doi:10.1016/j.jmb.2004.08.080 J. Mol. Biol. (2004) 343, 1195–1206

Principles of RNA Compaction: Insights from theEquilibrium Folding Pathway of the P4-P6 RNA Domainin Monovalent Cations

Keiji Takamoto1,2†, Rhiju Das3,4†, Qin He1,2, Sebastian Doniach3

Michael Brenowitz2,5, Daniel Herschlag4,6,7* and Mark R. Chance1,2,5*

1Department of Physiology andBiophysics, Albert EinsteinCollege of Medicine of YeshivaUniversity, New York, NY10461, USA

2Center for SynchrotronBioscience, Albert EinsteinCollege of Medicine of YeshivaUniversity, New York, NY10461, USA

3Department of PhysicsStanford University, StanfordCA 94305-4060, USA

4Department of BiochemistrySchool of Medicine, BeckmanCenter, Room B400, StanfordUniversity, Stanford, CA94305-5307, USA

5Department of BiochemistryAlbert Einstein College ofMedicine of Yeshiva UniversityNew York, NY 10461, USA

6Department of ChemistryStanford University, StanfordCA 94305, USA

7Biophysics Program, StanfordUniversity, Stanford, CA 94305USA

0022-2836/$ - see front matter q 2004 E

†K.T. & R.D. contributed equallyAbbreviations used: CE, 10 mM s

0.1 mM EDTA buffer; SAXS, small-aE-mail addresses of the correspon

herschla@cmgm.stanford.edu; mrc@

Counterions are required for RNA folding, and divalent metal ions such asMg2C are often critical. To dissect the role of counterions, we havecompared global and local folding of wild-type and mutant variants of P4-P6 RNA derived from the Tetrahymena group I ribozyme inmonovalent andin divalent metal ions. A remarkably simple picture of the foldingthermodynamics emerges. The equilibrium folding pathway in mono-valent ions displays two phases. In the first phase, RNA molecules that areinitially in an extended conformation enforced by charge–charge repulsionare relaxed by electrostatic screening to a state with increased flexibility butwithout formation of long-range tertiary contacts. At higher concentrationsof monovalent ions, a state that is nearly identical to the native folded statein the presence of Mg2C is formed, with tertiary contacts that involve baseand backbone interactions but without the subset of interactions thatinvolve specific divalent metal ion-binding sites. The folding modelderived from these and previous results provides a robust framework forunderstanding the equilibrium and kinetic folding of RNA.

q 2004 Elsevier Ltd. All rights reserved.

Keywords: RNA folding; electrostatic relaxation; tertiary interactionformation; compaction; P4–P6 domain

*Corresponding authors

Introduction

The discovery of catalysis by RNA in 1982 byCech and co-workers engendered excitement on

lsevier Ltd. All rights reserve

to this work.odium cacodylate,ngle X-ray scattering.ding authors:aecom.yu.edu

several fronts: an “RNA world” as a potentialsolution to the evolutionary problem of whethergenetic information (nucleic acids) or molecularfunction (proteins, according to the old view) camefirst;1 potential roles for RNA in modern-daybiology that extend beyond simply carrying infor-mation; and molecular questions of how a polymerfar simpler than proteins could fold into discretethree-dimensional structures and direct complexand highly efficient catalysis.2 Consideration of this

d.

1196 Principles of RNA Compaction

last problem led to an early focus on the roles of the2 0-hydroxyl group of RNA and of divalent metalions, both of which were believed to be required forRNA’s special functional properties. These viewshave matured over time.

There are many examples of RNA tertiarystructures with intimate roles for 2 0-hydroxylgroups, and, indeed, 2 0-hydroxyl groups havecatalytic roles.3,4 However, any absolute view ofthe necessity of the 2 0-hydroxyl group for catalysiswas vitiated by the discovery, through in vitroselection, of DNA catalysts,5 and these findingswere consistent with earlier work showing thatDNA of the corresponding sequence can adoptstructures similar to tRNA.6

Early work with catalytic RNAs demonstratedthat addition of divalent metal ions such as Mg2C

stimulated RNA structure formation and catalyticactivity.7 One extreme view was that all of RNAcatalysis could be reduced to the appropriateplacement of catalytic divalent metal ions.8 How-ever, Co(III) hexammine, an exchange-inert metalion, or high concentrations of monovalent cationswere shown to promote tertiary structure formationfor several RNAs,9–13 and even to promoteRNA catalysis in the absence of divalent metalions.9,11,14–16 In addition, group I ribozymes wereshown to form extensive tertiary contacts at highconcentrations of monovalent cations in the absenceof divalent cations.12,13,17 Nevertheless, Mg2C wasrequired to complete the folding of the group Icatalytic core required for activity.13

To distinguish the factors and forces that control

RNA folding, the NaC-promoted equilibrium fold-ing pathway of the P4-P6 domain of the Tetrahymenaintron has been dissected. The P4-P6 domain(Figure 1) was chosen for these studies, as itsthree-dimensional structure is known at atomicresolution with well-defined, specific Mg2C-bind-ing sites18,19 and its Mg2C-dependent folding hasbeen well-characterized.20–23 Yet, our initial studiesindicated that it collapsed upon addition of mono-valent ions alone. Further, this molecule is simpleenough, with two regions of long-range tertiarycontacts, to allow thorough testing of the roles ofthese interactions.We have coupled global and localprobes with mutagenesis to dissect the behavior ofthe P4-P6 domain in the presence of monovalentcations. The results, in conjunction with other recentstudies, allow us to draw general lessons about theforces that dictate the folding behavior of RNAmolecules.

Results

Preformed secondary structure in P4-P6

RNA ‘unfolded’ states often have stable second-ary structure, which simplifies study of foldingprocess. Dimethyl sulfate (DMS) and ribonucleaseT1 mapping of the isolated P4-P6 domain verifiedthe presence of intact secondary structure inconcentrations of NaC as low as 2 mM (data notshown; see also Jaeger et al.).24

Figure 1. The Tetrahymena ribo-zyme P4-P6 domain and mutantsexamined in this study. The base-pairing pattern present in theMg2C-induced native state isshown, and the two long-rangetertiary contacts are drawn asarrows. The boxes show themutant names used and the colorcoding used in subsequent Figures.The P5abc base-pairing pattern inthe absence of Mg2C is likely to beshifted in register by one nucleo-tide;47,49 this difference from thenative secondary structure doesnot affect the interpretations orconclusions of this study. Mutationof G174 to A, which favors the non-native secondary structure anddestabilizes Mg2C-induced fold-ing,49 has no detectable effect onthe NaC-induced compactionbehavior of P4-P6 (unpublishedresults).

Principles of RNA Compaction 1197

Two thermodynamic phases of NaC-inducedcompaction for P4-P6 RNA

The changes in size of the P4-P6 RNA as afunction of the concentration of sodium ions wereexamined using analytical ultracentrifugation(Figure 2(a)). The hydrodynamic radius, corre-sponding to the molecule’s compactness, decreasedin two major phases; the first phase occurs with amidpoint of 10–20 mM NaC,21 and the secondphase with a midpoint of w600 mM NaC. Tounderstand the forces driving these compactionevents, we compared the compaction behaviorof wild-type P4-P6 RNA to that of a seriesof mutants. Global information obtained byanalytical ultracentrifugation and small-angleX-ray scattering (SAXS) was combined withdetailed structural information from hydroxylradical footprinting to reveal the molecular basisof the compactions.

The first phase: electrostatic relaxation

Despite the reduction in hydrodynamic radius inthe first phase of compaction, there are minimalchanges in the hydroxyl radical reactivity of theRNA backbone (Figure 2(b)). Thus, the compactionoccurs without formation of stable tertiary contacts.To further test for potential roles of tertiary contactsin this phase of compaction, we monitored the sizeof a mutant form of P4-P6 lacking both long-rangetertiary interactions (Figure 1, Double mutant). Atransition coincident with that for wild-type RNAwas observed (Figure 2(a)), providing additionalevidence against the involvement of tertiary inter-actions in the first phase.

The hydrodynamic radius of 50 A estimated forthe starting state at a very low concentration of NaC

(!10 mM) is indicative of an extended RNAmolecule with the double helical elements splayedaway from one another (calculated value 53 A,versus 29 A for the folded molecule; see Materialsand Methods). This extended structure is consistentwith charge repulsion determining the orientationof the helices. The results described above suggeststrongly that the transition upon increasing theconcentration of NaC to 100 mM is a structuralrelaxation that is unaccompanied by formation oftertiary contacts. This relaxation is allowed becauseof enhanced screening of long-range electrostaticrepulsion between the RNA secondary structuralelements by the denser NaC counterion cloudpresent at higher concentrations of NaC.25 Figure2(c) shows the distribution of distances between allpairs of atoms within the molecule being probed,P(R), obtained from the SAXS experiments at100 mM NaC and from structural models. TheP4-P6 RNA in 100 mM NaC is less compact than itsnative state but considerably more compact thanpredicted for an extended state. The observedexperimental profile is the same as that predictedfor a relaxed ensemble of P4-P6 molecules in which

electrostatic repulsion is screened but no long-rangetertiary interaction is present (Figure 2(c)).Divalent cations also induce the relaxation. The

double mutant RNA examined at 1 mM Mg2C bySAXS has an Rg value and P(R) distributioncomparable to that observed at 100 mM NaC

(Figures 3(a) and 4(a)). Thus, the first phase ofcompaction reflects an electrostatic relaxationmediated by charge neutralization independent ofmetal ion valence.

The second phase: compaction induced bytertiary contact formation

At concentrations of NaC above 100 mM, furthercompaction beyond the relaxed state is observed(Figure 2(a)), accompanied by the appearance ofprotections from hydroxyl radical cleavage(Figure 2(b)). These observations indicate thatspecific interactions are formed within P4-P6 inthe presence of NaC. This compaction could bemediated through structure formation at the junc-tions between helices, formation of one or both ofthe long-range native tertiary contacts or the othernon-native tertiary contacts. The observation thatthe second phase of compaction is absent from amutant with both long-range tertiary contactsdisrupted strongly suggested that one or both ofthese contacts was responsible for the observedcompaction (Figure 2(a)).To examine the role of the individual tertiary

contacts, we investigated P4-P6 constructs withtertiary contacts disrupted separately (Figure 1).Disruption of the tetraloop or its receptor abolishedthe second phase, whereas disruption of the A-richbulge had no effect (Figure 4(a)). These resultssuggested a simple model in which NaC supportsformation of the tetraloop/tetraloop-receptor butnot the contact between the A-rich bulge, whichnormally binds Mg2C, and its docking site.The hydroxyl radical cleavage pattern for the

wild-type and mutant P4-P6 RNAs provided afurther test of the above model as well as adetailed structural comparison to the native state.Remarkably, hydroxyl radical protections for wild-type P4-P6 in high concentrations of NaC and inMg2C were indistinguishable throughout most ofits structure, including the “hinge” region betweenthe P4/P5/P6 coaxial stack and P5abc (residues126–128, 196–97 and 200–202) and thetetraloop/tetraloop-receptor (residues 153–155 and222–224; Figures 2(b), 4(b) and (c)). Thus, thestructure in the presence of a high concentrationof NaC appears to have the same overall fold as theMg2C state: parallel neighboring coaxial stacks ofP4/P5/P6 and P5abc juxtaposed by the hinge thatcovalently connects them and the tetraloop/tetra-loop-receptor that non-covalently joins these stacksat their distal ends.Mutation of the tetraloop/tetraloop-receptor

prevented the second phase of compaction, asnoted above, and abolished protection fromhydroxyl radicals (Figure 4(c) versus (e); see also

Figure 2. Compaction andtertiary structure formation inthe P4-P6 RNA as a function ofthe concentration of NaC. (a) P4-P6 compaction, monitored byanalytical centrifugation. Thehydrodynamic radius for thewild-type P4-P6 (black) is plottedand reveals two thermodynamicphases of compaction, with mid-points ofw20 mM andw600 mM.The first compaction phase ispresent in the double mutant(purple). The lines are shown asguides; note the logarithmic scalefor the concentration of NaC. Theexperimental errors are smallerthan the heights of the symbols.(b) Solvent exposure throughoutthe RNA backbone probed byhydroxyl radical footprinting.Footprinting patterns of the P4-P6 native state in 1 and 10 mMMg2C are shown for comparisonas separate bars at the right. The10 mM NaC state is the referencecondition (color white); the NaC

folding process shows a protec-tion pattern at high concentrationsof monovalent ion very similar tothat of the native state. Linearinterpolation is used to smooththe data between the experimen-tally measured concentrations ofNaC to allow better recognition ofthe transitions in this false-colorrepresentation. The measuredconcentrations are indicated atthe bottom of the map. The NaC

concentration scale matches thatin (a) to facilitate comparisons. (c)Small-angle X-ray scatteringmeasurements exhibit indistin-guishable distributions P(R) ofintramolecular distances R at100 mM NaC for wild-type P4-P6(black) and variants withmutations disrupting the L5b tet-raloop (L5b/UUCG; red), theJ6a/b tetraloop receptor (J6a/b-U; cyan), the A-rich-bulge (A-Bulge/U; green), and both thetetraloop and A-rich-bulge(double mutant; purple). Notethat overlap obscures most of thedata points. The P(R) distributionmeasured at 100 mM NaC isconsistent with a relaxedensemble of P4-P6 conformations(continuous line), but not a highlyextended state (dashed line) or thenative state (dotted line).

1198 Principles of RNA Compaction

Figure 3. Intermolecular distancedistributions of the P4-P6 RNAstates discussed in the main text.Symbols show experimental distri-butions overlaid on predictions forextended, relaxed, and folded(native) states. (a) The doublemutant (A-rich/U; L5b/UUCG) isrelaxed but not folded in 1 mMMg2C. (b) The wild-type RNA inhighNaC is compact but not native.(c) The wild-type RNA in 1 mMMg2C forms a conformation whosedistance distribution is consistentwith the native structure, asexpected. The experimental errorsare smaller than the heights of thesymbols.

Principles of RNA Compaction 1199

Supplementary Material, Figure S1). This result pro-vides additional evidence that the tetraloop/ tetra-loop-receptor interaction is formed and is requiredto stabilize the folded state in NaC. Conversely,

mutation of the A-rich bulge, which had no effect onNaC-induced compaction, also had no effect on theprotection pattern (Figure 4(c) versus (d)).The above results are consistent with the

Figure 4. Dissecting the secondphase of NaC-induced compactionof the P4-P6 domain. (a) SAXSmeasurements of the radius ofgyration for the wild-type (black),L5b/UUCG (red), J6a/b-U (cyan),A-Bulge/U (green), and doublemutant (purple) molecules.Dashed lines represent the radiusof gyration values of wild-type P4-P6 (black) and the double mutant(purple) in 1 mM Mg2C. Radius ofgyration values in 1 mM Mg2C forthe L5b/UUCG (41 A), J6a/b-U(39 A), and A-Bulge/U (41 A)molecules were within error of thedouble mutant value (39 A). (b)–(e)Hydroxyl radical protection pat-tern of wild-type P4-P6 in (b)10 mM Mg2C, (c) wild-type in 2 MNaC, (d) A-Bulge/Umutant in 2 MNaC, and (e) L5b/UUCG mutantin 2 M NaC. Protections are rela-tive to reactivity in 10 mM sodiumcacodylate buffer in the absence ofMg2C. For visual comparison, theprotection patterns for the statesare all displayed as color variationson the same backbone represen-tation of the P4-P6 native fold inMg2C, although conformationaldifferences are indicated by thedifferent protection patterns. Thebright or faint gold lines representthe interpreted presence orabsence of the tertiary contacts,respectively.

1200 Principles of RNA Compaction

involvement of Mg2C in establishing the structureof the A-rich bulge in the native state andcontributing to folding in Mg2C but not in NaC.As expected, protections surrounding the Mg2C

core and in its P4 docking site were diminishedgreatly in the presence of NaC compared to Mg2C

(residues 162–165, 169–170, 176–177, 211–214 inFigures 2(b), 4(b) and (c)). The protection pattern forthis high-NaC, near-native state indicates a tertiarystructure similar to that of the A-bulge/U mutantfolded either by NaC or Mg2C (see also Supplemen-tary Material, Figure S2).

Principles of RNA Compaction 1201

In summary, high concentrations of NaC alonesupport folding to a P4-P6 structure that is the samein nearly all respects as the native fold promoted byMg2C. Structural elements at the ends of thecoaxially stacked structural elements, the “hinge”region and the tetraloop/tetraloop-receptor, form inNaC, whereas interactions mediated by the A-richbulge that is directly involved in Mg2C-binding arenot present.

Discussion

Mg2C and RNA have long enjoyed a specialrelationship. Mg2C is ubiquitous in physiologicalenvironments, and there are clear roles for Mg2C inRNA structure and function. Nevertheless, it hasbecome increasingly clear that RNA can adoptnative structure and carry out function in decidedlynon-physiological ionic conditions: at high concen-trations of monovalent ions in the absence ofMg2C.9–15 Fundamental insights can be achievedby determining the effects of perturbations awayfrom physiological conditions and dissecting theunderlying causes of these effects. Here, we havedissected the equilibrium folding pathway for theP4-P6 RNA domain from the Tetrahymena group Iribozyme in monovalent cations using global andlocal structural probes in conjunction with targetedmutagenesis (Figure 5). Comparing these resultswith published studies of the structure and foldingof this RNA in the presence of Mg2C allows anincisive evaluation of the forces and featuresproposed to contribute to RNA folding.

RNA “collapse” preceding tertiary structureformation

Secondary structure formation

A simplifying feature of RNA folding, in

Figure 5. Proposed structural ensembles of the TetrahymenaNaC.

comparison to protein folding, is the ability toform “unfolded” states with stable secondarystructure and to then follow the adoption ofcompact tertiary structures by these preformedhelices. Compact intermediates have been observedin the extensive studies of the kinetic and thermo-dynamic folding pathways of several RNAs.17,26–29,57

For the RNase P RNA, Azoarcus and bI5 group Iintrons, the observed thermodynamic compactionwith increasing concentration of Mg2C appears toinvolve, at least in part, secondary structureformation beyond the extensive local duplexformation present in the starting state.28,30,31 Incontrast, DMS and RNase T1 mapping of the P4-P6domain in the presence of sodium ion concen-trations as low as 2 mM suggest that the secondarystructure is preformed, simplifying evaluation ofother compaction forces.In addition to inducing a net compaction in the

RNA, secondary structure formation greatly con-strains the RNA conformational ensemble, andthese constraints generally favor formation ofnative tertiary interactions and disfavor formationof many potential non-native secondary and ter-tiary interactions.32 Although this contribution tofolding was not probed directly in this study, thepreformed secondary structure is undoubtedly acritical feature in favoring subsequent P4-P6 RNAtertiary structure formation.

Electrostatic relaxation

Our equilibrium study, like much of the literatureprobing the kinetics and thermodynamics of metalion-dependent RNA folding, used an initial statewith very little salt present. For P4-P6 and thecomplete Tetrahymena ribozyme, electrostaticrepulsion dominates the interactions between thepreformed secondary structural elements at lowionic strength, forcing them into an extendedstructure (Figure 5(a)).12,33 Upon the addition of

ribozyme P4-P6 domain with increasing concentrations of

1202 Principles of RNA Compaction

cations sufficient to neutralize the charge repulsion,the RNA relaxes into a disordered family ofstructures in which the helices sample alternativeconformations that allow tertiary elements toencounter each other (Figure 5(b); see alsorefs33,51,52 ). This sampling yields a net observed“collapse” as the average molecular radius isdecreased. However, the number of accessibleconformations is predicted to increase via thisrelaxation process, and this prediction may betestable by thermodynamic and biophysical probesof the conformational flexibility of RNA.53–55

Electrostatic relaxation is an overall property ofthe P4-P6 structure in which the rigid chargedhelices are interconnected by multiple unpairedjunctions, and this relaxation is likely to be a generalproperty of the kinetics and thermodynamics ofnucleic acids. Indeed, we expect that unstructuredstates in vivo will be electrostatically relaxed ratherthan dominated by electrostatic repulsion asobserved in the low-salt conditions used in manyin vitro folding experiments.

Counterion-induced attraction

The theories of the counterion atmospheresurrounding nucleic acids provide possiblemechanisms of compaction beyond electrostaticrelaxation.34,35 Simulations of the ion atmosphereindicate the existence of a counterion-mediatedattractive force between nucleic acid elements andsuggest strongly that only counterions of highvalence can establish such a force.35 The observationof compaction of the P4-P6 RNA in the presence ofbothNaC andMg2C (Figures 2(a) and 3) provides noindication of a valence-dependent counterion corre-lation force playing a major role in RNA folding.33

Tertiary structure formation

Long-range tertiary contacts and specific metalion-binding sites

The group I introns from Tetrahymena andAzoarcus can fold into compact, near-native three-dimensional structures in the presence of highconcentrations of monovalent ions, although theaddition of Mg2C is required for proper folding ofthe conserved intron catalytic core.12,13 From thesestudies it remained unclear whether NaC alonecould induce significant folding of the isolated P4-P6 domain. In particular, the Mg2C cluster at thecore of the P4-P6 RNA19 was expected to becomemore important and possibly critical for thedomain’s folding outside the context of the multipleother ribozyme elements that encapsulate andstabilize the domain.56 We observe here a compac-tion of the isolated P4-P6 to nearly native dimen-sions upon addition of high concentrations of NaC,demonstrating that its metal ion core is not essentialfor formation of a compact fold.

The structure of the folded P4-P6 RNA in thepresence of a high concentration of NaC can be

understood from the simplest perspective of thepotential effects of monovalent cations. Moregenerally, all RNAs will experience enormousinternal electrostatic repulsion between backbonephosphate groups. As described above, moderateconcentrations of NaC can provide the non-specificelectrostatic screening required for relaxation;higher concentrations ofNaC can then screen chargerepulsion of phosphoryl oxygen atoms at the closeseparations observed in the native fold. Withsufficient screening, structural interactions that donot rely on binding of divalent metal ions in specificsites can be formed to give the specific tertiary fold.Although folding of the special tertiary motifs thatrequire Mg2C coordination is not required to form acompact fold, such metal ion-binding sites canprovide added stability to folded structures whendivalentmetal ions arepresent.Monovalent ions canalso stabilize the fold due to the presence of specificbinding sites, such as the one identified adjacent tothe tetraloop receptor of P4-P6.36,37

Local structure at junctions

The importance of structure formation at junc-tions between double helices for RNA folding hasbeen increasingly recognized.38,39 For the P4-P6RNA, mutants with changes within the junctionconnecting the P4/P5/P6 stack to the P5abc stack(Figure 1) require higher concentrations of Mg2C tofold.40 This junction also exhibits protections fromhydroxyl radical cleavage in the folded state,20 andthese protections are maintained in the NaC-foldedstate (Figures 2(b), and 4(b) and (c)).

Beyond their thermodynamic importance instabilizing RNA folds, junctions may play keyroles in folding kinetics.38 In the case of the P4-P6RNA, formation of specific tertiary contacts occurswith an observed rate constant of 1–2 sK1 whenMg2C-dependent folding is initiated from a low-saltcondition (!10 mM); increasing the initial concen-tration of monovalent ions to 20–30 mM increasesthe rate of folding at least 25-fold.21,22,41 Thethermodynamic transition from an extended stateto a relaxed state with a NaC midpoint of w10 mMcould account for this dramatic acceleration, withrelaxation of the junctions in the extended stateproviding one of the kinetic barriers to folding atlow concentrations of salt (Figure 5).

The slow folding of junctions may arise due tostructure formation that needs to be disrupted. Thelow-salt P4-P6 structure exhibits no change in thebackbone protection pattern of the extended andrelaxed states (Figure 2(b)) and may involvestacking of bases at the junctions. A similar kineticbarrier arising from junction structure has beenproposed for electrostatic relaxation in the full-length ribozyme based on time-resolved SAXSstudies.33 More generally, structures formed atjunctions can aid or hinder folding kinetics andthermodynamics, despite the typical focus in fold-ing on the residues that make specific long-rangeinteractions.

Principles of RNA Compaction 1203

A general picture for RNA folding: a balance offorces

The forces and features described in the preced-ing sections balance to provide the thermodynamicand kinetic pathways observed for RNA folding.Positioning via secondary structure formation,screening of electrostatic repulsion, long-rangetertiary interactions, junction structures and specificmetal ion-binding sites all play important roles inthe formation of a compact, stable P4-P6 RNA.Their balance leads to cooperative folding from anextended, unfolded state to the native state withincreasing concentrations of Mg2C. In the absenceof Mg2C, binding sites for divalent metal ions nolonger play a role, and the balance of the forcesshifts, revealing two additional states of P4-P6(Figure 5). As the concentration of NaC is increased,the RNA first compacts to an electrostaticallyrelaxed state and then adopts a near-native tertiarystructure lacking the metal ion core but stabilizedby Mg2C-independent tertiary interactions. Thesestates are normally hidden in the Mg2C equilibriumfolding pathway, but have been revealed in thepresence of Mg2C through the appropriatemutations of metal ion-binding sites and long-range contacts.

The ability to fold RNAs in the presence ofmonovalent ions provides an analytical tool thatwill aid in developing a complete quantitativeenergetic description of the forces that dictateRNA structure and dynamics. Such a descriptionis foundational, in turn, for developing a deepunderstanding of the thermodynamic and kineticmechanisms underlying the function of RNAmolecules involved in complex biologicalprocesses.

Materials and Methods

RNA preparation

Wild-type and mutant P4-P6 RNA constructs wereprepared by in vitro transcription from PCR-generatedDNA templates using phage T7 RNA polymerase, labeledat the 5 0-end with [g-32P]ATP using phage T4 polynucleo-tide kinase for footprinting experiments, and purified bydenaturing polyacrylamide gel electrophoresis.

Analytical ultracentrifugation

The RNA stock solutions in CE buffer (10 mM sodiumcacodylate, 0.1 mM EDTA, pH 7.0) were diluted into CEbuffer containing the indicated concentration of NaCl,heated to 95 8C for one minute, cooled on the bench for 30seconds and then incubated at 50 8C for 15–30 minutesprior to loading the centrifuge cells. The loaded cellsand rotor were allowed to temperature-equilibrate forw30 minutes in the centrifuge prior to initiating a run.Sedimentation velocity analyses were conducted usingthe absorption optics of a Beckman XL-I analyticalultracentrifuge at 25 8C, 6 mg/ml of RNA in double-sectorcells loaded into a Ti-60 rotor at a rotor speed of40,000 rpm. The sedimentation boundaries were analyzed

as described, assuming �nZ0:53 cm3=g.12,21 The sedimen-tation and diffusion coefficients decrease with increasingconcentration of NaC, consistent with compaction of theRNA. The presence of 0.1 mM EDTA in the ultracentri-fuge experiments ensures that micromolar multivalentmetal ion contaminants potentially associated with NaClstocks are not responsible for the observed transitions.

Hydroxyl radical footprinting

Radiolabeled RNA samples were denatured at 95 8C forone minute in CE buffer and then incubated at variousconcentrations of salt at 25 8C for 30 minutes. Fentonchemistry was carried out for 30 minutes at 25 8C withfinal concentrations of 2 mM Fe(NH4)2(SO4)2, 2.5 mMEDTA, and 6 mM sodium ascorbate without H2O2. Thereaction products were processed, separated by gelelectrophoresis and visualized using phosphor storageimaging.12,23 The resulting monovalent cation titrationautoradiograms were analyzed with single-nucleotideresolution.12,50 The reactivity data are presented as a false-color image following scaling to reference state, asfollows:12

Relative protectionZ 1KR

Rref

where R is the observed intensity and Rref is theintensity for the initial state in CE buffer. Footprinting ofthe low-sodium and high-sodium states of the wild-typeand A-Bulge/U constructs at independent laboratorieswith independently purified RNA led to indistinguish-able results. As in the ultracentrifuge experiments, thepresence of excess EDTA would serve to scavengemicromolar multivalent metal ion contaminants thatmight be present in the NaCl stocks.

Small-angle X-ray scattering (SAXS)

SAXS samples contained RNA concentrations of 1 or3 mg/ml of RNA or no RNA (for background measure-ments) with 50 mM sodium 3-(N-morpholino) propane-sulfonic acid, pH 7.0 (20 mM NaC) and variousconcentrations of sodium acetate. The reported SAXSprofiles exhibited no dependence on RNA concentration,indicating that aggregation is negligible. While no EDTAwas present in the SAXS experiments, the high concen-tration of phosphate (>2 mM) of negatively charged RNAensures that the micromolar levels of multivalent ioncontaminants that could be present in the sodium acetatewould be insufficient to influence a majority of the RNAmolecules present; the agreement in the dependence ofthe ultracentrifuge results on the concentration of NaC,for which EDTA was present, and the SAXS resultsprovides further strong evidence against an effect fromcontaminating multivalent metal ions. Charged couplingdevice (CCD) images were obtained at beamline 15-A atthe Photon Factory and beamline 12-ID at the AdvancedPhoton Source and processed into SAXS profiles usingstandard software packages.42,43 Complete NaC titrationsof the wild-type and A-Bulge/U constructs were carriedout at both beamlines with independently prepared RNAsamples and agreed within measurement error. Distancedistributions and radius-of-gyration (Rg) values wereobtained from SAXS profiles by Svergun’s regularizedtransformation.44 The resulting Rg values wereindistinguishable from those obtained by Guinieranalysis. SAXS profiles at low concentrations of NaC

(!100 mM) could not be used to obtain radii of gyration

1204 Principles of RNA Compaction

or intramolecular distance distributions because of anRNA concentration-dependent scattering signal fromintermolecular ordering. In fact, this ordering is aqualitative indicator of strong intermolecular Coulombicrepulsion, consistent with strong intramolecular repul-sion proposed for the RNA extended state under theseconditions.45 Under all other conditions, the observedscattering was independent of RNA concentration.Theoretical SAXS distributions and radius of gyration

(Rg) values for different RNA conformations, taking intoaccount ion atmosphere scattering effects,46 were simu-lated in Matlab routines. The native state structure wastaken from PDB coordinates 1GID,18 with RgZ29 A. Theextended state was taken to be a conformation with theorientation of P5abc adjusted to be at a maximal distancefrom the stacked P4/P5/P6 helices, corresponding to amaximum radius of gyration of 53 A. The relaxedensemble was simulated by uniformly sampling orien-tations of the P5abc subdomain relative to the P4/P5/P6helices, yielding an intermediate radius of gyration value,RgZ43 A. Shifting the base-pairing pattern of P5abc47 orintroducing flexibility at the P5abc and P4/P6 junctionsproduced no distinguishable changes in the predicteddistribution for the relaxed state at the coarse resolutionsampled by SAXS.

DMS and RNase T1 structure mapping

Dimethyl sulfate (DMS) footprinting was carried out asdescribed.17 For RNase T1 structure mapping, radio-labeled RNA was incubated with 2.5 nM RNase T1(Roche) for 20 minutes at 25 8C in buffer conditionsidentical with those used for X-ray scattering or underdenaturing conditions,48 and analyzed as in hydroxylradical footprinting reactions.

Acknowledgements

We thank K. Travers for insightful and helpfuldiscussion; K. Ito for expert user support and foruse of unpublished CCD software at the PhotonFactory through an agreement with the StanfordSynchrotron Radiation Laboratory (supported byNIH and DOE); and S. Seifert at the AdvancedPhoton Source (DOE contract no. W-31-109-Eng-38).This work was supported by NIH grants R01-GM-52348, P01-GM-66275 and an NSF fellowship (toR.D.).

Supplementary data

Supplementary data associated with thisarticle can be found, in the online version, atdoi:10.1016/j.jmb.2004.08.080

References

1. Joyce, G. F. & Orgel, L. E. (1993). Prospects forunderstanding the origin of the RNA world. In TheRNA World (Atkins, J. F., ed), Cold Spring HarborLaboratory Press, Cold Spring Harbor, NY, pp. 1–25.

2. Brion, P. & Westhof, E. (1997). Hierarchy anddynamics of RNA folding. Annu. Rev. Biophys. Biomol.Struct. 26, 113–137.

3. Yoshida, A., Shan, S., Herschlag, D. & Piccirilli, J. A.(2000). The role of the cleavage site 2 0-hydroxyl in theTetrahymena group I ribozyme reaction. Chem. Biol. 7,85–96.

4. Rupert, P. B. & Ferre-D’Amare, A. R. (2001).Crystal structure of a hairpin ribozyme–inhibitorcomplex with implications for catalysis. Nature,410, 780–786.

5. Li, Y. & Breaker, R. R. (1999). Deoxyribozymes: newplayers in the ancient game of biocatalysis. Curr. Opin.Struct. Biol. 9, 315–323.

6. Paquette, J., Nicoghosian, K., Qi, G. R., Beauchemin,N. & Cedergren, R. (1990). The conformation ofsingle-stranded nucleic acids tDNA versus tRNA.Eur. J. Biochem. 189, 259–265.

7. Pan, T., Long, D. M. & Uhlenbeck, O. C. (1993).Divalent metal ions in RNA dolding and catalysis. InThe RNAWorld (Atkins, J. F., ed), Cold Spring HarborLaboratory Press, Cold Spring Harbor, NY, pp. 271–302.

8. Yarus, M. (1993). Howmany catalytic RNAs? Ions andthe Cheshire cat conjecture. FASEB J. 7, 31–39.

9. Murray, J. B., Seyhan, A. A., Walter, N. G., Burke, J. M.& Scott, W. G. (1998). The hammerhead, hairpin andVS ribozymes are catalytically proficient in mono-valent cations alone. Chem. Biol. 5, 587–595.

10. Shelton, V. M., Sosnick, T. R. & Pan, T. (2001). Alteringthe intermediate in the equilibrium folding ofunmodified yeast tRNAPhe with monovalent anddivalent cations. Biochemistry, 40, 3629–3638.

11. O’Rear, J. L., Wang, S., Feig, A. L., Beigelman, L.,Uhlenbeck, O. C. & Herschlag, D. (2001). Com-parison of the hammerhead cleavage reactionsstimulated by monovalent and divalent cations.RNA, 7, 537–545.

12. Takamoto, K., He, Q., Morris, S., Chance, M. R. &Brenowitz, M. (2002). Monovalent cations mediateformation of native tertiary structure of theTetrahymena thermophila ribozyme. Nature Struct. Biol.9, 928–933.

13. Rangan, P. & Woodson, S. A. (2003). Structuralrequirement for Mg2C binding in the group I introncore. J. Mol. Biol. 329, 229–238.

14. Nesbitt, S., Hegg, L. A. & Fedor, M. J. (1997). Anunusual pH-independent and metal-ion-independentmechanism for hairpin ribozyme catalysis. Chem. Biol.4, 619–630.

15. Suga, H., Cowan, J. A. & Szostak, J. W. (1998).Unusual metal ion catalysis in an acyl-transferaseribozyme. Biochemistry, 37, 10118–10125.

16. Hertel, K. J., Peracchi, A., Uhlenbeck, O. C. &Herschlag, D. (1997). Use of intrinsic binding energyfor catalysis by an RNA enzyme. Proc. Natl Acad. Sci.USA, 94, 8497–8502.

17. Russell, R., Zhuang, X., Babcock, H. P., Millett, I. S.,Doniach, S., Chu, S. & Herschlag, D. (2002). Exploringthe folding landscape of a structured RNA. Proc. NatlAcad. Sci. USA, 99, 155–160.

18. Cate, J. H., Gooding, A. R., Podell, E., Zhou, K.,Golden, B. L., Kundrot, C. E. et al. (1996). Crystalstructure of a group I ribozyme domain: principles ofRNA packing. Science, 273, 1678–1685.

19. Cate, J. H., Hanna, R. L. & Doudna, J. A. (1997). Amagnesium ion core at the heart of a ribozymedomain. Nature Struct. Biol. 4, 553–558.

20. Murphy, F. L. & Cech, T. R. (1993). An

Principles of RNA Compaction 1205

independently folding domain of RNA tertiarystructure within the Tetrahymena ribozyme. Bio-chemistry, 32, 5291–5300.

21. Deras, M. L., Brenowitz, M., Ralston, C. Y., Chance,M. R. & Woodson, S. A. (2000). Folding mechanism ofthe Tetrahymena ribozyme P4-P6 domain. Biochemistry,39, 10975–11085.

22. Silverman, S. K., Deras, M. L., Woodson, S. A.,Scaringe, S. A. & Cech, T. R. (2000). Multiple foldingpathways for the P4-P6 RNA domain. Biochemistry, 39,12465–12475.

23. Uchida, T., He, Q., Ralston, C. Y., Brenowitz, M. &Chance, M. R. (2002). Linkage of monovalent anddivalent ion binding in the folding of the P4-P6domain of the Tetrahymena thermophila ribozyme.Biochemistry, 41, 5799–5806.

24. Jaeger, J. A., Zuker, M. & Turner, D. H. (1990). Meltingand chemical modification of a cyclized self-splicinggroup I intron: similarity of structures in 1 M NaC, in10 mM Mg2C, and in the presence of substrate.Biochemistry, 29, 10147–10158.

25. Draper, D. E. (2004). A guide to ions and RNAstructure. RNA, 10, 335–343.

26. Fang, X. W., Thiyagarajan, P., Sosnick, T. R. & Pan, T.(2002). The rate-limiting step in the folding of a largeribozyme without kinetic traps. Proc. Natl Acad. Sci.USA, 99, 8518–8523.

27. Rangan, P., Masquida, B., Westhof, E. & Woodson,S. A. (2003). Assembly of core helices and rapidtertiary folding of a small bacterial group I ribozyme.Proc. Natl Acad. Sci. USA, 100, 1574–1579.

28. Perez-Salas, U. A., Rangan, P., Krueger, S., Briber,R. M., Thirumalai, D. & Woodson, S. A. (2004).Compaction of a bacterial group I ribozyme coincideswith the assembly of core helices. Biochemistry, 43,1746–1753.

29. Buchmueller, K. L. &Weeks, K. M. (2003). Near nativestructure in an RNA collapsed state. Biochemistry, 42,13869–13878.

30. Fang, X., Pan, T. & Sosnick, T. R. (1999). A thermo-dynamic framework and cooperativity in the tertiaryfolding of a Mg2C-dependent ribozyme. Biochemistry,38, 16840–16846.

31. Chamberlin, S. I. & Weeks, K. M. (2003). Differentialhelix stabilities and sites pre-organized for tertiaryinteractions revealed by monitoring local nucleotideflexibility in the bI5 group I intron RNA. Biochemistry,42, 901–909.

32. Narlikar, G. J., Bartley, L. E. & Herschlag, D.(2000). Use of duplex rigidity for stability andspecificity in RNA tertiary structure. Biochemistry,39, 6183–6189.

33. Das, R., Kwok, L.W.,Millett, I., Bai, Y.,Mills, T. T., Jacob,J. et al. (2003). The fastest global events in RNAfolding: electrostatic relaxation and tertiary col-lapse of the Tetrahymena ribozyme. J. Mol. Biol. 332,311–319.

34. Murthy, V. L. & Rose, G. D. (2000). Is counteriondelocalization responsible for collapse in RNA fold-ing? Biochemistry, 39, 14365–14370.

35. Bloomfield, V. A. (1997). DNA condensation bymultivalent cations. Biopolymers, 44, 269–282.

36. Basu, S., Rambo, R. P., Strauss-Soukup, J., Cate, J. H.,Ferre-D’Amare, A. R., Strobel, S. A. & Doudna, J. A.(1998). A specific monovalent metal ion integral to theAA platform of the RNA tetraloop receptor. NatureStruct. Biol. 5, 986–992.

37. Conn, G. L., Gittis, A. G., Lattman, E. E., Misra, V. K. &

Draper, D. E. (2002). A compact RNA tertiarystructure contains a buried backbone–KC complex.J. Mol. Biol. 318, 963–973.

38. Hohng, S., Wilson, T. J., Tan, E., Clegg, R. M.,Lilley, D. M. & Ha, T. (2004). Conformationalflexibility of four-way junctions in RNA. J. Mol.Biol. 336, 69–79.

39. Lilley, D. M. (2000). Structures of helical junctions innucleic acids. Quart. Rev. Biophys. 33, 109–159.

40. Szewczak, A. A. & Cech, T. R. (1997). An RNA internalloop acts as a hinge to facilitate ribozyme folding andcatalysis. RNA, 3, 838–849.

41. Uchida, T., Takamoto, K., He, Q., Chance, M. R. &Brenowitz, M. (2003). Multiple monovalent ion-dependent pathways for the folding of the L-21Tetrahymena thermophila ribozyme. J. Mol. Biol. 328,463–478.

42. Seifert, S., Winans, R. E., Tiede, D. M. & Thiyagarajan,P. (2000). Design and performance of a ASAXSinstrument at the advanced photon source. J. Appl.Crystallog. 33, 782–784.

43. Amemiya, Y., Wakabayashi, K., Hamanaka, T.,Wakabayashi, T., Matsushita, T. & Hashizume, H.(1983). Design of a small-angle diffractometer usingsynchrotron radiation at the photon factory. Nucl.Instrum. Methods, 208, 471–477.

44. Svergun, D. I. (1992). Determination of the regu-larization parameter in indirect-transform methodsusing perceptual criteria. J. Appl. Crystallog. 25,495–503.

45. Skibinska, L., Gapinski, J., Liu, H., Patkowski, A.,Fischer, E.W. & Pecora, R. (1999). Effect of electrostaticinteractions on the structure and dynamics of a modelpolyelectrolyte. II. Intermolecular correlations.J. Chem. Phys. 110, 1794–1800.

46. Das, R., Mills, T. T., Kwok, L. W., Maskel, G. S., Millett,I. S., Doniach, S. et al. (2003). Counterion distributionaround DNA probed by solution X-ray scattering.Phys. Rev. Letters, 90, 188103.

47. Wu, M. & Tinoco, I. J. (1999). RNA folding causessecondary structure rearrangement. RNA, 5,1665–1674.

48. Donis-Keller, H., Maxam, A. M. & Gilbert, W. (1977).Mapping adenines, guanines, and pyrimidines inRNA. Nucl. Acids Res. 4, 2527–2538.

49. Silverman, S. K., Zheng, M., Wu, M., Tinoco, I., Jr &Cech, T. R. (1999). Quantifying the energetic interplayof RNA tertiary and secondary structure interactions.RNA, 5, 1665–1674.

50. Takamoto, K., Chance, M. R. & Brenowitz, M. (2004).Semi-automated, single-band peak-fitting analysis ofhydroxyl radical nucleic acid footprint autoradio-grams for the quantitative analysis of transitions.Nucl. Acids Res. 32, E119.

51. Heilman-Miller, S. L., Thirumalai, D. &Woodson, S. A.(2001). Role of counterion condensation in folding ofthe Tetrahymena ribozyme. I. Equilibrium stabilizationby cations. J. Mol. Biol. 306, 1157–1166.

52. Lee, N. & Thirumalai, D. (2000). Dynamics of collapseof flexible polyampholytes. J. Chem. Phys. 113, 5126.

53. Mikulecky, P. J. & Feig, A. L. (2004). Heat capacitychanges in RNA folding: application of perturbationtheory to hammerhead ribozyme cold denaturation.Nucl. Acids Res. 32, 3967–3976.

54. Rueda, D., Wick, K., McDowell, S. E. & Walter, N. G.(2003). Diffusely bound Mg2+ ions slightly reorientstems I and II of the hammerhead ribozyme toincrease the probability of formation of the catalyticcore. Biochemistry, 42, 9924–9936.

1206 Principles of RNA Compaction

55. Friederich, M.W., Vacano, E. & Hagerman, P. J. (1998).Global flexibility of tertiary structure in RNA: yeasttRNAPhe as a model system. Proc. Natl. Acad. Sci. USA.95, 3572–3577.

56. Lehnert, V., Jaeger, L., Michel, F. & Westhof, E. (1996).New loop–loop tertiary interactions in self-splicing

introns of subgroup IC and ID: a complete 3D modelof the Tetrahymena thermophila ribozyme. Chem. Biol. 3,993–1009.

57. Russell, R., Millett, I. S., Doniach, S. & Herschlag, D.(2000). Small angle X-ray scattering reveals a compactintermediate inRNAfolding.Nat. Struct. Biol.7, 367–370.

Edited by J. Doudna

(Received 15 July 2004; received in revised form 24 August 2004; accepted 26 August 2004)