insights into the catalytic mechanism of 16s rrna...

http://dx.doi.org/10.1016/j.jmb.2012.08.016 J. Mol. Biol. (2012) 423, 576–589

Contents lists available at www.sciencedirect.com

Journal of Molecular Biologyj ourna l homepage: ht tp : / /ees .e lsev ie r.com. jmb

Insights into the Catalytic Mechanism of 16S rRNAMethyltransferase RsmE (m3U1498) from Crystal andSolution Structures

Heng Zhang 1, Hua Wan2, 3, Zeng-Qiang Gao 1, Yong Wei 1, 4,Wen-Jia Wang 1, Guang-Feng Liu 1, Eleonora V. Shtykova 5,Jian-Hua Xu 1 and Yu-Hui Dong 1⁎1Beijing Synchrotron Radiation Facility, Institute of High Energy Physics, Chinese Academy of Sciences,19B, Yuquan Road, Beijing 100049, China2State Key Laboratory of Integrated Management of Pest Insects and Rodents, Institute of Zoology,Chinese Academy of Sciences, Beijing 100101, China3Graduate School of the Chinese Academy of Sciences, Beijing 100049, China4School of Life Sciences, University of Science and Technology of China, Hefei 230027, China5Institute of Crystallography, Russian Academy of Sciences, 59 Leninsky Prospekt, 117333 Moscow, Russia

Received 22 March 2012;received in revised form7 August 2012;accepted 20 August 2012Available online25 August 2012

Edited by J. Doudna

Keywords:RNA methylation;methyltransferase;PUA domain;trefoil knot;dimerization

*Corresponding author. E-mail addAbbreviations used: AdoMet, S-ad

SAXS, small-angle X-ray scattering;Bank; UMP, uridylic acid.

0022-2836/$ - see front matter © 2012 E

RsmE is the founding member of a new RNA methyltransferase (MTase)family responsible for methylation of U1498 in 16S ribosomal RNA inEscherichia coli. It is well conserved across bacteria and plants andmay play animportant role in ribosomal intersubunit communication. The crystal structurein monomer showed that it consists of two distinct but structurally relateddomains: the PUA (pseudouridine synthases and archaeosine‐specifictransglycosylases)-like RNA recognition and binding domain and theconserved MTase domain with a deep trefoil knot. Analysis of small-angleX-ray scattering data revealed that RsmE forms a flexible dimeric conforma-tion that may be essential for substrate binding. The S‐adenosyl‐L‐methionine(AdoMet)-binding characteristic determined by isothermal titration calorim-etry suggested that there is only oneAdoMetmolecule bound in the subunit ofthe homodimer. In vitromethylation assay of themutants based on the RsmE–AdoMet–uridylic acid complex model showed key residues involved insubstrate binding and catalysis. Comprehensive comparisons of RsmE withclosely relatedMTases, combinedwith the biochemical experiments, indicatedthat the MTase domain of one subunit in dimeric RsmE is responsible forbinding of one AdoMet molecule and catalytic process while the PUA-likedomain in the other subunit is mainly responsible for recognition of onesubstrate molecule (the ribosomal RNA fragment and ribosomal proteincomplex). The methylation process is required by collaboration of bothsubunits, and dimerization is functionally critical for catalysis. In general, ourstudy provides new information on the structure–function relationship ofRsmE and thereby suggests a novel catalytic mechanism.

© 2012 Elsevier Ltd. All rights reserved.

ress: [email protected]; AdoHcy, S-adenosyl homocysteine; MTase, methyltransferase;ITC, isothermal titration calorimetry; r-protein, ribosomal protein; PDB, Protein Data

lsevier Ltd. All rights reserved.

mailto:[email protected]

http://dx.doi.org/10.1016/j.jmb.2012.08.016

http://www.sciencedirect.com/science/journal/00222836

DELL

附注

“DELL”设置的“Marked”

577Catalytic Mechanism of RNA MTase RsmE

Introduction

Methylation of ribosomal RNA (rRNA) nucleo-tides mediated by methyltransferases (MTases)plays an important role in the biogenesis andactivity regulation of the ribosome, such as thefine-tuning of local rRNA structure, 30S subunitassembly, and antibiotic resistance.1–6 There arethree basic types of rRNA modifications: basemethylation, pseudouridylation, and 2′-O-methyla-tions. Base methylation is the most frequent type ofmodification in the rRNAs from bacteria.7,8 RsmE isa specific S‐adenosyl‐L‐methionine (AdoMet)-dependent MTase responsible for N3-methylationof U1498 of 16S rRNA in Escherichia coli.9 This site islocalized in the conserved region of the top of helix44 of 16S rRNA (Fig. 1), and its methylation wassuggested to play a role in intersubunit communi-cation. Moreover, it was reported that U1498directly interacts with the ribose–phosphate back-bone of the codon region,10,11 which is involved in

Fig. 1. (a) A schematic of the secondary structure of helices 4Biology of RNA Website (http://www.http://rna.ucsc.edu/rrelative location of m3U1498 (green sticks) in helix 44 (greensubunit (PDB ID: 2AW7; wheat).

forming an electronegative pocket that might bindfree nucleotides under certain conditions.10 ThersmE knockout (ΔrsmE) strain cannot competewhen grown together with wild-type cells, suggest-ing that this methylation plays an important role incell physiology.9

RsmE requires a highly structured ribonucleopro-tein particle (a fully assembled 30S ribosomesubunit) as a substrate for methylation, and thismethylation occurs late during 30S ribosomeassembly.12 However, methylation of U1498 invitro that proceeds very slowly with extremely lowkcat values was found, which has been presumeddue to the different actual substrate status betweenin vitro and in vivo or lost of a cofactor.12 Additionalproofs may be required to explain the “abnormal”phenomenon, such as from the view of its structure.RsmE is distinct from the SPOUT MTases familywith the trefoil knot structures and defines a newfamily of MTases characterized by its own specificmotifs.9 Moreover, the gene is highly conserved

4 and 45 of E. coli 16S rRNA from the Center for Molecularnacenter). m3U1498 is indicated by a green ellipse. (b) The) taken from the crystal structure of wild-type E. coli 30S

http://www.http://rna.ucsc.edu/rnacenter

†http://ekhidna.biocenter.helsinki.fi/dali_server

578 Catalytic Mechanism of RNA MTase RsmE

across bacteria and plants and may perform thesame function in other organisms. Although severalstructures of RsmE family proteins, such as Tm1380[Protein Data Bank (PDB) ID: 1Z85] from Thermotogamaritima, YggJ (PDB ID: 1NXZ) from Haemophilusinfluenzae, and YqeU (PDB ID: 1VHK) from Bacillussubtilis, have been determined, there are still nofunction studies on their structures. Moreover, theyhave been suggested as 2′-O-ribose MTases such asthe members of SpoU family,13 until RsmE fromE. coli was identified highly specific for m3U(3‐methyluridine) of 1498.Therefore, it is necessary to carry out functional

studies based on the structures of RsmE familymembers to gain insights into their catalyticprocess. Here, both crystal and solution structuresof E. coli RsmE were studied. Comparison of thesetwo structures, cofactor binding analyses, and invitro methylation assays of mutants involved incatalysis have now provided further understandingon the structure–function relationship of the con-served rRNA MTase and thereby suggest a novelcatalytic mechanism.

Results and Discussion

Overall structure

The structure E. coli RsmE was solved bymolecular replacement using the predicted MTaseHI0303 (PDB ID: 1VHY) from H. influenzae in spacegroup C2 as searching model and was refined to afinal R/Rfree factor of 0.21/0.25 at 2.25Å. Theasymmetric unit contains one RsmE molecule withoverall dimensions of ~95×44×75Å comprising 241residues with a C-terminal His tag and a total of 79water molecules.The structure of RsmE monomer similar to

HI0303 consists of two distinct but structurallyrelated domains: the putative RNA-binding domainformed by N-terminal sequence from Ile3 to Ile71and the conserved MTase domain formed byC-terminal sequence from Asp72 to Gly243, con-nected by a short linker region (Fig. 2a and b). Theputative RNA-binding domain consists of twistedfive β-sheets and one helix described as β1, β2, α1,β3, β4, and β5, similar to the PUA (pseudouridinesynthases and archaeosine‐specific transglycosy-lases) domain found in other RNA-bindingproteins.14 The MTase domain consists of fiveparallel β-sheets flanked by six helices (five α-helicesand one η-helix) described as β5, α2, β7, α3, α4, β8,η1, β9, α5, β10, and α6. Remarkably, the β10–α6segment is threaded through the β8–η1 loop in theC-terminus of this domain, forming a deep trefoilknot involved in AdoMet binding and catalysis(Fig. 2b and c), which has been found in manySPOUT MTases family members.15

Structural comparison of RsmE withrelated proteins

A DALI search† for globally similar proteins wasperformed within the PDB. Significant structuralsimilarity was found between RsmE and four othermembers of COG1385, which are all uncharacter-ized proteins solved by structural genomics consor-tia, thus far without published functional analyses.All of them are dimeric proteins with an α/β‐knotfold in the CTD (C-terminal domain).The PUA domains have been implicated to bind to

RNA with complex folded structures in manyproteins.14 Although the NTD (N‐terminal domain)of RsmE is similar to the PUA domains, itsarchitecture is not a typical PUA domain structureconsisting of two short helices and a β-sheet with sixmostly antiparallel β-strands. According to DALIstructure searches, the NTD shows no close struc-tural similarity with the typical PUA domains ofvarious RNA-binding proteins, such as pseudour-idine synthase TruB (PDB ID: 1R3E; r.m.s.d., 2.9Åfor 55 Cα atoms), rRNAMTase RlmI (PDB ID: 3C0K;r.m.s.d., 2.9Å for 53 Cα atoms), and 50S ribosomalprotein (r-protein) L25 (PDB ID: 1FEU; r.m.s.d., 2.9Åfor 59 Cα atoms). Superimposition of theses PUAdomains also showed that the PUA domain of RsmEhas diverged markedly (Fig. 3a–c). The structuralvariations in the PUA domain probably reflect theirfunction in recognizing distinct and specific sub-strates. The similarity to RNA-binding PUA do-mains and the positive surface charge distributionobserved in RsmE structures are suggestive fora conserved function of the PUA-like domain inRNA recognition.The DALI search and structural comparison of the

CTD of RsmE also showed apparent difference fromthe related domains of SPOUT family members:tRNA MTase TrmH (PDB ID: 1V2X; r.m.s.d., 3.1Åfor 192 Cα atoms), tRNA MTase TrmD (PDB ID:1P9P; r.m.s.d., 3.3Å for 236 Cα atoms), and theputative MTase YibK (PDB ID: 1MXI; r.m.s.d., 3.0Åfor 133 Cα atoms). However, their knot structuresinvolved in the catalytic domains are very similar(Fig. 3d–f), indicating that they may share universalcatalytic mechanisms.

Dimerization of RsmE in crystal

Although the crystals contain only one monomerper asymmetric unit, two monomers can form acompact dimer across a crystallographic 2‐fold axis,consisting of two subunits named subunit I (orange)and subunit II (green), respectively (Fig. 4a and b).The dimer is in an antiparallel fashion similar to E.coli TrmD but opposite to the members of SpoU

http://ekhidna.biocenter.helsinki.fi/dali_server

Fig. 2. (a) Structure-based sequence alignment for RsmE family including E. coli (E_coli), Klebsiella pneumoniae (K_pneumoniae), Yersinia pestis (Y_pestis), H.influenzae (H_influenzae), Burkholderia pseudomallei (B_pseudomallei), Corynebacterium pseudotuberculosis (C_pseudotuberculosis), Staphylococcus aureus (S_aureus),Midichloria mitochondrii (M_mitochondrii), and Pseudomonas stutzeri (P_stutzeri), performed using ClustalX (version 1.81) and ESPript 2.2. The conserved residues areboxed in blue, and identical conserved residues and low conserved residues are highlighted in red background and red letters, respectively. (b) Cartoon representationshowing the domain architecture of RsmE in monomer with helix in cyan, sheet in magenta, and loop in brown. (c) Stereo view of the electron density (in blue) in theactive‐site pocket containing the trefoil knot from a 2Fo−Fc map contoured at 2.0σ.

579Catalytic

Mechanism

ofRNAMTase

Rsm

E

image of Fig.�2

Fig. 3. Structural comparisons of NTD and CTD of RsmE with related proteins. (a–c) Structural comparison of NTD(green) with the PUA domain (cyan) of tRNA pseudouridine synthase TruB (PDB ID: 1R3E), rRNAMTase RlmI (PDB ID:3C0K), and the r-protein TL5 (PDB ID: 1FEU), respectively. (d–f) Structural comparison of CTD (green) with the catalyticdomain (containing the trefoil knot; magenta) of tRNAMTase TrmH (PDB ID: 1V2X), tRNAMTase TrmD (PDB ID: 1P9P),and the putative MTase YibK (PDB ID: 1MXI), respectively. The AdoMet or AdoHcymolecule is shown inmagenta sticks.


family MTases with parallel fashions.15 The buriedsurface area in the dimer interface is 1443Å2, whichis 11.3% of the total surface area per monomer(12,768Å2). The large dimer interface with extensiveinteractions is mainly formed by two α6 helices andβ10–α6 loops in the C-terminal catalytic domains ofthe two subunits (Figs. 2a and 4a). The residuesArg220, Thr226, Asp241, Lys98, and Arg74 fromsubunit II form direct interactions with Asp241′,Lys98′, Arg220′, Thr226′, and Pro219′, respectively(where primes denote amino acids from subunit I).Besides, there are indirect interactions of severalresidues such as Leu216 and Leu242′ and Val237and Arg238′ via well-ordered water molecules bytheir side chains.Dimerization has been shownvital for preservation

of the native structure, tRNA binding, and methyl-ation catalysis in α/β-knot MTases.15 For example, anotable loss observed in structure involving disrup-tion to the active site inmonomeric YibK, rendering itincapable of cofactor binding, was observed inmonomeric YibK.16 The interface in RsmE dimermay provide a platform for the catalytic process andcontribute to the stabilization of each subunit. Mostresidues on the dimer interface are conserved(Fig. 2a), which also supports the importance ofdimerization of RsmE from an evolutionary point ofview. More importantly, most of these interactingresidues are located in the trefoil knot of the catalytic

core domain of both subunits. The results showedthat molecular dimerization is required for catalysisby RsmE and that the trefoil knot plays an importantrole in the dimerization. Similar role has beenconfirmed in the knotted tRNA MTase TrmD fromE. coli. The extensive mutational analysis in TrmDrevealed that the mutations made outside thecatalytic region but at the dimer interface led toinactivation of the protein, and therefore, formationof a homodimer was required for activity.17

Solution structure of RsmE

The gel‐filtration chromatography demonstratedthat RsmE has a dimeric structure in solution12

similar to other members of SPOUT family. How-ever, the crystals obtained by us contain only onemonomer per asymmetric unit. To solve thiscontradiction, we performed small‐angle X‐rayscattering (SAXS) study of the intact structure ofRsmE in solution. It is reasonable to assume thatSAXS data analysis should also reveal the dimer-ization. Therefore, an additional purpose of theSAXS study was to compare the dimeric crystalstructure of RsmE above with that obtained bySAXS. It is especially important since its organiza-tion in solution and crystal may vary greatly due tospecial conditions at crystallization that differ from aphysiological state.

image of Fig.�3

Fig. 4. (a) Cartoon representation showing the dimer structure of RsmE consisting of subunit I (orange) and subunit II (green) and the interacting residues on theinterface. (b) A molecular surface representation of RsmE dimer, colored by its local electrostatic potential (blue, +7kT; red, −7kT).

581Catalytic

Mechanism

ofRNAMTase

Rsm

E

image of Fig.�4


Experimental SAXS curve after standard prelimi-nary processing (Materials andMethods) is presentedin Fig. 5a (curve 1). General structural parameterscalculated from the SAXS profile using Guinierapproximation and indirect Fourier transformation(program GNOM) are collected in Table S1.

Radii of gyration evaluated by different indepen-dent methods give practically the same values,which are close to the theoretical one determinedfrom the dimeric crystal structure by the programCRYSOL (25.8Å). It is the first evidence of theformation of a dimeric structure of RsmE in solution

Fig. 5. (a) Comparison of scatter-ing from monomeric and dimericstructures of RsmE: experimentalSAXS data (1), theoretical scatteringcurves computed by the programCRYSOL from the monomeric(2) and dimeric (3) crystal structuresof RsmE, and average model curvecomputed by the program EOM(4). Inset: distributions of Rg andDmax computed by the programEOM. (b) Structural modeling.Upper panel: scattering patternscorresponding to the DAMMINand GASBOR restorations are dis-placed one logarithmic unit downfor better visualization. For bothsets of curves: experimental SAXScurve (1), model scattering patterns(2), and smooth curve back trans-formed from p(r) and extrapolatedto zero scattering angle (3). Inset:the distance distribution functionsp(r) used at reconstructions. Lowerpanel: superposition of DAMMINand EOM models, the latter wascomputed using available high‐resolution structure of the RsmE(shown in ribbons PDB mode)(a), comparison of EOM model ofthe RsmE (b), and crystal dimericstructure of the protein (c) inspacefill PDB mode for better visu-alization of the difference of theirspatial organization.

image of Fig.�5


revealed by SAXS. The second one is estimation ofmolecular massM of the protein using the scatteringintensity at zero angle I0. Bovine serum albuminwith molecular mass of 66.5kDa was used as areference protein to calculateM of RsmE in solution.It proved to be 56.4kDa. Taking into account thatmolecular mass of the monomer unit of RsmE is27kDa, we can accept the existence of dimericconformation of RsmE. All other structural charac-teristics confirm the conclusion.Comparison of theoretical scattering patterns

from both crystal structures, monomeric and dimer-ic, with experimental SAXS profile from RsmEclearly indicates dimerization of the protein insolution (Fig. 5a, curves 2 and 3): the model curvefrom the available monomeric crystal structure ofRsmE differs considerably from the experimentalSAXS profile, while the crystallographic dimeryields good fit to it with average χ=2.6. It shouldbe noted that the SAXS method is low resolution,and it cannot describe and model the inner structureof biological macromolecules. Importantly, to getgood model fit to the experimental profile in therange of small angles close to the primary beam, thisscattering interval corresponds to the shape of thescattering object. Higher‐angle region can be dis-regarded in the calculation of the shape, anddiscrepancy can be significant in this region.Moreover, usage of high‐resolution crystal struc-tures at shape calculation leads to an increase of thediscrepancy at higher angles due to possibledifferences between inner conformations in crystaland in solution.Analysis of possible mutual arrangement of the

monomers in the dimeric structure of RsmE insolution was performed by the program EOM,which takes into account the coexistence of differentconformation contributing to the experimentalscattering pattern. These conformers are selectedusing a genetic algorithm from a pool containing alarge number of randomly generated models cover-ing the protein configurational space, that is, theprogram generates an ensemble of models, whosecombined theoretical scattering intensity best de-scribes the experimental SAXS data. Curve 4 inFig. 5a demonstrates average model fit (χ=2.3) tothe experimental SAXS profile obtained by theprogram EOM. Variability of the structures in thepool of the models is reflected in distributions of RgandDmax presented in the inset of Fig. 5a. Accordingto the EOM analysis, radius of gyration Rg of RsmEcan vary from 24 to 32Å with average value of26.6Å, and the maximal size Dmax of the proteinranges from 72 to 105Å with average Dmax of 80.1Å.Both these values are in good agreement with thedimensions obtained by other methods (Table S1).Variability of RsmE structure revealed by EOMpoints to a possible flexibility of its dimeric ensemblein solution. One of those obtained by EOM

structural configurations of RsmE is shown in thelower panel of Fig. 5b. Other models (not shown dueto their large number) have similar organization.Independent confirmation of the results above is

ab initio reconstruction of RsmE structure in solutionby the programs DAMMIN and GASBOR.Figure 5b, upper panel, demonstrates the results ofshape restorations along with the distance distribu-tion function p(r) used for the procedure (Fig. 5b,inset). The scattering patterns corresponding to theDAMMIN and GASBOR restorations are displacedone logarithmic unit down for better visualization.As can be seen from Fig. 5b, all model curves yieldvery good fits to the experimental data withdiscrepancy χ=1.5/1.6, and these obtained low‐resolution structural models are in a good agree-ment with both the crystallographic data and theEOM modeling (Fig. 5b, lower panel).Conformation of RsmE in solution revealed by

SAXS is rather flexible, which is probably a result ofthe absence of the substrate. This flexibility insolution can be essential for its more complicatedsubstrate recognition and catalysis. Compared toRsmE, although the homodimeric status of theSPOUT family members YibK and E. coli YbeAwere also confirmed by SAXS, their conformationsin solution are largely consistent with those incrystal.18

Thermodynamics of AdoMet binding by RsmE

The interaction of RsmE with AdoMet wascharacterized by isothermal titration calorimetry(ITC), and the integrated heat data could be fittedwell in the sequential binding site models (Fig. 6).Theoretically, the AdoMet-binding site is identical ineach subunit of homodimeric RsmE, which couldbind two AdoMet molecules sequentially. However,unexpectedly, the first association constant (Ka1=9.44±0.72×104M−1) is much higher than the secondone (Ka2=741±51M

−1) while the first enthalpy(ΔH1=−1444±37.7 cal/mol) is much lower thanthe second one (ΔH2=−2.98±0.19×104 cal/mol).The result indicated that the binding of two AdoMetmay be highly competitive and that access of thesecond AdoMet molecule will be blocked after thefirst one is bound. Therefore, only one AdoMetmolecule with one subsequent substrate moleculecould be bound by dimeric RsmE. Moreover, the Ka1value is significantly lower than those of severalrRNAMTases, such as Ka=2.09×10

5M−1 in RsmC19

and Ka=3.4×105M−1 in RlmI,20 indicating the

lower binding affinity of RsmE to AdoMet. Thatmay be one of the reasons that methylation ofU1498 occurs very slowly. Our result is consistentwith the conclusion that only one S‐adenosylhomocysteine (AdoHcy) molecule binds to eachYibK dimer, whose stoichiometry was also deter-mined by ITC experiments.16

Fig. 6. Thermodynamic analysis of RsmE titrated against the cofactor AdoMet. The raw data are presented on the toppanel (left, with the baseline subtracted), while the bottom panel shows the fitted titration curve (left). The filled dotsindicate the experimental data, the best fit to the experimental data was obtained from a nonlinear least‐squares methodof fitting using a double-site binding model depicted by a continuous line using the Origin software, and the parameterswere shown (right). Ka is the association constant, ΔH is the change in enthalpy, and ΔS is the change in entropy.


In vitro activity of RsmE mutants

Based on the ITC result (Fig. 6) and the similaritiesof RsmE with TrmH-AdoMet, TrmD-AdoHcy, andYibK-AdoHcy (Fig. 3d–f), we built a model ofRsmE–AdoMet–uridylic acid (UMP) complex. Inthis model, one AdoMet molecule and one UMPmolecule were docked sequentially into trefoil knotin the catalytic domain from the subunit II catalyticdomain of the dimeric RsmE (Fig. 7a). Both AdoMetand UMP showed generally favorable direct in-teractions with several conserved residues withoutbad steric clashes, respectively (Fig. 7b). Thehydrophobic Val221 could interact with bothligands among these residues. It is also worth notingthat the UMP could also directly interact with thehighly conserved Gln141′ from subunit I.These residues with direction interaction with

UMP and those residues (His27, Arg33, Lys59,Arg114, Lys126, and Lys133) predicted to beinvolved in rRNA binding13 were mutated toalanine to identify key residues in substrate recog-nition and catalysis by determining their MTaseactivities in vitro. The results showed that all mutantsexhibit reduced activities compared to the wild-typeRsmE (Fig. 7c). The highly conserved R223A locatedin the β10–α6 loop of catalytic site showed the most

severe loss of activity to 9% of the wild type,followed by T224A to 34%. Besides, Q141A showedthe loss of activity to 45%, suggesting its certain rolein cooperation of these two subunits in catalysisprocess. Among the mutant residues involved inrRNA binding, the highly conserved R33A andR114A also showed significant loss of activity to22% and 31% of the wild type, respectively. None ofthe mutant residues in vitro were completelyinactive, similar to the mutagenesis results of rRNAMTase RsmC,19 indicating that the catalytic reactioninvolving substrate binding is a complicated processrequired by the participation of several key residues.The electrostatic surface of the MTase structures

couldmost likely reflect the regionwhere the RNAorDNAsubstrateswill bind.On the electrostatic surfaceof the RsmE dimer (Fig. 4b), there is a long grooveconsisting mainly of positively charged conservedresidues, such as His27, Arg33, and Lys59. Inaddition, these residues are surrounded by manyhighly conserved residues on the surface of the twosubunits. Moreover, Arg33 and Lys59 are located inthe flexible region and these positively chargedresidues are close to the trefoil knot of the catalyticsite, whichmaymediate the binding of the negativelycharged substrate near the active site of RsmE. In ourstructure-based mutagenesis experiments, the roles

image of Fig.�6

Fig. 7. (a) Model of RsmE–AdoMet–UMP complex. One AdoMet molecule (red sticks) and one UMP molecule (cyansticks) were docked in sequence into the catalytic domain of subunit II (green) of dimeric RsmE. (b) The residues fromsubunit I (orange) and subunit II (green) directly interacting with AdoMet and UMP in the model. (c) In vitro MTaseactivities of the RsmE variants, determined colorimetrically as described in Materials and Methods. The MTase reactionwas carried out by incubating RsmEmutants with purified 30S subunit from E. coli K-12ΔrsmE strain and 100μM SAM at37°C for 1h. The activity is shown as the percentage of the wild-type MTase activity. Data presented are presented as theaverage (±standard error of the mean) from duplicate experiments.


of these positive residues located in subunit I, close tothe catalytic site of subunit II, were confirmed,although the mutations occur on both subunits.

Implication for the catalytic mechanism of RsmE

The role of the trefoil knot that forms AdoMet-bindingdomainand catalytic domainhasbeen studiedintensively in tRNAmethylation catalyzed by dimerictRNA-dependent SPOUT MTases.17,21–24 The novelRNA-dependent methylation mechanism based ondimerization for TrmH suggested that one monomersubunit binds AdoMet in the trefoil knot while theother serves as a tRNA-binding site in the dimer.23

Similar methylation pattern may be adopted by

RsmE, considering its similar trefoil knot structurewith TrmH. However, the real substrate of RsmE,consisting of rRNA and r-protein in vivo, is highlystructured and more complicated, and its methylationmechanisms may be somewhat different.Since the r-protein involved in the substrate of

RsmE remains unknown, we attempted to build anRsmE–AdoMet–rRNA (containing the r-protein)ternary complex model using the whole E. coli 30Ssubunit (PDB ID: 2AW7) with a number of low-resolution and high-resolution methods such asGRAMM and Z-DOCK. However, no reasonablesolutions were obtained. Then, we tried the modelusing variable lengths of rRNA fragments contain-ing U1498 with removal of r-protein and still cannot

image of Fig.�7


get improved results. The reason may be due to thestructural rearrangement of the access to U1498 inRsmE. The substrate rRNA may undergo significantconformation changes and then the target will flipout into the active site, as suggested by the flexibleconformation from the SAXS result. Similar obser-vations have been reported in TrmH and TrmD,17,22

where the substrate tRNA would undergo asignificant conformational change that disrupts theD loop. The docking model of rRNA MTase RsmF(m5C1407) onto the 30S subunit indicates that theC-terminal PUA domain and the N-terminal MTasedomain make several contacts with 16S rRNA andthe r-protein S12 and therefore explains why itssubstrate is only the assembled 30S subunit.25 Thesignificant similarity on the substrate preferencebetween RsmF and RsmE allows us to speculatethat RsmE can also interact with both rRNA andr-protein, which is also considered from similarlocation of their modifying target in helix 44 of 30Ssubunit (Fig. S1a). More importantly, although theRsmE–AdoMet–rRNA model was inactive (thedistance between the AdoMet methyl group andthe N3 atom of U1498 is ~12Å), interactions of bothsubunits of dimeric RsmE especially in positive-charged regions with the negative rRNA fragmentcan be obviously observed (Fig. S1b and c).

Conclusion

In the present study, we reported and comparedthe structure of RsmE in crystal with that in solution.The ITC experiment and structure-guided mutagen-esis revealed the AdoMet‐binding characteristic andkey residues involved in substrate binding andcatalysis. We can conclude that, during U1498methylation catalyzed by dimeric RsmE, theMTase domain of one subunit is responsible forthe binding of one AdoMet molecule and catalyticprocess while the PUA-like domain in the othersubunit is mainly responsible for recognition ofone substrate molecule (the rRNA fragment andr-protein complex). The methylation process isrequired by collaboration of both subunits, anddimerization is functionally critical for substraterecognition and catalysis by RsmE. These systematicstudies give us a deeper molecular understanding ofthe specific recognition mechanism of U1498 in 16SrRNA by MTase RsmE.

Materials and Methods

Cloning

The full length of rsmE was PCR-amplified from E. coliK-12 genomic DNA, cloned into the expression vectorpET21a with a non-cleavable C-terminal His6 tag (Nova-

gen, USA), transformed into E. coli DH5α cloning strain,and plated onto LB ampicillin plates. The plasmid wasisolated and transformed into an E. coli BL21 (DE3) starexpression strain (Invitrogen). Site-directed mutagenesisof rsmE was performed by a PCR-based techniqueaccording to the QuikChange site-directed mutagenesisstrategy (Stratagene) following the manufacturer's in-structions. The mutant genes were sequenced and foundto contain only the desired mutations.

Protein expression and purification

Bacterial cells were grown to mid-log phase in LBmediaat 37°C in the presence of 50mg/ml ampicillin. Inductionof the culture was then carried out with 0.3mM isopropyl-1-thio-β-D-galactopyranoside at 20°C. Cells were pelletedafter 20h by centrifugation at 8000rpm for 10min at 4°C.The cell pellet was resuspended in buffer A [20mM Tris,500mM NaCl, 5% (v/v) glycerol, 2mM β-mercaptoetha-nol, and 1mM PMSF (pH8.0)] and lysed by ultrasonifica-tion on ice. The cell debris and membranes were pelletedby centrifugation at 16,000rpm (R20A2 rotor, Hitachihigh-speed refrigerated centrifuge R21GIII) for 60min at4°C. The soluble C-terminally His6-tagged RsmE waspurified by affinity chromatography with nickel-nitrilo-triacetic acid resin (Bio-Rad). Untagged proteins wereremoved with buffer A containing 35mM imidazole.Recombinant RsmE was then eluted with buffer Acontaining 250mM imidazole. The protein was furtherpurified by gel filtration (Superdex 75, GE Healthcare)equilibrated in buffer B [20mM Tris, 100mM NaCl,5% (v/v) glycerol, and 2mM DTT (pH8.0)] using anÄKTA Purifier System (Amersham). Highly purifiedRsmE fractions were pooled and concentrated to 24mg/ml quantified using the Bio-Rad protein assay kit forcrystallization trials by ultrafiltration in an Amicon cell(Millipore, USA).

Protein crystallization

Initial crystallization condition of RsmE-His wasobtained under the 95# crystallization conditions in indexkit (Cat No. HR2-144; Hampton Research, USA) with thesitting‐drop vapor diffusion method at room temperatureafter 2days. The crystal quality was optimized byadjusting the concentration of the precipitant and buffer.The best crystal was obtained in solution containing 0.15Mpotassium thiocyanate and 24% (w/v) polyethylene glycolmonomethyl ether 2000 after 3–4days. Attempts at co-crystallizing RsmE with AdoMet or RsmE crystals soakedin AdoMet solution were not successful.

Data collection, crystal structure determination,and refinement

Diffraction data were collected on the beamline station1W2B of Beijing Synchrotron Radiation Facility. Beforedata collection, crystals were soaked for 5s in a cryopro-tectant consisting of 20% (v/v) glycerol in the crystalmother liquid and then flash-frozen in liquid nitrogen.The temperature was hold at 100K by liquid nitrogenduring data collection. Data were processed with theprogram HKL2000.26

Table 1. Data collection and refinement statistics

Data collectionWavelength (Å) 1.0Space group C2Unit cell parametersa, b, c (Å) 95.48, 43.68, 74.88β (°) 127.54Resolution (Å) 2.25 (2.29–2.25)a

Number of unique reflections 11,818 (596)Completeness (%) 99.9 (100)Redundancy 3.7 (3.5)Mean I/ơ (I) 29.54 (5.13)Molecules in the asymmetric unit 1Rmerge (%) 4.2 (20.9)

Structure refinementResolution range (Å) 47.71–2.25Rwork/Rfree (%) 20.9/25.4Average B‐factor (Å2)Main chain 41.62Side chain 44.68Waters 38.12Ramachandran plot (%)Most favored 96.7Allowed 3.3r.m.s.d.Bond lengths (Å) 0.010Bond angles (°) 1.246a The values in parentheses mean those of the highest‐

resolution shell.


The initial phases were calculated using the programPhaser27 with the predicted MTase HI0303 (PDB ID:1VHY) as the searching model. The phases from Phaserand the structure factors fromHKL2000 were combined inARP/wARP.28 Some biases were reduced manually inCoot.29 The program Phenix.refine30 was used to refinethe structure and append the water molecules.PROCHECK31 was used to validate the structure. Asummary of data collection and final refinement statisticsis listed in Table 1. The program PyMOL‡ was used toprepare structural figures.

SAXS measurement and data processing

Synchrotron SAXS measurements were performed atthe European Molecular Biology Laboratory on thestorage ring DORIS III of the Deutsches ElektronenSynchrotron (Hamburg) on the X33 beamline32 equippedwith a robotic sample changer33 and a PILATUSdetector (DECTRIS, Switzerland). The scattering wasrecorded in the range of the momentum transfer0.07bsb5.5nm−1, where s=(4πsinθ)/λ, 2θ is the scatter-ing angle, and λ=0.15nm is the X-ray wavelength. Themeasurements were carried out in a vacuum cuvettewith exposure times of 2min to diminish the parasiticscattering. The experimental scattering profiles from thesamples were corrected for the background scatteringfrom the solvent and were processed using standardprocedures and the program PRIMUS.34 The sampleswere measured at the four different solute concentra-

‡http://www.pymol.sourceforge.net/

tions, 3.0mg/ml, 5.0mg/ml, 7.0mg/ml, and 8.5mg/ml,with further extrapolation to zero concentration. Thelatter was used for structural analysis.General structural parameters such as radius of gyration

Rg, molecular massM, distance distribution functions p(r),and a maximal size of RsmE in solution were computedusingprogramsPRIMUS34 andGNOM.35 The p(r) functionwas further used for ab initio low‐resolution shaperestoration by programs DAMMIN36 and GASBOR.37

Theoretical scattering patterns I(s) from the availablehigh‐resolution coordinates of the protein were calculatedby a program CRYSOL.38 Additional structural informa-tion was obtained by the program EOM, which seeks todescribe experimental SAXS data using an ensemblerepresentation of atomic models.39

Isothermal titration calorimetry

ITCwas applied to quantitatively determine the bindingaffinity of RsmE to AdoMet. For the titration experiments,the protein was purified with the same method as aboveand was dialyzed against the buffer containing 50mM NaHepes (pH7.5), 0.15M NaCl, 5% (v/v) glycerol, 2mMMgCl2, and 2mM β-mercaptoethanol for 24h. AdoMetwas dissolved in the same buffer as above. The ITCexperiments were carried out using a high-sensitivity iTC-200 microcalorimeter from MicroCal (GE Healthcare) at20°C using 100μM AdoMet in the injector 2.5‐μM RsmE(quantified in the dimeric form) in the sample cell. Allsamples were thoroughly degassed and then centrifugedto get rid of precipitates. Injection volumes of 2μl perinjection were used for the different experiments, and forevery experiment, the heat of dilution for each ligand wasmeasured and subtracted from the calorimetric titrationexperimental runs for the protein. Consecutive injectionswere separated by 2min to allow the peak to return to thebaseline. Integrated heat data obtained for the ITCs werefitted in a double-site model using a nonlinear least-squares minimization algorithm to a theoretical titrationcurve, using the MicroCal Origin 7.0 software package.

In vitro methylation assays

30S ribosomal subunits were isolated from E. coli K-12ΔrsmE strain obtained from the ASKA recloned library[National BioResource Project (National Institute ofGenetics, Japan): E. coli] by the method as describedpreviously.9 Quantitation of subunits was determined byabsorbance at 260nm (1 A260 unit is equivalent to34.5pmol of 30S ribosomes). In vitro MTase assay wascarried out using a nonradioactive colorimetric assay kit(SAM510™; G Biosciences) with slight modification.MTase reaction was first carried out in a 20‐μl reactionmixture containing 0.8μM wild type and mutants ofRsmE, 1.5μM purified 30S ribosome subunit, and 100μMAdoMet and was incubated at 37°C for 1h. We added10μl of the reaction mixture to 90μl of “master mix”without SAM as suggested by the supplier. This reactionwas further incubated at 37°C. The rate of production ofhydrogen peroxide is monitored colorimetrically with 3,5‐dichloro-2-hydroxybenzene sulfonic acid at 510nm tomeasure the MTase reaction. A parallel control reaction tosubtract background where MTase assay was performed

http://www.pymol.sourceforge.net/


in the absence of AdoMet was done. Each mutant samplewas run in duplicate to ensure accuracy.

Molecularmodeling of theRsmE–AdoMet–UMPcomplex

The docking model was performed with AutoDock3.0.5.40 One AdoMet molecule and one UMP molecule(ligand) that was kept rigid were docked into the trefoilknot of RsmE (receptor) in sequence. Poses from eachround of docking were subsequently ranked, according tothe affinity scores that describe clashes of the ligands withthe receptor molecule and the proximity between AdoMetand UMP. The best-scoring poses were regarded as themost likely models.

PDB accession code

The atomic coordinates and structure factors of RsmEhave been deposited with the Research Collaboratory forStructural Bioinformatics PDBwith the identifier code 4E8B.

Acknowledgements

We are grateful to the staff of the beamline station3W1A of Beijing Synchrotron Radiation Facility andthe X33 beamline of European Molecular BiologyLaboratory for providing technical support and formany fruitful discussions. This work was supportedby the grants from the National Natural ScienceFoundation of China (10979005 and 31200552) andthe National Basic Research Program of China(2009CB918600 and 2012CB917203).

Supplementary Data

Supplementary data to this article can be foundonline at http://dx.doi.org/10.1016/j.jmb.2012.08.016

References

1. Ofengand, J. & Fournier, M. J. (1998). The pseudour-idine residues of rRNA: number, location, biosynthe-sis, and function. In Modification and Editing of RNA(Grosjean, H. & Benne, R., eds), pp. 229–253, ASMPress, Washington, DC.

2. Decatur, W. A. & Fournier, M. J. (2002). rRNAmodifications and ribosome function. Trends Biochem.Sci. 27, 344–351.

3. Douthwaite, S., Fourmy, D. & Yoshizawa, S.(2005). Fine-tuning of RNA functions by modifica-tion and editing. pp. 1–22, Springer-Verlag, Berlin,Germany.

4. Chow, C. S., Lamichhane, T. N. & Mahto, S. K. (2007).Expanding the nucleotide repertoire of the ribosomewith post-transcriptional modifications. ACS Chem.Biol. 2, 610–619.

5. Connolly, K., Rife, J. P. & Culver, G. (2008).Mechanistic insight into the ribosome biogenesisfunctions of the ancient protein KsgA. Mol. Microbiol.70, 1062–1075.

6. Xu, Z., O'Farrell, H. C., Rife, J. P. & Culver, G. M.(2008). A conserved rRNA methyltransferase regu-lates ribosome biogenesis. Nat. Struct. Mol. Biol. 15,534–536.

7. Ofengand, J. & Del Campo, M. (2004). Modifiednucleotides of Escherichia coli ribosomal RNA. InEcoSal—chapter 4.6.1 (online) Escherichia coli and Sal-monella: Cellular and Molecular Biology (Curtiss, R., ed.),ASM Press, Washington, DC.

8. Andersen, N. M. & Douthwaite, S. (2006). YebU is am5C methyltransferase specific for 16S rRNA nucle-otide 1407. J. Mol. Biol. 359, 777–786.

9. Basturea, G. N., Rudd, K. E. & Deutscher, M. P. (2006).Identification and characterization of RsmE, thefounding member of a new RNA base methyltransfer-ase family. RNA, 12, 426–434.

10. Berk, V., Zang, W., Pai, R. D. & Cate, J. H. (2006).Structural basis for mRNA and tRNA positioning onthe ribosome. Proc. Natl Acad. Sci. USA, 103,15830–15834.

11. Korostelev, A., Trakhanov, S., Laurberg, M. & Noler,H. F. (2006). Crystal structure of a 70S ribosome–tRNA complex reveals functional interactions andrearrangements. Cell, 126, 1065–1077.

12. Basturea, G. N. & Deutscher, M. P. (2007). Substratespecificity and properties of the Escherichia coli 16SrRNA methyltransferase, RsmE. RNA, 13, 1969–1976.

13. Forouhar, F., Shen, J., Xiao, R., Acton, T. B.,Montelione, G. T. & Tong, L. (2003). Functionalassignment based on structural analysis: crystalstructure of the YggJ protein (HI0303) of Haemophilusinfluenzae reveals an RNA methyltransferase with adeep trefoil knot. Proteins: Struct., Funct., Genet. 53,329–332.

14. Pérez-Arellano, I. & Gallego, J. (2007). The PUAdomain—a structural and functional overview. FEBSJ. 274, 4972–4984.

15. Tkaczuk, K. L., Dunin-Horkawicz, S., Purta, E. &Bujnicki, J. M. (2007). Structural and evolutionarybioinformatics of the SPOUT superfamily of methyl-transferases. BMC Bioinformatics, 8, 73.

16. Mallam, A. L. & Jackson, S. E. (2007). The dimerizationof an α/β-knotted protein is essential for structureand function. Structure, 15, 111–122.

17. Elkins, P. A., Watts, J. M., Zalacain, M., van Thiel, A.,Vitazka, P. R., Redlak, M. et al. (2003). Insights intocatalysis by a knotted TrmD tRNA methyltransferase.J. Mol. Biol. 333, 931–949.

18. Mallam, A. L., Onuoha, S. C., Grossmann, J. G. &Jackson, S. E. (2008). Knotted fusion proteins revealunexpected possibilities in protein folding. Mol. Cell,30, 642–648.

19. Sunita, S., Purta, E., Durawa, M., Tkaczuk, K. L.,Swaathi, J., Bujnicki, J. M. & Sivaraman, J. (2007).Functional specialization of domains tandemly dupli-cated within 16S rRNA methyltransferase RsmC.Nucleic Acids Res. 35, 4264–4274.

20. Sunita, S., Tkaczuk, K. L., Purta, E., Kasprzak, J. M.,Douthwaite, S., Bujnicki, J. M. & Sivaraman, J. (2008).Crystal structure of the Escherichia coli 23S rRNA: m5C

http://dx.doi.org/10.1016/j.jmb.2012.08.016


methyltransferase RlmI (YccW) reveals evolutionarylinks between RNAmodification enzymes. J. Mol. Biol.383, 652–666.

21. Ahn, H. J., Kim, H.W., Yoon, H. J., Lee, B. I., Suh, S.W.& Yang, J. K. (2003). Crystal structure of tRNA(m1G37) methyltransferase: insights into tRNA rec-ognition. EMBO J. 22, 2593–2603.

22. Nureki, O., Watanabe, K., Fukai, S., Ishii, R., Endo,Y., Hori, H. & Yokoyama, S. (2004). Deep knotstructure for construction of active site and cofactorbinding site of tRNA modification enzyme. Structure,12, 593–602.

23. Mosbacher, T. G., Bechthold, A. & Schulz, G. E. (2005).Structure and function of the antibiotic resistance-mediating methyltransferase AviRb from Streptomycesviridochromogenes. J. Mol. Biol. 345, 535–545.

24. Watanabe, K., Nureki, O., Fukai, S., Ishii, R., Okamoto,H., Yokoyama, S. et al. (2005). Roles of conserved aminoacid sequence motifs in the SpoU (TrmH) RNAmethyltransferase family. J. Biol. Chem. 280, 10368–10377.

25. Hallberg, B. M., Ericsson, U. B., Johnson, K. A.,Nordlund, P. & Erlandsen, H. (2006). The structure ofYebU, an RNA m5C methyltransferase from Escher-ichia coli reveals a C-terminal RNA-recruiting PUAdomain. J. Mol. Biol. 360, 774–787.

26. Otwinowski, Z. & Minor, W. (1997). Processing ofX-ray diffraction data collected in oscillation mode.Methods Enzymol. 276, 307–326.

27. McCoy, A. J., Grosse-Kunstleve, R. W., Adams, P. D.,Winn, M. D., Storoni, L. C. & Read, R. J. (2007). Phasercrystallographic software. J. Appl. Crystallogr. 40,658–674.

28. Perrakis, A., Morris, R. & Lamzin, V. (1999). Auto-mated protein model building combined with itera-tive structure refinement. Nat. Struct. Biol. 6, 458–463.

29. Emsley, P. & Cowtan, K. (2004). Coot: model-buildingtools for molecular graphics. Acta Crystallogr., Sect. D:Biol. Crystallogr. 60, 2126–2132.

30. Afonine, P. V., Grosse-Kunstleve, R. W. & Adams,P. D. (2005). The Phenix refinement framework. CCP4Newsletter. Number 42, Contribution 8.

31. Laskowski, R. A., MacArthur, M. W., Moss, D. S. &Thornton, J. M. (1993). PROCHECK: a program tocheck the stereochemical quality of protein structures.J. Appl. Crystallogr. 26, 283–291.

32. Roessle, M. W., Klaering, R., Ristau, U., Robrahn, B.,Jahn, D., Gehrmann, T. et al. (2007). Upgrade of thesmall-angle X-ray scattering beamline X33 at theEuropean Molecular Biology Laboratory Hamburg.J. Appl. Crystallogr. 40, s190–s194.

33. Round, A. R., Franke, D., Moritz, S., Huchler, R.,Fritsche, M., Malthan, D. et al. (2007). Automatedsample-changing robot for solution scattering exper-iments at the EMBL Hamburg SAXS station X33.J. Appl. Crystallogr. 41, 913–917.

34. Konarev, P. V., Volkov, V. V., Sokolova, A. V., Koch,M. H. J. & Svergun, D. I. (2003). PRIMUS: a WindowsPC-based system for small-angle scattering dataanalysis. J. Appl. Crystallogr. 36, 1277–1282.

35. Svergun, D. I. (1992). Determination of the regulari-zation parameter in indirect-transform methods usingperceptual criteria. J. Appl. Crystallogr. 25, 495–503.

36. Svergun, D. I. (1999). Restoring low resolutionstructure of biological macromolecules from solutionscattering using simulated annealing. Biophys. J. 76,2879–2886.

37. Svergun, D. I., Petoukhov, M. V. & Koch, M. H. (2001).Determination of domain structure of proteins fromX-ray solution scattering. Biophys. J. 80, 2946–2953.

38. Svergun, D. I., Barberato, C. & Koch, M. H. J. (1995).CRYSOL—a program to evaluate X-ray solutionscattering of biological macromolecules from atomiccoordinates. J. Appl. Crystallogr. 28, 768–773.

39. Bernado, P.,Mylonas, E., Petoukhov,M.V., Blackledge,M. & Svergun, D. I. (2007). Structural characterizationof flexible proteins using small-angle X-ray scattering.J. Am. Chem. Soc. 129, 5656–5664.

40. Morris, G. M., Goodsell, D. S., Halliday, R. S., Huey,R., Hart, W. E., Belew, R. K. & Olson, A. J. (1998).Automated docking using a Lamarckian geneticalgorithm and empirical binding free energy function.J. Comput. Chem. 19, 1639–1662.

insights into the catalytic mechanism of 16s rrna...

Documents