The SBP2 protein central to selenoprotein synthesis contacts the human ribosome at expansion segment 7L of the 28S rRNA

Hydroxyl radical probing of the 28S rRNA shows protection of discrete regions in expansion segment ES7L-E in SBP2-bound 80S and 60S human ribosomes

We next sought to determine the regions of the 28S rRNA that are protected by SBP2 in the ribosome. To do so, hydroxyl radical probing of the 28S rRNA was carried out with the 80S•CTSBP2 and 60S•CTSBP2 complexes. Hydroxyl radicals attack the C4′ atom of exposed ribose residues and thus induce strand scissions in the rRNA phosphodiester bonds (Wu et al. 1983). Positions of the scissions were detected by primer extension of 5′ ³²P-labeled oligodeoxyribonucleotides. Almost all of the about 5000 nucleotides of the human 28S rRNA sequence were analyzed, with the exception of three regions: 50 nucleotides and 80 nucleotides at the very 5′ and 3′ ends, respectively, and 230 nucleotides in expansion segment ES27L; this was due to the presence of highly GC-rich regions impeding primer hybridization. The location of the 28S rRNA regions protected by CTSBP2 against hydroxyl attack was mapped on the 28S rRNA 2D structure model derived from the cryo-EM structure of the human ribosome (Anger et al. 2013). The footprints reside in two regions of expansion segment ES7L whose accessibilities to hydroxyl radical attack were altered in the 60S•CTSBP2 complex vs. the free 60S subunits: Both regions were highly accessible in the free 60S subunits but displayed decreased accessibilities in the presence of CTSBP2 (Fig. 3A, cf. lane 2 with 3 and lane 9 with 10). These regions are 1133–1141, 1173–1186, and 1189–1190 at the apex of helices ES7L-F and ES7L-E, respectively (Fig. 3C). Using the SHAPE strategy, which leads to acylation of accessible 2′OH ribose residues, Caban and Copeland (2012) reported that SBP2 binding to the ribosome altered the reactivity of certain residues encompassing positions C4421 and U4419 in helix H89 and 4080–4166 in ES31L (numbering according to Anger et al. 2013). However, our footprinting experiments showed no differences in the accessibilities to hydroxyl radicals of these regions in the 60S•CTSBP2 complex (Supplemental Fig. S1), implying that at least the C4′ atoms of riboses (which are sensed by OH radicals) in those regions of H89 and ES31L are not involved in the CTSBP2-induced conformational changes. The same hydroxyl radical treatment was performed with the 80S•CTSBP2 complex to ask whether association of the 40S subunit could change the pattern of SBP2 protection on the 28S rRNA. To establish specificity of the SBP2 interactions with the ribosome, the same treatment was carried out with the 80S in the presence of the SBP2 mutant (MutSBP2) containing AAA^529–531 instead of TSL^529–531 and which is unable to bind the ribosome (Takeuchi et al. 2009). Supplemental Figure S2A (lane 8) showed that MutSBP2 was unable to generate a footprint, indicating that the footprints observed with ES7L-E were indeed due to the specific SBP2-ribosome interaction. Positions 1133–1141 in ES7L-F were not protected in the 80S ribosomal complex (Supplemental Fig. S2B); lack of protections in this region is likely due to structural rearrangements following subunit association. Again, as with the 60S•CTSBP2 complex, no change in the protection pattern was detected either in H89 or ES31L (Supplemental Fig. S2C,D).

Open in a separate window

FIGURE 3.

Direct hydroxyl radical probing of the human 28S rRNA in 60S•CTSBP2 (A) and 80S•CTSBP2 (B) complexes. The sites of the hydroxyl radical-induced cleavages were mapped on the 28S rRNA by reverse transcription; reverse transcriptase stopped 3′ to the nucleotides whose ribose moieties were subjected to hydroxyl radical attack. Lanes 2,3,9,10 contained hydroxyl radicals (^•OH). Lanes 1,8 lack hydroxyl radicals; lanes 2 and 9 lack CTSBP2. The extent of protection by CTSBP2 is shown on the left by a vertical bar indicating also the name of the protected helix. (C) The sites of protection identified in A and B are mapped onto the secondary structure of the 28S rRNA expansion segment ES7L (Anger et al. 2013); a blow-up of helices E and F of ES7L is displayed in the inset. Protections observed in both the 60S•CTSBP2 and 80S•CTSBP2 complexes are shown as closed circles, those found only in the 60S•CTSBP2 complex as open circles. Larger circles indicate higher protection. U,G,C,A lanes are sequencing markers.

Diepoxybutane cross-linking validates the CTSBP2–ES7L-E interaction

To validate the hydroxyl radical probing data, we asked whether ES7L-E is in close proximity to CTSBP2 in the 80S•CTSBP2 complex. Besides, as H89 was found to adopt a conformation change upon CTSBP2 binding (Caban and Copeland 2012), we sought to determine whether CTSBP2 contacts H89. To answer these questions, we analyzed by reverse transcription the 28S rRNA isolated from 80S ribosomes treated by diepoxybutane in the presence of either wild-type CTSBP2 or the SBP2 mutant MutSBP2. Nucleotides C1183, G1189, G1194, and G1195 were cross-linked to CTSBP2 (Fig. 4, cf. lanes 6 and 7). These positions overlap the area found to be protected against hydroxyl radical attack. The lack of reverse transcription stops with MutSBP2 at these positions argues for the specificity of the cross-links (Fig. 4, cf. lanes 7 and 8). Taking into account that diepoxybutane forms near 4 Å–length RNA-protein cross-links, one can conclude that the four identified nucleotides tightly surround CTSBP2 on the ribosome or even contact it. Remarkably, no nucleotide was found to cross-link in the H89 region (which displayed increased accessibility toward benzoyl cyanide in the 80S•CTSBP2 complex) (Caban and Copeland 2012), and in the ES7L-F region protected by CTSBP2 from hydroxyl radical attack in the 60S• CTSBP2 complex (Supplemental Fig. S3A,B).

Open in a separate window

FIGURE 4.

CTSBP2–ES7L-E cross-linking with diepoxybutane. Modified nucleotides are marked on the right of the gels. The modification site is one nucleotide prior to the stop (e.g., the stop observed at A1184 means that C1183 was modified); when the reverse transcription stop was observed at a G, this position was considered as modified because diepoxybutane reacts with the N7G atom which is not involved in Watson–Crick base-pairing: Reverse transcriptase only pauses. (Lane 7) Wild-type CTSBP2. Lane 8 contained the SBP2 mutant version MutSBP2. Lanes 6,7,8 contained diepoxybutane (Deb). Lane 5 lacked Deb. U,G,C,A lanes are sequencing markers. Cross-linked nucleotides (shown in gray circles) were mapped onto the secondary structure of ES7L-E (Anger et al. 2013).

Chemical probing of helices ES7L-E and H89 reveals conformation changes upon SBP2 binding to the human 60S subunit

The protections against hydroxyl radical attack observed at the sugar-phosphate backbone of helix ES7L-E led us to ask whether changes in the accessibility of the chemical groups of the bases could also be observed in this helix upon SBP2 binding to the 60S subunit. To this end, we performed chemical probing of the 28S rRNA in the 60S•CTSBP2 complex with kethoxal or dimethylsulfate (DMS). Kethoxal treatment creates a cyclic adduct between the N1 and N2 positions of unpaired guanines, DMS methylates N1 of adenines, and N3 of cytosines when they are unpaired as well. Modified positions were identified by the primer extension method. Figure 5A shows the result of the DMS and kethoxal treatments at ES7L-E. Three positions were modified by DMS, C1180, C1181, and C1182. C1182 is highly modified in the free and complexed subunit; however, only C1180 and C1181 displayed an increased reactivity in the 60S•CTSBP2 complex vs. the free subunit (Fig. 5A, cf. lanes 6 and 7). G1185 reactivity toward kethoxal is increased in the 60S•CTSBP2 complex compared with the free subunit (Fig. 5A, cf. lanes 13 and 14). As for ES7L-F, it displayed the same reactivity toward DMS and kethoxal in both the free 60S subunits and 60S•CTSBP2 complexes (Supplemental Fig. S4A).

Open in a separate window

FIGURE 5.

Chemical probing of the 28S rRNA in 60S•CTSBP2 complexes. Reverse transcription analysis of the 28S rRNA in the ES7L-E and H89 regions is displayed in A and B, respectively. Modified nucleotides (one nucleotide prior to the reverse transcriptase stops) are marked on the right of the gels. Reactions contained DMS (A,B, lanes 6,7) or kethoxal (Ket: A, lanes 13,14; B, lanes 9,10). Control lanes 5,12 (A) and 5,8 (B) did not contain DMS or kethoxal; CTSBP2 was not added in lanes 6,13 (A) and 6,9 (B). Modified positions are mapped onto the secondary structures of ES7L-E and H89 (Anger et al. 2013). Bases modified by DMS are in open circles, bases displaying increased reactivity toward DMS are displayed in a dark background; G1185 (boxed in ES7L-E) has an increased accessibility toward kethoxal modification. U,G,C,A lanes are sequencing markers. A gel compression led to stacking of the CG sequence at positions 1188–1189; resolution is better with the kethoxal gel.

As mentioned above, no protection against or increased reactivity toward hydroxyl radical attack was detected in ES31L or helix H89 upon binding of CTSBP2. As hydroxyl radicals attack the C4′ ribose, our findings do not rule out the possibility that chemical groups of bases in H89 and ES31 have their accessibility altered upon SBP2 binding. Chemical probing was therefore performed to examine this hypothesis. ES31L displayed the same reactivity toward kethoxal and DMS in both the free 60S subunits and 60S•CTSBP2 complexes, implying that the bases in this region are not involved in the CTSBP2-induced conformational changes of the 28S rRNA observed in Caban and Copeland (2012) (Supplemental Fig. S4B). In H89, A4414, A4415, A4422, A4424, and A4428 were modified by DMS (Fig. 5B). Among these, reactivities of A4414 and A4422 were higher in the 60S•CTSBP2 complex than in the free 60S subunit (Fig. 5B, cf. lanes 6 and 7), as a direct or indirect consequence of SBP2 binding to the ribosome. No significant increase of kethoxal modification was observed upon SBP2 binding (Fig. 5B, lanes 9,10).

Structure-based sequence alignment reveals the extreme evolutionary conservation of a part of helix ES7L-E 2D structure

The footprints that were observed in ES7L-E in the presence of CTSBP2 raised the question of whether the 2D structure of this helix is conserved in selenoprotein-making organisms to enable SBP2 to bind. To answer this question, we analyzed the sequences of several 28S rRNAs between positions 956 and 1284 encompassing helices ES7L-E and ES7L-F (numbering according to the human 28S rRNA sequence from Anger et al. [2013]). Sequences that could not be manually folded as ES7L-E because of high-sequence dissimilarities and/or the presence of blocks of extended deletions, were discarded. For example, the sequences of Anopheles gambiae (Cannone et al. 2002), Plasmodium falciparum (Cannone et al. 2002), Drosophila melanogaster (Anger et al. 2013), and Tetrahymena thermophila (Klinge et al. 2011) could not be considered because they possess much shorter ES7L sequences. Twelve organisms (10 vertebrates, one echinoderm, and one mollusk) were submitted to LocARNA (Will et al. 2007), a web-based tool, which simultaneously folds the input sequences and aligns them. Figure 6, A and B, reveals the perfect evolutionary conservation of the 2D structure of helix ES7L-E between positions 1170 and 1191 owing to a number of base covariations. The sequence of the G1171–C1190 and C1176–G1185 base pairs (human numbering) is conserved in the 12 organisms. Such a 2D structure conservation is indicative of a functional role of this helix.

Open in a separate window

FIGURE 6.

Structure-based alignment of expansion segment ES7L-E sequences in 28S rRNAs arising from 12 vertebrates and invertebrates. (A) The region boxed shows 2D structure conservation through base covariations of a section of helix ES7L-E. Color code: The hues show sequence conservation (red corresponds to absolutely conserved and green indicates the least conserved base pairs) and saturation decreases with the number of compatible base pairs, thus showing structural conservation (generated automatically by LocARNA). Parentheses indicate base pairs. Because most of the sequences are partial, numbering according to the human sequence was not possible. The first nucleotide in the sequence of Homo sapiens corresponds to C1148. (B) Secondary structure model of a section of the human ES7L-E helix supported by the base covariations shown in A. The number of covariations for each base pair is indicated on the right. Color code as in A and numbering as in Figure 5A. Open circles represent nonconserved nucleotides in the apical loop and base pairs that do not covary.

Purification of human ribosomal subunits

Human ribosomal 40S and 60S subunits were isolated from full-term placenta as described in Matasova et al. (1991). The concentration of subunits was determined assuming that one A₂₆₀-unit equals 50 pmol and 25 pmol in the case of 40S subunits and 60S subunits, respectively (Matasova et al. 1991). The activities of purified ribosomes were validated by poly-U-dependent synthesis of ¹⁴C-polyphenylalanine in vitro.

Formation of CTSBP2•ribosome complex

The 0.5 μM 60S subunits were incubated with 1.5 μM CTSBP2 in 50 μL of buffer A (20 mM HEPES-KOH at pH 7.5, 100 mM KCl, 2 mM MgCl₂, 1 mM DTT) at 25°C for 30 min. The complex was isolated by centrifugation in 10%–30% sucrose gradient in buffer A (rotor SW41 [Beckman Coulter], 23,000 rpm, 17 h, 4°C). Proteins in fractions containing ribosomal subunits were TCA precipitated. Pellets were dissolved and analyzed on 12% SDS-PAGE. After electrophoresis, the gel was stained with Coomassie BB R250 and analyzed using QuantityOne software.

Diepoxybutane cross-linking

CTSBP2•ribosome complexes were formed in 50 μL of buffer A as described above. Cross-linking was initiated by addition of 0.5% (v/v) diepoxybutane. The mixture was incubated for 45 min at 37°C. The cross-linking reaction was quenched by addition of 50 mM Tris-HCl (pH 7.5). After cross-link formation, excess of uncross-linked CTSBP2 was removed by centrifugation in 10%–30% sucrose gradient in buffer A containing 450 mM KCl (rotor SW41, 23,000 rpm, 17 h, 4°C). Fractions corresponding to 60S and 40S subunits were ethanol precipitated, dissolved in buffer A, blotted onto nitrocellulose membrane, and analyzed for CTSBP2 content with rabbit polyclonal anti-SBP2 antibodies, which were shown beforehand not to cross-react with ribosomal proteins (1/2500 dilution). Membranes were then treated with anti-rabbit HRP-conjugated secondary antibody (1/10000 dilution), revealed with the ECL-plus kit (GE Healthcare) and exposed to either X-ray film (GE Healthcare) or ChemiDoc XRS (BioRad). To analyze the cross-link distribution between rRNAs and proteins in 60S subunits, the corresponding subunit fraction prepared as described above was divided into two halves. One half was used for protein extraction with acetic acid according to Hardy et al. (1969), and total rRNA was isolated from another half by phenol extraction as in Graifer et al. (1994). Both ribosomal protein and total rRNA fractions were analyzed with anti-SBP2 antibody as described above. To analyze cross-link distribution between rRNAs in the 60S subunit, the corresponding subunit fraction prepared as described above was incubated for 30 min at 37°C in 0.5% SDS, 5 mM NaOH-EDTA (pH 7.5). The mixture was layered onto a 5%–20% sucrose gradient in 20 mM Tris-HCl (pH 7.5), 2 mM EDTA, 0.1% SDS, and centrifuged for 17 h at 27,000 rpm (rotor SW41). After centrifugation, fractions were ethanol precipitated, the pellets dissolved in a minimal amount of water, and blotted onto a nitrocellulose membrane. The SBP2 signal was detected as above.

Alternatively after quenching, the mixture was supplemented with 0.1% SDS, 4 mM EDTA, 0.05 mg/mL proteinase K, and incubated for 20 min at 37°C. rRNA was isolated by phenol extraction. Reverse transcription and product analysis were carried out as described in Malygin et al. (2013).

2-iminothiolane cross-linking

CTSBP2•ribosome complexes were formed as described above. CTSBP2-rRNA cross-links and SBP2-ribosomal proteins cross-links with 2-iminothiolane were performed as described (Kenny et al. 1979; Brimacombe et al. 1988). CTSBP2-rRNA cross-links and CTSBP2-ribosomal proteins cross-links were analyzed as described above.

Hydroxyl radical, kethoxal, and DMS reactions

Hydroxyl radical cleavage of 28S rRNA in 60S or 80S ribosomes (0.5 µM) and their complexes with CTSBP2 (1.5 µM CTSBP2) was performed as described in Malygin et al. (2013). Kethoxal and dimethylsulfate (DMS) modifications were performed according to Xu and Culver (2009) with the only modification that the final DMS concentration was 0.15% (v/v). The RNA was isolated by phenol extraction. For reverse transcription, 5′-³²P-labeled primers complementary to the human 28S rRNA sequence regions 196–212, 394–413, 508–525, 693–709, 735–753, 948–968, 1178–1195, 1221–1240, 1298–1315, 1435–1454, 1589–1605, 1792–1809, 1991–2008, 2160–2178, 2313–2331, 2406–2422, 2618–2639, 2795–2811, 2889–2907, 3047–3064, 3260–3277, 3626–3644, 3809–3829, 4036–4055, 4184–4200, 4361–4378, 4495–4515, 4540–4558, 4721–4739, 4880–4898, and 5010–5028 were used. Reverse transcription and product analysis were carried out as described in Malygin et al. (2013).

We thank C. Allmang (IBMC CNRS, Strasbourg, France) for the gift of materials and helpful discussions and Prof. E. Westhof (IBMC CNRS, Strasbourg, France) for his input and comments on the manuscript. We are grateful to A. Schweigert and S. Baudrey for technical assistance. O.K. was a cosupervised PhD student with the Institute of Chemical Biology and Fundamental Medicine, SB RAS, and the Université de Strasbourg, partly supported by the French government. This work was supported by the ARCUS and Supramolecular Chemistry programs (to G.K. and A.K.), the CNRS Laboratoire International Associé LIA NUCPROT (to A.K.), and the Russian Foundation for Basic Research (grant 12-04-93111-CNRSL_a to G.K.).