fS

Center for BehavioralNeuroscience, Department of

University, Atlanta, GA 30303,

received in revised form20 July 2009;accepted 23 July 2009

eroC2Hesiw

energetics via changes in temperature, pH, buffers, and salt concentrations,

u-spr

regulatory proteins Tat, Nef, and Rev.1,2 The genome.3,9,13,14 Rev initially binds RRE at a high

doi:10.1016/j.jmb.2009.07.066

Available online at wwwproduction of structural, enzymatic and accessoryviral proteins for the assembly of the progeny virusrequires the viral pre-mRNA to be diverted insufficient quantity from the host cell nuclear splicingmachinery and translocated to the cytoplasm asunspliced (~9 kb) and singly spliced (~4 kb)transcripts.35 This shift from the production and

affinity binding site, localized to the relatively smallstemloop structure RREIIB,10,11 followed by subse-quent oligomerisation of up to eight Rev moleculesthrough RNA Rev-responsive element (RRE) pro-tein contacts.12 This forms the nuclear export signal,which utilizes the cellular nuclear export machineryto translocate to the cytoplasm, where the dissoci-Available online30 July 2009

Edited by M. F. Summers

Introduction

In the early phases of the hciency virus type 1 (HIV-1) lifemRNA undergoes complete2 kb mRNA transcripts that acytoplasmic appearance of fully sand singly spliced mRNA trantransition from viral latency to t

Corresponding author. E-mail addrAbbreviations used: RRE, Rev-res

isothermal titration calorimetry.

0022-2836/$ - see front matter 2009 ERNA binding was observed to be substantially negative and could not beaccounted for by conventional solvent-accessible surface area models. Analternative model, based on the extent of hydrogen bond changes as a resultof differences in ligand-induced water displacement at the binding site,provided reasonable explanation of the trends in Cp, as well as H andS. Our studies show that incorporation of favorable interactions at thesolvent-excluded binding interface can be used to alleviate the unfavorableenthalpic penalties of displacing water molecules from the hydrated RNAsurface.

2009 Elsevier Ltd. All rights reserved.

Keywords: zinc finger; thermodynamics; solvent; specific heat; RNAprotein interactions

man immunodefi-cycle, the viral pre-licing resulting ine translated to the

HIV-1 replication cycle.3,68 The nuclear export ofthe viral pre-mRNA in unspliced and singly splicedforms is facilitated by the interaction between theregulatory protein Rev and the Rev-responsiveelement (RRE), a 234 nt untranslated RNA structurelocated within the env gene of the viral RNAReceived 25 March 2009; ITC and localized by NMR to a histidine on the znf -sheet. TheCp of znfUSAInterestingly, despite the large cationic charge on the znfs, the number ofinteractions with the RNA phosphate backbone was lower than intuitivelyexpected. The presence of binding induced protonation was established byBiology, Georgia Stateas well as znf and RNA mutations, by isothermal titration calorimetry.Thermodynamic Profiling oFinger Interactions

Subrata H. Mishra1,2, Alexander M.

1Departments of Chemistry andBiology, Georgia StateUniversity, Atlanta, GA 30303,USA2Brains and Behavior Program,

The interactions betwthe HIV regulatory pshowed that singlethe Rev peptide androle of amino acidcomplex formationpliced to unsplicedscripts marks thehe late phase of the

ess: mwg@gsu.edu.ponsive element; ITC,

lsevier Ltd. All rights reserveHIV RREIIB RNAZinc

pring1 and Markus W. Germann1

en the HIV Rev-responsive element (RRE) RNA andtein Rev, are crucial for the HIV life-cycle. Earlier, we2 zinc fingers (znfs) have the same binding site as

xhibit nanomolar affinity. In this study, the specificde chains and molecular processes involved withere investigated by perturbation of the binding

J. Mol. Biol. (2009) 393, 369382

.sciencedirect.comation of the complex releases the unspliced mRNAfor translation while Rev shuttles itself back to thenucleus for the next round of HIV mRNA export.6,8

Thus, the initial RevRREIIB interaction is vital forthe subsequent production of viral assembly com-ponents; efficient prevention of this interactioncould effectively abolish the production of infectiousvirions.

d.

The interaction between Rev and RREIIB has beencharacterized by biochemical, mutational and struc-tural methods.1517 NMR structural studies haveutilized an RREIIB analog RREIIBTR and theRev peptide, a 17 amino acid arginine-rich motif(Rev3450), to demonstrate that binding occurs inthe major groove of the RNA.18 Rev peptideRREIIBTR interaction induces the formation of twonew purine-purine base pairs (G-G and G-A).19

These base pairs are formed in the bulge in thestemloop IIB and are critical to Rev binding.An extensive list of strategies for inhibiting the

pathogenesis of the virus has been documented.20 Tointerfere with the essential RRERev interaction,these approaches use RNA-based strategies such asanti-sense RNA, RNA decoys, RNA aptamers,ribozymal siRNA, protein-based strategies thatinvolve transdominant negative proteins, chimericnucleases, intracellular antibodies, peptides,21,22 andsmall organic compounds.23,24 Earlier, we demon-strated that C2H2 zinc finger proteins (znfs) ZNF29and ZNF29G29R designed by phage display25 bindthe same RNA bulge that Rev utilizes with nanomo-lar affinity.26 Themajor groove is narrow in a regular

RNA helix, while the RREIIBmajor groove is openedat the bulge containing the purine-purine mis-matches. This facilitates access of -helical recogni-tion elements of Rev as well as znfs. Our currentstudies focus on understanding the energetics of theunderlying molecular processes that facilitate RNAznf recognition. We have studied the energeticperturbations to the znfRNA system resultingfrom mutations on ZNF29 and the RNA targetRREIIBTR by isothermal titration calorimetry (ITC).Effects of salt, pH, and temperature on the bindingand the role of solvent were investigated. Ourfindings suggest strategies for enhancing the bind-ing affinity of zinc finger proteins to the targetRNA.

Results and Discussion

Design of zinc finger and RNA mutants

The sequence of the zinc finger has beenoptimized by phage display, which produced a

NMg th 2wh

370 HIV RREIIB RNAZinc Finger InteractionsFig. 1. ZNF29 and RREIIBTR mutations. (a) ZNF2926displayed (side chains). The tip of znf is the turn connectinnumbering as described elsewhere.19 () RNA mutants wituracil. All mutations are single changes except G48G71

names of the RNA mutants are based on the position and thesubstitution are: RREIIBTRG50_2ap, RREIIBTRG70_2ap, RREIinvolving replacement of cytosine by uracil are: RREIIBTRC4R solution structure with the position of the mutationshe -sheet to the -helix. (b) RREIIBTR sequence with the-aminopurine substitutions; () mutation from cytosine toere both guanosines are replaced by 2-aminopurine. The

type of mutation. The mutations involving 2-aminopurineIBTRG48G71_2ap, and RREIIBTRG47_2ap. The mutations9U and RREIIBTRC69U.

Binding was monitored by NMR to ascertainsimilar interaction of the mutant znfs withRREIIBTR RNA (Fig. 2). While we observe new

Fig. 2. Imino proton spectra of free and znf-boundRREIIBTR RNA. The znf:RNA molar ratios of 1:1 wereachieved by titrating protein into RNA in ITC buffer (seeMaterials and Methods) at 298 K. The assignment of thefree RNA is based on published data,19 except for G55(GCAA loop), which was identified from the iminospectrum of a smaller RNA oligonucleotide representingthe upper stem (C51G67) (data not shown). The dottedlines indicate imino resonances that show minor shifts onznf addition or that do not shift at all. These resonancesrepresent base pairing of the upper stem, including theGCAA tetraloop (G53, G55, and G64) and the lower stem(G42, G76, and G77), thus precluding them from being apart of the binding region and consequently localizing the

371number of high-affinity RREIIB-binding proteins.25

Subsequently, we determined the NMR structuresof ZNF29 and the mutant ZNF29G29R,26 whichexhibits a higher affinity for the RNA. Using thisphage display-generated high-affinity scaffold, wenow seek to optimize RNA binding by changingadditional key residues to assess their contributionto binding. The mutant znfs are named on thebasis of the position of the mutation relative toZNF29 (Fig. 1a); the rationale involved in thechoice of the position and type of mutation are asfollows.

(I). -Sheet mutations. Histidine at position 6could be involved in stacking with an RNA base.The ZNF29H6A and ZNF29H6Kmutants test thispossibility. If the interaction involves aromaticstacking, reduced affinity is expected for bothmutations. The ZNF29R12A mutant queries therole of a charged polar side chain on the -sheet,to RNA binding.(II). Mutations at the tip of the zinc finger. TheZNF29N15A and ZNF29D16A mutants examinethe involvement of the amino acids at the tip ofthe zinc finger in RNA binding. Zinc fingers areknown to utilize their tips in DNA as well as RNAbinding.27,28

(III). Mutations on the -helix. The ZNF29N21Amutant examines if the asparagine at thisposition serves a role similar to that of theanalogous asparagine in the Rev peptide,18

which bridges the G47 A73 base pair in theRevRREIIBTR complex. The importance of thelength of the side chain was investigated usingZNF29N21Q. In addition, the double mutantZNF29N21Q29R was designed on the basis ofthe analysis of the binding affinity of the singlemutants.

None of the mutations involve zinc-coordinatingresidues or residues involved in the hydrophobicpacking of the zinc finger core. Nevertheless, wehave established that all mutants fold into thecharacteristic fold of zinc fingers from thepresence of signature chemical shifts (Phe14 H andHis27 2H)26,29 in 1D 1HNMR spectra in 2H2O (datanot shown). The constant structure of the scaffoldallows us to evaluate the effect of individual sidechains on the binding interaction.The wild type RNA in our binding study is

RREIIBTR, which is a modified version of theRREIIB RNA and has been shown to retain all ofthe elements necessary for specific binding of Revand Rev peptides.19 For ease of reference, regions ofthe RNA are referred to as the upper stem, middlestem, bulge and lower stem, as shown in Fig. 1b. TheRNA mutants probe the effects of base pairdisruption on znf binding. Four of these mutantsinvolve substitution of the guanosine in the middlestem (G50 and G70) and the bulge (G48G71 and

HIV RREIIB RNAZinc Finger InteractionsG47) with 2-aminopurine while the other twoinvolve substitution of the cytosine in the middlestem (C49, C79) with uracil.bound znfs to the region including the bulge and themiddle stem. The resonances at around 13.97 and 13.72were identified as U66 and U45, respectively, fromNOESY cross-peaks for the ZNF29-RREIIBTR complex

(data not shown). However, the absence of NOEs from thenew peaks (12.4512.75 and 11.75) to previously identifiedresonances prevented their identification.

and shifted imino proton resonances from the bulgeregions/middle stem for all mutants, other RNAregions are not affected. Specifically, the base pairsin the upper (G53G64), near the lower stem (G41C46) or the hairpin loop (G55) in all RREIIBTRcomplexes retain essentially the same chemical shiftas the free RNA. The middle stem/bulge region wasalso implicated in the binding site from steady-statefluorescence data (Supplementary Data S10). Thus,all the new znf mutants studied in this work havethe same binding site on the RNA, the bulge region,as determined earlier for ZNF29 and ZNF29G29R,26

enabling the comparison of their thermodynamicbinding parameters.

Binding affinities of znf mutantsRREIIBTR

Thermodynamic parameters (H, S, and Kd) forthe binding of zinc finger proteins to RREIIBTRweredetermined by ITC (Table 1). All binding isothermscould be fit to a single binding site model with a 1:1

plicate the involvement of H6 in RNA binding.The possibility of binding-linked protonation/deprotonation of this histidine was also investi-gated. The R12Amutation on the second -strandof ZNF29 caused a ~1.8-fold reduction in thebinding affinity. This was surprising, because znfsgenerally do not utilize the -sheet in their inter-actions with RNA.27,28 The imino spectra of theZNF29R12ARREIIBTR complex is essentiallysuperimposable to that of the control, suggestingan electrostatic contact between the R12 sidechain and the RNA phosphodiester backbone.(II). Mutations at the tip of the znf: The mutationsat the tip of znf, ZNF29N15A and ZNF29D16A,resulted in the lowest binding affinities (~2.2 and~ 3-fold lower for the N15A and D16Amutations,respectively). In addition, the imino proton peaksthat appeared on complex formation of these twomutants with RREIIBTR displayed distinct differ-ences in their chemical shift and intensitycompared to that of the control (Fig. 2). Thereduced affinity, along with the differences in the

.0

(nM)

21252910604540316101476446574682

e Kd=1also fros fromerformed in S

372 HIV RREIIB RNAZinc Finger Interactionsstoichiometry. The free energy of binding at 25 Cfor all znfRNA interactions had favorable enthalpyand entropy contributions. Changes in bindingaffinity for the znf mutants are based on the control,the RREIIBTRZNF29 interaction.

(I). -Sheet mutations: While a mutation toalanine (H6A) lowered the affinity ~1.4-fold, thelysinemutation (H6K) increased affinity ~1.6 fold.There are differences in the imino spectra ofZNF29H6K versusZNF29 RNAbinding,while thespectrum of ZNF29H6A is similar to that ofZNF29 (Fig. 2). Hence, the trend in the affinity ofthe H6 mutants cannot categorically rule outaromatic interactions, as mutation to lysine mayhave compensated for the absence of an aromaticinteraction. Nevertheless, these mutations im-

Table 1. Energetics of ZNF RNA binding at 298 K, pH 7

Protein RNA n Kd

ZNF29 RREIIBTR 1.02 149ZNF29H6A RREIIBTR 0.94 207ZNF29H6K RREIIBTR 1.01 95ZNF29R12A RREIIBTR 0.98 266ZNF29N15A RREIIBTR 1.09 322ZNF29D16A RREIIBTR 1.00 450ZNF29N21A RREIIBTR 0.94 232ZNF29N21Q RREIIBTR 0.94 80ZNF29G29R RREIIBTR 0.99 57ZNF29N21QG29R RREIIBTR 1.06 35ZNF29G29R RREIIBTRG70_2ap 1.00 671ZNF29G29R RREIIBTRC49U 1.07 510ZNF29G29R RREIIBTRG48G71_2ap 1.04 324ZNF29G29R RREIIBTRG47_2ap 1.02 303ZNF29G29R RREIIBTRG50_2ap 0.93 210ZNF29G29R RREIIBTRC69U 0.98 254

Binding affinities are expressed as dissociation constants (Kd) wherbinding isotherm in Origin 7.0 software. The errors reported areenergies were calculated from G=RT ln (Ka), and the entropie65 85 M, while RNA was 4.5 5 M. All experiments were p200 M -mercaptoethanol). Representative ITC traces are includ

were: ZNF29: 0.4, 29R: 0.4, H6K: 0.4, R12A: 0.4, N15A: 0.4, D16A:

a Errors in G and H were 0.02 kcal mol-1and 0.04 kcal mol-1, respecimino spectra of the complexes with RREIIBTR,indicates that these residues (N15 and D16) areinvolved directly in the formation of new basepairs in the complex.(III). Mutations on the -helix: Mutations atposition 21, N21A and N21Q, had oppositeeffects on the RNA binding affinity. While themutation to alanine reduced binding affinity~1.6-fold, there was a ~1.9-fold increase for N21Q.The imino proton spectrum of ZNF29N21ARNAwas similar to that of the control but there werenoticeable differences in the new imino protonpeaks for ZNF29N21QRREIIBTR. These com-parisons suggest that the side chain of N21 doesnot contact a base pair as N40 in the Revpeptide.18 The longer side chain of the N21Q

G (kcal mol1) H (kcal mol1) S (cal mol1)

9.30.1 7.20.1 7.29.10.1 5.30.1 139.60.2 4.30.1 17.79.00.0a 5.70.0a 118.90.1 8.20.2 2.38.70.1 8.00.2 2.49.00.1 6.70.2 8.19.70.2 7.80.2 6.39.90.1 5.50.1 14.610.20.2 7.60.2 8.68.410.1 10.10.5 -5.88.570.1 6.10.1 8.48.840.1 4.70.1 148.880.1 4.30.1 15.39.100.1 7.80.2 4.58.990.1 8.60.3 1.5

/Ka, the association constant obtained by curve-fitting the ITCm curve fitting to a 1:1 site binding model. The binding freeS=(H G) / T. ZNF concentrations were in the rangeed in ITC buffer (10 mM sodium phosphate, 100 mM NaCl,upplementary Data Fig. S1. The heats of dilution (kcal/mol)

0.4, H6A: +3.5, N21A: +2, N21Q: +2.tively.

but unfavorable entropy contribution. While all the

The value of nH was determined to be 0.24 at pH6.2 and 0.61 at pH 8.0 (Fig. 3a) and signifies anet uptake of protons. These values indicate thepresence of a single protonation site whose pKa isincreased due to binding. The pH range suggests theinvolvement of a histidine and was determined byNMR to be H6, because complex formation resulted

373RNA mutants displayed lower binding affinity toZNF29G29R than the wild type RNA, there is atrend depending on the position of the mutation.Any mutation disrupting the potential G70C49base pair (middle stem) in the complex caused adrastic reduction in znf binding affinity (~11.8-foldlower for RREIIBTRG70_2ap and ~ 8.9-fold lowerfor RREIIBTRC49U). The 2-aminopurine substitu-tions at the bulge (RREIIBTRG48G71_2ap,RREIIBTRG47_2ap) reduced znf binding affinity~56fold, while alteration of the G50C69 base pairhad the least deleterious effect on znf binding (~3.7and ~4.5-fold lower for RREIIBTRG50_2ap andRREIIBTRC69U, respectively). The trend in theenthalpy and entropy components of the free energyof interaction for both protein and RNAmutants arediscussed in further detail below.

Binding linked protonation and pH effects

The potential for binding-induced protonation wasexamined by determining the binding enthalpies atpH 6.2 and pH 8.0 using buffers with different heatsof ionization. The observed binding enthalpy forZNF29RREIIBTR is expressed as:31,32,39

DHiobs = DHint +DnH DHLp + DHib

1

where Hobsi is the enthalpy observed in the ITC

experiment at the respective pH and buffer, Hint isthe enthalpy of fully protonated znf binding RNA,nH is the number of protons taken up per mol ofcomplex formed, HLp is the enthalpy of binding-linked ligand protonation (all potential protonationsites) and Hb

i is the enthalpy of buffer ionization.For complex formation in buffers with different heatsof ionization at the same pH and temperature:mutant could result in altered proteinproteincontacts. In silico mutation (N21Q) of the freeZNF29 predicts a hydrogen bond between Q21and R17 (data not shown) (Swiss-PdbViewer30).Thus, the increased affinity of ZNF29N21Q couldpossibly arise from different positioning of theR17 side chain.

Binding affinities of RREIIBTRmutantsZNF29G29R

The thermodynamic binding parameters ofZNF29G29R to the RNA mutants detailed in Fig.1b are also given in Table 1. All RNAmutants boundZNF29G29R with 1:1 stoichiometry. The free energyof binding had favorable enthalpy and entropycontributions for all RNA mutantZNF29G29Rinteractions at 25 C, except for the RNA mutantwith a 2-aminopurine substitution at position G70(RREIIBTRG70_2AP), which has favorable enthalpy

HIV RREIIB RNAZinc Finger InteractionsDnH = DH2obs DH1obs

= DH2b DH1b 2in substantial chemical shift changes for H6 while nochange was observed for the zinc-coordinatinghistidines (Fig. 3c). In the free ZNF29, the pKa ofH6 is 6.7, determined from the change in the 2 13Cand 1H chemical shifts (Fig. 3b). Deprotonation ofH6 in the free ZNF29 is characterized by increasing13C and decreasing 1H chemical shifts (Supplemen-tary Data, S2). In the ZNF29RNA complex, the pKaof H6 is expected to be higher; unfortunately,precipitation at higher pH precluded a directmeasurement. However, the 13C chemical shiftdecrease and the 1H chemical shift increase (H6 2complex) supports a higher level of H6 protonationin the complex as compared to that of the free znf atpH 7 (Fig. 3c). Using the proton uptake determinedearlier, a simulation places the estimated pKa of H6in the complex around 8.3 (Fig. 3b), correspondingto an increase of 1.6 pKa units.If binding-induced protonation of the H6 side

chain is the only event contributing to the enthalpy,then the enthalpy of the ZNF29H6ARREIIBTRcomplex formation should remain constant in thepH range 6.28.0. However, this was not the case(Fig. 3d); there is an increase in the binding enthalpywith increasing pH for ZNF29H6A. Considering thepotential groups on znf and RNA that may beprotonated in this pH range, the most likelycandidate is the N-terminal NH3

+ for which a pKaof 8.0 is predicted (PROPKA).33 We note that theenthalpy of H6ARREIIBTR complex formationincreases linearly with lowered fractional proton-ation of this NH3

+ group (Fig. 3d). The N-terminus islocated near the tip of the zinc finger, which isknown to be involved in binding, because mutations(D16A, N15A) drastically affect the binding. Wetherefore hypothesize that interaction between theRNA and the tip of the znf is hampered by an intra-protein interaction involving the N-terminal NH3

+,possibly by an ion pair with D16. The znf tipRNAinteraction could involve disruption of this ion pair,resulting in an enthalpic penalty. In such a scenario,the increasing pH will result in lowered levels ofNH3

+ protonation and consequently less energy isexpended in breaking this ion pair on binding. As aresult, such an effect will manifest itself in moreexothermic enthalpies with increasing pH, as

The H6 2 chemical shifts in the complex could bedetermined only up to pH 8.8, beyond which precipitationprohibited measurements. Consequently, the end point ofthe pH titration of the complex could not be established.The pKa of H6 in the complex was estimated by iterative

use of pKa values in the HendersonHasselbalch equationthat reproduced proton uptake numbers determinedexperimentally (nH values from Fig. 3a).

374observed. The distinct chemical shift changes (13Cand 1H) of the N-terminal methionine (data notshown) upon binding RNA supports a change in theenvironment of M1. The ZNF29 binding enthalpy inFig. 3d is non-linear, because it involves twoprocesses, namely a linear contribution from ionpair disruption and a non-linear contribution frombinding-induced protonation. The correlation be-tween nH and pH is non-linear (Fig. 3b; Eq. (1)).The highest contribution to ZNF29 binding enthalpydue to pH-influenced processes is predicted to be at~pH 7.5, in agreement with the maximal protonuptake (nH) at this pH (Fig. 3b and d).

Fig. 3. Binding-linked protonation and pH effects. (a) TheRREIIBTR binding at pH 6.2 and pH8.0 are plotted againstare taken from the literature.39 The Hobs (kcal mol

1) v(cacodylate, ); and 4.470.07 (sodium phosphate, ), an(triethanolamine, ); 6.380.10 (EPPS, ) and 8.760.08determined to be 0.24 and 0.61 at pH 6.2 and pH 8 respecfractional change in H6 protonation for free ZNF29 was detchemical shift changes (chemical shifts at the respective pHfractional change in protonation is displayed as fully prorepresents the fit to the equation: Fractional change 10pHthe fractional protonation change calculated from the abovtheoretically calculatednH for a pKa shift from 6.7 (free) to 8.3are marked with . (c) Overlay of the 13C - 1H HSQC spectra fdistinct changes in chemical shift (1H2 and

13C2) for H6, while(d) The binding enthalpy (Hobs) and errors for ZNF29 () an(fp) of the N-terminal NH3

+ group (pKa 8.0). The pH values forcontinuous line represents a linear regression fit (Hobs=increasing pH and has a slope of 5.900.6 kcal mol-1 (Hipd, enwas calculated by the HendersonHasselbalch equation. (AddiSupplementary Data, S11).HIV RREIIB RNAZinc Finger InteractionsWe note that elevated pKa values for the RNAbases adenosine and cytidine have been reported forlocally crowded phosphodiester backbones inribozymes.34 However, pKa shifts for the RREIIBTRRNA were not considered, because the interactionsite does not contain such structural features and theCD spectra do not support major rearrangementupon binding (Fig. 6a).

Effects of salt

The extent of electrostatic contributions to thebinding free energy was assessed by evaluating the

experimentally determined enthalpies (Hobs) of ZNF29the enthalpy of buffer ionization (Hb). The Hb valuesalues at pH 6.2 are: 3.940.11 (MES, ); 4.960.08d those at pH 8 are: 5.060.04 (tricine, ); 4.450.08(sodium phosphate, ). The nH (Eq. (2)) values aretively, and indicate proton uptake on binding. (b) Theermined from the 1H (2) () and 13C (2) () fractionalvalues are provided in Supplementary Data, S2). The

tonated (0) to fully deprotonated (1). The broken linepKa=1 10pHpKa. The continuous black line representse equation for pKa=8.3. The continuous red line is the(bound). The experimentally determinednH, from Fig. 3aor the free (blue) and RREIIBTR-bound ZNF29 (red) showsthe zinc-coordinated histidines (H23 and H27) do not shift.d H6A () are plotted against the fractional deprotonationthe corresponding fractional deprotonation are shown. TheHint+ fpHipd) for H6A binding enthalpy (Hobs) withthalpy of ion pair disruption). The fractional deprotonationtional thermodynamic parameters for a and d are given in

affinities of ZNF29, ZNF29G29R and ZNF29R12Ato RREIIBTR under varying concentrations of salt(NaCl; Fig. 4a). The absence of significant conforma-tional change in the complex was confirmed from theobservation that the imino spectra of ZNF29-boundRREIIBTR are salt-independent (SupplementaryData, S3).The salt dependence was analyzed as:35

log Kobs = log NaCl =m0 + k 3where m is the number of ion pairs formed resultingin cation release from the RNA, is the fractionalneutralization of the RNA backbone phosphates bythermodynamically bound cations, and k is thefraction of ions released by the protein on binding.Binding of ZNF29G29R exhibited the highest level ofsalt-dependence, followed by ZNF29, while bindingof ZNF29R12A was salt-independent. For ZNF29and ZNF29G29R, a linear decrease of the bindingaffinities was observed with increasing concentra-tion of salt in the range 100 mM [NaCl]175 mM,

HIV RREIIB RNAZinc Finger InteractionsFig. 4. Salt dependence of znfRNA binding. (a)RREIIBTR binding affinity (Ka) for ZNF29, ZNFG2929R,and ZNF29R12A are plotted against the concentration ofsalt (NaCl). The linear correlation between salt dependenceand binding affinity has been analyzed by Eq. (3). All theabove experiments were conducted at 298 K with the samebuffer conditions (10 mM sodium phosphate, 200 M -mercaptoethanol, pH 7.00.02)with various concentrationsof NaCl. (b) Thermodynamic parameters of ZNF29G29RRREIIBTR binding determined with 100 mMNaF, NaCl orNaBr. The entropic contributions are expressed as TSobs

(298 K). Earlier studies showed that anions interactingweakly with a protein when released on DNA bindingmake the overall Hobs less exothermic.

37with log (Kobs) / log [NaCl] of 2.19 and 2.94,respectively, indicating net release of ions (Na+ andCl), while at lower concentrations of salt thebinding affinity did not increase linearly (below100 mM). Similar departures from linearity wereobserved in earlier studies of proteinDNA bindingat lower salt concentrations.36 This trend wasattributed to a change in the occupancy of ionbinding sites on the protein resulting from transferof the protein ion-binding surface from the bulksolution to a different ionic environment in thevicinity of the nucleic acid upon binding (highconcentration of cation, low concentration of anion).The presence of weak anion binding by the znf andsubsequent release on RNA binding was indicatedfrom the decreasing overall Hobs (less exothermic)and increasing overall Sobs, when the anion waschanged from a strongly hydrated (F) to a weaklyhydrated anion (Br) (Fig. 4b). Hence, changes in theoccupancy of the anion-binding sites on the znfwould explain the curvature of log (Kobs) versus log[NaCl], at low concentrations of salt.From the difference in log (Kobs) / log [NaCl] of

ZNF29 and ZNF29G29R and recognizing thatZNF29G29R has one additional charge, we estimate to be 0.75. This value of is reasonable for an oligomerlike RREIIBTR, which is expected to have a lower axialcharge density than a longhelix (0.88) due to end effectsas well as irregular charge density at the bulge.RNA binding by ZNF29G29R, ZNF29, and

ZN29R12A results in net release of 2.94, 2.19 and 0ion pairs, respectively, indicating that R12 in ZNF29and R29 in ZNF29G29R are involved in ionicinteractions with the RNA phosphate backbone.The linear extrapolation of the integrated form ofEq. (3) to 1 M NaCl can be used to calculate the freeenergy of binding in the absence of ion release(G0

0) (Supplementary Data, S4).35 For ZNF29 andZNF29G29RRNA binding, G0

0 values are 6.34and 5.91 kcal mol1 at 298 K and consequentlythe free energy contribution of ion release (cationsand anions, at 0.1 M salt, 298 K) are 2.96 and3.96 kcal mol1, respectively.

Temperature dependence of znfRNAbinding enthalpy

The temperature-dependence of Hobs for ZNF29and ZNF29G29R RNA binding has been evaluatedfrom 20 C to 35 C (Fig. 5). For ZNF29G29R, the freeenergy of binding is enthalpy and entropy-driven inthis temperature range. For ZNF29, the free energyof binding is enthalpy and entropy driven below 302K, while it is enthalpy-driven and entropy-opposedabove that. For both znfs, Gobs is nearly constant inthis temperature range. In cases when one or bothbiomolecules undergo thermally induced unfolding,the relationship between Hobs and temperature isnon-linear.38 Both ZNF29 and ZNF29G29R RNAbinding show linear temperature-dependence, con-

375firming that the components do not unfold in thestudied range. Moreover, the znfRNA complex, thefree znfs, and the free RNA have been independently

result in znf side chain rearrangements, this is notexpected to contribute substantially to Cp. Theimino spectra of the ZNF29 and ZNF29G29R RNAcomplexes are essentially superimposable, indicat-ing that the binding of both znfs results in the samechanges. Therefore, the formation of new base pairscannot account for the sizeable difference in theirCp values: jDDCpj = 129 cal mol1 K1 and are notexpected to be major contributors to the overall heatcapacity changes.A negative Cp in biomolecular interactions has

been linked to the extensive dehydration of non-

HIV RREIIB RNAZinc Finger InteractionsFig. 5. Enthalpyentropy compensation for the bind-

376confirmed to be stable at these temperatures, whichsimplifies the analysis (data not shown).The changes in the heat capacity of binding (Cp)

are 40340 and 27434 cal mol1 K1 for ZNF29and ZNF29G29R, respectively. These substantialchanges in binding heat capacity may arise fromseveral contributing factors, as discussed below,where the buffer ionizations can account for only asmall portion of Cp (28 cal

1 for sodium phos-phate buffer at pH 7).39

Major structural changes of either biomolecule orboth on complex formation can have significantcontributions to Hobs and consequently affectCp.

38 However, the CD spectra of free andcomplexed RNA show only minor differences,which suggests minimal RNA structural changesupon binding (Fig. 6a). In addition, chemical shiftsof the residues in the hydrophobic core have similarchemical shifts for the free and bound znf (Fig. 6b).Since the znf hydrophobic core is composed of sidechains from the -sheet as well as the -helix, thisindicates that on binding the znfs do not undergoany major conformational change. Thus, largestructural perturbations are not the cause of thesignificant value of Cp. While RNA binding will

ing of ZNF29 and ZNF29G29R to RREIIBTR. Thethermodynamic parameters for ZNF29G29R binding aregiven the following symbols: () H; () TS; and ()G; and those for ZNF29 are: () H; () TS; and ()G. The free energy was calculated as G=RT ln(Ka)and entropic contributions as TS=H G. The linearregression fit of the enthalpic dependence on temperatureyielded Cp for ZNF29 and ZNF29G29R binding to be40340 cal mol1 K1 and 27434 cal mol1 K1,respectively. A non-zero heat capacity is indicated by anonlinear correlation between ln (Ka) and 1/T (Supple-mentary Data S5). The symbols encompass the errorassociated. (All of the above data are given in Supple-mentary Data, S11).polar (Anp) compared to polar surfaces (Ap)(either by the conformational rearrangement of theinteracting molecules or by burial of non-polarinterfacial surfaces).40,41 The surface dehydrationcontribution to Cp can be described by theempirical correlation:

DCp aDAnp DAp 6where and are positive coefficients,4044 whileAnp and Ap are negative quantities becausethey represent burial of surface area. We haveestimated the values of net change in solvent-accessible surface area (SASA), Anp and Ap(protein and RNA) (Supplementary Data, S6, S7)using a 1.5 solvent probe.45 The estimated netAnpand Ap for ZNF29 RREIIBTR complex formationare 751 2 and 1888 2, respectively, while the netAnp and Ap values for ZNF29G29RRREIIBTRare 743 2 and 1983 2. On the basis of thesevalues, complex formation with either znf isexpected to bury more polar than non-polar SASA.Even using different values for and (Supple-mentary Data, S8), surface area burial is unable toexplain the substantial negative values of the changein observed heat capacity and the Cp for ZNF29versus ZNF29G29RRREIIBTR binding. Failure toaccount for negative Cp by conventional surfacearea models focusing on non-polar surface dehy-dration alone have been reported for other biomo-lecular interactions.46,47

It has been suggested that Cp contribution fromprotonation effects are compensated by opposingchanges in the buffer,49 and therefore can bedisregarded. However, recent studies have shownthat this assumption could be too simplistic.32,48

SASA calculations were performed as described inMaterials and Methods. The structures of the free and znf-complexed RNA are not available, and the valuespresented here serve only as guidelines to underscorethe extensive burial of polar over non-polar SASA. Evenwhen the SASA calculations were performed by unstack-ing the bases in the middle stem and the bulge, for the freeRREIIBTR structure, Cp calculated from semi-empiricalmodels (Supplementary Data S8) were not in the range ofexperimentally determined values. Intuitively too, thesurface of the interacting elements that would be buried

on complex formation, the -helix of the znfs and theRNA major groove at the bulge, involve more polar thannon-polar SASA.

HIV RREIIB RNAZinc Finger InteractionsWhile we have not evaluated contribution ofprotonation effects to Cp, we recognize that Cpfor ZNF29 versus ZNF29G29RRREIIBTR bindingcannot be explained by the temperature-dependenceof the induced protonation effect. Furthermore,trends in Hobs for znf and RNA mutational studiesprompted us to evaluate alternative models.

Cooper models explain Cp and trends inbinding enthalpy and entropy of mutant znfsand mutant RNAs

Large negative values of Cp for biomolecularinteractions have been rationalized by the involve-ment of water.50,51 In this model, trends in H andCp are evaluated on the basis of differences in theextent of hydrogen bond changes as a result ofdifferences in ligand-induced water displacement atthe binding site. Water is treated as: (1) solvating themacromolecular cavity and the ligand; (2) bulk water;and (3) non-displaced water on ligand binding,

Fig. 6. Absence of major structural rearrangementsfor the RNA and ZNF29 on binding. (a) CD spectra offree RREIIBTR RNA (dashed line) and ZNF29 com-plexed RNA (1:1 complex, continuous line) at aconcentration of 10 M. Spectra (20 scans) were acquiredin a 0.2 cm pathlength cuvette using a scanning speed of200 nm min1. The spectrum of free ZNF29 wassubtracted from that of the complexed RNA. (b) Overlayof the aromatic region of the 13C - 1H HSQC spectra forfree (blue) and RREIIBTR bound ZNF29 (red). Assign-ments are indicated. All spectra (CD and NMR) wereacquired at pH 7.0, 298 K in ITC buffer (see Materialsand Methods).where the occupancy of (1) and (2) are temperature-dependent, resulting in the following predictions:50

(a) The binding enthalpy of a ligand that dis-places more water molecules from themacromolecular cavity is less exothermic.

(b) Water involvement can result in negativeCp. Specifically, a ligand displacing morewater is predicted to have a less negativeCp. Each additional water molecule dis-placed increases Cp by +18 cal mol

1 K1.

Despite its simplicity, thismodel has proven usefulto rationalize H and Cp values for biomolecularinteractions in water.51

The salt-dependent behavior of ZNF29G29Rimplies that the terminal arginine makes an electro-static contact with the RNA phosphate backbone.Arginine interactions with the phosphate backboneare enthalpically favorable.52 Thus ZNF29G29Rbinding is expected to result in a more exothermicbinding compared to ZNF29. Yet, the opposite isobserved. We note that ZNF29G29R binding woulddisplace more water than ZNF29 and the observedbinding enthalpies are in agreement with the Coopermodel. Similarly, as predicted,Cp of ZNF29 bindingismore negative than that of ZNF29G29R. Thismodelpredicts (from |Cp|) ZNF29G29R displaces ~7more water molecules than ZNF29, in good agree-ment with the number estimated (7 ~ 8) from thedifference in their solvent-excluded volumes (110 3).We observe similar enthalpic trends for most of

the other ZNF29 mutants, signifying a similar rolefor the solvent (Fig. 7a). AlthoughN15 and D16 are apart of the znfRNA interactions, as evidenced fromimino proton data, the binding enthalpies of N15Aand D16A mutants are slightly more exothermic ascompared to ZNF29, instead of being lower.Similarly, ZNF29R12A binding is more exothermicthan that of ZNF29G29R despite the missingelectrostatic interaction (salt studies). The mutantsH6A and H6K cannot have binding-induced pro-tonation effects because they lack the histidine andyet mutation to alanine is more exothermic thanmutation to lysine. All the above comparisonsinvolve mutations from bulky to small aminoacids, which result in fewer displaced watermolecules on binding. Consequently, the decreaseddisplacement of water molecules results in a moreexothermic binding enthalpy, as predicted by theCooper model. In comparison to the ZNF29 bindingenthalpy, a mutation to smaller non-polar aminoacid thus has two opposing effects: an enthalpic lossdue to the absence of the potential favorableinteraction, compensated by an enthalpic gain dueto lower displacement of water. The bindingenthalpies of mutations that are secluded from theznfRNAsolvent interface (Fig. 7c), thus reflect thepotential interaction, as we observe in the case of the

377mutants N21A and N21Q.The enthalpic gain from the decreased displacement

of watermolecules to the bulk solvent is observed also

378for RNA mutations that affect the geometry of theRREIIBTR bulge. The change in the binding sitegeometry results in fewer displaced water moleculesand, consequently, is manifested as higher bindingenthalpies accompanied by lower entropic contribu-tions to the binding free energy (Fig. 7b). This effect ismore pronounced for modifications at the middlestem that alter the compact geometry of the bindingsite by preventing possible base pair formation (2-aminopurine substitutions).Modifications at the bulgeG48G71-2ap, G47_2ap resulted in less exothermicbinding, indicating their involvement at the RNAprotein interface devoid of solvent.While displacement of water molecules from the

hydrated RNA surface contributes to the bindingfree energy via the entropic gain, it is accompaniedby unfavorable enthalpic contributions, which miti-gates the net favorable impact on the overall bindingenergetics (Supplementary Data S9).

Conclusion

Here, we have performed a thermodynamicinvestigation of the znfRNA interactions utilizingisothermal titration calorimetry and NMR spectros-

Fig. 7. Enthalpic and entropic contributions to free bindingbars) of ZNF29 andmutants to RREIIBTR (a) and RNAmutantexothermicity. Entropic contributions (open bars) are expressedisotherm inOrigin. (c) Themolecular surface of the free ZNF29Gthe mutations rendered in solid colors. The protrusions fromsolvent interface affectingwater displacement on binding are sh26, and 28) surrounding N21 (yellow) is shown in transparent bHIV RREIIB RNAZinc Finger Interactionscopy. Earlier, we showed that znfRNA binding isdependent on the znf architecture.26 Ourcurrent studies focus on enumerating the role ofamino acid side chains in the RNAznf interactions.Although coupling of several molecular processeshampers the specific allocation of contributions tothe binding free energy, insight into the nature of theinteractions can still be obtained.The zinc finger motif is ubiquitous amongst

nucleic acid-binding proteins. Specifically, C2H2znfs belonging to TFIIIA have been shown to interactwith their RNA targets through multiple modes ofrecognition.53 Structural and biochemical studieshave determined that the residues at the tip of thehelix are critical to binding.53 Our studies show theimportance of the residues at these positions, N15and D16, to RNA binding. It was surprising toobserve the involvement of the -sheet residues (H6and R12) in znfRNA interactions, since zinc fingersdo not generally use any component of their -sheetin RNA binding.27,28,53 Both H6 and R12 residuesflank the -sheet (Fig. 7c), and binding-inducedprotonation of the H6 residue as well as salt studieson the ZNF29R12A mutant binding suggests theirrole in contacts to the phosphodiester backbone.These residues may serve to anchor the znf to the

energy of binding at 298 K. The binding enthalpies (filleds to ZNF29G29R (b) are displayed in the order of increasingas TS. Errors in H are from curve fitting of the binding29R (PDB ID 2AB7) (VMD)60 is shownwith the positions ofthe surface for mutational positions at the proteinRNAown in red. The surface of the helical residues (1719, 22, 24lue. The backbone of the znf is displayed as a cartoon trace.

proachmay be instrumental in improving the binding

cacodylate, 0.47 kcal mol ; tricine, 7.63 kcal mol ;1 1 39RNA in coordination with other interactions in thewidened RNA major groove. The removal of any ofthese anchors may introduce conformational flexi-bility of the RNA phosphate backbone at these sites,as evidenced by the increased binding entropy ofthese mutants (Fig. 7a) compared to ZNF29.The free energy of znfRNA association contains

contributions from several processes. Among them,the favorable contribution from ion release onbinding to the overall free energy accounts for ~32% and~40%of the observed total binding free energyfor ZNF29 and ZNF29G29R, respectively (100 mMNaCl, 298 K). Consequently, this indicates thatmolecular recognition between the znf and theRNA is driven by fewer electrostatic contacts on theRNA phosphate backbone than one might expect.Therefore znfRNA association is largely driven byfavorable contributions arising mainly from interac-tions in the RNA major groove. These favorableinteractions counteract the unfavorable contributionsfrom loss of translational and rotational degrees offreedom, which can be substantial.54 Unlike proteinDNA complexes, proteinRNA complexes have alower level of participation of the phosphate back-bone in interactions. Statistical analysis of proteinRNA complexes5557 has shown that H-bonds fromarginine/lysine side chains to the RNA 2-OHaccount for a significant portion of the molecularcontacts. We note that the screening of the znf phagedisplay library against the RREIIB target wasperformed in the presence of excess tRNA, whichmight have efficiently eliminated znf sequenceswhere recognition was predominantly driven byless specific RNA phosphate backbone interactions.The perturbations of the ZNF29RREIIBTR bind-

ing free energy, through znf or RNA mutations, islimited to a window of 1 kcal mol1, althoughsubstantial differences are observed in their enthal-pic and entropic components. Specifically, forZNF29 and ZNF29G29R binding, Cp as well asCp (ZNF29 versus ZNF29G29R) could not beexplained by conventional SASAmodels. The use ofCooper models with respect to dehydration at thebinding interface and their subsequent effect on therespective enthalpies, entropies as well as the natureof Cp, provided a satisfactory explanation. Addi-tionally, analysis of the enthalpy using buffers withdifferent heats of ionization demonstrated thepresence of binding-induced protonation.A more detailed understanding of the molecular

processes in this znfRNA system enabled the designof a znf with higher RREIIBTR binding affinity. Theinteraction of the terminal arginine (ZNF29G29R)with the RNA phosphate backbone, though favor-able, is alleviated by a substantial enthalpic penaltydue to displacement of solvent. The binding of themutant ZNF29N21Q, however, shows how suchenthalpic penalties can be mitigated with favorableinteractions at the binding interface that is secludedfrom the solvent. This information has been used

HIV RREIIB RNAZinc Finger Interactionssuccessfully to design the double mutantZNF29N21QG29R (Table 1),where both the enthalpicand entropic contributions to free energy are favor-triethanolamine, 8.02 kcalmol ; andEPPS, 5.15 kcalmol .

NMR spectroscopyaffinity of drugs guided by thermodynamic studies.

Materials and Methods

Proteins and RNA

Site-directed mutants of the ZF29 plasmid wereobtained using the QuickChange site-directed muta-genesis kit (Stratagene, La Jolla, CA) and transformed intoBL21DE3 pLys S cells (Novagen). Expression, purificationand characterization were done as described.26 Gel-purified RREIIB-TR RNA and the 2-aminopurine-modi-fied RREIIB-TR RNA sequences were obtained fromDharmacon Research, Inc. with 2 protection groups andtreated as described.26

Isothermal titration calorimetry (ITC)

All ITC experiments were done with a VP-ITC micro-calorimeter (MicroCal, LLC, Northampton, MA) with 10 laliquots of the protein added to RREIIBTR RNA every 400s. Unless stated otherwise, the ITC experiments were doneat 25 C in a standard ITC buffer (10 mM sodiumphosphate, 100 mM NaCl, 200 M -mercaptoethanol)with ZNF and RNA concentrations of 6585 M and4.55 M, respectively. ITC samples were degassed andthen adjusted to pH 7.0 at the appropriate temperaturerequired for the experiment. In all experiments, the pHdifference between the titrant and analyte was less than0.02. Equilibrium constants (Ka), H, and number ofbinding sites (n) were obtained from a one-site modelcurve fit using VPViewer2000 and Origin 7 (OriginLabCorp, Northampton, MA). The heat of dilution wassubtracted by applying a linear fit to the saturation portionof the titration curve. G and S were calculated from:

DG = RTlnKa =DH TDSwhere R is the gas constant (1.987 cal mol-1 K-1) and T isabsolute temperature (K). The specific heats (Cp) wereobtained from:

DCp = ADH=AT

The contribution from buffer ionization was calculatedfromnHCp,buffer,

32 wherenH is the number of protonstaken up/mol of complex, and Cp,buffer is the heatcapacity change from the deprotonation of the buffer.Buffer protonation measurements were recorded at

pH 6.2 in MES (2-(N-morpholino)ethanesulfonic acid),phosphate, and sodium cacodylate buffers and at pH 8.0 intricine, triethanolamine, EPPS (3-[4-(2-hydroxyethyl)-1-piperazinyl]propanesulfonic acid), and phosphate buffers.All buffers contained 10 mM buffer, 100 mM NaCl and200 M -mercaptoethanol. Published values for theheats of protonation for the buffers are as follows: MES,3.71 kcal mol1; phosphate, 1.22 kcal mol1; sodium

1 1ably increased, compared to ZNF29. Such an ap-

379NMR samples were prepared in the ITC buffer contain-ing 10% (v/v) 2H2O. Imino proton spectra were collected

probehead (Bruker). Data were processed with

XWINNMR 3.5. The 1D imino proton spectra for bothRREIIBTR RNA and RREIIBTRZNF complexes werecollected using a 1-1 jump and return pulse sequence.58

Heteronuclear single quantum correlation (HSQC) spectrausing echo-antiecho-TPPI and shaped pulses for the 13Cinversion were recorded using z gradients (2K512, 48scans, 1.5 s presaturation).59 The 2D data were processedwith a shifted sine bell (SSB) window function andtransformed with the following processing parameters:4K4K, SSB of 2 in f2 and f1. The 13C chemical shifts in theHSQC spectrum at pH 7 and 25 C was referencedindirectly to the 1H internal standard (DSS). This spectralreference was used to reference all the other HSQC spectra(different pH values).

Circular dichroism

All CD spectra were recorded on a Jasco J720 spectro-photometer from 200 nm to 330 nm at 200 nmmin1 with a1 nm bandwidth using a 0. 2 cm pathlength cuvette. A totalof 20 repetitive scans were averaged and smoothed by theSavitzkyGolay smoothing filter in the CD softwarepackage Jasco Spectra Manager v1.5.

Surface area calculations

SASA was calculated using a 1.5 solvent probe usingthe radii set of Richards et al.61 (Supplementary DataTables S6 and S7) and the program Surface Racer.45 Thechanges in area were calculated as:

DSASA = SASAcomplex SASAfree RNA + SASAfree protein

The PDB files 2AB3 and 2AB7 were used for the freeproteins ZNF29 and ZNF29G29R. The free RNA structureof RREIIBTR is not available. Therefore, the PDB file forrhe free RNA structure was created from the Rev-boundRREIIBTR18 (PDB ID 1ETF) by manually deleting the Revpeptide. Docking the free protein helix in the major grooveof the free RNA created the PDB file used for therespective complex. The AOH for ZNF29G29R (free andbound) is assumed to be the same as that for ZNF29.

Acknowledgements

This work was supported by grants from the NIH(AI/GM47459) and the Georgia Cancer Coalition.A.M.S. was supported by the Molecular Basis ofDisease Program at GSU. We are indebted to DrDavid Wilson for discussions and helpful sugges-tions in the preparation of this manuscript. We thankDunay Busto, Xiaoguang Qu and Yoshiko Santosofor help with protein production and purification.

Supplementary Dataover a range of 50 200 mM NaCl. All NMR experimentswere done at 25 C (unless stated otherwise) on a BrukerAvance 600 using a 5 mm QXI triple resonance z-gradient

380Supplementary data associated with this articlecan be found, in the online version, at doi:10.1016/j.jmb.2009.07.066References

1. O'Reilly, M.M., McNally, M. T. & Beemon, K. L. (1995).Two strong 5 splice sites and competing, suboptimal 3splice sites involved in alternative splicing of humanimmunodeficiency virus type 1 RNA. Virology, 213,373385.

2. Purcell, D. F. & Martin, M. A. (1993). Alternativesplicing of human immunodeficiency virus type 1mRNAmodulates viral protein expression, replication,and infectivity. J. Virol. 67, 63656378.

3. Pollard, V. W. & Malim, M. H. (1998). The HIV-1 Revprotein. Annu. Rev. Microbiol. 52, 491532.

4. Malim, M. H. & Cullen, B. R. (1991). HIV-1 structuralgene expression requires the binding of multiple Revmonomers to the viral RRE: implications for HIV-1latency. Cell, 65, 241248.

5. Pomerantz, R. J., Seshamma, T. & Trono, D. (1992).Efficient replication of human immunodeficiencyvirus type 1 requires a threshold level of Rev:potential implications for latency. J. Virol. 66,18091813.

6. Malim, M. H. & Cullen, B. R. (1993). Rev and the fateof pre-mRNA in the nucleus: implications for theregulation of RNA processing in eukaryotes.Mol. Cell.Biol. 13, 61806189.

7. Kim, S. Y., Byrn, R., Groopman, J. & Baltimore, D.(1989). Temporal aspects of DNA and RNA synthesisduring human immunodeficiency virus infection:evidence for differential gene expression. J. Virol. 63,37083713.

8. Fischer, U., Pollard, V. W., Luhrmann, R., Teufel, M.,Michael, M. W., Dreyfuss, G. & Malim, M. H. (1999).Rev-mediated nuclear export of RNA is dominant overnuclear retention and is coupled to the Ran-GTPasecycle. Nucleic Acids Res. 27, 41284134.

9. Hope, T. J. (1999). The ins and outs of HIV Rev. Arch.Biochem. Biophys. 365, 186191.

10. Kjems, J., Frankel, A. D. & Sharp, P. A. (1991). Specificregulation of mRNA splicing in vitro by a peptidefrom HIV-1 Rev. Cell, 67, 169178.

11. Tiley, L. S., Malim,M.H., Tewary, H. K., Stockley, P. G.& Cullen, B. R. (1992). Identification of a high-affinityRNA-binding site for the human immunodeficiencyvirus type 1 Rev protein. Proc. Natl Acad. Sci. USA, 89,758762.

12. Brice, P. C., Kelley, A. C. & Butler, P. J. (1999). Sensitivein vitro analysis of HIV-1 Rev multimerization. NucleicAcids Res. 27, 20802085.

13. D'Agostino, D. M., Felber, B. K., Harrison, J. E. &Pavlakis, G. N. (1992). The Rev protein of humanimmunodeficiency virus type 1 promotes polysomalassociation and translation of gag/pol and vpu/envmRNAs. Mol. Cell. Biol. 12, 13751386.

14. Felber, B. K., Hadzopoulou-Cladaras, M., Cladaras,C., Copeland, T. & Pavlakis, G. N. (1989). rev proteinof human immunodeficiency virus type 1 affects thestability and transport of the viral mRNA. Proc. NatlAcad. Sci. USA, 86, 14951499.

15. Hadzopoulou-Cladaras, M., Felber, B. K., Cladaras,C., Athanassopoulos, A., Tse, A. & Pavlakis, G. N.(1989). The rev (trs/art) protein of human immuno-deficiency virus type 1 affects viral mRNA andprotein expression via a cis-acting sequence in theenv region. J. Virol. 63, 12651274.

16. Bartel, D. P., Zapp, M. L., Green, M. R. & Szostak, J. W.

HIV RREIIB RNAZinc Finger Interactions(1991). HIV-1 Rev regulation involves recognition ofnon-Watson-Crick base pairs in viral RNA. Cell, 67,529536.

17. Kjems, J., Calnan, B. J., Frankel, A. D. & Sharp, P. A.(1992). Specific binding of a basic peptide from HIV-1Rev. EMBO J. 11, 11191129.

18. Battiste, J. L., Mao, H., Rao, N. S., Tan, R., Muhan-diram, D. R., Kay, L. E. et al. (1996). helix-RNAmajorgroove recognition in an HIV-1 rev peptide-RRE RNAcomplex. Science, 273, 15471551.

19. Battiste, J. L., Tan, R., Frankel, A. D. &Williamson, J. R.(1994). Binding of an HIV Rev peptide to Revresponsive element RNA induces formation of purine-purine base pairs. Biochemistry, 33, 27412747.

20. Nielsen, M. H., Pedersen, F. S. & Kjems, J. (2005).Molecular strategies to inhibit HIV-1 replication.Retrovirology, 2, 10.

21. Peled-Zehavi, H., Horiya, S., Das, C., Harada, K. &Frankel, A. D. (2003). Selection of RRE RNA bindingpeptides using a kanamycin antitermination assay.RNA, 9, 252261.

22. McColl, D. J., Honchell, C. D. & Frankel, A. D. (1999).Structure-based design of an RNA-binding zincfinger. Proc. Natl Acad. Sci. USA, 96, 95219526.

23. Zapp,M. L., Young,D.W.,Kumar,A., Singh, R., Boykin,D.W., Wilson,W. D. & Green, M. R. (1997). Modulationof the Rev-RRE interaction by aromatic heterocycliccompounds. Bioorg. Med. Chem. 5, 11491155.

24. Chapman, R. L., Stanley, T. B., Hazen, R. & Garvey,E. P. (2002). Small molecule modulators of HIV Rev/Rev response element interaction identified by randomscreening. Antivir. Res. 54, 149162.

25. Friesen, W. J. & Darby, M. K. (2001). Specific RNAbinding by a single C2H2 zinc finger. J. Biol. Chem. 276,19681973.

26. Mishra, S. H., Shelley, C. M., Barrow, D. J., Jr, Darby,M. K. & Germann, M. W. (2006). Solution structuresand characterization of human immunodeficiencyvirus Rev responsive element IIB RNA targetingzinc finger proteins. Biopolymers, 83, 352364.

27. Berg, J. M. (2003). Fingering nucleic acids: the RNAdid it. Nat. Struct. Biol. 10, 986987.

28. Hall, T. M. (2005). Multiple modes of RNA recognitionby zinc finger proteins. Curr. Opin. Struct. Biol. 15,367373.

29. Lee, M. S., Mortishire-Smith, R. J. & Wright, P. E.(1992). The zinc finger motif. Conservation ofchemical shifts and correlation with structure. FEBSLett. 309, 2932.

30. Guex, N. & Peitsch, M. C. (1997). SWISS-MODEL andthe Swiss-PdbViewer: an environment for compara-tive protein modeling. Electrophoresis, 18, 27142723.

31. Baker, B. M. & Murphy, K. P. (1996). Evaluation oflinked protonation effects in protein binding reactionsusing isothermal titration calorimetry. Biophys. J. 71,20492055.

32. Nguyen, B., Stanek, J. & Wilson, W. D. (2006).Binding-linked protonation of a DNA minor-grooveagent. Biophys. J. 90, 13191328.

33. Li, H., Robertson, A. D. & Jensen, J. H. (2005). Veryfast empirical prediction and rationalization of proteinpKa values. Proteins: Struct. Funct. Genet. 61, 704721.

34. Tang, C. L., Alexov, E., Pyle, A. M. & Honig, B. (2007).Calculation of pKas in RNA: on the structural originsand functional roles of protonated nucleotides. J. Mol.Biol. 366, 14751496.

35. Record, M. T., Jr, Lohman, M. L. & De Haseth, P.(1976). Ion effects on ligand-nucleic acid interactions.J. Mol. Biol. 107, 145158.

36. Fried, M. G. & Stickle, D. F. (1993). Ion-exchange

HIV RREIIB RNAZinc Finger Interactionsreactions of proteins during DNA binding. Eur. J.Biochem. 218, 469475.37. Lohman, T. M., Overman, L. B., Ferrari, M. E. &Kozlov, A. G. (1996). A highly salt-dependententhalpy change for Escherichia coli SSB protein-nucleic acid binding due to ion-protein interactions.Biochemistry, 35, 52725279.

38. Privalov, P. L., Jelesarov, I., Read, C. M., Dragan, A. I.& Crane-Robinson, C. (1999). The energetics of HMGbox interactions with DNA: thermodynamics of theDNA binding of the HMG box from mouse sox-5.J. Mol. Biol. 294, 9971013.

39. Fukada, H. & Takahashi, K. (1998). Enthalpy and heatcapacity changes for the proton dissociation of variousbuffer components in 0.1 M potassium chloride.Proteins: Struct. Funct. Genet. 33, 159166.

40. Spolar, R. S. & Record, M. T., Jr (1994). Coupling oflocal folding to site-specific binding of proteins toDNA. Science, 263, 777784.

41. Spolar, R. S., Livingstone, J. R. & Record, M. T., Jr(1992). Use of liquid hydrocarbon and amide transferdata to estimate contributions to thermodynamicfunctions of protein folding from the removal ofnonpolar and polar surface from water. Biochemistry,31, 39473955.

42. Gomez, J. & Freire, E. (1995). Thermodynamicmapping of the inhibitor site of the aspartic proteaseendothiapepsin. J. Mol. Biol. 252, 337350.

43. Gomez, J., Hilser, V. J., Xie, D. & Freire, E. (1995). Theheat capacity of proteins. Proteins: Struct. Funct. Genet.22, 404412.

44. Ren, J., Jenkins, T. C. & Chaires, J. B. (2000). Energeticsof DNA intercalation reactions. Biochemistry, 39,84398447.

45. Tsodikov, O. V., Record, M. T., Jr & Sergeev, Y. V.(2002). Novel computer program for fast exactcalculation of accessible and molecular surface areasand average surface curvature. J. Comput. Chem. 23,600609.

46. Bergqvist, S., Williams, M. A., O'Brien, R. & Ladbury,J. E. (2004). Heat capacity effects of water moleculesand ions at a protein-DNA interface. J. Mol. Biol. 336,829842.

47. Holdgate, G. A., Tunnicliffe, A., Ward, W. H., Weston,S. A., Rosenbrock, G., Barth, P. T. et al. (1997). Theentropic penalty of ordered water accounts for weakerbinding of the antibiotic novobiocin to a resistantmutant of DNA gyrase: a thermodynamic andcrystallographic study. Biochemistry, 36, 96639673.

48. Kozlov, A. G. & Lohman, T. M. (2000). Largecontributions of coupled protonation equilibria tothe observed enthalpy and heat capacity changes forssDNA binding to Escherichia coli SSB protein.Proteins: Struct. Funct. Genet. 4(Suppl.), 822.

49. Freire, E. (1995). Thermal denaturation methods inthe study of protein folding. Methods Enzymol. 259,144168.

50. Cooper, A. (2005). Heat capacity effects in proteinfolding and ligand binding: a re-evaluation of the roleof water in biomolecular thermodynamics. Biophys.Chem. 115, 8997.

51. Clarke, C., Woods, R. J., Gluska, J., Cooper, A.,Nutley, M. A. & Boons, G. J. (2001). Involvement ofwater in carbohydrate-protein binding. J. Am. Chem.Soc. 123, 1223812247.

52. Mascotti, D. P. & Lohman, T. M. (1997). Thermody-namics of oligoarginines binding to RNA and DNA.Biochemistry, 36, 72727279.

53. Lu, D., Searles, M. A. & Klug, A. (2003). Crystal

381structure of a zinc-finger-RNA complex reveals twomodes of molecular recognition. Nature, 426, 96100.

54. Dunitz, J. D. (1995). Win some, lose some: enthalpy-entropy compensation in weak intermolecular inter-actions. Chem. Biol. 2, 709712.

55. Jones, S., Daley, D. T., Luscombe,N.M., Berman,H.M.& Thornton, J. M. (2001). Protein-RNA interactions: astructural analysis. Nucleic Acids Res. 29, 943954.

56. Bahadur, R. P., Zacharias, M. & Janin, J. (2008).Dissecting protein-RNA recognition sites. NucleicAcids Res. 36, 27052716.

57. Treger, M. & Westhof, E. (2001). Statistical analysis ofatomic contacts at RNA-protein interfaces. J. Mol.Recognit. 14, 199214.

58. Plateau, P. & Gueron, M. (1982). Exchangeable protonNMR without base-line distorsion, using new strong-pulse sequences. J. Am. Chem. Soc. 104, 73107311.

59. Kupce, E. (2001). Applications of adiabatic pulses inbiomolecular nuclear magnetic resonance. MethodsEnzymol. 338, 82111.

60. Humphrey, W., Dalke, A. & Schulten, K. (1996).VMD: visual molecular dynamics. J. Mol. Graph. 14,2728.

61. Richards, F. M. (1977). Areas, volumes, packing andprotein structure. Annu. Rev. Biophys. Bioeng. 6,151176.

382 HIV RREIIB RNAZinc Finger Interactions

Thermodynamic Profiling of HIV RREIIB RNAZinc Finger InteractionsIntroductionResults and DiscussionDesign of zinc finger and RNA mutantsBinding affinities of znf mutantsRREIIBTRBinding affinities of RREIIBTR mutantsZNF29G29RBinding linked protonation and pH effectsEffects of saltTemperature dependence of znfRNA binding enthalpyCooper models explain Cp and trends in binding enthalpy and entropy of mutant znfs and mutan.....

ConclusionMaterials and MethodsProteins and RNAIsothermal titration calorimetry (ITC)NMR spectroscopyCircular dichroismSurface area calculations

AcknowledgementsSupplementary DataReferences