Selective binding of the TATA box-binding protein to the TATA box-containing promoter: analysis of structural and energetic factors.

We report the results of an energy-based exploration of the components of selective recognition of the TATA box-binding protein (TBP) to a TATA box sequence that includes 1) the interaction between the hydrophobic Leu, Pro, and Phe residues of TBP with the TA, AT, AA, TT, and CG steps, by ab initio quantum mechanical calculations; and 2) the free energy penalty, calculated from molecular dynamics/potential of mean force simulations, for the conformational transition from A-DNA and B-DNA into the TA-DNA form of DNA observed in a complex with TBP. The GTAT, GATT, GAAT, and GTTT tetramers were explored. The results show that 1) the discrimination of TA, AT, AA, TT, or CG steps by TBP cannot rest on their interaction with the inserting Phe side chains; 2) the steric clash between the bulky and hydrophobic Pro and Leu residues and the protruding -NH2 group of guanine is responsible for the observed selectivity against any Gua-containing basepair; 3) the Pro and Leu residues cannot selectively discriminate among TA, AT, AA, or TT steps; and 4) the calculated energy required to achieve the TA-DNA conformation of DNA that is observed in the complex with TBP appears to be a key determinant for the observed selectivity against the AT, AA, and TT steps. The simulations also indicate that only the TA step can form a very efficient interbase hydrogen bond network in the TA-DNA conformation. Such an energetically stabilizing network is not achievable in the AA and TT steps. While it is viable in the AT step, structural constraints render the hydrogen bonding network energetically ineffective there.     

Binding mechanisms of TATA box-binding proteins: DNA kinking is stabilized by specific hydrogen bonds

One of the common mechanisms of DNA bending by minor groove-binding proteins is the insertion of protein side chains between basepair steps, exemplified in TBP (TATA box-binding protein)/DNA complexes. At the central basepair step of the TATA box TBP produces a noticeable decrease in twist and an increase in roll, while engaging in hydrogen bonds with the bases and sugars. This suggests a mechanism for the stabilization of DNA kinks that was explored here with ab initio quantum mechanical calculations and molecular dynamics/potential of mean force calculations. The hydrogen bonds are found to contribute the energy necessary to drive the conformational transition at the central basepair step. The Asn, Thr, and Gly residues involved in hydrogen bonding to the DNA bases and sugar oxygens form a relatively rigid motif in TBP. The interaction of this motif with DNA is found to be responsible for inducing the untwisting and rolling of the central basepair step. Notably, direct readout is shown not to be capable of discriminating between AA and AT steps, as the strength of the hydrogen bonds between TBP and the DNA are the same for both sequences. Rather, the calculated free energy cost for an equivalent conformational transition is found to be sequence-dependent, and is calculated to be higher for AA steps than for AT steps.     

Signals for TBP/TATA box recognition

The TATA box-binding protein (TBP) recognizes its target sites (TATA boxes) by indirectly reading the DNA sequence through its conformation effects (indirect readout). Here, we explore the molecular mechanisms underlying indirect readout of TATA boxes by TBP by studying the binding of TBP to adenovirus major late promoter (AdMLP) sequence variants, including alterations inside as well as in the sequences flanking the TATA box. We measure here the dissociation kinetics of complexes of TBP with AdMLP targets and, by phase-sensitive assay, the intrinsic bending in the TATA box sequences as well as the bending of the same sequence induced by TBP binding. In these experiments we observe a correlation of the kinetic stability to sequence changes within the TATA recognition elements. Comparison of the kinetic data with structural properties of TATA boxes in known crystalline TBP/TATA box complexes reveals several "signals" for TATA box recognition, which are both on the single base-pair level, as well as larger DNA tracts within the TATA recognition element. The DNA bending induced by TBP on its binding sites is not correlated to the stability of TBP/TATA box complexes. Moreover, we observe a significant influence on the kinetic stability of alteration in the region flanking the TATA box. This effect is limited however to target sites with alternating TA sequences, whereas the AdMLP target, containing an A tract, is not influenced by these changes.     

How Proteins Recognize the TATA Box

The crystal structure of a complex of human TATA-binding protein with TATA-sequence DNA has been solved, complementing earlier TBP/DNA analyses fromSaccharomyces cerevisiaeandArabidopsis thaliana. Special insight into TATA box specificity is provided by considering the TBP/DNA complex, not as a protein molecule with bound DNA, but as a DNA duplex with a particularly large minor groove ligand. This point of view provides explanations for: (1) why T·A base-pairs are required rather than C·G; (2) why an alternation of T and A bases is needed; (3) how TBP recognizes the upstream and downstream ends of the TATA box in order to bind properly; and (4) why the second half of the TATA box can be more variable than the first.     

DNA sequence preferences of several AT-selective minor groove binding ligands.

We have examined the interaction of distamycin, netropsin, Hoechst 33258 and berenil, which are AT-selective minor groove-binding ligands, with synthetic DNA fragments containing different arrangements of AT base pairs by DNase I footprinting. For fragments which contain multiple blocks of (A/T)4 quantitative DNase I footprinting reveals that AATT and AAAA are much better binding sites than TTAA and TATA. Hoechst 33258 shows that greatest discrimination between these sites with a 50-fold difference in affinity between AATT and TATA. Alone amongst these ligands, Hoechst 33258 binds to AATT better than AAAA. These differences in binding to the various AT-tracts are interpreted in terms of variations in DNA minor groove width and suggest that TpA steps within an AT-tract decrease the affinity of these ligands. The behaviour of each site also depends on the flanking sequences; adjacent pyrimidine-purine steps cause a decrease in affinity. The precise ranking order for the various binding sites is not the same for each ligand.     

DNA and protein footprinting analysis of the modulation of DNA binding by the N-terminal domain of the Saccharomyces cerevisiae TATA binding protein

Recombinant full-length Saccharomyces cerevisiae TATA binding protein (TBP) and its isolated C-terminal conserved core domain (TBPc) were prepared with measured high specific DNA-binding activities. Direct, quantitative comparison of TATA box binding by TBP and TBPc reveals greater affinity by TBPc for either of two high-affinity sequences at several different experimental conditions. TBPc associates more rapidly than TBP to TATA box bearing DNA and dissociates more slowly. The structural origins of the thermodynamic and kinetic effects of the N-terminal domain on DNA binding by TBP were explored in comparative studies of TBPc and TBP by "protein footprinting" with hydroxyl radical (*OH) side chain oxidation. Some residues within TBPc and the C-terminal domain of TBP are comparably protected by DNA, consistent with solvent accessibility changes calculated from core domain crystal structures. In contrast, the reactivity of some residues located on the top surface and the DNA-binding saddle of the C-terminal domain differs between TBP and TBPc in both the presence and absence of bound DNA; these results are not predicted from the crystal structures. A strikingly different pattern of side chain oxidation is observed for TBP when a nonionic detergent is present. Taken together, these results are consistent with the N-terminal domain actively modulating TATA box binding by TBP and nonionic detergent modulating the interdomain interaction.