HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

J. Biol. Chem., Vol. 277, Issue 52, 50676-50682, December 27, 2002

This Article Abstract Full Text (PDF) All Versions of this Article: 277/52/50676    most recent M208540200v1 Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Fried, M. G. Articles by Lu, P. Search for Related Content PubMed PubMed Citation Articles by Fried, M. G. Articles by Lu, P. Social Bookmarking                      What's this?

Role of Hydration in the Binding of lac Repressor to DNA^*

Michael G. Fried^§, Douglas F. Stickle^¶, Karen Vossen Smirnakis, Claire Adams, Douglas MacDonald^**, and Ponzy Lu^**

From the  Department of Biochemistry and Molecular Biology, Penn State University College of Medicine, Hershey, Pennsylvania 17033, the ^¶ Department of Pathology and Microbiology, Nebraska Medical Center, Omaha, Nebraska 68198, the  Department of Medicine, Massachusetts General Hospital, Boston, Massachusetts 02114, and the ^** Department of Chemistry, University of Pennsylvania, Philadelphia, Pennsylvania 19104

Received for publication, August 20, 2002, and in revised form, October 5, 2002

	ABSTRACT

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

The osmotic stress technique was used to measure changes in macromolecular hydration that accompany binding of wild-type Escherichia coli lactose (lac) repressor to its regulatory site (operator O1) in the lac promoter and its transfer from site O1 to nonspecific DNA. Binding at O1 is accompanied by the net release of 260 ± 32 water molecules. If all are released from macromolecular surfaces, this result is consistent with a net reduction of solvent-accessible surface area of 2370 ± 550 Å². This area is only slightly smaller than the macromolecular interface calculated for a crystalline repressor dimer-O1 complex but is significantly smaller than that for the corresponding complex with the symmetrical optimized O_sym operator. The transfer of repressor from site O1 to nonspecific DNA is accompanied by the net uptake of 93 ± 10 water molecules. Together these results imply that formation of a nonspecific complex is accompanied by the net release of 165 ± 43 water molecules. The enhanced stabilities of repressor-DNA complexes with increasing osmolality may contribute to the ability of Escherichia coli cells to tolerate dehydration and/or high external salt concentrations.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

The control of transcription initiation involves the binding of gene regulatory proteins to regulatory and competing genomic DNA sequences. The stability and specificity of these interactions depend on their solution environment. Important variables include salt concentration and identity (1-3), pH (4, 5), pressure (6-8), and accessible volume (9, 10). In addition, changes in hydration accompany macromolecular interactions (reviewed in Refs. 8, 11, and 12). For those interactions in which the hydration change is large, the free energies of interaction depend sensitively on the activity of water (a_H2O).

The intimate association of protein and DNA surfaces is accompanied by the displacement of water molecules associated with those surfaces. In addition, allosteric changes that extend beyond the contact surfaces can alter the solvent-accessible surface areas of protein and DNA and, thus, the numbers of associated water molecules. Any water molecules bound or released in these transactions are reactants or products, respectively, in the binding reaction. Changes in the number of thermodynamically associated water molecules can be detected and quantitated by the osmotic stress technique (12-14), using small, neutral solutes (osmolytes) that are typically excluded from the volumes immediately adjacent to macromolecular surfaces (11, 15). With three caveats, the dependence of the affinity of protein for DNA (K_obs) on water activity is a measure of the net change in the number of water molecules that are associated with the participating macromolecules. These are as follows: (i) that K_obs should not be significantly perturbed by the differential interaction of osmolytes with reactants and products; (ii) that volume exclusion by osmolytes should not significantly alter K_obs; and (iii) that changes in solvent properties that indirectly affect binding affinity (for example, the dielectric coefficient) should not account for changes in K_obs.

Here we report the use of osmotic stress to measure the water stoichiometries of lac repressor binding to its primary lac promoter regulatory site (O1) and to nonspecific DNA. In the absence of low molecular weight inducers (e.g. allolactose), lac repressor binds its regulatory sites (operators O1, O2, and O3 (16)) and promoter-bound RNA polymerase (17) and inhibits one or more initial step(s) of the transcription process (17-19). In the cell, nonspecific DNA competes with target sequences for lac repressor binding. The distribution of repressor between regulatory and nonspecific sequences determines the occupancy of operator O1, and hence the transcriptional activity of the lac promoter (20-22). As shown below, under standard in vitro conditions, this distribution depends on the water activity.

In principle, the value of the water stoichiometry may contain information about structural changes that accompany binding. If the water stoichiometry of a reaction reflects the net association or dissociation of water molecules from the macromolecular surface, its value should be proportional to the sum of changes in solvent-accessible surface areas of reaction participants. For proteins, it has been estimated that an inner hydration layer water molecule occupies 9 ± 1 Å² of surface (14). We use this value, with caveats, to obtain an estimate of the change in solvent-accessible surface area accompanying lac repressor-DNA interaction. This estimated change is similar to one calculated for an isosteric model of the interaction, using the protein and DNA structures present in the co-crystal solved by Bell and Lewis (23). This similarity is surprising, because both repressor and operator DNA have been shown to undergo allosteric transitions during sequence-specific binding (cf. Refs. 23-26).

	EXPERIMENTAL PROCEDURES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Reagents-- Betaine, ethylene glycol, propylene glycol, triethylene glycol, polyethylene glycols 400 and 8000, acrylamide, N,N'-methylenebisacrylamide, and bacterial alkaline phosphatase were purchased from Sigma. Acetamide and polyethylene glycol 1450 were from Eastman Kodak. Phage T4 polynucleotide kinase was purchased from New England Biolabs; [-³²P]ATP was from Amersham Biosciences.

Protein and DNA Samples-- Two preparations of wild-type lac repressor were kindly provided Dr. Kathleen Matthews. The protein was >95% pure as judged by SDS-PAGE and >50% active in lac operator binding (27). Repressor concentrations were determined spectrophotometrically using ₂₈₀ = 2.2 × 10⁴ M¹ cm¹/repressor subunit (28). Protein concentrations given throughout this paper refer to the species active in sequence-specific DNA binding.

Plasmid pMM02 DNA was purified as described (29). A 77-bp restriction fragment, containing one copy of the wild-type lac operator, was resolved from vector DNA by cleavage with endonucleases HindIII and HpaII and purified by preparative gel electrophoresis (30). DNA molecules were labeled at 5' termini with ³²P (31). Genomic DNA from Escherichia coli was purchased from Sigma. Samples were sonicated, deproteinized, and fractionated as described (32). All DNA samples were dialyzed extensively against 10 mM Tris (pH 8.0 at 21 ± 1 °C), 1 mM EDTA prior to use. DNA concentrations were determined spectrophotometrically using ₂₆₀ = 1.3 × 10⁴ M¹ cm¹/bp.

Formation and Detection of Protein-DNA Complexes-- Binding reactions were carried out at 21 (±1) °C in solutions containing 10 mM Tris-HCl (pH 8.0), 1 mM EDTA, 250 mM KCl, and 0.1 mg of bovine serum albumin/ml, supplemented with osmolyte (acetamide, betaine, ethylene glycol, propylene glycol, triethylene glycol, and polyethylene glycols) as appropriate. Mixtures were equilibrated for 1 h. Duplicate samples incubated for longer periods gave identical results, indicating that equilibrium had been attained. Samples were analyzed by native gel electrophoresis as described¹ (29, 33-35). For the lac repressor-DNA system, these conditions have been shown to allow accurate quantitation of free and bound DNA species (29, 67).

Analysis of Binding-- For determination of K_obs, the simple mechanism shown in Equation 1 was assumed.

(Eq. 1)

Here R represents repressor; O represents operator DNA; R·O the repressor-operator complex; and K_obs = [R·O]/[R][O]. Values of [O] and [R·O] were determined by quantitation of the ³²P associated with the corresponding gel bands. Values of K_obs were obtained by fitting Equation 2 to the binding data.

(Eq. 2)

Here Y = [R·O]/([O] + [R·O]), and [R]_t and [O]_t are the total concentrations of repressor and operator DNA molecules, respectively (29).

In binding competition assays, repressor-operator complexes were titrated with unlabeled E. coli genomic DNA, establishing the equilibrium shown in Equation 3,

(Eq. 3)

in which R, O, and D represent repressor, operator, and competing DNA sites, respectively. At the ith step of the titration, the specificity ratio is given by Equation 4,

(Eq. 4)

Here [R·O] and [O] are measured quantities. The concentrations [R·D] and [D] were calculated using [R·D] = [R·O]_o  [R·O]_i and [D] = [D]_o  m[R·D], in which [R·O]_o and [D]_o are equal to the initial concentrations of R·O and D; [R·O]_i is equal to the measured concentration of R·O at the ith titration step, and m is equal to the number of base pairs occupied by the protein (33, 36).

Osmometry and Densimetry-- Sample osmolalities were measured with a vapor pressure osmometer (41). In aqueous solutions, osmolality (Os) and water activity are related by lna_H2O = Os , in which is the molal volume of water, and osmolality is given in units of osmol kg¹. Solution densities were measured as a function of osmolyte concentration using a Mettler densitometer, operating at 21 °C. Apparent partial specific volumes, _app, were calculated using the relationship (37) shown in Equation 5,

(Eq. 5)

in which  is the solution density; _o is the solvent density; and c is the osmolyte concentration (all in grams per milliliter). The value of _app extrapolated to infinite dilution was taken as the partial specific volume, (37). Molar volumes were calculated according to V_mol = ·M_r.

Error Analysis-- 95% confidence intervals for fitted parameters were estimated by the method of Broderson et al. (38).

	RESULTS

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Effects of Neutral Solutes on the Binding of lac Repressor to the Primary Lac Operator-- The binding of repressor (R) to operator (O) in an aqueous solution containing a neutral solute (S) can be represented by Equation 6,

(Eq. 6)

in which _H2O and _S are the stoichiometric coefficients of water and solute, respectively. Following standard approaches (14, 40), it can be shown that ln K_obs depends on lna_H2O as shown in Equation 7 (57).

(Eq. 7)

Timasheff and co-workers (11, 15) have shown that many low molecular weight osmolytes are excluded from the volumes immediately adjacent to proteins (i.e. that these volumes are occupied preferentially by water molecules). Thus, for a process involving a large change in solvent-accessible surface area, in a solution containing such a "non-interacting" solute, one might expect _H2O to contribute more importantly to ln K_obs/lna_H2O than does _S. As described below, this expectation can be tested by comparing values of ln K_obs/lna_H2O obtained with different structurally and chemically distinct osmolytes.

The electrophoresis mobility shift assay (33, 34) was used to detect the binding of wild-type lac repressor to its primary lac operator site, O1 (Fig. 1A). Under conditions of low binding saturation, the predominant complex formed contained one repressor tetramer, occupying site O1 (33, 35, 41). Isotherms for the binding of lac repressor in a buffer containing 250 mM KCl and 0 or 5% v/v ethylene glycol are shown in Fig. 1B. In the absence of osmolyte, the value of K_obs (1.1 ± 0.1 × 10¹⁰ M¹) is in good agreement with values obtained at similar [salt] (29, 42, 43). The value obtained in the presence of 5% ethylene glycol (K_obs = 3.4 ± 0.5 × 10¹¹ M¹) is similar to ones obtained at similar [salt], in the presence of 5% glycerol (29, 44) or 5% Me₂SO (45).

View larger version (35K): [in this window] [in a new window]   Fig. 1.   A, electrophoretic mobility shift analysis: titration of lac operator DNA with lac repressor protein. Addition of lac repressor to a solution containing the free 32P-labeled lac promoter fragment (F) resulted in the formation of a 1:1 repressor-operator complex (R). All reactions were carried out in 10 mM Tris-HCl, pH 8.0, 1 mM EDTA, 250 mM KCl, and 0.1 mg of bovine serum albumin/ml. All samples contained 4.2 × 1010 M lac operator DNA. Samples b-k also contained 1.1 × 1011 M, 2.1 × 1011 M, 1.1 × 1010 M, 2.7 × 1010 M, 5.3 × 1010 M, 1.1 × 109 M, 2.1 × 109 M, 5.3 × 109 M, 1.1 × 108 M, and 2.1 × 108 M lac repressor, respectively. B, representative isotherms for the binding of lac repressor to the 77-bp lac operator fragment. The fraction of lac DNA bound (Y) is plotted as a function of the log10 of the free [repressor].  and , binding reactions carried out in the absence of added osmolyte;  and , binding reactions carried out in the presence of 5% v/v ethylene glycol. The solid curves were calculated using values of Kobs = 1.1 ± 0.1 × 1010 M1 (no added osmolyte) and Kobs = 3.4 ± 0.5 × 1011 M1 (5% v/v ethylene glycol) obtained by fitting the binding data as described in the text.

The inclusion of acetamide, betaine, ethylene glycol, propylene glycol, or triethylene glycol in the reaction mixture increases the value of K_obs. For these solutes, ln K_obs is linearly correlated with lna_H2O, with the consensus value of ln K_obs/lna_H2O = 260 ± 32 (Fig. 2). In principle, several factors can contribute to the observed increase in K_obs, including changes in water activity, changes in solvent polarity, and direct solute-macromolecule interactions. An important criterion for discrimination between direct solute interactions with macromolecules and osmotic effects is the sensitivity of the value of ln K_obs/lna_H2O to changes in the identity of the solute (12, 57). The number of solute molecules associated with each macromolecule and the change in that number with protein-DNA interaction (_S) should depend on the identity of the solute. On the other hand, a purely osmotic effect should be independent of solute identity (12, 14). Closely similar values of ln K_obs/lna_H2O were obtained for the five low molecular weight solutes examined (Table I). This could occur only if the product _S·(ln a_S/lna_H2O) was the same, within error, for all solutes tested or if all values of _S·(ln a_S/lna_H2O) were much less than that of _H2O (Equation 7). Since chemically distinct solutes associate to different degrees with proteins (11, 15) and are likely to do so with nucleic acids as well, it seems improbable that _S·(ln a_S/lna_H2O) is the same for all the osmolytes tested. We conclude that preferential macromolecule-osmolyte interactions contribute much less to ln K_obs/lna_H2O than do changes in hydration.

View larger version (27K): [in this window] [in a new window]   Fig. 2.   Dependence of Kobs for the lac repressor-operator O1 DNA binding reaction on the water activity of solutions containing the indicated osmolytes. Values of ln Kobs/lnaH2O for solutions containing ethylene glycol (), propylene glycol (), triethylene glycol (), acetamide (), and betaine (). The error bars represent 95% confidence limits, estimated as described in the text. The solid line is a least squares fit to the combined data, returning ln Kobs/lnaH2O = 260 ± 32.

A second possibility is that the increase in K_obs is a consequence of decrease in the solution dielectric coefficient with osmolyte concentration. If this were the case, ln K_obs would be expected to scale identically with dielectric coefficient for all osmolytes tested. Shown in Fig. 3 are values of ln K_obs as a function of dielectric coefficient for the osmolytes acetamide, betaine, and ethylene glycol. Since the slopes differ markedly for these osmolytes, changes in solvent polarity cannot account for the common effects of different osmolytes on K_obs.

View larger version (18K): [in this window] [in a new window]   Fig. 3.   Dependence of Kobs for the lac repressor-operator binding reaction on the dielectric coefficients of solutions containing acetamide (), ethylene glycol (), and betaine (). Dielectric coefficients were calculated from dielectric increments tabulated by Cohn and Edsall (73). The solid lines are least squares fits to the combined acetamide and ethylene glycol data sets and to the betaine data set. Error bars represent 95% confidence limits.

A third possibility is that the observed changes in K_obs are due to volume exclusion by the osmolytes. A hallmark of such effects is that they increase strongly with solute molar volume (10). Shown in Fig. 4 is the dependence of ln K_obs/lna_H2O on ln V_mol, the natural log of solute molar volume, measured in our assay buffer, for solutes ranging from M_r = 59 (acetamide) to M_r ~8000 (polyethylene glycol 8000). Values of ln K_obs/lna_H2O depend strongly on ln V_mol for the large solutes, but become essentially independent of ln V_mol for the small ones. Shown in Fig. 4, inset, a linear extrapolation of ln K_obs/lna_H2O as a function of V_mol to V_mol = 0 gives a limiting value, 272 ± 17, that is indistinguishable from the consensus value obtained with small osmolytes (260 ± 32). This suggests that volume exclusion has a significant influence on the value of ln K_obs/lna_H2O only when the osmolytes are large (for this system, V_mol  130 ml/mol).

View larger version (22K): [in this window] [in a new window]   Fig. 4.   Dependence of apparent water stoichiometry H2O(app) on solute molar volume. The data are for the binding of lac repressor to the 77-bp lac operator fragment at 21 ± 1 °C in 10 mM Tris-HCl, pH 8.0, 1 mM EDTA, 250 mM KCl, and 0.1 mg of bovine serum albumin/ml. Molar volumes of solutes were calculated from solution densities as described in the text. Error bars representing 95% confidence limits in ln Kobs/lnaH2O are plotted for all data but are obscured by the symbols for some points.

Taken together, the results presented above support the interpretation that ln K_obs/lna_H2O = 260 ± 32, obtained with small osmolytes, is a measure of the net change in macromolecular hydration (_H2O) that accompanies interaction of the lac repressor with the lac operator site O1. This result is in reasonable agreement with a previous estimate (_H2O = 210 ± 30) obtained from analysis of the [salt] dependence of operator binding (46). Such large values indicate that water is by far the dominant stoichiometric component of the binding reaction. If we assume that all water molecules contributing to _H2O occupy the inner hydration layer when associated with a macromolecule, and if a water molecule in the inner hydration layer occupies 9 ± 1 Å² of macromolecular surface as proposed for protein hydration (14), then the value of _H2O specifies the change in solute-excluding, solvent-accessible macromolecular surface area (A_W) that accompanies complex formation. Summarized in Table I, the effects of low molecular weight solutes on K_obs are consistent with the estimate A_W = 2370 ± 550 Å². As shown below, this value is only slightly smaller than the area occluded in the repressor-DNA interface on formation of the complex with operator O1.

Burial of Solvent-accessible Surfaces in the Repressor-Operator Interface-- The crystal structures of lac repressor dimer complexes with the "ideal" O_sym and wild-type O1 operators (23, 25) provide us with models from which to estimate the amount of solvent-accessible surface area that is buried at the repressor-operator interface. Atomic coordinates were obtained from the Protein Data Bank² (PDB files 1EFA and 1JWL (47)), and solvent-accessible surface areas were calculated using the Connolly algorithm (48) as implemented in the program GRASP version 1.25 (49). A probe radius of 1.4 Å (equal to that of the oxygen atom in a water molecule) was used. Models of the solvent-inaccessible surfaces of DNA present in the repressor dimer-operator O1 and repressor dimer-O_sym complexes are shown in Fig. 5. The change in solvent-accessible surface area on formation of each complex (A_S) was calculated using Equation 8,

View larger version (99K): [in this window] [in a new window]   Fig. 5.   Models of the solvent-inaccessible surfaces of DNA present in repressor dimer-operator O1 and repressor dimer-Osym complexes. Top, van der Waals surfaces of the wild-type operator 1 (O1) DNA obtained from its complex with a lac repressor dimer. Bottom, van der Waals surfaces of the ideal operator (Osym) DNA obtained from its complex with a lac repressor dimer. The red regions represent molecular surfaces that are inaccessible to solvent in the complexes, determined using GRASP version 1.25 (49). The gray regions represent surfaces that are accessible to solvent in the complexes. Coordinates of the repressor-O1 and repressor-Osym crystal structures (PDB identifiers 1JWL and 1EFA) were obtained from the Protein Data Bank (47).

(Eq. 8)

Here A_RO is the solvent-accessible surface area of the complex, and A_R and A_O are the corresponding areas calculated for models of free repressor and DNA molecules in which their conformations are identical to those found in the complex. It follows that these calculations do not take into account changes in the solvent-accessible surface areas of the protein and the DNA that results from conformational changes that occur during binding. In addition, they use structural data for the dimeric form of lac repressor, so they cannot take into account conformational differences that may distinguish dimeric and tetrameric forms of the protein or changes that take place at the dimer-dimer interface. With these caveats in mind, the values of A_S for the repressor (dimer)-O_sym and repressor (dimer)-O1 complexes are 3340 and 2600 Ų, respectively. Comparable values have been found for the DNA complexes of other regulatory proteins and restriction endonucleases (Table II). The value of A_S for the O1 complex is in striking agreement with that estimated from the water stoichiometry of the binding reaction of tetrameric, wild-type repressor with operator O1 (A_W = 2370 ± 550 Å²; see above). However, as summarized in Table II, such agreement has not been found for all protein-DNA systems. The source of this discrepancy may lie in the fact that both allostery and the burial of interacting protein and DNA surfaces contribute to the net change in _H2O, whereas the calculation of A_S from a single, static crystal structure does not include allosteric transitions.

The Ratio of K_S/K_N Increases with Decreasing Water Activity-- To regulate transcription, lac repressor partitions between regulatory and nonspecific DNA sites in an inducer-dependent manner (cf. Refs. 16 and 50). The sequence-specific interaction with lac operator O1 is of high affinity and is accompanied by a net release of 7-9 ions (4, 45, 46, 51). The nonspecific interaction with genomic DNA is significantly weaker and is accompanied by a net release of 9-12 ions (46, 51). These contrasts suggest that the change in solvent-accessible surface area accompanying DNA binding may differ for specific and nonspecific binding reactions. If the corresponding water stoichiometries differ, the distribution of repressor between regulatory (specific) and genomic (nonspecific) sites will depend on water activity.

To test these notions, we performed binding competition assays (33, 36) to measure the specificity ratio K_S/K_N as a function of water activity. Here K_S is the association constant for lac operator O1, and K_N is the population-average association constant for genomic DNA. Shown in Fig. 6A, solutions containing ³²P-labeled lac operator DNA and repressor were titrated with unlabeled E. coli genomic DNA. The transfer of lac repressor from the operator to competitor reduces the amount of ³²P-labeled repressor-operator complex (R) and increases the amount of free ³²P-labeled DNA (F). Analyzed according to Equation 4, these data gave a value of K_S/K_N = 4.7 (± 1.5) × 10⁵, in reasonable agreement with measurements made previously under similar buffer conditions (33, 46). Results of competition experiments in which a_H2O was varied by inclusion of acetamide, ethylene glycol, glycerol, and sucrose are shown in Fig. 6B. The consensus value of ln(K_S/K_N)/lna_H2O is 95 ± 11; values for individual osmolytes are summarized in Table III. Taken with this competition data, the value _H2O = 260 ± 32 for formation of the specific repressor-operator complex implies that the formation of a nonspecific repressor-DNA complex is accompanied by the net release of 165 ± 43 water molecules. This range of values is significantly less negative than that for the specific interaction, consistent with the notion that the reduction in solvent-accessible surface area accompanying nonspecific binding (A_W) is less than that for a specific interaction.

View larger version (39K): [in this window] [in a new window]   Fig. 6.   A, electrophoretic mobility shift analysis: binding competition assay. Repressor·O1 complexes (R) were incubated with increasing concentrations of unlabeled E. coli DNA. Transfer of repressor to the competitor results in the liberation of lac operator DNA (F). All reactions were carried out in 10 mM Tris-HCl, pH 8.0, 1 mM EDTA, 250 mM KCl, 0.1 mg of bovine serum albumin/ml, and 5% ethylene glycol. All samples contained 2.2 × 1011 M lac operator DNA and 3.4 × 1011 M repressor protein. Competing DNA concentrations in samples b-j were, respectively, 3.9 × 107 M, 3.9 × 106 M, 1.3 × 105 M, 7.5 × 105 M, 1.3 × 104 M, 4.2 × 104 M, 1.8 × 103 M, 3.5 × 103 M, and 9.7 × 103 M (base pairs). B, the distribution of lac repressor between lac O1 and nonspecific E. coli DNA depends on water activity. Graph of ln(KS/KN) as a function of lnaH2O in solutions containing the indicated osmolytes: ethylene glycol (), propylene glycol (), and betaine (), respectively. Error bars represent 95% confidence limits.

	DISCUSSION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

The binding of lac repressor to DNA brings together initially hydrated surfaces and is accompanied by allosteric transitions in both molecules (23, 25). Typically, protein-DNA interactions reduce the net surface area accessible to solvent (Table II) (52, 53) and result in a release of water (however see Ref. 54 for an interesting exception). The liberation of water molecules provides a powerful entropic driving force for such reactions. The osmotic stress method introduced by Parsegian and co-workers (12, 13, 55) allows measurement of the change in the number of water molecules present in system compartments that are inaccessible to neutral solutes. These compartments include cavities and molecular interfaces from which solutes are sterically excluded and surfaces that are preferentially hydrated (8, 12).

By using this method, we have found that ln K_obs/lna_H2O = 260 ± 32 for the binding of E. coli lac repressor to lactose operator O1, in osmolyte solutions. This value was nearly independent of V_mol for small osmolytes (acetamide, ethylene glycol, propylene glycol, betaine, and triethylene glycol). However, when larger osmolytes were tested (polyethylene glycols 400, 1450, and 8000), ln K_obs/lna_H2O increased with increasing V_mol, as expected if volume exclusion contributed significantly to K_obs. We conclude that solute molar volume-dependent perturbations of K_obs can be detected by our assay method but that they are not significant, under our conditions, when small osmolytes (V_mol  130 ml/mol) are used. We found that ln K_obs/lna_H2O exhibited a different dependence on solution dielectric coefficient in reactions in which betaine was the osmolyte than it did in reactions in which acetamide or ethylene glycol was the osmolyte. This outcome is inconsistent with models in which changes in K_obs are a direct consequence of changes in the solution dielectric coefficient. Finally, the agreement among values of ln K_obs/lna_H2O obtained with chemically diverse osmolytes argues that direct solute-macromolecule interactions are not the dominant source of changes in K_obs. Together, these results support our interpretation that ln K_obs/lna_H2O is a measure of _H2O, the change in the number of macromolecule-associated, solute-excluding water molecules that occurs on complex formation.

The water stoichiometry of the repressor-O1 interaction (_H2O = 260 ± 32) is more negative than that of the gal repressor-operator (180  _H2O  100 (56)) or that of EcoRI with its canonical site (_H2O approximately 160 (8)). It is significantly more negative than that found for the binding of CAP to its primary lac promoter site (_H2O = 79 ± 11 (57)). Although these values span a significant range, water is by far the largest known stoichiometric component in these reactions. The release of large numbers of water molecules is compatible with the proposition (52, 53) that DNA binding is driven, in part, by the removal of large amounts of non-polar surface from water. However, our understanding of the relationship between the change in solvent-accessible surface area and the experimentally derived water stoichiometry is only approximate. If all water molecules contributing to the measured value of _H2O occupy the inner hydration layer when associated with a macromolecule, and if each water molecule occupies 9 ± 1 Å² of macromolecular surface (14), the value of _H2O = 260 ± 32 obtained for lac repressor implies that the change in solute-excluding, solvent-accessible surface area (A_W) is equal to 2370 ± 550 Å². However, not all water molecules need be released from macromolecular surfaces. If any water molecules are released from outer hydration layers, the area buried on complex formation should be less than or equal to that occupied by a monolayer containing _H2O water molecules. In addition, the estimate that one water molecule occupies 9 ± 1 Å of macromolecular surface is based on results obtained with proteins (see Ref. 14 and references cited therein). The area occupied by a water molecule on a highly charged DNA surface may be smaller as a consequence of electrostriction (58). These considerations suggest that the largest value of A_W is an upper limit for the net change in solvent-accessible surface area.

Both allostery and the burial of complementary protein and DNA surfaces contribute to the net change in _H2O. The calculated change in solvent-accessible surface area in the repressor-O1 interface (A_S) is 2600 Ų. A monolayer of water occupying a surface of this size would contain ~290 water molecules, a value consistent with the measured number of water molecules released (260 ± 32). This similarity raises the intriguing possibility that the allosteric transitions that accompany site-specific binding, including re-organization of the amino-terminal domains of lac repressor (25), and a 40-45° DNA bend (23, 25)) result in little net change in the number of solute-excluding water molecules.

E. coli cells can survive under conditions of high external osmolarity (reviewed in Ref. 59). Among the adaptive responses of these cells to increases in external osmotic pressure are the intracellular accumulation of potassium and glutamate ions (60-65) and loss of water to the environment, resulting in decreased cytoplasmic volume (9). A consequence of increased salt concentration is a significant reduction in the stability of protein-DNA complexes (cf. Refs. 1, 66, and 67)). Despite this effect, gene regulatory proteins appear to function appropriately in osmotically stressed cells grown at high extracellular salt concentrations (64). To resolve this paradox, Record and co-workers (9) proposed that macromolecular crowding caused by the reduction of cytoplasmic volume favors regulatory protein-DNA interactions. The results presented here suggest that the reduction of cytoplasmic water activity (equivalent to elevated osmotic pressure) may further enhance the stability of protein-DNA assemblies.

To regulate gene expression, lac repressor partitions between regulatory and competing genomic DNA-binding sites (16, 20, 50). Differences in binding mechanism and affinity (cf. Refs. 46 and 51) suggest that the water stoichiometries of sequence-specific and nonspecific DNA binding reactions should differ.³ Our results support models in which a net release of water accompanies nonspecific DNA binding but that the number of water molecules released (_H2O = 165 ± 43) is smaller than that found for operator O1 binding (_H2O = 260 ± 32). Similar patterns have been reported for gal repressor (56) and for CAP protein (57). The differential sensitivity of these DNA binding reactions to changes in water activity raises the possibility that water release may play a role in regulating the specificity of protein-DNA interactions.⁴ A change in the distribution of gene-control proteins between regulatory and nonspecific genomic sites could provide bacteria with a direct means of sensing and responding to changes in environmental osmolarity.

	ACKNOWLEDGEMENTS

We thank Dr. Kathleen Matthews for providing lac repressor protein and Dr. Gang Liu for pMM02 DNA.

	FOOTNOTES

^* This work was supported in part by funds from the Pennsylvania State University Life Science Consortium.The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

^§ To whom correspondence should be addressed: Dept. of Biochemistry and Molecular Biology, Penn State University College of Medicine, P. O. Box, 850, Hershey, PA 17033. Tel.: 717-531-5250; Fax: 717-531-7072; E-mail: mfried@psu.edu.

Published, JBC Papers in Press, October 11, 2002, DOI 10.1074/jbc.M208540200

¹ For these analyses, both gel and running buffer were 20 mM Tris acetate, 2 mM EDTA, 250 mM KCl.

² The Protein Data Bank can be accessed at the following web address: www.rcsb.org/pdb.

³ In this context, it is worth noting that differences are even detectable in the interactions of the repressor with the closely related O_sym and O1 operators (23).

⁴ Whereas a reduction in intracellular water activity will stabilize all interactions with _H2O < 0, sequence-specific interactions will be stabilized to a greater degree if the number of waters released on specific binding exceeds that released on non-specific binding.

	REFERENCES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

1.	Record, M. T., Jr., Lohman, T. M., and deHaseth, P. L. (1976) J. Mol. Biol. 107, 145-158[CrossRef][Medline] [Order article via Infotrieve]
2.	Leirmo, S., Harrison, C., Cayley, D. S., Burgess, R. R., and Record, M. T. (1987) Biochemistry 26, 2095-2101[CrossRef][Medline] [Order article via Infotrieve]
3.	Record, M. T., and Mossing, M. C. (1987) in RNA Polymerase and the Regulation of Transcription (Reznikoff, W. S., ed) , pp. 61-83, Elsevier Science Publishing Co., Inc., New York
4.	Barkley, M., Lewis, P., and Sullivan, G. (1981) Biochemistry 20, 3842-3851[CrossRef][Medline] [Order article via Infotrieve]
5.	Overman, L. B., and Lohman, T. M. (1994) J. Mol. Biol. 236, 165-178[CrossRef][Medline] [Order article via Infotrieve]
6.	Foguel, D., and Silva, J. L. (1994) Proc. Natl. Acad. Sci. U. S. A. 91, 8244-8247[Abstract/Free Full Text]
7.	Royer, C. A. (1995) Methods Enzymol. 259, 357-377[Medline] [Order article via Infotrieve]
8.	Robinson, C. R., and Sligar, S. G. (1995) Methods Enzymol. 259, 395-426[Medline] [Order article via Infotrieve]
9.	Cayley, S., Lewis, B., Guttman, H., and Record, M. T. (1991) J. Mol. Biol. 222, 281-300[CrossRef][Medline] [Order article via Infotrieve]
10.	Zimmerman, S. B., and Minton, A. P. (1993) Annu. Rev. Biophys. Biomol. Struct. 22, 27-65[Medline] [Order article via Infotrieve]
11.	Timasheff, S. N. (1993) Annu. Rev. Biophys. Biomol. Struct. 22, 67-97[Medline] [Order article via Infotrieve]
12.	Parsegian, V. A., Rand, R. P., and Rau, D. C. (1995) Methods Enzymol. 259, 43-94[Medline] [Order article via Infotrieve]
13.	Parsegian, V. A., Rand, R. P., Fuller, N. L., and Rao, D. C. (1986) Methods Enzymol. 127, 400-416[Medline] [Order article via Infotrieve]
14.	Colombo, M. F., Rao, D. C., and Parsegian, V. A. (1992) Science 256, 655-659[Abstract/Free Full Text]
15.	Arakawa, T., and Timasheff, S. N. (1985) Biophys. J. 47, 411-414[Abstract/Free Full Text]
16.	Muller-Hill, B. (1996) The Lac Operon , Walter de Gruyter & Co., Berlin
17.	Straney, S. B., and Crothers, D. M. (1987) Cell 51, 699-707[CrossRef][Medline] [Order article via Infotrieve]
18.	Lee, J., and Goldfarb, A. (1991) Cell 66, 793-798[CrossRef][Medline] [Order article via Infotrieve]
19.	Schlax, P. J., Capp, M. W., and Record, M. T., Jr. (1995) J. Mol. Biol. 245, 331-350[CrossRef][Medline] [Order article via Infotrieve]
20.	von Hippel, P. H., Revzin, A., Gross, C. A., and Wang, A. C. (1974) Proc. Natl. Acad. Sci. U. S. A. 71, 4808-4812[Abstract/Free Full Text]
21.	von Hippel, P. H., Revzin, A., Gross, C. A., and Wang, A. C. (1975) in Protein-Ligand Interactions (Sund, H. , and Blauer, G., eds) , pp. 270-288, Walter de Gruyter & Co., Berlin
22.	Stickle, D. F., Vossen, K. M., Riley, D. A., and Fried, M. G. (1994) J. Theor. Biol. 168, 1-12[CrossRef][Medline] [Order article via Infotrieve]
23.	Bell, C. E., and Lewis, M. (2001) J. Mol. Biol. 312, 921-926[CrossRef][Medline] [Order article via Infotrieve]
24.	Zwieb, C., Kim, J., and Adhya, S. (1989) Genes Dev. 3, 606-611[Abstract/Free Full Text]
25.	Lewis, M., Chang, G., Horton, N. C., Kercher, M. A., Pace, H. C., Schumacher, M. Z., Brennan, R. G., and Lu, P. (1996) Science 271, 1247-1254[Abstract]
26.	Kalodimos, C. G., Boelens, R., and Kaptein, R. (2002) Nat. Struct. Biol. 9, 193-197[Medline] [Order article via Infotrieve]
27.	Fried, M. G., and Daugherty, M. A. (2001) J. Biol. Chem. 276, 11226-11229[Abstract/Free Full Text]
28.	Butler, A. P., Revzin, A., and von Hippel, P. H. (1977) Biochemistry 16, 4757-4765[CrossRef][Medline] [Order article via Infotrieve]
29.	Stickle, D. F., Liu, G., and Fried, M. G. (1994) Eur. J. Biochem. 226, 869-876[Medline] [Order article via Infotrieve]
30.	Fried, M. G., and Crothers, D. M. (1984) J. Mol. Biol. 172, 241-262[Medline] [Order article via Infotrieve]
31.	Maxam, A., and Gilbert, W. S. (1977) Proc. Natl. Acad. Sci. U. S. A. 74, 560-565[Abstract/Free Full Text]
32.	Fried, M. G., and Bloomfield, V. A. (1984) Biopolymers 23, 2141-2155[CrossRef][Medline] [Order article via Infotrieve]
33.	Fried, M. G., and Crothers, D. M. (1981) Nucleic Acids Res. 9, 6505-6525[Abstract/Free Full Text]
34.	Garner, M. M., and Revzin, A. (1981) Nucleic Acids Res. 9, 3047-3060[Abstract/Free Full Text]
35.	Hudson, J. M., and Fried, M. G. (1990) J. Mol. Biol. 214, 381-396[CrossRef][Medline] [Order article via Infotrieve]
36.	Fried, M. G. (1989) Electrophoresis 10, 366-376[CrossRef][Medline] [Order article via Infotrieve]
37.	Durschlag, H. (1986) in Thermodynamic Data for Biochemistry and Bio/Technology (Hinz, H.-J., ed) , pp. 45-128, Springer-Verlag, Berlin
38.	Broderson, R., Nielsen, F., Christiansen, J., and Andersen, K. (1987) Eur. J. Biochem. 169, 487-495[Medline] [Order article via Infotrieve]
39.	Sprague, E. D., Larrabee, C. E., and Halsall, H. B. (1980) Anal. Biochem. 101, 175-181[CrossRef][Medline] [Order article via Infotrieve]
40.	Wyman, J. (1964) Adv. Protein Chem. 19, 223-275[Medline] [Order article via Infotrieve]
41.	Vossen, K. M., Stickle, D. F., and Fried, M. G. (1996) J. Mol. Biol. 255, 44-54[CrossRef][Medline] [Order article via Infotrieve]
42.	Whitson, P. A., Olson, J. S., and Matthews, K. S. (1986) Biochemistry 25, 3852-3858[CrossRef][Medline] [Order article via Infotrieve]
43.	Moraitis, M. I., Xu, H., and Matthews, K. S. (2001) Biochemistry 40, 8109-8117[CrossRef][Medline] [Order article via Infotrieve]
44.	Mossing, M. C., and Record, M. T., Jr. (1985) J. Mol. Biol. 186, 295-305[CrossRef][Medline] [Order article via Infotrieve]
45.	Winter, R., and von Hippel, P. (1981) Biochemistry 20, 6948-6960[CrossRef][Medline] [Order article via Infotrieve]
46.	Ha, J.-H., Capp, M. W., Hohenwalter, M. D., Baskerville, M., and Record, M. T., Jr. (1992) J. Mol. Biol. 228, 252-264[CrossRef]