Selenocysteine Insertion Sequence (SECIS)-binding Protein 2 Alters Conformational Dynamics of Residues Involved in tRNA Accommodation in 80 S Ribosomes*

Background: Selenocysteine incorporation requires unique translation factors that interact with the ribosome. Results: SECIS-binding protein 2 alters ribosome conformation at two discrete sites. Conclusion: The selenocysteine incorporation reaction requires ribosome modifications in a region known to be required for tRNA accommodation. Significance: Identifying the ribosomal dynamics required for Sec incorporation will enhance our understanding of the translation elongation reaction. Sec-tRNASec is site-specifically delivered at defined UGA codons in selenoprotein mRNAs. This recoding event is specified by the selenocysteine insertion sequence (SECIS) element and requires the selenocysteine (Sec)-specific elongation factor, eEFSec, and the SECIS binding protein, SBP2. Sec-tRNASec is delivered to the ribosome by eEFSec-GTP, but this ternary complex is not sufficient for Sec incorporation, indicating that its access to the ribosomal A-site is regulated. SBP2 stably associates with ribosomes, and mutagenic analysis indicates that this interaction is essential for Sec incorporation. However, the ribosomal function of SBP2 has not been elucidated. To shed light on the functional relevance of the SBP2-ribosome interaction, we screened the functional centers of the 28 S rRNA in translationally competent 80 S ribosomes using selective 2′-hydroxyl acylation analyzed by primer extension (SHAPE). We demonstrate that SBP2 specifically alters the reactivity of specific residues in Helix 89 (H89) and expansion segment 31 (ES31). These results are indicative of a conformational change in response to SBP2 binding. Based on the known functions of H89 during translation, we propose that SBP2 allows Sec incorporation by either promoting Sec-tRNASec accommodation into the peptidyltransferase center and/or by stimulating the ribosome-dependent GTPase activity of eEFSec.

Protein synthesis involves the sequential incorporation of amino acids by the translational apparatus in response to sense codons in the messenger RNA (mRNA). In eukaryotes, amino-acyl-tRNA molecules (aa-tRNAs) 3 are delivered to the ribosomal A-site by the elongation factor, eEF1A (EF-Tu in prokaryotes) in a ternary complex (TC) with GTP. An exception arises in the case of the 21st amino acid, selenocysteine (Sec), which is encoded by a nonsense UGA codon in selenoprotein mRNAs. Sec is biosynthesized on a specialized tRNA (tRNA Sec ) that is recognized and delivered to the ribosome by a dedicated elongation factor, eEFSec (1,2). However, the eEFSec TC is not sufficient to decode the UGA codon indicating that, unlike the canonical TC, its access to the A-site is restricted.
Interestingly, the serylated precursor from which Sec is biosynthesized (Ser-tRNA Sec ) is capable of suppressing the UGA codon in vitro (3). Ser-tRNA Sec is presumably bound and delivered by eEF1A because eEFSec exhibits marked specificity for Sec-tRNA Sec (1,2,4). This suggests that the inability of the eEFSec TC to suppress UGA codons randomly is an intrinsic feature of eEFSec, consistent with the presence of an additional fourth domain relative to the canonical elongation factor eEF1A. Canonical aa-tRNAs have been shown to bind the ribosomal A-site with an equal affinity despite the molecular diversity inherent in the various aa-tRNAs. Interestingly, this uniform binding is lost when deacylated tRNAs are used as substrates, suggesting that the amino acid also contributes to the thermodynamic stability of this interaction and highlighting the fact that the ribosome is not blind to the identity of the amino acid (5). Thus, it is also possible that the Sec residue contributes to this regulation by destabilizing the Sec-tRNA Sec in the absence of the other components of the Sec incorporation complex. Sec incorporation is specified by the Sec insertion sequence (SECIS) element in the 3Ј-untranslated region (UTR) of selenoprotein mRNAs and the SECIS-binding protein, SBP2 (6,7). However, the mechanism through which this information is transmitted to the ribosome is unknown. SBP2 comprises a dispensable N-terminal domain, a central Sec incorporation domain (SID), and a C-terminal RNA binding domain (RBD); together, the SID and the RBD are referred to as CTSBP2. Aside from its essential function in binding the SECIS, SBP2 also interacts with the ribosome, making it a likely candidate for communicating this specificity information. Consistent with this notion, extensive mutagenesis of the RBD in CTSBP2 led to the identification of a discrete stretch of amino acids ( 647 RFQDR 651 ) that when mutated to penta-alanine resulted in a decrease in ribosome binding, and Sec incorporation activity (11). Importantly, this SBP2 mutant retained near wild-type levels of SECIS binding activity allowing us to conclude that ribosome binding is an essential component of the Sec incorporation activity of SBP2.
Despite this finding, we subsequently reported that when expressed as separate proteins and added to a Sec incorporation assay in equimolar amounts, the SID and the RBD retain near wild-type levels of Sec incorporation activity in vitro but lose the ability to associate stably with ribosomes (9). In addition, our attempts to detect an SBP2-SECIS-ribosome complex have been unsuccessful (10,11). Together with the fact that the SID and RBD as separate proteins could be formaldehyde crosslinked to ribosomes (9), these results indicate that transient SBP2-ribosome interactions are sufficient to promote Sec incorporation and suggest that SBP2 has distinct functions on the SECIS and the ribosome.
Using selective 2Ј-hydroxyl acylation analyzed by primer extension (SHAPE), we demonstrate that CTSBP2 specifically enhances the reactivity of residues U4071, and C4073 in the apical loop of Helix 89 (H89), and between G3802 and C3812, and G3758 and U3768 on expansion segment 31 (ES31; mouse numbering). Additionally, we demonstrate that the SID and RBD are also able to alter the reactivity of these same residues in ES31, and to a lesser extent H89, indicating that functional interactions persist despite the loss of stable ribosome binding. Based on the known roles of H89, we propose that SBP2-induced conformational changes in this helix promote Sec incorporation by affecting Sec-tRNA Sec accommodation and/or by stimulating the ribosome-dependent GTPase activation of eEFSec.

EXPERIMENTAL PROCEDURES
Constructs-The CTSBP2 (amino acids 399 -846), the SID (amino acids 399 -586), and the RBD (amino acids 586 -846) were all cloned into pTrcHis (Invitrogen) by TOPO-TA cloning, to generate constructs containing an N-terminal Xpress/ His tag. Rat ribosomal protein L30 (rpL30) in pET200 (Invitrogen) was kindly provided by Donna Driscoll. The penta-alanine mutant versions of CTSBP2 and RBD were generated by sitedirected mutagenesis of the CTSBP2 and the RBD constructs in pTrcHis. All of the constructs were sequenced completely by automated DNA sequencing.
Expression of Recombinant Proteins-All of the recombinant proteins in this study were purified as described by Kinzy et al. (10). All of the proteins were buffer-exchanged into 1ϫ SHAPE buffer lacking MgCl 2 (80 mM Hepes-KOH, pH 7.5, 100 mM KCl, 2 mM DTT) using a P30 gel filtration column (Bio-Rad) or by overnight dialysis.
Purification of 80 S Ribosomes-80 S ribosomes were purified as described, with minor modifications (12). Briefly, 50 ml of crude rabbit reticulocyte lysate (Green Hectares) was thawed, and one EDTA-free tablet of complete protease inhibitor mixture (Roche Applied Science) was added. Ribosomes were pelleted by ultracentrifugation at 100,000 ϫ g for 2.5 h at 4°C using a Sorvall S80-AT3 rotor. Ribosome pellets were first washed with Buffer A (50 mM Tris-HCl, pH 7.5, 0.25 M sucrose, 1 mM DTT, and 0.1 mM EDTA) and then gently resuspended in high salt buffer (Buffer A ϩ 0.5 M KCl) using 0.1-0.2 of the original lysate volume. The ribosome solution was allowed to stir for 30 -60 min at 4°C before layering over a 1 M sucrose cushion in high salt buffer (2:1 ratio of ribosome solution to sucrose pad) and centrifuging at 278,000 ϫ g for 2 h at 4°C. 80 S ribosomes were gently resuspended in 1ϫ SHAPE buffer (80 mM Hepes-KOH, pH 7.5, 100 mM KCl, 2 mM DTT, and 0.5 mM MgCl 2 ). The concentration of 80 S ribosomes was determined using an extinction coefficient at A 260 of 50 M Ϫ1 cm Ϫ1 . 80 S ribosomes were stored in small aliquots at Ϫ80°C. Ribosome function was validated by luciferase mRNA translation in ribosome-depleted rabbit reticulocyte lysate.
SHAPE Reactions-SHAPE reactions were generated by incubating 10 pmol of 80 S ribosomes in the presence or absence of 5 M CTSBP2 at 37°C for 5 min. Alternatively, SHAPE reactions contained 10 pmol of ribosomes and 10 M SID, RBD, or rpL30. All reactions were performed in 1ϫ SHAPE buffer (80 mM Hepes-KOH, pH 7.5, 100 mM KCl, 2 mM DTT, and 0.5 mM MgCl 2 ) in a final volume of 100 l. Benzoyl cyanide (BzCN; Sigma) or vehicle (DMSO) was added to a final concentration of 80 mM. rRNA was extracted using the RNeasy kit (Qiagen) according to the manufacturer's instructions. Ribosomal RNA was eluted with RNasefree water and either used immediately for primer extension or stored at Ϫ80°C.
Primer Extension and Data Analysis-1 pmol of modified rRNA was incubated with 12.5 pmol of ROX-labeled oligonucleotide (Invitrogen) in a total volume of 6.0 l at 80°C and slow cooled to 25°C using a thermocycler. 3 l of 4:1:1 buffer (5ϫ Superscript III buffer:100 mM DTT:10 mM dNTP mix) was added and incubated at 52°C for 1 min before adding 200 units of Superscript III (Invitrogen) and incubating for 20 min at 52°C. DNA fragments were ethanol-precipitated in the presence of 20 g of glycogen as a carrier, and DNA pellets were resuspended in 10 l of deionized formamide (Fermentas). Capillary electrophoresis was performed in house using an ABI 3130XL sequencer. GeneScan TM LIZ®500 was added to each reaction and used as both a size standard to determine the relative position of SHAPE generated peaks and as an internal control to normalize peak areas due to variability in the injection volumes. Capillary electrophoresis-generated traces were analyzed using DAx software (Van Mierlo software consultancy). The positions selected for discussion were altered in an SBP2-dependent fashion with 100% reproducibility for a minimum of five independent experiments in all cases except where the SID and RBD domains were tested as independent domains, in which case three independent experiments were performed.
Sequencing Reactions and Gel Electrophoresis-For sequencing reactions, rRNA was extracted from 80 S ribosomes using TRIzol (Invitrogen) according to the manufacturer's instructions and used as a template for primer extension. Primer extension was performed by mixing 1 g of rRNA with 2 l of end-labeled primer in a total volume of 6 l. Sequencing reactions also contained 1 l of ddNTP (10 mM; Roche Applied Science). Reactions were incubated at 80°C for 2 min and slow cooled to 25°C in a thermocycler. Next, 3 l of 4:1:1 buffer (5ϫ Superscript III buffer:100 mM DTT:10 mM dNTP mix) and 200 units of Superscript III was added, and reactions were incubating for 20 min at 52°C. 10 l of formamide gel-loading buffer (80% (w/v)) deionized formamide, 10 mM EDTA, pH 8.0, 1 mg/ml xylene cyanol, and 1 mg/ml bromphenol blue) was added directly to the primer extension reactions and incubated at 95°C for 5 min. 3.5 l of each reaction was resolved on a 6% denaturing polyacrylamide gel and analyzed by phosphorimaging.

RESULTS
Novel SBP2-dependent Alterations in SHAPE Reactivity in 28 S rRNA-SBP2 binds stably to the large subunit of the ribosome, and preliminary studies indicate that this interaction is mediated by rRNA (13). However, the precise location of SBP2 on the ribosome has not been determined, nor has the mechanistic contribution of this interaction during Sec incorporation been elucidated. The large subunit of the ribosome has two major functional centers: (i) the GAC, and (ii) the peptidyl- transferase center (PTC). The GAC is formed by H43 and H44, and along with the universally conserved SRL at the apical loop of H95, is responsible for binding and stimulating the GTPase activity of the translational GTPases, including initiation factor eIF5B (IF2 in prokaryotes), the canonical elongation factors eEF1A and eEF2 (EF-G in prokaryotes), and presumably eEFSec (SelB in prokaryotes). Each of these functional centers is shown in Fig. 1 as a diagram adapted from the Comparative RNA website (14) (Fig. 1). The PTC is the catalytic core of the ribosome; it is responsible for binding and correctly aligning the tRNA substrates in the A-and P-sites and catalyzing peptide bond formation. Due to the large size of 28 S rRNA, we chose to focus our studies on these areas because of their direct role in aa-tRNA selection during canonical translation elongation in all three domains of life. To shed light on the ribosomal function of SBP2, we used SHAPE to probe the 28 S rRNA for SBP2-dependent changes in the reactivity of residues at or near these regions to determine whether SBP2 could induce a unique rRNA conformation that would specify the recruitment of the eEFSec TC.
Probing 28 S Functional Centers-SHAPE makes use of a class of hydroxyl-selective electrophiles to analyze nucleotide dynamics in RNA (15). We used the compound BzCN because of its fast reactivity and hence its potential to capture transient conformational changes (16). Salt-washed 80 S ribosomes were purified from rabbit reticuloctye lysate, and the translational activity of these ribosomes was qualitatively assessed by supplementing rabbit reticuloctye lysate depleted of endogenous ribosomes and testing their ability to translate a luciferase reporter mRNA (data not shown). 80 S ribosomes were treated with 80 mM BzCN in the presence or absence of CTSBP2. Alternatively, ribosomes were treated with vehicle (DMSO) to distinguish primer extension stops arising from RNA cleavages or excessive secondary structure from those resulting from modified RNA. Modified or vehicle-treated rRNA was extracted and used as a template for primer extension using ROX-labeled oligonucleotides downstream of the regions indicated in Fig. 1. DNA fragments were precipitated and resolved by capillary electrophoresis in the presence of an internal size standard (GS500 LIZ). The internal size standard was used to approximate the position of reactive residues and to normalize for variability in the injection volumes of the different samples resolved by capillary electrophoresis. Normalization of the data in this manner significantly reduced any global fluctuation in SHAPE reactivity across samples as evidenced by the high degree of overlap in traces (see Figs. 2, 4, 6, and 7). Visual inspection of the normalized traces revealed that the lowest variability corresponded to a region ϳ50 to ϳ150 bases from the 5Ј end of the primer; for this reason, we limited our analysis to these regions only. Helix 89 Modifications in Presence of SBP2-Previous chemical probing studies have correlated a conformational change in the SRL with the selective binding of EF-G over EF-Tu in pretranslocation state ribosomes (17). Additionally, movement of the GAC has also been implicated in EF binding specificity (18). In our analysis, there were no differences in the reactivity of residues in the GAC or the SRL in the presence of SBP2 (data not shown). Instead, we consistently observed an SBP2-dependent enhancement in the reactivity of two nucleotides in H89 ( Fig. 2A). Because the reactive residues observed in H89 were in close proximity to the end of the primer, we repeated the primer extension with a second primer located further downstream and confirmed the enhanced reactivity at these positions (Fig. 2B). To confirm the specificity of these results, we performed a control experiment with the ribosome bindingdefective SBP2 mutant 647 RFQDR 651 , which lacks Sec incorporation activity as well as the ability to interact stably with ribosomes. In the presence of 647 RFQDR 651 , the reactivity of the aforementioned residues was significantly reduced (compare Fig. 2, A with C). As an additional specificity control, we tested the ability of ribosomal protein L30 (rpL30) to induce similar modifications. Both SBP2 and rpL30 belong to the L7Ae/L30 protein family, and both have been shown to competitively bind to the SECIS element in vitro (8). In the presence of rpL30, no enhanced sensitivity to BzCN was detected (compare Fig. 2, B  and D). To identify the exact nucleotide positions in H89 that are affected, cDNA fragments were generated with a radioactively labeled primer and resolved on a 6% sequencing gel with dideoxy sequencing reactions performed in parallel (Fig. 3A). The residues in H89 displaying enhanced reactivity in the pres-ence of CTSBP2 were C4073, and U4071. The location of these residues in the mouse 28 S rRNA secondary structure map and on the high resolution structure of the 70 S ribosome from Thermus thermophilus is indicated in Fig. 3, B and C. We specifically chose this three-dimensional structure because it contains the post-translocation state of the ribosome with the EF-Tu TC in the A-site and thus represents the conformational state of the ribosome that the eEFSec TC is expected to interact with during Sec incorporation. The structure depicted in Fig.  3C has the archaeal eEFSec homolog, aSelB, and the human Sec-tRNA in place of the EF-Tu TC.
Expansion Segment Modifications-Mammalian 28 S rRNA is ϳ1800 nucleotides longer than that in bacteria, and these "extra" highly divergent sequences are termed expansion segments. One of these lies between the PTC and the L1 stalk in eukaryotes and thus was covered by our screen of that region. Interestingly, two striking SBP2-dependent enhancements in BzCN reactivity were detected (Fig. 4A). In control experiments, neither the 647 RFQDR 651 mutant nor rpL30 was able to induce the same enhancements (Fig. 4, B and C). We were unable to obtain unambiguous sequencing data within ES31, presumably due to the high GC content in this region (Ͼ80% in mouse 28 S rRNA). Comparison of the reactive nucleotide positions in H89 identified by gel electrophoresis with those obtained by capillary electrophoresis using the LIZ size standards showed a small margin of error up to 6 bases, which suggests that the latter provides a reliable approximation of the reactive nucleotide position. Based on this observation, we estimate the reactive residues in ES31 to be ϳG3802-C3812 and ϳG3758 -U3768. Fig. 5 illustrates the relative position of H89 A, sequencing lanes are indicated by A, U, C, or G and were generated from rRNA extracted from ribosomes. Primer extension reactions were generated from rRNA extracted from ribosomes incubated with DMSO, or BzCN in the presence or absence of CTSBP2. Residues U4071 and C4073 (E. coli U2473 and C2475, respectively), which display enhanced reactivity in the presence of CTSBP2, are indicated by the blue box. B, mouse 28 S secondary structure map of H89 with bases U4071 and C4073 indicated. C, PyMOL-generated image of the eEFSec homolog, archaeal SelB (Protein Data Bank code 1WB3), and the human tRNA Sec (Protein Data Bank code 3A3A), aligned over the crystal structure of the EF-Tu TC bound to the 70 S ribosome (2WRQ and 2WRR) using the align command. Domain I (orange), II (teal), III (magenta), and IV (blue) of eEFSec are also highlighted. H89, which is projected from the PTC (white), is colored in red, and residues U2473 and C2475 are shown as blue spheres. The P-site tRNA (dark gray) and the SRL (purple) are also indicated for reference.

SBP2 Alters Ribosome Conformation
and ES31 on the crystal structure of the yeast ribosome (29). Although ES31 sequence is not conserved in eukaryotes, the size and overall secondary structure are similar.
SHAPE Reactivity in H89 and ES31 with SID/RBD-We have shown previously that when the SID and the RBD of SBP2 are expressed as separate proteins they retain near wild-type levels of Sec incorporation activity despite losing the ability to stably bind the ribosome in vitro (9). To determine whether the SID and the RBD could interact transiently with the ribosome, we repeated our SHAPE analysis of H89 and ES31 in the presence of wild-type SID and RBD, together or individually. In addition, we tested the 647 RFQDR 651 to penta-alanine (RBD-647 RFQDR 651 ) mutant version of RBD as a control for specificity. When wild-type SID and RBD were incubated with ribosomes, we observed a subtle, but reproducible increase in the reactivity of nucleotides U4071 and C4073 that was absent when the same experiment was repeated with wild-type SID and RBD-647 RFQDR 651 (compare Fig. 6, A with B). However, wild-type SID and RBD promoted a strong increase in the reactivity of ES31 that was completely abolished when the ribosomes were incubated with wild-type SID and RBD-647 RFQDR 651 (compare Fig. 6, C with D). When ribosomes were incubated with SID alone, no changes in nucleotide reactivity were observed in H89 or ES31 (data not shown). Interestingly, however, the RBD alone was sufficient to specifically enhance the reactivity of nucleotides in ES31 (compare Fig. 7, A  with B). These results confirm that when present in trans, the SID and RBD are still able to interact with the ribosome, despite our inability to detect a stable interaction in vitro.

DISCUSSION
Canonical aa-tRNA Selection and Sec Incorporation-A controversial and unresolved mechanistic question regarding Sec incorporation is how eEFSec delivers the Sec-tRNA Sec to the ribosome selectively at the UGA codons present in selenoprotein mRNAs. In other words, what prevents the random incorporation of Sec at UGA nonsense codons, and how do the other components of the Sec incorporation complex, namely SBP2 and/or the SECIS element, overcome this to direct its site-specific incorporation? Unraveling this conundrum is essential to our understanding of selenoprotein synthesis and how this process is further regulated by additional cis-and trans-acting factors in vivo.
The canonical aa-tRNA selection pathway has been extensively characterized in prokaryotes and is predicted to follow a similar mechanism in eukaryotes (20). This multistep pathway begins with the docking of a TC to the factor binding site of the ribosome, which allows the aa-tRNA to sample the A-site decoding center in a distorted conformation referred to as the A/T state (21)(22)(23)(24). Cognate codon-anticodon interactions induce conformational changes in the ribosome and the TC that lead to GTPase activation (25). Subsequently, ribosomestimulated GTP hydrolysis induces domain rearrangements in   MARCH 23, 2012 • VOLUME 287 • NUMBER 13

JOURNAL OF BIOLOGICAL CHEMISTRY 10669
the EF that decreases its affinity for aa-tRNA, resulting in its departure from the ribosome (26). This allows the aa-tRNA to move through a corridor formed by H89 -H92 of the large subunit rRNA and into the PTC, where peptide bond formation occurs (27). This newly acquired aa-tRNA conformation is designated the A/A state, and the transition from the A/T to the A/A state is referred to as accommodation. During accommodation, the codon-anticodon interaction is interrogated once again in a separate proofreading step where near cognate aa-tRNAs that are retained during the initial selection step are rejected from the A-site.
When the inability of the eEFSec TC to randomly suppress UGA codons is put into this context, it becomes clear that at least one of the steps in the aa-tRNA delivery pathway is regulated. During canonical translation elongation, the fidelity of aa-tRNA selection is controlled largely by the identity of the anticodon and the induced fit conformations that follow codon recognition (28). Because the anticodon in the Sec-tRNA Sec is cognate with respect to the UGA codon, it is likely that regulation occurs prior to codon recognition (i.e. TC binding), and/or directly after (i.e. GTPase activation or accommodation).
Our SHAPE analysis showed a specific set of SBP2-dependent changes in the reactivity of nucleotides ϳG3758 -U3768 and ϳG3802-C3812 in ES31, and U4071 and C4073 in H89, consistent with a conformational change in these regions. Interestingly, our data indicate that the RBD alone was suffi-cient for the enhanced reactivity in ES31, but we see a large discrepancy in the reactivity in H89 when comparing CTSBP2 and the SID/RBD. Our interpretation for this result is that the putative conformational change in H89 is dependent on an allosteric signal between the SID and the RBD, which occurs more readily when the two domains are physically linked, thus increasing the likelihood of capturing this conformational state. Below, we speculate on the functional relevance of these SBP2induced conformational rearrangements.
Expansion Segment 31 and E-site/A-site Allostery-ES31 is a eukaryotic-specific expansion segment located near the E-site of the ribosome. The functional role of ES31 during translation is unknown, but the crystal structure of the Saccharomyces cerevisiae 80 S ribosome indicates that it forms part of a eukaryotic-specific intersubunit bridge (eukaryotic bridge 8) with a previously unidentified small subunit ribosomal protein near the L1 stalk (H76 -H78) (29). This intersubunit bridge was also observed in the structure of the mammalian 80 S ribosome obtained by cryoelectron microscopy (30). The L1 stalk has been implicated in the movement and release of deacylated tRNAs from the E-site (31,32). In addition, E-site clearance is allosterically modulated by the tRNA status of the A-site (33). This reciprocal linkage between the A-and E-sites initially described in bacteria has also been observed in the ribosomes of higher eukaryotes (34). Given its location near the L1 stalk, the authors suggest that eukaryotic bridge 8 may be important in restoring the ribosome conformation competent for aa-tRNA binding in eukaryotic ribosomes (30). Consistent with the allosteric communication between the A-and E-sites, a deletion that shortened the size of H89 led to an increase in the reactivity of nucleotide A2225 (Escherichia coli numbering/mouse A3822) in H75 (35). Interestingly, H75 is located at the base of ES31 in eukaryotes (Fig. 1), further implicating this region in A site/E site allostery.
Helix 89 and aa-tRNA Accommodation-H89 is located near the A-site pocket where it forms part of the accommodation corridor that directs the movement of aa-tRNAs to the PTC following GTP hydrolysis. A computer simulation of the accommodation pathway revealed that nucleotide U2473 (E. coli numbering/mouse U4071) in H89 makes direct contacts with the elbow region of the aa-tRNA in the early stages of the A/T to A/A transition and continuously associates with the backbone of the aa-tRNA throughout this process (27). Yet another simulation showed that H89 is displaced during accommodation, and it was suggested that residues in this helix can acquire alternative conformations for different aa-tRNAs (36,37). Comparison of x-ray structures of the 70 S ribosome with cognate or near cognate tRNA reveal a network of rRNA helices (including H89) and ribosomal protein contacts that stabilize the elbow of cognate tRNA in the A/A state (38). The absence of these stabilizing contacts when a near cognate tRNA is bound to the A-site may explain why these tRNAs are rejected during proofreading. Unlike canonical aa-tRNAs, the Sec-tRNA Sec is unusually large, has a reduced number of posttranscriptional modifications, a long variable arm, as well as a highly reactive selenium atom. Given its unique nature, a possible function for SBP2 may be to modulate the conformation of H89 to allow the accommodation and stable binding of the Sec-tRNA Sec to the ribosomal A-site.
Helix 89 and GTPase Activation-Aside from its function in tRNA accommodation, H89, which is projected from the PTC toward the elongation factor binding site, also interacts with IF2 and its eukaryotic homolog eIF5B (39,40). Chemical probing studies show that IF2 protects nucleotides A2476 and A2478 (E. coli numbering/mouse A4074 and A4076) on the apical loop of H89, and cryo-EM studies indicate that these contacts are mediated through domain IV of IF2 (39,41). Interestingly, a C2475G (mouse C4075G) mutation in H89 caused a decrease in the affinity of IF2 for the ribosome but increased its GTP hydrolysis activity; this same mutation had no impact on the function of the elongation factors EF-Tu and EF-G (35). This indicates that interactions between IF2 domain IV and H89 can influence the GTPase activation of IF2, possibly by modulating the interaction between the G-domain and the SRL. This interdomain communication is not unprecedented because the nucleotide state of the G-domain has been shown to alter the position of IF2 domain IV nearly 90 Å away (42).
The crystal structure of aSelB revealed that, despite its sequence homology with EF-Tu in domains I-III, it has an overall shape that strikingly resembles that of IF2/eIF5B (43). Interestingly, both eEFSec and IF2 recognize a specialized tRNA (Sec-tRNA Sec and fMet-tRNA Met , respectively), and both factors have an additional, flexible, C-terminal extension (domain IV) relative to EF-Tu. Despite the low sequence homology between domain IV in IF2 and aSelB, these domains are structurally, and thus possibly functionally similar.
This information is suggestive of a role for SBP2 in modulating the conformation of H89 and promoting the ribosome-dependent GTPase activation of eEFSec through its fourth domain. This possibility is appealing because it may allow SBP2 to enhance eEFSec function without interfering with the activity of the canonical elongation factors during translation. Further evidence, albeit indirect, for GTPase activation being a limiting step, comes from the Sec incorporation mechanism in prokaryotes where interactions between the prokaryotic SECIS and SelB domain IV are required for its robust ribosome-dependent GTPase activation (44). It is important to note, however, that the fourth domains of eEFSec and SelB share no sequence or structural homology, and unlike the prokaryotic SECIS, the eukaryotic SECIS is predicted to be a kink-turn (19). In addition, despite multiple attempts, we have been unable to observe ribosome-or SBP2-dependent stimulation of eEFSec GTPase activity, likely because eEFSec lacks an intrinsic ribosome binding activity. Thus, the fundamental and so far intractable barrier to testing the mechanism of SBP2 in Sec incorporation is the lack of a Sec incorporation system composed of purified components. A, electrophoretic traces obtained from reactions containing ribosomes treated with DMSO (blue), or BzCN in the presence (red) or absence (green) of the RBD. cDNA fragments were generated from reactions using primer 3928 to cover ES31. B, same as A except reactions were performed with the RBD 647 RFQDR 651 penta-alanine mutant (purple). Arrows indicate the same nucleotide positions displaying enhanced reactivity in the presence of CTSBP2 seen in Fig. 3A. Each trace is representative of three independent SHAPE experiments.

CONCLUSION
Our SHAPE analysis of the SBP2-ribosome interaction has led to the identification of a specific set of rRNA residues in H89 and ES31 that are modified by SBP2 as reflected by their increased reactivity to BzCN. As discussed above, the literature on H89 provides interesting insights on the putative ribosomal function(s) of SBP2 during Sec incorporation. Although the scarcity of information on ES31 makes it difficult to predict the relevance of modifications in this region, the reciprocal linkage between the A-and E-sites provide a rational explanation for the observed modifications in this region despite their large distance from the A-site. These results invoke a more dynamic role for SBP2 during Sec incorporation than that of a static anchor that simply recruits eEFSec to the SECIS element. The mechanistic framework that is beginning to emerge is consistent with a ribosome-based function for SBP2, but how this fits in with its role in SECIS binding and eEFSec recruitment is not clear. A deeper understanding of the basic mechanism of eukaryotic Sec incorporation awaits elucidation of the precise functional role of both of these interactions and the interplay between them.