Structural and thermodynamic evidence for a stabilizing role of Nop5p in S-adenosyl-L-methionine binding to fibrillarin.

In Archaea, fibrillarin and Nop5p form the core complex of box C/D small ribonucleoprotein particles, which are responsible for site-specific 2'-hydroxyl methylation of ribosomal and transfer RNAs. Fibrillarin has a conserved methyltransferase fold and employs S-adenosyl-l-methionine (AdoMet) as the cofactor in methyl transfer reactions. Comparison between recently determined crystal structures of free fibrillarin and fibrillarin-Nop5p-AdoMet tertiary complex revealed large conformational differences at the cofactor-binding site in fibrillarin. To identify the structural elements responsible for these large conformational differences, we refined a crystal structure of Archaeoglobus fulgidus fibrillarin-Nop5p binary complex at 3.5 A. This structure exhibited a pre-formed backbone geometry at the cofactor binding site similar to that when the cofactor is bound, suggesting that binding of Nop5p alone to fibrillarin is sufficient to stabilize the AdoMet-binding pocket. Calorimetry studies of cofactor binding to fibrillarin alone and to fibrillarin-Nop5p binary complex provided further support for this role of Nop5p. Mutagenesis and thermodynamic data showed that a cation-pi bridge formed between Tyr-89 of fibrillarin and Arg-169 of Nop5p, although dispensable for in vitro methylation activity, could partially account for the enhanced binding of cofactor to fibrillarin by Nop5p. Finally, assessment of cofactor-binding thermodynamics and catalytic activities of enzyme mutants identified three additional fibrillarin residues (Thr-70, Glu-88, and Asp-133) to be important for cofactor binding and for catalysis.

2Ј-O-Methylation is one of the most frequent modifications on specific nucleotides within rRNA and other classes of RNA. In vertebrates, more than one hundred 2Ј-O-methyl groups have been identified in rRNA, most of which occur at highly conserved locations within functionally important regions (1). The methyltransferase (MTase) 1 that is responsible for 2Ј-Omethylation of the majority of rRNA nucleotides is the box C/D small ribonucleoprotein particle (snoRNP) which in addition to methylating rRNA also methylates mRNA (2,3) and small nuclear RNA (4 -7) and processes precursor rRNA into mature rRNAs.
Box C/D snoRNPs require assembly of box C/D snoRNAs with a set of nucleolar proteins that include fibrillarin (Nop1p in yeast), Nop56/58p, and 15.5-kDa proteins (Snu13p in yeast). In Archaea, homologs of box C/D snoRNPs (box C/D sRNPs) consist of box C/D sRNAs, fibrillarin, Nop5p (a single homolog of Nop56/58p), and L7Ae proteins (3,8). The box C/D RNAs are responsible for the recognition of methylation target RNAs via base pairing of their antisense regions with the target. The s(no)RNP proteins contain the actual methyltransferase activity. The protein subunit that catalyzes the methyl transfer reactions in s(no)RNPs is believed to be fibrillarin. The first clue that linked the methylation activity of s(no)RNPs to fibrillarin came from the work of Tollervey et al. (9) on yeast, where the nop1.3 allele, a temperature-sensitive mutant of the nop1 gene coding for yeast fibrillarin, exhibited a strongly inhibited precursor rRNA methylation phenotype at non-permissive temperatures. The second evidence came from the crystal structure of an archaeal fibrillarin homolog from Methanococcus jannaschii (MJ), which revealed the conservation of the AdoMet-binding fold that is common to all known MTase structures (10). We recently determined a co-crystal structure of fibrillarin complexed with Nop5p from Archaeoglobus fulgidus (AF), which contained a bound AdoMet at the predicted binding site in fibrillarin (11). This result establishes further the role of fibrillarin as the methyltransferase.
Similar to the previously known AdoMet-dependent MTases, fibrillarin contains the same set of conserved motifs that are important to the binding of AdoMet (see Fig. 1). These include motifs I-IV at the carboxyl ends of beta strands ␤1-␤4 as defined by Cheng and Roberts (12) in studying DNA MTases. Consistent with their important functional roles in catalysis, three temperature-sensitive mutants of yeast fibrillarin occurred within the conserved AdoMet-binding motifs. For instance, nop1.2, nop1.3, and nop1.7 mutations occurred in motif III, motif I, and motifs II and IV, respectively (9). Furthermore, by using an in vitro reconstituted archaeal sRNP enzyme, Omer et al. (13) demonstrated that substitutions of two amino acids within motifs I and III of Sulfolobus solfataricus fibrillarin (A85V and P129V) resulted in completely or partially abolished methylation activities, whereas sRNP assembly was unaffected. These data suggest that fibrillarin has a function similar to other previously known AdoMet-dependent MTases.
However, recent structural and biochemical evidence suggests that fibrillarin alone binds weakly to AdoMet. There are currently three known fibrillarin crystal structures, all from Archaea, MJ (10), AF (11), and Puroccocus furiosus (PF) (14). Among these three structures, only that in complex with AF Nop5p had the cofactor bound at the predicted AdoMet-binding site. Interestingly, in both structures of fibrillarin without the bound Nop5p (MJ and PF), motif I and the loop connecting ␤1 and ␣1 of fibrillarin adopted conformations that would exclude the binding of AdoMet (10,14). This clearly raised the possibility that Nop5p has a functional role in cofactor binding by modulating the conformation of AdoMet-binding motifs in fibrillarin. However it is also possible that the binding of the cofactor itself induces the conformational change of the cofactor binding pocket in fibrillarin.
In the co-crystal structure of AF fibrillarin-Nop5p complex bound with AdoMet (holocomplex), a number of conserved fibrillarin residues were observed to interact directly with AdoMet ( Fig. 1). Glu-88 forms two hydrogen bonds with the ribose hydroxyl groups of AdoMet. Thr-70 also forms a hydrogen bond with the carboxyl group of AdoMet. Asp-133 is situated near the positively charged thiomethyl group and thus may facilitate cofactor binding through favorable electrostatic interactions. Finally, Tyr-89 establishes an aromatic stacking interaction with the adenine ring of the cofactor. The opposing side of the phenol ring of Tyr-89 closely packs against a positive Nop5p residue, Arg-169. This structural arrangement between Tyr-89 of fibrillarin and Arg-169 of Nop5p creates a strong cation-interaction as evident from the large negative electrostatic and van der Waals energies between them (E es ϭ Ϫ5.2 kcal/mol, E vdw ϭ Ϫ2.8 kcal/mol) (computed by CAPTURE) (15). Notably, Tyr-89 would be completely exposed to solvent without forming the cation-interaction. Tyr-89 is well conserved among all known fibrillarin proteins. Arg-169 is also conserved among the Nop family of proteins, suggesting that this cationinteraction is preserved in all homologous complexes of fibrillarin and Nop5p. The importance of this cation-bridge formed between fibrillarin and Nop5p may play a critical role in stabilizing the association of the cofactor with fibrillarin.
In this work, we examined the requirement of Nop5p for cofactor binding by combining structural, thermodynamic, and functional analyses. To discern the effect on fibrillarin conformational changes in the cofactor binding pocket, we refined the crystal structure of Nop5p-fibrillarin complex without soaked cofactors at 3.5 Å. We also established isothermal titration calorimetry conditions, which allowed us to directly compare thermodynamic parameters of AdoMet binding to different fibrillarin complexes. By monitoring the changes in heat as the cofactor was titrated into a protein solution, we obtained binding constants and enthalpy changes of AdoMet binding to the Nop5p-fibrillarin complex, to mutant fibrillarin Nop5p complexes, and to fibrillarin. Furthermore, to assess the functional importance of the cofactor interacting residues in both fibrillarin and Nop5p, we compared in vitro methylation activities of the wild-type Nop5p-fibrillarin complex and its mutants where several AdoMet-binding residues were disrupted. These structural, thermodynamic, and functional studies support the stabilizing role of Nop5p in cofactor binding to fibrillarin and clearly identify residues in fibrillarin that are directly involved in cofactor binding and catalysis.

EXPERIMENTAL PROCEDURES
Protein Purification-The wild-type Nop5p-fibrillarin protein complex and fibrillarin alone were purified as described previously by Aittaleb et al. (11). All the mutations were performed within co-expressing fibrillarin and Nop5p genes using the QuikChange mutagenesis kit from Stratagene. The mutants were purified in a similar procedure as that for the wild type. All proteins were subjected to gel filtration on a Superdex S200 column (Amersham Biosciences) in a buffer containing 20 mM Tris, pH 8.0, 5% glycerol, 1.0 M NaCl, 5 mM ␤-mercaptoethanol, and 0.5 mM EDTA. Gel filtration profiles for all mutant fibrillairin-Nop5p complexes were similar to that of the wild-type complex, suggesting no misfolding of the mutant proteins.
Structure Refinement-During earlier structure determination of the fibrillarin-Nop5p complex, we collected a multiple wavelength anomalous diffraction data set from a crystal of fibrillarin-Nop5p complex containing seleno-methionine without soaked cofactors. This data set allowed phase determination. The final structure, however, was refined against the diffraction data set collected from cofactor-soaked fibrillarin-Nop5p crystal because of its higher resolution. To elucidate the structural conformation around the cofactor binding site in the absence of a bound cofactor, we have now refined the coordinates of fibrillarin-Nop5p complex against the data set collected at the selenine K-edge. This data set has the best statistics among the three data sets (11). Refinement was carried out using crystallography NMR software (CNS) (16) by including all 10 selenine atoms and keeping the Bijvoet pairs separated. The statistics of structural refinement are listed in Table I. Structure comparison of the apocomplex with earlier structures is presented in Fig. 2.
Isothermal Titration Calorimetry-Titration experiments were performed by isothermal titration calorimetry (ITC) using a VP-ITC microcalorimeter (Microcal, Inc., Northampton, MA) interfaced with a computer. The titration calorimeter consists of 1.45 ml of sample cell containing a macromolecule solution and a matched thermal reference cell filled with water. AdoMet or S-adenosyl-L-homocystein (AdoHcy) was dissolved in the same buffer used for the protein to be titrated. Prior to the experiment, samples were filtered and degassed under vacuum for 10 min in a Thermo Vac system (Microcal). The sample cell was filled with the working buffer (for dilution heat control) or with the protein to be characterized. Titrations with the cofactor (25ϫ protein concentration) consisted of a preliminary 2-l injection (not to be considered in data analysis) followed by twenty-nine 5 l injections with at least 4-min intervals between injections. All runs were made at constant stirrer speed of 310 rpm, and all experiments were performed at 30°C. The heat caused by the dilution of the cofactor was subtracted from the experimental data conducted with the protein. Protein concentrations were determined by UV absorption using the theoretical extinction coefficients computed from the amino acid sequences (⑀ 280 ϭ 48360 M Ϫ1 cm Ϫ1 ). The evolved heat peaks were integrated and then fitted to a theoretical titration curve of a single binding site model by non-linear least squares to yield ⌬H 0 (molar enthalpy change in kcal/mol), K a (binding constant in molar Ϫ1 ) and n (stoichiometry ratio).
All isotherms show negative deflection indicative of an exothermic reaction (see Fig. 3, A and B). A separate run with cofactor titrating into buffer showed insignificant endothermic and equal sized dilution peaks, suggesting that at this concentration no aggregation of the cofactor in buffer occurs (Fig. 3B). Despite the fact that Nop5p-fibrillarin complex forms a homodimer as previously observed in crystal and in solution (11,18), ITC data clearly exhibited the characteristics of single set of identical sites. Therefore, all data were processed using the standard One Set of One Site model as implemented in the Microcal Origin software based on the Wiseman isotherm (17), where Q is the heat content of the solution, K a is the binding constant, and M t is the bulk concentration of the protein in volume V 0 , X t is the bulk concentration of the ligand, and n is the number of binding sites. During actual fitting, the heat released for each injection increment was used with the correction of displaced volume caused by each injection. Therefore, the enthalpy change (⌬H 0 ) and the binding constant (K a ) were directly obtainable from the experiments after data processing. The free energy and entropy were subsequently calculated using ⌬G 0 ϭ ϪRTlnK a and T⌬S 0 ϭ Ϫ⌬G 0 ϩ⌬H 0 , respectively. The resulting binding parameters (binding constant K a , molar Gibbs free energy, ⌬G 0 , molar enthalpy ⌬H 0 , and molar entropy ⌬S 0 ) of the proteins with AdoMet and with AdoHcy (D133A mutant only), and the standard deviations are summarized in Table II.
Electrostatic Potential Calculations-To assess the electrostatic contribution of Asp133 to cofactors binding, electrostatic potentials were calculated with the hybrid boundary element and finite difference nonlinear Poisson-Boltzmann (PB) algorithm (19). The atomic coordinates of the wild-type Nop5p-fibrillarin complex (Protein Data Bank accession number 1NT2) were employed, and no missing residues or hydrogen atoms were added. For the D133A mutant Nop5p-fibrillarin complex, the Asp-133 residue was substituted for alanine followed by stereochemistry regularization using the O program (20). The dielectric constant of the solute was set to 2 and the solvent to 80. The temperature and ionic strength of the solution was fixed at 298 Kelvin and 0.1 M NaCl, respectively. The Parse parameter set (21) was used to assign van der Waals radii to atoms. The formal charge set was employed with a charge of Ϫ1e assigned to aspartate, glutamate, and the C-terminal residues. A charge of ϩ1e was assigned to lysine and arginine residues. Histidine residues were assigned a charge of ϩ0.5e. The net charge of the wild-type and mutant Nop5p-fibrillarin complexes is ϩ6.5e and ϩ7.5e, respectively. The solvent-excluded molecular surface was employed to define the solute-solvent boundary based on a 1.4-Å solvent probe radius. No ion exclusion region was considered.
The MSMS molecular surface program (22) was employed to triangulate the solvent-excluded surface and thus generate the boundary elements (ϳ90,000 triangles) used in the linear PB solution. The finite difference cubic grid contained 128 3 nodes. The total extent of the three-dimensional uniform Cartesian grid is four times the largest dimension of the solute. Computed electrostatic potential surfaces for both the wild-type and the D133A mutant are displayed in Fig. 4.
In Vitro Methylation Assay-In vitro methylation assays were carried out according to a published procedure described previously (18). Briefly, all enzymes were prepared to be free of ribonuclease contamination before being assembled for catalysis. Tritium-AdoMet (Sigma) was used as the methyl donor in the reaction, which permitted us to monitor the methylation reaction by counting the retained radioactivity caused by the target RNA on DE81 ion exchange filter. Wild-type Nop5p-fibrillarin or a mutant was mixed with L7Ae protein prior to being added to a pre-annealed RNA mixture containing both the AF sR3 box C/D guide RNA (rna.wustl.edu/snoRNAdb/) and a complementary target RNA oligo to the guide sequence upstream of box D sequences. The radioactivity retained on the DE81 filter at each reaction time point was quantified by scintillation counter. The production of methylation product was plotted in Fig. 5 along with least-square fitted progress curves.

Structural Comparison of AdoMet-binding Sites in Nop5pfree and Nop5p-bound Fibrillarins-
The fibrillarin portion of the AF Nop5p-fibrillarin complex structure (11) was superimposed with those of MJ fibrillarin (10) and of PF fibrillarin (14). The backbone root mean square differences were 1.095 Å (191 C ␣ atoms) for MJ fibrillarin and 0.934 Å (144 C ␣ atoms) for PH fibrillarin. The most significant difference within the core structure of the three fibrillarins is in motif I and II where the cofactor binds (Fig. 2, B and C). In particular, the backbone geometry of motif II loop in the two free fibrillarin structures differ from each other and from the fibrillarin bound with Nop5p and AdoMet. Each free fibrillarin adopted a conformation that would exclude the binding of the cofactor. In both structures, the loops connecting ␤2 and ␣2 traversed through where the ribose moiety of AdoMet would lie. Most dramatically, the aromatic residues (Phe-106 in PF fibrillarin and Tyr-106 in MJ fibrillarin) completely swung away from where they could establish the favorable base stacking interactions with the cofactor adenine ring and protruded into the solvent region (Fig. 2, B and C). This striking difference in backbone geometry and key binding residues between the two free fibrillarin structures and that bound with cofactor suggested an intrinsic structural flexibility of this region in fibrillarin that could potentially hinder the optimal binding of the cofactor.
The Apocomplex Structure of Nop5p-Fibrillarin Closely Resembles the Holocomplex-To discern whether the observed conformational differences between free fibrillarin structures and that bound with cofactor and Nop5p are induced by Nop5p binding or by cofactor binding, we refined the crystal structure of Nop5p-fibrillarin in the absence of soaked AdoMet (apocomplex). The refined apocomplex of Nop5p-fibrillarin was superimposed with the holocomplex reported previously (11). Sig-maA-weighted 3F o Ϫ 2F c and F o Ϫ F c maps were computed by using the observed amplitude and phases resulted from refined coordinates. Both maps clearly showed an absence of a bound cofactor at the predicted AdoMet binding site ( Fig. 2A). Fig. 2A also showed that the fibrillarin residues covering the AdoMet binding site overlap well with those in holocomplex (209 C␣ atoms, root mean square difference, 0.209 Å). In particular, Tyr-89 in the apocomplex did not deviate from its conformation in the holocomplex where it established a favorable stacking interaction with the cofactor adenine ring. This is in contrast to the large conformation differences of the Tyr-89-equivalent residues in the two free fibrillarin structures. We interpret this result to indicate that binding of Nop5p alone is sufficient to stabilize the structural conformation of fibrillarin for cofactor binding.
Thermodynamic Difference between Cofactor Binding to Nop5p-Fibrillarin Complex and to Fibrillarin Alone-We used isothermal titration calorimetry to carry out a comparative and quantitative study of cofactor binding to fibrillarin alone and to Nop5p-fibrillarin complex (Fig. 3). It can be seen in Table II that the overall Gibbs free energy of AdoMet binding to free fibrillarin was 10.5 M and to the Nop5p-fibrillarin complex was 2.7 M. Thus AdoMet binds ϳ4-fold more strongly to Nop5p-fibrillarin than to fibrillarin alone because of more favorable contribution from binding enthalpy but less favorable contribution from entropy. Both processes were enthalpy driven in agreement with the observed polar nature of the AdoMet-binding site. Thus if assuming a similar process of solvent re-arrangement in cofactor binding to either fibrillarin or to Nop5p-fibrillarin complex, the increased binding entropy suggested a less ordered final structure when cofactor bound to fibrillarin alone. This interpretation is again consistent with the structural observation that the Tyr-89-equivalent aromatic residues were unable to establish base stacking interaction in the two free fibrillarin structures (Fig. 2).
To further establish the role of the cation-formed between Tyr-89 of fibrillarin and Arg-169 of Nop5p in cofactor binding, we titrated the fibrillarin-Nop5p complexes containing either the Y89A or R169A mutation with AdoMet. The Y89A mutant showed clearly impaired binding affinity for AdoMet in terms of its dissociation constant K d (Table II). There was a significant increase in binding enthalpy (Ϫ4.7 kcal/mol versus Ϫ11.7 kcal/ mol in wild type) and a large gain in the binding entropy T⌬S 0 term (1.9 kcal/mol versus Ϫ4.0 kcal/mol in wild type) when compared with AdoMet binding to the wild-type complex. These thermodynamic results are consistent with a strong role of the phenol ring of Tyr-89 in restricting the bound conformation of AdoMet by maintaining favorable base-stacking interactions with the adenine moiety of the cofactor.
Titration of AdoMet to the R169A mutant revealed a thermodynamic energy compensation, which resulted in no change in the Gibbs free energy, ⌬G 0 . A significant increase in enthalpy (ϳ3 kcal/mol) was balanced by a gain in entropy (ϳ3 kcal/mol). This result may be interpreted as indicating that the decrease in binding potential energy because of the removal of Arg-169 (thus the cation-interaction) was compensated for by an increase in conformational flexibility of the final cofactorenzyme complex, assuming the single amino acid mutation has a negligible effect on protein dynamics and on solvent reorganization.
To further demonstrate that AdoMet binds to fibrillarin specifically to Motif I, II, and IV residues, we investigated thermodynamics of AdoMet binding to three Nop5p-fibrillarin mutant complexes. Each of the complexes contains a single mutation at Thr-70, Glu-88, and Asp-133 in fibrillarin, respectively. These three residues were implicated as the cofactor binding residues by the recent crystal structure of the holocomplex (11).
Thr-70 is located at the end of ␤1 strand (motif I). It forms a hydrogen bond with the carboxyamide group of the methionyl group. Glu-88 is at the end of ␤2 strand (motif II), and it forms hydrogen bonds with the hydroxyl groups of AdoMet (Fig. 1). These specific interactions are expected to stabilize cofactor binding. Surprisingly, mutations of either Thr-70 or Glu-88 to alanine (T70A or E88A) did not weaken the binding of AdoMet for the enzyme (Table II). T70A had slightly better affinity than that of the wild-type complex. The thermodynamic parameters exhibited enthalpy-entropy compensation similar to that observed for R169A mutant. The increase of binding enthalpy (ϳ2 kcal/mol) was compensated by an increase in entropic term T⌬S 0 (ϳ2.3 kcal/mol) when compared with the wild-type binding. Given the observed hydrogen bond between Thr-70 and the carboxyamide group of AdoMet, these thermodynamic results of binding are suggestive of conformational flexibilities of cofactor binding to T70A. The thermodynamics of AdoMet binding to E88A exhibited an opposing enthalpy-entropy compensation effect than that observed for T70A. A slight decrease in binding entropy (ϳ1 kcal/mol) was compensated for by a decrease in binding enthalpy (ϳ1.5 kcal/mol), suggesting a complicated effect of Glu-88 mutation on AdoMet binding. Under- standing the physical-chemical nature of these mutations requires combined results from high resolution structures of the mutants and the dynamic studies on proteins and the ligand.
In contrast, mutation of Asp-133 to alanine had a profound effect on cofactor binding affinity. The binding dissociation constant (90 M) was reduced nearly 40-fold from that of the wild-type Nop5p-fibrillarin complex (2.7 M) primarily from a large increase in enthalpy of binding (Table II). Asp-133 in fibrillarin and the spatially equivalent residues in fibrillarin homologs were implicated in the catalytic step involving deprotonation of the 2Ј-OH group of the substrate RNA (23,24).
Asp-133 is strictly conserved in fibrillarin and is in proximity of the thiomethyl group of the AdoMet. It appeared that the negatively charged carboxyl group of Asp-133 could be the major stabilizing factor for the positively charged sulfonium ion on AdoMet, as the binding affinity of D133A to the neutral cofactor AdoHcy was improved substantially (14.9 M) (Table  II).
To assess electrostatic contributions to the binding of AdoMet to the Nop5p-fibrillarin complex, we computed electrostatic potentials at the molecular surface of the Nop5p-fibrillarin complex from numerical solutions of the nonlinear Poisson-Boltzmann equation. Fig. 4 displays color-coded electrostatic potential surfaces of Nop5p-fibrillarin in the immediate vicinity of the bound AdoMet. The computed surface potentials clearly showed a charge complementarily at the cofactor binding site in that the carboxyl group of the AdoMet molecule is located within a positively charged pocket, whereas the adenine moiety is within a negatively charged pocket. The computed electrostatic potentials also predicted that the positively charged sulfonium sulfur interacted with the negatively charged Asp-133, as the same negative potential near the sulfonium sulfur in the wild-type protein was absent in the D133A mutant (Fig. 4). This interpretation is in agreement with the reduced binding affinity of AdoMet for D133A (Table II).
Methylation Activities of the Wild-type and the Mutant Nop5p-Fibrillarin Complexes-To fully understand the functional roles of each of the conserved AdoMet-interacting residues, we further carried out in vitro methylation assays using a reconstituted sRNP enzyme developed previously (18) and tested the ability of the mutant enzymes in catalyzing 2Ј-Omethylation on a target RNA oligomer that was complementary to the guide sequence upstream of box D. The progress curves for methylation reaction catalyzed by the wild-type Nop5p-fibrillarin complex and the mutant are plotted in Fig. 5. Each curve shows an average of at least four duplicated reactions. Fig. 5 clearly shows that whereas the Y89A and R169A mutants maintained some moderate activities, T70A, E88A, and D133A exhibited non-detectable methylation activities in our assays. These results further support an important functional role played by Asp-133, Thr-70, and Glu-88 in catalytic reactions. The abolished activity in D133A and the strict conservation of the Asp-133 residue are consistent with the proposed role of an aspartic residue in catalysis (25). However, having previously demonstrated that it is the COOH domain of Nop5p that binds the guide RNA (11), we could not rule out the possibility that the investigated fibrillarin residues (Thr-70, Glu-88, and Asp-133) are involved in binding the target RNA substrate. Thus, the observed reduction in catalytic activities caused by their mutations could be a result of impaired target substrate binding. DISCUSSION Understanding the function of ribosomal RNA processing and nucleotide modification requires detailed knowledge of the structure and thermodynamics of the box C/D snoRNP assembly. Archaeal box C/D sRNPs are ribonucleoprotein assemblies that selectively methylates 2Ј-hydroxyl groups of rRNA and tRNA. The core protein of box C/D sRNPs, fibrillarin, is responsible for the actual methylation reaction by transferring the methyl group from the methyl donor, AdoMet, to the target 2Ј-OH group on RNA. Fibrillarin forms a tight complex with another core protein, Nop5p, as a dimer of two heterodimers during initial assembly. The potential role of Nop5p in assisting cofactor binding to fibrillarin is investigated in this work by structural, thermodynamic, and in vitro methylation studies. Structural comparison of fibrillarin from different archaeal organisms revealed that the protein backbone and a critical FIG. 2. Structural comparison of fibrillarin suggests large conformational changes at the cofactor binding site. A, the holocomplex of fibrillarin-Nop5p (AF fibrillarin (blue) and AF Nop5p (green)) was superimposed with the apocomplex (AF fibrillarin (orange) and AF Nop5p (gray)). SigmaA-weighted 3F o Ϫ 2F c map at 1.0 was displayed at the cofactor binding site for the apocomplex, indicating the absence of a bound cofactor molecule. B, the holocomplex of fibrillarin-Nop5p (AF fibrillarin (blue) and AF Nop5p (green)) was superimposed with PH fibrillarin (gray). Phe-106 of PH fibrillarin is highlighted and labeled. C, the holocomplex of fibrillarin-Nop5p (AF fibrillarin (blue) and AF Nop5p (green)) was superimposed with MJ fibrillarin (cyan). Tyr-106 of MJ fibrillarin is highlighted and labeled.
aromatic residue near the cofactor binding site in free fibrillarin adopt a conformation that is different from that when Nop5p is bound. These alternative conformations in fibrillarin are unfavorable for efficient cofactor binding. Binding thermodynamics from ITC measurement indeed showed a reduced binding affinity of cofactor for fibrillarin in the absence of bound Nop5p. Mutational studies identified a cation-pair formed between Tyr-89 in fibrillarin and Arg-169 in Nop5p as one important structural element to facilitate the required conformation change at fibrillarin active site for AdoMet binding.
Despite the structural evidence that Thr-70 and Glu-88 bind AdoMet specifically, mutation of these two residues resulted in a slight increase in binding affinities. Such "super binding" The solvent-excluded surface with a solvent probe radius of 1.4 Å was used to define the solute-solvent boundary. The electrostatic potential ranges from yellow to red to white to blue to green, where yellow and red are negative potential, white is neutral, and blue and green are positive potential. The electrostatic potential range is Ϫ5 kcal/mol/e (yellow) to ϩ5 kcal/mol/e (green). mutants were also observed previously in a solid-phase assay of cofactor interaction with the vaccinia virus cap-dependent 2Ј-O-methyltransferase, VP39, where two motif IV mutants, although inactive, resulted in stronger binding coefficients of AdoMet (24). In another AdoMet-MTase interaction study by the equilibrium dialysis method, the loss of the 2Ј-hydroxyl group on AdoMet ribose moiety did not affect the optimal binding of the cofactor to a nucleolar 2Ј-O-methyltransferase (26). Because of the limitation in these binding assays, no explanatory insight could be provided on the structural and thermodynamic principles responsible for the gain in binding affinities upon removal of the interacting groups. Our analyses on directly measured thermodynamic quantities of cofactor binding to the T70A and E88A mutants offer some explanations on the nature of interactions. The large decrease in binding enthalpy of the cofactor to the E88A mutant suggests that the energetic costs in formation of the specific hydrogen bonds between Glu-88 and the cofactor may be compensated by breakage of hydrogen bonds formed between Glu-88 and solvent molecules or with nearby polar residues. In vitro methylation assays using the wild-type enzyme and those containing sitespecific mutations at active sites highlight the functional role of three conserved residues, Thr-70, Glu-88, and Asp-133, in catalysis.
Currently, little is known about the catalytic mechanism of s(no)RNPs. In general, MTases facilitate the methyl transfer reaction by restricting both AdoMet and the methylation target molecule in close proximity, which enables the thiomethyl group of the methionine moiety to be reactive toward polarizable nucleophiles (nitrogen, oxygen, sulfur, or activated carbons) (27). After donating the methyl group, AdoMet is converted into AdoHcy (12). Sequence alignment and structure superimposition of fibrillarin with other 2Ј-O-methylation MTases revealed a conserved KDK triad at the site of methyl transfer (28). Substitution of these three residues in the bacterial MTase RrmJ with alanine had deleterious effects on its methylation activity toward the 23 S rRNA substrate (23). This led to the proposal of a general base type of reaction mechanism employed by RrmJ in which a critical lysine residue facilitates nucleophilic attack by deprotonating the 2Ј-OH of the target RNA (23). In AF fibrillarin, the proposed catalytic triad residues correspond to Lys-42, Asp-133, and Lys-162. Both Lys-42 and Asp-133 are within 3.5 Å of the thiomethyl carbon of the bound AdoMet, which signifies their roles as catalytic residues. Activation of the 2Ј-hydroxyl could thus be occurring through the similar general base type of mechanism proposed for RrmJ. Asp-133 in fibrillarin could act as the general base by deprotonating the 2Ј-OH group on the target RNA during catalysis.
Our mutagenesis studies represent the initial efforts in probing the functional role of the potential catalytic residue Asp-133. Although the impaired catalytic activity of D133A is supportive of its role as a catalytic residue, the thermodynamic data on cofactor binding strongly imply its direct involvement in stabilization of cofactor binding. Several structural mechanisms are possible to rationalize the observed stabilization effect of Asp-133 on the bound AdoMet molecule. First, the negatively charged carboxyl side chain may form favorable interaction with the positively charged sulfonium sulfur of the cofactor. Our computed electrostatic potential surfaces of the wild-type and the D133A mutant proteins and the drastically lower binding affinity of the D133A mutant for AdoMet (which contains the sulfonium charge) than from AdoHcy (which lacks the sulfonium charge) are supportive of this mechanism. Second, the carboxylate group of Asp-133 may form a specific interaction with the amino group of AdoMet mediated by a water molecule. In the crystal structure of Nop5p-fibrillarin complex bound with AdoMet, a continuous electron density from the Asp-133 carboxylate group to the amino group was observed (data not shown) that could be attributed to a network of hydrogen bonds mediated by a water molecule, thus supporting this mechanism. Third, Asp-133 may form a salt bridge with the nearby Lys-42 residue, which could restrict the side chain of Lys-42 to an orientation that provides less steric hindrance to the sulfonium methyl group on AdoMet. Evidence supporting this mechanism is found in the binding study with various AdoMet analogs to the nucleolar 2Ј-O-methyltransferase (26) where substitution of an ethyl group for the sulfonium methyl greatly reduced the binding affinity of the cofactor analog. Although we favor the first mechanism that explains the role of Asp-133 in stabilizing the cofactor, additional studies are required to distinguish these mechanisms.
Previous biochemical studies show that Nop5p is essential in interacting with guide sRNA and in organizing a symmetric protein scaffold for sRNP assembly (18). Together with the thermodynamic and catalytic studies presented here, it is clear that Nop5p has important functional roles both as a scaffold protein in bridging the catalytic subunit to the substrate-guide RNA duplex and as an accessory protein for cofactor binding and catalysis.
FIG. 5. Methylation progression curves obtained from in vitro methylation assay on wild-type and mutant Nop5p-fibrillarin complexes. Each reaction mixtures contained Nop5p-fibrillarin or a mutant complex, L7Ae protein, box C/D guide RNA, target RNA oligo complementary to box C/D, and H 3 -labeled AdoMet cofactor. The incorporation of the methyl group was monitored by counting the retained radioactivity on extensively washed DE81 ion exchange filters at different intervals of time. The experimentally measured activities were normalized to be zero at t ϭ 0 min and were fitted to a typical progress curve.