Structural analysis reveals the characteristic features of Mtr4, a DExH helicase involved in nuclear RNA processing and surveillance

Abstract

Mtr4 is a conserved RNA helicase that functions together with the nuclear exosome. It participates in the processing of structured RNAs, including the maturation of 5.8S ribosomal RNA (rRNA). It also interacts with the polyadenylating Trf4-Air2 heterodimer to form the so-called TRAMP (Trf4-Air2-Mtr4 Polyadenylation) complex. TRAMP is involved in exosome-mediated degradation of aberrant RNAs in nuclear surveillance pathways. We report the 2.9-Å resolution crystal structure of Saccharomyces cerevisiae Mtr4 in complex with ADP and RNA. The structure shows a central ATPase core similar to that of other DExH helicases. Inserted in the DExH core is a region characteristic of Mtr4 orthologues that folds into an elongated stalk connected to a β-barrel domain. This domain shows unexpected similarity to the KOW domain of L24, a ribosomal protein that binds 23S rRNA. We find that indeed the KOW domain of Mtr4 is able to bind in vitro transcribed tRNA^iMet, suggesting it might assist in presenting RNA substrates to the helicase core. The interaction of Mtr4 with Trf4-Air2 is mediated not by the stalk/KOW insertion but by the DExH core. We find that in the context of the TRAMP complex, the DExH core functions independently in vitro as an RNA helicase and a protein-binding platform. Mtr4 has thus evolved specific structural and surface features to perform its multiple functions.

The eukaryotic RNA exosome is an essential, ubiquitous, and conserved protein complex that acts in virtually all pathways where ribonucleic acids undergo 3′–5′ processing or degradation (reviewed in ref. 1). In particular, the exosome participates in the discrete trimming during the processing of structured RNA precursors, in the turnover of both precursor and mature coding RNAs and in quality control pathways that detect and degrade aberrant RNA molecules (reviewed in ref. 2). The exosome is a complex of 10 different subunits (reviewed in ref. 3). Work on the yeast exosome proteins has shown that the RNase activities of the core complex reside in a single subunit, Rrp44, which contains a 3′–5′ exoribonuclease activity in the C-terminal domain and an endonucleolytic cleavage activity in the N-terminal domain (4–8). The other nine core subunits are catalytically inert, but are nevertheless essential for the viability of yeast cells (9). The catalytically inactive core of the eukaryotic exosome has a ring-like architecture (4) that functions both to recruit auxiliary proteins and to bind RNA substrates by threading them to the Rrp44 nuclease (reviewed in refs. 2 and 3). An important consequence of this arrangement of subunits is that the architecture restricts and fine-tunes access to the exonuclease site (10). Indeed, the relatively poor ribonuclease activity of the exosome in vitro has been observed since its discovery (9), prompting the quest for activating factors that would account for the rapid and processive activity that the complex shows in vivo.

Mtr4 (also known as Dob1p) is an auxiliary factor required for most of the functions of the exosome in the yeast nucleus (reviewed in refs. 11 and 12). It is an RNA-dependent ATPase that is localized both in the nucleolus and nucleoplasm (13, 14). In vitro, Mtr4 is capable of unwinding RNA duplexes in the 3′–5′ direction (15, 16). In vivo, mutation or depletion of Mtr4 results in defects in ribosome biogenesis similar to those observed in strains depleted of the exosome associated nuclease Rrp6 or of core exosome subunits (14). In particular, the absence of functional Mtr4 impairs the trimming of the 7S rRNA precursor to the mature 5.8S rRNA and the degradation of an excised pre-rRNA spacer fragment (14). A similar accumulation of 3′-extended 5.8S rRNA precursors is observed upon knockdown of the orthologue in human cells (17). Mtr4 is highly conserved across eukaryotes, with 50% sequence identity shared between the yeast and human proteins.

Mtr4 participates not only in the processing of various structured RNAs including snRNAs and snoRNAs (18), but also in nuclear RNA surveillance pathways that involve the so-called TRAMP (Trf4-Air2-Mtr4 Polyadenylation) complex (reviewed in ref. 11). TRAMP is formed by the direct interaction of Mtr4 with the noncanonical poly(A)polymerase Trf4 and its cofactor, the putative RNA-binding protein Air2 (19–21). The complex acts by adding a short single-stranded poly(A) overhang to the 3′ end of aberrant or unstable transcripts such as a hypomethylated initiator tRNA (tRNA^iMet) (19, 21, 22). The helicase is thought to unwind the structured regions of tRNA^iMet and present it to the exosome as a substrate for degradation, although it is unclear how the association between Mtr4 and the exosome is achieved.

Mtr4 shares more than 30% sequence identity with Ski2, a putative RNA helicase required for all known functions of the exosome in the cytoplasm (23). Both are DExH-box ATPases with a RecA-fold core related to that of archaeal Hel308, a DNA helicase whose structure has been determined with a partially unwound DNA duplex (24). To address what are features characteristic of Mtr4 orthologues and their regulation, we have determined the crystal structure of Mtr4 in complex with RNA and characterized its mode of interaction with Trf4-Air2.

Results and Discussion

Crystal Structure Determination of Yeast Mtr4-Δ80.

We identified by limited proteolysis the stable and conserved RNA-binding region of Saccharomyces cerevisiae Mtr4 encompassing residues 81–1073 (Mtr4-Δ80) (Figs. S1 and S2). To crystallize Mtr4-Δ80, we incubated the purified protein with an A₁₀ RNA oligo and ADP∶AlF_x. We obtained crystals that diffracted to 2.9-Å resolution and solved the structure with a multiwavelength anomalous diffraction (MAD) experiment on a crystal grown from selenomethionine-substituted protein. The structure was refined to 2.9-Å resolution to an Rfree of 24.8%, Rfactor of 19.9%, and good stereochemistry (see Table S1 for data and refinement statistics). Mtr4 folds into a globular DExH ATPase core and features an insertion that forms a protruding structural unit (Fig. 1). The crystal asymmetric unit contains two independent molecules. Molecule A includes most of the Mtr4 polypeptide chain, with the exception of residues 363–391. Molecule B has well-defined electron density for the DExH core, which is essentially identical to that of molecule A. However, residues 672 to 809 in the protruding region and three loops are disordered in molecule B and could not be modeled. Five nucleotides of the A₁₀ RNA oligo and ADP are also present in both molecules of the asymmetric unit. Here, we describe molecule A unless otherwise specified.

Fig. 1.

Crystal structure of yeast Mtr4-Δ80. (A) View of the S. cerevisiae Mtr4-Δ80 structure in two orientations. The DExH core is formed by the two RecA domains (in light blue) together with the N-terminal β-hairpin (in blue), the WH domain (in yellow), and the helical bundle domain (in pink). An insert in the sequence of the WH domain folds into a long stalk (in orange) and a globular KOW domain (in red). The helices in the stalk domain are labeled at Left. ADP and five nucleotides of a single-stranded poly(A) RNA are shown (in ball-and-stick representation in black) bound at the DExH core. The N terminus and C terminus are labeled (residues 81 and 1073). This and all other structure figures were generated with PyMOL (Delano Scientific). (B) Schematic representation of the domain organization of Mtr4. The domains are colored as in A.

Architecture of the Central DExH Core.

The DExH core of Mtr4 is formed by four domains (RecA-1, RecA-2, winged-helix, and helical bundle) that interact to form a globular structure of dimensions 55 × 55 × 35 Å. The structural analysis reveals that the architecture of the core has remarkable similarity to the corresponding domains of the DExH helicases Hel308 (24) and Prp43 (25). Hence, we refer to it as the DExH core. The similarity of Mtr4 and Hel308 beyond the RecA region was not previously detected by sequence analysis probably due to the presence of the insertion in the middle of the winged-helix domain (Fig. 1B).

The two RecA-like domains (RecA-1, and RecA-2, light blue in Fig. 1) pack against each other to bind RNA and ADP. An unusual feature of Mtr4 is a 60-residue segment N-terminal to RecA-1 that folds into a long β-hairpin (dark blue in Fig. 1). The N-terminal β-hairpin packs with the β-strands against RecA-1 and latches onto RecA-2. A linker of 15 residues connects RecA-2 to the winged-helix (WH) domain (yellow in Fig. 1A). The WH domain packs against an 8-helix-bundle domain (pink in Fig. 1), which in turn contacts the cleft between the two RecA domains. A related helical bundle region was first described as the ratchet domain in the structure of Hel308 because of its proposed role in translocating the nucleic acid (24) and has subsequently been found in the RNA splicing helicases Prp43 (25) and Brr2 (26). Finally, the conserved hydrophobic residues at the C terminus of Mtr4 interact at an apolar patch between RecA-1 and RecA-2. The relative conformation of the two RecA domains of Mtr4 is stabilized on one side (near the ADP) by the N-terminal β-hairpin and on the other (near the RNA) by the helix-bundle domain and the C-terminal tail (Fig. 1A Right). Mutants known to have phenotypes in vivo (M540I in Mtr4-20 and C942Y in Mtr4-1) map to the DExH core and are likely to indirectly perturb RNA binding (Fig. S3A).

Architecture of the Stalk/KOW Insertion Domain.

Mtr4 features an insertion of 270 amino-acid residues that protrude from the globular DExH core as a separate structural unit (Fig. 1A Left). This region is inserted between the third and fourth helices of the WH domain (Fig. S1). The insert starts with two α-helices oriented roughly perpendicular with respect to each other (helices α1 and α2, orange in Fig. 1A Left). It continues into a globular domain (in red) and ends with two α-helices (α3 and α4) that are coiled around the preceding helices α1 and α2. The helical portion of the insert thus forms an L-shaped stalk extending for about 40 Å from the DExH core. The globular portion of the insert (red in Fig. 1A) has a β-barrel architecture characterized by two long loops (Fig. S1). A DALI search (27) revealed unexpected structural similarity to Tudor, SH3, and KOW (Kyrpides–Ouzounis–Woese) (28) domains (Z scores of about 4.5). Whereas the similarity to the Tudor and SH3 domains appears restricted to the overall fold, the similarity to KOW proteins extends to residues that have been implicated in macromolecular interactions (see below). We therefore refer to the globular domain of the Mtr4 insertion as a KOW domain.

In one of the Mtr4 molecules in the asymmetric unit (molecule A), the KOW domain is in close proximity to the DExH core. The two long loops of the KOW domain contact the helical bundle and the RecA-2 domains, but this intramolecular interaction is most likely due to crystal packing. In molecule B, the KOW domain is disordered. Thus, the KOW domain appears to be a flexible unit that could sample the conformational space around the end of the stalk. The stalk helices α1 and α4 that emerge from the DExH core are in a similar orientation in the two molecules of the asymmetric unit (rmsd 2.5 Å), whereas helices α2 and α3 that connect to the KOW domain deviate more in their conformation (rmsd 7.5 Å) (Fig. S3B). Although the stalk is not a rigid unit, its conformational flexibility appears to be restricted by intramolecular contacts at the hinge regions (Fig. S3B).

ADP and RNA Binding by the DExH Core of Mtr4.

The two RecA domains of Mtr4 pack against each other forming a cleft that is lined by the conserved motifs typical of the DExH/DEAD family of SF2 helicases (reviewed in ref. 29). A DALI search finds that the relative orientation of the two domains is most similar to the close conformation observed in the structures of the DExH proteins Hel308 (24) and Prp43 (25) (in complex with a partially unwound DNA substrate and with ADP, respectively) and to the DEAD-box protein Dbp5 in complex with AMPPNP and RNA (30). ADP binds at one side of the interdomain cleft, with the adenine ring sandwiched between Arg547 and Phe148 and with the adenine amino group recognized by Gln154 (that is part of the Q motif). Although the crystallization experiments were carried out in the presence of ADP-AlFx, we did not observe ordered electron density for the AlFx moiety.

RNA binds at the opposite side of the interdomain cleft, with the 5′ end at RecA-2 and the 3′ end at RecA-1 (Fig. 2). Binding of the five ribonucleotides occurs at similar binding pockets used by other DEAD/DExH helicases. In particular, nucleotides 2 to 5 in the Mtr4 structure superpose with the first four unpaired bases of the DNA substrate in the Hel308 structure (position +1 to +4) (Fig. 2). The most 5′ nucleotide (nucleotide 1) is instead in a different conformation as compared to DNA-bound Hel308, where the corresponding position in the nucleic acid is the last base pair before the duplex is unwound [position -1 (24)]. Nucleotide 1 packs against Trp524 and Gly526. These residues, together with Lys523 and Arg530, form a β-hairpin that is conserved at the same structural position in Hel308, where it promotes strand separation (24) (Fig. 2 and Fig. S4). An equivalent unwinding β-hairpin is present in the structure of the DExH helicase Prp43 (25). Thus, the overall mode of substrate binding and unwinding is likely similar in DExH helicases.

Fig. 2.

RNA binding by the DExH core of Mtr4. The structure of Mtr4 in complex with ssRNA is shown at Left (A). As a comparison, the structure of Hel308 bound to a partially unwound duplex DNA (PDB ID code 2P6R) is shown after optimal superposition at Right (B). The domains in the Hel308 structure are colored as the corresponding domains in the DExH core of Mtr4. The close-up views show the region around the unwinding β-hairpin. The numbering of the deoxyribonucleotides of the product strand in the Hel308 structure is from ref. 24. The product strand being translocated through the ATPase is in black, whereas in gray is the complementary strand in the Hel308 structure.

The DExH Core Recruits Trf4-Air2.

The presence of the stalk/KOW insertion (hereon defined as SK) is a specific feature of Mtr4 orthologues. We therefore tested whether the SK region (residues 618–873) is required for binding Trf4-Air2, a specific interacting partner of Mtr4. We expressed and purified a heterodimer containing the conserved poly(A)-polymerase domain of Trf4 (residues 111–490) and the conserved zinc-knuckle region of Air2 (residues 1–180). Both Mtr4 f.l. and Mtr4-Δ80 formed a ternary complex with Trf4^111–490-Air2^1–180 that could be isolated by size-exclusion chromatography (Fig. 3A and Fig. S2C). Next, we engineered a mutant of Mtr4-Δ80 lacking the SK region (Mtr4-Δ80-ΔSK). This mutant mimics the structure of Hel308, where a loop is present at the corresponding position of the SK insertion in the DExH core. Despite the significant removal of amino-acid residues, this mutant appears to be properly folded (melting temperature of 48 ºC for Mtr4-Δ80-ΔSK and of 46 ºC for Mtr4-Δ80, Fig. S2A). The Mtr4-Δ80-ΔSK mutant formed a ternary complex with Trf4^111–490-Air2^1–180 by size-exclusion chromatography (Fig. 3A and Fig. S2C), indicating that the DExH core of Mtr4 rather than the stalk/KOW insertion mediates the formation of the TRAMP complex.

Fig. 3.

The DExH core of Mtr4 is a binding platform for Trf4-Air2. (A) TRAMP complex formation by size-exclusion chromatography (Superdex 200 HR 10/30, GE Healthcare). The SDS-PAGE Coomassie-stained gel includes samples of the peak fractions indicated. The overlay of the corresponding chromatograms is shown in Fig. S2C. (B) The ATPase assay was performed in the presence of an A₃₅ RNA with Mtr4-Δ80 either alone (blue triangles), Trf4^{111–490 DADA}-Air2^1–180 alone (gray diamonds), or both in complex together (blue squares). The activity of Mtr4-Δ80-ΔSK is also shown (green circles). The data represent mean values and standard deviations from three independent experiments. The proteins used in the ATPase assay were analyzed by SDS-PAGE and are shown in Fig. S2D.

We next tested whether binding of Trf4-Air2 regulates the ATPase activity of Mtr4. For this experiment, we used a catalytically inactive mutant of Trf4 (DADA mutant) that retains Mtr4 binding but is unable to contribute directly to ATP hydrolysis (20) (Fig. 3B, gray diamonds). Mtr4-Δ80 showed ATPase activity in the presence of an A15 RNA substrate (Fig. 3B, blue triangles). Adding Trf4^DADA-Air2 did not affect the RNA-dependent ATPase activity of Mtr4 (Fig. 3B, blue squares). Inspection of the molecular surface reveals the presence of patches of residues conserved in Mtr4 orthologues and not in other helicases (Fig. S5). These conserved patches are far from the RNA/ADP binding sites and are possible targets for Trf4-Air2 binding.

The KOW Domain of Mtr4 Shares Similarities with rRNA-Binding Proteins.

The β-barrel structure in the insertion domain of Mtr4 shows intriguing similarities with the KOW domains present in ribosomal proteins. In addition to conserved residues important for the structural integrity of the domain, the KOW domains of bacterial L24 and eukaryotic L26 and L27 proteins contain an invariant glycine residue present in a loop region (28). Structural studies of the ribosome have shown that the invariant glycine of L24 makes a base contact with ribosomal RNA (rRNA) (31, 32). In addition, rRNA is contacted by positively charged residues present in the long loops that characterize the L24 β-barrel. The β-barrels of Mtr4 and L24 superpose with a rmsd of 2.6 Å. The superposition places Gly686 in yeast Mtr4 at the similar structural position of Gly15 in Escherichia coli L24 (yellow, Fig. 4A). Yeast Mtr4 also features conserved positively charged residues at similar positions as rRNA-binding residues of E. coli L24 (Fig. 4A).

Fig. 4.

Mtr4 has a KOW domain that contributes to RNA binding. (A) Structural similarity between the KOW domain of Mtr4 (in red) and the ribosomal protein L24 (in dark gray, PDB ID code 3I1N). E. coli L24 is shown bound to 23S rRNA (light gray). In yellow is the conserved glycine residue characteristic of KOW domains. Conserved positively charged residues of yeast Mtr4 and RNA-binding residues of L24 at similar structural positions are highlighted. Some extended loop regions outside of the KOW core have been omitted for clarity. (B) The KOW domain is involved in RNA binding. Gel-mobility assay conducted with a labeled tRNA^iMet probe and the purified recombinant proteins indicated above the lanes (at increasing concentrations of 0.45, 1.5, 45, and 150 μM protein). The position of the free RNA probe (lanes 1 and 14) and of the RNA-bound complexes (asterisk) is shown at Right. The proteins used in the gel-shift assay that were analyzed by SDS-PAGE are shown in Fig. S2B (Right).

To test whether the KOW domain of Mtr4 contributes to the RNA-binding properties of Mtr4, we used an electrophoretic gel-mobility retardation assay with in vitro transcribed tRNA^iMet, which lacks modifications and therefore mimics the physiological hypomodified tRNA^iMet substrate of Mtr4. Mtr4-Δ80 efficiently bound tRNA^iMet even in the absence of ATP (Fig. 4B, lanes 2–5), consistent with previous reports that nucleic acid binding by DExH proteins does not depend on the presence of ATP (16, 24, 29). Mtr4-Δ80-SK showed reduced RNA binding in the gel-shift assay (Fig. 4B, lanes 6–9) and a lower RNA-dependent ATPase activity (Fig. 3B, green circles) as compared to Mtr4-Δ80. When expressed and purified separately, the KOW domain (residues 666–818) bound tRNA^iMet (Fig. 4B, lanes 10–13).

Upon superposition of the Mtr4 and Hel308 structures, we find that in this crystal form the KOW domain is positioned near the duplex region of the nucleic acid bound to Hel308 (Fig. 2). In the conformation observed in molecule A, the long β3–β4 loop of the KOW domain would clash against the two unpaired bases of the unwound 5′-3′ strand (Fig. S4). However, the flexibility of the KOW domain would allow it to adopt a different conformation relative to the ATPase domain upon nucleic acid binding.

Conclusions

Mtr4 is a conserved helicase of the DExH family that cooperates with the eukaryotic nuclear exosome in RNA processing and degradation. Three aspects are particularly important for its biological functions: the binding and unwinding of RNAs, the interaction with Trf4-Air2 to form the TRAMP complex, and the interaction with the exosome. What are the features of Mtr4 that support these activities?

The RNA-binding and -unwinding activities of Mtr4 require the DExH core domain. The conformation of the two RecA domains we observe in the DExH core is similar to that previously observed with the DExH helicases Hel308 and Prp43 either in the unbound form or bound to nucleic acid or ADP (24, 25). The active conformation of the two RecA domains appears to be stabilized even in the absence of ligands by intramolecular interactions with the additional regions that form the DExH core (the N-terminal β-hairpin, the winged-helix, and the helical bundle domains). Although interdomain movements in Mtr4 are likely to occur upon ATP hydrolysis and RNA translocation, we do not expect major conformational changes of the RecA domains as those observed for example between the active and inactive states of the DEAD-box protein eIF4AIII (33, 34). The difference in the extent of conformational changes between the on and off states might rationalize why in DEAD-box proteins binding of ATP and nucleic acid is cooperative, whereas in DExH proteins it is uncoupled.

Formation of the TRAMP complex is also mediated by the DExH core. Although it might be expected that Trf4-Air2 discriminates Mtr4 from other DExH helicases by binding to the most distinctive region of the molecule (i.e., the stalk/KOW insertion), it instead binds at the most common region. This is striking because Mtr4 appears to be only a docking platform for Trf4-Air2. It has previously been reported that Mtr4 does not affect the polyadenylation activity of Trf4-Air2 (19). Here, we find that Trf4-Air2 does not affect the ATPase activity of Mtr4. Thus, the Trf4-Air2 binding and helicase properties of Mtr4 are apparently uncoupled despite being mediated by the same DExH core. The characteristic insertion of Mtr4 contributes instead to RNA binding. The insertion contains a fold that is structurally similar to KOW domains of ribosomal RNA-binding proteins. We find that the KOW domain of Mtr4 is indeed able to bind a structured RNA such as in vitro transcribed tRNA^iMet. Given the function of Mtr4 in processing structured RNAs and given its conformational flexibility around the end of the stalk domain, it is tempting to speculate that the KOW domain might assist the helicase in recognizing structured RNAs and present them to the unwinding β-hairpin of the DExH core. The Ski2 helicase is also predicted to have an insertion in the DExH core, although little similarity can be found at the sequence level with the KOW domain of Mtr4. Ski2 functions together with the exosome in the cytoplasm, and as such it encounters different RNA substrates.

Wheras the 5′ end of the short single-stranded RNA bound in the Mtr4 structure is positioned at the top of the DExH core, near the unwinding β-hairpin, the 3′ end exits at the base of the core. In the Hel308 structure, the 3′ end of the unwound product strand wraps around a helix-loop-helix (HLH) domain that packs against the DExH core. Mtr4 does not have the equivalent of the Hel308 HLH domain. We speculate that an equivalent binding to the RNA 3′ end might be exerted by exosome proteins. We find that the surface of Mtr4 from which the 3′ end protrudes is well conserved in the sequences of Mtr4 and Ski2 orthologues (Fig. S5), suggesting it might function as a docking site for a common binding partner of the two helicases. How such an interaction might occur is a question for future studies.

Materials and Methods

Protein Expression and Purification.

Mtr4-Δ80 was expressed as a 6x-His-tagged protein (cleavable with Tobacco Etch Virus TEV protease) in E. coli BL21(DE3) cells or in B834 cells for selenomethione substitution. It was purified using a Ni-nitrilotriacetate (NTA) affinity step followed by TEV cleavage and heparin Sepharose chromatography (GE Healthcare). The final purification step by size-exclusion chromatography (Superdex 200, GE Healthcare) was carried out in 20 mM Hepes pH 7.5, 150 mM NaCl, 5 mM MgCl₂, and 2 mM DTT. The same protocol was used to purify Mtr4 f.l., Δ80-ΔSK, and KOW. The complex of S. cerevisiae Trf4-Air2 was obtained by coexpression in E. coli of TEV-cleavable His-tagged Trf4^111–490 and untagged Air2^1–180. The complex was affinity-purified on Ni-NTA beads (GE Healthcare) followed by TEV cleavage, heparin and size-exclusion chromatography. The Trf4^DADA mutation was engineered with the QuikChange kit (Stratagene), verified by DNA sequencing and purified like the wild type.

Crystallization and X-Ray Structure Solution.

Mtr4-Δ80 was concentrated to 40 mg/mL after gel filtration and incubated with 1 mM ADP∶AlF_x and a 1.2 M excess of A₁₀ RNA for 10 min at 22 ºC. Crystals of the native protein were grown by sitting-drop vapor diffusion at 18 ºC using 50 mM MES pH 6.0, 200 mM ammonium acetate, and 20% (wt/vol) PEG 3350 in the reservoir. The crystals were cryoprotected with the reservoir solution supplemented with 20% ethylene glycol prior to data collection at 100 K. All diffraction data were collected at the Swiss Light Source (SLS) PXII beamline (Villigen, Switzerland) and processed using XDS (35). Phase information was obtained by a three-wavelength MAD experiment with a selenomethionine-substituted crystal. SHELXD (36) was used to locate the selenium sites and SHARP (37) to calculate the phases. The initial electron density map was improved using solvent flipping and noncrystallographic symmetry averaging with DM (38). The model was built using Coot (39) and refined using Refmac5 (40) against the native data to 2.9-Å resolution.

ATPase Assay.

The ATPase reactions were carried out in 50 mM Hepes, pH 7.5, 100 mM NaCl, 10 mM MgCl₂, 1 mM DTT, 10% (vol/vol) glycerol and 0.1 mg/mL BSA. We added 1.5 pmol of Mtr4 with or without equimolar amounts of Trf4-Air2 proteins in a 20-μL reaction volume containing 2 mM ATP, 2 pmol A₁₅ RNA, and traces of [γ-³²P]ATP. Reactions were incubated for 0, 30, 60, and 90 min at 30 ºC and stopped by adding 400 μL of ice-cold acid-washed charcoal (Sigma) in 10 mM EDTA. After 30 min of centrifugation at 16,000 g, the supernatants containing γ-³²P were counted using a Packard Tri-carb 2100TR Liquid Scintillation Analyzer.

Gel-Shift Assay.

The coding sequence for S. cerevisiae tRNA^iMet preceded by a T7 promoter sequence was cloned in a pUC19 vector by assembly of overlapping phosphorylated oligonucleotides. RNA was transcribed using the T7 RNA polymerase MEGAshortscript kit (Ambion) in the presence of [α-32P]UTP and purified on a 10% (wt/vol) polyacrylamide gel containing 7 M Urea. For the gel-shift assay, 0.5 pmol labeled tRNA^iMet was mixed with 4.5, 15, 45, or 150 pmol protein in a 10-μL reaction containing 20 mM Hepes at pH 7.5, 100 mM potassium acetate, 30 mM KCl, 5% (vol/vol) glycerol, 5 mM magnesium diacetate, 0.1% (vol/vol) NP-40, 2 mM DTT, and 30 μg/mL heparin as a nonspecific competitor. The mixtures were incubated 1 h at 4 °C before adding 2 μL 50% (vol/vol) glycerol containing 0.25% (wt/vol) xylene cyanole. Samples were run on a 6% (wt/vol) polyacrylamide gel at 4 °C and visualized by phosphorimaging (GE Healthcare).