Probing the kinetics of formation of the bacteriophage MS2 translational operator complex: identification of a protein conformer unable to bind RNA

doi:10.1006/jmbi.2000.4355 available online at http://www.idealibrary.com on J. Mol. Biol. (2001) 305, 1131±1144

Probing the Kinetics of Formation of theBacteriophage MS2 Translational Operator Complex:Identification of a Protein Conformer Unable toBind RNA

Hugo Lago, Andrew M. Parrott, Tim Moss, Nicola J. Stonehouseand Peter G. Stockley*

Astbury Centre for StructuralMolecular Biology, Faculty ofBiological Sciences, Universityof Leeds, Leeds LS2 9JT, UK

Present address: A. M. Parrott, DBiochemistry & Molecular Biology,School, Newark, NJ 07109, USA.

Abbreviations used: TR, translatcoat protein; 2AP, 2-deoxy, 2-amino

E-mail address of the [email protected]

0022-2836/01/051131±14 $35.00/0

We have investigated the kinetics of complex formation between bac-teriophage MS2 coat protein subunits and synthetic RNA fragmentsencompassing the natural translational operator site, or the consensussequences of three distinct RNA aptamer families, which are known tobind to the same site on the protein. Reactions were assayed usingstopped-¯ow ¯uorescence spectroscopy and either the intrinsic trypto-phan ¯uorescence of the protein or the signals from RNA fragments site-speci®cally substituted with the ¯uorescent adenosine analogue 20-deoxy,2-aminopurine. The kinetics observed were independent of the ¯uoro-phore being monitored or its position within the complex, indicating thatthe data report global events occurring during complex formation. Com-petition assays show that the complex being formed consists of a singlecoat protein dimer and one RNA molecule. The binding reaction is atleast biphasic. The faster phase, constituting 80-85 % of the amplitude, isa largely diffusion driven RNA-protein interaction (k1 � 2 � 109 Mÿ1 sÿ1).The salt dependence of the forward reaction and the similarities of theon-rates of lower-af®nity RNA fragments are consistent with a diffusion-controlled step dominated by electrostatic steering. The slower phase isindependent of reactant concentration, and appears to correspond to iso-merisation of the coat protein subunit(s) prior to RNA binding(kiso � 0.23 sÿ1). Measurements with a coat protein mutant (Pro78Asn)show that this phase is not due to cis-trans isomerisation at this residue.The conformational changes in the protein ligand during formation of anRNA-protein complex might play a role in the triggering of capsid self-assembly and a model for this is discussed.

# 2001 Academic Press

Keywords: bacteriophage MS2; RNA; ¯uorescence; 2-aminopurine;stopped-¯ow kinetics; aptamers
*Corresponding author
Introduction

Sequence-speci®c RNA-protein complexes playvital regulatory roles throughout cellular metab-olism (Draper, 1995). Despite their importance,there is relatively little information on the kinetics

epartment ofNew Jersey Medical

ional repression; CP,purine.

ing author:

of complex formation (Metzger et al., 1997; Draper,1999). Here, we report our results from stopped-¯ow ¯uorescence measurements of RNA-proteininteraction in one of the paradigmatic model sys-tems of sequence-speci®c recognition, namely theMS2 bacteriophage translational repression (TR)complex (Stockley et al., 1994; Witherell et al.,1991).

RNA bacteriophages have evolved complexregulatory mechanisms to control the temporaltranslation of their polycistronic genomic RNAs.One such event is the translational repression ofthe phage replicase cistron by coat protein (CP)molecules. This occurs via sequence-speci®c inter-

# 2001 Academic Press

http://www.idealibrary.com

mailto:[email protected]

1132 Stopped-¯ow Kinetics of RNA-Protein Interaction

action of a phage coat protein dimer and a stem-loop operator site of just 19 nt, located in the 50¯anking sequence of the replicase gene. This com-plex sequesters the ribosome binding site and theinitiating AUG, preventing translation of thedownstream message. In addition, this complexmarks the genomic RNA for speci®c packaginginto the phage shell by promoting self-assembly,i.e. it serves as an assembly initiation complex(Ling et al., 1970; Beckett & Uhlenbeck, 1988;Beckett et al., 1988). The isolated 19 nt operatorfragment (Figure 1) contains all the information topromote sequence-speci®c interaction with the coatprotein in vitro. It has been studied intensively inorder to understand its key recognition features(Carey & Uhlenbeck, 1983; Talbot et al., 1990;Stockley et al., 1995; Witherell et al., 1991; Peabody,1993; Peabody & Lim 1996; Lago et al., 1998).

Our understanding of this recognition event atthe molecular level is primarily based on thewealth of recent high-resolution X-ray crystallogra-phy. Structures are now available for the wild-type, T � 3 phage particle (ValegaÊrd et al., 1990;Golmohammadi et al., 1993); RNA-free capsids,including proteins mutant at the RNA binding site,produced by over-expression of a recombinant coatprotein gene in the absence of genomic phageRNA (van den Worm et al., 1998); a non-assem-bling coat protein dimer (Ni et al., 1995); and thestructures of several operator capsids in which the19 nt operator fragment or sequence variants havebeen soaked into crystals of RNA-free capsids,where they make sequence-speci®c contacts toevery coat protein dimer in the protein shell(ValegaÊrd et al., 1994, 1997; Grahn et al., 1999). Theresults with natural operators and variants havealso been extended by determination of the struc-

Figure 1. Structures of the RNAs used. The secondary stshown, from left to right: (i) the wild-type operator (TR) and1998). The positions where 20-deoxy, 2-aminopurine was intrin bold. Numbering is relative to the start of the MS2 replicaon X-ray crystal structures (ValegaÊrd et al., 1994, 1997; Conveaptamers contain a non-Watson-Crick base-pair (boxed). The

tures of a series of RNA aptamers selected from30N sequence pools (Hirao et al., 1998), which alsobind to the same site on the coat protein (Converyet al., 1998; Rowsell et al., 1998). Comparison ofthese two types of complexes has revealed the keymolecular recognition elements. Solution structuresof the unbound RNA operator, calculated fromNMR data (Borer et al., 1995), suggest that thespecies bound by the protein is at best only aminor component of a set of conformers in equili-brium, con®rming earlier predictions based onchemical reactivity of the various functional groups(Talbot et al., 1991). It is clear that complex for-mation with coat protein involves conformationalchanges from the dominant form of the RNA frag-ment in solution (Parrott et al., 2000).

Coat proteins in the T � 3 phage shell show theexpected three conformers for a quasi-equivalentstructure, the conformational changes being princi-pally restricted to the loop of polypeptide connect-ing the F and G b-strands, the FG-loop. RNAbinding by disassembled coat protein dimers in sol-ution leads to changes in the intrinsic ¯uorescenceintensity of the protein, suggesting that this is corre-lated with conformational changes (Beckett &Uhlenbeck, 1988). Here, we have used this ¯uor-escence signal and that from operator derivativessite-speci®cally substituted with 20-deoxy, 2-amino-purine (2AP; Parrott et al., 2000) to investigate thekinetics of RNA-protein complex formation.

Results

Preparation of the reaction components

In 20 mM acetic acid and at protein concen-trations below micromolar levels, the wild-type

ructures of the RNA stem-loops used in this study arethe aptamers; (ii) F5; (iii) F6; and (iv) F7 (see Hirao et al.,oduced to replace the original nucleotide are highlightedse gene (�1). The secondary structures shown are basedry et al., 1998; Rowsell et al., 1998). Both F5 and F7 RNAstructure of 2-aminopurine is shown inset.

Stopped-¯ow Kinetics of RNA-Protein Interaction 1133

coat protein is stable as aggregates no higher thandimers (Beckett & Uhlenbeck, 1988). RNA oligonu-cleotides used in binding experiments included thewild-type TR sequence and the consensussequences of three aptamer families which all bindto the same site on the protein (see Figure 1, Hiraoet al., 1998; Convery et al., 1998; Rowsell et al.,1998). The aptamers are numbered relative to thewild-type operator for ease of comparison. Thesites of substitution with 20-deoxy, 2-aminopurine,e.g. ÿ4(2AP)TR with 2AP substituted at positionÿ4, are all adenosine bases in the unmodi®edsequence and, at least for the TR derivatives, canbe substituted by 20-deoxynucleotides withoutsigni®cantly affecting the apparent af®nity for coatprotein (Stockley et al., 1995). The apparent af®-nities (Kd) for coat protein binding were deter-mined using the assay based on the intrinsic¯uorescence of the coat protein described (Beckett& Uhlenbeck, 1988; Lago et al., 1998; Parrott et al.,2000) (see Table 1).

The stopped-flow assay

The RNA and protein samples were prepared in0.2� TMK buffer, a standard used previously forbinding measurements in this system (Talbot et al.,1990). For the determination of forward rate con-stants, equal concentrations of coat protein dimerand RNA (one mole of CP2 binds one mole ofRNA (Spahr et al., 1969; Beckett & Uhlenbeck,1988)) were used for single-mix stopped-¯owexperiments, and the changes in ¯uorescence inten-sity observed.

Association kinetics

Typical ¯uorescence traces for the reaction moni-tored by either the signal from the protein or fromthe RNA derivatives ÿ4(2AP)TR and ÿ4(2AP)F5are shown in Figure 2. Over the ®rst 30 seconds,the reaction was clearly biphasic as judged by bothtypes of probe. The ®rst phase, constitutingroughly 80-85 % of the total amplitude, being lar-gely complete by 40 ms; the second much slowerphase occurring over 30 seconds. Under the con-ditions used here (0.2� TMK buffer and 200 nMreactant concentration), �30-35 % of the total

Table 1. The results of the competition experiments

Complex Competitor kÿ1 (sÿ

CP2: ÿ4(2AP)TR TR or F5 2.2 (�21CP2: F6 ÿ10(2AP)TR 0.16 (�1CP2: ÿ10(2AP)F6 TR or F5 0.11 (�1CP2: F7 ÿ10(2AP)TR 0.52 (�1P78N: ÿ4(2AP)TR TR or F5 1.9 (�14

kÿ1 is the back rate for the dissociation of CP2:RNA complexes dbrium dissociation constant of the complexes between MS2 coat prolibrium ¯uorescence measurements (Lago et al., 1998; Parrott et aextended incubation to allow equilibration the effect of the proteinratio kÿ1/k1 does not correspond to Kd, although the relative differen

change in ¯uorescence amplitude upon bindingoccurs in the dead time of the instrument. Thebaselines correspond to either the intrinsic protein¯uorescence before binding (as in the cases of TR,and the aptamers F5, F6, F7) or to the ¯uorescenceof the 2AP derivatives (such as ÿ4(2AP)TR) beforebinding to MS2 coat protein.

The amount of amplitude change within thedead time reduced to 4 % or less with the shorterRNAs, such as F6 or F7, which are 14 nt long andtherefore contain fewer negative charges than TRor F5 (Figure 3). In addition, increasing the saltconcentration of the binding buffer also reducedthe amount of unobserved amplitude to <4 %(compare Figures 2(b) and 3(d)). An assumptionwas therefore made that the amplitude changeswithin the dead time correspond to the earlieststages of the observed fast phase, rather than to adistinct kinetic phase, allowing normalisation ofthe data. The strong dependence on ionic strengthis typical of the binding rates of some protein-pro-tein and protein-DNA interactions (Karshikov et al.,1992; Schreiber & Fersht, 1993; Pontius, 1993)where association constants approach the diffusionlimit. Accordingly, we ®tted the data for the fasterphase to a second order reaction model (equations(1) and (2)) using the baselines as the start ofthe fast phase. The data for the slower phasewere ®tted to a ®rst order (single exponential)model. The data ®t well to these kinetic models(Figures 2-4), yielding a second order rate constantfor the biomolecular reaction of k1 � 1 � 109 to2 � 109 Mÿ1 sÿ1 (for ÿ4(2AP)TR), which is consist-ent with a diffusion limited process. The slowerphase is suggestive of a conformational change oran isomerisation event with a kiso � 0.23 sÿ1

(�4.25 %). The rates of both phases were the samewhen measured either by changes in the ¯uor-escence of CP or the ÿ4TR probe, suggesting thatthe same bimolecular process of binding RNA toprotein was being monitored in both cases.

The dependence of the rates on reagent concen-trations between 25 nM and 200 nM were thenexamined (Figure 4). The apparent rate of the ®rstphase was concentration dependent, whereas therate of the second phase was not. Similar resultsfor both phases were observed when usingÿ10(2AP)TR as probe over similar concentration

1) kÿ1/k1 (nM) Kd (nM)

%) 1.03 (�25 %) 3.297 %) 0.31 (�21 %) 1.540 %) 0.21 (�15 %) 0.695 %) 2.08 (�16 %) 12.3%) 0.91 (�23 %) 2.85

etermined using the competitor RNAs shown. Kd is the equili-teins (wild-type or P78N) and the RNAs, determined from equi-l., 2000). Under these conditions of protein molar excess andisomerisation would be expected to be minimal. As a result theces in these values between samples are similar.

Figure 2. Stopped-¯ow kinetics of MS2 coat protein binding to the RNA fragments. The kinetic traces show abiphasic time-dependent decrease (a) in intrinsic protein ¯uorescence (excitation at 280 nm and emission >330 nm),or increase (b) in ÿ4(2AP)TR or (c) in ÿ4(2AP)F5 ¯uorescence (excitation at 307 nm and emission >330 nm), aftermixing 200 nM (®nal concentration) CP2 and 200 nM TR, ÿ4(2AP)TR or ÿ4(2AP)F5, respectively. The buffer usedwas 0.2� TMK. Continuous lines are the non-linear least-squares ®t to a second order reaction for the fast phase(equation (1)), and to a single exponential for the slow phase. The break points, here and throughout, indicate thetime limits of the ®ts to each phase and the small panels below each time-course show the residuals of these ®ts.


Figure 3. In¯uence of RNA chain length and ionic strength on the initial rate. The Figure shows the traces for thefast phase (k1) of the interaction of CP2 with the aptamers (a) F5 (18 nt), (b) F6 (14 nt) and (c) F7 (14 nt) in 0.2� TMK.(d) The kinetic trace for binding to the ÿ4(2AP)TR derivative in higher ionic strength buffer (20 mM Tris, 256 mMKCl, 2 mM MgCl2) is also displayed. The kinetics were followed by observing changes in intrinsic protein ¯uor-escence (for F5, F6 and F7) or by changes in the ¯uorescence of the 2(AP) moiety in the ÿ4(2AP)TR derivative. Thecontinuous lines correspond to the non-linear ®t to equation (1). For comparison, the equivalent traces for the inter-action of coat protein with TR (19 nt) or ÿ4(2AP)TR in low ionic strength buffer (0.2� TMK) are shown in Figure 2(a)and (b).


ranges. The remaining probes, i.e. F5, F6, F7,ÿ10(2AP) F5, and ÿ10(2AP)F6 (Figure 1), were stu-died at 200 nM ®nal concentration and the bindingdata analysed as described above (Table 2). Withthe exceptions of the smaller F6 and F7 RNA frag-ments, the values of k1 were very similar and diffu-sion limited. The values of kiso were essentiallyidentical for all the RNAs. The slower phase wasmore dif®cult to follow using protein ¯uorescence,due to low signal to noise ratios. A similar problemwas encountered for the ÿ10(2AP)F6 probe, whichonly shows a small ¯uorescence intensity changeupon binding CP.

It appears that all the reagents show similar kin-etics of binding, suggesting that similar events,independent of the position of the ¯uorophore, aretaking place in each case. It is tempting to specu-late that the changes correspond to capsid assem-bly triggered by the RNA-protein interaction.

However, this appears not to be the case for anumber of reasons. Firstly, the slower phasedetected in the ¯uorescence experiment is not con-centration dependent, which assembly would haveto be. Secondly, the faster phase shows bimolecularkinetics, suggestive of the formation of a simpleRNA-CP2 complex rather than a larger proteinaggregate, which would show more complex kin-etics. Finally, transmission electron microscopy(TEM) of samples of the coat protein-TR mixture,at the same concentrations used in the stopped-¯ow ¯uorimeter, together with controls of capsidand dissociated coat protein alone, are consistentwith the absence of capsid formation under theseconditions for far longer than the data acquisitionperiod in the ¯uorimeter (data not shown). Theseexperiments, therefore, reveal events taking placeduring the formation of the initial sequence-speci®cRNA-protein complex.

Figure 4. Concentration dependence of the kineticphases. The rates of the fast (k1, main graph) and slow(kiso, inset) phases of CP2 binding to ÿ4(2AP)TR werecalculated from the change in the RNA ¯uorescence.Experiments were performed using equal concentrationsof both reagents over a concentration range for k1 from25 nM to 200 nM and for kiso from 25 nM to 800 nM.Each datum point is the estimated value of kapp for eachconcentration, obtained by non-linear least-squares ®tusing equation (1). The straight line is the best ®t toequation (2), with k1 � 2.07 � 109 Mÿ1 sÿ1. The insetshows the rate of the slow phase. Error bars representthe standard deviation of the estimates of k1 correspond-ing to at least three different experiments.


Ionic strength dependence of the interaction

The effect of ionic strength on the rate constantswas investigated by varying the KCl concentrationof the buffer used. The experiments were carriedout by monitoring the ¯uorescence signal of twoTR derivatives, ÿ4(2AP)TR and ÿ10(2AP)TR, andthe intrinsic ¯uorescence of the protein with theunderivatised TR fragment. The results for

Table 2. Kinetics of the RNA-protein interaction

Protein RNA k1

Wild-type TRÿ4(2AP)TRÿ10(2AP)TRF5ÿ4(2AP)F5ÿ10(2AP)F5F6ÿ10(2AP)F6F7

P78N TRÿ4(2AP)TRÿ10(2AP)TR

The calculated values of k1 and kiso for the interaction of coat proteorder rate constant (see Materials and Methods, equation (1)) corresprotein (wild-type or P78N) to the RNA fragments. kiso is the ®rst orobserved in the association reactions. Errors, here and in Table 1,measurements and each measurement is the average of a maximum

ÿ4(2AP)TR are shown in Figure 5(a). Theseare typical of those seen with the other RNAs.The bimolecular step was strongly affected byionic strength (k1 varying between 2 � 109 to7.5 � 106 Mÿ1 sÿ1 ) as the concentration of KClincreased from 0 to 0.5 M, with �80 % of thereduction happening before the concentrationreached 0.15 M. As pointed out above, this is typi-cal of reactions where association constantsapproach the diffusion limit (Karshikov et al., 1992;Schreiber & Fersht, 1993; Pontius, 1993). In thesesystems, the interacting components are chargedand it has been proposed that the electrostatic®elds around the molecules orient the binding sur-faces prior to collision (electrostatic steering). Uponbinding, reorientation leading to ®nal docking hap-pens in milliseconds or less (Berg & von Hippel,1985; von Hippel & Berg, 1989). The MS2-TR sys-tem is consistent with this view, the TR beingnegatively charged and the RNA-binding surfaceof the CP2 being more positively charged than therest of the molecule.

In contrast, as shown inset in Figure 5(a), theslower phase displays little variation upon increas-ing the ionic strength, which together with the lackof a protein concentration effect (Figure 4), is con-sistent with a reaction step which does not directlyinvolve either an RNA-protein or a protein-proteininteraction.

Temperature dependence of the interaction

The temperature dependence of both phaseswas studied in the range from 3 to 39 �C, in 0.2�TMK, using ÿ4(2AP)TR as a probe. The observedsecond order rate constant became too fast above26 �C to be measured reliably, so only data in therange 3-25 �C were used. The results can be seenin Figure 5(b). Activation parameters were calcu-lated for the bimolecular phase from thesedata, giving an Arrhenius activation energy,Ea � 57.78(�8.75) kJmolÿ1 and a �H��

� 10ÿ9 (Mÿ1sÿ1) kiso (sÿ1)

1.98 (�18 %) 0.23 (�11 %)2.13 (�14 %) 0.23 (�7 %)1.29 (�19 %) 0.24 (�6 %)1.61 (�21 %) 0.21 (�12 %)1.32 (�15 %) 0.24 (�11 %)1.92 (�16 %) 0.26 (�8 %)0.51 (�13 %) n.d.0.52 (�11 %) n.d.0.25 (�13 %) 0.21 (�11 %)2.13 (�19 %) 0.23 (�14 %)2.09 (�14 %) 0.23 (�8 %)1.12 (�9 %) 0.22 (�7 %)

in with the different RNA fragments are shown. k1 is the secondponding to the fast phase observed in the binding of MS2 coatder rate constant (equation (2)) corresponding to the slow phasecorrespond to the standard deviation of at least three differentof 25 runs in the stopped-¯ow apparatus.

Figure 5. Salt concentration and temperature depen-dence of the interaction. (a) Salt concentration depen-dence of the rate constants for coat protein binding toÿ4(2AP)TR are shown. Final concentrations after mixingwere 200 nM for both reactants in 20 mM Tris-HCl(pH 7.4), 2 mM MgCl2, 0-500 mM KCl. Buffer solutionswith increasing ionic strength were made volumetricallyusing appropriate KCl stocks solutions. (b) Temperaturedependence of the rate constants for the reactiondescribed above. The data were ®tted by non-linearregression to equation (3). In each graph the main panelshows k1 and the inset kiso.


55.24(�8.45) kJmolÿ1, �S�� 122.56(�15.8) JKÿ1

molÿ1 and a value for �G�� 18.23(�3.59)kJmolÿ1 at 25 �C.

These results suggest that the association rateconstant for the formation of the complex CP2:TRis enthalpically determined with a positive entropyterm. The entropy term for a bimolecular reactionwould be expected to be negative (Espenson,1995), but there are other examples of protein-protein, diffusion limited interactions, which aresimilarly enthalpically driven with positive entropy

terms (e.g. the colicin E9-IM9 interaction, seeWallis et al., 1995a,b). Release of water moleculesupon binding of CP2 and TR, or aptamers, maycontribute to the increase in entropy observed. Inaddition, the low value for the free energy of acti-vation, �G�� 18.53 kJmolÿ1, is consistent withpredictions for a diffusion limited reaction(�G�� 16 kJmolÿ1; Gutfreund, 1972; Geeves &Gutfreund, 1982; Pontius, 1993).

For the slower phase, it was possible to followthe reaction easily between 3 and 36 �C. Above thistemperature, the ¯uorescence change upon bindingbecame very small and was indistinguishable fromthe noise. The results are shown in the inset toFigure 5(b). An Arrhenius activation energy(Ea) for the slow phase of 73.12(�5.13) kJmolÿ1

was calculated from the data, and a �H�� 70.64(�6.35) kJmolÿ1, �S�� ÿ7.07(�0.84) JKÿ1

molÿ1 and �G�� 72.78(�8.25) kJmolÿ1 at 25 �C.These data show that there is little loss in entropyin the slower phase of the CP2-TR interaction.

Origin of the slow phase

As demonstrated above, the slower phase of theCP2:TR interaction is independent of reactant con-centration and of ionic strength, both features con-sistent with a ®rst order isomerisation that occursbefore RNA binding. If this is correct, then anexcess of CP2 over TR should reduce the ampli-tude, but not the rate, of this phase. Conversely, anexcess of TR over the CP2 concentration shouldhave no effect. The results of such titration exper-iments (Figure 6) show that under conditions ofmolar excess of ÿ4(2AP)TR the slow step is presentas usual with similar amplitude (�15-20 % of thetotal normalised intensity change), but in molarexcess of CP2, the slow step disappears; the totalamplitude changes being similar in both exper-iments as expected. Similar results (not shown)were obtained with the ÿ10(2AP)TR derivative.These results con®rm that the conformationalchange occurs in the MS2 coat protein before itbinds to the RNA.

Nature of the slow phase

One possible protein isomerisation event such asthe one described above would be proline cis-transisomerisation. The rate and Arrhenius activationenergy of the slower phase (0.23 sÿ1 and73 kJ molÿ1) are both well within the reportedranges for such reactions (Schmid, 1986; Kiefhaber,1995). From X-ray crystallographic studies, it isknown that the MS2 CP subunit in the T � 3capsid exists in three quasi-equivalent conformers,A, B and C, giving rise to two types of dimer, A/Band C/C (ValegaÊrd et al., 1990). The primarydifferences between them are in the conformationof the loop connecting the F and G b-strands(FG-loop), which is in extended conformations inthe A and C subunits, but is folded back towardsthe core of the subunit in the B conformer. Proline

Figure 6. Probing the origin of the slower phase. Themain Figure shows the stopped-¯ow kinetic tracederived from changes in ¯uorescence of 2AP in theregion of the slow phase after mixing a tenfold molarexcess of ÿ4(2AP)TR (2000 nM) with coat protein dimer(200 nM) (residual plot shown below). The observedrate was kiso � 0.24 sÿ1. The inset displays a similar traceafter mixing a tenfold molar excess of coat proteindimer (2000 nM) with ÿ4(2AP)TR (200 nM). The signalwas essentially constant under these conditions.


78 (P78) exists as the trans isomer in A andC monomers, but is cis in B subunits(Golmohammadi et al., 1993). Thus one possibleexplanation for the isomerisation seen in the kinetictraces might be interconversion of the P78 residue.Dissociation of the capsid into CP2 in acetic acid(see Materials and Methods) might allow somethese residues to re-equilibrate leading to theappearance of the slow phase in the interactionwith the operator RNA.

Previously, we characterised a site-directedmutant protein, P78N, which is still able to formT � 3 capsids, and also bind TR in vivo and in vitro(Stonehouse et al., 1996; Lago et al., 1998; Hill et al.,1997). The N78 residue is trans in all subunits. Wetherefore repeated the stopped-¯ow measurementsusing the P78N mutant protein in order to explorewhether Pro78 isomerisation was the rate-limitingstep in the slower phase. The kinetics of RNAbinding by this mutant to the 2-aminopurinederivatives ÿ4(2AP)TR and ÿ10(2AP)TR were stu-died. The results for the ÿ4(2AP)TR are shown inFigure 7. No difference was observed between thekinetics of the mutant and wild-type protein(Table 2), suggesting that isomerisation of P78 isnot responsible for the slower phase.

Further kinetic characterisation ofthe interaction

The RNA-protein interaction was further charac-terised by obtaining values for the back rates. Dis-

sociation kinetics have often been used to studydisplacement reactions in which a ligand is chasedwith an excess of a second ligand with equal, ormore often higher, af®nity (Eccleston, 1987;Bagshaw & Trentham, 1974). Here, the dissociationrates of the CP2 complexes of ÿ4(2AP)TR,ÿ10(2AP)F6, F6 and F7 were studied by chasing apreformed complex with a 10 to 25-fold molarexcess of TR, F5 or ÿ10(2AP)TR as competitors. Ineach case, either the decrease in ¯uorescence signalfollowing the displacement of a (2AP)RNA deriva-tive, such as ÿ4(2AP)TR by TR or F5, or theincrease in ¯uorescence signal following the displa-cement of an unlabelled aptamer by ÿ10(2AP)TR,was used to monitor the dissociation.

For the CP2: ÿ4(2AP)TR complex, for example,CP2 and ÿ4(2AP)TR were mixed and incubated for25 ms, 100 ms or ®ve minutes (in the case of theCP2:ÿ4(2AP)TR complex, a seven day incubationtime was also tested) and subsequently chasedwith an equal volume of 10 to 25-fold molar excessof TR or F5 aptamer. For longer incubations(5®ve minutes) the CP2 and the RNA were mixedmanually and then incubated for the set timebefore chasing the complex formed with TR or F5in the stopped-¯ow apparatus set in single mixingmode. For shorter incubations (25 or 100 ms), thestopped-¯ow apparatus was set in double sequen-tial mixing mode. The CP2 and ÿ4(2AP)TR were®rst premixed in the instrument and then chasedwith TR or F5. The delay (incubation time)between premixing and the chase was set to be 25or 100 ms. Results of the competition experimentsfor CP2:ÿ4(2AP)TR complexes can be seen inFigure 8. Experiments were conducted using 400,200, or 50 nM concentrations of CP2: ÿ4(2AP)TRcomplexes, and 10 to 25-fold molar excess of com-petitors. No apparent reactant concentrationdependence was observed and the same resultswere obtained with both short and long preincu-bation times. Similar results were obtained for thecoat protein complexes with ÿ10(2AP)F6, F6 andF7 (see Table 1).

The amplitude changes between formation anddissociation of the complexes were nearly the sameand were independent of the identity of the com-petitor used. The bulk of the dissociation curvecould be ®tted to a single exponential decay,which was independent of the pre-incubation timebetween 25 ms and ®ve minutes. (The one exper-iment with a seven day incubation of ÿ4(2AP)TRalso did not alter the result.) These results suggestthat the off-rate corresponds to the dissociationrate constant, kÿ1, of the RNA-protein complexes.A second, very slow phase of magnitude equal to5-10 % of the total amplitude of the reaction, wasapparent after completion of this bulk dissociationreaction. With longer pre-incubation times, someself-association between complexes may occur togenerate higher order species which may be inter-mediates in the formation of capsids. In the compe-tition assay described these high order complexesmight need to dissociate before the displacement

Figure 7. RNA binding by theP78N mutant protein. Stopped-¯owkinetics of the P78N CP2 binding toÿ10(2AP)TR. The kinetic tracesshow the biphasic time-dependentincrease in ÿ10(2AP) ¯uorescence(excitation at 307 nm and emission>330 nm) after mixing 200 nMP78N CP2 and 200 nM ÿ10(2AP)TR(®nal concentrations). The calcu-lated rates are: k1 � 1.12 � 109 Mÿ1

sÿ1 and kiso � 0.23 sÿ1. The detailsof the plots and data analysis areas described in the legend toFigure 2.


reaction can take place. If that were the case, veryshort pre-incubation times (25-100 ms) would beexpected to reduce the amplitude of this phasebecause fewer higher order complexes would beformed under these conditions. This is not the casehowever. The minor phase was also unaffected bythe length of the pre-incubation. An alternativeexplanation is that at the higher RNA concen-

Figure 8. Estimating the dissociation rate. The kineticsof chasing the coat protein:ÿ4(2AP)TR complex with a25-fold molar excess of TR are shown for an equimolar,single-mixing, stopped-¯ow experiment in which400 nM of CP2 and 400 nM of ÿ4(2AP)TR were mixedmanually, incubated for ®ve minutes, and then chasedwith an excess of TR. Data were ®tted to a single expo-nential with a dissociation rate constant of 2.5 sÿ1. Theupper trace corresponds to the ¯uorescence signal of thecomplex, in the absence of the chaser (TR), whilst thelower trace corresponds to the ¯uorescence signal of theRNA in the absence of either the protein or the chaser.The change in amplitude upon dissociation was, within5-10 %, the same as the difference in ¯uorescence signalsbetween the complex (upper trace) and the free RNA(lower trace).

trations used in the competition assays somedimerisation of the competitor RNA moleculesoccurs, as has been reported previously (Borer et al.,1995).

Discussion

We have shown that similar interaction kineticsare observed for all the RNA derivatives and thetwo coat protein sequences tested, and that theposition and nature of the ¯uorophore has no sig-ni®cant effect on the apparent rates. This suggeststhat, despite their structural differences and theirvariation in af®nity for coat protein (Parrott et al.,2000), all of the RNAs bind by a similar mechan-ism. The ¯uorescence changes, therefore, monitorthe global molecular recognition reaction ratherthan individual structural rearrangements in thecomponents of the complex. The kinetics of themajor phase of the forward reaction ®t well to asimple second order model in the range of concen-trations tested, as expected for a bimolecular pro-cess. Competition assays have allowed us to probethe reverse reaction for some of the RNAs, whichappears to have essentially the same total ampli-tude change as observed for formation of therespective complexes. Since this reaction ®ts wellto a single exponential decay and is indifferent toextended pre-incubation times up to ®ve minutes,it appears that no signi®cant assembly beyond theinitial RNA-coat protein dimer complex occursunder these conditions. This is consistent with theresults of light-scattering and TEM experimentsduring a reassembly time-course, in which speciesthe size of capsids are not observed in the mixturesuntil at least ten minutes after mixing (data notshown). Note, the ¯uorescence amplitudes ana-lysed here only refer to the active protein andRNA molecules in the mixture.

The complex being formed in the fast phase ofthe reaction appears to be equivalent to ``complexI'' described by other groups (Spahr et al., 1969;


Knolle & Hohn, 1975; Beckett & Ulhenbeck, 1988).The results are consistent with a one-step bindingprocess of the form:

CP2 � TRÿ!k1

ÿkÿ1CP2 : TR

where CP2 represents the coat protein dimer.The measured values of k1 are close to those

expected for a diffusion-controlled reaction and,not surprisingly, show little variation between thedifferent RNAs. The diffusion-controlled upperlimit for collision between a coat protein dimerand a TR operator fragment can be calculatedusing the von Smoluchowski's equation (Daune,1999). Estimating the sum of the reaction radii forthe coat protein dimer and TR molecules to be�100 AÊ , and using values for the diffusion con-stants at 20 �C of dimers and TR RNA determinedby analytical ultracentrifugation (data not shown)of 6.44 � 10ÿ11 m2 sÿ1 and 1.22 � 10ÿ10 m2 sÿ1,respectively, yields a value for the diffusion-con-trolled upper limit of 1.4 � 108 Mÿ1 sÿ1. Theobserved rates for all the RNA probes tested in0.2� TMK buffer vary between 0.25 � 109 to2.0 � 109 Mÿ1 sÿ1 , suggesting that k1 is diffusionlimited. Consistent with this view is the fact thatthe rate constant is strongly dependent on ionicstrength, as expected for a bimolecular reaction inwhich electrostatic steering plays a crucial role(Berg & von Hippel, 1985; von Hippel & Berg,1989). Additionally, the free energy of activation issmall, again consistent with a reaction whichapproaches the diffusion limited rate.

The dissociation rate constants calculated fromthe competition assay data show variation betweenthe different RNA ligands tested, as expected fromtheir differing af®nities. In particular, there are sig-ni®cant increases in off-rate for the ÿ4(2AP)TRcompared to the aptamers F6, ÿ10(2AP)F6 and F7.These results are consistent with the crucial role ofthe adenosine at ÿ4 in making intermolecular con-tacts with the protein, i.e. the hydrogen bondsbetween threonine A45 and N7; serine A47 to N1,and a double contact from the exo-cyclic 6-aminogroup to threonine A45 and threonine A59(ValegaÊrd et al., 1994, 1997; van den Worm et al.,1998). The F7 aptamer also has a higher off-rate(Table 2) than the F6 aptamer consistent with itslack of the ÿ10 adenosine residue, which contrib-utes the following intermolecular contacts; hydro-gen bonds via N1 and the 6-amino group tothreonine B45, and via N3 to serine B47.

The kinetic data for the forward reaction alsoexhibited a much slower phase with a rate thatwas independent of both reactant and salt concen-trations. The titration experiments (Figure 6) estab-lished that this phase was due to events takingplace in the protein component prior to binding toRNA, rather than rearrangement of the RNA-pro-tein complex once formed. Similar conclusionswhere reached by Jia et al. (1996), for the inter-

action of T7 RNA polymerase with its promoter.One interpretation could be that this phase re¯ectsdimerisation of a fraction of monomeric coat pro-tein, which may form during the acetic acid treat-ment. However, this seems to be unlikely, sincedimerisation would be expected to be both concen-tration and salt dependent. Ultracentrifugation ofthe coat protein stock solutions in dilute acetic acidalso suggested that only dimers (Mw,app � 28,300)were present (data not shown). Beckett &Uhlenbeck (1988) report that in the assay buffer(TMK) the Kd for dimer formation is between0.1-1.0 nM. This suggests that at the lowest concen-trations used here (25 nM), at most, only 6 % of thedimers would be dissociated and under the stan-dard assay conditions that ®gure would drop to�2 %. In a few experiments not reported here theupper concentration limit of the reactants wasraised to 2000 nM, and these assays showed simi-lar phases to those reported above. These resultswould appear to rule out dimerisation as the causeof the slow phase.

Alternatively, it could be argued that coat pro-tein subunits exist in 20 mM acetic acid (pH 4) inconformational equilibrium, between native-likedimers, which bind RNA rapidly, and non-native-like dimers which must refold before bindingRNA. The CP2 solution was incubated in 0.2�TMK (pH 7.4) for at least one hour before use.These are conditions in which the MS2 capsid isvery stable. If partially unfolded coat protein sub-units exist at pH 4, they most likely would refoldto native like dimers at the higher pH. Thissuggests that the isomerisation detected here isrelated to the RNA binding event rather thanbeing an artefact of acid dissociation of the capsid.There is now widespread evidence that confor-mational changes commonly occur in the proteinligand of an RNA-protein complex as RNA-bind-ing proceeds (Draper, 1999). Here, the role that theTR RNA plays in triggering phage capsid self-assembly is suggestive that the conformationalchange/refolding might be related to initiation ofcapsid assembly. The simplest model, which inte-grates all the kinetic data, reported here would be:

CP�2ÿ!kiso

ÿkÿisoCP2 � TR

ÿ!k1

ÿkÿ1CP2 : TR

where CP2* represents a non-binding isomer of thecoat protein.

Whatever the origin of the isomerisation, X-raycrystallographic data provide a plausible expla-nation of why a fraction of the coat protein couldbe unable to bind the RNA. When operator frag-ments are soaked into RNA-free T � 3 shells theybind to every coat protein dimer. At C/C dimersboth orientations of the RNA occur but at A/Bdimers the orientation is unique, the loop regionbinding to the A subunit and the ÿ10A/bulgeregion to the B subunit. Modelling a rigid bodyrotation of the RNA in such complexes suggests


that in the opposite orientation on the A/B dimerthe RNA would clash sterically with the backboneof alanine B80 (ValegaÊrd et al., 1997). This obser-vation is further supported by experiments inwhich RNA fragments encompassing just the loopregion and two base-pairs of the stem are soakedinto capsid crystals. Such fragments bind only to Aand C subunits and not to B (Grahn et al., 1999).Assuming that the only protein conformers of anysigni®cance are A, B or C-like species, and thatthere is an equal probability that any subunit willadopt one of these, we can predict the amount ofB/B-like dimers, which would be expected to beunable to bind RNA, to be one-ninth (�1/3 � 1/3)or 11 % of the total coat protein subunits.This value is close to the percentage of the totalamplitude change occupied by the slower phase(�15-20 %).

Table 1 shows the ratio of kÿ1/k1 of selectedRNAs and the Kd values obtained by equilibrium¯uorescence titrations (Lago et al., 1998; Parrottet al., 2000). In a one-step binding mechanism theratio of kÿ1/k1 would be expected to be the sameas the Kd. In this case the values are similar andthe relative differences between samples are alsovery similar, but the presence of the isomerisationstep precludes direct comparison. In addition,there are experimental problems associated withthis approach to Kd measurements, since the highaf®nities make precise estimations dif®cult.

These results may be the key to understandingthe molecular mechanism underlying the assemblypathway of the capsid, since it is the ®rst directlink between RNA-binding and coat protein con-formation. Initially, it was tempting to speculatethat the conserved P78 residue, which is at thebeginning of the FG-loop region and is a cis pep-tide in B subunits but the more normal trans at Aand C subunits, was the molecular switch control-ling assembly (ValegaÊrd et al., 1990;Golmohammadi et al., 1993). Subsequently, weshowed that the N78P substituent also supportedformation of T � 3 shells with all bonds trans(Stonehouse et al., 1996), although in an infectiousclone this mutant was dead (Hill et al., 1997).A plausible explanation for the requirement forproline at position 78 in vivo is that it creates partof the binding site for the maturation protein(Paranchych, 1975), the A protein; B-type P78 resi-dues clustering along the pore at the particle 5-foldaxis. The similar behaviour of the N78 variant inthe stopped-¯ow assay suggests, however, thatthis proline residue at least is not rate limiting.Conversion of B subunits to A or C conformers, aswell as requiring the proline isomerisation, alsodisrupts two hydrogen bonds to a charged residue(E76) (Golmohammadi et al., 1993), consistent withthe large activation energy of this step. Furtherwork with mutant proteins is underway to identifythe origin of the slower phase.

An important issue raised by these results is therole, if any, of RNA binding in capsid assembly.We and others (Beckett & Ulhenbeck, 1988) have

shown that the RNA operator stem-loop stimulatesassembly from acid dissociated subunits at sub-stoichiometric ratios, although the protein alonewill also support assembly at higher concen-trations. This is consistent with the earlier obser-vations that in vivo and in vitro there is packagingspeci®city (Ling et al., 1970; Knolle & Hohn, 1975).Such sequence-speci®c RNA interactions would berequired by assembly mechanisms involvingnucleation, which have been proposed before forthe RNA phages (Spahr et al., 1969; Knolle &Hohn, 1975; Beckett & Ulhenbeck, 1988) and forT � 3 RNA plant viruses (Sorger et al., 1986). Sincefor the RNA phages, T � 3 shell assembly can bestimulated by increasing the protein concentrationin the absence of the TR RNA or simply by theaddition of polyanions (Knolle & Hohn, 1975), thecomplex formed here may function by stabilising aprotein sub-assembly which is an obligate inter-mediate on the pathway to T � 3 shell formation.Conformational changes in the FG-loop appear tooffer an obvious pathway by which this mightoccur. In the fr system, however, it has been shownthat T � 3 shells can still form with subunits hav-ing FG-loops truncated by four residues at the endof the loop (Axblom et al., 1998). Reassembly isalso largely unaffected by a number of amino acidsubstitutions in this region (Stockley et al., 1993,1994). Deletion of single residues, which has theeffect of putting neighbouring loops out of registerwith each other is, however, very deleterious forassembly. It is therefore unclear whether FG-loopconformations play any role on the pathway tocapsid formation. Examination of the X-ray struc-ture of the protein shell suggests that the appropri-ate curvature of the ®nal capsid can be speci®edby contacts around the particle quasi-3-fold axis(Liljas, personal communication), but again thisdoes not imply that such a species is on the assem-bly pathway. Characterisation of RNA-proteincomplexes with higher protein stoichiometries anddirect observation of the behaviour of the FG-loopsduring RNA binding will be required to clarifythese issues and such experiments are beingplanned.

Materials and Methods

Purification of proteins

Recombinant RNA-free MS2 capsids were prepared asdescribed (Mastico et al., 1993; Stonehouse & Stockley,1993). Protein concentrations were measured by UVspectrophotometry, using extinction coef®cients at280 nm estimated as described by Gill & von Hippel(1989). Disassembled coat protein was prepared by aceticacid extraction of the capsids as described by Sugiyamaet al. (1967), and kept on ice in 20 mM acetic acid(pH � 3.4) until use.

Synthesis/purification of RNA operators

Solid-phase synthesis, deprotection and puri®cationof RNA operators was carried out as described (Murray


et al., 1994). The modi®ed base precursor,50-O-(dimethoxytrityl)-N-2-(dimethylaminomethylidene)-deoxypurineriboside-30-O-[(2-cyanoethyl)-N,N-isopropyl]-phosphoramidite was purchased from Glen Research(Virginia). Three derivatives of the wild-type operator(TR) sequence (19 nt) were synthesised, as were fourderivatives of the F5 aptamer (18 nt) and two derivativesof the F6 aptamer (14 nt) (Figure 1). Derivativesÿ4(2AP)TR and ÿ10(2AP)TR refer to the wild-typeoperator sequence containing a single 20-deoxy, 2-amino-purine substitution at adenosine positions ÿ4 or ÿ10,respectively. Derivatives ÿ4(2AP)F5 and ÿ10(2AP)F5,and ÿ10(2AP)F6 refer to equivalent substitutions withinthe F5 and F6 aptamer sequences, respectively. Concen-trations were determined by UV spectrophotometry,using estimated nucleotide and dinucleotide extinctioncoef®cients at 260 nm (see Puglisi & Tinoco, 1989) ande260 � 1000 Mÿ1 cmÿ1 for 2AP in aqueous solutions,pH > 3.63 (Fox et al., 1958).

The known secondary structures (Rowsell et al., 1998)of wild-type operator (TR), and the F5 and F6 aptamersare illustrated in Figure 1. All RNAs were characterisedby electrospray mass spectrometry, which con®rmedtheir expected molecular masses. Their apparentaf®nities, Kd, for coat protein were determined using the¯uorescence assay described (Lago et al., 1998; Parrottet al., 2000) (see Table 1).

Stopped-flow fluorescence measurements

All experiments were carried out using an AppliedPhotophysics SX17.MV stopped-¯ow instrument. Trypto-phan ¯uorescence was excited at 280 nm, while 2APderivatives were excited at 307 nm. Excitation andemmision slits were set to 2 mm. Baselines of all reac-tants and buffers were taken before and after each exper-iment, and data were corrected accordingly. For eachexperiment, 12 traces each consisting of 4000 datumpoints were averaged and analysed using the softwareincluded with the instrument and the commercial pro-gram SigmaPlot. Temperature was controlled via an NBSwater-bath and was 20 �C, unless stated otherwise.

Protein samples for stopped-¯ow experiments wereprepared by dilution of aliquots from a 20 mM aceticacid stock of dissociated MS2 CP into 0.2� TMK buffer(100 mM Tris-HCl (pH 7.4), 80 mM KCl, 10 mM MgCl2).Samples were then incubated at �20 �C for an hourbefore use. MS2 protein concentrations were kept belowmicromolar level (dimer) to avoid formation of higheroligomers (Beckett & Uhlenbeck, 1988). RNAs werediluted into 0.2� TMK buffer from aqueous stocks andtheir concentrations adjusted using A260.

Data analysis

The binding curves were analysed as two phases suf®-ciently well resolved that each could be treated indepen-dently. A fraction of the intensity change observedoccurs in the dead time of the instrument and wasassumed to be a part of the faster kinetic phase. Datawere normalised to 100 % of the total intensity change.The association rate was too fast to allow use of pseudo-®rst order conditions. The stopped-¯ow traces weretherefore ®tted directly to a second order irreversiblekinetic equation (1) (Eccleston, 1987; Capellos & Bielski,1972; Espenson, 1995), which is appropriate for theRNA-protein interaction under the experimental con-ditions used. The apparent rate of reaction, kapp, under

the special conditions in which the initial concentrationsof each reactant are equal, is related to the second orderrate constant by equation (2) (Eccleston, 1987):

A� B!k1AB

where [A] � [B] at t � 0; A � CP2; B � RNA

F�t� � Fo � F1�k1�A0�t=�1� k1�A0�t�� 1�

kapp � k1�A0� �2�(i.e. the inverse of the reaction half-time) where F is the¯uorescence signal at time t, Fo is the initial ¯uorescencevalue, before mixing reactants, F1 is the total change in¯uorescence for the reaction, k1 is the true second orderconstant, and [A0] is the initial concentration of eachreactant.

For the slower kinetic phase, and for dissociation reac-tions, single exponentials were used to analyse the data.The statistical quality of the ®ts throughout were judgedby the randomness of the residual plots.

Thermodynamic parameters were calculated by non-linear least-squares ®tting using SigmaPlot (Jandel). TheArrhenius activation energy, Ea, and the transition stateparameters were calculated from equations (3) and (4),and their standard derivatives (Walmsley, 1996):

k � A exp�ÿEa=RT� �3�

k � �kbT=h� exp�ÿ�S��=R� exp��H�� RT� �4�where k is the rate constant (k1 or kiso), A is the pre-exponential factor of the Arrhenius equation, Ea is theArrhenius activation energy, R is the gas constant, Tis the temperature, h is Planck's constant, and kb isBoltzmann's constant.

Acknowledgements

We thank Professors Tony Clarke (Bristol/ImperialCollege) and Nick Price (Glasgow), and Professor SheenaRadford, for their encouragement and helpful commentson the kinetic analysis of the RNA-protein interaction.We acknowledge the work of Dr David Scott, who setup the stopped-¯ow assay. We thank Andy Baron andDr Alison Ashcroft for ultracentrifugation and massspectrometry data, and Jenny Baker for expert technicalassistance. This work was supported, in part, by the UKBBSRC and MRC, and by the Leverhulme Trust. Workin the Astbury Centre is also supported by the North ofEngland Structural Biology Centre (NESBiC).

References

Axblom, C., Tars, K., Fridborg, K., Orna, L., Bundule,M. & Liljas, L. (1998). Structure of phage fr capsidswith a deletion in the FG loop: implications forviral assembly. Virology, 249, 80-88.

Bagshaw, C. R. & Trentham, D. R. (1974). The character-isation of myosin-product complexes and ofproduct-release steps during the magnesiumion-dependent adenosine triphosphatase reaction.Biochem. J. 141, 331-349.


Beckett, D. & Uhlenbeck, O. C. (1988). Ribonucleopro-tein complexes of R17 coat protein and a transla-tional operator analog. J. Mol. Biol. 204, 927-938.

Beckett, D., Wu, H.-N. & Uhlenbeck, O. C. (1988). Rolesof operator and non-operator RNA sequences inbacteriophage R17 capsid assembly. J. Mol. Biol.204, 939-947.

Berg, O. G. & von Hippel, H. (1985). Diffusion-controlled macromolecular interactions. Annu. Rev.Biophys. Biophys. Chem. 14, 131-160.

Borer, P. N., Lin, Y., Wang, S., Roggenbuck, M. W.,Gott, J. M., Ulhenbeck, O. C. & Pelczer, I. (1995).Proton NMR and structural features of a 24-nucleo-tide RNA hairpin. Biochemistry, 34, 6488-6503.

Capellos, C. & Bielski, B. H. J. (1972). Kinetics Systems,John Wiley & Sons, Inc., New York.

Carey, J. & Uhlenbeck, O. C. (1983). Kinetic andthermodynamic characterization of the R17 coatprotein-ribonucleic acid interaction. Biochemistry, 22,2610-2615.

Convery, M. A., Rowsell, S., Stonehouse, N. J.,Ellington, A. D., Hirao, I., Murray, J. B., Peabody,D. S., Phillips, S. E. V. & Stockley, P. G. (1998).Crystal structure of an RNA aptamer-protein com-plex at 2.8 AÊ resolution. Nature Struct. Biol. 5, 133-139.

Draper, D. E. (1995). Protein-RNA recognition. Annu.Rev. Biochem. 64, 593-620.

Draper, D. E. (1999). Themes in RNA-protein recog-nition. J. Mol. Biol. 293, 255-270.

Daune, M. (1999). Molecular Biophysics: Structures inMotion, Oxford University Press, Oxford, UK.

Eccleston, J. F. (1987). Stopped-¯ow spectrophotometrictechniques. In Spectrophotometry and Spectro¯uorime-try: A Practical Approach (Harris, D. A. & Bashford,C. L., eds), pp. 137-164, IRL Press, Oxford, UK.

Espenson, J. H. (1995). Chemical Kinetics and ReactionMechanism, McGraw Gill, New York, USA.

Fox, J. J., Wempen, I., Hampton, A. & Doerr, I. L.(1958). Thiation of nucleosides. I. synthesisof 2-amino-6 mercapto-9-b-d-ribofuranosylpurine(``thioguanosine'') and related purine nucleosides.J. Am. Chem. Soc. 80, 1669-1675.

Geeves, M. A. & Gutfreund, H. (1982). The use of press-ure perturbations to investigate the interaction ofrabbit muscle myosin subfragment 1 with actin inthe presence of MgADP. FEBS Letters, 140, 355-359.

Gill, C. & von Hippel, P. H. (1989). Calculation of pro-tein extinction coef®cients from amino acidsequence data. Anal. Biochem. 182, 319-326.

Golmohammadi, R., ValegaÊ rd, K., Fridborg, K. & Liljas,L. (1993). The re®ned structure of bacteriophageMS2 at 2.8 AÊ resolution. J. Mol. Biol. 234, 620-639.

Grahn, E., Stonehouse, N. J., Murray, J. B., van denWorm, S., ValegaÊrd, K., Fridborg, K., Stockley, P. G.& Liljas, L. (1999). Crystallographic studies of RNAhairpins in complexes with recombinant MS2 cap-sids: implications for binding requirements. RNA, 5,131-138.

Gutfreund, H. (1972). Enzymes: Physical Principles, Wiley-Interscience, London, UK.

Hill, R. H., Stonehouse, N. J., Fonseca, S. A. & Stockley,P. G. (1997). Viable phage from cDNA fragments ofthe MS2 genome: analysis of MS2 coat proteinmutants in vivo reveals a possible function for theconserved proline-78. J. Mol. Biol. 266, 1-7.

Hirao, I., Spingola, M., Peabody, D. & Ellington, A. D.(1998). The limits of speci®city: an experimental

analysis with RNA aptamers to MS2 coat proteinvariants. Mol. Divers. 4, 75-89.

Jia, Y., Kumar, A. & Patel, S. S. (1996). Equilibrium andstopped-¯ow kinetics of interaction between T7RNA polymerase and its promoters measured byprotein and 2-aminopurine F¯uorescence changes.J. Biol. Chem. 271, 30451-30458.

Karshikov, A., Bode, W., Tulinsky, A. & Stone, S. R.(1992). Electrostatic interactions in the association ofproteins: an analysis of the thrombin-hirudin com-plex. Protein Sci. 1, 727-735.

Kiefhaber, T. (1995). Protein folding kinetics. MethodsMol. Biol. 40, 313-341.

Knolle, P. & Hohn, T. (1975). Morphogenesis of RNAphages. In RNA Phages (Zinder, N., ed.), pp. 147-202, Cold Spring Harbor Laboratory Press, ColdSpring Harbor, NY.

Lago, H., Fonseca, S. A., Murray, J. B., Stonehouse, N. J.& Stockley, P. G. (1998). Dissecting the key recog-nition features of the MS2 bacteriophage transla-tional repression complex. Nucl. Acids Res. 26, 1337-1344.

Ling, C. M., Hung, P. P. & Overby, L. R. (1970). Inde-pendent assembly of Qb and MS2 phages in doublyinfected E. coli. Virology, 40, 920-929.

Mastico, R. A., Talbot, S. J. & Stockley, P. G. (1993).Multiple presentation of foreign peptides on thesurface of an RNA-free spherical bacteriophage cap-sid. J. Gen. Virol. 74, 541-548.

Metzger, A. U., Bayer, P., Willbold, D., Hoffmann, S.,Frank, R. W., Goody, R. S. & Rosch, ?. (1997). Theinteraction of HIV-1 Tat(32-72) with its target RNA:a ¯uorescence and nuclear magnetic resonancestudy. Biochem. Biophys. Res. Commun. 201, 820-828.

Murray, J. B., Collier, A. K. & Arnold, J. R. P. (1994). Ageneral puri®cation procedure for chemically syn-thesized oligoribonucleotides. Anal. Biochem. 218,177-184.

Ni, C.-Z., Syed, R., Kodandapani, R., Wickersham, J.,Peabody, D. S. & Ely, K. R. (1995). Crystal structureof the MS2 coat protein dimer: implications forRNA binding and virus assembly. Structure, 3, 255-263.

Parancych, W. (1975). Attachment, ejection and pen-etration stages of the RNA phage infectious process.In RNA Phages (Zinder, N., ed.), pp. 85-112, ColdSpring Harbor Laboratory Press, Cold SpringHarbor, NY.

Parrott, A. M., Lago, H., Adams, C. J., Ashcroft, A. E.,Stonehouse, N. J. & Stockley, P. G. (2000). RNAaptamers for the MS2 bacteriophage coat proteinand the wild-type RNA operator have similar sol-ution behaviour. Nucl. Acids Res. 28, 489-497.

Peabody, D. S. (1993). The RNA binding site of bacterio-phage MS2 coat protein. EMBO J. 12, 595-600.

Peabody, S. & Lim, F. (1996). Complementation of RNAbinding site mutations in MS2 coat protein hetero-dimers. Nucl. Acids Res. 24, 2352-2359.

Pontius, B. W. (1993). Close encounters: why unstruc-tured, polymeric domains can increase rates ofspeci®c macromolecular association. Trends Biochem.Sci. 18, 181-186.

Puglisi, J. D. & Tinoco, I., Jr (1989). Absorbance meltingcurves of RNA. Methods Enzymol. 180, 304-325.

Rowsell, S., Stonehouse, N. J., Convery, M. A., Adams,C. J., Ellington, A. D., Hirao, I., Peabody, D. S.,Stockley, P. G. & Phillips, S. E. V. (1998). Crystalstructures of a series of RNA aptamers complexed


to the same protein target. Nature Struct. Biol. 5,970-975.

Schmid, F. X. (1986). Fast-folding and slow-foldingforms of unfolded proteins. Methods Enzymol. 131,70-82.

Schreiber, G. & Fersht, A. R. (1993). Interaction of bar-nase with its polypeptide inhibitor Bbarstar studiedby protein engineering. Biochemistry, 32, 5145-5150.

Sorger, P. K., Stockley, P. G. & Harrison, S. C. (1986).The structure and assembly of turnip crinkle virusII: mechanism of assembly in vitro. J. Mol. Biol. 191,639-658.

Spahr, P. F., Farber, M. & Gesteland, R. F. (1969). Bind-ing site on R17 RNA for coat protein. Nature, 222,455-458.

Stockley, P. G., Stonehouse, N. J., Walton, C., Walters,D. A., Medina, C., Macedo, M. B., Hill, H. R.,Goodman, S. T. S., Talbot, S. J., Tewary, H. K.,Golmohammadi, R., Liljas, L. & ValegaÊrd, K. (1993).Molecular mechanism of RNA-phage morphogen-esis. Biochem. Soc. Trans. 21, 627-633.

Stockley, P. G., Stonehouse, N. J. & ValegaÊrd, K. (1994).The molecular mechanism of RNA phage morpho-genesis. Int. J. Biochem. 26, 1249-1260.

Stockley, P. G., Stonehouse, N. J., Murray, J. B.,Goodman, S. T. S., Talbot, S. J., Adams, C. J., Liljas,L. & ValegaÊ rd, K. (1995). Probing sequence-speci®cRNA recognition by the bacteriophage MS2 coatprotein. Nucl. Acids Res. 15, 575-594.

Stonehouse, N. J. & Stockley, P. G. (1993). Effect ofamino acids substitutions on the thermal stability ofMS2 capsids lacking genomic RNA. FEBS Letters,334, 355-359.

Stonehouse, N. J., ValegaÊrd, K., Golmohammadi, R., vanden Worm, S., Walton, C., Stockley, P. G. & Liljas,L. (1996). Crystal Structures of MS2 capsids withmutations in the subunit FG loop. J. Mol. Biol. 256,330-339.

Sugiyama, T. & Nakada, D. (1967). Control of trans-lation of MS2 RNA cistrons by MS2 coat protein.Biochemistry, 57, 1774-1750.

Talbot, S. J., Goodman, S., Bates, S. R. E., Fishwick,C. W. G. & Stockley, P. G. (1990). Use of syntheticoligoribonucleotides to probe RNA-protein inter-actions in the MS2 translational operator complex.Nucl. Acids Res. 18, 3521-3528.

Talbot, S. J., Medina, G., Haneef, I., Fishwick, C. W. G.& Stockley, P. G. (1991). Hyper-reactivity ofadenines and conformational ¯exibility of a transla-tional repression site. FEBS Letters, 283, 159-164.

ValegaÊrd, K., Liljas, L., Fridborg, K. & Unge, T. (1990).The three-dimensional structure of the bacterialvirus MS2. Nature, 345, 36-41.

ValegaÊrd, K., Murray, J. B., Stockley, P. G., Stonehouse,N. J. & Liljas, L. (1994). Crystal structure of anRNA bacteriophage coat protein-operator complex.Nature, 371, 623-626.

ValegaÊrd, K., Murray, J. B., Stonehouse, N. J., van denWorm, S., Stockley, P. G. & Liljas, L. (1997). Thethree dimensional structures of two complexesbetween recombinant MS2 capsids and RNA oper-ator fragments reveal sequence-speci®c protein-RNA interactions. J. Mol. Biol. 270, 724-738.

van den Worm, S. H. E., Stonehouse, N. J., ValegaÊrd, K.,Murray, J. B., Walton, C., Fridborg, K., Stockley,P. G. & Liljas, L. (1998). Crystal structures of MS2coat protein mutants in complex with wild-typeRNA operator fragments. Nucl. Acids Res. 26, 1345-1351.

von Hippel, P. H. & Berg, O. G. (1989). Facilitated targetlocation in biological systems. J. Biol. Chem. 264,675-678.

Wallis, R., Moore, G. R., James, R. & Kleanthous, C.(1995a). Protein-protein interactions in colicin E9DNase-immunity protein complexes. 1. Diffusion-controlled association and femtomolar binding forthe cognate complex. Biochemistry, 34, 13743-13750.

Wallis, R., Leung, K., Pommer, A. J., Videler, H., Moore,G. R., James, R. & Kleanthous, C. (1995b). Protein-protein interactions in colicin E9 DNase-inmmunitycomplexes. 2. Cognate and noncognate interactionsthat span the millimolar to femtomolar af®nittyrange. Biochemistry, 34, 13571-13579.

Walmsley, A. (1996). Physical factors affecting enzymeactivity. In Enzymology LabFax (Engel, P. C., ed.),chapt. 5, pp. 190-196, BIOS Scienti®c Publishers &Academic Press, Oxford and San Diego.

Witherell, G. W., Gott, J. M. & Uhlenbeck, O. C. (1991).Speci®c interaction between RNA phage coatproteins and RNA. Prog. Nucl. Acid Res. Mol. Biol.40, 185-220.

Edited by J. Karn

(Received 9 August 2000; accepted 21 November 2000)

Probing the kinetics of formation of the bacteriophage MS2 translational operator complex: identification of a protein conformer unable to bind RNA

Documents