RNA Sequence Elements Required for High Affinity Binding by the Zinc Finger Domain of Tristetraprolin

Tristetraprolin (TTP) binds AU-rich elements (AREs) encoded within selected labile mRNAs and targets these transcripts for rapid cytoplasmic decay. RNA binding by TTP is mediated by an ∼70-amino acid domain containing two tandemly arrayed CCCH zinc fingers. Here we show that a 73-amino acid peptide spanning the TTP zinc finger domain, denoted TTP73, forms a dynamic, equimolar RNA·peptide complex with a 13-nucleotide fragment of the ARE from tumor necrosis factor α mRNA, which includes small but significant contributions from ionic interactions. Association of TTP73 with high affinity RNA substrates is accompanied by a large negative change in heat capacity without substantial modification of RNA structure, consistent with conformational changes in the peptide moiety during RNA binding. Analyses using mutant ARE substrates indicate that two adenylate residues located 3–6 bases apart within a uridylate-rich sequence are sufficient for high affinity recognition by TTP73 (Kd <20 nm), with optimal affinity observed for RNA substrates containing AUUUA or AUUUUA. Linkage of conformational changes and binding affinity to the presence and spacing of these adenylate residues provides a thermodynamic basis for the RNA substrate specificity of TTP.

acceleration of deadenylation, where the poly(A) tail is progressively shortened in a 3Ј 3 5Ј direction prior to extremely rapid degradation of the mRNA body (reviewed in Ref. 4).
The mRNA-destabilizing activity of AREs is mediated by association of cytoplasmic trans-acting factors (reviewed in Ref. 5). Some factors, like members of the Hu family of RNA-binding proteins, prevent mRNA degradation (6,7), whereas others, like AUF1 (8 -10) and KSRP (11), promote rapid decay of AREcontaining transcripts. Association of tristetraprolin (TTP) with ARE-containing mRNAs also promotes their rapid cytoplasmic catabolism, involving acceleration of deadenylation rates (12,13). The significance of this mechanism is apparent from a TTP knock-out mouse model, where enhanced stability of tumor necrosis factor ␣ (TNF␣) mRNA in macrophages from TTP-deficient mice (14) produces constitutive enhancement of circulating TNF␣, ultimately leading to development of a systemic inflammatory syndrome (15).
TTP is the prototype of the CCCH family of eukaryotic tandem zinc finger proteins (reviewed in Ref. 16). Interactions with selected AU-rich RNA substrates occur through the zinc finger domain, which is necessary and sufficient for high affinity RNA recognition (12). Preliminary characterization of TTP substrate selectivity suggests a preference for RNA sequences containing AUUUA motifs, common among AREs from mRNAs encoding cytokines/lymphokines and inflammatory mediators (17)(18)(19). However, the apparent selectivity of TTP for a subset of ARE targets, coupled with the existence of several AREbinding proteins promoting common functions (i.e. mRNA destabilization), underscores the need to understand the biophysical basis for RNA substrate preferences of these trans-acting factors.
In this study, we continue our examination of sequencespecific RNA recognition by trans-acting, ARE-binding factors by evaluating interactions between the zinc finger domain of TTP and model ARE substrates. A previous study employing a synthetic 73-amino acid peptide containing the tandem zinc fingers of TTP, termed TTP73, demonstrated that specific RNA⅐peptide complexes could be formed in solution using RNA substrates spanning regions of the ARE from TNF␣ mRNA (20). Here we have used the interaction between the TTP73 peptide and a series of short RNA oligonucleotide substrates to characterize the thermodynamics, dissociation kinetics, and substrate requirements of TTP73⅐RNA complexes. While providing information about the dynamics and physical forces contributing to ARE recognition by the TTP zinc finger domain, this work also presents thermodynamic evidence supporting preliminary NMR data that indicated a conformational change in the TTP73 peptide accompanying association with high affinity RNA substrates (20). Subsequent studies using mutant RNA substrates provide evidence that conformational changes in the peptide⅐RNA complex are coupled to high affinity binding and are linked to the presence of two adenylate residues within a U-rich RNA sequence, thus establishing a thermodynamic basis for discrimination of RNA substrates by TTP. Furthermore, the moderately rapid dissociation kinetics of RNA complexes containing the TTP zinc finger domain support a model of regulated mRNA turnover involving dynamic competition between various ARE-binding proteins in the cytoplasm.

EXPERIMENTAL PROCEDURES
RNA Substrates-Sequences of RNA oligonucleotides used in this study are listed in Table I. Synthesis, deprotection of 2Ј-hydroxyl groups, and purification of RNA substrates were performed by Dharmacon Research (Lafayette, CO) or Integrated DNA Technologies (Coralville, IA). Dried RNA pellets were resuspended in 10 mM Tris⅐HCl (pH 8.0) and quantified by A 260 /A 495 as described previously (21,22). For RNA substrate Fl-ARE-AU 3 A-Cy5, contributions of the cyanine-5 (Cy5) moiety to A 260 and A 495 were considered negligible. Where indicated, 5Ј-hydroxyl RNA substrates were radiolabeled to specific activities of 3-5 ϫ 10 3 cpm/fmol using T4 polynucleotide kinase and [␥-32 P]ATP as described (21).
Synthetic Peptide Containing the TTP Tandem Zinc Finger Domain-A 73-amino acid peptide corresponding to residues 102-174 of human TTP (GenBank TM accession number NP_003398, Fig. 1A) was synthesized, purified, and refolded by Albachem Ltd. (Edinburgh, Scotland, UK), as described previously (20), and is termed TTP73. Lyophilized TTP73 peptide was resuspended in 10 mM Tris⅐HCl (pH 8.0) containing 2 mM dithiothreitol and frozen in aliquots at Ϫ80°C. Recovered peptide was quantified by Coomassie Blue-stained SDS-PAGE against a titration of bovine serum albumin. The isoelectric point and pH dependence of net charge for the TTP73 peptide were calculated using the pI application of the Biology Workbench version 3.2 (San Diego Supercomputer Center; www.workbench.sdsc.edu).
Gel Mobility Shift Assays (GMSAs)-Binding reactions including varying concentrations of TTP73 peptide and 0.2 nM 32 P-labeled RNA substrate were assembled in a final volume of 10 l containing 10 mM Tris⅐HCl (pH 8.0), 100 mM KCl, 2 mM dithiothreitol, 0.1 g/l bovine serum albumin, and 10% glycerol. Heparin (0.2 g/l) was included to inhibit nonspecific RNA binding activity. Where indicated, ZnCl 2 (5 M) or EDTA (0.5 mM) were also included in binding reactions. Prior to addition of RNA substrates, the TTP73 peptide was incubated for 20 min at room temperature in the binding mixture to permit coordination of Zn 2ϩ . Preliminary experiments indicated significantly poorer RNA binding activity for TTP73 when this Zn 2ϩ coordination step was omitted (data not shown). Upon introduction of RNA substrates, binding reactions were incubated on ice for 15 min and then fractionated by electrophoresis through 6% (40:1 acrylamide/bisacrylamide) native gels at 4°C. Following electrophoresis, gels were dried, and the location of reaction products was identified by PhosphorImager scan (Amersham Biosciences).
Fluorescence Anisotropy Assays-For quantitative analyses of RNAprotein binding equilibria, a fluorescence anisotropy-based assay was employed, similar to that described previously (21,23). Briefly, fluorescein (Fl)-labeled RNA substrates were incubated with TTP73 in reactions similar to those described for GMSAs (above), except that glycerol was omitted; final volume was brought to 100 l, and binding reactions were performed at 25°C unless otherwise noted. Fluorescence anisotropy was measured with a Beacon 2000 Fluorescence Polarization System (Panvera, Madison, WI) using fluorescein excitation ( ex ϭ 490 nm) and emission ( em ϭ 535 nm) filters. Association of TTP73 with Fl-RNA substrates was detected by an increase in anisotropy of Fl-RNA emission, because of restricted segmental motion and retarded rotational correlation time in the RNA⅐protein complex relative to the free RNA substrate (23)(24)(25).
In all experiments reported in this work, protein binding did not induce a change in the fluorescence quantum yield of Fl-RNA substrates. As such, the measured fluorescence anisotropy (A t ) of each binding reaction was interpreted as a function of the intrinsic anisotropy (A i ) and fractional concentration (f i ) of each fluorescent species using Equation 1 (24,26).  (23).
Here K represents the equilibrium association constant, and A R and A PR represent the intrinsic anisotropy values of the free and proteinassociated Fl-RNA substrates, respectively. A R was measured as the anisotropy of each RNA substrate in the absence of TTP73 (n Ն 3), whereas all other constants were solved by nonlinear least squares regression of A t versus [TTP73] using PRISM version 3.0 (GraphPad, San Diego). To measure RNA⅐peptide complex dissociation rates, binding reactions containing TTP73 and Fl-RNA substrates were assembled as described above with anisotropy measured at equilibrium (t ϭ 0). Following addition of a 5000-fold excess of unlabeled RNA, anisotropy was measured in intervals of 15 s, with five measurements taken for each time point. Thermodynamic Analyses of Peptide:RNA Binding Equilibria-Enthalpic (⌬H 0 ) and entropic (T⌬S 0 ) contributions to the free energy of TTP73 binding to RNA substrates were estimated from the temperature dependence of equilibrium binding constants by van't Hoff plots of ln(K) versus 1/T. This plot is linear when ⌬H 0 is independent of temperature (27,28) and may be resolved by Equation 3, where R is the gas constant (1.987 ϫ 10 Ϫ3 kcal⅐mol Ϫ1 ⅐K Ϫ1 ).
When significantly nonlinear, ln(K) versus 1/T plots were resolved using Equation 4, where ⌬C 0 P,obs represents the change in molar heat capacity, and T H and T S represent characteristic temperatures at which enthalpy and entropy, respectively, contribute no net energy to formation of the peptide⅐RNA complex (28). Solution of ⌬C 0 P,obs , T H , and T S then allowed the contributions of enthalpy and entropy to the energy of peptide⅐RNA complex formation to be calculated as a function of temperature using Equations 5 and 6, respectively (28).  Fl-ARE 13 Fl-AUUUAUUUAUUUA ARE 13 AUUUAUUUAUUUA Fl-ARE 9 Fl-UUAUUUAUU Fl-ARE 7 Fl-UAUUUAU Fl-U 13 Fl-UUUUUUUUUUUUU Fl-UUUUAUUUAUUUU-Cy5 R␤ UGGCCAAUGCCCUGGCUCACAAAUACCACUG a "Fl" and "Cy5" indicate the positions of the fluorescein and cyanine-5 moieties, respectively, conjugated to applicable RNA substrates.
FRET efficiency (E FRET ) is dependent on the scalar distance (r) between the donor and acceptor by Equation 7.
R 0 is the Förster distance for the donor-acceptor pair or the scalar distance yielding FRET efficiency of 50% (29,30). A value of R 0 for the Fl-Cy5 dye pair linked to single-stranded RNA was calculated using Equation 8.
Here 2 is the orientation factor for dipole-dipole coupling; n is the refractive index of the medium; Q D is the quantum yield of the donor in the absence of the acceptor; and J() is the overlap integral between donor emission and acceptor absorbance (29,30). For the Fl-ARE-AU 3 A-Cy5 RNA substrate, 2 was assumed to be 2/3, consistent with random orientation of donor and acceptor, because Fl and Cy5 were conjugated to the RNA moiety using 6-and 3-carbon linkers, respectively. Further support for this approximation is given by the very low anisotropy of fluorescein emission (A R ϭ 0.022 Ϯ 0.002) of the Fl-ARE-AU 3 A substrate at 25°C. Because all fluorescence measurements were performed in dilute aqueous solution, n was set to 1.333. Q D was measured as 0.40 using the Fl-ARE-AU 3 A substrate as described (31), with 3-aminofluoranthene in Me 2 SO (Q ϭ 0.32 at 25°C) as a reference standard (32). The overlap integral J() was calculated from the fluorescence emission spectrum of Fl (F D ()) and the absorbance spectrum of Cy5 (⑀ A ()) using Equation 9 (data not shown).
From these data, J() resolved to 1.49 ϫ 10 15 M Ϫ1 ⅐cm Ϫ1 ⅐nm 4 , yielding R 0 of 47 Å. Fluorescence emission spectra were measured using a Cary Eclipse fluorescence spectrophotometer (Varian Instruments, Walnut Creek, CA) with a Peltier temperature control cell ( ex ϭ 485 nm, 10 nm bandwidth) at 25°C. E FRET values were calculated from paired measurements of blank-corrected donor emission (518 nm) using RNA substrates containing either the FRET donor alone (Fl-ARE-AU 3 A; F D ) or the donor-acceptor pair (Fl-ARE-AU 3 A-Cy5; F DA ) by Equation 10 (30,33). 13 Substrate-Previously we demonstrated that TTP73 could bind a 20-nucleotide fragment of the ARE from TNF␣ mRNA with high affinity (20). Whereas this equilibrium was largely consistent with 1:1 stoichiometry of peptide:RNA, a minor but detectable complex consistent with two protein binding events was detected at peptide concentrations above 5 nM. However, we and others (17) have detected interactions between the TTP zinc finger domain and RNA substrates as small as 9 -13 nucleotides. Accordingly, this study focused on the recognition of small (Յ13 nucleotides) RNA ligands by TTP73, to minimize the potential for multiple peptide binding events.

The Zinc Finger Domain of TTP Forms a Dynamic, Equimolar Complex with the ARE
The RNA substrate ARE 13 encodes a sequence contained within the TNF␣ ARE. By GMSA, TTP73 formed a single RNA⅐peptide complex with the radiolabeled ARE 13 ligand in a concentration-dependent manner (Fig. 1B). Binding was absolutely dependent on the presence of Zn 2ϩ , verifying that zinc finger-independent mechanisms did not make significant contributions to RNA binding activity in this system. Also, RNA binding activity was dependent on the ARE sequence, because TTP73 did not form detectable complexes with the unrelated R␤ RNA substrate. Using the fluorescence anisotropy assay, TTP73 peptide binding to a fluorescein-labeled ARE 13 sub-strate (Fl-ARE 13 ) was well resolved by the binary equilibrium binding model (Equation 2), indicated by a strong coefficient of determination (R 2 ϭ 0.9901) and random distribution of residuals about the regression solution. For assembly of the TTP73⅐Fl-ARE 13 complex, the bimolecular association constant, K, resolves to 2.8 Ϯ 0.5 ϫ 10 8 M Ϫ1 (mean Ϯ nϪ1 , n ϭ 4), yielding a dissociation constant (K d ϭ 1/K) of 3.6 nM. Consistent with the GMSA experiments (Fig. 1B), TTP73 binding to the Fl-ARE 13 substrate was absolutely dependent on Zn 2ϩ , because no RNA binding activity was detected in reactions lacking the cation (Fig. 2B). Interaction between TTP73 and this RNA substrate was also relatively dynamic in solution, with off-rate analyses yielding an apparent first-order dissociation constant (k off ) of 3.2 Ϯ 0.4 ϫ 10 Ϫ2 s Ϫ1 at 25°C (regression solution Ϯ 95% confidence interval), corresponding to a complex half-life of 22 s in solution (Fig. 3).
Contribution of Ionic Interactions to Stability of the TTP73⅐Fl-ARE 13 Complex-Because of the polyanionic nature of RNA ligands, association with RNA-binding proteins often includes significant contributions from ionic interactions (34,35). Several features of TTP73 suggested that ion pairs might play a significant role in stabilizing its association with AUrich RNA substrates. (i) The TTP73 peptide has a predicted isoelectric point of 9.34 and is expected to exhibit a net charge of ϩ5 at pH 8. (ii) Coordination of two Zn 2ϩ ions contributes additional positive charge to the peptide. (iii) Several amino acid residues with basic side chains are located within each zinc finger motif. (iv) NMR structural analysis of an ARE⅐peptide complex containing the zinc finger domain of a TTP-related peptide, TIS11d, shows four basic amino acid side chains in close proximity to the RNA phosphodiester backbone (36). These basic residues are conserved in the zinc finger domain of TTP. To evaluate the net influence of ion pairs on the stability of TTP73⅐Fl-ARE 13 ribonucleoprotein (RNP) complexes, equilibrium binding constants were measured under conditions of increasing ionic strength by varying the concentration of KCl from 50 to 500 mM, and plotted as log(K) versus Ϫlog[K ϩ ] (Fig.  4). The slope of this plot is influenced by the number ion pairs formed during RNA⅐protein complex assembly (37). The salt dependence of TTP73⅐Fl-ARE 13 RNP formation yielded ϪѨlog(K)/Ѩlog[K ϩ ] ϭ 1.1 Ϯ 0.2 (regression solution Ϯ95% confidence interval). This value is similar to RNP complexes containing the trp RNA-binding attenuation protein (38), RNase A (35), and Hsp70 (39) but significantly lower than that expected for RNPs with large ionic components, including U1A (40) and ribosomal protein S8 (37) bound to cognate RNA substrates (ϪѨlog(K)/Ѩlog[K ϩ ] Ͼ3). The modest value of this parameter indicates that ionic interactions make significant but relatively minor contributions to the stability of the TTP73⅐Fl-ARE 13 RNP, confirming that nonelectrostatic mechanisms, possibly including base-specific contacts or stacking interactions, are critical for assembly of this RNA⅐protein complex.
RNA Sequence Elements Contributing to High Affinity Binding of the TTP73 Peptide-AREs constitute a diverse collection of mRNA-destabilizing sequences and may be recognized by many different cytoplasmic RNA-binding proteins (4,5). Some ARE-binding factors, including AUF1 (23, 41), Hsp70 (39), HuA (18), and HuD (17), appear to show little selectivity among ARE templates, other than a general preference for U-rich sequences. However, some recent studies (17)(18)(19) indicate that TTP binding may be selective for RNA substrates containing AUUUA motifs, whose appearance is restricted to a subset of known AU-rich mRNA-destabilizing sequences. To identify specific RNA sequence elements contributing to TTP association with U-rich RNA sequences, the binding affinity of the TTP73 peptide for a series of mutant ARE substrates was measured using the fluorescence anisotropy assay. Data generated through these experiments (Table II) revealed three principal RNA sequence features contributing to optimal TTP73 binding. First, optimal TTP73 binding requires an RNA substrate greater than 7 nucleotides in length (cf. Fl-ARE 7 versus Fl-ARE 9 ) and may be effectively attained with substrates as small as nine nucleotides (cf. Fl-ARE 9 versus Fl-ARE 13 and Fl-ARE-AU 3 A). Second, optimal binding of TTP73 requires two adenylate residues within the U-rich motif (cf. Fl-ARE-AU 3 U and Fl-ARE-UU 3 A versus Fl-ARE-AU 3 A; Fl-ARE 9 -AU 3 U and Fl-ARE 9 -UU 3 A versus Fl-ARE 9 ), with each adenylate residue contributing to total binding affinity (i.e. K TTP73⅐Fl-ARE 13 complexes were prepared by incubation of TTP73 peptide (20 nM) with Fl-ARE 13 substrate (0.2 nM). At equilibrium, the fluorescence anisotropy of the mixture was measured (t ϭ 0). Following addition of a 5000-fold excess of unlabeled ARE 13 substrate, anisotropy measurements were taken as a function of time (solid circles), with each point representing the mean Ϯ nϪ1 of three independent reactions. Data were analyzed by nonlinear regression using a first-order exponential decay function (solid line). Specificity is shown using a 5000-fold excess of unlabeled R␤ substrate as competitor, which was unable to displace the Fl-ARE 13 ligand (open circles). for Fl-U 13 Ͻ Fl-ARE-AU 3

U and Fl-ARE-UU 3 A Ͻ Fl-ARE-AU 3 A). A cytidine residue may not substitute for either adenylate within the AUUUA motif (cf. Fl-ARE-AU 3 C versus Fl-ARE-AU 3 U; Fl-ARE-CU 3 A versus Fl-ARE-UU 3 A), and guanosine makes only minor contributions to TTP73 binding affinity in these positions (cf. Fl-ARE-AU 3 G versus Fl-ARE-AU 3 U; Fl-ARE-GU 3 A versus Fl-ARE-UU 3 A). Third, TTP73
binding is highly sensitive to the spacing between adenylate residues, showing optimal affinity for RNA substrates containing the canonical ARE motif AU 3 A. However, TTP73 also displayed strong affinity for substrates containing AU 4 A, and intermediate affinity for those containing AU 2 A and AU 5 A, relative to substrates containing a single adenylate residue (i.e. K for Fl-ARE-AU 3 A Ͼ Fl-ARE-AU 4 A Ͼ Fl-ARE-AU 2 A, Fl-ARE-AU 5 A Ͼ Fl-ARE-AU 3 U, Fl-ARE-UU 3 A). By contrast, placement of a single uridylate between two adenylate residues inhibits TTP73 peptide binding relative to poly(U) substrates containing a single A (cf. Fl-ARE-AUA versus Fl-ARE-AU 3 U and Fl-ARE-UU 3 A). Together, these data indicate that U-rich RNA sequences containing the motif AU n A (n ϭ 2-5) may serve as moderate to high affinity (K d Ͻ20 nM at 25°C) TTP-binding sites.
Specific RNA Substrates Induce a Significant Negative Change in Heat Capacity upon TTP73⅐RNA Complex Formation-Contributions of enthalpy and entropy to the stability of TTP73⅐RNA complexes were evaluated by the temperature dependence of equilibrium binding constants. A van't Hoff plot of ln(K) versus 1/T for TTP73 binding to the Fl-ARE 13 substrate was clearly nonlinear (Fig. 5A), indicating that ⌬H 0 varied with temperature. Nonlinear regression using Equation 4 indicated that formation of the TTP73⅐Fl-ARE 13 RNP was accompanied by a large negative change in heat capacity (Table III). Solutions of ⌬H 0 and T⌬S 0 versus T revealed that changes in enthalpy and entropy accompanying TTP73⅐Fl-ARE 13 complex formation are strongly temperature-dependent (Fig. 5B). Below 19°C (292 K), TTP73 binding to the Fl-ARE 13 substrate is driven entirely by favorable changes in entropy (⌬S 0 Ͼ 0), which compensate for unfavorable enthalpic changes (⌬H 0 Ͼ 0). Between 19 and 24°C (297 K), both entropy and enthalpy contribute to complex stability, whereas above 24°C, formation of the TTP73⅐Fl-ARE 13 RNP is driven entirely by enthalpy, which must overcome unfavorable entropic changes in this system. In general, negative changes in molar heat capacity resulting from noncovalent interactions between proteins and nucleic acids are indicative of conformational changes in protein and/or nucleic acid components (27,28,42). Previously, we demonstrated (20) that the C-terminal zinc finger of the TTP73 peptide is largely unordered in solution but becomes ordered following association with high affinity RNA substrates. Based on these observations, we contend that the large negative value of ⌬C 0 P,obs associated with TTP73⅐Fl-ARE 13 RNP complex formation is at least partly due to structural remodeling of the C-terminal zinc finger of TTP73.
To determine whether changes in heat capacity accompanying TTP73⅐RNA complex formation might be coupled to specific RNA sequence elements, additional van't Hoff analyses were performed using selected mutant ARE substrates (Fig. 6). Tested RNA ligands Ն9 nucleotides in length containing the canonical AUUUA motif (Fl-ARE 9 , Fl-ARE-AU 3 A) exhibited significantly nonlinear relationships between ln(K) and 1/T and yielded thermodynamic constants that did not significantly differ from those describing TTP73 binding to the Fl-ARE 13 substrate (Table III). By contrast, no significant change in heat capacity accompanied formation of the TTP73⅐Fl-ARE 7 complex (⌬C P,obs 0 ϭ 0.0 Ϯ 0.3 kcal/mol, data not shown). Together, these experiments indicated that RNA substrates containing a single AUUUA motif were able to direct a change in heat  capacity upon TTP73⅐RNA complex formation but that some minimal number of uridylate residues flanking this motif was also required.
Additional mutant RNA substrates addressed the potential of each adenylate residue within the AUUUA motif to contribute to ⌬C P,obs 0 . Replacement of the 3Ј-adenylate with uridylate (Fl-ARE-AU 3 U) completely abrogated the change in heat capacity upon TTP73 binding (⌬C P,obs 0 ϭ Ϫ0.4 Ϯ 0.4 kcal/mol, data not shown), permitting a linear solution for ln(K) versus 1/T. However, association of TTP73 with the 5Ј A3 U substitution mutant substrate (Fl-ARE-UU 3 A) yielded a large negative ⌬C P,obs 0 (Table III) but also displayed a severe temperature dependence that precluded any significant RNA binding activity above 30°C (K d Ͼ 500 nM). From these observations, we infer that the 3Ј-adenylate residue within the AUUUA motif is essential for the observed change in heat capacity upon TTP73 binding. By contrast, the 5Ј-adenylate is dispensable for this change in heat capacity but strongly prevents binding at elevated temperatures, when enthalpic contributions to binding energy are required to overcome the entropic penalty of conformational change (i.e. Fig. 5B).
Conformational Changes Accompanying TTP73 Binding to a High Affinity RNA Substrate Do Not Involve Significant Alteration of RNA Structure-Changes in heat capacity resulting from TTP73 binding to high affinity RNA substrates Fl-ARE 13 , Fl-ARE 9 , and Fl-ARE-AU 3 A (Table III) are consistent with conformational changes in one or more components of these TTP73⅐RNA complexes. Although a previous NMR study (20) indicated that the inherently unfolded C-terminal zinc finger of TTP73 became ordered as a result of interactions with AUUUA-containing RNA substrates Ն9 nucleotides in length, it was unclear whether the RNA moiety also experienced significant conformational changes during peptide⅐RNA complex formation. To determine whether TTP73 may alter local RNA structure, FRET was employed to measure changes in the distance between the 5Ј and 3Ј termini of a high affinity RNA target as a function of TTP73 peptide binding.
The RNA substrate Fl-ARE-AU 3 A-Cy5 (Fig. 7A) incorporates a FRET donor (Fl) and acceptor (Cy5) at its 5Ј-and 3Ј-ends, respectively (Table I). The emission spectrum of this substrate in the absence of TTP73 revealed characteristic peaks for both Fl (518 nm) and Cy5 (667 nm) (Fig. 7B, solid line). Digestion of the RNA substrate with RNase A resulted in increased donor emission with a concomitant decrease in acceptor emission (Fig. 7B, dotted line), indicating that the distance (r) between the 5Ј-and 3Ј-ends of the intact Fl-ARE-AU 3 A-Cy5 RNA substrate was within the detection limit of FRET. Based on Fl emission from RNA substrates Fl-ARE-AU 3 A-Cy5 (F DA ; Fig.  7C, solid line) and Fl-ARE-AU 3 A (F D ; Fig. 7D, solid line), E FRET for the unbound Fl-ARE-AU 3 A-Cy5 substrate was calculated as 0.34 Ϯ 0.01 by Equation 10. However, addition of TTP73 (500 nM) did not appreciably alter the emission spectra of ARE-AU 3 A substrates containing the donor fluorophore alone (Fl-ARE-AU 3 A; Fig. 7D) or the donor-acceptor pair (Fl-ARE-AU 3 A-Cy5; Fig. 7E). Samples spanning a titration of TTP73 peptide concentrations yielded no significant change in E FRET (Fig. 7F),  consistent with inter-fluorophore distances of 52-54 Å for all cases, based on E FRET values from 0.31 to 0.35. Evaluation of TTP73 binding to the Fl-ARE-AU 3 A-Cy5 substrate by changes in Fl anisotropy verified that peptide binding was not inhibited by the 3Ј-Cy5 modification (data not shown). Because the distance between the RNA 5Ј and 3Ј termini does not detectably change as a result of TTP73 peptide binding, we conclude that gross alterations in RNA conformation do not substantially contribute to changes in system heat capacity accompanying TTP73⅐RNA complex formation. These findings do not, however, preclude the possibility that minor, localized alterations in RNA structure may contribute to changes in heat capacity upon TTP73 binding to selected RNA substrates. DISCUSSION AREs direct the rapid decay of many cellular mRNAs, yet the sequence diversity of these mRNA-destabilizing sequences and the plethora of proteins known to interact with them suggest that competitive or combinatorial binding events may contrib-ute to the regulation of ARE function. The former possibility is supported by observations that several ARE-binding factors, including the mRNA-destabilizing factor AUF1 (21), the RNAbinding domain of TTP (this work), Hsp70 (39), and the mRNA-stabilizing factor HuD (17) all bind AU-rich RNA substrates through moderately dynamic mechanisms, with dissociative half-times varying from 10 to 120 s. By contrast, some other sequence-specific nucleic acid-binding proteins, including U1A (40) and the TATA-binding protein (43), exhibit significantly slower dissociation kinetics (t1 ⁄2 Ͼ20 min). Accordingly, it is conceivable that cellular control of ARE-directed mRNA decay kinetics may include modulation of the cytoplasmic concentration, RNA binding activity, or even the RNA-binding dynamics of selected ARE-binding proteins.
Although many proteins associate with AU-rich mRNA substrates, emerging data indicate that different ARE-binding factors display a spectrum of distinct but overlapping RNA sequence preferences. From comparisons of thermodynamic parameters describing binding of the TTP73 peptide with both high and low affinity RNA substrates, we have shown that high affinity interactions between the TTP zinc finger domain and ARE-like RNA substrates involve several features. First, the two adenylate residues contained within the canonical AUUUA motif are required for high affinity binding, but some variation is permitted in the spacing between these residues. Second, uridylate residues flanking the AUUUA motif contribute to peptide binding. Together, these RNA substrate requirements for high affinity TTP73 binding account for the following results: (i) in vitro RNA selection experiments using recombinant TTP, where a preference for AUUUA motifs flanked by uridylate residues was observed (19); (ii) the ability of AREs from different mRNAs to compete for TTP binding activity in cell extracts (18). In the latter study, only RNA competitors containing one or more AUUUA or AUUUUA motifs flanked by uridylate residues were able to effectively displace TTP from a high affinity ARE substrate.
Recently, an NMR study using the tandem zinc finger domain of a TTP-related protein, TIS11d, indicates that both zinc fingers display similar structures upon binding an ARE substrate and interact via a 3Ј 3 5Ј polarity with tandem UAUU motifs (36). Extrapolation of the TIS11d⅐ARE complex model to the TTP73 peptide is supported by the strong binding affinity (K d Ͻ4 nM) and negative changes in heat capacity observed for binding events between TTP73 and all tested RNA substrates containing tandem UAUU motifs (Fl-ARE 13 , Fl-ARE 9 , and Fl-ARE-AU 3 A). However, additional mechanistic details of ARE recognition by TTP may be extracted from binding studies involving mutant ARE substrates and support a model whereby conformational changes, most likely including the peptide C-terminal zinc finger, are coupled to the RNA substrate sequence and contribute to optimal RNA binding activity. First, high affinity RNA binding requires a conformational transition denoted by a change in heat capacity. Given (i) that the RNA substrate does not exhibit a significant structural change upon TTP73 binding (Fig. 7), and (ii) previous observations that structural ordering of the TTP73 C-terminal zinc finger requires association with high affinity RNA substrates (20), we conclude that conformational changes in the C-terminal zinc finger domain contribute significantly to the large negative change in heat capacity accompanying TTP73⅐RNA complex formation. Second, induction of conformational changes likely requires peptide-RNA contacts involving both zinc finger domains. The N-terminal finger of TTP may interact independently with RNA substrates of the type UUUAUUU; however, binding affinity is very poor (K d ϭ 4.7 M) (44). TTP73 binding to the Fl-ARE-AU 3 U substrate yielded improved sta- FIG. 7. Influence of TTP73 binding on local RNA structure measured by FRET. A, schematic of an RNA substrate labeled with Fl at its 5Ј-end and Cy5 at the 3Ј-end. Following excitation at 485 nm, the Fl moiety may release energy by quantum emission ( max ϭ 518 nm) or by nonquantum events, including excitation of Cy5 by FRET. B, emission spectra of the Fl-ARE-AU 3 U-Cy5 RNA substrate (2 nM) prior to (solid line) and following (dotted line) digestion with RNase A (1 g/ml, 20 min, 25°C). Positions of Fl and Cy5 emission peaks are indicated. C and D, emission spectra of the Fl-ARE-AU 3 U (C) and Fl-ARE-AU 3 U-Cy5 (D) RNA substrates prior to (solid line) and following (dotted line) addition of TTP73 peptide (500 nM). E, values of E FRET for the Fl-ARE-AU 3 U-Cy5 RNA substrate as a function of TTP73 concentration, solved as described under "Experimental Procedures." Each point indicates the mean Ϯ spread of two independent experiments. bility (Table II) but without detectable changes in heat capacity. Although this RNA substrate contains a UAUU motif, which accounts for all contacts with a single zinc finger in the case of TIS11d (36), it is possible that the paucity of upstream uridylate residues precludes selected peptide-RNA interactions required for conformational change. By contrast, TTP73 binding to Fl-ARE-UU 3 A was associated with a change in heat capacity but without the improvement in binding affinity observed with the Fl-ARE 13 , Fl-ARE 9 , or Fl-ARE-AU 3 A substrates. As such, we propose that the association of the Nterminal zinc finger with the UAUU motif and accessibility of the C-terminal finger to a significant U-rich sequence is sufficient to promote conformational changes indicated by the decrease in heat capacity. However, the high entropic cost of this transition requires a compensatory improvement in binding enthalpy, which is, at least in part, dependent on contact between the C-terminal finger and an adenylate residue within the upstream U-rich sequence. Based on the TIS11d⅐RNA structure, hydrogen bonds form between the upstream adenylate and the C-terminal zinc finger at two sites on the base (36). Most interesting, one such contact common to both zinc finger-UAUU interactions utilizes N-7 of adenosine as the hydrogen bond acceptor, which may explain why A3 G substitutions (Table II; ⌬⌬G ϭ 1.3 kcal/mol), which would not disrupt this contact, were energetically less costly than A3 C/U substitutions (⌬⌬G ϭ 1.7 to 2.0 kcal/mol) at either position. Finally, the moderate binding affinity (K d Ͻ20 nM) of TTP73 for RNA substrates containing adenylate residues separated by 2-5 uridylate residues suggests that the peptide linker between the zinc fingers also possesses some degree of flexibility.
In conclusion, the data presented in this work, together with the current structural model of the TIS11d⅐ARE complex (36), suggest that binding of the N-terminal zinc finger of TTP with a UAUU (or closely related) motif and interactions between the C-terminal finger and upstream U-rich RNA sequences are required to induce conformational changes in the C-terminal peptide domain. However, a second adenylate residue located 3-6 bases 5Ј of the first is required to stabilize the RNA⅐peptide complex (K d Ͻ20 nM), with optimal affinity observed for RNA substrates containing AUUUA or AUUUUA. Within the context of the complete TTP peptide, it remains possible that sequences flanking the zinc finger domain may alter the energetics and/or dynamics of RNP complex formation on AU-rich RNA substrates. However, the studies described herein establish a thermodynamic basis for RNA substrate specificity by the RNA-binding domain of TTP and define sequence features permitting high affinity RNA recognition by this protein.