Inosine

ABSTRACT The formation of sequence-specific complexes of TATA binding protein (TBP) with the minor groove of DNA results in the burial of large nonpolar surfaces and the exclusion of water from these interfaces. The release of water is thus expected to provide a significant entropic driving force for formation of the transcription-preinitiated complexes mediated by the binding of TBP to specific sequences. In this article are described equilibrium-binding studies of Saccharomyces cerevisiae TBP to 14 bp oligonucleotides bearing either the tightly bound and efficiently transcribed adenovirus major late promoter (TATAAAAG) or its inosine-substituted derivative (TITIIIIG) as a function of neutral osmolyte concentration. These two DNA sequences present the same pattern of minor groove hydrogen-bond donors and acceptors to the protein. TBP-DNA complex formation was monitored by steady-state fluorescence resonance energy transfer measurements of the oligonucleotides end-labeled with fluorescein (donor) and TAMRA (acceptor). Correct interpretation of the results obtained with the inosine-substituted sequence required careful consideration of the optical properties of the dyes as a function of osmolyte concentration to demonstrate that the relative change in the end-to-end distances for TATAAAAG- and TITIIIIG-bearing oligonucleotides is the same upon TBP binding. Although the affinity of TBP is slightly greater for the adenosine compared with the inosine-substituted TATA sequence in the absence of osmolyte, the end-to-end distances of the bound DNA in complex with TBP, the enthalpic and electrostatic components of binding, are identical within experimental precision. However, ~18 additional molecules of water are released upon TBP binding the TATAAAAG as compared with the TITIIIIG sequence resulting in an entropic advantage to the binding of the natural promoter sequence. These results are considered with regard to differences in the flexibility and hydration of the two DNA sequences.

INTRODUCTION

The stability and sequence specificity of protein-DNA complexes is dependent on thermodynamic variables including temperature, pressure, pH, ion type, and ion concentration (Parsegian et al., 2000; Record, Jr. et al., 1985, 1998; Sidorova and Rau, 2001). Atomic resolution structures have been solved for a growing number of sequence-specific complexes of proteins with DNA, including many gene regulatory proteins. These structures show intimate association of site-specifically bound proteins with the DNA with water excluded from the macromolecular interface. The replacement of contacts with water by protein-DNA contacts is observed in some of the crystallographic studies (Pabo and Sauer, 1992).

The hydration shell that surrounds proteins and DNA in solution is perturbed upon formation of macromolecular complexes. The release of water bound to the interacting surfaces of the reactants into the bulk solution provides an entropic driving force for complex formation. Some of the complexes whose atomic resolution structures have been solved have also been the subjects of detailed thermodynamic investigations. These biochemical studies are uniformly consistent with the displacement of water from protein-DNA interfaces (Fried et al., 2002; Jen-Jacobson et al., 2000; Parsegian et al., 1995, 2000; Record, Jr. et al., 1998; Sidorova and Rau, 1996; Vossen et al., 1997).

A measure of water release upon complex formation can be obtained by osmotic stress measurements (Parsegian et al., 1995; Sidorova and Rau, 1996). This approach uses small neutral solutes (perturbing solutes; see Record, Jr. et al., 1998) that do not bind with high affinity to the reactants whose concentration affects the thermodynamic activity of the water (the osmotic effect) and the biopolymers (preferential interactions). Since preferential interactions are dependent upon the chemical composition of the solute, these effects can be differentiated by measurements conducted using multiple osmolytes. Although it has been suggested that the osmotic-stress technique provides only a lower boundary to the number of water molecules released due to steric exclusion (Courtenay et al., 2000), comparative studies provide a way to circumvent this limitation. For example, osmotic-stress analysis of the binding of a protein to specific and nonspecific sequences conducted under otherwise identical experimental conditions allows the effect of steric exclusion to be extracted from the two sets of data (Fried et al., 2002; Sidorova and Rau, 1996, 2001).

The assembly of the TATA binding protein (TBP) to promoter DNA is required for the initiation of eukaryotic gene transcription by all three RNA polymerases. The biological function of TBP is clearly linked to its specific interaction with a myriad array of other transcription control proteins and, at least for RNA polymcrase II, to its specific binding to particular sequences of DNA. Crystallographic studies have shown that TBP binds to the minor groove of DNA exposing a broad flat surface to the protein and dramatically bending the DNA (Burley and Roeder, 1996). The binding and bending of the bound DNA has been investigated in solution by gel permutation (Starr et al., 1995), gel phasing (Bareket-Samish et al., 2000), and fluorescence resonance energy transfer (FRET) (Parkhurst et al., 1996, 1999; Powell et al., 2001, 2002). In addition, time-resolved FRET studies conducted as a function of the TATA box sequence revealed dramatic differences in both the end-to-end distances and the distribution of distances for the TBP-bound DNA that are osmolyte-sensitive for lower affinity sequences (Wu et al., 2001a,b).

A comparative approach has been used in this study to explore the role of dehydration in the sequence-specific binding of the Saccharomyces cerevisiae TATA binding protein (TBP) to the high affinity sequence TATAAAAG and its inosine-substituted derivative (TITIIIIT). These sequences present an identical pattern of hydrogen-bond donors and acceptors in the minor groove that contribute to TBP binding (Lee et al., 1991; Starr and Hawley, 1991) and both sequences support the similar level of transcription activity (Lee et al., 1997). Despite this similarity, these two sequences have very different physical properties. Indeed, the inspiration for these experimental studies was a computational analysis of these two sequences and their interaction with TBP, particularly with regard to hydration and water release upon complex formation (Pastor et al., 1999). The sequence of adenosine is calculated to bind 19 more water molecules in the minor groove as compared with that of inosine. Thus, comparison of TBP binding to the two sequences is an ideal context within which to explore the entropic consequences of water release upon the TBP-DNA complex formation.

The accurate analysis of the steady-state FRET intensity changes was critical to the conduct of these studies due to the small number of water molecules released upon complex formation. In addition, careful consideration was given to the effect of osmolyte on the spectroscopic parameters of the fluorescent dyes attached to the oligonucleotides. A generally applicable formalism is presented for the calculation of the efficiency of energy transfer that accounts for the properties of a particular donor-acceptor pair.

MATERIALS AND METHODS

Protein, DNA, and other reagents

S. cerevisiae TBP was expressed in Escherichia coli and purified as previously described (Parkhurst et al., 1996; Petri et al., 1995). An extinction coefficient of 13.4 × 10^sup 3^ M^sup -1^ × cm^sup -1^ was used to calculate the concentration of the protein. Triethylene glycol (TEG) and methyl glucoside (methyl-alpha-d-glucopyranoside; MG) were purchased from Sigma (St. Louis, MO) and used without additional purification. The osmolal concentration of TEG and MG were calculated from the dependence of their osmotic pressure on molal concentration (N. Sidorova and D. Ran, personal communication). Labeled and unlabeled DNA oligonucleotides were obtained from TriLink Biotechnologies (San Diego, CA). The concentration of oligonucleotides was determined using extinction coefficients calculated from the known basepair composition correcting for the absorbance of fluorescein or TAMRA at 260 nm.

The strategy used in this study of creating a double-labeled duplex from singly-labeled complementary oligonucleotides differs from the doublelabeled single strand annealed with the unlabeled complementary strand used in previous studies of TBP interactions (Parkhurst et al., 1996, and subsequent publications). Creation and purification of the double-labeled oligonucleotide is synthetically demanding. Other FRET methods are based on the hybridization of single-strand, singly-labeled oligonucleotides followed by purification of the double-labeled duplex (e.g., Clegg, 1992). The present approach is based on a titration of the donor-labeled oligonucleotide with the acceptor-labeled complement to determine the equimolar ratio of two strands of DNA within the resultant duplex. This approach does not require additional purification of the DNA probe, is not sensitive to errors in the stoichiometry of the two DNA strands, and is very accurate. As will be presented in Results, it was the unexpected fluorescence behavior of the inosine-substituted oligonucleotide that prompted an analysis of the properties of the single- and double-labeled duplexes in the absence of TBP and the development of this approach.

The fluorescence of singly versus doubly labeled oligonucleotides

The fluorescence emission spectra of the fluorescein and TAMRA singly-labeled oligonucleotides were measured before and after annealing with unlabeled complementary strands. An increase in the emission of fluorescein at 520 nm was observed for the adenosine-bearing oligonucleotide in contrast to the decrease measured for the inosine-bearing one (compare Fig. 1, A and B, curves 1 and 3). A decrease in the fluorescence at 580 nm upon duplex formation was observed for both of the TAMRA-labeled oligonucleotides (Fig. 1, A and B, curves 4 and 5). These results reflect a DNA duplex-dependent sequence-specific interaction of fluorescein with the inosine-containing oligonucleotide. Saturation of each of the two singly-labeled duplexes with TBP yielded no detectable changes in their emission spectra (Fig. 1 A and B, curves 2 and 6).

In contrast, when fluorescence of the doubly-labeled TATAAAAG duplex is measured, a new shoulder at 580 nm in the fluorescence spectrum appears due to resonance energy migration (Fig. 2; see also Parkhurst et al., 1996). The total emission (Fig. 2, curve 1) at 580 nm is composed of two components-donor (fluorescein) emission due to direct excitation, and the emission of the acceptor (TAMRA). To resolve the portion of the overall fluorescence caused by the acceptor, the emission spectrum of the donor (Fig. 1 A, curve 2) was normalized to the emission spectrum of the doublelabeled duplex at 520 nm, where only the donor emits (Fig. 2, curve 2). The difference spectrum that results from the subtraction of spectrum 2 from 1 yields spectrum 3 (Fig. 2). The 580-nm fluorescence emission in difference spectrum 3 (denoted dF^sub 490/580^) consists of two components-direct TAMRA excitation and FRET. This parameter dF^sub 490/580^ is used for all analyses of FRET efficiency. The perfect match of this difference spectrum (3) with the normalized spectrum of TAMRA-labeled duplex (5) demonstrates the absence of interactions except FRET between the donor and acceptor within the duplex (Fig. 2). Thus, measurement of the fluorescencc of the double-labeled duplex DNA al 580 nm provides a measure of the FRET efficiency, E, with this pair of donors and acceptors.

The procedure used in this study involves the annealing of singly-labeled oligonucleolides to make a doubly-labeled duplex. A titration is used to establish a proper ratio of annealed donor/acceplor-labeled oligonucleotides. The equilibrium between oligonucleotides bearing a donor (D) and an acceptor (A) existing in the solution along titration is shown in Scheme 1. Fig. 3 shows the results of a titration of the fluorescein-labeled oligonucleotide (F14) with the T14 complement over the concentration range that spans zones I-IV (Scheme 1) in which TAMRA emission at 580 nm was monitored after excitation at cither 490 or 555 nm. Excitation at 490 nm results in FRET and intrinsic TAMRA fluorescence (dF^sub 490/580^) whereas excitation at 555 nm results in only TAMRA emission. However, dF^sub 490/580^ is not a unique measure of E during the course of a titration since it includes contributions from both FRET and the increasing concentration of TAMRA. Titration of fluorescein-labeled single-strand oligonucleotide with the TAMRA-labeled complement is performed to extract the FRET component from dF^sub 490/550^ (Fig. 3).

In Fig. 3 A, line 1 depicts the TAMRA fluorescence change upon excitation of 490 nm (zone II in Scheme 1) reflecting changes due to both FRET and the direct excitation of TAMRA within the duplex. Line 2 depicts the TAMRA fluorescence change due to increasing concentration of the free acceptor strand (zone IV in Scheme 1). The stoichiomelric T14/F14 duplex (zone III in Scheme 1) occurs at the intersection of lines 1 and 2. Fig. 3 B depicts the TAMRA fluorescence change due as duplex (line 3) or mixture of duplex with single-strand (line 4) upon direct excitation at 555 nm, conditions under which FRET docs not contribute to the observed fluorescence. As above, the intersection of these two lines indicates the formation of the stoichiometric T14/F14 duplex.

Determination of the relative portion of TAMRA fluorescence due to FRET

Measurement of FRET in TBP-DNA binding titrations

Titrations such as shown in Fig. 3 were conducted for each set of individually single-labeled and the double-labeled duplex stock solutions prepared tor protein binding studies. The concentration at which the labeled complementary oligonucleotides was equimolar, the value of R^sub ET^ for protein-free DNA, and the ratio S^sub 2^/S^sub 4^ required for the determination of R^sub ET^ for TBP-bound DNA were determined from that titrations. A ratio of 0.95 of TAMRA to fluorescein-labeled single-strand oligonucleotide was used to anneal double-labeled duplexes in TBP-binding experiments to insure that excess TAMRA-labeled single-strand DNA was not present. All of the data required for the calculation of R^sub ET^ is obtained for each concentration of TBP at pairs of excitation/emission wavelengths of 490/520, 490/580, and 555/580 nm. R^sub ET^ was calculated (Eq. 5) from the change in TAMRA fluorescence accompanying TBP binding as a function of TBP concentration and is the parameter used to measure TBP binding. (The details of this procedure are presented in Online Supplement 3.)

TBP-DNA titrations

S. cerevisiue TBP is monomeric under the experimental conditions of this study (Daugherty et al., 1999, 2000) and a single monomer binds to a TATA Box sequence (Khrapunov et al., 2002; Parkhurst et al., 1996, 1999; Petri et al., 1995, 1998). For the equilibrium P^sub f^ + O^sub f^ [Lef-right arrow] PO, P^sub f^ and O^sub f^ are the concentrations of free TBP and DNA, respectively, and PO is the concentration of the TBP-DNA complex. The concentration of PO is K^sub a^ × P^sub f^ × O^sub f^ = K^sub a^(P^sub tot^ - PO) × (O^sub tot^ - PO), where P^sub tot^ and O^sub tot^ are the total concentrations of protein and DNA, respectively, and K^sub a^ is the equilibrium association constant. Since the concentration of the fluorescently-labeled oligonucleolide probe is sufficiently high relative to the binding affinity of the complex, the simplifying assumption [ligand]^sub total^ [asymptotically =] [ligand]^sub free^. cannot be used in the analysis of the titrations of the present study.

RESULTS

The structures of TBP complexes with TATAAAAG and TITIIIIG

R^sub ET^ for TAMRA fluorescence was separately determined using Eq. 5 for the adenosine- and inosine-containing probes from a series of titrations of T14 against F14 (Fig. 3, for example). Upon titration of the two double-labeled duplexes by TBP, the emission spectrum upon excitation at 490 nm changes due to FRET (Fig. 4, insert; Parkhurst et al., 1996). In contrast, neither the fluorescein fluorescence at 520 nm nor the TAMRA fluorescence at 580 nm changes upon titration of the single-labeled duplexes with TBP (Fig. 1).

Titrations of the double-labeled A-AdMLP and I-AdMLP probes with TBP yield R^sub ET^ as a function of TBP concentration for each sequence (Fig. 4; Table 1). Differences between the two free probes are observed (Table 1). In addition, the maximum value of R^sub ET^ measured upon the addition of saturating concentrations of TBP (Fig. 4, dashed lines) are different for the adenosine- and inosine-containing probes. As discussed below, we conclude these differences are due to differences in the critical distance, R^sub o^, not the end-to-end distances of the two probes.

Many factors influence R^sub o^ for a given pair of probes (Eq. 3). We note that upon duplex formation the fluorescein fluorescence increases for A-AdMLP and decreases for I-AdMLP (Fig. 1) and that there are differences in the spectral overlap integral (Table 1, column [epsilon]^sub T^, 555/[epsilon]^sub F^, 490). However, R^sub o^ is constant for a given oligonucleotide as a function of TBP binding (Wu et al., 2001a). Therefore the relative protein-induced change of end-to-end distance can be calculated for both oligonucleotides from Eq. 2 using E^sub f^ and E^sub b^ (Table 1). The relationships between end-to-end distances of DNA free (R^sub f^) and bound to TBP (R^sub b^), R^sub f^ = 1.17 . R^sub b^ and R^sub f^ = 1.16 . R^sub b^ for A-AdMLP and I-AdMLP, respectively, are identical within experimental error and in excellent agreement with the results obtained for an A-AdMLP oligonucleotide in which the top strand was double-labeled and annealed with the unlabeled compliment (Parkhurst et al., 1996; Wu et al., 2001a).

A change in either R^sub o^ or in R results in a change in the value of E (Eq. 2). Consider that an A-tract, such as that present in the A-AdMLP sequence, induces curvature to the helical axis of DNA (Crothers and Drak, 1992; Davis et al., 1999). An overall helix axis bend of 19° (calculated as the radius of curvature of the DNA) was measured in an NMR solution study of a dodecamer containing a 6-bp A-tract (MacDonald et al., 2001). For a 14-bp oligonucleotide, this angle of curvature yields the length ratio between bent and linear DNA, L^sub b^/L^sub l^ of 0.996, assuming a simple isotropic bending rod model of DNA (Wildeson and Murphy, 2000). A 0.4% change in end-to-end length is not detectable by our assay. Since the curvature of I-AdMLP is greater than that of random-sequence DNA but less that of an A-tract of comparable length (Koo and Crothers, 1987; Shatzky-Schwartz et al., 1997), the difference between the end-to-end distances of the A-AdMLP and I-AdMLP probes will be undetectable in our measurements. Thus, differences in the critical distance, R^sub o^, must be responsible for the difference in the measured value of E between the I-AdMLP and A-AdMLP oligonucleotide duplexes (Fig. 4, Table 1).

Comparison of the electrostatic, enthalpic, and entropic components of TBP binding to A-AdMLP and I-AdMLP

To ascertain whether electrostatics differentially contribute to TBP binding the adenosine- and inosine-substituted TATA Box sequences, a series of TBP-binding isotherms were determined as a function of KCl concentration (Fig. 5). This linkage analysis provides an estimate of the number of the counterions released upon TBP binding (Record, Jr. et al., 1998). The slopes of the data shown Fig. 5 for A-AdMLP and I-AdMLP are identical within experimental error, 1.64 ± 0.27 and 1.65 ± 0.35, respectively. These values are less than that previously determined for the A-AdMLP within a long DNA restriction fragment (3.5; Petri et al., 1998). This difference is unsurprising due to documented electrostatic end-effects upon binding to oligonucleotides (Zhang et al., 1999) and the potential for protein-DNA contacts with the DNA surrounding a specific binding sequence (Holbrook et al., 2001).

Partitioning of the enthalpic and entropic contributions to the [Delta]G° of binding was accomplished using the van't Hoff equation d ln K/d ln(1/T) = [Delta]H°/R for TBP-binding isotherms obtained as a function of temperature at 100 mM KCl (Fig. 6). The observed standard enthalpy changes, [Delta]H°, determined for the A-AdMLP and I-AdMLP sequences of 19.3 ± 2.9 and 18.5 ± 1.8 kcal M^sup -1^, respectively, are identical within experimental error. The linearity of these van't Hoff analysis over this limited range of temperatures is consistent with previous studies conducted using DNA restriction fragments (Petri et al., 1995, 1998) or oligonucleotides (Parkhurst et al., 1999).

DISCUSSION

Utilization of the parameter R^sub ET^

Two FRET approaches are presently used to study the structure of nucleic acids. A variety of approaches have been developed and published for steady-state FRET (see Cantor and Schimmel, 1980; Clegg, 1992; Lakowicz, 1999, for general discussions). However, uncertainties remain with this approach with regard to issues such as the variability induced by particular experimental conditions and the differences in probe concentration among samples. Steady-state methodology relates the measured average donor-acceptor distance to a change in the structure, curvature, and bending of DNA consistent with the helical geometry of the DNA molecule (Clegg, 1992; Gelfand et al., 1999; Stuhmeier et al., 2000; Toth et al., 1998; Wildeson and Murphy, 2000). Time-resolved methodology exploits the distribution of donor-acceptor distances and has been used to study changes in the DNA structure depending on sequence, extent of hybridization, and protein binding (Klostermeier and Millar, 2001; Parkhurst et al., 1996; 2001-2002; Wu et al., 2001a). The mean donor-acceptor distance and the dispersion of the distribution are extracted from the distribution of donor-acceptor distances accessible to time-resolved studies. The time-resolved method is particularly informative due to its physical connection to DNA dynamics in solution.

An issue general to the quantitative interpretation of both approaches is their dependence upon the critical distance, R^sub o^, which in turn depends on the orientation factor, [kappa], of a donor-acceptor pair and J, the spectral overlap integral (Eq. 3). Variation in the published values of R^sub o^ for the fluorescein-TAMRA donor-acceptor pair reflect the different contributions of these physical parameters (Parkhurst et al., 1996; Stuhmeier et al., 2000, 1997; Tom et al., 1998; Wildeson and Murphy, 2000; Wu et al., 2001a; Yang and Millar, 1996). Although the orientation of fluorescein is generally observed to be random, its behavior depends on the chemistry of linker connecting it to the DNA (Hill and Royer, 1997).

Substitution of inosine for adenine in the interior of an oligonucleotide resulted in opposite changes in the fluorescence of fluorescein attached to the DNA upon hybridization despite the identity of the terminal bases (Fig. 1; see Materials and Methods). The behavior of DNA-linked TAMRA is more problematic, including the observations of high anisotropy (Fig. 6; see also Clegg, 1992; Stuhmeier et al., 2000, 1997) and quenching by adjacent guanines (Edman et al., 1996; Eggeling et al., 1998; Stuhmeier et al., 2000, 1997; Wennmalm et al., 1997). The fluorescence quantum yield of both fluorescein and TAMRA depends on the 5'-basepair to which they are linked (Hill and Royer, 1997; Yang and Millar, 1996) and whether the dye is attached to 3' or 5' end of the DNA (unpublished observations). Single molecule studies also reveal influences of DNA sequence on TAMRA fluorescence (Edman et al., 1996; Eggeling et al., 1998; Wennmalm et al., 1997). Thus, only upon careful consideration of R0 (e.g., Klostermeier and Millar, 2001; Parkhurst et al., 2001-2002; Stuhmeier et al., 2000) can reliable structural parameters be extracted from FRET measurements (Stuhmeier et al., 2000, 1997; Wu et al., 2001a; Yang and Millar, 1996).

However, the determination of relative changes in DNA structure requires only the accurate measurement of relative changes in the efficiency, E, of energy migration. The parameter R^sub ET^ (Eq. 5) is simply E multiplied by a proportional coefficient, F^sub abs^, that is specific for a donor-acceptor pair (Eq. 6). The use of the R^sub ET^ for monitoring equilibrium and structural changes of macromolecules has several advantages. 1), It is a titration approach that directly determines the stoichiometric ratio of the labeled oligonucleotides that comprise the duplex. Since RET is insensitive to excess donor-labeled oligonucleotide, measurements can be made at acceptor-donor ratios of unity or less (0.85-0.95), avoiding an excess of the acceptor-labeled oligonucleotide in the reaction mixture. 2), Absorption measurements, which may be a source of error due to uncertainty in the extinction coefficient of the tethered dyes, are not required. 3), The fluorescence intensities at different combinations of wavelengths required for determination of R^sub ET^ are acquired in the same experiment for one particular sample. 4), Since intensity ratios rather than absolute values are utilized, R^sub ET^ is insensitive to fluctuations in experimental conditions or probe concentrations. 5), Probe purification after annealing of the single strands and measurement of the concentration of double-labeled duplex is not required. 6), R^sub ET^ facilitates measurement of relative changes in the conformation of DNA when R^sub 0^ is unchanged.

The relative structures of TBP-TATAAAAG and TBP-TITIIIIG complexes

The sequence-specific binding of TBP to DNA is considered a defining example of indirect readout in which macromolecular structure and conformational accommodation are key elements of binding. The TBP-DNA interaction displays unusual functional as well as structural characteristics. Its high-equilibrium binding affinity masks unusually slow rates of DNA association and dissociation (Hoopes et al., 1992; Parkhurst et al., 1996; Perez-Howard et al., 1995; Petri et al., 1995) and a complex two-intermediate reaction pathway (Parkhurst et al., 1999; Powell et al., 2001).

DNA deformability is believed to play an important role in protein-DNA interactions (Sivolob and Khrapunov, 1995). The binding and stability of TBP with the extreme bend introduced into the bound DNA by TBP plays an important, if not dominant, role in its ability to facilitate transcription initiation. TBP binds with higher affinity to DNA that is prebent toward the major groove (Parvin et al., 1995), presumably favoring TBP's minor groove interaction. Cyclization kinetics studies of TBP-DNA complexes suggests that the AdMLP TATA box DNA is bent in a direction opposite to the bend induced by TBP and is anisotropically flexible (Davis et al., 1999), although the latter conclusion has been questioned (Pastor and Weinstein, 2000). It is claimed that overall DNA flexibility correlates with transcriptional activity (Qian et al., 2001), although the extent of this correlation seems minimal. In contrast, a strong correlation was reported between the solution bend angles of DNA within the TBP-DNA complex and relative transcription activity (Wu et al., 2001a). However, further study has revealed a complex dependence of TBP-induced bend angle upon the TATA box sequence and osmolyte concentration (Wu et al., 2001b). Thus, comparisons of transcriptional activity with TATA box sequences must consider this myriad array of factors.

The feature of the TIT-for-TAT (Pastor et al., 1999) substitution that stimulated the present experimental analysis is that substitution of inosine for adenine changes the chemical structure of the major but not the minor groove (Saenger, 1984). Long before atomic resolution structures of TBP were solved, substitution of inosine for adenosine was used to establish that TBP recognized and bound the minor groove of DNA; these studies concluded that the interaction of TBP with the sequences CICJIII and TATAAAA were indistinguishable (Starr and Hawley, 1991). Efficient preinitiation complex formation and transcription initiation is supported by inosinc:cytosine basepair-substituted TATA box sequences (Lee et al., 1997).

This study demonstrates that the end-to-end distance (R^sub bound^/R^sub free^) of the A-AdMLP and I-AdMLP DNA complexed to TBP is indistinguishable despite the higher flexibility and lower affinity of the I-tract compared with A-tract (Table 1). The sensitivity of R^sub bound^/R^sub free^ to end-to-end distance of DNA changes is shown in studies of subtle and not-so-subtle DNA bends including those modeled as smoothly bent (MacDonald et al., 2001; Wildeson and Murphy, 2000), single-kinked (Hardwidge et al., 2002), and double-kinked (Wu et al, 2001a). The equivalence of the TBP-induced DNA bend inferred from the I-AdMLP and A-AdMLP data are consistent with the observed correlation between solution bend angles and transcription activity of the TBP-DNA complex (Wu et al., 2001a).Thus, deformability of DNA can be considered as an important (Parkhurst et al., 2001-2002) but not the sole factor leading to larger TBP-induced DNA bend angles. Both structural and thermodynamic features of TBP-DNA complexes modulate the affinity of the protein and DNA for each other (Table 1).

A role for water release in the stability of the TBP-DNA complex

The detailed thermodynamic analysis (Figs. 5 and 6) shows that neither the enthalpic nor electrostatic components of binding contribute noticeably to the differential affinity of TBP for a native TATA sequence compared with an inosine-substituted sequence. The osmotic-stress and preferential-interactions approaches to describing the effect of neutral solutes on biochemical equilibria are extensively discussed in the literature (Courtenay et al., 2000; Parsegian et al., 2000; Tang, 2002; Timasheff, 1998). The central point of the discussion is the unknown distribution of cosolute at macromolecule surfaces as well as in the bulk and local domains of the solution. Record and co-workers have proposed measurement of "iso-osmolal preferential interaction coefficients" for each solute (Courtenay et al., 2000; Record, Jr., et al., 1998). Although this approach is attractive, because of its clear physicochemical basis it is difficult to implement. From a practical point of view, the difference between osmotic stress and preferential interactions methodologies reduces to substitution of Nw in Eq. 12 with N^sub w^ * (K^sub p^ - 1) where K^sub p^ is the local-bulk partition coefficient for a non-electrolyte solute (Courtenay et al., 2000; Parsegian et al., 1995, 2000; Tang, 2002; Timasheff, 1998). Both approaches are equivalent when K^sub p^ is 0. If K^sub p^ is not 0, data such as was shown in Fig. 11 report only the lower limit of N^sub w^.

By conducting our experiments identically except for the oligonucleotide substitution of inosine for adenosine, we have circumvented this ambiguity since the quantity of interest is the difference in water release between the two binding reactions. Rau and co-workers developed a similar differential method by using a competitive assay to compare DNA sequence-specific and nonspecific protein binding. The amount of water sequestered by two nonspecific protein-DNA complexes-the synthetic polymer poly (dI-dC).poly (dI-dC) and the oligonucleotide differing from the specific recognition site by only two basepairs-is closely similar (Parsegian et al., 2000; Sidorova and Rau, 1996, 2001). As noted above, the chemical constituents in the minor grooves of the A-AdMLP and I-AdMLP sequences are identical, as are the changes in their end-to-end distances induced by TBP binding. Since the protein is the same in both sets of experiments, the water-accessible surfaces of the separate reactants and the TBP-DNA complex are presumably the same for the A-AdMLP and I-AdMLP TBP-binding reactions.

The minor groove of the DNA within the TBP-DNA complex is completely dehydrated in the co-crystal complexes (Kim et al., 1993a,b). Molecular dynamics simulations have shown that both enthalpic and entropic forces contribute to the stabilization of the TBP-DNA complex (Pastor et al., 2000). The enthalpic penalty associated with the dehydration of the DNA minor groove accompanying TBP binding is compensated for by the increased entropy resulting from water release. Thus, the entropic advantage is predicted to be higher for A-AdMLP relative to I-AdMLP based upon the excess of 19 water molecules bound within the minor groove of A-AdMLP that are displaced upon TBP binding (Pastor et al., 1999). This prediction is consistent with the greater affinity of TBP for A-AdMLP than for I-AdMLP (Fig. 10, Table 1).

Approximately 18 water molecules are released upon binding TBP to the A-AdMLP as compared with slight binding (in the range of experimental error for the associate constant determination) of water to the I-AdMLP sequence (Fig. 11). That difference is in a good agreement with the theoretically predicted excess of 19 water molecules bound in the minor groove of A-AdMLP (Pastor et al., 1999). The difference in amount of water released that accompanies TBP binding to two promoters can be largely attributed to the steric exclusion of water from the interacting surface of the TBP-DNA complex. An excess of water in the minor groove of the TATA box is thermodynamically expensive (Dunitz, 1994) and its release upon TBP binding makes the TATA box energetically preferential. Thus, it appears that despite the structural advantage of the TITIIIIG sequence with regard to its greater flexibility, the binding of TBP to the natural promoter binding is favored by entropic advantage due to the release of water upon complex formation.

SUPPLEMENTARY MATERIAL

Online supplements to this article can be found by visiting BJ Online at http://www.biophysj.org.

We thank Elizabeth Jamison for assistance in conducting some of the binding measurements and for the preparation of the TBP used in these studies. We also thank Donald Rau and Nina Sidorova for sharing information on the physical-chemical features of osmolytes, helpful discussions, and Kay Parkhurst and Larry Parkhurst for critically reading the manuscript.

This work was supported by grant GM39929 from the National Institutes of Health.

REFERENCES

Bareket-Samish, A., I. Cohen, and T. E. Haran. 2000. Signals for TBP/ TATA box recognition. J. Mol. Biol. 299:965-977.

Burley, S. K., and R. G. Roeder. 1996. Biochemistry and structural biology of transcription factor IID (TFIID). Anna. Rev. Biochem. 65:769-799.

Cantor, C. R., and P. R. Schimmel. 1980. Biophysical Chemistry. W. H. Freeman and Company, San Francisco, CA.

Clegg, R. M. 1992. Fluorescence resonance energy transfer and nucleic acids. Meth. Enzymol. 211:353-388.

Courtenay, E. S., M. W. Capp, C. F. Anderson, and M. T. Record, Jr. 2000. Vapor pressure osmometry studies of osmolyte-protein interactions: implications for the action of osmoprotectants in vivo and for the interpretation of "osmotic stress" experiments in vitro. Biochemistry. 39:4455-4471.

Crothers, D. M., and J. Drak. 1992. Global features of DNA structure by comparative gel electrophoresis. Meth. Enzymol. 212:46-71.

Daugherty, M. A., M. Brenowitz, and M. G. Fried. 1999. The TATA-binding protein from Saccharomyces cerevisiae oligomerizes in solution at micromolar concentrations to form tetramers and octamers. J. Mol. Biol. 285:1389-1399.

Daugherly, M. A., M. Brenowitz, and M. G. Fried. 2000. Participation of the ammo-terminal domain in the self-association of the full-length yeast TATA binding protein. Biochemistry. 39:4869-4880.

Davis, N. A., S. S. Majee, and J. D. Kahn. 1999. TATA box DNA deformation with and without the TATA box-binding protein. J. Mol. Biol. 291:249-265.

Dunitz, J. D. 1994. The entropic cost of bound water in crystals and biomolecules. Science. 264:670.

Edman, L., U. Mets, and R. Rigler. 1996. Conformational transitions monitored for single molecules in solution. Proc. Natl. Acad. Sci. USA. 93:6710-6715.

Eggeling, C., J. R. Fries, L. Brand, R. Gunther, and C. A. Seidel. 1998. Monitoring Conformational dynamics of a single molecule by selective fluorescence spectroscopy. Proc. Natl. Acad. Sci. USA. 95:1556-1561.

Fried, M. G., D. F. Stickle, K. V. Smirnakis, C. Adams, D. MacDonald, and P. Lu. 2002. Role of hydration in the binding of LAC reprcssor to DNA. J. Biol. Chem. 277:50676-50682.

Gelfand, C. A., G. E. Plum, S. Mielewczyk, D. P. Remeta, and K. J. Breslauer. 1999. A quantitative method for evaluating the stabilities of nucleic acids. Proc. Natl. Acad. Sci. USA. 96:6113-6118.

Hardwidge, P. R., J. Wu, S. L. Williams, K. M. Parkhurst, L. J. Parkhurst, and L. J. Maher, III. 2002. DNA bending by bZIP charge variants: a unified study using electrophoretic phasing and fluorescence resonance energy transfer. Biochemistry. 41:7732-7742.

Hill, A. V. 1910. The possible effects of the aggregation of the molecules of hemoglobin on the dissociation curves. J. Physiol. (Lond.). 40:iv-vii.

Hill, J. J., and C. A. Royer. 1997. Fluorescence approaches to study of protein-nucleic acid complexation. Meth. Enzymol. 278:390-416.

Holbrook, J. A., O. V. Tsodikov, R. M. Saecker, and M. T. Record, Jr. 2001. Specific and non-specific interactions of integration host factor with DNA: thermodynamic evidence for disruption of multiple IHF surface salt-bridges coupled to DNA binding. J. Mol. Biol. 310:379-401.

Hoopes, B. C., J. F. LeBlanc, and D. K. Hawley. 1992. Kinetic analysis of yeast TFIID-TATA box complex formation suggests a multi-step pathway. J. Biol. Chem. 267:11539-11547.

Jen-Jacobson, L., L. E. Engler, and L. A. Jacobson. 2000. Structural and thermodynamic strategies for site-specific DNA binding proteins. Struct. Fold. Des. 8:1015-1023.

Khrapunov, S., N. Pastor, and M. Brenowitz. 2002. Solution structural studies of the Saccharomyces cerevisiae TATA binding protein (TBP). Biochemistry. 41:9559-9571.

Kim, J. L., D. B. Nikilov, and S. K. Burley. 1993a. Co-crystal structure of TBP recognizing the minor groove of a TATA element. Nature. 365:520-527.

Kim, Y. C., J. H. Geiger, S. Hahn, and P. B. Sigler. 1993b. Crystal structure of a yeast TBP/TATA-box complex. Nature. 365:512-520.

Klostermeier, D., and D. P. Millar. 2001. Time-resolved fluorescence resonance energy transfer: a versatile tool for the analysis of nucleic acids. Biopolymers. 61:159-179.

Koo, H. S., and D. M. Crothers. 1987. Chemical determinants of DNA bending at adenine-thymine tracts. Biochemistry. 26:3745-3748.

Lakowicz, J. R. 1999. Principles of Fluorescence Spectroscopy. Kluwer Academic/Plenum Press, New York.

Lankas, F., T. E. Cheatham III, N. Spackova, P. Hobza, J. Langowski, and J. Sponer. 2002. Critical effect of the N2 amino group on structure, dynamics, and elasticity of DNA polypurine tracts. Biophys. J. 82:2592-2609.

Lee, D. K., M. Horikoshi, and R. G. Roeder. 1991. Interaction of TFIID in the minor groove of the TATA element. Cell. 67:1241-1250.

Lee, D. K., K. C. Wang, and R. G. Roeder. 1997. Functional significance of the TATA element major groove in transcription initiation by RNA polymerase II. Nucleic Acids Res. 25:4338-4345.

MacDonald, D., K. Herbert, X. Zhang, T. Pologruto, P. Lu, and T. Polgruto. 2001. Solution structure of an A-tract DNA bend. J. Mol. Biol. 306:1081-1098.

Morrison, L. E., and L. M. Stols. 1993. Sensitive fluorescence-based thermodynamic and kinetic measurements of DNA hybridization in solution. Biochemistry. 32:3095-3104.

Pabo, C. O., and R. T. Sauer. 1992. Transcription factors: structural families and principles of DNA recognition. Annu. Rev. Biochem. 61:1053-1095.

Parkhurst, K. M., M. Brenowitz, and L. J. Parkhurst. 1996. Simultaneous binding and bending of promoter DNA by the TATA binding protein: real lime kinetic measurements. Biochemistry. 35:7459-7465.

Parkhurst, K. M., R. M. Richards, M. Brenowitz, and L. J. Parkhurst. 1999. Intermediate species possessing bent DNA are present along the pathway to formation of a final TBP-TATA complex. J. Mol. Biol. 289:1327-1341.

Parkhurst, L. J., K. M. Parkhurst, R. Powell, J. Wu, and S. Williams. 2001-2002. Time-resolved fluorescence resonance energy transfer studies of DNA bending in double-stranded oligonucleotides and in DNA-protein complexes. Biopolymers. 61:180-200.

Parsegian, V. A., R. P. Rand, and D. C. Rau. 1995. Macromolecules and water: probing with osmotic stress. Meth. Enzymol. 259:43-94.

Parsegian, V. A., R. P. Rand, and D. C. Rau. 2000. Osmotic stress, crowding, preferential hydration, and binding: a comparison of perspectives. Proc. Natl. Acad. Sci. USA. 97:3987-3992.

Parvin, J. D., R. J. McCormick, P. A. Sharp, and D. A. Fisher. 1995. Pre-bending of a promoter sequence enhances affinity for the TATA-binding factor. Nature. 373:724-727.

Pastor, N., A. D. MacKerell, Jr., and H. Weinstein. 1999. TIT for TAT: the properties of inosine and adenosine in TATA box DNA. J. Biomol. Struct. Dyn. 16:787-810.

Pastor, N., and H. Weinstein. 2000. Theoretical Biochemistry: Processes and Properties of Biological Systems. L. A. Eriksson, editor. Elsevier Science, New York.

Pastor, N., H. Weinstein, E. Jamison, and M. Brenowitz. 2000. A detailed interpretation of OH radical footprints in a TBP-DNA complex reveals the role of dynamics in the mechanism of sequence-specific binding. J. Mol. Biol. 304:55-68.

Perez-Howard, G. M., P. A. Weil, and J. M. Beechem. 1995. Yeast TATA binding protein interaction with DNA: fluorescence determination of oligomeric state, equilibrium binding, on-rate, and dissociation kinetics. Biochemistry. 34:8005-8017.

Petri, V., M. Hsieh, and M. Brenowitz. 1995. Thermodynamic and kinetic characterization of the binding of the TATA binding protein to the adenovirus E4 promoter. Biochemistry. 34:9977-9984.

Petri, V., M. Hsieh, E. Jamison, and M. Brenowitz. 1998. DNA sequence-specific recognition by the Saccharomyces cerevisiae "TATA" binding protein: promoter-dependent differences in the thermodynamics and kinetics of binding. Biochemistry. 37:15842-15849.

Powell, R. M., K. M. Parkhurst, M. Brenowitz, and L. J. Parkhurst. 2001. Marked stepwise differences within a common kinetic mechanism characterize TATA-binding protein interactions with two consensus promoters. J. Biol. Chem. 276:29782-29791.

Powell, R. M., K. M. Parkhurst, and L. J. Parkhurst. 2002. Comparison of TATA-binding protein recognition of a variant and consensus DNA promoters. J. BM. Chem. 277:7776-7784.

Qian, X., D. Strahs, and T. Schlick. 2001. Dynamic simulations of 13 TATA variants refine kinetic hypotheses of sequence/activity relationships. J. Mol. Biol. 308:681-703.

Record, M. T., Jr., C. F. Anderson, P. Mills, M. Mossing, and J. H. Roe. 1985. Ions as regulators of protein-nucleic acid interactions in vitro and in vivo. Adv. Biophys. 20:109-135.

Record, M. T., Jr., W. Zhang, and C. F. Anderson. 1998. Analysis of effects of salts and uncharged solutes on protein and nucleic acid equilibria and processes: a practical guide to recognizing and interpreting polyelectrolyte effects, Hofmeister effects, and osmotic effects of salts. Adv. Prot. Chem. 51:281-353.

Saenger, W. 1984. Principles of Nucleic Acid Structure. C. R. Cantor, editor. Springer-Verlag, New York, Berlin, Heidelberg, Tokyo.

SantaLucia, J., Jr. 1998. A unified view of polymer, dumbbell, and oligonucleotide DNA nearest-neighbor thermodynamics. Proc. Natl. Acael. Sci. USA. 95:1460-1465.

Shatzky-Schwartz, M., N. D. Arbuckle, M. Eisenstein, D. Rabinovich, A. Bareket-Samish, T. E. Haran, B. F. Luisi, and Z. Shakked. 1997. X-ray and solution studies of DNA oligomers and implications for the structural basis of A-tract-dependent curvature. J. Mol. Biol. 267:595-623.

Sidorova, N. Y., and D. C. Rau. 1996. Differences in water release for the binding of EcoRI to specific and nonspecific DNA sequences. Proc. Natl. Acad. Sci. USA. 93:12272-12277.

Sidorova, N. Y., and D. C. Rau. 2001. Linkage of EcoRI dissociation from its specific DNA recognition site to water activity, salt concentration, and pH: separating their roles in specific and non-specific binding. J. Mol. Biol. 310:801-816.

Sivolob, A. V., and S. N. Khrapunov. 1995. Translational positioning of nucleosomes on DNA: the role of sequence-dependent isotropic DNA bending stiffness. J. Mol. Biol. 247:918-931.

Starr, D. B., and D. K. Hawley. 1991. TFIID binds in the minor groove of the TATA box. Cell. 67:1231-1240.

Starr, D. B., B. C. Hoopes, and D. K. Hawley. 1995. DNA bending is an important component of site-specific recognition by the TATA binding protein. J. Mol. Biol. 250:434-446.

Stuhmeier, F., A. Hillisch, R. M. Clegg, and S. Diekman. 2000. Fluorescence energy transfer analysis of DNA structures containing several bulges and their interaction with CAP. J. Mol. Biol. 302:1081-1100.

Stuhmeier, F., J. B. Welch, A. I. Murchie, D. M. Lilley, and R. M. Clegg. 1997. Global structure of three-way DNA junctions with and without additional unpaired bases: a fluorescence resonance energy transfer analysis. Biochemistry. 36:13530-13538.

Tang, K. E. S., and V. A. Bloomfield. 2002. Assessing accumulated solvent near a macromolecule solute by preferential interaction coefficients. Biophys. J. 82:2876-2891.

Timasheff, S. N. 1998. In disperse solution, "osmotic stress" is a restricted case of preferential interactions. Proc. Natl. Acad. Sci. USA. 95:7363-7367.

Toth, K., V. Sauermann, and J. Langowski. 1998. DNA curvature in solution measured by fluorescence resonance energy transfer. Biochemistry. 37:8173-8179.

Vossen, K. M., R. Wolz, M. A. Daugherty, and M. G. Fried. 1997. Role of macromolecular hydration in the binding of the Escherichia coli cyclic AMP receptor to DNA. Biochemistry. 36:11640-11647.

Wennmalm, S., L. Edman, and R. Rigler. 1997. Conformational fluctuations in single DNA molecules. Proc. Natl. Acad. Sci. USA. 94: 10641-10646.

Wildeson, J., and C. J. Murphy. 2000. Intrinsic bending in GGCC tracts as probed by fluorescence resonance energy transfer. Anal. Biochem. 284:99-106.

Wu, J., K. M. Parkhurst, R. M. Powell, M. Brenowitz, and L. J. Parkhurst. 2001a. DNA bends in TATA-binding protein-TATA complexes in solution are DNA sequence-dependent. J. Biol. Chem. 276:14614-14622.

Wu, J., K. M. Parkhurst, R. M. Powell, and L. J. Parkhurst. 2001b. DNA sequence-dependent differences in TATA-binding protein-induced DNA bending in solution are highly sensitive to osmolytes. J. Biol. Chem. 276:14623-14627.

Yang, M., and D. P. Millar. 1996. Conformational flexibility of three-way DNA junctions containing unpaired nucleotides. Biochemistry. 35:7959-7967.

Yang, M., and D. P. Millar. 1997. Fluorescence resonance energy transfer as a probe of DNA structure and function. Methods Enzymol. 278:417-444.

Zhang, W., H. Ni, M. W. Capp, C. F. Anderson, T. M. Lohman, and M. T. Record, Jr. 1999. The importance of Coulombic end effects: experimental characterization of the effects of oligonucleotide flanking charges on the strength and salt dependence of oligocation (L^sup 8+^) binding to single-stranded DNA oligomers. Biophys. J. 76:1008-1017.

Sergei Khrapunov and Michael Brenowitz

Department of Biochemistry, Albert Einstein College of Medicine, Bronx, New York

Submitted April 11, 2003, und accepted for publication September 10, 2003.

Address reprint requests to Sergei Khrapunov, Tel.: 718-430-3180; E-mail: khraps@aecom.yu.edu; or Michael Brenowitz, Tel.: 718-430-3179, Fax: 718-430-8565; E-mail: brenowit@aecom.yu.edu.

© 2004 by the Biophysical Society

0006-3495/04/01/371/13 $2.00

Copyright Biophysical Society Jan 2004
Provided by ProQuest Information and Learning Company. All rights Reserved