 [2,19]. An insight into the CTD structure has been obtained by NMR , crystallography [21,22] or SAXS (small-angle X-ray scattering)  analyses. The stalk has been visualized on the bacterial ribosome crystallized in complex with EF-G. Maps at a lower resolution [5.0 Å (1 Å=0.1 nm)] allowed placement of L10 and four copies of the N-terminal domain of L12 in the model as polyalanine chains, showing for the first time a substantial fragment of the stalk on the ribosome .
In the present paper we report the low-resolution structural model of the yeast ribosomal stalk determined by SAXS. The structure encompasses the P0 fragment (residues 199–312) with two P1–P2 heterodimers (described in yeast as two independent P1A–P2B and P1B–P2A dimers) representing the entire stalk, including the C-termini of the P-proteins which constitute the functional part of the stalk. The structure was modelled on to the ribosomal particle, giving the first insight into the complete architecture of the ribosome with the stalk.
Preparation of yeast 80S ribosomal particles and stalk complexes was performed as described previously . The protein concentration was measured according to the Bradford method using a Bio-Rad protein assay kit, or by absorbance at 280 nm using a molar absorption coefficient calculated from the amino acid composition . Two P-protein complexes were purified, TH199 and TH230, with the compositions ΔP0199–312–(P1A–P2B)–(P1B–P2A) and ΔP0230–312–(P1B–P2A) respectively.
The synchrotron radiation X-ray scattering data from solutions of the ribosomal stalk complexes were collected on the X33 beamline of EMBL (European Molecular Biology Laboratory) on the storage ring DORIS III [DESY (Deutsches Elektronen Synchrotron), Hamburg, Germany]. For each complex, several solute concentrations in the range 1–15 mg/ml were measured. To monitor for radiation damage, two successive 2 min exposures of protein solutions were compared and no significant changes were observed. The data were normalized to the intensity of the transmitted beam and radially averaged; the scattering of the buffer was subtracted and the difference curves were scaled for protein concentration. The low-angle data measured at lower protein concentrations were extrapolated to infinite dilution and merged with the higher concentration data to yield the final composite scattering curves. The data-processing steps were performed using the program package PRIMUS . The forward scattering, I(0), and the radius of gyration, Rg, were evaluated using the Guinier approximation  assuming that at very small angles (s<1.3/Rg) the intensity is represented as I(s)=I(0) exp[−(sRg)2/3]. These parameters were also cross-checked with the values computed from the entire scattering patterns using the indirect transform package GNOM , also providing the pair distribution function of the particle, p(r), and the maximum size, Dmax. The excluded volume of the hydrated particle was computed from the small-angle portion of the data (s<3.0 nm−1) using the equation described previously . Before this analysis, an appropriate constant was subtracted from each data point to force the s−4 decay of the intensity at higher angles following Porod's law for homogeneous particles .
Ab initio shape determination
The ‘shape scattering’ curves were further used to generate low-resolution ab initio shapes of the pentameric (TH199) and trimeric (TH230) stalk complexes by DAMMIN . In total, 12 DAMMIN runs were performed to check the stability of the solution, and the results were superimposable with each other. These models were averaged to determine common structural features and to select the most typical shapes using the programs DAMAVER  and SUPCOMB .
Fitting the stalk structure into ribosomal cryo-EM maps
For visualization of a ribosome stalk model the EMDB database (http://EMDataBank.org) was screened in a search for maps containing strong stalk occupancy. The cryo-EM density maps of 80S particles from S. cerevisiae (EMD-1668)  and Canis familiaris (EMD-1480)  were selected. They were aligned in UCSF Chimera  and filtered to 10 Å. A simulated density map of the S. cerevisiae ribosome at 10 Å was created from PDB files (3IZB, 3IZF, 3IZE and 3IZS)  using the program VMD . The stalk P-proteins (P0, P1–P2) were simulated separately from PDB using the surface option in the UCSF Chimera program.
RESULTS AND DISCUSSION
One of the major obstacles for the high-resolution analysis of the stalk is its dynamic nature [11,38], which hampers crystallographic and cryo-EM analyses. The use of NMR is hindered by the size of the complex, which usually exceeds 50 kDa; and preparation of the complex is challenging as three or five [in yeast, P0–(P1A–P2B)–(P1B–P2A)] components have to be tightly assembled. To provide insight into the stalk structure, we used SAXS, which provides low-resolution data only, but overcomes the limitations of the other methods. The stalk was obtained from S. cerevisiae ribosomal particles, using a procedure which allowed isolation of natively assembled complexes . Two complexes were used: (i) pentameric TH199, which has a truncated P0 protein with the rRNA-binding domain removed and only the P-domain left intact (residues 199–312), and is able to bind the two independent P1A–P2B and P1B–P2A dimers; and (ii) trimeric TH230, ΔP0230–312–(P1B–P2A), which is a further deletion form of the TH199 complex containing a truncated variant of P0 able to bind only one dimer, P1B–P2A. Both complexes were subjected to SAXS analysis yielding the processed experimental data shown in Figure 1(A). The Guiner plots (Figure 1A, inset) were linear, suggesting that the protein complexes form monodisperse solutions containing single molecular species. The estimated molecular mass of the pentameric complex (55±7 kDa) agreed with the value calculated from the primary sequence (56.42 kDa). For the truncated variant, the experimental value of 40±6 kDa exceeded the calculated value (31.18 kDa). However, the excluded particle volume of the truncated variant (Figure 1B) scaled well with that of the full complex, indicating that the truncated complex was not aggregated. The distance distribution functions, p(r), of the complexes (Figure 1A) were bell-shaped functions typical for globular particles. The distribution of the TH199 complex was more skewed, pointing to an elongated particle with a cross-section of approximately 2.5 nm [corresponding to the maximum of the p(r)]. The trimeric TH230 complex displayed a more symmetrical p(r) function and revealed the same maximum diameter (13 nm) as the truncated variant, suggesting that the TH230 complex ‘folds back’ into the TH199 complex.
Overall structure determination
Low-resolution models of the TH199 and TH230 complexes were reconstructed ab initio from the experimental scattering patterns. The models obtained for the two complexes provided good fits to the experimental data (Figure 1A). The TH199 structure had a flattened anisometric elongated shape with three domains, belonging presumably to the individual dimers and to ΔP0 (Figure 1C). The model of the TH230 complex reflected part of the TH199 pentamer lacking one domain (Figure 1D). These relationships were visualized by aligning the two structures (Figure 2). TH230 could be fitted into the TH199 complex (Figure 2B), showing that the missing domain corresponds to the P1A–P2B dimer. The relative position of the P1A–P2B dimer was verified by aligning the model of the TH199 complex with the structural model of the P1A–P2B dimer previously determined by SAXS . The fitting of the dimer within the pentamer (Figure 2A) showed that P1A–P2B occupies the edge of the stalk. The location of the dimer was also confirmed by rebuilding the framework of the TH199 pentamer using TH230 and the P1A–P2B dimer (Figure 2C). Therefore the architecture of the stalk can be visualized as two dimers arranged in parallel, brought into contact by their respective interaction with P0. As has been shown for the stalk from Pyrococcus horikoshii, the individual dimers are attached to P0 by its helical spine  and are arranged one after the other on P0. However, the archaeal structure lacks the C-termini of the P-proteins and represents the core of the stalk, in contrast with the present SAXS model. Given the low resolution of the present model, the P1–P2 C-termini cannot unequivocally be located, but a tentative assignment of the C-terminal part of P0 can be made. Considering that the P-domain of the P0 protein forms the helical spine responsible for binding of the dimers, we propose that helices are located in the main body of the SAXS structure, and that the protruding structure may correspond to the flexible C-terminus, forming a distinctive beak (Figure 2).
Modelling of the stalk complex on the 80S ribosome
Knowing the position of the stalk on the ribosome may help in understanding the actions of the GAR. Using the above model of the stalk we built a complete picture of the eukaryotic ribosome (Figure 3). In the cryo-EM data of eukaryotic ribosomes the stalk is missing  or is represented by a residual structure . On the basis of homology with the bacterial structures, the stalk can be assigned to the helical spine of the L10 protein and the core of the L12 dimers . Also, in recent crystallographic analysis of the yeast ribosome the stalk is visible as a residual fragment represented by a fragment of P0 protein (residues 3–107 and 182–221) and two P1A/P2B proteins (residues 1–46 and 5–51 respectively) . As discussed above, the main obstacle in stalk visualization is its dynamic nature, which also hampers analyses by cryo-EM, a technique which is sensitive to structural fluctuations. With regard to cryo-EM, a structure of a complex is a result of averaging the information from all the individual particles taken for analysis. This is good for alignable stable parts, such as the core of the ribosomal large subunit, where it provides more detail. However, an averaging of the stalk in multiple positions results in the observed smudge and lack of high-resolution features. The further away from the stabilized protein–RNA interaction of the stalk base P0/L10 protein, the less information is available, therefore the resolution of the whole stalk in the absence of a stabilizing factor is highly unlikely, especially the CTDs which are considered unstructured. Therefore in all cryo-EM reconstructions of the 80S ribosome or 60S subunit, the stalk represented is an undescribed entity, and it should be stressed that this is the largest ribosomal element to escape from the structural characterization on the ribosomal particle.
Since this particular element of the stalk is structurally conserved between bacterial and archaeal/eukaryotic structures, as shown by the recent structure of an archeal stalk complex , it is likely that the eukaryotic structure visible on cryo-EM maps belongs to the same structural entity. Therefore the EMDB database was screened for maps containing the strongest stalk occupancy, and the S. cerevisiae  and C. familiaris  cryo-EM density maps were chosen (Figure 3). After filtering down to a resolution of 10 Å, these were aligned with the present stalk from the SAXS model. Additionally, a simulated density map of S. cerevisiae ribosome at 10 Å was created from PDB files (3IZB, 3IZF, 3IZE and 3IZS) using the program VMD (Figure 3B). Initial inspection suggested that the stalk visible in all densities was a sufficient size to accommodate the SAXS model. The latter can be fitted into the cryo-EM density at the extended stalk region in both maps (Figures 3A and 3C). In order to precisely locate the stalk elements, we reconstructed this structure using several individual components, using the bacterial and archeal L10 rRNA-binding domain [2,17]. Additionally, the mode of L10–L12 interaction was built on the basis of structures in a non-ribosomal state of archeal P0–P1 from P. horikoshii  and P0 from Methanococcus jannaschii . These structures complement each other, as the P. horikoshii structure represents the rRNA-binding domain and the CTD of P0, with fragments of P1 dimers, whereas M. jannaschii data comprises the rRNA-binding domain and domain II, characteristic elements for archeal/eukaryotic P0, but lacks P1/P2 termini. By combining these structures we constructed a stalk model on the yeast cryo-EM ribosome (Figure 3B). The model was aligned with the stalk SAXS structure. The protruding part of the reconstituted model fitted with the solution stalk structure. In our model, the C-termini of P1/P2 proteins face outward. The single C-terminus of P0 (beak) in the SAXS model protrudes from the reconstituted model and represents its continuation. However, as we have proposed for the bacterial L12 protein , the yeast P1/P2 C-termini, having a flexible hinge, may sample a large volume to efficiently recruit translation factors or RIP (ribosomal inactivating proteins) . Therefore our SAXS model and its proposed ribosomal location represent a single steady-state conformational position.
Przemyslaw Grela, Michal Gajda and Jean-Paul Armache performed the experiments and collected the data; Roland Beckmann and Dmitri I. Svergun helped with the interpretation of the data; Dawid Krokowski prepared the yeast strains and helped with data interpretation; Nikodem Grankowski carried out a critical reading of the paper before submission; and Marek Tchorzewski designed the experiments, interpreted the data and wrote the paper.
This work was supported by the Ministry of Science and Higher Education, Poland [grant number N302 061034]. D.S. acknowledges support from the BMBF (Federal Ministry of Education and Research) Research Grant SYNC-LIFE [contract number 05K10YEA].
Abbreviations: CTD, C-terminal domain; EF-G, elongation factor G; EM, electron microscopy; GAR, GTPase-associated region; SAXS, small-angle X-ray scattering; tGTPase, translational GTPase
- © The Authors Journal compilation © 2012 Biochemical Society