Motif III in Superfamily 2 “Helicases” Helps Convert the Binding Energy of ATP Into a High-Affinity RNA Binding Site in the Yeast DEAD-Box Protein Ded1

doi:10.1016/j.jmb.2009.12.025 J. Mol. Biol. (2010) 396, 949–966

Available online at www.sciencedirect.com

Motif III in Superfamily 2 “Helicases” Helps Convert the Binding Energy of ATP into a High-Affinity RNA Binding Site in the Yeast DEAD-Box Protein Ded1

Josette Banroques1,2,3, Monique Doère3, Marc Dreyfus1, Patrick Linder3 and N. Kyle Tanner1,3⁎

1Institut de Biologie Motif III in the putative helicases of superfamily 2 is highly conserved in Physico-chimique, CNRS UPR both its sequence and its structural context. It typically consists of the 9073 in association with the sequence alcohol–alanine–alcohol (S/T-A-S/T). Historically, it was thought Université Paris VII, Paris to link ATPase activity with a “helicase” strand displacement activity that 75005, France disrupts RNA or DNA duplexes. DEAD-box proteins constitute the largest family of superfamily 2; they are RNA-dependent ATPases and ATP- 2Centre de Génétique dependent RNA binding proteins that, in some cases, are able to disrupt Moléculaire, CNRS FRE 3144, short RNA duplexes. We made mutations of motif III (S-A-T) in the yeast Gif-sur-Yvette 91198, France DEAD-box protein Ded1 and analyzed in vivo phenotypes and in vitro 3Département de Microbiologie properties. Moreover, we made a tertiary model of Ded1 based on the et Médecine Moléculaire, Centre solved structure of Vasa. We used Ded1 because it has relatively high Médical Universitaire, Geneva ATPase and RNA binding activities; it is able to displace moderately stable 1211, Switzerland duplexes at a large excess of substrate. We find that the alanine and the threonine in the second and third positions of motif III are more important Received 6 August 2009; than the serine, but that mutations of all three residues have strong received in revised form phenotypes. We purified the wild-type and various mutants expressed in 8 December 2009; Escherichia coli. We found that motif III mutations affect the RNA-dependent accepted 14 December 2009 hydrolysis of ATP (kcat), but not the affinity for ATP (Km). Moreover, Available online mutations alter and reduce the affinity for single-stranded RNA and 21 December 2009 subsequently reduce the ability to disrupt duplexes. We obtained intragenic suppressors of the S-A-C mutant that compensate for the mutation by enhancing the affinity for ATP and RNA. We conclude that motif III and the γ binding energy of -PO4 of ATP are used to coordinate motifs I, II, and VI and the two RecA-like domains to create a high-affinity single-stranded RNA binding site. It also may help activate the β,γ-phosphoanhydride bond of ATP. © 2009 Elsevier Ltd. All rights reserved. Keywords: RNA helicase; ATPase; molecular motor; Saccharomyces cerevisiae; Edited by A. Pyle RecA like

Introduction

The putative helicases of superfamily 2 (SF2) are an ubiquitous group of enzymes that are associated *Corresponding author. IBPC, CNRS UPR Pierre et Marie with all processes involving RNA and DNA. SF2 Curie, 75005 Paris, France. E-mail address: proteins are closely related to those of superfamily 1 [email protected]. (SF1), some of which are known processive DNA Abbreviations used: SF2, superfamily 2; SF1, helicases that are involved in DNA replication and 1,2 superfamily 1; ssRNA, single-stranded RNA; AMP-PNP, repair. These NTPases (generally ATPases) are adenosine 5′-(β,γ-imino)triphosphate; PDB, Protein Data characterized by a highly conserved structural core Bank; SD-Leu plate, synthetic minimal medium plate consisting of two linked RecA-like domains that lacking leucine; 5-FOA, 5-fluoroorotic acid; EMSA, contain seven or more conserved motifs involved in electrophoretic mobility shift assay. nucleotide triphosphate and nucleic acid binding,

0022-2836/$ - see front matter © 2009 Elsevier Ltd. All rights reserved. 950 Motif III in Superfamily 2 “Helicases” interdomain interactions, and NTPase activity.1,2 an essential gene in the yeast Saccharomyces cerevi- The largest SF2 family comprises DEAD-box siae. It was first identified as an intragenic suppres- proteins, which are RNA-dependent ATPases and sor of a PRP8 mutant, which implied a role in ATP-dependent RNA binding proteins. These mRNA splicing,22 but it was also implicated in the proteins are associated with all processes involving transcription of polymerase III RNAs,23 and to have RNA from transcription to decay, and each family a general role in translation initiation24,25 and 40S member is typically involved in a unique ribosome scanning.26 Finally, it was implicated in P- – process.3 7 They contain nine conserved motifs body formation and RNA degradation27 and in (Q, I, Ia, Ib, and II–VI), a conserved GG sequence yeast L-A virus synthesis.28 Ded1 is closely related between motifs Ia and Ib, and a conserved QxxR to a subfamily of DEAD-box proteins involved in sequence (where x is any residue) between motifs developmental regulation,29–31 including the Dro- IV and V.3,4,7,8 Although commonly known as sophila Vasa protein for which the crystal structure DEAD-box “helicases,” they have not been shown was solved in the presence of RNA and adenosine to have a processive unwinding activity of double- 5′-(β,γ-imino)triphosphate (AMP-PNP).12 stranded RNA, and they are only able to displace Motif III is defined by the sequence serine– short duplexes, typically at very high protein/ alanine–threonine (S-A-T) in the DEAD-box family. duplex ratios in vitro.1,7,9 We mutated the three positions in DED1 and There are now a large number of solved crystal analyzed the in vivo phenotypes and in vitro structures of DEAD-box proteins, including several properties of the proteins expressed in and – with bound ligands.10 15 These data have verified or purified from Escherichia coli. We find that muta- clarified the roles of the different conserved motifs tions of motif III are poorly tolerated in vivo, and and features. Thus, the Q motif is involved in that these mutants affect the hydrolysis of ATP adenine recognition. Motifs I and II, also called (kcat) and the affinity for ssRNA, but not the Walker motifs A and B, are involved in the binding affinity for ATP (Km). This subsequently reduces of the α and β phosphates of the ATP. Motifs Ia, Ib, the ability of the proteins to displace nucleic acid IV, and V; the conserved sequences GG and QxxR; duplexes. It appears that motif III is critical for and a residue next to motif II are involved in single- aligning—and perhaps activating—the γ phos- stranded RNA (ssRNA) binding. Motifs III, V, and phate of ATP, for coordinating the different motifs, VI are involved in binding the α, β,andγ and for helping to convert the binding energy of phosphates of ATP.2,4,7,8 ATP into a high-affinity RNA binding site. Thus, The enzymatic roles of specific residues within the mutations of motif III in Ded1 affect the observed motifs of DEAD-box proteins were derived from “helicase” activity indirectly by altering and mutagenic and biochemical analyses. The highly reducing RNA binding, and this may be true for conserved glutamic acid of motif II, which is other characterized SF2 proteins as well. conserved in other SF2 and SF1 families as well, is β γ thought to activate the , -phosphoanhydride bond Results of ATP through a coordinated water molecule, in association with a conserved histidine residue of motif VI and a bound magnesium ion.4,16 A similar Sequence alignments in and around motif III mechanism of activation was proposed for the SF2 Dengue NS3 protein.17 The role of motif III is more Sequence alignments of 699 unique DEAD-box complicated; it has been suggested to link ATPase proteins were performed as described in Materials and “helicase” activities because mutations in and Methods and are shown in Fig. 1a. Motif III mammalian eIF4A eliminated strand displacement contained a serine in the first position 89% of the activity without affecting ATP hydrolysis or RNA time, while a threonine occurred 11% of the time. binding.18 In contrast, similar mutations in yeast The second position of alanine was 100% conserved, Has1 have reduced affinities for ATP and RNA and and there was virtually always a threonine in the reduced ATPase activity, which resulted in reduced last position, although a serine was found in 0.4% of strand displacement activity.5 Mutations of motif III the sequences. In contrast, the sequences flanking also dissociated the ATPase activity from the strand motif III were more variable, but there was a displacement activity in the related yeast DEAH-box tendency for hydrophobic residues to precede and, protein Prp22, but the RNA binding affinities were to a lesser extent, follow motif III. We examined the not determined.19 Finally, motif III was proposed to solved crystal structures of Vasa, DDX19B (human serve as a relay for ATP binding and hydrolysis and Dbp5), Mss116, and eIF4AIII in the presence of for the binding of DNA in the SF1 Rep protein, but bound RNA and AMP-PNP to better understand its – the character of motif III and its interactions with the structural context;10 15 motif III occurred at the end other motifs in SF1 are significantly different from of a β-sheet strand that subsequently turned into a those of SF2.20,21 loop, and this was probably why there was a We were interested in obtaining mutants of motif preference for hydrophobic groups preceding the III in the DEAD-box protein Ded1 that dissociated motif (Fig. 1a). The loop structure then turned into a the different activities, so that we could better helix, but there were differences between Vasa, understand the interrelationships between the DDX19B, Mss116, and the different eIF4AIII crystal motifs and the role of the protein in vivo. DED1 is structures in the size of the loop and helix. Motif III in Superfamily 2 “Helicases” 951

Fig. 1. Sequence alignments show that motif III in SF2 proteins is highly conserved. (a) Nearly 700 unique DEAD-box protein sequences were aligned as previously described.32 The sequence shown below is from Ded1. The structural context of motif III is based on the solved crystal structure of Vasa, where L is a loop, S is a β-sheet strand, and H is a helix. Alanine of motif III interacts γ with -PO4 of AMP-PNP through a water molecule in the Vasa structure. (b) Sequence alignments of various families of SF2 proteins. Upper-case letters are residues, and lower-case letters are classes ofresidues:a,aromatic;c, charged; h, hydrophobic; l, ali- phatic; o, serine or threonine; p, polar; t, alanine or glycine; (−) negative; (+) positive; (•)any residue. The consensus was arbi- trarily taken at 60% of the aligned sequences.

We examined other solved crystal structures of over 99% of the DEAH-box proteins. The other DEAD-box proteins in the Protein Data Bank families, such as Ski2 and Hef, showed more (PDB)†. We found the structural context of motif variability, but the consensus was always alcohol– III to be remarkably conserved, with the first residue alanine–alcohol, with threonine being largely pre- being almost always part of a β-sheet, followed by a ferred in the third position. The SecA and Snf2 loop of four (4.3±0.7) and then a helix. The latter families were exceptions; the consensus was threo- helix was followed by a loop, a β-sheet strand, and nine–glycine–threonine in nearly 100% of the then a loop that connected domain 1 with domain 2. sequences (Fig. 1b). Despite this remarkable conser- Although there were differences in loop size in the vation, the sequences flanking motif III were highly presence and in the absence of ligands, as for Mss116 variable between families. Nevertheless, the struc- and eIF4AIII, these differences were not correlated tural context found in the DEAD-box family was with the ligands per se. Instead, they seemed to also true for the solved crystal structures of all the reflect the variability between the various solved other SF2 families examined in the PDB. Some crystal structures. families, such as SecA and Hef, had longer loops We did BLAST and sequence alignments on other (generally six to seven), and the viral NS3 proteins SF2 families as well and found motif III to be highly had loops that connected directly with domain 2. conserved between families (Fig. 1b). For example, But the context of motif III (β-sheet loop) was almost motif III was serine–alanine–threonine (S-A-T) in inevitably the same. However, the alcohol– tiny– alcohol sequence of motif III was not universally conserved in SF2 families or even in the DEAD-box † http://www.rcsb.org/ family; the UL9 protein of the herpes simplex virus 952 Motif III in Superfamily 2 “Helicases” had a D-A-T sequence,33 and the cyanobacteria cold yeast, replated on SD-Leu plates, and restreaked shock DEAD-box protein CrhC, which was not on 5-FOA plates to verify the phenotypes. We obtained in the original BLAST, had the sequence sequenced 10 of the recovered wild-type and F-A-T.34 These results suggested that the first pseudo-wild-type plasmids, 13 of the conditional position of motif III was somewhat malleable in SF2. mutants, and 5 plasmids from cells that did not grow on 5-FOA plates. For position T343, we found Growth phenotypes of motif III mutants that about 35% of the clones were wild type or pseudo wild type, 60% failed to support growth, We randomly mutagenized the first codon and only 5% were unable to grow at low or high (corresponding to S341) and the third codon (T343) temperatures (Fig. 2a). We sequenced 13 of the of motif III in DED1, transformed yeast cells with the recovered wild-type or pseudo-wild-type plasmids, DNA, selected the transformed cells on synthetic 6 of the conditional mutants, and 10 plasmids of the minimal medium plates lacking leucine (SD-Leu lethal mutants. plates), and then streaked the resulting colonies on Figure 2b summarizes the results obtained in these 5-fluoroorotic acid (5-FOA) plates to eliminate cells screens. Consistent with sequence alignments, we retaining the plasmid-encoded copy of the wild-type noticed that the DED1 mutants that supported DBP1 gene, which complemented DED1 deletion growth to about the same level as wild type when overexpressed. We used yeast cells deleted for contained either a threonine or a serine in the first the chromosomal copies of both DED1 and DBP1 or third position of the motif, respectively, although genes in all our screens. About 270 colonies were S341T grew slightly slower than the wild type at all tested in each case on 5-FOA plates at 18 °C, 30 °C, the temperatures tested and T343S was slightly cold and 36 °C. For position S341, we found that 50% of sensitive. However, the first position tolerated a the clones were wild type or pseudo wild type, number of other amino acids that allowed the cells about 20% of the clones failed to grow at any to grow more or less than the wild type. Two temperature, and about 30% showed conditional categories of conditional mutants were seen accord- growth (Fig. 2a). The plasmid DNA was recovered ing to the strength of the phenotypes. The mutants from representative colonies, retransformed into S341A, S341G, and S341C had weak slow-growth

Fig. 2. Motif III mutations have strong phenotypes. (a) Cultures of yeast cells, deleted for chromosomal copies of DED1 and DBP1 (ded1∷HIS dbp1∷KANMX6)and transformed with leucine plasmids containing ded1 with the indicated mutations, were serially diluted (by factors of 10) and then spotted on minimal medium plates containing 5-FOA to eliminate cells retaining the uracil plasmid containing the wild-type copy of DBP1.Plates were incubated at the indicated temperatures. (Ø) A control leucine plasmid lacking any insert. Note that the few isolated colonies seen for some lanes (SAC at 18 °C and SAI) were pseudo revertants that resulted from recombination between plasmids. (b) Résumé of the growth phenotypes of the tested mutants. CS, cold sensitive; TS, sensitive to high temperatures. Motif III in Superfamily 2 “Helicases” 953 phenotypes and were considered pseudo wild type. mutated S201 and T203 in eIF4A to alanine and In contrast, the S341H, S341L, S341M, and S341Q obtained the same results: S201A was cold sensitive, mutants were more severe, and they were unable to while the T203A mutant failed to support growth grow at 18 °C and grew more slowly at 30 °C and (data not shown). In contrast, when we independently 36 °C. Three mutants (S341R, S341D, and S341P) mutated T188 and T190 in Dbp8 to alanines, we were unable to grow at any temperature. Thus, obtained wild-type growth at all the temperatures bulky and charged groups were less tolerated than tested, although the combined double mutant grows small residues in the first position. two times slower at 30 °C (data not shown).35 We only obtained the conditional mutation T343C However, Dbp8 was more tolerant of mutations in for the third position for all six clones isolated, general, relative to other essential DEAD-box proteins, indicating an essential role for an alcohol at this and its ATP-dependent enzymatic activity per se may position. Nevertheless, the T343C mutant was not have been essential in vivo (J.B., unpublished data). severe, and it failed to support growth at 18 °C (Fig. 2a). None of the other mutants were viable Isolation of intragenic suppressors of the (T343A, T343I, T343F, T343Y, T343M, T343E, T343K, conditional T343C Ded1 mutant and T343N). We also site-specifically mutated the 100% conserved A342 into a glycine; this mutant We decided to investigate gene-encoded (intra- grew very poorly at 30 °C or 36 °C and was unable genic) changes that compensate for mutations in to grow at 18 °C. Finally, we constructed a double motif III and that might reveal something about how mutant where both S341 and T343 were changed to other residues potentially interact with motif III or alanines (S341A T343A). This mutant failed to about its role in the activities of Ded1. We chose the support growth at all temperatures. ded1 T343C mutant because cells transformed with This study was not intended to be an exhaustive this plasmid were unable to grow at 18 °C or lower screen; we were mostly interested in obtaining temperatures. We performed a PCR mutagenesis of conditional mutations for subsequent analyses. How- the plasmid-encoded gene containing the SAC ever, different codons were obtained for the same mutation as described in Materials and Methods. residues, indicating that the oligonucleotides were We obtained 12 clones out of the 250 screened clones sufficiently randomized. None of the mutants showed that grew better at the selected temperatures. We dominant-negative phenotypes in the presence of the obtained the intragenic suppressors S118P and wild-type DED1. Protein expression of the mutants E143G twice; F205L, F242L, K301R, Y359H, Y359N, was confirmed by Western blot analysis (data not and M421V once; and two plasmids that contained shown). We concluded from these in vivo analyses two mutations in addition to T343C (L152S+T370A that the third position of motif III was more important and I360T+S590P). We separated these latter muta- than the first, but that an alcohol residue (S, T, and, to tions by subcloning them into the appropriate a lesser extent, C) was preferred for both positions. regions of the original ded1-T343C plasmid; only We were interested in knowing how general these T370A and I360T enhanced the growth of the motif phenotypes were in other essential yeast DEAD-box III mutant, while L152S and S590P did not affect proteins. We previously showed that, in Has1, the growth. Thus, we obtained 10 unique intragenic first-position S228A mutant was cold sensitive, and mutations that suppressed the T343C mutation to the third position T230A was lethal.5 In this study, we various extents (Fig. 3).

Fig. 3. Intragenic suppressors of SAC mutation. Cells containing the wild-type gene (SAT), T343C mutation (SAC), and T343C mutation with intragenic suppressors (+residue) were plated as described in Fig. 2. 954 Motif III in Superfamily 2 “Helicases”

The M421V, E143G, S118P, F205L, and F245L typically polar (83.7%) in sequence alignments, suppressors only slightly enhanced growth at 30 °C but a glycine naturally occurred 10.0% of the time. and 36 °C, and poorly suppressed the cold-sensitive F205L was found in a variable loop region phenotype. K301R moderately enhanced growth at between motifs I and Ia. F242L was adjacent to all temperatures. Oddly, T370A moderately sup- motif Ia; it was a leucine in only 4.4% of the pressed the cold-sensitive phenotype, but enhanced sequences, while a polar group was present in the high-temperature sensitivity. The best suppres- 56.7%. K301R was close to motif II, and it was an sors were I360T, Y359H, and Y359N, suggesting that arginine in 11.3% of the sequences, charged in this region of the protein was particularly important. 67.7% and polar in 95.3%. The position occupied Interestingly, the Y359H suppressor grew better at by Y359 was typically hydrophobic (92.3%) and all temperatures, while Y359N grew best at higher often a proline (69.2%); it was an asparagine in temperatures. 2.0% of the sequences and never a histidine. We mapped the positions of the suppressors on Similarly, I360T was a hydrophobic residue in the three-dimensional model of Ded1 (Fig. 4b). All 75.8% of the sequences and a threonine in only 3%. suppressors, except for S118P, were located on the T370A was located close to a variable loop, but outer shell of the two RecA-like domains and were was typically polar (57.1%). Finally, the position outside the conserved motifs. Nevertheless, nearly corresponding to M421V was polar (65.4%), and a all the suppressors mapped to residues that had valine was present in only 0.4% of the sequences. functionally conserved side chains among the 700 None of the suppressors were close to residues aligned DEAD-box sequences. The S118P suppres- predicted to interact with the RNA ligand, and sor was located in the highly variable amino only S118P was positioned close enough to terminus in a region that was predicted to be potentially affect the interactions with AMP-PNP. unstructured. The E143G mutant was adjacent to However, the E143G mutation could alter the the highly conserved, isolated, aromatic group of conserved helix–loop–helix–loop structure of the Q the DEAD-box Q motif;36,37 this position was motif involved in adenine recognition.

Fig. 4. Three-dimensional modeling of Ded1. The tertiary structure of Ded1 was modeled based on the solved crystal structure of chain A of Vasa as described in Supplementary Material. (a) The structure of Vasa (orange) superimposed on the ribbon model of Ded1 (gray with colored motifs). Vasa is shown with 40% transparency to facilitate viewing of the two structures. The conserved motifs are as indicated, and the bound ligands (AMP-PNP and RNA) are from the Vasa structure and are shown as stick models. The bound Mg2+ ligand is shown as a space-filling model. (b) Positions of the suppressors of the motif III T343C mutation (boxed) on the model of Ded1. The structure is shown from the back of that shown in (a), and the ribbons outside the conserved motifs are shown with 40% transparency for viewing the mutant residues (sticks). (c) Alignment of the Vasa and Ded1 proteins that was generated with the Swiss-Model program (see Supplementary Material). The conserved motifs and sequences are indicated above the alignment, and the positions of the T343C mutation and corresponding suppressors are shown below (boxed). Motifs are colored as shown in (a) and (b). Motif III in Superfamily 2 “Helicases” 955

Reduced RNA-dependent ATPase activities of Ded1 is unknown.32,36,37 We then determined the motif III mutants Michaelis–Menten kinetic parameters at 30 °C (Table 1). We subcloned representative ded1 mutants into a A very low enzymatic performance (kcat/Km) in pET expression vector containing a carboxyl-termi- vitro, relative to the wild type, was observed for the nal His6 tag. This allowed us to express the proteins lethal (T343A and T343I) and the poorly growing in E. coli and to purify them on nickel affinity (T343C and A342G) ded1 mutants at the second and columns. We were able to successfully obtain the third positions. They all had between 5.6% and 8.0% wild-type Ded1, various motif III mutants, and the of wild-type activity. This result confirmed the SAC mutant with six of the different suppressors at importance of these two positions in motif III. high purity and yield (data not shown). Moreover, Similar changes from the first-position S341 to we purified the Y359H and Y359N suppressors cysteine and alanine had less profound effects (24– separated from the original SAC mutation. Finally, 55% wild-type activity), which again was well we purified the mutants Y359K and Y359P—the correlated with the in vivo phenotypes. These former to determine whether an amine-containing mutants grew nearly as well as the wild type, and residue was important at this position and the latter thepercentageofactivitycorrespondedtothe because a proline residue was found in this position severity of the growth defect. Oddly, the S341L nearly 70% of the time in sequence alignments. The mutant had better ATPase activity than the T343C expression and purification characteristics of the mutant (12% versus 6.8%), but it grew more slowly proteins were similar to those of wild type, than T343C at higher temperatures (both were cold indicating that there were no major structural sensitive); the leucine mutation may have destabi- alterations. However, we verified this by subjecting lized the protein. Thus, the first position, which the mutant proteins SAC, SAA, and AAA to partial showed more sequence variability in the alignments, trypsin digestion in the presence and in the absence was less critical than the second and third positions, of ligands; there were no significant differences in which were more highly conserved. The lethal cleavage patterns with those of the wild-type double mutant ded1 S341A T343A showed only protein (data not shown). 2.9% of the ATPase activity of the wild type with We then tested the mutants and the suppressors 1 mM ATP, and we were unable to determine its for their capacity to hydrolyze ATP in the enzymatic performance (data not shown). Interest- presence of RNA by using a molybdate–Malachite ingly, the effects were almost entirely on the rate of Green assay to monitor the production of free hydrolysis of the ATP (kcat) rather than on the phosphate. We used total yeast RNA, as previ- binding affinities (Km), which showed only small ously described, because the in vivo substrate for differences for all the mutants.

Table 1. Kinetic parameters for RNA-dependent ATPase activities

−1 a −3 −1 −1 b c d e Km (mM) kcat (min ) kcat/Km ×10 (M min ) % Wild type Phenotypes EMSA Wild type (SAT) 0.37±0.03 280±30 740±70 100 WT +++ AAT 0.23±0.03 94±9 410±60 55 ∼WT ++ CAT 0.44±0.06 79±8 180±30 24 ∼WT ND LAT 0.53±0.10 46±5 87±17 12 CS and SG +/− SGT 0.35±0.02 21±2 60±6 8.0 bSG +/− SAA 0.52±0.05 22±2 41±4 5.6 NG +/− SAC 0.48±0.05 24±2 51±5 6.8 CS and SG + SAI 0.57±0.04 30±4 51±6 6.9 NG +/− SAS 0.58±0.04 110±10 190±20 26 ∼WT +++ AAA ND ND ND ∼3NG+/− SAC+F242L 0.40±0.08 31±3 78±16 10 CS and ∼SG ++ SAC+K301R 0.21±0.02 25±3 120±10 16 CS and ∼SG ++ SAC+Y359H 0.093±0.013 31±3 330±50 44 ∼WT ++ SAC+Y359N 0.23±0.03 26±3 110±20 15 CS ++ SAC+I360T 0.26±0.03 35±4 140±10 18 CS ++ SAC+T370A 0.095±0.006 26±3 270±30 37 ∼CS and TS ++ Y359H 0.063±0.002 150±20 2400±200 320 WT +++ ½ Y359N 0.18±0.01 150±20 820±90 110 WT +++ ½ Y359K 0.20±0.01 190±20 960±100 130 WT ND Y359P 0.18±0.02 200±20 1100±100 150 WT ND Values were derived from nonlinear fits to the Michaelis–Menten equation using the mean values of at least three independent experiments. Errors are standard deviations from the mean. a A minimum of 10% error was assumed for the protein concentrations. b The largest errors of kcat or Km were carried forward. c Enzymatic performance (kcat/Km) relative to the wild-type protein. AAA value was estimated using 1 mM ATP. d WT, wild-type growth; ∼WT, pseudo-wild-type growth; SG, slow growth; ∼SG, slightly slow growth; bSG, very slow growth; NG, no growth; CS, cold sensitive; ∼CS slightly cold sensitive; TS, sensitive to high temperatures. e AMP-PNP-dependent EMSAs of RNA. Note that the SAA, SAI, SGT, and AAA mutants did not form discrete retarded products. EMSA and strand displacement activities were correlated. 956 Motif III in Superfamily 2 “Helicases”

The suppressors of the T343C mutation improved Motif III mutants had reduced ATP-dependent enzymatic performance by 47% to nearly 650%, affinity for RNA although the best suppressor was only 44% of the wild-type activity. Oddly, this was almost entirely We then investigated the RNA binding affinities of due to an increased affinity (lower Km) for ATP. the proteins by electrophoretic mobility shift assays Moreover, the best suppressor Y359H showed (EMSAs).38,39 We previously showed that Ded1 has a nearly the same properties in isolation of the higher affinity for ssRNA substrates in the presence of T343C mutation: it had nearly 6-fold higher affinity AMP-PNP, which is a nonhydrolyzable analog of for ATP relative to that for the wild type, while kcat ATP, than in the absence of a nucleotide or in the was reduced by only 2-fold. The Y359K, Y359N, and presence of ADP.32 In this set of experiments, we used Y359P mutants, in isolation, improved Km by about a constant concentration of a 25-nt-long ssRNA 2-fold and had similar values of kcat as Y359H. Thus, (RNA01) and varied the protein concentration as motif III mutations affected the hydrolysis of ATP described in Materials and Methods. rather than the binding, and the suppressors In the absence of nucleotide or in the presence of compensated for reduced hydrolysis by enhancing ADP, all of the motif III mutants showed weak the affinity for ATP. binding to ssRNA, similar to binding to the wild-type We used highly saturating concentrations of protein, although the AAA mutant was particularly total yeast RNA (0.5 μg/μl) for RNA-dependent weak (Fig. 5). All of the motif III mutants showed ATPase assays, based on the wild-type Ded1. strongly reduced affinities for ssRNA in the presence However, it was possible that the mutants had a of AMP-PNP that was correlated with the in vivo very poor affinity for the RNA and that our assay phenotypes and in vitro ATPase activities. The SGT was not saturating for these proteins. We were not mutant was often retained in the well of the gel (25– able to use more than 1.5 μg/μl total RNA in our 50% of retarded material), suggesting that it might assays because the components tended to precip- aggregate or that it was less stable (data not shown). itate under the reaction conditions. Thus, it was To facilitate these analyses, we typically counted possible that the reduced rates of hydrolysis (kcat) everything that migrated above the free ssRNA as were indirectly a result of a reduced affinity for the being retarded. Therefore, our estimates actually RNA. overestimated the binding affinities because smearing

Fig. 5. EMSAs of RNA incubated with different Ded1 constructs. 32P-end-labeled RNA01 was incubated with different concentrations (nM) of the indicated Ded1 proteins in the absence of a cofactor (−nt), in the presence of ADP (+ADP), or in the presence of AMP-PNP (+PNP), and electrophoretically separated on a 6% nondenaturing polyacrylamide gel at 4 °C. Free, RNA01 migration in the absence of protein; Ori, the bottom of the loading well of the gel. Bar graphs show the percentage of radioactivity migrating more slowly than the free RNA01. Note that the SAT+Y359H panel was chosen because it best showed the three distinct slower-migrating bands; typically, it had slightly stronger binding affinity than the wild type, as determined in Hill plots with more extensive data points (data not shown). Motif III in Superfamily 2 “Helicases” 957 above the free ssRNA was counted similarly as rupted a hairpin structure that formed around the discrete bands migrating slowly in the polyacryl- hybridization site. Substrates had 25-nt-long ssRNA amide gel. Blunt-ended double-stranded RNA and landing sites and 16-bp duplexes. Both 5′ and 3′ single-stranded DNAs showed a weak affinity for the duplexes were functional substrates. We used a 20- wild-type protein, independent of AMP-PNP.32 fold or 25-fold excess of a DNA trap (α-Hyb1 DNA) As shown in Fig. 5, motif III mutations affected that hybridized to the 16-nt-long 32P-labeled Hyb1 both the affinity and the character of ssRNA binding oligonucleotide that was released and prevented it in the presence of AMP-PNP. Typically, the wild- from reannealing onto the ssRNA template (Fig. 6a). type Ded1 showed three distinct bands that were However, these experiments were qualitatively and clearly separated from the free ssRNA and that quantitatively similar to those obtained when an increased with the protein concentration. In con- excess of unlabeled Hyb1 (which hybridized to form trast, the most benign motif III mutation (AAT) an identical but unlabeled duplex) was used.32 In showed only one major band that smeared upward both cases, the single-stranded regions of the at the higher protein concentrations and that unlabeled RNA products were competitors for the corresponded to the faster-migrating band of the binding site on the protein that increased during the wild-type Ded1. The other mutants showed binding course of the reactions. Thus, the observed displace- characteristics that corresponded to the phenotypes. ment rates decreased during the course of the For example, the SAA and SAI mutants, which did reactions. not support growth, did not form a discrete band, We tested both a DNA version and an RNA while the SAC mutant, which was slow growing version of Hyb1 (Fig. 6a); the DNA–RNA duplex was and cold sensitive, formed a single weak band nearly completely displaced after 5 min at 30 °C with (Fig. 5). The AAA mutant showed very little AMP- a 50-fold excess of substrate to wild-type Ded1. In PNP-dependent binding; it did not support growth. contrast, the RNA–RNA duplex, which was calcu- The suppressors Y359H and T370A (Fig. 5) and the lated to have a Gibbs free energy that was 5.5 kcal/ suppressors F242L, K301R, Y359N, and I360T (data mol more stable (ΔG°=−25.3 kcal/mol) than that of not shown) enhanced the RNA binding affinity of the DNA–RNA duplex (ΔG°=−19.8 kcal/mol) the SAC mutant in direct relation to their in vivo under standard conditions, was only partially dis- phenotypes and in vitro ATPase activities (kcat/Km), placed after 60 min with stoichiometric concentra- but they did not restore the characteristic profile of tions of the duplex and the protein (Fig. 6b; wild-type the binding (a single major band instead of three Hyb1 RNA/K06 RNA). However, this more than bands). In contrast, the Y359H suppressor in 100-fold difference in activities was not entirely isolation showed essentially wild-type binding attributable to the difference in the free energies of properties with three major bands in the presence the duplexes because the DNA trap formed an of ssRNA and AMP-PNP (Fig. 5). unusually weak duplex with the released Hyb1 We determined the binding affinities of wild-type RNA, which was calculated to be 8.3 kcal/mol less Ded1, Y359H, and Y359N in the presence and in the stable than the RNA–RNA duplex due to weak rUU/ absence of AMP-PNP using the Hill plot, as dAA base pairs (ΔG°=−17.0 kcal/mol).40 Indeed, previously described (data not shown).32 There 25% of the [32P]Hyb1 RNA/K06 RNA duplex would was significant day-to-day experimental variability, reform in the presence of a hundredfold excess of α- and the differences were small, but the isolated Hyb1 DNA when heat denatured and slow cooled in suppressors consistently showed a higher affinity the absence of protein, while only about 5% of [32P] for the ssRNA than for the wild type in the presence Hyb1 DNA/K06 RNA duplex would reform with and in the absence of AMP-PNP. The medium value only a 25-fold excess of α-Hyb1 DNA (data not was about a 30% enhancement. There was no shown). Hence, we primarily used the DNA/RNA evidence that the wild-type or isolated suppressor duplexes in our subsequent analyses. proteins were functioning cooperatively as multi- We then assayed our various motif III mutants mers or that there were multiple independent and suppressors (Fig. 6b). We used a GKT-to-GAT binding sites. Thus, these results were consistent mutant of the P-loop of motif I as control because with the interpretation for the Y359 suppressors that this mutant showed only insignificant ATPase the enhanced affinity for ATP was indirectly a result activity, and it was expected to significantly disrupt of a slightly enhanced affinity for ssRNA, although the α and β phosphate interactions of ATP with the cumulative effects could not be ruled out. protein; this mutant showed slight ATP-independent strand displacement at a 10-fold excess over the Reduced strand displacement activity of motif III DNA–RNA substrate. The various motif III mutants mutants showed intermediate activities that were strongly correlated with their ATPase (kcat/Km) and RNA We analyzed the various purified proteins by a binding activities. Thus, the double AAA mutant strand displacement assay, as described in Materials showed weak strand displacement at a 10-fold and Methods. We used different substrates that excess of protein over duplex, while the SAC mutant formed duplexes on either the 5′ end or the 3′ end of showed a relatively good strand displacement with the ssRNA “landing” site for the proteins, including a 5-fold excess of the duplex over protein. The a modified version of K01 (K06) that gave less suppressors enhanced the strand displacement of protein-independent displacement because it dis- the SAC mutation according to their ATPase and 958 Motif III in Superfamily 2 “Helicases”

Fig. 6. Strand displacement activity of the various Ded1 mutants. (a) Three of the duplexes used in these assays. 32P-labeled Hyb1 DNA or RNA was hybridized to R1 or K06 RNAs as shown. α-Hyb1 DNA is complementary to Hyb1, and it acts as a trap to prevent rehybridization of [32P]Hyb1 on the RNAs. (b) Strand displacement activity of 50 nM duplex by the various Ded1 constructs, in the presence (1 mM) or in the absence of ATP. Samples were incubated for the indicated times (in minutes) at 30 °C, and then the material was electrophoretically separated and analyzed as described in Materials and Methods. Note that the protein concentrations were calibrated to give similar amounts of displacement, when possible, with the same amount of duplex; therefore, they reflect relative differences in activity between the wild-type protein and the mutants. The duplexes used for each panel are as indicated. (‖) The displaced [32P]Hyb1 trapped by α-Hyb1.

RNA binding activities, with SAC+Y359H being virtually identical interactions were seen in the about the same as the wild type (Fig. 6b and data not solved crystal structures of DDX19B and Mss116 shown). However, there appeared to be no correla- and in the two solved structures of eIF4AIII in the tion with the characteristic binding profile (forma- presence of bound ligands (PDB coordinates 2db3, – tion of three discrete bands) of ssRNA binding, as 2hyi, 3i5x, 3i5y, 3i61, 3i62, 2j0s, 3fht, and 3g0h).10 15 seen with the wild-type protein. Moreover, the in Serine 432 of motif III in Vasa (S341 in Ded1) vivo phenotypes were more severe than expected formed hydrogen bonds with D402 of motif II and from the strand displacement activities, relative to with the peptide backbone of T434 of motif III. The previous mutants in the Q motif and motif IV.32,36 peptide backbone of S432 interacted with the The AAA mutant, which did not support growth, backbone of A287 that was two residues upstream had approximately the same ATP-dependent strand of motif I, although this interaction was not seen in displacement activity as the wild-type yeast eIF4A one of the crystal structures of eIF4AIII (PDB with the same substrate.41 The SAA mutant enzy- coordinate 2j0s), DDX19B (PDB coordinate 3g0h), matically displaced a 5-fold excess of duplex even and Mss116 (PDB coordinate 3i5x); the residues though it contained a lethal mutation. This sug- were slightly outside the limits of detection for a gested that this assay imperfectly reflected the in hydrogen bond (N3.3 Å). The peptide backbone of vivo enzymology of Ded1 with respect to motif III. the central A433 of motif III hydrogen bonded γ with -PO4 of AMP-PNP through a water mole- Structural context of motif III cule, and this interaction was shared with E400 of γ motif II that is thought to activate -PO4 in The structural context of motif III was highly association with H575 of motif VI.4,16 Threonine conserved in all the solved SF2 crystal structures 434 of motif III formed hydrogen bonds with D402 examined; it was situated at the end of a β-sheet of motif II and with H575 of motif VI; this latter strand that goes into a loop. For most SF2 proteins, interaction was not seen in one of the DDX19B motif III was an integral part of domain 1; however, structures (PDB coordinate 3fht), but the threonine in the viral NS3 proteins, it was part of a flexible and the histidine were only 2.90 Å apart. Thus, linker that went directly into domain 2.21 Figure 7b motif III was centrally located to bring together shows motif III interactions in the solved crystal elements of motifs I, II, and VI, as well as to help γ structure of Vasa with the bound ligands, but position -PO4 of ATP. Motif III in Superfamily 2 “Helicases” 959

and AMP-PNP ligands as Vasa and eIF4AIII,17 which are similar to those of DDX19B.13,15 Mss116 forms more extensive interactions with the RNA, both within the core and with the carboxyl terminus.14 Moreover, the NS3 proteins show the same characteristic cation–π interactions as the DEAD-box proteins between a conserved arginine of motif VI and a conserved phenylalanine of motif IV.32 Thus, even though the NS3 motif III was part of the flexible linker between domains, it showed the same types of interactions described above for Vasa in nearly all the ligand states, including the hydrolyzed form of ADP and PO4. This was true even though the equivalent residues in motifs II and VI were histidine and glutamine, respectively, instead of aspartic acid and histidine, and motif III had the sequence TAT. ATP was hydrolyzed when diffused into RNA-NS3 crystals, which indicated that the crystals were catalytically active. The interactions in NS3 seemed to be independent of the presence of nucleotide; most of the structural differences are RNA dependent in other regions of the protein.17 The sole exception was the structure bound to RNA and free PO4, which showed no interaction between the alanine of motif III and PO4 through water. These types of interactions are also true for Mss116 in the presence of ADP and BeFx or 14 AlFx (PDB coordinates 3i61 and 3i62) and for eIF4AIII in the presence of ADP and AlFx (PDB 42 coordinate 3ex7). BeFx is isomorphous to a phosphate group; thus, ADP-BeFx is considered a ground-state ATP analog, while AlFx is thought to mimic the planar phosphate transition state.44 We concluded that the motif III interactions were highly conserved in DEAD-box proteins and in many other SF2 families as well, even if the identity of specific residues varied between families. Indeed, when we independently changed the aspartic acid of motif II to a histidine and the histidine of motif VI into a glutamine in Ded1, such as in DEAH-box and Fig. 7. Interactions of motif III in the solved crystal NS3 proteins, we obtained nearly wild-type growth; structure of Vasa. (a) Overview of the relationship between however, the interactions also were genetically the conserved motifs and the substrates. The conserved linked because combining either mutation with the motifs are shown as ribbons, and the bound ligands (AMP- SAC mutation of motif III gave a lethal phenotype (J. PNP and oligouridine U2–U6) are shown as sticks. Domains B., unpublished data). Thus, although the residues 1 and 2 are shown as half-ellipses, and the overall could vary, the nature of the interactions of motif III relationship is similar to the structure shown in Fig. 4a. (b) remained the same in many different SF2 proteins. Interactions of motif III. Only interactions involving motif III are shown, and the equivalent residues in Ded1 are shown in parentheses. The two water molecules are shown as small Discussion red dots for clarity, important residues are shown as sticks, and hydrogen-bond interactions are shown as dotted green lines. H Oc has been proposed to catalyze the hydrolysis of Motif III was previously thought to link ATPase 2 “ ” ATP, while H2Or was proposed to enhance the nucleophi- activity with a helicase activity in SF2 14,42,43 2+ 5,18,19,45 licity of H2Oc. The bound Mg ligand is shown as a proteins. In our hands, it appears that the gray sphere. Glutamine 400 (E307) of motif II (not shown) role of motif III in the DEAD-box protein Ded1 is γ makes contacts with -PO4 of AMP-PNP through one of the more complex than this. The mutants affect the same water molecules (H Or) as A433 (A342) of motif III. 2 hydrolysis of ATP (kcat) and the binding affinity and binding characteristics for ssRNA; this subsequently reduces the observed strand displacement activity. The only SF2 crystal structure solved in various There are few effects on the binding affinity for ATP reaction states with the various bound ligands was (Km), even though the central alanine of motif III 17 γ that of the Dengue virus NS3 protein. This protein makes contact with -PO4 of ATP through water in forms nearly identical interactions with the RNA the solved crystal structures. The observed in vitro 960 Motif III in Superfamily 2 “Helicases” activities correlate well with in vivo phenotypes, types and in vitro properties are closely correlated, which indicate that they also are important in the cell. so it also is unlikely that the suppressors primarily However, the mutant phenotypes that we obtained affect interactions with other cofactors. The sup- are more severe than those obtained in other motifs of pressors alter the predicted electrostatic potential of Ded1 that we have studied. For example, the mutant the protein surface of Ded1, and this could enhance F405Y in motif IV of Ded1 has more than 10-fold less protein–protein or protein–RNA contacts (data not ATPase activity but nearly wild-type growth,32 while shown). However, we have no functional (enzymat- the LAT mutation of motif III has 8-fold less activity ic) evidence that Ded1 dimerizes or oligomerizes, but barely grows. This emphasizes the central role and the subunits of any such oligomer would have that motif III plays in the in vivo activity of Ded1. This to work independently. Moreover, no dimerization effect seems to be primarily at the level of the is observed in proteome screens in vivo.46,47 On the characteristics of RNA binding. All the mutants that other hand, the suppressors Y359H, Y359N, I360T, failed to form discrete bands in the presence of AMP- and K301R could enhance RNA binding by intro- PNP either did not support growth (SAA, SAI, and ducing polar or positively charged residues, but this AAA) or supported barely detectable growth (SGT; presupposes a much larger RNA binding site than is Fig. 1 and Table 1). In contrast, the strand displace- currently known. This nevertheless remains a real ment activity in vitro was more closely correlated possibility for Ded1 because they could expand the with the overall AMP-PNP-dependent binding affin- preexisting site or create new interactions that ity for the RNA (smearing); for example, the SAA facilitate ssRNA binding. This would explain why mutant is able to enzymatically displace a 5-fold they enhance the affinity for ssRNA without excess of duplex (as opposed to at least a 50-fold restoring the wild-type binding profile (Fig. 6b). excess by the wild type) in the presence of ATP even though it is not viable in vivo (Figs. 5 and 6). Previous characterization of motif III in other SF2 proteins Role of the intragenic suppressors of SAC mutation There is a notable divergence of observations for the in vivo and in vitro effects of the mutations of The suppressors mapped almost entirely on the motif III on different SF2 proteins. An SAT-to-LAT outer shell of the “helicase” core throughout the two mutation in the related DDX3 in humans has RecA-like domains (Fig. 4b and c). The only significantly reduced ATPase and strand displace- exception was the S118P suppressor that mapped ment activities, as seen for Ded1.48 Changing SAT to 26 residues upstream of the isolated highly con- AAA in the mammalian DEAD-box protein eIF4A served phenylalanine of the Q motif in a region that increased the binding affinity (lower Km) for ATP, 36,37 is likely to be flexible. The E143G suppressor increased Vmax, but eliminated the strand displace- mapped next to this phenylalanine, which is further ment activity; RNA binding is unchanged, but this evidence for the critical, but as yet unknown, role of was based indirectly on the level of RNA-stimulated the phenylalanine in the enzymatic activity of UV cross-linking of α-[32P]ATP to the protein.18 The DEAD-box proteins. Nearly all the suppressors equivalent yeast eIF4A protein has a cold-sensitive mapped in positions with functionally conserved phenotype for the AAT mutation and a lethal amino acids (in nearly 700 aligned sequences), and phenotype for SAA (this study). Likewise, the they often involved changing a hydrophobic residue mutations AAT and SAA in the yeast DEAD-box into a polar or charged group, or vice versa. All the protein Has1 have cold-sensitive or lethal pheno- analyzed suppressors compensate for the reduced types, respectively; 2-fold or 4-fold reduced affinities catalytic efficiency of the SAC mutant by enhancing for ssRNA, respectively; 2-fold reduced affinities for the affinity for ATP (Km) rather than by restoring the ATP (higher Km); 2-fold lower rates of hydrolysis 5 rates of hydrolysis (kcat; Table 1). They could do this (kcat); and reduced strand displacement activities. directly by modifying the ATP binding site or In the yeast Dbp8 protein, the mutations AAT and indirectly by enhancing the binding affinity for SAA are pseudo wild type (this study), and the AAA ssRNA. The latter is at least partially the case for the mutant undergoes slow growth at 30 °C.35 Y359H and Y359N suppressors because both confer A lethal phenotype is obtained when SAT is an enhanced affinity for ssRNA in the presence and changed to LAT in the yeast DEAD-box protein in the absence of AMP-PNP. This is consistent with Sub2.49 The mitochondria yeast DEAD-box protein the cooperative ATP-dependent RNA binding that Mss116 likewise has reduced ATPase and strand we observe. displacement activity when SAT is changed to However, none of the suppressors map close to AAA.50,51 The RNA binding affinity is similar to the RNA binding site identified in the solved crystal that for the wild type, but Mss116 has an expanded structures of Vasa, DDX19B, Mss116, and eIF4AIII, RNA binding site that may have reduced the and only S118P is in a flexible loop that is close to the magnitude of the mutant effects.14,51 Similar results – adenine.10 15 They are all located on the exterior are seen with YxiN, which has a secondary high- solvent-exposed shell of the core, so it is unlikely affinity binding site.52 Mutants of the yeast Prp28 that they indirectly affect the ligand binding sites protein, where the central alanine is changed to (ATP or ssRNA) by altering the conformations of the valine or tryptophan, are wild type or cold sensitive, RecA-like domains. Moreover, the in vivo pheno- respectively;53 however, the ATPase activity of Motif III in Superfamily 2 “Helicases” 961

Prp28 may not have been essential under the assay of a proton relay system that enhances the nucleo- conditions.32 This also may explain why more than philicity of the attacking water molecule.14,42,43 The half of the proteins having a LAT mutation in a large glutamic acid is thought to activate the β,γ-phos- screen of DEAD-box, DEAH-box, and Ski2 proteins phoanhydride bond with a water molecule, in involved in ribosomal processing in yeast were wild association with the conserved histidine of motif VI type: the endogenous wild-type genes were under in DEAD-box proteins (or glutamine in other the control of tetracycline-repressible promoters that families) and a bound Mg2+.4,16,17 The alanine may have been leaky enough to partially compen- seems to help position the functional groups, but it sate for the mutations (J.B., unpublished data).54,55 may also be involved in activating the β,γ-phos- This would clarify the weak phenotypes of some of phoanhydride bond. This explains why only alanine the mutations in the other conserved motifs as well. —and, to a lesser extent, glycine—is used because a The in vivo phenotypes of mutations of serine or larger side group would cause steric hindrance with γ threonine in the SAT motif of the SF2 DEAH-box -PO4; with residues in motifs I, II, and VI; and with proteins Prp2, Prp16, Prp22, and Prp43 in yeast are the bound water molecule. – similarly either slow growing or cold sensitive.19,56 58 The role of serine or threonine in the first position Mutations of either threonine in the TAT sequence of in SF2 helicases also seems clear. Serine, threonine, viral NPH-II eliminate the strand displacement and cysteine are the only residues that can form activity and significantly reduce the ATPase hydrogen bonds with the peptide backbone of the activity.45 Likewise, similar mutants of the NS3 third residue and maintain the sharp bend of motif protein of hepatitis C virus have reduced ATPase III seen in SF2 proteins (but not in SF1). The first and strand displacement activities.59,60 Finally, a residue further positions motif III relative to motif II threonine-to-serine mutation in the DAT sequence by making contacts with the second aspartic acid of of the UL9 protein of herpes simplex virus has motif II in DEAD-box proteins (or a histidine in reduced replication and reduced ATPase activity, other families). There also are peptide backbone without affecting the affinity for ATP.33,61 Changing interactions between the alcohol and an amino acid the alanine to serine also reduces replication, while a just upstream of motif I that could help position threonine-to-alanine mutation destroys replication motif I relative to motifs II and III. However, the competence.33 side-chain interactions of the first-position serine are Even though there are significant differences in not critical because neutral residues (A and G) are observations and interpretations, there appears to be tolerated under our assay conditions. a common theme for the mutations of motif III that The third-position threonine links motifs II, III, is found in the various SF2 proteins. Alterations in and VI in SF2 proteins through the interactions of either alcohol often result in cold-sensitive slow- the motif II aspartic acid (histidine in other families) growing or lethal phenotypes, and the third- and the motif VI histidine (glutamine). This threo- position alcohol is more critical than the first- nine plays a critical role in enzymatic activity position alcohol. Mutations affect ATPase and because only mutants with an alcohol at this strand displacement activities, and they have less position are viable in Ded1. Nevertheless, it is not profound effects on the binding affinity of ATP. clear why threonine is preferred at this position in Mutations may or may not affect the overall binding nature because serine can form the same interac- affinities for nucleic acids (RNA or DNA). However, tions. Moreover, the SAS mutant supports nearly most of these results are compatible with our wild-type growth and has good enzymatic activities observations on Ded1, and the apparent differences under our experimental conditions. Hydroxyl alco- probably reflect the characteristics of specific pro- hols at positions 1 and 3 are probably preferred teins. For example, some proteins, such as the because they are weakly acidic, are good nucleo- DEAD-box protein YxiN, have an additional high- philes, and readily form hydrogen bonds. In affinity RNA binding site that is separable from the contrast, we expect proteins substituted with RecA-like core and therefore independent of ATP cysteines to form weaker hydrogen bonds and to binding.52,62 Some proteins, such as Mss116, have be more sensitive to pH. more expansive RNA binding sites that may hide DEAD-box proteins are thought to use ATP the effects of altered binding around the enzymatic hydrolysis to recycle enzymatic forms rather than site in vitro.14 In other cases, the ATPase activity per to drive the ssRNA binding or strand displacement se may not be necessary for in vivo viability. activities.7,63,64 We have previously shown that Ded1 binds ATP and ADP with approximately the What is the role of motif III? same affinities (Km and Ki, respectively), and we were able to isolate mutants that altered the affinity Motif III is defined by the sequence alcohol–tiny– for one without altering the affinity for the other.36 alcohol. The question remains: Why are the charac- In yeast eIF4A, we showed that ADP actually binds teristic features of motif III so highly conserved in SF2 more tightly than ATP even though there are no γ- 37 proteins? The role of the conserved alanine seems PO4 interactions. Therefore, the binding energy of γ β γ clear: it forms hydrogen bonds with -PO4 of ATP the and phosphates is used not only to activate through a bound water molecule that is also shared the phosphoanhydride bond but also to constrain with the highly conserved glutamic acid of motif II; the different motifs and the two RecA-like domains this water molecule has been proposed to act as part of the protein core, so that they can bind RNA with a 962 Motif III in Superfamily 2 “Helicases” much higher affinity. This explains why mutations 1yks, 1ymf, 1z3i, 1z63, 1z6a, 2bmf, 2db3, 2f55, 2fsf, 2fsg, of motif III in mammalian eIF4A actually enhance 2fsh, 2fsi, 2fz4, 2g9n, 2gxu, 2hxy, 2hyi, 2i4i, 2j0q, 2j0s, 2oxc, the binding affinity for ATP because the other 2p6r, 2p6u, 2pl3, 2v6i, 2v6j, 2va8, 2vbc, 2vlx, 2z0m, 3b7g, elements are no longer constrained.18 RNA binding 3ber, 3bor, 3bxz, 3ews, 3fhc, 3fht, 3g0h, 3i5x, 3i5y, 3i61, would further activate the β,γ-phosphoanhydride 3i62, and 8ohm. bond for the hydrolysis reaction to occur by an activated water molecule, presumably through Cloning, vectors, and strains distortion of the bond or stabilization of a transition All yeast manipulations were performed using standard state. In accordance with this idea, an AAA 68 mutation in YxiN reduces the cooperative binding techniques. The DED1 coding region (1850 bp), cloned and coupling energies of ATP and RNA, even into the SpeI and XhoI sites of Bluescript plasmid (Stratagene), was used for all PCR amplifications and though the overall relationship between domains 1 “ ” “ ” clonings. Mutations were introduced by a two-step over- and 2 ( closed and open forms) is not different lapping fusion PCR, using PFU polymerase as previously from that of the wild-type protein in the presence described.36,37 Mutations were introduced with oligonu- 52 and in the absence of ligands. In Förster resonance cleotides completely randomized for the three positions of energy transfer analyses, YxiN maintains the the codon at either amino acid position 341 or amino acid “closed” conformation in association with RNA position 343. The two pools of amplified DNAs were and with AMP-PNP, ADP-BeFx, ADP-AlFx, and digested with AccI and StyI, gel purified, and cloned into the equivalent sites of DED1 in the Bluescript plasmid. ADP-MgFx, although the latter two analogs do not support RNA unwinding, indicating that there are Additional oligonucleotides were used to specifically subtle conformational differences.65 insert either alanine or cysteine at position 341, and a glycine at position 342. A double mutant with alanine in This would be consistent with the proposed both positions 341 and 343 was also made. SpeI-XhoI mechanisms on how catalytic antibodies and fragments corresponding to the entire coding region of enzymes use ligand binding energies to stimulate DED1 were then subcloned into the p415-PL-ADH vector 66,67 reactions by stabilizing the transition states. (ARS-CEN-LEU).37 Ded1 yeast plasmids were transformed ATP binding to the protein would create a high- into the Δded1/Δdbp1 strain (ded1∷HIS3 dbp1∷KANMX6) affinity RNA binding site, but RNA binding to the containing the DBP1 open reading frame cloned into the 69 protein would subsequently activate the β,γ-phos- multicopy p416-TEF plasmid; using the DBP1 plasmid phoanhydride bond for hydrolysis. The resulting instead of DED1 minimized recombination between plasmids and the appearance of pseudo revertants. For ADP-bound form would then have a low affinity for – the RNA, and translocation or release of the RNA each screen, we selected about 260 280 transformants that could grow on SD-Leu plates, streaked them on a 5-FOA- would occur. Motif III is ideally situated to containing medium, and then tested them for their ability coordinate the ATP and the different motifs for to grow at 16 °C, 18 °C, 30 °C, and 36 °C. Selected these conformational changes. Thus, previously constructs were recovered by plasmid rescue, their published claims that motif III links the ATPase phenotypes were verified by retransforming yeast, and and “unwinding” activities need to be reconsidered then they were sequenced to identify the mutation. Protein to reflect the possibility that the unwinding activity expression was verified by Western blot analysis, with was indirectly effected by an altered or reduced antibodies against the HA tag present in the constructs. affinity for the nucleic acid ligand within the enzymatic site, or to reflect the possibility that Isolation of intragenic suppressors RNA binding and strand displacement are concur- rent, and that the ATPase activity is only used to Intragenic suppressors of the conditional DED1 T343C recycle the protein from a high-affinity conforma- mutation were obtained by mutagenic PCR using Taq tion to a low-affinity conformation for ligands. DNA polymerase and two oligonucleotides that hybridized at either end of the ded1 T343C open reading frame. Two to four independent cycles of mutagenic PCR were Materials and Methods first performed using 6.5 mM MgCl2, 0.5 mM MnCl2, 1 mM dCTP and dTTP, and 0.2 mM dATP and dGTP. Then, 40 PCR cycles were performed under normal Sequence alignments and phylogeny conditions using 2% of the mutagenized PCR as template. This library of DNA fragments was digested BLAST and sequence alignments with ClustalXXL were with SpeI and XhoI and subcloned into the equivalent performed as previously described.32 The Bork Web site sites of the p415-ADH vector. The DNA was amplified was used to determine the consensus sequence‡, and the in E. coli and then used to transform the Δded1/Δdbp1 ExPASy Stataln tool§ was used to quantify the results. We yeast strain containing the DBP1 plasmid. We restreaked examined the structural context of motif III in the solved 250 clones on 5-FOA plates at 16 °C, 18 °C, 30 °C, and crystal structures of various SF2 families using the 36 °C to eliminate cells retaining the wild-type DBP1 following PDB coordinates: 1cu1, 1d9z, 1db3, 1fuu, 1hei, plasmid. The plasmids that supported growth were 1hv8, 1m74, 1nkt, 1nl3, 1oyw, 1oyy, 1q0u, 1qde, 1qva, recovered and subjected to restriction digestion map- 1s2m, 1t6n, 1tf2, 1tf5, 1vec, 1wp9, 1wrb, 1xti, 1xtj, 1xtk, ping. All the clones that showed wild-type growth were eliminated because they were recombinants with the Dbp1 plasmid, and we ended up with 12 clones that grew better than the original mutant. We verified the ‡ http://coot.embl.de/Alignment/consensus.html phenotypes of the plasmids by retransforming yeast, and § http://www.expasy.ch/tools/stataln.html then we sequenced the DNA. Motif III in Superfamily 2 “Helicases” 963

Protein expression and purification Strand displacement assay

The AgeI fragments of DED1 containing the motif III We used various duplex substrates, including K01– mutations were cloned into the equivalent sites of the pET- Hyb1 and R1–Hyb1, that had been previously 22b expression vector (Novagen) containing the wild-type described.32,36 We also used a new 45-nt-long substrate, DED1. The Rosetta(DE3) E. coli strain (Novagen) was used K06, that was similar to K01, except for the single-stranded for expressing the proteins. Expression and purification region had an additional uridine and had a cytidine were performed as previously described.32,36 Protein changed to a uridine (5′ GGG CUA GCA CCG UAA AGC concentrations were determined by the Bio-Rad Protein AAG UUA AUU CAA AAC AAA ACA AAA GCU 3′, Assay, using bovine serum albumin as standard. The where underlined characters hybridized to the oligonu- proteins were judged to be 90–95% pure on Laemmli SDS cleotide and boldface characters were different from K01); polyacrylamide gels. this was performed to destabilize a hairpin structure that could compete for the Hyb1 binding site. This construct gave less background displacement than the K01–Hyb1 RNA-dependent ATPase assays duplex. The same 16-nt-long oligonucleotide (Hyb1) hybridized to either the 5′ end (K01 and K06) or the 3′ We used a colorimetric assay based on molybdate– end (R1) of the RNAs. We used both DNA and RNA Malachite Green and used whole-yeast RNA (Type III variants of Hyb1 (Hyb1 DNA and Hyb1 RNA) that were Sigma) and ATP (GE-Pharmacia) as substrates, as labeled on the 5′ ends with 32P. The DNA–RNA duplexes previously described for Ded1.36 In brief, reactions had a calculated Gibbs free energy of −19.8 kcal/mol,40,70 were incubated at 30 °C for various times and stopped while the RNA–RNA duplexes had −25.3 kcal/mol.71 by making the reaction mix 4.5 mM in ethylenediami- However, the actual melting temperatures of the different netetraacetic acid and placing it on ice. Data were duplexes varied due to competing intramolecular base- analyzed using Kaleidagraph 4.0.4 (Synergy). Time pairings of the RNAs. courses at each ATP concentration were usually Under our reaction conditions, with a large excess of repeated three independent times. Time courses also substrate over the enzymes, most of the displaced Hyb1 were carried out in the absence of RNA to determine would reanneal on the RNAs. Consequently, we generally the background hydrolysis of the ATP and in the used a 20-fold to 25-fold excess of an unlabeled 18-nt-long absence of ATP to monitor contribution due to RNA oligonucleotide complementary to Hyb1 (α-Hyb1 DNA) degradation. In both cases, the signals were only a as trap to prevent reannealing of the displaced [32P]Hyb1. small percentage of the reactions including ATP and The α-Hyb1 DNA–Hyb1 DNA duplex had a calculated RNA; these background signals were subtracted from Gibbs free energy of −18.4 kcal/mol, and the α-Hyb1 subsequent calculations of the reaction rates. The mean DNA–Hyb1 RNA duplex was −17.0 kcal/mol; the lower- values and standard deviations from the mean were than-expected free energy of the latter was due to the calculated with Kaleidagraph. unusually low stability of the two sets of rUU/dAA base pairs.40 Reactions were carried out at 30 °C for various times, – Measuring protein RNA affinity by EMSA the products were separated by 15% polyacrylamide gel electrophoresis at 4 °C under nondenaturing conditions, We used a 5′ 32P-end-labeled 25-nt-long RNA that was and the corresponding bands were quantified with a chemically synthesized (RNA01; PAGE purified; Dhar- Cyclone PhosphoImager and Optiquant software and macon), as previously described.32 In brief, we used a analyzed with Kaleidagraph. Because the proteins had a constant RNA concentration (5 nM) and varied the protein low intrinsic affinity for the RNA substrates, we often saw concentration. The protein and RNA were incubated ATP-independent strand displacement that was directly together in 20-μl volumes for 20 min on ice with no proportional to the protein concentration. It was low for nucleotide, with 5 mM ADP (Sigma), or with 5 mM AMP- the Ded1 wild-type protein, which was used at a low PNP (Roche). RNasin (1 U/μl; Promega) was added to protein–duplex ratio, but we had significant background reactions containing AMP-PNP because of RNase con- for many of the poorly active mutants that were used at tamination in the product. Loading buffer was then much higher protein concentrations. Moreover, there was added, and the products were separated by electrophore- some strand exchange with the oligonucleotide trap, sis on 6% nondenaturing polyacrylamide gels run at 4 °C. which resulted in an apparent protein-independent Data were quantified with either a Cyclone (Packard) or displacement during the time course. This background Typhoon Trio (Amersham Biosciences) PhosphoImager activity was subtracted from the results. and ImageQuant software (Packard), and then analyzed with Kaleidagraph. Most mutants poorly bound to the RNA, and binding values could not be determined. For the wild type and for some suppressors in isolation of other mutations, we used a more extensive range of Acknowledgements protein concentrations so that we could fit the data to the Hill equation using a single binding site, as previously We thank Ronald Bock, Tien-Hsien Chang, and described (although similar results were obtained when 32 Paul Lasko for the kind gifts of plasmids. We are the number of binding sites was left unfixed). To very much indebted to Costa Georgopoulos for his facilitate comparisons, we counted all of the radioactivities undying support for our endeavors. This work was above the free oligonucleotide as protein-specific binding even though there were significant qualitative differences. supported by the Swiss National Science Founda- Background smearing of radioactivity, from the lane tion, the Canton of Geneva, the Centre National de without added protein, was subtracted from the results. la Recherche Scientifique, and ANR Program Blanc All binding assays were generally conducted multiple grant 08-BLAN-0086-02 to M.D. N.K.T. was sup- times for each protein. ported by the Swiss Institute of Bioinformatics and 964 Motif III in Superfamily 2 “Helicases” as a CDD Chercheur of Centre National de la 16. Story, R. M., Li, H. & Abelson, J. N. (2001). Crystal Recherche Scientifique. J.B. also was supported by a structure of a DEAD box protein from the hyperther- mophile Methanococcus jannaschii. Proc. Natl Acad. Sci. Short-Term European Molecular Biology Organiza- 98 – tion Fellowship. USA, , 1465 1470. 17. Luo, D., Xu, T., Watson, R. P., Scherer-Becker, D., Sampath, A., Jahnke, W. et al. (2008). Insights into RNA unwinding and ATP hydrolysis by the flavivirus Supplementary Data NS3 protein. EMBO J. 27, 3209–3219. 18. Pause, A. & Sonenberg, N. (1992). Mutational analysis Supplementary data associated with this article of a DEAD box RNA helicase: the mammalian can be found, in the online version, at doi:10.1016/ translation initiation factor eIF-4A. EMBO J. 11, j.jmb.2009.12.025 2643–2654. 19. Schwer, B. & Meszaros, T. (2000). RNA helicase dynamics in pre-mRNA splicing. EMBO J. 19, 6582–6591. References 20. Korolev, S., Hsieh, J., Gauss, G. H., Lohman, T. M. & Waksman, G. (1997). Major domain swiveling 1. Pyle, A. M. (2008). Translocation and unwinding revealed by the crystal structures of complexes of E. mechanisms of RNA and DNA helicases. Annu. Rev. coli Rep helicase bound to single-stranded DNA and Biophys. 37, 317–336. ADP. Cell, 90, 635–647. 2. Singleton, M. R., Dillingham, M. S. & Wigley, D. B. 21. Korolev, S., Yao, N., Lohman, T. M., Weber, P. C. & (2007). Structure and mechanism of helicases and Waksman, G. (1998). Comparisons between the nucleic acid translocases. Annu. Rev. Biochem. 76,23–50. structures of HCV and Rep helicases reveal structural 3. Bleichert, F. & Baserga, S. J. (2007). The long unwinding similarities between SF1 and SF2 super-families of road of RNA helicases. Mol. Cell, 27,339–352. helicases. Protein Sci. 7, 605–610. 4. Cordin, O., Banroques, J., Tanner, N. K. & Linder, P. 22. Jamieson, D. J., Rahe, B., Pringle, J. & Beggs, J. D. (2006). The DEAD-box protein family of RNA (1991). A suppressor of a yeast splicing mutation helicases. Gene, 367,17–37. (prp8-1) encodes a putative ATP-dependent RNA 5. Rocak, S., Emery, B., Tanner, N. K. & Linder, P. (2005). helicase. Nature, 349, 715–717. Characterization of the ATPase and unwinding 23. Thuillier, V., Stettler, S., Sentenac, A., Thuriaux, P. & activities of the yeast DEAD-box protein Has1p and Werner, M. (1995). A mutation in the C31 subunit of the analysis of the roles of the conserved motifs. Saccharomyces cerevisiae RNA polymerase III affects Nucleic Acids Res. 33, 999–1009. transcription initiation. EMBO J. 14, 351–359. 6. Silverman, E., Edwalds-Gilbert, G. & Lin, R. J. (2003). 24. Chuang, R. Y., Weaver, P. L., Liu, Z. & Chang, T. H. DExD/H-box proteins and their partners: helping (1997). Requirement of the DEAD-box protein ded1p RNA helicases unwind. Gene, 312,1–16. for messenger RNA translation. Science, 275, 7. Tanner, N. K. & Linder, P. (2001). DExD/H box RNA 1468–1471. helicases: from generic motors to specific dissociation 25. de la Cruz, J., Iost, I., Kressler, D. & Linder, P. (1997). functions. Mol. Cell, 8, 251–262. The p20 and Ded1 proteins have antagonistic roles in 8. Caruthers, J. M. & McKay, D. B. (2002). Helicase eIF4E-dependent translation in Saccharomyces cerevi- structure and mechanism. Curr. Opin. Struct. Biol. 12, siae. Proc. Natl Acad. Sci. USA, 94, 5201–5206. 123–133. 26. Berthelot, K., Muldoon, M., Rajkowitsch, L., Hughes, 9. Jankowsky, E. & Fairman, M. E. (2007). RNA J. & McCarthy, J. E. (2004). Dynamics and processivity helicases—one fold for many functions. Curr. Opin. of 40S ribosome scanning on mRNA in yeast. Mol. Struct. Biol. 17, 316–324. Microbiol. 51, 987–1001. 10. Andersen, C. B., Ballut, L., Johansen, J. S., Chamieh, 27. Beckham, C., Hilliker, A., Cziko, A. M., Noueiry, A., H., Nielsen, K. H., Oliveira, C. L. et al. (2006). Structure Ramaswami, M. & Parker, R. (2008). The DEAD-box of the exon junction core complex with a trapped RNA helicase Ded1p affects and accumulates in DEAD-box ATPase bound to RNA. Science, 313, Saccharomyces cerevisiae P-bodies. Mol. Biol. Cell, 19, 1968–1972. 984–993. 11. Bono, F., Ebert, J., Lorentzen, E. & Conti, E. (2006). The 28. Chong, J. L., Chuang, R. Y., Tung, L. & Chang, T. H. crystal structure of the exon junction complex reveals (2004). Ded1p, a conserved DExD/H-box translation how it maintains a stable grip on mRNA. Cell, 126, factor, can promote yeast L-A virus negative-strand 713–725. RNA synthesis in vitro. Nucleic Acids Res. 32, 12. Sengoku, T., Nureki, O., Nakamura, A., Kobayashi, S. 2031–2038. & Yokoyama, S. (2006). Structural basis for RNA 29. Johnstone, O., Deuring, R., Bock, R., Linder, P., Fuller, unwinding by the DEAD-box protein Drosophila Vasa. M. T. & Lasko, P. (2005). Belle is a Drosophila DEAD- Cell, 125, 287–300. box protein required for viability and in the germ line. 13. von Moeller, H., Basquin, C. & Conti, E. (2009). The Dev. Biol. 277,92–101. mRNA export protein DBP5 binds RNA and the 30. Tarn, W. Y. & Chang, T. H. (2009). The current cytoplasmic nucleoporin NUP214 in a mutually understanding of Ded1p/DDX3 homologs from yeast exclusive manner. Nat. Struct. Mol. Biol. 16, 247–254. to human. RNA Biol. 6,17–20. 14. Del Campo, M. & Lambowitz, A. M. (2009). Structure 31. Rosner, A. & Rinkevich, B. (2007). The DDX3 of the yeast DEAD box protein Mss116p reveals two subfamily of the DEAD box helicases: divergent wedges that crimp RNA. Mol. Cell, 35, 598–609. roles as unveiled by studying different organisms 15. Collins, R., Karlberg, T., Lehtio, L., Schutz, P., van den and in vitro assays. Curr. Med. Chem. 14, 2517–2525. Berg, S., Dahlgren, L. G. et al. (2009). The DEXD/H- 32. Banroques, J., Cordin, O., Doere, M., Linder, P. & box RNA helicase DDX19 is regulated by an α-helical Tanner, N. K. (2008). A conserved phenylalanine of switch. J. Biol. Chem. 284, 10296–10300. motif IV in superfamily 2 helicases is required for Motif III in Superfamily 2 “Helicases” 965

cooperative, ATP-dependent binding of RNA sub- 51. Solem, A., Zingler, N. & Pyle, A. M. (2006). A DEAD strates in DEAD-box proteins. Mol. Cell. Biol. 28, protein that activates intron self-splicing without 3359–3371. unwinding RNA. Mol. Cell, 24, 611–617. 33. Martinez, R., Shao, L. & Weller, S. K. (1992). The 52. Karow, A. R. & Klostermeier, D. (2009). A conserved helicase motifs of the herpes simplex virus conformational change in the helicase core is type 1 origin-binding protein UL9 are important for necessary but not sufficient for RNA unwinding function. J. Virol. 66, 6735–6746. by the DEAD box helicase YxiN. Nucleic Acids Res. 34. Chamot, D., Magee, W. C., Yu, E. & Owttrim, G. W. 37, 4464–4471. (1999). A cold shock-induced cyanobacterial RNA 53. Chang, T. H., Latus, L. J., Liu, Z. & Abbott, J. M. helicase. J. Bacteriol. 181, 1728–1732. (1997). Genetic interactions of conserved regions in the 35. Daugeron, M. C. & Linder, P. (2001). Characterization DEAD-box protein Prp28p. Nucleic Acids Res. 25, and mutational analysis of yeast Dbp8p, a putative 5033–5040. RNA helicase involved in ribosome biogenesis. 54. Bernstein, K. A., Granneman, S., Lee, A. V., Man- Nucleic Acids Res. 29, 1144–1155. ickam, S. & Baserga, S. J. (2006). Comprehensive 36. Cordin, O., Tanner, N. K., Doere, M., Linder, P. & mutational analysis of yeast DEXD/H box RNA Banroques, J. (2004). The newly discovered Q motif of helicases involved in large ribosomal subunit biogen- DEAD-box RNA helicases regulates RNA-binding esis. Mol. Cell. Biol. 26, 1195–1208. and helicase activity. EMBO J. 23, 2478–2487. 55. Granneman, S., Bernstein, K. A., Bleichert, F. & 37. Tanner, N. K., Cordin, O., Banroques, J., Doere, M. & Baserga, S. J. (2006). Comprehensive mutational Linder, P. (2003). The Q motif: a newly identified analysis of yeast DEXD/H box RNA helicases motif in DEAD box helicases may regulate ATP required for small ribosomal subunit synthesis. Mol. binding and hydrolysis. Mol. Cell, 11, 127–138. Cell. Biol. 26, 1183–1194. 38. Cann, J. R. (1998). Theoretical studies on the mobility- 56. Hotz, H. R. & Schwer, B. (1998). Mutational analysis of shift assay of protein–DNA complexes. Electrophoresis, the yeast DEAH-box splicing factor Prp16. Genetics, 19, 127–141. 149, 807–815. 39. Lane, D., Prentki, P. & Chandler, M. (1992). Use of gel 57. Martin, A., Schneider, S. & Schwer, B. (2002). Prp43 is retardation to analyze protein–nucleic acid interac- an essential RNA-dependent ATPase required for tions. Microbiol. Rev. 56, 509–528. release of lariat-intron from the spliceosome. J. Biol. 40. Sugimoto, N., Nakano, S., Katoh, M., Matsumura, A., Chem. 277, 17743–17750. Nakamuta, H., Ohmichi, T. et al. (1995). Thermody- 58. Plumpton, M., McGarvey, M. & Beggs, J. D. (1994). A namic parameters to predict stability of RNA/DNA dominant negative mutation in the conserved RNA hybrid duplexes. Biochemistry, 34, 11211–11246. helicase motif ‘SAT’ causes splicing factor PRP2 to 41. Barhoumi, M., Tanner, N. K., Banroques, J., Linder, P. stall in spliceosomes. EMBO J. 13, 879–887. & Guizani, I. (2006). Leishmania infantum LeIF protein 59. Kim, D. W., Kim, J., Gwack, Y., Han, J. H. & Choe, J. is an ATP-dependent RNA helicase and an eIF4A-like (1997). Mutational analysis of the hepatitis C virus factor that inhibits translation in yeast. FEBS J. 273, RNA helicase. J. Virol. 71, 9400–9409. 5086–5100. 60. Tai, C. L., Pan, W. C., Liaw, S. H., Yang, U. C., Hwang, 42. Nielsen, K. H., Chamieh, H., Andersen, C. B., L. H. & Chen, D. S. (2001). Structure-based mutational Fredslund, F., Hamborg, K., Le Hir, H. & Andersen, analysis of the hepatitis C virus NS3 helicase. J. Virol. G. R. (2009). Mechanism of ATP turnover inhibition in 75, 8289–8297. the EJC. RNA, 15,67–75. 61. Marintcheva, B. & Weller, S. K. (2001). Residues 43. Dittrich, M. & Schulten, K. (2005). Zooming in on ATP within the conserved helicase motifs of UL9, the hydrolysis in F1. J. Bioenerg. Biomembr. 37, 441–444. origin-binding protein of herpes simplex virus-1, are 44. Petsko, G. A. (2000). Chemistry and biology. Proc. Natl essential for helicase activity but not for dimerization Acad. Sci. USA, 97, 538–540. or origin binding activity. J. Biol. Chem. 276, 45. Gross, C. H. & Shuman, S. (1998). The nucleoside 6605–6615. triphosphatase and helicase activities of vaccinia virus 62. Karginov, F. V., Caruthers, J. M., Hu, Y., McKay, D. B. NPH-II are essential for virus replication. J. Virol. 72, & Uhlenbeck, O. C. (2005). YxiN is a modular protein 4729–4736. combining a DEx(D/H) core and a specific RNA- 46. Gavin, A. C., Aloy, P., Grandi, P., Krause, R., Boesche, binding domain. J. Biol. Chem. 280, 35499–35505. M., Marzioch, M. et al. (2006). Proteome survey reveals 63. Chen, Y., Potratz, J. P., Tijerina, P., Del Campo, M., modularity of the yeast cell machinery. Nature, 440, Lambowitz, A. M. & Russell, R. (2008). DEAD-box 631–636. proteins can completely separate an RNA duplex 47. Tarassov, K., Messier, V., Landry, C. R., Radinovic, S., using a single ATP. Proc. Natl Acad. Sci. USA, 105, Serna Molina, M. M., Shames, I. et al. (2008). An in vivo 20203–20208. map of the yeast protein interactome. Science, 320, 64. Liu, F., Putnam, A. & Jankowsky, E. (2008). ATP 1465–1470. hydrolysis is required for DEAD-box protein recycling 48. Yedavalli, V. S., Neuveut, C., Chi, Y. H., Kleiman, L. & but not for duplex unwinding. Proc. Natl Acad. Sci. Jeang, K. T. (2004). Requirement of DDX3 DEAD box USA, 105, 20209–20214. RNA helicase for HIV-1 Rev-RRE export function. 65. Aregger, R. & Klostermeier, D. (2009). The DEAD box Cell, 119, 381–392. helicase YxiN maintains a closed conformation during 49. Zhang, M. & Green, M. R. (2001). Identification and ATP hydrolysis. Biochemistry, 48, 10679–10681. characterization of yUAP/Sub2p, a yeast homolog of 66. Houk, K. N., Leach, A. G., Kim, S. P. & Zhang, X. the essential human pre-mRNA splicing factor (2003). Binding affinities of host–guest, protein– hUAP56. Genes Dev. 15,30–35. ligand, and protein–transition-state complexes. 50. Del Campo, M., Tijerina, P., Bhaskaran, H., Mohr, S., Angew. Chem. Int. Ed. Engl. 42, 4872–4897. Yang, Q., Jankowsky, E. et al. (2007). Do DEAD-box 67. Wentworth, P., Jr. & Janda, K. D. (2001). Catalytic proteins promote group II intron splicing without antibodies: structure and function. Cell Biochem. unwinding RNA? Mol. Cell, 28, 159–166. Biophys. 35,63–87. 966 Motif III in Superfamily 2 “Helicases”

68. Guthrie, C. & Fink, G. R. (1991). Guide to yeast 70. Bommarito, S., Peyret, N. & SantaLucia, J., Jr. genetics and molecular biology. In Methods in Enzy- (2000). Thermodynamic parameters for DNA mology (Abelson, J. N. & Simon, M. I., eds), Methods in sequences with dangling ends. Nucleic Acids Res. Enzymology, vol. 194, pp. Academic Press, San Diego. 28, 1929–1934. 69. Mumberg, D., Muller, R. & Funk, M. (1995). Yeast vectors 71. Turner, D. H., Sugimoto, N. & Freier, S. M. (1988). for the controlled expression of heterologous proteins in RNA structure prediction. Annu. Rev. Biophys. Biophys. different genetic backgrounds. Gene, 156,119–122. Chem. 17, 167–192.