D Small Ribonucleoprotein Particles

doi:10.1016/j.jmb.2003.08.012

J. Mol. Biol. (2003) 333, 295–306

Functional Requirement for Symmetric Assembly of Archaeal Box C/D Small Ribonucleoprotein Particles Rumana Rashid1†, Mohamed Aittaleb1†, Qiong Chen1 Katharina Spiegel2, Borries Demeler3 and Hong Li1* 1 Department of Chemistry and Biochemistry, Institute of Molecular Biophysics, Florida State University, Tallahassee FL 32306, USA 2

Keck Biophysics Facility Northwestern University Evanston, IL 60208, USA 3 Department of Biochemistry University of Texas Health Science Center at San Antonio San Antonio, TX 78284, USA

Box C/D small ribonucleoprotein particles (sRNPs) are archaeal homologs of small nucleolar ribonucleoprotein particles (snoRNPs) in eukaryotes that are responsible for site specific 20 -O-methylation of ribosomal and transfer RNAs. The function of box C/D sRNPs is characterized by stepwise assembly of three core proteins around a box C/D RNA that include fibrillarin, Nop5p, and L7Ae. The most distinct structural feature in all box C/D RNAs is the presence of two conserved box C/D motifs accompanied by often a single, and sometimes two, antisense elements located immediately upstream of either the D or D0 box. Despite this asymmetric distribution of antisense elements, the bipartite feature of the box C/D motifs appears to be in pleasing agreement with a recently reported three-dimensional structure of the core protein complex between fibrillarin and Nop5p. This investigates functional implications of the symmetric features both in box C/D RNAs and in the fibrillarin –Nop5p complex. Site-directed mutagenesis was employed to generate box C/D RNAs lacking one of the two box C/D motifs and a mutant fibrillarin – Nop5p complex deficient in self-association. The ability of the mutated components to assemble and to direct methyl transfer reactions was assessed by gel mobility-shift, analytical ultracentrifugation, and in vitro catalysis studies. The results presented here suggest that, while a box C/ D sRNP is capable of asymmetrical assembly, the symmetries in both the box C/D RNA and in the fibrillarin –Nop5p complex are required for efficient catalysis. These findings underscore the importance of functional assembly in methyl transfer reactions. q 2003 Elsevier Ltd. All rights reserved.

*Corresponding author

Keywords: RNP assembly; 20 -O-methylation; rRNA biogenesis; RNA methyltransferase; small guide RNA

Introduction Eukaryotic organisms utilize an unprecedented mechanism to modify and process their ribosomal RNAs (rRNAs). In the nucleolus, where biogenesis of cytoplasmic ribosomes takes place, hundreds of small nucleolar ribonucleoprotein particles (snoRNPs) exist to catalyze site-specific modification and processing of nascent ribosomal RNA † R.R. and M.A. contributed equally to this work. Abbreviations used: snoRNP, small nucleolar ribonucleoprotein particle; snoRNA, the RNA molecule in an snoRNP; snRNA, small nuclear RNA; AF, Archaeoglobus fulgidus; SV, sedimentation velocity; vHW, van Holde & Weischet; SE, sedimentation equilibrium. E-mail address of the corresponding author: [email protected]

transcripts.1 – 4 An snoRNP contains multiple components, each of which plays a specific functional role in enzyme assembly and catalysis. The RNA molecule in an snoRNP (snoRNA) functions as a “guide” in substrate recognition through the formation of an RNA duplex with the target RNA while snoRNP proteins catalyze the actual modification or processing reactions. There are two major families of guide snoRNAs as defined by their conserved internal sequence elements (boxes): the box C/D and box H/ACA snoRNAs. The majority of box C/D snoRNAs guide sitespecific 20 -O-methylation of pre-rRNA ribose sugars by base-pairing with pre-rRNA at selected sites of modification. By using the same substrate recognition strategy, the box H/ACA snoRNAs direct site-specific pseudouridination of rRNA nucleotides. In addition to function in rRNA

0022-2836/$ - see front matter q 2003 Elsevier Ltd. All rights reserved.

296

modification, box C/D snoRNPs methylate small nuclear RNAs (snRNAs) U6,5 and U5,6 as well as mRNAs.7,8 Box H/ACA snoRNPs have been shown to guide pseudouridination of U2 snRNA9 in eukaryotic cells, suggesting the diverse functional roles of snoRNPs in modifying all types of RNAs. Homologs of box C/D snoRNPs have been identified in all sequenced archaeal genomes, and some of them have been shown to function in methylation of ribosomal and transfer RNA in Archaea.10 – 14 The function of an snoRNP is characterized by step-wise assembly of snoRNP proteins around the conserved box sequences of an snoRNA. The central feature in box C/D RNAs is the presence of two homologous sets of sequence motifs: two C-like boxes, C and C0 , with the sequence RUGAUGA (where R stands for any purine) and two D-like boxes, D and D0 , with the sequence CUGA. Box C (C0 ) and D (D0 ) sequences act synergistically in methyl transfer reactions,1 and in maintaining snoRNA stability and nuclear localization.15,16 Frequently one, but sometimes two, antisense elements complementary to the ribosomal RNA sequences are located immediately upstream of either the D, or the D0 , (or both) boxes. The antisense element forms a duplex with the target RNA and guides 20 -O-methylation of the target RNA nucleotide exactly five base-pairs away from the cytidine base of the adjacent D box. This has been referred to as the N þ 5 rule.17 Each of the two box C/D motifs is presumed to constitute the required signature for a methyltransferase to bind in proximity to the target nucleotide and has been shown to function independently of each other in yeast.18 Despite the high degree of sequence conservation in the C/D box among all known snoRNAs, box C0 /D0 sequences frequently are not conserved. Because of this lack of consensus, and yet the sequences upstream of the box D0 still function as guides, the two box C/D motifs may have different mechanism for recruiting the methyltransferases. Understanding the roles in snoRNP assembly of (1) the interaction between the snoRNP proteins and the individual box C/D motifs, and (2) the synergism between the two motifs will enable understanding of how these components are essential to the snoRNP’s enzymatic function. In yeast, four proteins have been identified to be the core components of snoRNPs, fibrillarin (Nop1p), Nop58p, Nop56p, and Snu13p. Fibrillarin is the most likely candidate for the catalytic subunit because of its structural homology to the known methyltransferases.13,19,20 Nop58p and Nop56p have close sequence homology to each other and were shown to be essential nucleolar proteins associated with the box C/D RNAs.21 – 23 In Archaea, most of the sequenced archaeal genomes contain open reading frames that encode the archaeal homologs of these eukaryotic proteins: fibrillarin, Nop5p, and L7Ae. However, archaeal proteins are generally smaller and lack some

Assembly and Catalytic Function of Box C/D sRNPs

eukaryote-specific domains such as the GAR domain in fibrillarin and the KExD domain in Nop56p and Nop58p. The first clue concerning the function of Nop56p/58p came from structural studies of their archaeal homologs. A recently determined crystal structure of the fibrillarin– Nop5p protein complex from Archaeoglobus fulgidus (AF) revealed the mode of fibrillarin – Nop5p interaction as well as that of Nop5p self-association.24 A coiled-coil domain of Nop5p brings together two Nop5p molecules, each of which is joined to one fibrillarin molecule through its NH2 domain. The unpredicted quaternary structure for these two core pro˚ teins placing two catalytic subunits nearly 80 A apart appeared to suggest two active sites. The symmetric disposition of the two core proteins fits well with the bipartite feature in archaeal box C/D RNAs, lending the first structural support for a two-site assembly model. Previous evidence for the two-site model was provided by Cahill et al.25 Their chemical cross-linking study of microinjected U25 RNA into Xenopus oocytes with snoRNP proteins placed the Nop56p –fibrillarin proteins in proximity with the box C0 /D0 sequences and the Nop58p– fibrillarin proteins with the box C/D sequences. The structural feature central to the two-site model is the quaternary association of the fibrillarin –Nop5p complex. Although the symmetric dimer of two AF fibrillarin –Nop5p heterodimers can be assumed to interact satisfactorily with the bipartite structure in archaeal box C/D RNA, the functional implication of such oligomerization in sRNP assembly and catalysis is unknown. Indeed, current biochemical results on eukaryotic snoRNP proteins have painted a complicated picture of an asymmetric but inter-dependent binding pattern. Co-immunoprecipitation studies show that yeast Nop58p binds to box C/D RNA independent of fibrillarin, while Nop56p binding to box C/D RNA requires the presence of fibrillarin.21 – 23 In yeast, the fact that Nop56p is not required for snoRNA stability, while Nop58p is, suggests the absence of Nop56p from the terminal stem region.26 In the nucleotide analog interference mapping and in vitro binding studies using yeast U37 box C/D RNA in Xenopus oocytes, the 15.5 kDa protein was required to bind to only one of the two box C/D motifs for snoRNP assembly in vivo.27 These data indicate a differential interaction of the core proteins with box C/D RNAs that is supported by the different degree of nucleotide conservation in the two box C/D motifs. However, these in vivo data are difficult to confirm in vitro until purified enzyme components can be obtained. L7Ae is the archaeal homolog of the 15.5 kDa protein in human and Snu13p in yeast.28,29 The role of L7Ae and its eukaryotic homologs in initiating the assembly of the ribonucleoprotein enzyme has been demonstrated clearly in a number of systems. It is the first to bind, followed by subsequent

297

Assembly and Catalytic Function of Box C/D sRNPs

assembly of fibrillarin and Nop56p/Nop58p proteins.29,30 Data from a number of laboratories have established that L7Ae and its homologs interact specifically with the conserved box C and box D sequences.25,28,29,31 This binding site was predicted to form a kink-turn motif similar to those formed by the 50 stem-loop of U4 snRNA32 and the 23 S rRNA.33 This initial interaction established between Snu13p and the box C/D RNA is required for assembly of fibrillarin and Nop56p/Nop58p proteins through a mechanism that is currently not understood. In light of the facts that (1) the AF fibrillarin –Nop5p complex forms a symmetric dimer in solution and (2) all archaeal sRNAs contain two box C/D motifs, the holoenzyme appears to be assembled in a bipartite form that would require two bound L7Ae proteins. However, the bipartite assembly model would contradict the results from the nucleotide analog interference mapping and binding studies of the eukaryotic snoRNPs in which only a single L7Ae homolog, the 15.5 kDa protein, was found to be bound at the box C/D motif of U37 snoRNA.27 A detailed study on L7Ae binding stoichiometry and its influence on subsequence assembly are required to reconcile these differences. In order to address the functional requirement for the two-site model in archaeal sRNPs, we carried out in vitro assembly and catalysis studies with purified AF sRNP components. We show that at low concentrations, the L7Ae protein interacts independently with the two box C/D motifs and binds to a single box C/D motif, while at high concentrations, L7Ae binds to both sites. Surprisingly, the sRNP enzyme assembled in the presence of low concentrations of L7Ae is nearly as active as the wild-type enzyme. The symmetric fibrillarin –Nop5p complex is able to assemble with box C/D RNAs composed of either one or both box C/D motifs bound to L7Ae. However, box C/D RNAs containing both box C/D and box C0 /D0 motifs show the highest levels of activity in guiding methylation reactions. A fibrillarin – Nop5p mutant that is deficient in self-association exhibits a nearly wild-type level of assembly activity. In contrast, its catalytic efficiency was reduced greatly, suggesting a functional requirement for the symmetric assembly of the fibrillarin– Nop5p complex in methyl transfer reactions. On the basis of the data presented, we propose that self-association between the two Nop5p –fibrillarin complexes is the mechanism responsible for recruiting the methyltransferase, fibrillarin, to the less conserved box C0 /D0 motif in the absence of L7Ae binding.

Results

gel mobility-shift assay that allowed detection of each specific archaeal sRNP complex. We selected two predicted box C/D RNAs in AF, sR1 and sR3† as gel shift substrates. Note that sR3 sRNA is embedded within the intron of precursor tRNAtrp (pre-tRNAtrp) and contains two antisense elements complementary to the pre-tRNAtrp itself. This sRNA was shown experimentally to guide cismethylation of U34 and C39 of tRNAtrp.12 The archaeal homolog of Snu13p, L7Ae, was cloned from AF genomic DNA with an N-terminal polyhistidine tag and was purified as described in Materials and Methods. Purified recombinant L7Ae protein shifted both sR1 and sR3 RNAs strongly and specifically. AF fibrillarin and Nop5p proteins were expressed and co-purified as a single complex from bacterial cells.24 We carried out binding titrations of radioactively labeled sR3 and sR1 RNAs with the L7Ae protein and monitored gel mobility-shift pattern of the RNAs due to L7Ae binding. As shown in Figure 1, the L7Ae-containing lanes in sR3 (Figure 1(B)) showed two slowly migrating complexes that were dependent upon the concentration of L7Ae used in titration. The same pattern was seen in sR1 titrations (data not shown). This gel shift pattern is consistent with the lower shifted band being the 1:1 molar ratio L7Ae:RNA complex and the upper band being the 2:1 molar ratio L7Ae:RNA complex. This shows that L7Ae bound to one box C/D site at low concentration and to both sites of a box C/D RNA at higher concentrations. We tested this conclusion by repeating L7Ae binding titrations with two mutant sR3 RNAs. The mutations were within the two tandem GA pairs shown previously to be detrimental for L7Ae binding.28,31 The adenosine nucleotides A46 in box C and A72 in box C0 (Figure 1(A)), presumed to form the second GA pair from the flipped-out U in the proposed kink-turn motif were mutated to cytidine, and the mutants were designated sR3 boxC(A/C) and sR3 boxC0 (A/C), respectively. The resulting mutants contained one defective and one functional kink-turn motif for binding to L7Ae. As expected, the gel shift patterns of both mutants with L7Ae exhibited the 1:1 complex shifts over the entire range of concentrations used in titration, consistent with the interpretation that a single L7Ae molecule bound to these mutant RNAs (Figure 1(B)). The binding affinities of L7Ae for either sR3 mutant did not appear to be different, although the shifted sR3 boxC0 (A/C) complex migrated slightly differently from the shifted sR3 boxC(A/C) complex, possibly due to a different RNA conformation. When the full-length pretRNAtrp RNA was used in binding reactions, only a single shifted band was observed throughout the titration concentration range (data not shown). This is likely the result of the large pre-tRNAtrp

AF L7Ae protein can interact with both box C/D and box C0 /D0 motifs The snoRNP assembly process was studied via a

† http://rna.wustl.edu/snoRNAdb/Archaea/ Afu-align.html

298

Assembly and Catalytic Function of Box C/D sRNPs

Figure 1. L7Ae interacts with both box C/D motifs. A, Primary and predicted secondary structure of the box C/D sR3 RNA that was used in the experiment. Conserved nucleotides in boxes C and D are shown by bold letters. Mutated nucleotides are indicated by blocked letters and the names of the corresponding mutants are labeled. B, In vitro transcribed, 32P-labelled box C/D RNAs were incubated with increasing amounts of the AF L7Ae protein. Resulting RNP complexes were resolved on a non-denaturing 7.5% (w/v) polyacrylamide gel. The positions of free RNA and RNP complexes are indicated by arrows. In lanes 2– 12, RNAs were incubated with L7Ae protein at concentrations of 0.12, 0.50, 1.0, 2.0, 2.5, 3.0, 3.5, 4.0, 5.0, 6.0, and 7.0 mM, respectively. The 1:1 and 1:2 RNA:L7Ae complexes are labeled.

(126 nt) preventing the gel resolution of either titration complex. Consistent with this interpretation, a single shifted band was observed in binding titrations of U37 snoRNA (, 99 nt) with the human 15.5 kDa protein.27 The importance of the GA pair in mediating the RNA’s kink-turn interaction with L7Ae is already established; this binding study confirms it further. More significantly, it demonstrates clearly that L7Ae interacts with the two box C/D motifs independently. In contrast to the functional importance of the A nucleotides in box C and C0 sequences, mutations of the A nucleotides in either box D or box D0 (sR3 box D(A/C) and sR3 boxD0 (A/C)) had little effect on L7Ae binding (Figure 1(B)). Again, two specific RNP complexes were observed, suggesting a twosite binding model for L7Ae that is dependent on its concentration. This finding is consistent with the fact that methylation activity has a looser requirement for box D0 than for box C0 sequences.25 However, despite the fact that L7Ae binding to box C/D RNA tolerated A to C changes in D boxes, removing the exocyclic amine group from A89 in box D of U37 RNA had completely abolished binding of 15.5 kDa to U37.27 This suggests a leading

role of proper hydrogen bonding in 15.5 kDa-box C/D RNA interactions. Symmetric assembly of L7Ae on box C/D RNA is not required for fibrillarin – Nop5p assembly Next, we investigated whether the subsequent assembly of the fibrillarin– Nop5p complex requires the 2:1 molar ratio L7Ae:RNA complex in formation of a bipartite particle. We employed the same sR3 mutants with single L7Ae-binding sites in this experiment. Our hypothesis was that if the fibrillarin –Nop5p dimer required the presence of two L7Ae proteins for assembly, then the fibrillarin –Nop5p dimer would not further gel shift from the L7Ae –sR3 mutant RNA complexes. Following the stepwise assembly process, we titrated the RNA with L7Ae and then subjected the L7Ae – sR3 mutant RNA complexes to binding reactions with the fibrillarin– Nop5p complex. Surprisingly, both sR3 mutant RNAs were found to be fully capable of assembly with the fibrillarin – Nop5p symmetric dimer complex in the presence of only one L7Ae (Figure 2), indicating that a symmetric 2:1 molar ratio L7Ae:RNA complex is not required

Assembly and Catalytic Function of Box C/D sRNPs

299

Figure 2. Asymmetric assembly of sRNPs. The two sR3 mutants, sR3 boxC(A/C) and sR3 boxC0 (A/C), lacking one binding site for L7Ae were incubated with increasing amounts of the fibrillarin – Nop5p complex. Lanes 1– 6 contain 2 nM sR3 boxC(A/C) RNA and lanes 7– 12 contain 2 nM sR3 boxC0 (A/C) RNA. These results show that the fibrillarin –Nop5p complex is able to associate with asymmetric box C/D RNAs that interact with a single L7Ae protein. We note that the free sR3 boxC(A/C) RNA lane (lane 1) showed several bands that likely resulted from alternative RNA conformers. However, these alternative conformers converged to a single L7Ae– RNA complex upon the addition of L7Ae. RNP1 corresponds to the 1:1 RNA– L7Ae complex and RNP2 corresponds to the RNA– L7Ae– fibrillarin – Nop5p complex.

in binding of the symmetric dimer of fibrillarin – Nop5p. A box C/D RNA bound with a single L7Ae molecule is sufficient in assembly with the fibrillarin –Nop5p symmetric dimer. The binding affinity of fibrillarin – Nop5p to both of the sR3boxC(A/C) or the sR3-boxC0 (A/C) mutants appears the same. This result, together with the fact that many eukaryotic box C/D RNAs lack the consensus sequences in their box C0 /D0 motifs and are thus unable to associate with 15.5 kDa protein,30 lends further support to the asymmetric assembly model with respect to 15.5 kDa protein binding proposed earlier.27 Dimerization of Nop5p by the coiled-coil domain is not required in its interaction with box C/D RNA In the AF fibrillarin –Nop5p crystal, the complex forms a symmetric dimer of two fibrillarin – Nop5p heterodimers.24 The dimerization interface involves the packing of two coiled-coils against each other that is described frequently in terms of ridges into grooves. Based on the large buried solvent-accessible surface at the interface, it was predicted that the fibrillarin– Nop5p complex undergoes rapid self-association in solution.24 We investigated the functional requirement for oligomerization of the fibrillarin – Nop5p complex by combining analytic ultracentrifugation and mutagenesis experiments. A series of sedimentation velocity (SV) experiments was first performed for the wild-type fibrillarin –Nop5p protein in a number of buffer conditions at 30 8C over a range of protein concentrations. Sedimentation velocity data were analyzed both by the integral sedimentation coefficient distribution GðsÞ obtained from the van Holde & Weischet (vHW) analysis,34,35 and by direct boundary modeling using a distribution of Lamm

equation solutions cðsÞ:36 While the GðsÞ distribution offers a model-independent approach by separating diffusional boundary spreading from the sedimentation boundary spreading, the cðsÞ plot often exhibits fine structures in sedimentation coefficient distributions. As illustrated in Figure 3, both GðsÞ and cðsÞ plots for the wild-type fibrillarin– Nop5p complex reveal a clear concentrationdependent behavior in the fibrillarin– Nop5p complex that is indicative of self-association. There are primarily two sedimentation species of the fibrillarin – Nop5p complex, one at , 5.3 S and the other at , 7.7 S. By taking the advantage of the available three-dimensional coordinates for the fibrillarin – Nop5p complex, we computed its hydrodynamic properties including s20,w using a bead-modeling methodology.37 We found that the , 5.3 S species agrees well with the computed s20,w value (5.45 S) based on the symmetric dimer of the fibrillarin – Nop5p complex. The exact nature of the , 7.7 S species is unclear but its formation seems to be disrupted by the presence of glycerol, because it was absent from another SV run when 5% (v/v) glycerol was added to the protein buffer (data not shown). The SV results indicate that under the conditions used for this experiment, the fibrillarin – Nop5p forms the same symmetric dimer of two heterodimers (, 5.3 S) as observed in the crystal with the tendency to self-associate into a tetramer of two symmetric dimers, forming a complex of , 7.7 S. These data indicate that the symmetric dimer of the fibrillarin –Nop5p complex exhibits no tendency to dissociate into single heterodimers within the concentration range (1.6 –16.5 mM) used for the SV experiments. To better understand the self-association behavior of the fibrillarin –Nop5p complex, sedimentation equilibrium (SE) data were obtained under the same buffer conditions as those used for SV runs, employing concentrations corresponding

300

Assembly and Catalytic Function of Box C/D sRNPs

Figure 3. Coiled-coil domain mediates self-association of fibrillarin – Nop5p. Sedimentation coefficient distribution of the wild-type fibrillarin –Nop5p is plotted at the bottom and that of the L97K/I112E double mutant is plotted at the top. The amplitudes of GðsÞ and cðsÞ for both protein samples were scaled vertically to an arbitrary unit for the purpose of comparison. The two values (3.66 S and 5.45 S) as indicated on the s20,w axis were computed s20,w values for the monomeric and dimeric fibrillarin – Nop5p, respectively, based on the crystal structures by using the program HYDROPRO. The chemical equilibrium for each protein complex is indicated on corresponding plots. These results clearly

301

Assembly and Catalytic Function of Box C/D sRNPs

to initial A230 and A280 values of 0.3. SE data could be fit adequately to a model for an ideal monomer– dimer equilibrium with 110,600 Da as the molecular mass of the monomer and a 6.0 mM dissociation constant (Figure 4(A) and (B)). The fitted molecular mass for the monomer species corresponds well with the calculated molecular mass of the fibrillarin –Nop5p symmetric dimer (111,746 Da). In order to observe the fibrillarin– Nop5p heterodimer species in solution and to understand its function in sRNP assembly and catalysis, we selectively disrupted the coiled-coil interaction interface. We mutated Leu97 to lysine and Ile112 to glutamate in Nop5p and named it the L97K/I112E double mutant. We subjected the mutant protein complex to both SV and SE experiments under conditions identical with those used for the wild-type. The GðsÞ; cðsÞ plots and the SE results of the double mutant were then compared to those of the wildtype (Figures 3 and 4). As expected, SV data for the double mutant exhibited clear concentrationdependent, self-dissociation behavior between the two main species, 3.6 S and 5.0 S, not seen in the wild-type protein. The 5.0 S species likely corresponds to the symmetric dimer and is formed only at higher concentrations of protein (Figure 3). The 3.6 S species fit well with the s20,w values computed by the bead-model using the coordinates for the single fibrillarin – Nop5p heterodimer (3.66 S). In agreement with the SV data, SE analysis showed a good fit to ideal monomer – dimer equilibrium with 58,450 Da as the molecular mass for the monomeric species, which would be the heterodimer with the calculated molecular mass of 55,873 Da (Figure 4(C) and (D)). The dissociation constant was determined to be 9.03 mM. These results demonstrate clearly that the coiled-coil domain is responsible for dimerization of two fibrillarin –Nop5p heterodimers in solution. Next, we compared the assembly of wild-type and double mutant fibrillarin –Nop5p with the L7Ae –sR3 RNA complex. Interestingly, the coiledcoil double mutant was able to assemble with the L7Ae –RNA complex at nearly wild-type affinity, as indicated by its gel shift with the L7Ae – sR3 RNA (Figure 5). The double mutant-bound complex migrated faster than that shifted by the wildtype fibrillarin – Nop5p, reflecting the smaller size of a bound single fibrillarin –Nop5p heterodimer. A minor band in the double mutant shifting lane was seen at the same molecular mass as that of the shifted wild-type complex, which likely resulted from the presence of a small amount of dimeric species in the binding reactions. When the double mutant fibrillarin – Nop5p complex was incubated with sR3 boxC(A/C) and sR3 boxC0 (A/C), the double mutant again was able to

shift the L7Ae-bound RNA (Figure 5). We conclude that symmetric dimerization of fibrillarin – Nop5p is not required in sRNP assembly. This is supported by our previous results in mapping the RNA binding surface on the fibrillarin –Nop5p complex, in which the COOH domain of Nop5p was found to be essential in binding RNA.24 Thus, a minimum sRNP assembly is devoid of any internal symmetry and consists of one L7Ae molecule, one fibrillarin– Nop5p heterodimer, and a single box C/D RNA motif. Functional requirements for symmetric assembled core proteins in catalysis The assemblies of known molecular stoichiometry created from site-directed mutagenesis provide an opportunity to assess how sRNP assembly dictates methyltransferase reaction. We employed a methylation assay method similar to that used for DNA methylation as described.38 In each methylation assay, sRNP complex contained fibrillarin – Nop5p dimer complex (or the L97K/I112E mutant), L7Ae, and one of the following guide RNAs: wild-type sR3, sR3 boxC(A/C), or sR3 boxC0 (A/C). Two RNA oligomers of 12– 14 nucleotides in length that are complementary to the two guide RNA regions upstream of box D and box D0 in sR3, respectively, were used as the methylation targets. Only one target RNA was used at a time. Each sRNP complex was incubated with [methyl-3H]S-adenosyl-L -methionine ([3H]AdoMet) at 70 8C. Aliquots of reaction mixture terminated at specific time-points were spotted onto an anionexchange filter (DE81, Whatman). The methylated RNA that was retained on the positively charged filters after an extensive wash was counted in a scintillation counter. Figure 6 shows the time-courses of methylation reactions for various sRNP assemblies. The methylation activities towards the target RNA complementary to upstream of the box D in sR3 RNA catalyzed by the wild-type enzyme and the wildtype sR3 RNA with different concentrations of L7Ae protein were compared. A small difference in activities was observed between the reactions containing either a low or a high concentration of L7Ae. This suggests that a single bound L7Ae molecule is sufficient to initiate the assembly of catalytically competent sRNPs. However, at high concentrations of L7Ae, where the wild-type sR3 binds two L7Ae molecules, the sRNP enzyme has the highest level of activity (Figure 6). In contrast to the flexible requirement for L7Ae binding stoichiometry, the symmetric dimer of the fibrillarin– Nop5p complex is essential in catalysis. As shown in Figure 6, the L97K/I112E double

show that the double mutant dissociated from the symmetric dimer to two heterodimers at micromolar concentrations while the wild-type protein remained as a tightly associated symmetric dimer at the lowest concentration used for the SV experiments.

302

Assembly and Catalytic Function of Box C/D sRNPs

Figure 4. Sedimentation equilibrium results indicate a monomer – dimer association for both the wild-type fibrillarin – Nop5p and the L97K/I112E double mutant. In (B) and (D) combined molecular mass versus radius squared are plotted with smooth curves generated from simulation using a monomer– dimer equilibrium. Fitting residuals are displayed in (A) and (C) for the wild-type and the double mutant data, respectively.

mutant using the wild-type sR3 RNA as guide showed greatly declined methylation activity towards the same methylation target oligonucleotide associated with box D of the sR3 RNA. We thus conclude that the symmetric dimer is required for in vitro methylation activity. Surprisingly, the bipartite feature in box C/D RNA was shown to be equally important in catalysis. We tested the guiding ability of the two mutant RNAs that contain an A to C mutation in their respective box C or C0 . We assembled reaction mixtures containing the wild-type fibrillarin – Nop5p complex, L7Ae proteins, either the sR3 boxC0 (A/C) or sR3 boxC(A/C) mutant RNA, and the target RNA complementary to the sequences upstream of box D or D0 respectively. Both sR3 boxC0 (A/C) RNA and sR3 boxC(A/C) showed deficiency in guiding methylation reactions (Figure 6), suggesting the importance of the bipartite structure of box C/D RNAs in catalysis. Interestingly, the sR3 boxC(A/C) mutant showed

significantly less activity than the sR3 boxC0 (A/C) mutant. This result could indicate that the box C/D motif has, perhaps, a leading role in recruiting the fibrillarin –Nop5p complex to the target RNA in order to facilitate methylation at both sites.

Discussion A central issue in sRNP assembly and catalysis is the functional requirement for the internal (pseudo) symmetries present both in box C/D RNA and in the fibrillarin –Nop5p complex. Evidence is presented that sRNPs are capable of assembly into minimal complexes devoid of internal (pseudo) symmetries. However, the presence of the internal (pseudo) symmetries in box C/D RNA and in the fibrillarin – Nop5p is critical for efficient methyl transfer reactions. Figure 7 summarizes the catalytic properties of differently assembled sRNPs resulting from our study. It is

Figure 5. Self-association of Nop5 is not required for RNA-binding. Comparison of RNA-binding properties of wild-type and the L97K/ I112E double mutant fibrillarin – Nop5p complex was carried out against pre-formed L7Ae – sR3 (wild-type and mutant) complexes. Lanes 2 – 4 contain binding reactions with preformed L7Ae – sR3 RNA (sR3 WT) complexes. Lane 5 – 7 contain L7Ae– sR3 boxC(A/C) RNA complexes. The wild-type and mutant fibrillarin/Nop5p protein concentrations were identical (4 mM). RNP1 corresponds to the complex of L7Ae with RNA. RNP2 corresponds to the L7Ae– RNA – fibrillarin – Nop5p heterodimer complex. RNP3 corresponds to the L7Ae –RNA – fibrillarin – Nop5p symmetric dimer complex.

Assembly and Catalytic Function of Box C/D sRNPs

Figure 6. Symmetric assembly is required for efficient catalysis. Time-courses of in vitro methylation activities as measured by retained CPM on DE81 filters were plotted for reactions catalyzed by the wild-type fibrillarin – Nop5p and the double-mutant with the sR3 and its mutants. The wild-type þ high L7Ae and the wildtype þ low L7Ae reactions contained all wild-type components, plus 1.5 mM and 0.15 mM L7Ae, respectively. In the double mutant reaction, the L97K/I112E double mutant replaced the wild-type fibrillarin – Nop5p. In the sR3 boxC(A/C) and the sR3 boxC0 (A/C) reaction, the wild-type sR3 RNA was replaced by either of the two mutant RNAs and the target RNA oligonucleotide complementary to the guide sequences upstream of box D0 or D was used correspondingly.

clear that although the bipartite feature in box C/D RNA and in the fibrillarin– Nop5p complex is not required for assembly, lack of symmetry in either the guide RNA or the fibrillarin– Nop5p complex reduces methylation activities significantly. The functional requirement for the inherent

303

2-fold symmetry of fibrillarin –Nop5p in assembly and catalysis was scrutinized by working with a mutant of Nop5p deficient in self-association, L97K/I112E. Gel mobility-shift experiments showed clearly that the L97K/I112E double mutant was able to associate as a single heterodimer with the L7Ae –RNA complexes that contain one or two box C/D motifs, suggesting that self-association between two fibrillarin –Nop5p heterodimers is not required for their interaction with box C/D RNAs. However, our in vitro catalysis assay showed that the double mutant had markedly reduced activity in methylation when compared with the wild-type enzyme. The molecular mechanism underlying the requirement for the symmetric dimer in catalysis is not clear. One simple explanation is that a single heterodimer is unable to orient the box C/D RNA and the guidetarget duplex with respect to the active site due to unpacking of the two coiled-coil domains, which could profoundly change the dynamic and structural behavior of the RNA-binding domain. This explanation does not necessarily require that both fibrillarin– Nop5p heterodimers in the symmetric dimer bind to the two box C/D motifs simultaneously. Alternatively, a more complex explanation might be that concurrent binding of the symmetric dimer on both box C/D motifs facilitates conformational changes in both RNA and protein subunits required for catalysis to take place. Although there is no direct evidence to suggest that the symmetric dimer of fibrillarin – Nop5p binds simultaneously to both box C/D motifs, previous cross-linking results from the eukaryotic snoRNP homologs had placed Nop56p near the box C0 /D0 motif and Nop58p near the box C/D motif, supporting the latter explanation. Further structural work on intact sRNP and snoRNP particles is necessary in order to resolve these issues.

Figure 7. Observed sRNP assemblies and their catalytic competence. Observed sRNP assemblies are divided into three groups, the dimeric, monomeric, and asymmetric assemblies. The dimeric assemblies include those with single bound L7Ae and are depicted hypothetically with the two fibrillarin – Nop5p complexes interacting with the two box C/D motifs simultaneously. The monomeric assemblies were observed with the L97K/I112E double mutant. The asymmetric assemblies were observed with the sR3 RNA mutants.

304

L7Ae initiates the assembly of fibrillarin – Nop5p with box C/D RNA. The function of the L7Ae protein in box C/D sRNP assembly and catalysis was investigated. Our results have confirmed the previously known function of L7Ae in initiation of sRNP assembly and have established its concentration-dependent stoichiometry in binding to box C/D RNA. More importantly, we found that one L7Ae molecule bound at either of the two box C/D motifs is sufficient to initiate subsequent assembly of the wild-type fibrillarin –Nop5p dimer and the L97K/I112E double mutant. Furthermore, when large populations of box C/D RNA are bound with a single L7Ae molecule (low concentration of L7Ae), the methylation activity was found to be reduced only slightly from when L7Ae is doubly bound (high concentration of L7Ae), suggesting an asymmetric assembly of L7Ae in a functional sRNP. This result is consistent with the nucleotide analog interference mapping and in vitro binding studies on the human homolog of L7Ae, the 15.5 kDa protein, where convincing evidence has been presented that suggests a single bound molecule of 15.5 kDa at box C/D sequences.27 It remains to be elucidated, then, how the fibrillarin –Nop56p (Nop5p) complex interacts with box C0 /D0 sequences in the absence of the 15.5 kDa (L7Ae) protein. In summary, the symmetric nature of the fibrillarin –Nop5p complex and the box C/D RNAs has been demonstrated to be functionally important. Evidence presented here suggests that the minimal sRNP assembly that is competent in catalysis consists of one L7Ae molecule, one symmetric dimer of fibrillarin – Nop5p, and a bipartite box C/D RNA. Verification of this model requires determination of the entire complex in the presence of a target RNA.

Materials and Methods Expression and purification of AF proteins Expression and purification of wild-type and mutant AF fibrillarin – Nop5p proteins have been described.24 AF L7Ae was PCR amplified from AF genomic DNA and was cloned into the pET-13b vector. Harvested cells expressing L7Ae were sonicated in buffer A (25 mM sodium phosphate buffer (pH 7.4), 5% (v/v) glycerol, 1 M KCl) and were heated for 15 minutes at 70 8C. After centrifugation, the supernatant was treated with poly (ethyleneimine) and ammonium sulfate to remove nonspecific RNAs. The pellet containing L7Ae was resuspended in buffer B (20 mM Tris –HCl (pH 8.0), 5% (v/v) glycerol, 0.7 M KCl, 5 mM b-mercaptoethanol, 5 mM imidazole) and the solution cleared by centrifugation at 15,000 rpm (17,216 g) for 30 minutes. The supernatant was loaded onto a Ni-NTA column (Qiagen) equilibrated with buffer B followed by washing with washing buffer (buffer B containing 25 mM imidazole). The bound L7Ae proteins were eluted by an elution buffer (buffer B containing 270 mM imidazole) and were subjected to gel-filtration on a Superdex S200 column (Amersham Pharmacia) in buffer C (20 mM Tris – HCl (pH 8.0), 5%

Assembly and Catalytic Function of Box C/D sRNPs

(v/v) glycerol, 0.7 M KCl, 5 mM b-mercaptoethanol, 0.5 mM EDTA). Gel mobility-shift experiment Gel mobility-shift assays were performed essentially as described.24 pUC18 vector, carrying the sR3 gene flanked by the bacteriophage T7 promoter and HindIII sequences was linearized with HindIII enzyme. sR3 RNA was obtained by standard in vitro transcription reactions using phage T7 RNA polymerase in the presence of [32P]CTP (800 Ci/mmol; ICN Biomedicals) as described.24 The sR3 boxC(A/C), sR3 boxC0 (A/C), sR3 boxD(A/C), and sR3 boxD0 (A/C) mutants were generated in the same pUC18 vector by changing the first adenosine residue in each of the boxes to cytidine (Figure 1(A)). sR1 RNA was transcribed from a synthetic DNA template carrying the phage T7 promoter sequence. The RNP complexes between trace amount of 32 P-labeled RNA and sRNP proteins were resolved on a non-denaturing 7.5% (w/v) polyacrylamide gel and were visualized from an exposed phosphorimager screen by the ImageQuant software of the Storm system. Analytical ultracentrifugation experiments Sedimentation velocity and equilibrium experiments were conducted with a Beckman Optima XL-A analytical ultracentrifuge equipped with UV absorbance optics at the Keck Biophysics facility (Northwestern University). Protein samples were made in a buffer containing 25 mM sodium phosphate (pH 6.88) and 1 M NaCl, which served also as the reference buffer. Equilibrium and velocity data were analyzed with the UltraScan version 6.1 software†. Hydrodynamic corrections for buffer conditions were made according to data published by Laue et al.39 and as implemented in UltraScan. The partial specific volume of the peptides was estimated as described,40 and as implemented in UltraScan. Sedimentation equilibrium experiments were performed at 30 8C and rotor speeds ranging between 11,000 rpm and 27,000 rpm. Samples were spun in six-channel Epon/ charcoal centerpieces in the AN-50-TI rotor. Scans were collected at equilibrium at either 230 nm or 280 nm in radial step mode with 0.001 cm step-size setting and 20-point averages. Multiple loading concentrations ranging between 0.3 and 0.7 absorbance unit were measured at the given wavelength, data exceeding 0.9 absorbance unit were excluded from the fit. For both the wild-type and the double mutant, data in the concentration range between 0 mM and 16 mM were examined. Since multiple wavelengths were used in the global fit, extinction coefficients for each wavelength needed to be measured. To this end, wavelength scans were performed between 220 nm and 340 nm for each concentration and fit to a global extinction profile using UltraScan. The extinction profile was normalized with the extinction coefficient at 280 nm by estimation from the protein sequence, as described41 and as implemented in UltraScan. Sedimentation velocity experiments were performed at 30 8C and 45,000 rpm. Samples were spun in two-channel Epon/charcoal centerpieces in the AN-50-TI rotor. Scans were collected continuously at both 230 nm and 280 nm in radial continuous mode with 0.001 cm step-size setting and no averaging until all material was pelleted. † http://www.ultrascan.uthscsa.edu

305

Assembly and Catalytic Function of Box C/D sRNPs

Loading concentrations at either wavelength ranged between 0.7 and 0.9 absorbance unit. Activity assay We employed an RNA methylation assay modified from a procedure used in enzyme kinetic study of the phage T4 Dam DNA-(N6-adenine) methyltransferase.42 In a typical reaction, fibrillarin – Nop5p (or the L97K/ I112E mutant) and L7Ae were mixed to reach final concentrations of 0.46 mM and 1.88 mM, respectively, in a buffer containing 20 mM Tris – HCl (pH 8.0), 5% (v/v) glycerol before adding a mixture of pre-annealed guide and target RNAs. The target RNA oligo complementary to the sequence upstream of box D was used except for the sR30 boxC(A/C) mutant, where the target RNA oligomer complementary to the sequence upstream of box D0 was used. The total reaction mixture was then divided into 40 ml samples and placed into a 70 8C waterbath. To initiate the methylation reaction, 0.6 ml of [3H]AdoMet (55 Ci/mmol; ICN Biomedicals) was added to each sample, and the reactions were terminated at 10, 20 and 40 minutes by chilling on ice. Each reaction sample was then spotted directly onto a DE81 anion-exchange 2.3 cm filter disc (Whatman). Each filter was washed extensively with 3 £ 30 ml of 20 mM NH4HCO3 solution containing 30 mM unlabeled AdoMet (Sigma), 2 £ 30 ml of water and finally with 30 ml of 75% (v/v) ethanol. The dried filters were counted in a scintillation cocktail. Each progress curve was an average from at least four independently duplicate experiments. In order to ensure the integrity of RNA molecules throughout the assay, mock reactions in the absence of labeled AdoMet were carried out and the presence of RNA molecules was assessed by loading the reaction mixture at the end of 40 minutes onto a denaturing polyacrylamide gel containing urea.

7.

8. 9.

10.

11. 12.

13. 14. 15.

16. 17.

Acknowledgements This work was supported by Florida Biomedical Research grant BM002 (H.L.), NIH grant R01 GM66958-01 (H.L.) and NSF grant DBI-9974819 (B.D.). We thank members of the Li group for helpful discussions and S. Hattman for an in vitro DNA methylation assay protocol.

18.

19.

20.

References 1. Maxwell, E. S. & Fournier, M. J. (1995). The small nucleolar RNAs. Annu. Rev. Biochem. 64, 897– 934. 2. Lafontaine, D. L. & Tollervey, D. (1998). Birth of the snoRNPs: the evolution of the modification-guide snoRNAs. Trends Biochem. Sci. 23, 383– 388. 3. Weinstein, L. B. & Steitz, J. A. (1999). Guided tours: from precursor snoRNA to functional snoRNP. Curr. Opin. Cell Biol. 11, 378– 384. 4. Kiss, T. (2001). Small nucleolar RNA-guided posttranscriptional modification of cellular RNAs. EMBO J. 20, 3617– 3622. 5. Tycowski, K. T., You, Z. H., Graham, P. J. & Steitz, J. A. (1998). Modification of U6 spliceosomal RNA is guided by other small RNAs. Mol. Cell, 2, 629– 638. 6. Jady, B. E. & Kiss, T. (2001). A small nucleolar guide

21. 22. 23.

24.

RNA functions both in 20 -O-ribose methylation and pseudouridylation of the U5 spliceosomal RNA. EMBO J. 20, 541–551. Cavaille, J., Buiting, K., Kiefmann, M., Lalande, M., Brannan, C. I., Horsthemke, B. et al. (2000). Identification of brain-specific and imprinted small nucleolar RNA genes exhibiting an unusual genomic organization. Proc. Natl Acad. Sci. USA, 97, 14311– 14316. Terns, M. P. & Terns, R. M. (2002). Small nucleolar RNAs: versatile trans-acting molecules of ancient evolutionary origin. Gene Expr. 10, 17 – 39. Zhou, H., Chen, Y. Q., Du, Y. P. & Qu, L. H. (2002). The Schizosaccharomyces pombe mgU6-47 gene is required for 20 -O-methylation of U6 snRNA at A41. Nucl. Acids Res. 30, 894– 902. Speckmann, W. A., Li, Z. H., Lowe, T. M., Eddy, S. R., Terns, R. M. & Terns, M. P. (2002). Archaeal guide RNAs function in rRNA modification in the eukaryotic nucleus. Curr. Biol. 12, 199– 203. Dennis, P. P., Omer, A. & Lowe, T. (2001). A guided tour: small RNA function in Archaea. Mol. Microbiol. 40, 509– 519. Gaspin, C., Cavaille, J., Erauso, G. & Bachellerie, J. P. (2000). Archaeal homologs of eukaryotic methylation guide small nucleolar RNAs: lessons from the Pyrococcus genomes. J. Mol. Biol. 297, 895– 906. Omer, A. D., Lowe, T. M., Russell, A. G., Ebhardt, H., Eddy, S. R. & Dennis, P. P. (2000). Homologs of small nucleolar RNAs in Archaea. Science, 288, 517– 522. Bachellerie, J. P., Cavaille, J. & Huttenhofer, A. (2002). The expanding snoRNA world. Biochimie, 84, 775 –790. Peculis, B. A. & Steitz, J. A. (1994). Sequence and structural elements critical for U8 snRNP function in Xenopus oocytes are evolutionarily conserved. Genes Dev. 8, 2241– 2255. Terns, M. P., Grimm, C., Lund, E. & Dahlberg, J. E. (1995). A common maturation pathway for small nucleolar RNAs. EMBO J. 14, 4860– 4871. Omer, A. D., Ziesche, S., Decatur, W. A., Fournier, M. J. & Dennis, P. P. (2003). RNA-modifying machines in archaea. Mol. Microbiol. 48, 617– 629. Kiss-Laszlo, Z., Henry, Y. & Kiss, T. (1998). Sequence and structural elements of methylation guide snoRNAs essential for site-specific ribose methylation of pre-rRNA. EMBO J. 17, 797– 807. Wang, H., Boisvert, D., Kim, K. K., Kim, R. & Kim, S. H. (2000). Crystal structure of a fibrillarin homologue from Methanococcus jannaschii, a hyperthermo˚ resolution. EMBO J. 19, 317– 323. phile, at 1.6 A Niewmierzycka, A. & Clarke, S. (1999). S-Adenosylmethionine-dependent methylation in Saccharomyces cerevisiae. Identification of a novel protein arginine methyltransferase. J. Biol. Chem. 274, 814– 824. Lafontaine, D. L. & Tollervey, D. (1999). Nop58p is a common component of the box C þ D snoRNPs that is required for snoRNA stability. RNA, 5, 455– 467. Lafontaine, D. L. & Tollervey, D. (2000). Synthesis and assembly of the box C þ D small nucleolar RNPs. Mol. Cell. Biol. 20, 2650– 2659. Gautier, T., Berges, T., Tollervey, D. & Hurt, E. (1997). Nucleolar KKE/D repeat proteins Nop56p and Nop58p interact with Nop1p and are required for ribosome biogenesis. Mol. Cell. Biol. 17, 7088– 7098. Aittaleb, M., Rashid, R., Chen, Q., Palmer, J. R., Daniels, C. J. & Li, H. (2003). Structure and function of archaeal box C/D sRNP core proteins. Nature Struct. Biol. 10, 256– 263.

306

Assembly and Catalytic Function of Box C/D sRNPs

25. Cahill, N. M., Friend, K., Speckmann, W., Li, Z. H., Terns, R. M., Terns, M. P. & Steitz, J. A. (2002). Sitespecific cross-linking analyses reveal an asymmetric protein distribution for a box C/D snoRNP. EMBO J. 21, 3816– 3828. 26. Venema, J. & Tollervey, D. (1999). Ribosome synthesis in Saccharomyces cerevisiae. Annu. Rev. Genet. 33, 261– 311. 27. Szewczak, L. B., DeGregorio, S. J., Strobel, S. A. & Steitz, J. A. (2002). Exclusive interaction of the 15.5 kD protein with the terminal box C/D motif of a methylation guide snoRNP. Chem. Biol. 9, 1095–1107. 28. Kuhn, J. F., Tran, E. J. & Maxwell, E. S. (2002). Archaeal ribosomal protein L7 is a functional homolog of the eukaryotic 15.5 kD/Snu13p snoRNP core protein. Nucl. Acids Res. 30, 931– 941. 29. Omer, A. D., Ziesche, S., Ebhardt, H. & Dennis, P. P. (2002). In vitro reconstitution and activity of a C/D box methylation guide ribonucleoprotein complex. Proc. Natl Acad. Sci. USA, 99, 5289– 5294. 30. Watkins, N. J., Dickmanns, A. & Luhrmann, R. (2002). Conserved stem II of the box C/D motif is essential for nucleolar localization and is required, along with the 15.5 K protein, for the hierarchical assembly of the box C/D snoRNP. Mol. Cell. Biol. 22, 8342– 8352. 31. Watkins, N. J., Segault, V., Charpentier, B., Nottrott, S., Fabrizio, P., Bachi, A. et al. (2000). A common core RNP structure shared between the small nucleoar box C/D RNPs and the spliceosomal U4 snRNP. Cell, 103, 457– 466. 32. Vidovic, I., Nottrott, S., Hartmuth, K., Luhrmann, R. & Ficner, R. (2000). Crystal structure of the spliceosomal 15.5 kD protein bound to a U4 snRNA fragment. Mol. Cell, 6, 1331– 1342. 33. Klein, D. J., Schmeing, T. M., Moore, P. B. & Steitz, T. A. (2001). The kink-turn: a new RNA secondary structure motif. EMBO J. 20, 4214– 4221. 34. Van Holde, K. E. & Weischet, W. O. (1978). Boundary analysis of sedimentation-velocity experiments with

monodisperse and paucidisperse solutes. Biopolymers, 17, 1387– 1403. Demeler, B., Saber, H. & Hansen, J. C. (1997). Identification and interpretation of complexity in sedimentation velocity boundaries. Biophys. J. 72, 397– 407. Schuck, P., Perugini, M. A., Gonzales, N. R., Howlett, G. J. & Schubert, D. (2002). Size-distribution analysis of proteins by analytical ultracentrifugation: strategies and application to model systems. Biophys. J. 82, 1096– 1111. Garcia De La Torre, J., Huertas, M. L. & Carrasco, B. (2000). Calculation of hydrodynamic properties of globular proteins from their atomic-level structure. Biophys. J. 78, 719– 730. Evdokimov, A. A., Zinoviev, V. V., Malygin, E. G., Schlagman, S. L. & Hattman, S. (2002). Bacteriophage T4 Dam DNA-[N6-adenine]methyltransferase. Kinetic evidence for a catalytically essential conformational change in the ternary complex. J. Biol. Chem. 277, 279– 286. Laue, T. M., Shah, B. D., Ridgeway, T. M. & Pelletier, S. L. (1992). Computer-aided interpretation of analytical sedimentation data for proteins. In Analytical Ultracentrifugation in Biochemistry and Polymer Science (Harding, S. E., Rowe, A. J. & Honton, J. C., eds), Royal Society of Chemistry, UK. Cohn, E. J. & Edstall, J. T. (1943). Proteins, Amino Acids and Peptides as Ions and Dipolar Ions, chapt. 4, p. 157, Reihnold, New York. Gill, S. C. & von Hippel, P. H. (1989). Calculation of protein extinction coefficients from amino acid sequence data. Anal. Biochem. 182, 319–326. Malygin, E. G., Zinoviev, V. V., Petrov, N. A., Evdokimov, A. A., Jen-Jacobson, L., Kossykh, V. G. & Hattman, S. (1999). Effect of base analog substitutions in the specific GATC site on binding and methylation of oligonucleotide duplexes by the bacteriophage T4 Dam DNA-[N6-adenine] methyltransferase. Nucl. Acids Res. 27, 1135– 1144.

35. 36.

37.

38.

39.

40. 41. 42.

Edited by J. Doudna (Received 22 May 2003; received in revised form 15 July 2003; accepted 7 August 2003)