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.

Site-specific DNA binding of architectural protein integration host factor (IHF) is involved in formation of functional multiprotein-DNA assemblies in Escherichia coli, while non-specific binding of IHF and other histone-like proteins serves to structure the nucleoid. Here, we report an isothermal titration calorimetry study of the thermodynamics of binding IHF to a 34 bp fragment composed entirely of the specific H' site from lambda-phage DNA. At low to moderate [K(+)] (60-100 mM), strong competition is observed between specific and non-specific binding as a result of a low specificity ratio (approximately 10(2)) and a very small non-specific site size. In this [K(+)] range, both specific and non-specific binding are enthalpy-driven, with large negative enthalpy, entropy and heat capacity changes and binding constants that are insensitive to [K(+)]. Above 100 mM K(+), only specific binding is observed, and both the binding constant and the magnitudes of enthalpy, entropy and heat capacity changes all decrease strongly with increasing [K(+)]. When interpreted in the context of the structure of the specific complex, the thermodynamics provide compelling evidence for a previously unrecognized design principle by which proteins that form extensive binding interfaces with nucleic acids control binding constants, binding site sizes and effects of temperature and ion concentrations on stability and specificity. We propose that up to 22 of the 23 IHF cationic side-chains that are located within 6 A of DNA phosphate oxygen atoms in the complex, are masked in the absence of DNA by pairing with anionic carboxylate groups in intramolecular salt-bridges (dehydrated ion-pairs). These salt-bridges increase in stability with increasing temperature and decreasing [K(+)]. To explain the unusual thermodynamics of IHF-DNA interactions, we propose that both specific and non-specific binding at low [K(+)] require disruption of salt-bridges (as many as 18 for specific binding) whereupon many of the unmasked charged groups hydrate and the cationic groups interact with DNA. From structural or thermodynamic parallels with IHF, we propose that large-scale coupling of disruption of protein salt-bridges to DNA binding is significant for other large-interface DNA wrapping proteins including the nucleosome, lac repressor core tetramer, RNA polymerase core protein, HU and SSB.

Thermodynamic analysis of interactions between denaturants and protein surface exposed on unfolding: interpretation of urea and guanidinium chloride m-values and their correlation with changes in accessible surface area (ASA) using preferential interaction coefficients and the local-bulk domain model.

A denaturant m-value is the magnitude of the slope of a typically linear plot of the unfolding free energy change DeltaG degrees (obs) vs. molar concentration (C(3)) of denaturant. For a given protein, the guanidinium chloride (GuHCl) m-value is approximately twice as large as the urea m-value. Myers et al. (Protein Sci 1995;4:2138-2148) found that experimental m-values for protein unfolding in both urea and GuHCl are proportional to DeltaASA(corr)(max), the calculated maximum amount of protein surface exposed to water in unfolding, corrected empirically for the effects of disulfide crosslinks: (urea m-value/DeltaASA(corr)(max)) = 0.14+/-0.01 cal M(-1) A(-2) and (GuHCl m-value/DeltaASA(corr)(max)) = 0.28+/-0.03 cal M(-1) A(-2). The observed linearity of plots of DeltaG degrees (obs) vs. C(3) indicates that the difference in preferential interaction coefficients DeltaGamma(3) characterizing the interactions of these solutes with denatured and native protein surface is approximately proportional to denaturant concentration. The proportionality of m-values to DeltaASA(corr)(max) indicates that the corresponding DeltaGamma(3) are proportional to DeltaASA(corr)(max) at any specified solute concentration. Here we use the local-bulk domain model of solute partitioning in the protein solution (Courtenay et al., Biochemistry 2000;39:4455-4471) to obtain a novel quantitative interpretation of denaturant m-values. We deduce that the proportionality of m-value to DeltaASA(corr)(max) results from the proportionality of B(1)(0) (the amount of water in the local domain surrounding the protein surface exposed upon unfolding) to DeltaASA(corr)(max). We show that both the approximate proportionality of DeltaGamma(3) to denaturant concentration and the residual dependence of DeltaGamma(3)/m(3) (where m(3) is molal concentration) on denaturant concentration are quantitatively predicted by the local-bulk domain model if the molal-scale solute partition coefficient K(P) and water-solute exchange stoichiometry S(1,3) are independent of solute concentration. We obtain K(P,urea) = 1.12+/-0.01 and K(P,GuHCl) = 1.16+/-0.02 (or K(P,GuH+) congruent with 1.48), values which will be useful to characterize the effect of accumulation of those solutes on all processes in which the water-accessible area of unfolded protein surface changes. We demonstrate that the local-bulk domain analysis of an m-value plot justifies the use of linear extrapolation to estimate ( less, similar 5% error) the stability of the native protein in the absence of denaturant (DeltaG(o)(o)), with respect to a particular unfolded state. Our surface area calculations indicate that published m-values/DeltaASA ratios for unfolding of alanine-based alpha-helical oligopeptides by urea and GuHCl exceed the corresponding m-value/DeltaASA ratios for protein unfolding by approximately fourfold. We propose that this difference originates from the approximately fourfold difference (48% vs. 13%) in the contribution of polar backbone residues to DeltaASA of unfolding, a novel finding which supports the long-standing but not universally accepted hypothesis that urea and guanidinium cation interact primarily with backbone amide groups. We propose that proteins which exhibit significant deviations from the average m-value/DeltaASA ratio will be found to exhibit significant deviations from the expected amount and/or average composition of the surface exposed on unfolding.

Wrapping of flanking non-operator DNA in lac repressor-operator complexes: implications for DNA looping.

In our studies of lac repressor tetramer (T)-lac operator (O) interactions, we observed that the presence of extended regions of non-operator DNA flanking a single lac operator sequence embedded in plasmid DNA produced large and unusual cooperative and anticooperative effects on binding constants (Kobs) and their salt concentration dependences for the formation of 1:1 (TO) and especially 1:2 (TO2) complexes. To explore the origin of this striking behavior we report and analyze binding data on 1:1 (TO) and 1:2 (TO2) complexes between repressor and a single O(sym) operator embedded in 40 bp, 101 bp, and 2514 bp DNA, over very wide ranges of [salt]. We find large interrelated effects of flanking DNA length and [salt] on binding constants (K(TO)obs, K(TO2)obs) and on their [salt]-derivatives, and quantify these effects in terms of the free energy contributions of two wrapping modes, designated local and global. Both local and global wrapping of flanking DNA occur to an increasing extent as [salt] decreases. Global wrapping of plasmid-length DNA is extraordinarily dependent on [salt]. We propose that global wrapping is driven at low salt concentration by the polyelectrolyte effect, and involves a very large number (>/similar 20) of coulombic interactions between DNA phosphates and positively charged groups on lac repressor. Coulombic interactions in the global wrap must involve both the core and the second DNA-binding domain of lac repressor, and result in a complex which is looped by DNA wrapping. The non-coulombic contribution to the free energy of global wrapping is highly unfavorable ( approximately +30-50 kcal mol(-1)), which presumably results from a significant extent of DNA distortion and/or entropic constraints. We propose a structural model for global wrapping, and consider its implications for looping of intervening non-operator DNA in forming a complex between a tetrameric repressor (LacI) and one multi-operator DNA molecule in vivo and in vitro. The existence of DNA wrapping in LacI-DNA interactions motivates the proposal that most if not all DNA binding proteins may have evolved the capability to wrap and thereby organize flanking regions of DNA.

Enthalpy and heat capacity changes for formation of an oligomeric DNA duplex: interpretation in terms of coupled processes of formation and association of single-stranded helices.

The thermodynamics of self-assembly of a 14 base pair DNA double helix from complementary strands have been investigated by titration (ITC) and differential scanning (DSC) calorimetry, in conjunction with van't Hoff analysis of UV thermal scans of individual strands. These studies demonstrate that thermodynamic characterization of the temperature-dependent contributions of coupled conformational equilibria in the individual "denatured" strands and in the duplex is essential to understand the origins of duplex stability and to derive stability prediction schemes of general applicability. ITC studies of strand association at 293 K and 120 mM Na+ yield an enthalpy change of -73 +/- 2 kcal (mol of duplex)-1. ITC studies between 282 and 312 K at 20, 50, and 120 mM Na+ show that the enthalpy of duplex formation is only weakly salt concentration-dependent but is very strongly temperature-dependent, decreasing approximately linearly with increasing temperature with a heat capacity change (282-312 K) of -1.3 +/- 0.1 kcal K-1 (mol of duplex)-1. From DSC denaturation studies in 120 mM Na+, we obtain an enthalpy of duplex formation of -120 +/- 5 kcal (mol of duplex)-1 and an estimate of the corresponding heat capacity change of -0.8 +/- 0.4 kcal K-1 (mol of duplex)-1 at the Tm of 339 K. van't Hoff analysis of UV thermal scans on the individual strands indicates that single helix formation is noncooperative with a temperature-independent enthalpy change of -5.5 +/- 0.5 kcal at 120 mM Na+. From these observed enthalpy and heat capacity changes, we obtain the corresponding thermodynamic quantities for two fundamental processes: (i) formation of single helices from disordered strands, involving only intrastrand (vertical) interactions between neighboring bases; and (ii) formation of double helices by association (docking) of single helical strands, involving interstrand (horizontal and vertical) interactions. At 293 K and 120 mM Na+, we calculate that the enthalpy change for association of single helical strands is approximately -64 kcal (mol of duplex)-1 as compared to -210 kcal (mol of duplex)-1 calculated for duplex formation from completely unstructured single strands and to the experimental ITC value of -73 kcal (mol of duplex)-1. The intrinsic heat capacity change for association of single helical strands to form the duplex is found to be small and positive [ approximately 0.1 kcal K-1 (mol of duplex)-1], in agreement with the result of a surface area analysis, which also predicts an undetectably small heat capacity change for single helix formation.

DNA footprints of the two kinetically significant intermediates in formation of an RNA polymerase-promoter open complex: evidence that interactions with start site and downstream DNA induce sequential conformational changes in polymerase and DNA.

Kinetic studies of formation and dissociation of open-promoter complexes (RPo) involving Esigma70 RNA polymerase (R) and the lambdaPR promoter (P) demonstrate the existence of two kinetically significant intermediates, designated I1 and I2, and facilitate the choice of conditions under which each accumulates. For such conditions, we report the results of equilibrium and transient DNase I and KMnO4 footprinting studies which characterize I1 and I2. At 0 degreesC, where extrapolation of equilibrium data indicates I1 is the dominant complex, DNA bases in the vicinity of the transcription start site (+1) do not react with KMnO4, indicating that this region is closed in I1. However, the DNA backbone in I1 is extensively protected from DNase I cleavage; the DNase I footprint extends approximately 30 bases downstream and at least approximately 40 bases upstream from the start site. I1 has a short lifetime (

Quantitative analysis of multiple-hit footprinting studies to characterize DNA conformational changes in protein-DNA complexes: application to DNA opening by Esigma70 RNA polymerase.

Formation of many site-specific protein-nucleic acid complexes involves sequential conformational changes subsequent to initial binding which create functionally active assemblies. Characterization of population distributions and structural characteristics of intermediate and product conformations is necessary to understand both the mechanisms and the thermodynamics of these processes. For these purposes, here we develop the quantitative method of multiple hit footprinting (MHF), where chemical or enzymatic probing is performed as a function of either concentrations of the footprinting agent and/or time of exposure to it, in the multiple hit regime where many of the population or subpopulation of reactive DNA molecules are modified at more than one site. Properly controlled MHF experiments yield both the population distribution of different conformers and reactivity rate constants of the footprinting agent at all reactive positions in each conformer, which may be interpreted in terms of the accessibility of the site or the local concentration of the reagent. MHF experiments are particularly well-suited for dissecting effects at sites where unbound DNA is non-reactive and bound DNA is reactive with base-specific probes (e.g. KMnO4, DMS). We suggest that this method will also be applicable to analysis of enhancements in reactivity of other footprinting agents (e.g. DNase I, HO.). To demonstrate the utility of the MHF analysis, we quantify fragment distributions and individual site reactivities from multiple-hit KMnO4 footprinting of the non-template strand of Esigma70 RNA polymerase-lambdaPR promoter DNA complexes populated at binding equilibrium at 37 degreesC and transiently populated at a fixed time after a temperature downshift from 37 degreesC to 0 degreesC. For this system, a MHF analysis directly addresses the following questions: (i) what fraction of the population of promoter DNA molecules is open in the vicinity of the transcription start site (RPo) both at 37 degreesC and (transiently) after a downshift to 0 degreesC; (ii) does opening of the start site region in RPo occur entirely in one mechanistic step at the lambdaPR promoter and (iii) does the structure of RPo vary with temperature? In addition, we use the MHF-determined population distribution of KMnO4-reactive (RPo) and non-reactive promoter DNA to normalize the biphasic kinetics of decay of RPo to free promoter DNA after a 37 degrees to 0 degreesC temperature downshift, and thereby characterize the kinetics of the conformational changes involved in forming RPo.

Thermodynamics of the interactions of lac repressor with variants of the symmetric lac operator: effects of converting a consensus site to a non-specific site.

What are the thermodynamic consequences of the stepwise conversion of a highly specific (consensus) protein-DNA interface to one that is nonspecific? How do the magnitudes of key favorable contributions to complex stability (burial of hydrophobic surfaces and reduction of DNA phosphate charge density) change as the DNA sequence of the specific site is detuned? To address these questions we investigated the binding of lac repressor (LacI) to a series of 40 bp fragments carrying symmetric (consensus) and variant operator sequences over a range of temperatures and salt concentrations. Variant DNA sites contained symmetrical single and double base-pair substitutions at positions 4 and/or 5 [sequence: see text] in each 10 bp half site of the symmetric lac operator (Osym). Non-specific interactions were examined using a 40 bp non-operator DNA fragment. Disruption of the consensus interface by a single symmetrical substitution reduces the observed equilibrium association constant (K(obs)) for Osym by three to four orders of magnitude; double symmetrical substitutions approach the six orders in magnitude difference between specific and non-specific binding to a 40 bp fragment. At these adjacent positions in the consensus site, the free energy effects of multiple substitutions are non-additive: the first reduces /deltaG(obs)o/ by 3 to 5 kcal mol(-1), approximately halfway to the non-specific level, whereas the second is less deleterious, reducing /deltaG(obs)o/ by less than 3 kcal mol(-1). Variant-specific dependences of K(obs) on temperature and salt concentration characterize these LacI-operator interactions. In general, binding constants and standard free energies of binding both exhibit characteristic extrema near 290 K. As a consequence, both the enthalpic and entropic contributions to stability of Osym and variant complexes change from positive (i.e. entropy driven) at lower temperatures to negative (i.e. enthalpy driven) at higher temperatures, indicating that the heat capacity change upon binding, deltaC(obs)o, is large and negative. In general, /deltaC(obs)o/ decreases as the specificity and stability of the variant complex decreases. Stabilities of complexes of LacI with Osym and all variant operators are strongly [salt]-dependent. Binding constants for the variant complexes exhibit a power-dependence on [salt] that is larger in magnitude (i.e. more negative) than for Osym, but no obvious trend relates changes in contributions from the polyelectrolyte effect and the observed reductions in stability (delta deltaG(obs)o). These variant-specific thermodynamic signatures provide novel insights into the consequences of converting a consensus interface to a less specific one; such insights are not obtained from comparisons at the level of delta deltaG(obs)o. We propose that this variant-specific behavior arises from a strong effect of operator sequence on the extent of induced conformational changes in the protein (and possibly also in the DNA site) which accompany binding.

Cooperative and anticooperative effects in binding of the first and second plasmid Osym operators to a LacI tetramer: evidence for contributions of non-operator DNA binding by wrapping and looping.

The interaction of lac operator DNA with lac repressor (LacI) is a classic example of a genetic regulatory switch. To dissect the role of stoichiometry, subunit association, and effects of DNA length in positioning this switch, we have determined binding isotherms for the interaction of LacI with a high affinity (Osym) operator on linearized plasmid (2500 bp) DNA over a wide range of macromolecular concentrations (10(-14) to 10(-8) M). Binding data were analyzed using a thermodynamic model involving four equilibria: dissociation of tetramers (T) into dimers (D), and binding of operator-containing plasmid DNA (O) to dimers and tetramers to form three distinct complexes, DO, TO, and TO2. Over the range of concentrations of repressor, operator, and salt (0.075 M K+ to 0.40 M K+) investigated, we find no evidence for any significant thermodynamic effect of LacI dimers. Instead, all isotherms can be interpreted in terms of just two equilibria, involving only T and the TO and TO2 complexes. As a reference binding equilibrium, which we propose must approximate the DO binding interaction, we compare the plasmid Osym results with our extensive studies of the binding of a 40 bp Osym DNA fragment to LacI. On this basis, we obtain a lower bound on the LacI dimer-tetramer equilibrium constant and values of the equilibrium constants for formation of TO and TO2 complexes. At a salt concentration of 0.40 M, the Osym plasmid binding data are consistent with a model with two independent and identical binding sites for operator per LacI tetramer, in which the binding to a site on the tetramer is only slightly more favorable than the reference binding interaction. Increasingly large deviations from the independent-site model are observed as the salt concentration is reduced; binding of a second operator to from TO2 becomes strongly disfavored relative to formation of TO at low salt concentrations (0.075 to 0.125 M). In addition, binding of both the first and second plasmid operator DNA molecules to the tetramer becomes increasingly more favorable than the reference binding interaction as [K+] is reduced from 0.40 M to 0.125 M. At 0.075 M K+, however, the strength of binding of the second plasmid operator DNA to the LacI tetramer is dramatically reduced; this interaction is much less favorable than binding the first plasmid operator DNA, and becomes much less favorable than the reference binding interaction. We propose that these differences arise from changes in the nature of the TO and TO2 complexes with decreasing salt concentration. At low salt concentration, we suggest the hypothesis that flanking non-operator sequences bind non-specifically (coulombically) by local wrapping, and that distant regions of non-operator DNA occupy the second operator-binding site by looping. We propose that wrapping stabilizes both 1:1 and 2:1 complexes at low salt concentration, and that looping stabilizes the 1:1 complex but competitively destabilizes the 2:1 TO2 complex at low salt concentration. These effects must play a role in adjusting the stability and structure of the LacI-lac operator repression complex as the cytoplasmic [K+] varies in response to changes in extracellular osmolarity.

Compensating effects of opposing changes in putrescine (2+) and K+ concentrations on lac repressor-lac operator binding: in vitro thermodynamic analysis and in vivo relevance.

Ion concentrations (K+, Glu-) in the cytoplasm of growing Escherichia coli cells increase strongly with increases in the osmolarity of a defined growth medium. While in vitro experiments demonstrate that the extent of protein-nucleic acid interactions (PNAI) depends critically on salt concentration, in vivo measurements indicate that cells maintain a relatively constant extent of PNAI independent of the osmolarity of growth. How do cells buffer PNAI against changes in the cytoplasmic environment? At high osmolarity, the increase in macromolecular crowding which accompanies the reduction in amount of cytoplasmic water in growing cells appears quantitatively sufficient to compensate for the increase in [K+]. At low osmolarity, however, changes in crowding appear to be insufficient to compensate for changes in [K+], and additional mechanisms must be involved. Here we report quantitative determinations of in vivo total concentrations of polyamines (putrescine(2+), spermidine(3+)) as a function of osmolarity (OsM) of growth, and in vitro binding data on the effects of putrescine concentration on a specific PNAI (lac repressor-lac operator) as a function of [K+]. The total concentration of putrescine in cytoplasmic water decreases at least eightfold from low osmolarity (approximately 64 mmol (l H2O)-1 at 0.03 OsM) to high osmolarity (approximately 8 mmol (l H2O)-1 at 1.02 OsM). Over this osmotic range the total [K+] increases from approximately 0.2 mol (l H2O)-1 to approximately 0.8 mol (lH2O)-1. We find that the effect of putrescine concentration on the repressor-operator interaction in vitro is purely competitive and is quantitatively described by a simple competition formalism in which lac repressor behaves a a specific-binding oligocation (ZR = 8+/-3). We demonstrate that this thermodynamic result is consistent with a structural analysis of the number of positively charged side-chains on two DNA binding domains of repressor which interact with the phosphodiester backbone of the operator site. Since this oligocation character of the binding surface of DNA-binding proteins appears to be general, we propose the competitive effects of putrescine and K+ concentrations on the strength of specific binding are general. At low osmolarity, compensating changes in putrescine and K+ concentration in response to changes in external osmolarity provide a general mechanism for E. coli to vary cytoplasmic osmolarity while maintaining a constant extent of PNAI.