Pre-Steady-State and Stopped-Flow Fluorescence Analysis Of

doi:10.1006/jmbi.2001.5413availableonlineathttp://www.idealibrary.comon J. Mol. Biol. (2002) 317, 21±40

Pre-steady-state and Stopped-flow Fluorescence Analysis of Escherichia coli Ribonuclease III: Insights into Mechanism and Conformational Changes Associated With Binding and Catalysis

FrankE.CampbellJr1{,AdamG.Cassano1{,VernonE.Anderson2 andMichaelE.Harris1*

1Center for RNA Molecular To better understand substrate recognition and catalysis by RNase III, we Biology and examined steady-state and pre-steady-state reaction kinetics, and changes in intrinsic enzyme ¯uorescence. The multiple turnover cleavage of a 2Department of Biochemistry model RNA substrate shows a pre-steady-state burst of product for- Case Western Reserve mation followed by a slower phase, indicating that the steady-state reac- University School of Medicine tion rate is not limited by substrate cleavage. RNase III catalyzed Cleveland, OH 44106, USA hydrolysis is slower at low pH, permitting the use of pre-steady-state kinetics to measure the dissociation constant for formation of the

enzyme-substrate complex (Kd 5.4(Æ0.6) nM), and the rate constant for 1 phosphodiester bond cleavage (kc 1.160(Æ0.001) min , pH 5.4). Isotope incorporation analysis shows that a single solvent oxygen atom is incorporated into the 50 phosphate of the RNA product, which demonstrates that the cleavage step is irreversible. Analysis of the pH dependence of

the single turnover rate constant, kc, ®ts best to a model for two or more titratable groups with pKa of ca 5.6, suggesting a role for conserved acidic residues in catalysis. Additionally, we ®nd that kc is dependent on the pKa value of the hydrated divalent metal ion included in the reaction, providing evidence for participation of a metal ion hydroxide in catalysis, potentially in developing the nucleophile for the hydrolysis reaction. In order to assess whether conformational changes also contribute to the enzyme mechanism, we monitored intrinsic tryptophan ¯uorescence. During a single round of binding and cleavage by the enzyme we detect a biphasic change in ¯uorescence. The rate of the initial increase in ¯uorescence was dependent on substrate concentration yielding a second- order rate constant of 1.0(Æ0.1) Â 108 M1 s1, while the rate constant of the second phase was concentration independent (6.4(Æ0.8) s1; pH 7.3). These data, together with the unique dependence of each phase on divalent metal ion identity and pH, support the hypothesis that the two ¯uorescence transitions, which we attribute to conformational changes, correlate with substrate binding and catalysis. # 2002 Elsevier Science Ltd. Keywords: ribonuclease III; endonuclease; phosphodiesterase; enzyme *Corresponding author kinetics; stopped-¯ow ¯uorescence

Introduction

The enzyme ribonuclease III (EC 3.1.26.3) was the ®rst speci®c ribonuclease to be discovered and 1 {These two authors contributed equally to this work. puri®ed. Althoughatthetimetheenzyme'srole Abbreviations used: dsRBD, double-stranded RNA in cell function was unclear, it has since been binding domain; RRM, RNA recognition motif. shown that RNase III plays multiple roles in RNA E-mail address of the corresponding author: processing and gene expression by recognizing [email protected] differentRNAsubstrates.2Theparticipationof

0022-2836/02/010021±20 $35.00/0 # 2002 Elsevier Science Ltd. 22 RNase III Kinetics and Mechanism

RNase III in numerous aspects of RNA metabolism these sequences had little effect on multiple turn- and the general features of substrate binding are overcleavageofamodelsubstratebyRNaseIII.17 well established; however, a detailed understand- Interestingly, a re-examination of the sequence ing of the mechanisms of molecular recognition data revealed an exclusion of certain sequences at and catalysis by Escherichia coli RNase III, the de®ned positions relative to the cleavage site; archetypical example of this class of enzyme, is inclusion of these ``disfavored'' sequences in a lacking. modelsubstrateinhibitedbinding.18Thus,exclu- A hallmark of RNase III is its involvement in the sion of non-cognate RNAs appears to be an processing of multiple cellular RNAs. RNase III important aspect of substrate recognition; however, plays an important role in rRNA processing by the precise interactions between RNase III and its cleaving both strands of duplex RNA formed by substrates are not well understood. Clearly, a base-pairing between the 50 and 30 sequences ¯ank- detailed understanding of the kinetics and thermo- ing the mature rRNA sequences, generating the dynamics of recognition by RNase III will be essen- immediateprecursorsto16Sand23SrRNAs.3,4 tial for understanding discrimination between Additionally, RNase III removes some bacterial cognate and non-cognate RNAs. rRNA introns, producing the fragmented rRNAs of Biochemical and structural studies of RNA-pro- RhodobactercapsulatusandSalmonellatyphimurium.5 tein interactions over the past decade revealed that The enzyme also recognizes additional, non-riboso- recognition is most often accompanied by confor- mal RNA substrates in vivo, including speci®c mational changes coupled to formation of speci®c mRNAs from E. coli, and bacteriophages l and complexes; however, the functional signi®cance of T7,2wherecleavageislinkedtoregulationofgene this dynamic behavior has only been de®ned in a expression.6,7Ineukaryotes,RNaseIIIisinvolved fewinstances.19±21Issuesofconformational in the maturation of U2 and U5 small nuclear dynamics appear to be particularly relevant for RNAs, and a large subset of small nucleolar dsRBD-containing proteins such as RNase III. RNAs,8,9inadditiontoitsfunctioninrRNA Bevilacqua and colleagues recently showed that processing.10Recently,aproteincontaining dsRBD binding induces a more A-form helical geo- sequence motifs found in all RNase III enzymes metryinRNAscontaininginternalbulges.22Given was shown to participate in the initial step of gene that most RNase III substrates in vivo also have silencingbyRNAinterference.11Thus,detailed irregular structures, it is relevant to ask whether analysis of the kinetic and enzymatic behavior of corresponding conformational changes involving bacterial RNase III will have general signi®cance the dsRBD occur in substrate recognition by RNase for this broad family of enzymes. III. The importance of this issue is underscored by E. coli RNase III is a homodimer of a 226 amino evidence that changes in enzyme conformation are acid (26 kDa) polypeptide, and like most proteins coupled to speci®city in other enzymes that recog- that interact with RNA, the enzyme appears to be nize complex macromolecular substrates such as composed of multiple functional domains. The amino-acyltRNAsynthetasesandrestriction amino acid sequence of the monomer suggests a enzymes.23±25 folded protein consisting of two domains. The RNase III catalyzes the hydrolysis of the P-30O C-terminal domain consists of a single ca 70 amino phosphodiester bond in RNA generating 50 phos- acid residue double-stranded RNA binding phate and 30 hydroxyl termini, and requires diva- domain(dsRBD).12,13ThelargerN-terminal lentmetalion(s)foractivity.26Thus,RNaseIII domain is thought to contain at least a portion of belongs to a broad class of phosphodiesterase the monomer active site as well as the dimer inter- enzymes, including large ribozymes, polymerases face. Sequence alignment of RNase III orthologues and restriction endonucleases, that use active site identi®es a collection of conserved amino acids metal ions for catalysis of phosphoryl transfer including several acidic residues in the N-terminal reactions.27Basedonstructuralandkineticevi- domain,14thatarebelievedtobeinvolvedinfor- dence, models for catalysis by these enzymes that mation of the enzyme active site. involve one or more active site metal ions have RNase III usually, but not always, cleaves within beenproposed.27,28Amechanismwithtwoions double-stranded RNA leaving a two-nucleotide 30 coordinated to the reactive phosphoryl can be used overhang.2Inspectionofthesecondarystructures to describe the general features of metal ion depen- of E. coli RNase III substrates indicates that strand dent phosphodiesterase catalysis. Here, one ion cleavage sites are ¯anked by 10-14 nucleotides of coordinates to the bridging oxygen atom of the base-paired sequence resulting in a 20-30 nucleo- leaving group while the second interacts with the tide double helical substrate. Deletion studies are attackingnucleophile.29Althoughthe``twometal consistent with a role for helical structure in sub- ion'' mechanism appears to hold for the polymer- straterecognition,15butmostinvivosubstrates ase-derivedexonucleasessuchasKlenow,30and have more complex structures. Alignment of ca 70 has provided a useful framework for exploring known RNase III cleavage sites reveals weakly mechanisms of enzymatic phosphoryl transfer, conserved sequence motifs adjacent to sites of there is evidence that only a single metal ion is cleavage.16However,substratescandeviatesigni®- required for catalysis by some phos- cantly from the consensus, yet are apparently phodiesterases.31,32Additionally,recentanalysesof recognized ef®ciently in vivo, and mutations within group I ribozyme catalysis indicate that at least RNase III Kinetics and Mechanism 23

Figure 1. Puri®cation and analysis of E. coli RNase III. (a) Analysis of puri®ed RNase III enzyme by SDS-PAGE. Lane 1, whole cell lysate taken prior to induction; lane 2, whole cell lysate taken four hours after induction with IPTG; lanes 3 and 5, Benchmark protein size standards (Gibco-BRL); lanes 4 and 6, 1 mg and 20 mg, respectively, of the ®nal puri®ed RNase III preparation. (b) Analysis of the dimerization state of RNase III by gel-®ltration chromatography. The Ve values for a series of molecular mass standards and the void volume, V0 are indicated. The notation of ``dimer'' and ``monomer'' indicate the positions at which material of 52 kDa and 26 kDa, respectively, would be predicted to elute based on a standard curve obtained from the Ve values shown. The elution peak observed for the RNase III sample corresponds to an apparent molecular mass of 62 kDa. (c) Secondary structure of the R1.1 RNA substrate. The primary cleavage site is indicated by an arrow. (d) Speci®city of RNase III cleavage and visualization of reaction products. RNase III cleavage reactions were performed under pre-steady-state conditions at pH 5.4 with 5 nM R1.1 RNA and 2 mM RNase III. The reactions corresponding to lanes 3 through 9 contained uniformly labeled substrate, while the reaction of lane 10, as well as the markers in lanes 1, 2, 11, and 12 utilized (50-32P) end-labeled R1.1 RNA. The reaction in lane 3 contained no RNase III, and was incubated for one hour at 37 C. Lanes 4-9 correspond to aliquots taken from a single reaction after 10, 30, 50, 90 seconds, ®ve minutes and one hour. The reaction containing end-labeled RNA in lane 10 was also incubated for one hour. Lanes 1 and 11 correspond to partial alkaline hydrolysis of end-labeled R1.1 RNA, and lanes 2 and 12 show partial ribonuclease T1 digests of the same material. The position and size of the precursor substrate (60 nt), the 50 cleavage product (47 nt) and the 30 cleavage product (13 nt) are indicated. three distinct metal ions interact with the reactive manganese (Mn2), and this mutant enzyme is less phosphoryl,33andproteinenzymeswithasmany susceptible to inhibition by high concentrations of activesitemetalshavealsobeendescribed.34 Mn2.36Thus,itappearsthat,likeothermetal The importance of magnesium (Mg2) ions in dependentphosphodiesterases,27acidicresidues RNase III catalysis is well established, and recent may play a role in positioning metal ions involved evidence also supports a role for metal ions in stain RNase III function. However, the number and bilizingtheenzyme-substratecomplex.35Interest- af®nity of metal ions required for binding and cat- ingly, the speci®city of metal ion participation in alysis, the roles they play in catalytic mechanism, bindingdiffersfromthosemetalionbindingsites and the enzyme residues involved in metal ion important for catalysis, since calcium (Ca2) sup- interactions are not well de®ned. ports binding, but not catalysis. Evidence for the A fundamental impediment to understanding participation of speci®c RNase III residues in metal the mechanisms of recognition and catalysis by ion interactions comes from the observation that RNase III is that it is not known whether modi®- mutation of one of the conserved acidic residues cations in enzyme or substrate structure that have within the catalytic domain (Glu117) results in an thus far been examined directly affect substrate enzyme that can bind substrate, but promotes cata- recognition, catalysis, or some other aspect of the lysisatagreatlyreducedrate.35Thedeleterious reaction pathway. Importantly, the contributions of effects of an aspartic acid substitution at Glu117 conformational changes, which can complicate the can be somewhat ameliorated by the presence of interpretationofstructure-functionstudies,37have 24 RNase III Kinetics and Mechanism not been assessed. To address these issues and to provide a context for exploring the molecular mechanisms of substrate recognition and catalysis, we examined the steady-state and pre-steady-state reaction kinetics of RNase III cleavage of a com- monly used model substrate. Here we provide evidence that the steady-state cleavage rate does not directly re¯ect substrate cleavage, but rather is limited by a slow step subsequent to catalysis. Pre- steady-state reaction kinetics, however, provide a means for directly monitoring both substrate binding and catalysis. This framework permitted several insights into the RNase III reaction mechanism, including an assessment of the number of acidic residues involved in catalysis and evidence for the participation of a metal hydroxide in the transition state. Furthermore, stopped-¯ow ¯uorescence studies revealed conformational changes associated with both binding and catalysis by RNase III. Together, the results provide signi®- cant new insight into the kinetic and dynamic features of substrate recognition and catalysis by E. coli RNase III.

Results Analysis of steady-state reaction kinetics RNase III was over-produced in E. coli and puri- ®edasdescribed.26,38,39Thepuri®edmaterialhas mobility corresponding to the expected molecular mass of 26 kDa with few additional proteins (<1- 2 %) detected by Coomassie blue staining Figure 2. Steady-state kinetic analysis of R1.1 RNA (Figure1(a)).Furtheranalysisoftheenzymeprep- cleavage at neutral pH. (a) Dependence of the initial aration by gel-®ltration chromatography demon- rate on substrate concentration. Representative data for strates that all puri®ed enzyme is in the dimeric multiple turnover reactions performed at pH 7.5 (Tris- form(Figure1(b)).R1.1RNA,awell-characterized HCl) using (50-32P) end-labeled R1.1 substrate are model substrate derived from bacteriophage T7 shown. Substrate concentrations were varied from wasutilizedinouranalysis(Figure1(c)).Under 10 nM to 640 nM and enzyme concentration varied relatively low (150 mM KCl) monovalent ion con- from 62.5 pM to 4 nM, maintaining [S]/[E>] 5 40 and ditions, RNase III cleaves a single phosphodiester keeping the fraction cut in each reaction under 10 %. bond in the 60 nt RNA generating a 47 nt 50 frag- The duration of each time-course varied from two min- 0 40 utes to 100 minutes depending on the enzyme concen- mentanda13nt3 fragment. Figure1(d)shows tration. Rates at each substrate concentration were separation by denaturing gel electrophoresis of determined by a linear ®t of the fraction substrate con- precursor and product RNA fragments in samples sumed versus time. The data are ®t to the Michaelis- from a time course of RNase III cleavage of uni- Menton equation; v Vmax/(1 (Km/[S])). The inset formly radiolabeled R1.1 RNA. Comparison of the shows the same data plotted as v versus v/[S] (Eadie- cleavage products obtained using uniformly Hofstee) and ®t to v Vmax Km (v/[S]). Both plots 0 labeled and 5 end-labeled R1.1 RNA identi®es the provide equivalent values for Vmax and Km 0 1 47 nt fragment as the 5 cleavage product (Vmax 1.16 min and Km 42 nM). (b) Analysis of the (Figure1(d)).Themobilityofthisfragmentrelative magnitude of the apparent pre-steady-state burst. A plot to partial digests generated using hydroxide and of moles of product formed versus time is shown for multiple turnover reactions performed at pH 8.0 (Tris- RNase T is consistent with cleavage at the correct 1 HCl) using (50-32P) end-labeled R1.1 substrate. The data phosphodiester bond. Importantly, these data are ®t to a linear equation and extrapolated to the y demonstrate that quanti®cation of the time and axis. The amplitude of the burst (p) is shown at the left. enzyme-dependent generation of accurately Squares indicate data from a reaction containing 40 nM cleaved products is not complicated by secondary RNase III and 200 nM R1.1 RNA; circles depict data cleavages, or by contaminating endonucleases in from a reaction containing 20 nM RNase III and 100 nM the enzyme preparation. R1.1 RNA. The ratio of the y-intercepts is 2. First, we determined the steady-state kinetic parameters kcat and Km for the multiple turnover cleavage of the R1.1 substrate by RNase III. Multiple turnover reactions were performed over a range of RNase III Kinetics and Mechanism 25 substrate concentrations, and for each, the plots of shownaspinFigure2(b),increasesindirectpro- product accumulation versus time were ®t to a line- portion to enzyme concentration; in all cases, dou- arequation.Figure2(a)showsaplotoftheseinitial bling the enzyme concentration doubled the rates versus substrate concentration from cleavage apparent burst amplitude{. Taken together, the reactions containing 10-640 nM R1.1 RNA. The relatively slow multiple turnover rate and the pre- data ®t well to both hyperbolic and linearized sence of a pre-steady-state burst of product for- (Eadie-Hoftsee) versions of the Michaelis-Menten mation indicate that a slow step after the ®rst 1 equation and give a kcat value of 1.16 min and round of binding and cleavage, perhaps product Km of 42 nM. These measurements differ some- release or an enzyme conformational change, is 1 what from values of 3.8 min for kcat, and 335 nM rate limiting for the multiple turnover reaction. for Km that were recently reported using the R1.1 36 substrate. Onesourceofthisdifferencecould Pre-steady-state kinetic analysis at low pH arise from the manner in which the reactions are performed, since in the present case reactions were Because the steady-state kinetics do not appear initiated by mixing enzyme and substrate that to provide direct information about binding and were equilibrated separately in reaction buffer; catalysis, it is essential to de®ne reaction conditions whereas, previous analyses have involved starting where the apparent reaction rate re¯ects the rate of reactions by addition of Mg2 to pre-mixed substrate cleavage, and where the behavior of enzyme and substrate. In addition, differences in active enzyme-substrate complexes can be directly reaction conditions, including monovalent ion observed. Since hydroxide is presumably the identity could contribute to these variations in kin- nucleophile in the RNase III catalyzed hydrolysis etic behavior. reaction, lowering the reaction pH should result in Given that the steady-state kinetic parameters a slower catalytic rate, and allow the rate of clea- rarely are a direct re¯ection of enzyme catalytic vage of the bound substrate to be determined rateandsubstrateaf®nity,41,42weexaminedthe directlyusingmanualmethodology.Figure3(a) multiple turnover reaction kinetics more closely. shows that at pH 5.4 the single turnover cleavage We noted that plots of product accumulation versus rate over a range of enzyme concentrations was time did not extrapolate back to the origin slow enough to be determined manually. The data (Figure2(b)).Thisbehaviorsuggeststhataburstof for relatively short reaction times ®t well to a product formation precedes the steady-state single-exponential function; however, longer time- accumulation of product, which would indicate courses are best described by a double-exponential that slow steps after a single round of binding and function. At saturating concentrations of enzyme cleavage are responsible for the linear portion of we measure a cleavage rate constant of ca 1 min1 the reaction. Accordingly, we measured the appar- for the fast phase with amplitude of 50-70 % of the ent amplitude of the burst as a function of enzyme total substrate, and a rate constant of ca 0.1 min1 concentration. If the burst represents a single for the slower phase with amplitude of ca 5-10 %. round of binding and cleavage, then its amplitude The remaining 20-40 % of the substrate that does should be proportional to enzyme concentration. not undergo cleavage is presumably in a confor- The accumulation of products from reactions con- mation that is not reactive. taining 20 nM RNase III and 100 nM R1.1 RNA, or Next, we examined the dependence of the fast 40 nM enzyme and 200 nM substrate are shown in phase (kobs)onenzymeconcentration(Figure3(b)). Figure2(b).Inbothcasesthedata®twelltoaline- Since these experiments are conducted at enzyme ar equation; however, the lines clearly do not concentrations in excess of substrate, and since the extrapolate to the origin. Importantly, we ®nd that fast cleavage rate is easily isolated by ®tting the the amplitude of the intercept with the y axis, data to an equation that accounts for the slowly cleaving population, the presence of non-pro- ductive, or misfolded RNA substrates should not { Assuming the two-step mechanism, p [E]0(k1/ in¯uence the analysis of pre-steady-state kinetic (k k ))2 where k is the apparent ®rst-order rate 1 2 1 data. We ®nd that at low concentrations of RNase constant for the single turnover reaction, k2 is the rate constant for a slow second step that regenerates the free III (1-35 nM) the single turnover rate is dependent enzyme, and [E] is the concentration of enzyme. If all on enzyme concentration, but at higher concen- 0 trations the observed rate constant becomes inde- substrates bind productively, and if k14k2, then p becomes a measure of active enzyme concentration. We pendent of enzyme concentration. This behavior is ®nd that the burst amplitudes are ca 40 % of the molar consistent with a second-order reaction followed amount of enzyme. However, in single turnover by an irreversible ®rst-order reaction, in which the experiments performed at saturating enzyme ®rst step is formation of the E-S complex and the concentrations a signi®cant fraction (20-30 %) of the second step re¯ects chemical cleavage as shown in substrateisnotconsumed(seebelow,Figures4and5), scheme 1: providing evidence for an unreactive population of R1.1 RNAs. In this case the burst amplitude re¯ects the k1 k fraction of bound substrates that are able to undergo E S E Á S !2 E Á P cleavage rather than the fraction of enzyme molecules k1 that were active. Thus, 40 % is only a minimum estimate of the fraction of active enzyme. Scheme 1 26 RNase III Kinetics and Mechanism

Figure 3. Analysis of pre-steady- state reaction kinetics at low pH. (a) Measurement of single turnover rate constant at low pH. Reactions contain 0.2 nM radiolabeled R1.1 RNA substrate and increasing concentrations of enzyme; open circles, 1 nM; open squares, 2 nM; open triangles, 7 nM; ®lled squares, 35 nM; and ®lled circles, 500 nM. The corresponding rates for the predominant fast phase were 0.126 min1, 0.204 min1, 0.678 min1, 1.08 min1 and 1.122 min1, respectively, determined by ®tting the data to a double-exponential equation as described in Materials and Methods. (b) Dependence of the single turnover rate constant on enzyme concentration. The rate constant for the fast phase from single turnover time-courses of

R1.1 cleavage (kobs) is plotted versus enzyme concentration. The data are ®t to a model for a second-order reaction followed by an irreversible ®rst-order reaction as described in the text and in Materials and 1 Methods. The experiment shown here resulted in values for k2 and Kapp of 1.158 min and 6.08 nM, respectively. 1 The reported values for k2andKapp(k21.16(Æ0.01)min ;Kapp5.4(Æ0.6)nM)inthetextandTable1aretheaver- age of three independent determinations with standard errors of less than 14 %. (c) Pulse-chase analysis of the relative magnitudes of the rate constants for substrate cleavage (k2) and dissociation (k1). Reactions (pH 5.1) contained 2 mM RNase III and 5 nM radiolabeled R1.1 RNA. The fraction of reacted substrate is plotted versus time for the reaction that was not chased (open circles); a reaction that was chased with 40 mM cold substrate before mixing enzyme and substrate (open squares); and a reaction that was chased with 40 mM cold substrate at two minutes after initiation of the reaction (®lled circles). Data for each reaction are ®t to either a single (open squares) or double-exponential function (open and ®lled circles). (d) Isolated plot of fraction reacted substrate versus time for data obtained after the cold chase. Data are ®t to a single-exponential function.

In scheme 1, kobs [E]k2/([E] Kapp), where chase, all of the labeled substrate will dissociate Kapp ((k2 k1)/k1). A plot of kobs versus enzyme and the reaction rate will decrease signi®cantly. concentration ®ts well to this mechanism and Figure3(c)showsaplotofproductaccumulation yields a single turnover rate constant (k2) at pH 5.4 versus time for control reactions where either no 1 of1.16(Æ0.01)min andaKapp of 5.4(Æ0.6) nM chase occurred, or the reaction was chased with an (Figure3(b)). excess of cold substrate (40 mM) before mixing

The parameter Kapp value is in¯uenced by sub- enzyme and substrate, and for a reaction that was strate af®nity, as well as the rate of reaction of the chased after two minutes. We ®nd that in reactions enzyme-substrate complex. To examine the extent containing 5 nM radiolabeled substrate, after a to which these two aspects of enzyme behavior chase with 40 mM cold substrate at two minutes contribute to Kapp, we measured the relative rates the rate of product accumulation decreased rapidly 1 of substrate dissociation (k1) and cleavage (k2) to0.01min (Figure3(d)).Thisrateisverysimilar using a pulse-chase approach. First, a population to reactions that were ``chased'' before the start of of enzyme-substrate complexes containing radio- the reaction that yielded a rate constant of labeled RNA was allowed to accumulate prior to 0.02 min1. Therefore, we conclude that the rate of addition of an excess of non-radiolabeled substrate. substrate dissociation is signi®cantly faster (>ten- If the rate of substrate dissociation is signi®cantly fold) than the rate of cleavage of the bound sub- slower than the rate of cleavage (if k24k1), then a strate. Since by de®nition Kapp ((k2 k1)/k1), chase with cold substrate will have no effect on the then if k25k1, then Kapp becomes an approximate rate of product accumulation, since all of the measurement of the dissociation constant (Kd enzyme-substrate complexes formed prior to the k1/k1) for substrate binding. As indicated above, chase will go on to form product without dissociat- we ®nd that Kapp for the single turnover reaction is ing. On the other hand, if the rate of dissociation is 5.4(Æ0.6) nM, which is similar to the values for Kd faster than cleavage (if k25k1), then after the measured previously using gel-mobility shift RNase III Kinetics and Mechanism 27

35,43 experiments at pH 7.5 (Kd1.814.4nM). Thus, the pre-steady-state approach provides a means for measuring both substrate af®nity and the rate of cleavage of the bound substrate.

Irreversibility of the cleavage step, and dependence on pH and divalent metal ion identity Based on the chemistry of non-enzymatic phos- phoryltransferreactions,44±46andmechanisticstu- diesofphosphoryltransferenzymes,47,48the RNase III catalyzed reaction almost certainly pro-

ceeds in a single step via a concerted SN2 type mechanism. The termini of the products formed by RNase III cleavage of RNA are a 50 phosphate and a30 hydroxyl. Based on these products the mechanism is thought to involve irreversible nucleophilic attack of a solvent water molecule on the phosphate at the cleavage site with the 30 oxygen atomastheleavinggroup(Figure4(a)).However, the chemical details of this mechanism, such as its irreversibility and the incorporation of solvent water have not been tested. Additionally, the con- cept that enzymes act by stabilization of reaction transition states means that enzymes catalyze both the forward and reverse reactions of a speci®c chemical transformation. Moreover, it has been argued that enzymes are most ef®cient when the internalreactionequilibriumisnearunity.49Thus, although the equilibrium for phosphodiester hydrolysis is expected to lie strongly toward products, in the context of the specialized environment of enzyme active sites this situation may not ensure kinetic irreversibility. In fact, the hydrolysis 3 of ATP to ADP and PO4 has been shown to be reversibleinATPases.50Togainabetterunder- standing of mechanism and set the stage for subsequent exploration of the catalytic strategies employed by RNase III we employed a simple iso- topic incorporation analysis employing 18O- enriched solvent. A key prediction of the mechanism proposed above is the incorporation of one, and only one, solvent water atom in the newly generated 50 phos- phateonthe30cleavageproduct(Figure4(a)).To test this prediction we performed the RNase III 16 18 cleavage reaction in a mixture of H2 O and H2 O.

thesized control pGp. The value of m/z, or the mass-to- charge ratio, in this experiment is identical with the mass. The monoisotopic mass (442u) is indicated as M and higher molecular mass species due to the presence of natural abundance isotopes are indicated by M 1 and M 2. (c) Mass spectrum of pGp containing the reacted phosphate from RNase III cleavage. Reactions Figure 4. Solvent 18O incorporation during the RNase were performed under multiple turnover conditions as III reaction. (a) Proposed mechanism for the phospho- described in Materials and Methods. The individual diester hydrolysis reaction catalyzed by RNase III. species are indicated as in (b). Material due to incorpor- Nucleotides that are 50 and 30 to the reactive phosphate ation of a single 18OatM 2, as well as a small amount are shown. The solvent nucleophile is shown as a half- of material at M 4 (due to natural abundance isotopes) ®lled circle. (b) Mass spectrum of enzymatically syn- is indicated. 28 RNase III Kinetics and Mechanism

Performing the reaction in 18O-enriched water atom should result in incorporation of a single 18O in the reacted phosphate. Any signi®cant reversal of the reaction will result in the incorporation of additional 18O atoms in the product phosphate upon subsequent re-hydrolysis events. This is because, in the case of reversibility, the free rotation of the 50 phosphate would allow 18Otobe incorporated as a non-bridging oxygen atom in ligated RNA which upon subsequent cleavage results in two 18O atoms in the resulting product. Subsequent to the reaction, the 50 phosphate of the 30 cleavage product was puri®ed as the 50 phosphate of a 30,50 guanosine-bis-phosphate (pGp) containing the guanosine residue located at the N(1) position relative to the cleavage site as described in Materials and Methods. The number of 18O atoms incorporated during the hydrolysis reaction was measured by electrospray ionization quadrupole mass spectrometry. Figure4(b)showsthemasspro®leofacontrol population of pGp derived by treatment of 30 guanosine monophosphate with polynucleotide kinase and ATP. The major peak at 442u is consistent with the monoisotopic mass of pGp (M). The peak at 443u (M 1) is due to the presence of natural abundance 13C and 15N. Also, we observe a measurable amount of material at 444u (M 2), which results from the 0.2 % natural abundance 18O among the 11 oxygen atoms present in pGp. The contrasting mass pro®le from pGp containing the 50 phosphate generated in the RNase III clea- vagereactionisshowninFigure4(c).Aswiththe control pGp, a major peak is observed at 442u, and a lesser amount of material with a mass of M 1 Figure 5. (a) Dependence of the single turnover rate is observed due to natural abundance isotopes. In constant on pH. The log of the single turnover rate con- contrast to the control pGp sample, there is a dra- stant, kobs, is plotted versus reaction pH. Each data point matic increase in the material at M 2 due to the is the average of three independent determinations. incorporation of a single 18O atom from solvent Error bars represent one standard deviation. The data during the RNase III catalyzed hydrolysis reaction. are ®t to models describing the protonation of a single Additional material at 445u is due to natural abun- titratable group (broken line), and two or three groups dance 13C and 15N in the pGp containing the incor- with the same pKa value (dotted line and continuous porated 18O. Importantly, only background line, respectively) as described in the text. Experimental conditions are described in Materials and Methods. amountsofpGpwithmassofM4(compareto (b) Dependence of the single turnover rate constant on M2inFigure4(b))areobservedthatresultfrom metal ion identity. Single turnover assays were per- 18 the presence of natural abundance levels of Oin formed as described in Materials and Methods. Reac- the other ten oxygen atoms in pGp. These results tions contained either MgCl2, MnCl2, NiCl2 or CoCl2 at clearly demonstrate that the RNase III catalyzed 10 mM. These reactions were performed in MRB at hydrolysis reaction involves the incorporation of pH 5.4 with 5 nM R1.1 RNA and 2 mM RNase III. The a single 18O atom from solvent, which we inter- data shown here represent the average of two indepen- pret as indicating that the reaction is kinetically dent experiments; similar experiments under slightly irreversible. different buffer conditions yielded similar results. The log of the single turnover rate (k ) is plotted versus the Any understanding of enzyme catalytic mechan- obs pKa value of the hydrated form of the metal ion ism must necessarily include an examination of the includedinthereaction(datafromFeig&Uhlenbeck).56 role of proton transfer. Consequently, we explored The error in the value for the rate constant determined the extent to which the single turnover rate at in individual experiments with Mg2,Mn2 and Ni2 2 saturating enzyme concentration (k2) is dependent was less than 7 %. Reactions containing Co were con- onpH.Figure5(a)showsaplotofthelogofthe sistently too fast to measure accurately by manual pipet- single turnover rate constant versus pH at values ting, and so the rate constants measured in Co2 had between pH 5.1 and pH 6.1. As the pH of the reac- errors from 35 to 45 %. Individual data points represent tion increases the reaction rate also increases, until averages and error bars represent one standard devi- 2 2 2 above pH 6.1 the rate becomes too fast to measure ation. The values for Ni ,Mn and Mg ®t to a linear manually. However, it is clear that the reaction equation with a slope of 0.34 (continuous line). A ®t of all of the data to a linear equation yielded a line with slope of ca 0.5 (broken line). RNase III Kinetics and Mechanism 29 rate becomes less dependent on hydroxide concen- Accordingly, we determined the single turnover tration at more basic pH values. The simplest rate constant for reactions conducted in the pre- model for this behavior is that protonation of a sence of Mg2,Mn2, nickel (Ni2) and cobalt functional group in the enzyme-substrate complex (Co2), all of which have been shown to support inhibits catalysis: theRNaseIIIcleavagereaction.26Asreportedpre- viously, two additional metal ions, cadmium 2 2 Ka (Cd ) and Ca , which were examined as controls, AH A H did not support R1.1 cleavage by RNase III (data Scheme 2 not shown). Interestingly, the ionic radius of the divalent metal ions that do not support RNase III where AH is the inactive form of the enzyme, A catalysis are 0.95 AÊ or greater, whereas the ionic is the active, deprotonated form and the equili- radii of Mg2,Mn2,Ni2 and Co2 are all smaller brium between the protonated and deprotonated (0.72 AÊ , 0.67 AÊ , 0.66 AÊ and 0.75 AÊ , respectively), forms is described by Ka (see Materials and suggesting that there may be a steric limitation on Methods). However, the data ®t poorly to an ion size in one or more essential divalent metal ion equation describing only a single titratable group binding sites in the enzyme-substrate complex. in that the rate decreases much more sharply at AsshowninFigure5(b)we®ndthatthepKa acidic pH values than predicted (broken line, value of the metal ion hydrate correlates with Figure5(a)).Next,weconsideredtwotitratable changes in the single turnover rate constant deter- 2 2 2 2 groups with differing Ka values: mined with either Mg ,Mn ,Ni or Co .We observe a clear relationship between the log of the

Ka1 Ka2 single turnover rate constant and the pKa value of HAH HA H A2 H the hydrated divalent metal ion in the reaction; as

Scheme 3 the metal ion pKa value decreases, the associated rate constant progressively increases. For Mg2, where HAH and HA are inactive, protonated Mn2 and Ni2 the data ®t very well to a linear forms and A2 represents the active deprotonated equation with a slope of 0.34 (continuous line, form of the enzyme-substrate complex. A good Figure5(b)).Interestingly,reactionsperformed theoretical ®t to the data was obtained with an with Co2 were signi®cantly faster than would be equation describing this model assuming two predicted from the relationship between rate and different pKa values for the two titratable groups; pKa value for the other metal ions tested; however, however, one of the values was consistently too the relatively fast rate with Co2 was more dif®cult low to be considered reasonable (pKa < 2). The to measure precisely. Including the data obtained data also ®t well to a similar model assuming that usingCo2inthelinear®t(dottedline,Figure5(b)) both of the titratable groups have the same pKa yields a slope of 0.5, but with a signi®cantly poorer value (pKa5.9;dottedline,Figure5(a)).Interest- ®t. Although it is not possible to draw a direct line- ingly, the best ®t was obtained using a variation of ar relationship for all of the data, there is clearly a this model (see Materials and Methods) that correlation between the observed rate in different assumed three titratable groups with the same pKa metal ions and the pKa value of the metal ion value (pKa5.6;continuousline,Figure5(a)). hydrate. To further explore the relationship between the Taken together, the pH dependence of k2 and the rate constant k2 and the substrate cleavage step we correlation between this kinetic parameter and determined its dependence on metal ion identity. metal ion identity both suggest that under acidic High-resolution crystal structures of several metal pH conditions the rate measured at saturating ion dependent phosphodiesterase enzymes, enzyme concentration re¯ects the chemical step of together with chemical and geometric consider- the reaction. Importantly, we detected no lag in ations indicate that, in general, a water molecule is product formation at any of the enzyme concen- activated to carry out nucleophilic attack at the trations tested, which further supports the con- phosphorus atom via interactions with divalent clusion that there are unlikely to be slow steps, metalions.27Inthesecases,thenucleophileispart such as conformational changes, occurring after of the inner hydration sphere of a metal ion, binding and before the chemical step. However, whereby coordination lowers the pKa value of the kinetic analysis cannot exclude the possibility that water and therefore increases the concentration of changes in enzyme and substrate conformation themorenucleophilicmetal-hydroxidespecies.30,51 might occur very rapidly compared with binding If the ability of a metal ion to stimulate catalysis is and catalysis, or that such changes are in rapid proportional to the concentration of M2(OH)at equilibrium relative to these events. the active site, then metal ions with lower pKa values should have higher cleavage activities due Stopped-flow fluorescence analysis to the increased presence of M2(OH). Thus, an observed relationship between metal ion pKa To examine more closely the possibility that con- values and metal ion cleavage activity has been formational changes were involved in RNase III taken as evidence that a solvated metal hydroxide binding and catalysis, we utilized intrinsic trypto- participatesincatalysis.52 phan ¯uorescence as a reporter of enzyme confor- 30 RNase III Kinetics and Mechanism mation in stopped-¯ow experiments. The RNase III Both the fast and slow changes in tryptophan monomer has a single tryptophan residue at 139 ¯uorescence ®t well to single-exponential functions that is located between the dsRBD and the catalytic individually, or a double-exponential function in domains of the enzyme; therefore, this residue is combination. The ®t to a double-exponential func- 1 not expected to be directly involved in the RNA- tion yields a rate constant (kA) of 55(Æ5) s for the 1 protein interface. Thus, any change in intrinsic fast phase and a rate constant (kB) of 6.4(Æ0.8) s ¯uorescence is likely to be due to conformational for the slow phase at pH 7.3, 10 mM Mg2, 0.8 mM changes in the protein that alter the environment R1.1RNAand1.25mMRNaseIII(Table1).The of Trp139. Single turnover reactions were per- amplitude of the change in ¯uorescence is rela- formed with concentrations of enzyme and sub- tively small, which places a constraint on the range strate that were at least tenfold higher than Kd. of enzyme and substrate concentrations that can be Figure6showsthatfollowingmixingofRNaseIII examined, but the ¯uorescence signal can be with the R1.1 substrate at pH 7.3 in the presence of readily detected under the single turnover con- Mg2 there is a rapid increase in relative ¯uor- ditionsdescribedforFigure6. escence in the ®rst 0.1 second that is followed by a To assess whether the changes in ¯uorescence slower decrease in ¯uorescence that takes place were due to bimolecular association of RNase III over a time-scale from zero to one second. Impor- with the R1.1 substrate or due to unimolecular tantly, no change in ¯uorescence was observed rearrangements subsequent to binding, we when either enzyme or substrate was mixed with measured kA and kB values at a series of substrate buffer, or when buffer was examined alone concentrations(Figure7).Atconcentrationsof (Figure6,anddatanotshown).Thus,weconclude enzyme and substrate that are greater than tenfold that the changes in ¯uorescence are unlikely to be over Kd, a unimolecular rearrangement occurring due to mixing or dilution artifacts, but in fact after substrate binding is not expected to show any re¯ect changes in enzyme conformation occurring signi®cant dependence on the concentration of sub- during binding and/or catalysis. No other pro- strate. On the other hand, conformational changes cesses were observed over intermediate time-scales associated with binding should display concen- (0-60 seconds) (data not shown). Since the apparent trationdependence.AsshowninFigure7,thefast pre-steady-state burst observed in multiple turn- increase in ¯uorescence re¯ected by the rate con- over experiments takes place in less than ten stant, k , is dependent on the concentration of R1.1 seconds, we conclude that the rapid increase and A slower decrease encompass all of the changes in protein ¯uorescence that occur during a single round of binding and cleavage by RNase III.

Figure 7. Dependence of kA and kB on substrate con- Figure 6. Changes in intrinsic tryptophan ¯uorescence centration. Reactions contained 0.25 mM RNase III, a during the single turnover cleavage of R1.1 RNA by range of R1.1 RNA substrate concentrations, 10 mM MgCl and Tris reaction buffer (pH 7.3). Rate constants RNase III. Reactions were performed in Tris reaction 2 k and k were obtained from a ®t to a double-exponen- buffer (pH 7.3) (see Materials and Methods) including A B tial function. Rate constants represent the average of at 10 mM MgCl2 and contained 1.25 mM RNase III and either 0.8 mM R1.1 RNA (open circles), or buffer solution least three experiments with error bars representing one standard deviation. Data for k and k are depicted by with no substrate (crosses). The data for the reaction A B ®lled circles and squares, respectively. The data for k containing substrate are ®t to a double-exponential func- A are ®t to a linear equation yielding a second-order rate tion and the resultant rate constants are reported in 8 1 1 Table1. constant of 1.0 Â 10 M s . RNase III Kinetics and Mechanism 31

Table 1. Effects of pH and divalent metal ion on the single turnover rate constant, k2, and the rate constants for changes in ¯uorescence, kA and kB pH 7.3 pH 6.1 pH 5.8 Mg2 Ca2 Mg2 Mg2

1 a k2(s ) n/d00.097Æ0.0160.057Æ0.002 1 b c kA(s ) 55Æ576Æ 7 70 Æ 852Æ 10 1 kB (s ) 6.4 Æ 0.8 0 0.17 Æ 0.06 0.083 Æ 0.007 a The single turnover rate constant, k2, was determined at saturating concentrations of RNase III at pH 6.1 and 5.8 as shown in Figure5. b Rate constants for both ¯uorescence transitions, kA and kB, were determined using 1.25 mM RNase III and 0.8 mM R1.1 RNA as described in Materials and Methods in the presence of either 10 mM Mg2 or Ca2 as indicated. c Data were ®t to a double-exponential equation.

RNA. Fitting these data to a linear equation yields In order to test how closely the ¯uorescence 8 1 1 asecond-orderrateconstantof1Â10 M s , transition represented by kB corresponds to the rate suggesting that kA involves a one-step process of substrate cleavage, we examined whether kB associated with binding that occurs 10-100-fold was signi®cantly slower in reactions performed slowerthanthediffusioncontrolledlimit.42Incon- underacidicconditions(pH5.8and6.1;Table1). trast, the subsequent decrease in ¯uorescence rep- AsshowninFigure9(a)andTable1thereisasig- resented by k was independent of substrate B ni®cant decrease in kB as the reaction pH becomes concentration over the range tested. The lesser more acidic, similar to the dependence observed magnitude of kB relative to kA and its lack of for the single turnover catalytic rate. We ®nd that dependence on R1.1 RNA concentration argue that at pH 6.1 there is a ca 40-fold decrease in kB rela- kB re¯ects a unimolecular event that occurs sub- tive to measurements performed at pH 7.3 sequent to formation of the enzyme-substrate com- 1 1 (kB 0.17 s (pH 6.1) versus kB 6.4 s (pH 7.3)), plex such as catalysis. and an almost 80-fold decrease when the reaction To test the hypothesis that kA is associated with was performed at pH 5.8 (k 0.083 s1 (pH 5.8)). the binding step of the RNase III single turnover B In contrast, the rate constant kA describing the reaction and that kB re¯ects the cleavage step, we ¯uorescence transition associated with binding tested their sensitivity to alternative divalent metal was found to be essentially pH independent with 35 ions.Itiswellestablished thatbothbindingand values of ca 55 s1,70s1 and 52 s1 measured at cleavage of the R1.1 RNA substrate are dependent pH 7.3, 6.1 and 5.8, respectively. Thus, the depen- on the presence of Mg2. In contrast, Ca2 clearly dence of kB, and the relative independence of kA on supports substrate binding at af®nities that are reaction pH provided additional evidence that essentially identical with those measured in the these parameters are associated with catalysis and presence of Mg2; however, Ca2 is unable to sup- substrate binding, respectively. portcatalysis.AsshowninFigure8(a),both¯uor- A direct comparison between the single turn- escence transitions are observed in reactions over catalytic rate constant, k , and the ¯uor- containing the optimal divalent metal ion, Mg2.In 2 escence transition k illustrates how closely these contrast, when the stopped-¯ow ¯uorescence anal- B ysis was carried out with Ca2 only the rapid twophenomenaarecoupled(Figure9(b)).Given the large dependence of both k2 and kB on reac- increase in ¯uorescence represented by kA was observed. No subsequent decrease in intrinsic ¯u- tion conditions a direct comparison of these rate orescence was detected, even in reactions moni- constants required performing parallel exper- toredforuptooneminute(Figure8(b),anddata iments. Thus, essentially identical reaction mix- not shown). Importantly, we ®nd that the rate con- tures of 0.8 mM RNase III and 1.25 mM R1.1 2 RNA at pH 6.1 were analyzed either by stant for kA measured for Ca is similar to that determined in the presence of Mg2 stopped-¯ow ¯uorescence, or by product Ca 1 Mg 1 accumulation using a trace amount of radio- (kA 76(Æ7) s and kA 55(Æ5)s )(Table1). Additionally, when divalent metal ions were labeledRNA.Figure9(b)showsthatthedata omitted from the reaction, no changes in enzyme from both analyses ®t single-exponential func- ¯uorescence were detected. The observation that tions and yield rate constants that are remark- 1 1 the occurrence of the changes in ¯uorescence rep- ably similar (kB 0.16 s and k2 0.13 s ). Thus, the conformational change re¯ected by k resented by kA and kB are dependent on divalent B is not only sensitive to pH, as would be metal ions, and the fact that the second phase (kB) is observed in the presence of Mg2, but not Ca2, expected for a transition associated with RNase provides further support for the conclusion that kA III catalysis, but also occurs with a rate constant re¯ects events associated with substrate binding, that is practically indistinguishable from that while kB is associated in some manner with cataly- derived from measuring the accumulation of sis. cleaved RNA product. 32 RNase III Kinetics and Mechanism

Figure 9. Dependence of kB on pH. (a) Changes in ¯uorescence observed in reactions containing 1.25 mM

RNase III, 0.8 mM R1.1 RNA, 10 mM MgCl2 and MRB at either pH 5.8 (open circles) or pH 6.1 (open squares). The data are ®t to a single-exponential function, since the initial rapid increase in ¯uorescence is too fast to be observed at this time-scale. Resultant rate constants are

listedinTable1.(b)CorrespondenceofkB and the rate constant for product formation determined manually using radiolabeled substrate. Crosses represent ¯uorescence data obtained at pH 6.1, circles represent data from a parallel reaction that contained identical concentrations of enzyme and substrate, as well as a trace concentration of radiolabeled RNA. The data for both assays are ®t to a single-exponential function.

Figure 8. Metal ion dependence of kA and kB. Each panel shows the change in intrinsic ¯uorescence detected in reactions containing 1.25 mM RNase III, 0.8 mM R1.1 RNA, Tris reaction buffer (pH 7.3), and the divalent metal ion indicated. (a) 10 mM MgCl2; (b) 10 mM Ca2; (c) no metal ion. For the reaction contain- Discussion ing Mg2 the data are ®t to a double-exponential function, and for reactions containing Ca2 a single- These studies allow several conclusions to be exponential function was used. The resulting rate con- drawn concerning the mechanism and reaction kin- stantsarereportedinTable1. etics of E. coli RNase III cleavage of the R1.1 model substrate. (1) The multiple turnover reaction rate, which has generally been used to assess catalysis by this enzyme, does not directly re¯ect the rate of cleavage of the bound substrate. (2) At low pH the single turnover rate at saturating enzyme re¯ects the rate of cleavage of the bound substrate, and RNase III Kinetics and Mechanism 33 under these conditions substrate association and which the enzyme is susceptible to product inhi- dissociation are in equilibrium relative to the rate bition. Given that substrate dissociation appears to 1 of RNA cleavage. (3) The chemical step is irrevers- befast(>10min ){relativetothekcat, it is likely ible, involving incorporation of a single solvent that product dissociation might also be similarly oxygen atom during the reaction. (4) The rate of fast. Therefore, we hypothesize that an enzyme RNA cleavage is sensitive to pH, re¯ecting a conformational change, either before or after pro- requirement for at least two functional groups in duct dissociation, is rate limiting for multiple turn- the transition state that can be deprotonated in the over. Thus, because of the limitations of the accessible pH range. (5) One role of divalent metal multiple turnover kinetics, it appears from the data ions, which are essential for the catalytic step, is to presented here that a clear understanding of the provide a metal-bound hydroxide that may act as mechanisms of RNase III catalysis is best gained the nucleophile in the hydrolysis reaction. (6) Con- from analysis of pre-steady-state kinetics. formational changes accompany binding and cata- The ability to monitor the rate of cleavage of the lysis, and the transition associated with RNA bound substrate at low pH permitted several basic cleavage appears to be concomitant with catalysis. features of the RNase III catalyzed RNA cleavage The bases for these deductions are discussed in reaction to be de®ned. We ®nd that protonation of turn, as well as the implications they have for the two or more functional groups in the enzyme-sub- function of RNase III. strate complex results in inhibition of catalysis. In general, the multiple turnover cleavage rate Interestingly, a highly conserved feature of the cat- has been used to evaluate the effects of mutation, alytic domain of RNase III is the presence of sev- substrate modi®cation and changes in reaction con- eral acidic residues. Thus, it is tempting to ditions on catalysis by RNase III. However, we speculate that one or more of these conserved ®nd evidence for a pre-steady-state burst of pro- amino acid residues are involved in the sensitivity duct formation followed by a slower accumulation of the cleavage reaction to decreasing pH{. What due to turnover. Additionally, the single turnover then, might be the role of these functional groups rate constant for substrate cleavage at saturating whose participation in catalysis depends on depro- enzyme concentration is faster (ca 6 min1 (pH 6.1); tonation? The proposed mechanism of RNase III 1 hydrolysis involves deprotonation of water to 380 min (pH 7.3)) than the relatively slow kcat (1.16 min1 (pH 7.5)). Thus, the rate of product for- yield the hydroxide ion nucleophile, and protonation of the 30 oxygen atom leaving group. Thus, it mation during multiple turnover appears to re¯ect might be expected that protonation of a single a slow step occurring after catalysis. In this situ- functional group (involved in nucleophilic acti- ation, it is dif®cult to draw straightforward con- vation) would inhibit catalysis; however, models clusions from structure-function studies based on that invoke two or more basic functional groups multiple turnover kinetics, since V and K max m best describe the data. A similar sensitivity to pH values under these conditions are dependent on involving ionization of two carboxylic acid resi- the rate constant for the rate determining step, and dues involved in nucleophilic activation is do not directly re¯ect enzyme binding or catalysis. observedintheDNAendonucleaseEcoRV.54Itis None of the studies presented here directly proposed that the proximity of the two carboxy- addresses exactly what step after substrate clea- lates results in an increase in the ability of one of vage is rate limiting for the multiple turnover reac- these residues to accept a proton from the nucleo- tion. It may be that slow release of product, or phile. The preponderance of carboxylate residues enzyme isomerization, limits the steady-state clea- in the active sites of other phosphodiesterases vage rate, given that conformational changes suggests that this mechanism may be general; clearly occur during the reaction. In order to better however, a second important role for these resi- understand this issue it will be necessary to exam- duesthatwouldalsobesensitivetoprotonationis ine in detail the pathway for dissociation of the coordinationofactivesitemetalions.30,55Inthis two RNA products and measure the extent to scenario the inhibition of enzyme activity at low pH would be due to weakened metal ion inter- { Based on the pulse-chase experiment presented in actions rather than an inability to perform proton Figure3,therateofsubstratedissociationisestimated transfer. It may be possible to distinguish between to be at least tenfold faster than the rate of reaction of these alternatives by employing the pre-steady- the enzyme-substrate complex. Under the conditions state approaches described here to examine the used in the pulse-chase analysis the single turnover rate metal ion dependence of the reaction rate under was ca 1 min1 indicating a lower limit for the substrate different pH conditions. dissociation rate of ca 10 min1. While it is well established that RNase III { Here the conserved glutamate and aspartate requires divalent metal ions for activity, the roles residues in RNase III are given the conventional of metal ions in catalysis remain unde®ned. Gener- designation as acidic, although in practical terms their contribution is usually that of Brùnsted bases, that is, in ally, it is accepted that metal ions play a direct role their ability to accept a proton at neutral pH. Thus, if in catalysis by RNase III similar to other divalent the contribution of the conserved glutamate and metal ion dependent phosphodiesterases such as aspartate residues is to act as anions, then protonation restriction enzymes, RNase H, and large will engender inhibition.53 ribozymes.27Oneproposedroleformetalionsin 34 RNase III Kinetics and Mechanism catalysis by this class of enzymes involves inter- usingbothstructure-functionrelationships,44and action with the nucleophile, which serves to lower heavy-atomisotopeeffects,46whichdemonstrate the pKa value, thereby resulting in a higher frac- that phosphodiester hydrolysis reactions are tionation to hydroxide for the bound water at neu- always concerted. Additionally, analysis of the tralpH.51Intriguingly,we®ndthatthepresenceof stereochemistry of catalysis by phosphodiesterases hydrated metal ions with pKa values lower than such as restriction enzymes shows that cleavage that of Mg2 results in a higher catalytic rate, and occurswithinversionofcon®guration.58,59Thus, that there is a signi®cant correlation between the for these enzymes the reaction occurs in a single log of the single turnover rate constant and metal step and is irreversible. To test the mechanism and ion pKa. The lower the pKa value of an individual reversibility of the RNase III catalyzed reaction we 18 divalent ion, the greater the fraction of the examined incorporation of solvent H2 O into the hydrated ion that will be in the form of metal products of the reaction. Similar approaches have hydroxide at a given pH. If the ability of a metal beenparticularlyusefulininvestigatingthemech- ion to stimulate catalysis is proportional to the con- anismsofATPases,60,61aswellasotherphosphoryl centration of metal hydroxide in the active site, transferenzymes.62±64Asanticipated,we®ndthat 18 then metal ions with lower pKa values will have a single O atom is incorporated into the newly highercleavagerates.52Theobservationoffaster generated 50 phosphate. This result is consistent RNase III catalytic rates in metal hydrates of with a single-step irreversible hydrolysis mechan- decreasing pKa values argues that a metal ion ism. hydroxide is required for catalysis by this enzyme. In addition to advancing our understanding of One attractive hypothetical role for this metal the RNase III mechanism the present study also hydroxide is in generating the hydroxide ion that provides, for the ®rst time, an assessment of con- is incorporated as the nucleophile in the hydrolysis formational changes during substrate binding. reaction. Given the correlation between reaction Conformational changes appear to be a common rate and metal ion pKa it is interesting to note that aspect of RNA-protein interactions, and stopped- the slope of a linear ®t to the data is less than 1 ¯ow ¯uorescence has been successfully used to (m 0.34-0.5). The most general interpretation is study RNA-protein dynamics in several case- that the ability of metal ion hydrate to stimulate s,23,24,65,66aswellasforunderstandingRNA the reaction does not depend merely on concen- dynamics.67±69We®ndthatthereareconfor- tration, in which case the expected slope would be mational transitions associated with both binding unity. A smaller dependence on pKa is expected if andcatalysisbyRNaseIII(Figure10).Thecon- the activating metal ion hydrate is acting as the clusion that the initial, rapid increase in ¯uor- nucleophile in the reaction because there is a sig- escence upon mixing enzyme and substrate is ni®cant although not absolute dependence of phos- associated with substrate binding is based on the phodiester hydrolysis reactions on the basicity of concentration dependence of the rate constant of 44 the nucleophile (bnuc0.3). Thus,therate this parameter. In addition, this transition shows a enhancement from an increase in concentration of metal ion dependence that is similar to that the metal-bound hydroxide is partially offset by its demonstrated for substrate binding. The concen- decreased reactivity relative to free hydroxide, tration dependence of kA yields a second-order rate 51 8 1 1 which has a higher pKa value. Analternative constant of 1 Â 10 M s , which is less than the possibility is that a metal ion hydrate is important diffusion controlled limit for macromolecular inter- for formation of structure in the enzyme substrate actions.42Similar¯uorescencetransitionsuponfor- complex. Given the importance of metal ion inter- mation of RNA-protein complexes occurring at actionstoRNAstructure,56andevidenceofmetal equally fast rates have also been observed in sub- ion interactions in the R1.1 RNA internal bulge strate binding by aminoacyl-tRNA synthetases andloop,57itmaybethatahydratedmetalion, (4Â107M1s1),23,24bindingoftherestriction binding in the deprotonated form to the enzyme- enzymeEcoRVtoitssubstrate(5Â107M1s1),70 substrate complex, makes a signi®cant contribution binding of the termination factor Rho to RNA to catalysis. This alternative seems less likely, since a much greater dependence on pKa (m ca 1) than isobserved(m0.34-0.5;Figure5(b))wouldbe expected, as all of the effect on reaction rate would come from the increase in metal hydroxide concentration. Based on these arguments and information from orthologous enzymes, our current model of the chemical reaction includes a metal ion hydroxide as the reactive nucleophile. The mechanism of the RNase III catalyzed 0 Figure 10. Kinetic scheme for single turnover cleavage hydrolysis reaction is thought to be SN2 with the 3 of R1.1 RNA by RNase III. The R1.1 RNA is depicted as oxygen atom as the leaving group and a solvent a stem-loop. RNase III is depicted as gray circles, hexa- water molecule or hydroxide acting as the nucleo- gons or ovals. The changes in enzyme conformation phile. This perspective is based on analysis of non- indicated by the changes in ¯uorescence are depicted enzymatic phosphodiester hydrolysis reactions using these different shapes. RNase III Kinetics and Mechanism 35

(8Â108M1s1),65andintheinteractionofMS2 RNA or DNA substrates, RNase III has to correctly coat protein with its cognate RNA stem-loop identify cognate substrates in what has been 9 1 1 66 (2Â10 M s ). Thus,itseemslikelythatkA termed the ``molecular distraction'' of a pool of represents a rapid transition occurring either con- structurallysimilarnon-cognateRNAs.76 comitantly with, or just after the initial interaction Obviously, errors in recognition by cellular DNase betweensubstrateandenzyme(Figure10). and RNase enzymes would have ruinous conse- It is widely assumed that at least part of the quences for cell function. One attractive hypothesis interface between RNase III and its substrates is that the conformational changes observed here involves the dsRBD motif found in all RNase III are relevant to the central issue of how RNase III enzymes. Interestingly, there appears to be few recognizes structurally diverse substrates. large-scale differences in the three-dimensional structure of the free dsRBD relative to the com- plexeswithRNA.19,71,72However,thetwoavail- Materials and Methods able structures show some differences in the RNA- protein interface, making it dif®cult to predict how Overexpression, purification and quantification of the dsRBD is involved in substrate recognition by E. coli RNase III RNase III. The single tryptophan residue moni- RNase III was overexpressed in HMS174(DE3)/ tored in the experiments described here is located pET11a-RNC cells and puri®ed based on previously between the conserved catalytic domain of RNase describedprotocols.26,38,39Inductionofexpression III and the dsRBD. Interdomain linkers, such as resulted in synthesis and accumulation of the 26 kDa those found between the two RNA recognition RNase III monomer polypeptide. After cell lysis, RNase motifs (RRMs) of U1A and poly(A) binding protein III was isolated as inclusion bodies, which were washed show the greatest change in structure between the and subsequently solubilized in a solution containing freeandboundprotein.73,74Thus,itmaybethat 8 % (w/v) ammonium sulfate. This material was applied Trp139 is fortuitously positioned to be a good to a column of AG poly(I)-poly(C) agarose (Amersham reporter of inter-domain conformation. Pharmacia Biotech) (5 ml bed volume, 2.8 cm Â 1.5 cm) equilibrated in buffer containing 0.5 M NH Cl. The col- The energetic penalties involved in binding large 4 umn was washed with 1 M NH Cl, and RNase III was RNA substrates, which are associated with break- 4 eluted with 2 M NH4Cl. During chromatography, absor- ing interactions found in the unbound state, sub- bance of the eluate was monitored at 280 nm. Individual strate distortion and conformational restrictions, fractions were analyzed by SDS-PAGE, and the RNase are likely to be signi®cant relative to the positive III-containing fractions were pooled. SDS-PAGE was car- contribution of binding interactions. The complex- ried out using 12 % (w/v) polyacrylamide gels according ities of these issues for interpretation of structure- tostandardmethods.77ThegelwasstainedwithBiosafe function studies are illustrated by recent analysis Coomassie G250 (BioRad), destained with water and air ofbindingbythewell-studiedRRM.37,75Thus,the dried between cellophane sheets. The pooled fractions observation of conformational changes associated were concentrated twofold in Slide-A-Lyzer 7000 MWCO cassettes (Pierce) by brief exposure to Slide-A- with RNase III substrate binding has a signi®cant Lyzer Concentrating Solution (Pierce), and were sub- implication for structure-function studies of this sequently dialyzed against enzyme storage buffer, ESB enzyme. It may be dif®cult to deconvolute whether (50 % (v/v) glycerol, 0.5 M KCl, 30 mM Tris-HCl changes in the binding equilibrium, due to (pH 8.0), 0.1 mM EDTA, 0.1 mM DTT) in the same mutation or modi®cation, result from changes in cassettes. The ®nal material was stored in aliquots at the structure of the free enzyme or substrate, the 20 C. Six grams of cells yielded 11 mg of RNase III bound complex, or some intermediate. that was >95 % pure as judged by Coomassie G250 stain- In addition to the rapid increase in ¯uorescence ing of an overloaded SDS-PAGE lane. associated with binding we detect a slower phase The concentration of RNase III was determined using that is concomitant with cleavage of the bound a calculated molar extinction coef®cient based on the substrate(Figure10).Theobservationisofinterest cysteine, tyrosine and tryptophan content of E. coli RNaseIII.78,79Thecalculatedmolarextinctioncoef®cient both because it provides insight into the dynamics 1 1 at 280 nm is 13,490 M cm (A1% 5.3) for the RNase of RNase III and because the stopped-¯ow ¯uor- III monomer (26,980 M1 cm1 for the dimer). This value escence approach provides a convenient continu- differs by 2.4-fold from the extinction coef®cient reported ous assay for both binding and catalytic events. previously;26however,theprocedureusedtomeasure Clearly, combination of this approach with pre- this value has not been well described. Since calculated steady-state kinetics using radiolabeled substrate molar extinction coef®cients have generally proven to be studies will provide a powerful method for explor- reliable,evenwhenatoddswithmeasuredvalues,79the ing molecular recognition and catalysis by RNase value of 26,980 M1 cm1 was employed. III. Interestingly, precedence for the occurrence of For analytical gel ®ltration chromatography, a Super- conformational changes associated with catalysis dex 75 (Amersham Pharmacia Biotech) gel ®ltration within RNA-protein complexes is found in both column (129 ml bed volume, 64.5 cm Â 1.6 cm) was employed. The separation properties of the column were aminoacyl-tRNAsynthetaseenzymes24andclea- 70 determined by calibration with proteins of known mol- vagebytherestrictionenzymeEcoRV. Itis ecular mass. Speci®cally, aprotinin (6500 Da), cyto- of interest to note that similar issues of molecular chrome c (12,400 Da), carbonic anhydrase (29,000 Da), recognition may be a consideration for RNase III and bovine serum albumin (66,000 Da) were utilized as as well. As with most enzymes that interact with molecular mass standards and blue dextran 36 RNase III Kinetics and Mechanism

(2,000,000 Da) as a marker for the void volume (V0). The aration for gel puri®cation, 15 ml of formamide gel load- column was equilibrated in 50 mM Tris-HCl (pH 8.0), ing buffer, GLB (96 % (v/v) formamide, 10 mM EDTA, 150 mM KCl at a ¯ow rate of 0.5 ml/minute. The mar- 20 mM Tris-HCl (pH 8.0), 0.025 % (w/v) xylene cyanol kers were dissolved in the same buffer including 5 % gly- and 0.025 % (w/v) bromophenol blue), was added. The cerol and applied to the column in a volume of 1 ml at RNA was electrophoresed through a 15 % (w/v) poly- the same ¯ow rate. Elution of the protein standards was acrylamide 7 M urea gel, visualized by UV shadowing, followed by monitoring absorbance at 280 nm. The excised, and eluted overnight in RNA elution buffer elution volume (Ve) for each protein was measured and (0.3 M sodium acetate, 0.5 % (w/v) SDS, 10 mM Tris- used to calculate a standard curve (molecular mass ver- HCl (pH 8.0), 1 mM EDTA). The eluted RNA was sus Ve/V0). The apparent molecular mass of RNase III, extracted twice with phenol/chloroform, once with as well as predicted elution volumes for the dimeric and chloroform, and precipitated with ethanol. After centrifu- monomericformsoftheenzymepresentedinFigure1(b) gation, the RNA pellet was washed twice with 80 % were calculated from the standard curve. For analytical (v/v) ethanol, air dried, and resuspended in TE. The runs, 400 mg of RNase III in 0.1 ml of ESB was applied to concentration was determined by measuring absorbance the Superdex 75 column that had been equilibrated in a at 260 nm. This protocol routinely resulted in yields of buffer equivalent to ESB but with 5 % glycerol instead of 10-20 nmol R1.1. For larger-scale synthesis, multiple 50 % glycerol. 400 ml reactions were employed. To prepare substrate RNAs for (50 32P) end-labeling the transcription reactions were primed with guanosine. Guanosine-primed RNAs Synthesis and purification of RNA were generated using the conditions described above, PCR was utilized to create a DNA fragment contain- but with the following changes: 1 mM each ATP, CTP, ing a T7 promoter upstream of the R1.1 RNA coding UTP and GTP, 6 mM guanosine were used. Also, sup- sequence suitable for cloning into a plasmid vector. The plemental addition of MgCl2 was omitted. fragment also contained a unique BbsI site so that tran- To prepare (50-32P) end-labeled R1.1 RNA, 200 pmol of scription by T7 RNA polymerase from a plasmid linear- guanosine-primed R1.1 RNA was incubated at 37 C for ized at the BbsI site would yield the R1.1 RNA. one hour in a reaction containing 1 Â polynucleotide Speci®cally, T7R1.1 (50 AAGAAGGTCA ATCATAAAGG kinase buffer (New England Biolabs), 400 pmol of CCACTCTTGC GAATGACCTT GAGTTTGTCC CTC [g-32P]ATP (ICN, end-labeling grade), 100 units of Super- TACTCCC TATAGTGAGT CGTATTA 30), which asin, and 80 units of T4 polynucleotide kinase (New Eng- encodes the coding strand of the T7 promoter upstream land Biolabs) in a total volume of 200 ml. The RNA was of the sequence complementary to the R1.1 RNA, was processed and gel puri®ed as described above except used as the template for PCR using the following pri- that visualization was by autoradiography. To prepare mers: R1.1EcoT7For (50 CGGAATTCTA ATACGACTCA uniformly 32P-labeled R1.1 RNA, transcription reactions CTATAGGG 30), and R1.1BbsBamRev (50 CGGGATCC- (100 m) contained 1Â transcription buffer, 1 mM each GA AGACCGAAGA AGGTCAATCA TAAAGGCC 30). ATP, CTP, and GTP, 0.1 mM UTP, 100 mCi [a-32P]UTP Oligodeoxyribonucleotides were obtained from Oligos, (ICN), 2.5 mg pT7R1.1/BbsI, 40 units Superasin, 25 units Etc. A 100 ml PCR reaction containing 1 pmol of T7R1.1 of yeast pyrophosphatase and 100 units of T7 RNA poly- and 100 pmol each of R1.1EcoT7For and R1.1BbsBamRev merase. Reactions were incubated for four hours at 37 C was used to generate suf®cient DNA for cloning. Plati- in a dry incubator. RNA was gel-puri®ed, processed, num pfx DNA polymerase (Gibco-BRL) was used in the and quanti®ed as described above. RNA was visualized reaction according to the manufacturer's instructions. by autoradiography. The 101 bp double-stranded PCR product was puri®ed Partial alkaline hydrolysis reactions to generate gel from a 2 % TAE agarose gel and digested with EcoRI and markers contained 25 mg of yeast tRNA (Gibco), 5 ml BamHI (New England Biolabs), which cut at each end of 0.5 M carbonate buffer (pH 9.5) and (50-32P) end-labeled the PCR product. The resulting fragment was ligated RNA in a total volume of 50 ml. The reaction was incu- into similarly cut pUC18, and used to transform the bated at 95 C for ®ve minutes (empirically optimized), DH5a strain of E. coli (Gibco-BRL). Clones were isolated, chilled on ice, then neutralized with 16 ml of 0.2 M HCl, screened by restriction digest and sequenced using stan- and an equal volume of formamide GLB was added. 77 dardmolecularcloningprocedures, togeneratethe Partial ribonuclease T1 digestion reactions contained 5 mg pT7R1.1 plasmid for in vitro transcription of the R1.1 of yeast tRNA, 10 mlofT1 buffer (20 mM sodium acetate RNA substrate. buffer (pH 5), 7 M urea, 1 mM EDTA), (50-32P) end- R1.1 RNA was synthesized by T7 RNA polymerase as labeled RNA, and 30 units of ribonuclease T1 (Roche) in run off transcripts from pT7R1.1 that was linearized by a volume of 14 ml. The reaction was incubated for digestion with BbsI (pT7R1.1/BbsI). The product was a 15 minutes at 50 C (empirically optimized) and 60 nt RNA molecule with the sequence: 50 GGGAGUA- quenched by the addition of 14 ml of ATA stop mix (for- GAG GGACAAACUC AAGGUCAUUC GCAAGA- mamide GLB/100 mM aurin tricarboxylic acid, 4:1). GUGG CCUUUAUGAU UGACCUUCUU 30. In vitro transcription reactions contained T7 RNA polymerase transcription buffer (Ambion), 16 mM MgCl2,5mM Analysis of RNase III reaction kinetics each ATP, CTP, UTP, GTP (Promega), 10 mg pT7R1.1/ BbsI, 160 units of Superasin (Ambion), 100 units of yeast Reactions contained RNase III and R1.1 RNA at the inorganic pyrophosphatase (Sigma) and 400 units T7 concentrations indicated in the Figure legends and typi- RNA polymerase (Ambion) in a volume of 400 ml. Tran- cally in a ®nal reaction volume of 50 ml. Substrate and scription reactions were incubated overnight at 37 Cin enzyme solutions were prepared separately. The sub- a dry incubator. Reactions were phenol/chloroform- strate solutions were prepared in Mes reaction buffer extracted and RNA was precipitated in the presence of (MRB; 50 mM Mes (Calbiochem) (pH as indicated in the 0.3 M sodium acetate and ethanol. The RNA was recov- Figure legends), 0.1 mM DTT, 0.1 mM EDTA, 0.01 % (v/ ered by centrifugation, dried, and resuspended in 15 ml v) NP40 and 194 mM KCl) and substrate RNA at twice of TE (10 mM Tris-HCl (pH 8.0), 1 mM EDTA). In prep- the desired ®nal concentration. The substrate solution RNase III Kinetics and Mechanism 37 was heated to 95 C for three minutes and cooled to reported is the pH of the ®nal reaction mixture taken at 37 C over ®ve minutes. After an additional ®ve minutes 37 C. Buffers for determination of the pH dependence of at 37 C, 50 mM MgCl2 in 1 Â MRB (preheated to 37 C) the single turnover rate were generated as follows. Suf®- was added to give a ®nal concentration of 10 mM MgCl2 cient components of 3 Â Mes core solution (240 mM KCl, and a ®nal volume of 25 ml. This solution was incubated 150 mM Mes, 0.3 mM DTT, 0.3 mM EDTA) were com- for an additional 20-60 minutes at 37 C prior to initiat- bined for a ®nal solution volume of 280 ml; however, the ing the reaction. Enzyme solutions were composed of working volume prior to adjusting pH was maintained

1 Â MRB, 10 mM MgCl2 and twice the ®nal enzyme con- at 200 ml. The pH values of aliquots of this solution centration desired for individual reactions. Enzyme were adjusted, at room temperature, through measured dilutions were made in ESB. Enzyme solution volumes additions of 10 M KOH, to give a range of pH values were 25 ml. The stock enzyme solution contributed no from pH 5 to pH 6.5. Since the K concentration in each more than 10 % of this volume and ESB was added to solution varied, KCl was added to normalize the K con- control for differences in volume of enzyme added. The centration to reach a concentration of 114 mM. Solutions enzyme solution was preheated for ®ve minutes at 37 C were ®ltered through 0.2 mm ®lters and stored at 20 C. prior to initiating the reaction. Reactions were initiated Each 3 Â Mes solution (corresponding to different pH by adding an equal volume (usually 25 ml) of substrate values) was used to create a 30 ml mock reaction that solution to the enzyme solution and mixed rapidly using was identical with actual reactions except that enzyme gentle pipetting. At each desired time-point, a 2 ml ali- storage buffer was used in place of enzyme. Each mock quot was removed from the reaction and added to a reaction was equilibrated at 37 C, and its pH deter- tube containing 4 ml of formamide GLB containing mined to give the ®nal pH of the reaction. 100 mM EDTA. The samples were loaded directly on a The following equations were used to ®t the pH 15 % polyacrylamide 7 M urea gel and electrophoresis dependenceoftheRNaseIIIcleavagereaction.42Fora continued until the bromophenol blue dye reached the single ionizable group the observed rate constant is bottom of the gel. Gels were dried and RNA bands were expected to vary with pH as described by: quanti®ed by phosphorimager analysis using Molecular Dynamics screens and instrumentation and using Image- kA kobs 3 Quant software. 1 10 pKapH The dependence of the single turnover rate constant on enzyme concentration was determined as follows. where kobs is the observed rate constant at a particular Reactions were performed as described above in MRB pH, kA is the activity of the deprotonated species and it is assumed that protonated species are inactive. For sys- (pH 5.4), 10 mM MgCl2, 0.2 nM R1.1 substrate and 1 nM to 2 mM RNase III ([E]/[S] 5 5). For each enzyme con- tems with two or more ionizable groups the following centration, the fraction of substrate reacted was quanti- equation was used: ®ed at each time-point, and these data were displayed as a plot of the fraction of the total substrate reacted versus kA kobs n 4 time. The data were ®t to a double-exponential function: 1 n 10 pKapH 10 pKapH

k k where kobs is the observed rate constant at a particular P=Stotal A Bea tCeb t 1 pH, kA is the activity of the deprotonated species and n is the number of ionizable groups with the same pK . where A represents the reaction endpoint, B and C rep- a resent amplitudes of the two phases, ka and kb represent the rate constants of the two phases, and t is the time of Isotope incorporation measurements incubation. The value for the fast phase, kobs, was plotted versus enzyme concentration and the data points are ®t Isotope incorporation reactions were performed in to equation (2) (see Results): MRB (pH 5.1) including 10 mM MgCl2. The reactions contained 150 mM R1.1 substrate and 5 mM RNase III in kobs Ek2= EKapp 2 a volume of 200 ml, and were incubated overnight at 37 C. The 18O content of the water was enriched to 18 where kobs is the observed pseudo-®rst-order rate con- 40 % using 97 % O water (Isotec). The reacted phos- 0 0 stant, [E] is the enzyme concentration, k2 is the cleavage phate was puri®ed in the form of 5 ,3 -guanosine-bis- rate constant at saturating enzyme concentration and phosphate (pGp) as follows. The 30 product of the reac- Kapp ((k2 k1)/k1) according to Scheme 1. The tion was puri®ed using denaturing PAGE, eluted and reported values for k2 and Kapp are the average of three precipitated as described above. The recovered product determinations. was digested to mononucleotides with RNase T2 Pulse-chase experiments were performed as described (Sigma). The pGp containing the reactive phosphate was above in MRB (pH 5.1) and 10 mM MgCl2, with 5 nM puri®ed away from nucleotide monophosphates by (50-32P) end-labeled R1.1 substrate and 2 mM RNase III. HPLC using a C18 reverse phase column (Phenomenex) Reactions were chased after two minutes with the run isocratically in 50 mM diisoprolyethylamine and 1 % addition of excess, unlabeled R1.1 RNA (40 mM) that (v/v) acetic acid (pH 4.3). The fractions containing pGp was prepared in parallel with labeled RNA substrate. were desalted using multiple rounds of lyophilization Parallel control reactions were carried out without a and resuspension in water. For mass spectrometric chase, and with the chase added at the start of the reac- analysis, pGp was resuspended in a mixture of water tion. Products were quanti®ed and the data analyzed as methanol and acetic acid (50:50:1, by volume) at a described above. concentration of 100 mM. Mass spectra were taken using The dependence of the single turnover rate constant a Finnigan TSQ quadrupole mass spectrometer with an on pH was determined using 5 nM R1.1 substrate and electrospray ion source. Standard pGp was synthesized 2 mM RNase III as described above. The reaction rate at by treating 30-guanosine monophosphate with poly- saturating enzyme was measured at six different pH nucleotide kinase and ATP and puri®ed as described values: 5.1, 5.3, 5.4, 5.6, 5.8, and 6.1. In all cases, the pH above. 38 RNase III Kinetics and Mechanism

Stopped-flow fluorescence analysis 7. Moller-Jensen, J., Franch, T. & Gerdes, K. (2001). Temporal translational control by a metastable RNA Fluorescence analysis was performed on an Applied structure. J. Biol. Chem. 276, 35707-35713. Photophysics p* 180 spectrometer. Reactions were per- 8. Abou Elela, S. & Ares, M., Jr (1998). Depletion of formed in MRB (pH 5.8 or 6.1) or Tris reaction buffer yeast RNase III blocks correct U2 30 end formation (MRB, but with 30 mM Tris-HCl (pH 7.3) in place of and results in polyadenylated but functional U2 Mes and 80 mM KCl). Reactions additionally contained snRNA. EMBO J. 17, 3738-3746. (10 mM) the indicated metal ion chloride. Temperature 9. Chanfreau, G., Legrain, P. & Jacquier, A. (1998). was regulated at 37.0(Æ0.1) C using a circulating water Yeast RNase III as a key processing enzyme in small bath. Reactions were initiated by injecting equal volumes nucleolar RNAs metabolism. J. Mol. Biol. 284, 975- of substrate and enzyme into a ¯ow cell. Concentrations 988. given for RNase III and R1.1 are the concentrations after 10. Abou Elela, S. A., Igel, H. & Ares, M., Jr. (1996). mixing. Excitation at 290 nm was performed using a RNase III cleaves eukaryotic preribosomal RNA at a variable wavelength monochromator and a 75 W mer- U3 snoRNP-dependent site. Cell, 85, 115-124. cury-xenon lamp. Fluorescence emission was detected at 11. Bernstein, E., Caudy, A. A., Hammond, S. M. & 90 using a 320 nm cut-off ®lter. The ¯uorescence time- Hannon, G. J. (2001). Role for a bidentate ribonu- courses shown are the average of between two and clease in the initiation step of RNA interference. six individual time-courses. Data exhibiting single- Nature, 409, 363-366. phase behavior were ®t to a single-exponential equation 12. Kharrat, A., Macias, M. J., Gibson, T. J., Nilges, M. function: & Pastore, A. (1995). Structure of the dsRNA binding domain of E. coli RNase III. EMBO J. 14, 3572- y A Bek t 5 3584. 13. St. Johnston, D., Brown, N. H., Gall, J. G. & Jantsch, where y is the relative ¯uorescence, A is the reaction M. (1992). A conserved double-stranded RNA-bind- endpoint, B represents the amplitude, k represents the ing domain. Proc. Natl Acad. Sci. USA, 89, 10979- exponential rate constant and t is the time after injection. 10983. Data exhibiting biphasic behavior were ®t to a double- 14. Mian, I. S. (1997). Comparative sequence analysis of exponential function according to equation (1), above. ribonucleases HII, III, II PH and D. Nucl. Acids Res. 25, 3187-3195. 15. Chelladurai, B., Li, H., Zhang, K. & Nicholson, A. W. (1993). Mutational analysis of a ribonuclease III processing signal. Biochemistry, 32, 7549-7558. 16. Krinke, L. & Wulff, D. L. (1990). The cleavage speci- Acknowledgments ®city of RNase III. Nucl. Acids Res. 18, 4809-4815. 17. Chelladurai, B. S., Li, H. & Nicholson, A. W. (1991). We thank Dr Mark Caprara, Dr Pieter deHaseth and the members of the Harris laboratory for advice and A conserved sequence element in ribonuclease III critical reading of the manuscript. We gratefully processing signals is not required for accurate in vitro acknowledge the assistance of Stephen Ingalls for mass enzymatic cleavage. Nucl. Acids Res. 19, 1759-1766. spectrometric analysis. We thank Dr Allen W. Nicholson 18. Zhang, K. & Nicholson, A. W. (1997). Regulation of for generously providing the bacterial strain used in ribonuclease III processing by double-helical puri®cation of wild-type E. coli RNase III. A.G.C. was sequence antideterminants. Proc. Natl Acad. Sci. supported by an HHMI graduate fellowship. USA, 94, 13437-13441. 19. Perez-Canadillas, J. M. & Varani, G. (2001). Recent advances in RNA-protein recognition. Curr. Opin. Struct. Biol. 11, 53-58. References 20. Williamson, J. R. (2000). Induced ®t in RNA-protein recognition. Nature Struct. Biol. 7, 834-837. 1. Robertson, H. D., Webster, R. E. & Zinder, N. D. 21. Leulliot, N. & Varani, G. (2001). Current topics in (1968). Puri®cation and properties of ribonuclease III RNA-protein recognition: control of speci®city and from Escherichia coli. J. Biol. Chem. 243, 82-91. biological function through induced ®t and confor- 2. Nicholson, A. W. (1999). Function, mechanism and mational capture. Biochemistry, 40, 7947-7956. regulation of bacterial ribonucleases. FEMS Micro- 22. Zheng, X. & Bevilacqua, P. C. (2000). Straightening biol. Rev. 23, 371-390. of bulged RNA by the double-stranded RNA-bind- 3. Dunn, J. J. & Studier, F. W. (1973). T7 early RNAs ing domain from the protein kinase PKR. Proc. Natl and Escherichia coli ribosomal RNAs are cut from Acad. Sci. USA, 97, 14162-14167. large precursor RNAs in vivo by ribonuclease 3. 23. Baltzinger, M. & Holler, E. (1982). Kinetics of acyl Proc. Natl Acad. Sci. USA, 70, 3296-3300. transfer ribonucleic acid complexes of Escherichia coli 4. Young, R. A. & Steitz, J. A. (1978). Complementary phenylalanyl-tRNA synthetase. A conformational sequences 1700 nucleotides apart form a ribonu- change is rate limiting in catalysis. Biochemistry, 21, clease III cleavage site in Escherichia coli ribosomal 2460-2467. precursor RNA. Proc. Natl Acad. Sci. USA, 75, 3593- 24. Ibba, M., Johnson, C. M., Hennecke, H. & Fersht, 3597. A. R. (1995). Increased rates of tRNA charging 5. Burgin, A. B., Parodos, K., Lane, D. J. & Pace, N. R. through modi®cation of the enzyme-aminoacyl- (1990). The excision of intervening sequences from adenylate complex of phenylalanyl-tRNA synthe- Salmonella 23 S ribosomal RNA. Cell, 60, 405-414. tase. FEBS Letters, 358, 293-296. 6. Regnier, P. & Grunberg-Manago, M. (1990). RNase 25. Baldwin, G. S., Sessions, R. B., Erskine, S. G. & III cleavages in non-coding leaders of Escherichia coli Halford, S. E. (1999). DNA cleavage by the EcoRV transcripts control mRNA stability and genetic restriction endonuclease: roles of divalent metal ions expression. Biochimie, 72, 825-834. in speci®city and catalysis. J. Mol. Biol. 288, 87-103. RNase III Kinetics and Mechanism 39

26. Li, H. L., Chelladurai, B. S., Zhang, K. & Nicholson, substrate binding and cleavage by Escherichia coli A. W. (1993). Ribonuclease III cleavage of a bacterio- ribonuclease III. Nucl. Acids Res. 29, 1915-1925. phage T7 processing signal. Divalent cation speci- 44. Kirby, A. J. & Younas, M. (1970). The reactivity of ®city, and speci®c anion effects. Nucl. Acids Res. 21, phosphate esters. Reactions of diesters with nucleo- 1919-1925. philes. J. Chem. Soc. (B), 1165-1172. 27. Wilcox, D. E. (1996). Binuclear metallohydrolases. 45. Herschlag, D. & Jencks, W. P. (1989). Phosphoryl Chem. Rev. 96, 2435-2458. transfer to anionic oxygen nucleophiles. Nature of 28. Cowan, J. A. (1997). Metal-mediated hydrolysis of the transition state and electrostatic repulsion. J. Am. biological phosphate esters. J. Biol. Inorg. Chem. 2, Chem. Soc. 111, 7587-7596. 168-176. 46. Hengge, A. C., Aleksandra, T. E. & Cleland, W. W. 29. Steitz, T. A. & Steitz, J. A. (1993). A general two- (1995). Studies of transition-state structures of phos- metal-ion mechanism for catalytic RNA. Proc. Natl phoryl transfer reactions of phosphodiesters of Acad. Sci. USA, 90, 6498-6502. p-nitrophenol. J. Am. Chem. Soc. 117, 5919-5926. 30. Beese, L. S. & Steitz, T. A. (1991). Structural basis 47. Knowles, J. R. (1980). Enzyme-catalyzed phosphoryl for the 30-50 exonuclease activity of Escherichia coli transfer reactions. Annu. Rev. Biochem. 49, 877-919. DNA polymerase I: a two metal ion mechanism. 48. Hengge, A. C. (2001). Isotope effects in the study of EMBO J. 10, 25-33. enzymatic phosphoryl transfer reactions. FEBS 31. Mol, C. D., Kuo, C. F., Thayer, M. M., Cunningham, Letters, 501, 99-102. R. P. & Tainer, J. A. (1995). Structure and function 49. Burbaum, J. J., Raines, R. T., Albery, W. J. & of the multifunctional DNA-repair enzyme exonu- Knowles, J. R. (1989). Evolutionary optimization of clease III. Nature, 374, 381-386. the catalytic effectiveness of an enzyme. Biochemis- 32. Vipond, I. B., Baldwin, G. S. & Halford, S. E. (1995). try, 28, 9293-9305. Divalent metal ions at the active sites of the EcoRV 50. Wu, D. & Boyer, P. D. (1986). Bound adenosine 0 0 and EcoRI restriction endonucleases. Biochemistry, 5 -triphosphate formation, bound adenosine 5 - 34, 697-704. diphosphate and inorganic phosphate retention, and 33. Shan, S., Yoshida, A., Sun, S., Piccirilli, J. A. & inorganic phosphate oxygen exchange by chloro- plast adenosinetriphosphatase in the presence of Herschlag, D. (1999). Three metal ions at the active 2 2 site of the Tetrahymena group I ribozyme. Proc. Natl Ca or Mg . Biochemistry, 25, 3390-3396. Acad. Sci. USA, 96, 12299-12304. 51. Herschlag, D. & Jencks, W. P. (1990). Catalysis of 34. Hos®eld, D. J., Guan, Y., Haas, B. J., Cunningham, the hydrolysis of phosphorylated pyridines by Mg(OH) : a possible model for enzymatic phos- R. P. & Tainer, J. A. (1999). Structure of the DNA phoryl transfer. Biochemistry, 29, 5172-5179. repair enzyme endonuclease IV and its DNA com- 52. Dahm, S. C., Derrick, W. B. & Uhlenbeck, O. C. plex: double-nucleotide ¯ipping at abasic sites and (1993). Evidence for the role of solvated metal three-metal-ion catalysis. Cell, 98, 397-408. hydroxide in the hammerhead cleavage mechanism. 35. Li, H. & Nicholson, A. W. (1996). De®ning the Biochemistry, 32, 13040-13045. enzyme binding domain of a ribonuclease III proces- 53. Cornish-Bowden, A. J. (1995). Fundamentals of sing signal. Ethylation interference and hydroxyl Enzyme Kinetics, pp. 180-181, Portland Press Ltd, radical footprinting using catalytically inactive London. RNase III mutants. EMBO J. 15, 1421-1433. 54. Stanford, N. P., Halford, S. E. & Baldwin, G. S. 36. Sun, W. & Nicholson, A. W. (2001). Mechanism of (1999). DNA cleavage by the EcoRV restriction endo- action of Escherichia coli ribonuclease III. Stringent nuclease: pH dependence and proton transfers in chemical requirement for the glutamic acid 117 side- catalysis. J. Mol. Biol. 288, 105-116. 2 chain and Mn rescue of the Glu117Asp mutant. 55. Viadiu, H. & Aggarwal, A. K. (1998). The role of Biochemistry, 40, 5102-5110. metals in catalysis by the restriction endonuclease 37. Kranz, J. K. & Hall, K. B. (1998). RNA binding med- BamHI. Nature Struct. Biol. 5, 910-916. iates the local cooperativity between the beta-sheet 56. Feig, A. L. & Uhlenbeck, O. C. (1999). The role of and the C-terminal tail of the human U1A RBD1 metal ions in RNA biochemistry. In The RNA World protein. J. Mol. Biol. 275, 465-481. (Gesteland, R. F., Cech, T. R. & Atkins, J. F., eds), 38. Dunn, J. J. (1976). RNase III cleavage of single- 2nd edit., pp. 287-320, Cold Spring Harbor stranded RNA. Effect of ionic strength on the ®del- Laboratory Press, Cold Spring Harbor, NY. tiy of cleavage. J. Biol. Chem. 251, 3807-3814. 57. Schweisguth, D. C., Chelladurai, B. S., Nicholson, 39. Chen, S. M., Takiff, H. E., Barber, A. M., Dubois, A. W. & Moore, P. B. (1994). Structural characteriz- G. C., Bardwell, J. C. & Court, D. L. (1990). ation of a ribonuclease III processing signal. Nucl. Expression and characterization of RNase III and Acids Res. 22, 604-612. Era proteins. Products of the rnc operon of Escheri- 58. Connolly, B. A., Eckstein, F. & Pingoud, A. (1984). chia coli. J. Biol. Chem. 265, 2888-2895. The stereochemical course of the restriction endonu- 40. Nicholson, A. W., Niebling, K. R., McOsker, P. L. & clease EcoRI-catalyzed reaction. J. Biol. Chem. 259, Robertson, H. D. (1988). Accurate in vitro cleavage 10760-10763. by RNase III of phosphorothioate-substituted RNA 59. Grasby, J. A. & Connolly, B. A. (1992). Stereochemi- processing signals in bacteriophage T7 early mRNA. cal outcome of the hydrolysis reaction catalyzed by Nucl. Acids Res. 16, 1577-1591. the EcoRV restriction endonuclease. Biochemistry, 31, 41. Northrop, D. B. (1998). On the meaning of Km and 7855-7861. V/K in enzyme kinetics. J. Chem. Ed. 75, 1153-1157. 60. Bagshaw, C. R., Trentham, D. R., Wolcott, R. G. & 42. Fersht, A. (1998). Structure and Mechanism in Protein Boyer, P. D. (1975). Oxygen exchange in the Science, pp. 155-168, W.H. Freeman & Co., New gamma-phosphoryl group of protein-bound ATP York. during Mg2-dependent adenosine triphosphatase 43. Calin-Jageman, I., Amarasinghe, A. K. & Nicholson, activity of myosin. Proc. Natl Acad. Sci. USA, 72, A. W. (2001). Ethidium-dependent uncoupling of 2592-2596. 40 RNase III Kinetics and Mechanism

61. Abend, A., Garrison, P. N., Barnes, L. D. & Frey, 70. Baldwin, G. S., Vipond, I. B. & Halford, S. E. (1995). P. A. (1999). Stereochemical retention of the con- Rapid reaction analysis of the catalytic cycle of the ®guration in the action of Fhit on phosphorus-chiral EcoRV restriction endonuclease. Biochemistry, 34, substrates. Biochemistry, 38, 3668-3676. 705-714. 62. Cho, Y. K., Matsunaga, T. O., Kenyon, G. L., 71. Bycroft, M., Grunert, S., Murzin, A. G., Proctor, M. Bertagnolli, B. L. & Cook, P. F. (1988). Isotope & St Johnston, D. (1995). NMR solution structure of exchange as a probe of the kinetic mechanism a dsRNA binding domain from Drosophila Staufen of pyrophosphate-dependent phosphofructokinase. protein reveals homology to the N-terminal domain Biochemistry, 27, 3320-3325. of ribosomal protein S5. EMBO J. 14, 3563-3571. 63. Zhang, Z. Y., Malachowski, W. P., Van Etten, R. L. 72. Ramos, A., Grunert, S., Adams, J., Micklem, D. R., & Dixon, J. E. (1994). Nature of the rate-determining Proctor, M. R., Freund, S. et al. (2000). RNA recog- steps of the reaction catalyzed by the Yersinia nition by a Staufen double-stranded RNA-binding protein-tyrosine phosphatase. J. Biol. Chem. 269, domain. EMBO J. 19, 997-1009. 8140-8145. 73. Handa, N., Nureki, O., Kurimoto, K., Kim, I., 64. Walker, H., Ferretti, J. A. & Stadtman, T. C. (1998). Sakamoto, Y., Muto, Y. & Yokoyama, S. (1999). Isotope exchange studies on the Escherichia coli sele- Structural basis for recognition of the tra mRNA nophosphate synthetase mechanism. Proc. Natl Acad. precursor by the Sex-lethal protein. Nature, 398, 579- Sci. USA, 95, 2180-2185. 585. 65. Kim, D. E. & Patel, S. S. (2001). The kinetic pathway 74. Deo, R. C., Bonanno, J. B., Sonenberg, N. & Burley, of RNA binding to the Escherichia coli transcription S. K. (1999). Recognition of polyadenylate RNA by termination factor Rho. J. Biol. Chem. 276, 13902- the poly(A)-binding protein. Cell, 98, 835-845. 13910. 75. Reyes, C. M. & Kollman, P. A. (2000). Structure and 66. Lago, H., Parrott, A. M., Moss, T., Stonehouse, N. J. & Stockley, P. G. (2001). Probing the kinetics of for- thermodynamics of RNA-protein binding: using mation of the bacteriophage MS2 translational oper- molecular dynamics and free energy analyses to ator complex: identi®cation of a protein conformer calculate the free energies of binding and confor- unable to bind RNA. J. Mol. Biol. 305, 1131-1144. mational change. J. Mol. Biol. 297, 1145-1158. 67. Silverman, S. K. & Cech, T. R. (1999). RNA tertiary 76. Jen-Jacobson, L. (1997). Protein-DNA recognition folding monitored by ¯uorescence of covalently complexes: conservation of structure and binding attached pyrene. Biochemistry, 38, 14224-14237. energy in the transition state. Biopolymers, 44, 153- 68. Qin, P. Z. & Pyle, A. M. (1997). Stopped-¯ow ¯uor- 180. escence spectroscopy of a group II intron ribozyme 77. Sambrook, J., Fritsh, E. F. & Maniatis, T. (1989). reveals that domain 1 is an independent folding unit Molecular Cloning: A Laboratory Manual, 2nd edit., with a requirement for speci®c Mg2 ions in the ter- Cold Spring Harbor Laboratory Press, Cold Spring tiary structure. Biochemistry, 36, 4718-4730. Harbor, NY. 69. Bevilacqua, P. C., Li, Y. & Turner, D. H. (1994). 78. Nashimoto, H. & Uchida, H. (1985). DNA sequen- Fluorescence-detected stopped ¯ow with a pyrene cing of the Escherichia coli ribonuclease III gene and labeled substrate reveals that guanosine facilitates its mutations. Mol. Gen. Genet. 201, 25-29. docking of the 50 cleavage site into a high free 79. Gill, S. C. & von Hippel, P. H. (1989). Calculation of energy binding mode in the Tetrahymena ribozyme. protein extinction coef®cients from amino acid Biochemistry, 33, 11340-11348. sequence data. Anal. Biochem. 182, 319-326.

Edited by J. Karn

(Received 22 October 2001; received in revised form 10 January 2002; accepted 10 January 2002)