Multimeric Self-assembly Equilibria Involving the Histone-like Protein H-NS

The thermodynamic parameters affecting protein-protein multimeric self-assembly equilibria of the histone-like protein H-NS were quantified by “large zone” gel-permeation chromatography. The abundance of the different association states (monomer, dimer, and tetramer) were found to be strictly dependent on the monomeric concentration and affected by physical (temperature) and chemical (cations) parameters. On the basis of the results obtained in this study and the available structural information concerning this protein, a mechanism is proposed to explain the association behavior also in relation to the functional properties of the protein.

zation of the bacterial nucleoid.
In the present study, we have undertaken a quantitative thermodynamic analysis of the process of H-NS oligomerization using "large zone" gel permeation chromatography (17) of in vivo 35 S-labeled protein. Our approach allows quantification of the assembly equilibria to be determined down to the nanomolar concentration range. Our results show that oligomerization of the protein is strictly dependent on protein concentration (in the range of 10 Ϫ8 to 10 Ϫ6 M) and strongly affected by physical (temperature) and chemical (cation) effectors. From the results obtained, we propose a structural model for the H-NS association equilibria and discuss it in light of the in vivo functional role of this protein.

Preparation of H-NS and [ 35 S]H-NS-H-NS was hyperproduced in E.
coli pop3184 harboring pPLc11 carrying hns (12). Cells were grown at 30°C in low sulfate medium containing per liter: 11.2 g of K 2 HPO 4 , 3.0 g of KH 2 PO 4 , 1 g of NH 4 Cl, 0.5 g of NaCl, 0.05 g of MgSO 4 ⅐7H 2 O, 0.02 g of Mg-acetate, 0.2 mg of CaCl 2 , 0.6 mg of FeCl 3 , 5 mg of thymine, 5 mg of adenine, 5 mg of uracil, 0.02 g of each amino acid except for methionine, 1 mg of biotin, 1 mg of vitamin B 1 , 8 g of glucose, and 50 mg of ampicillin. When the culture reached A 500 ϭ 0.6, the cells were transferred to 42°C for 15 min to inactivate the temperature-sensitive repressor, and incubation was continued for 30 min at 37°C. When radioactive protein was prepared [ 35 S]sulfate (Amersham Pharmacia Biotech, 1000 Ci/mmol) was added to the cultures to a final concentration of 5-6 C/ml just before induction at 42°C. After an additional 30 min incubation at 37°C, the cells were harvested and H-NS was purified as described previously (1). The purity and homogeneity of the protein was checked by SDS-polyacrylamide electrophoresis (Fig. 1), and the concentration of the protein was determined as described (18).
"Large-zone" Gel-permeation Chromatography-Large-zone analytical gel-permeation chromatography was performed according to Beckett et al. (19) using a 1 ϫ 30 cm Sephadex G-100 column (C 10/40, Amersham Pharmacia Biotech) equipped with a thermostatic jacket. The column, kept at constant temperature (Ϯ 0.02 K) using an external bath, was equilibrated in 20 mM Tris-HCl, pH 7.2, containing 200 mM NaCl and loaded with 45 ml of H-NS in the equilibrium buffer at concentrations ranging between 10 Ϫ6 and 10 Ϫ8 M (see "Results") obtained by mixing a fixed amount of radiolabeled protein with variable amounts of non-labeled H-NS. The column was then eluted with the equilibrium buffer at a constant flow rate of 0.5 ml/min, controlled by a peristaltic pump connected to the effluent tubing. Fractions of 0.5 ml were collected directly into scintillation vials, mixed with 4 ml of scintillation fluid (LCS Safety Mixture, Baker Analyzed reagent), and counted in a ␤-scintillation counter. The column was calibrated using, as standards, a solution of 100 mM bovine serum albumin, and 100 mM horse heart myoglobin dissolved in 20 mM Tris-HCl, pH 7.2, containing 200 mM NaCl. The single elution profiles were determined spectrophotometrically at 280 nm.
Data Analysis-In the large-zone chromatography, the ␤-emission count for each fraction eluted at a volume V e represents the raw data. From this collected two-dimensional data array (cpm versus V e ), the centroid elution volumes V e were numerically determined, as shown in Fig. 2, and the difference between the two areas, above and below the leading boundary, has been used as a searching function. The algorithm recursively searches for the V e value at which the searching function is minimized, and the areas of interest are determined by trapezoidal integration (19,20). The functional relationship between the first moments (centroids) V e and the total protein concentration C T can be used to determine the stoichiometries and thermodynamic constants for the polysteric equilibria. In the simple case of a monomer-dimer-tetramer equilibrium, the centroid V e values can be expressed as follows, where V 1 , V 2 , and V 4 are the centroid volumes for the monomeric, dimeric, and tetrameric species, respectively; and M 1 , M 2 , and M 4 are the equilibrium molar concentrations of the oligomeric species.
The total protein concentration (as a monomer) can be related to the different assembled populations according to the mass-balance: C T ϭ M 1 ϩ 2 M 2 ϩ 4 M 4 . The dimerization and tetramerization equilibrium constants K 2 and K 4 are defined as: K 2 ϭ M 2 /((M 1 ) 2 ) and K 4 ϭ M 4 / ((M 1 ) 4 ). We can deduce the monomer equilibrium concentration M 1 at a given C T as the value satisfying the equality: The experimental data of the dissociation curve (V e versus C T ) can be analyzed using Equation 1, where the M 1 , M 2 , and M 4 values are replaced by the appropriate numerical solutions of Equation 2 found by bisection (19).
The temperature effect on the oligomerization process can be taken into account substituting the relevant equilibrium constants with their classical Taylor expansion counterparts around a reference temperature ( ϭ 293 K), where ϭ 1/T and o ϭ 1/ V e is defined as the elution volume where the difference is minimal between the areas above (A) and below (B) the leading boundary . This approach is coincident with a van't Hoff analysis, where ⌬H is temperature-dependent. The thermodynamic Gibbs energy ⌬G 2 , relative to the dimerization process, is defined as ⌬G 2 ϭ ϪRT ln K 2 , and the same is for ⌬G 4 referring to the tetramerization, ⌬G 4 ϭ ϪRT ln K 4 . Therefore, the stepwise quantity, relative to the dimer-tetramer association, is ⌬G 4 Ϫ 2 ⌬G 2 . From the equilibrium constants K 2 and K 4 , the fractions Y i (i ϭ 1, 2, and 4) for the monomeric, dimeric, and tetrameric species, respectively, can be deduced as follows:

RESULTS
Gel Permeation Chromatography-The thermodynamic characterization of the oligomerization process of the regulatory protein H-NS has been achieved in this study applying the "large-zone" gel permeation approach because it has been proved to be a valid and rigorous method to determine the fundamental thermodynamic properties of polysteric equilibria in a macromolecule (17,19). From the elution profile collected at different H-NS concentrations ( Fig. 3 shows a typical elution profile obtained with 1 mM H-NS), the centroid elution volume V e at each protein concentration was calculated according to Valdes and Ackers (17) (see Data Analysis under "Experimental Procedures"). When lower concentrations of protein were used, the centroid elution volume was shifted toward higher values, indicating a decrease in the average molecular mass and a dissociation of the molecular species present in solution. The H-NS dissociation curve obtained by plotting the largezone centroid elution volume (calculated as described above) as a function of the total molar concentration C T (as monomer) is presented in Fig. 4A. The data, analyzed according to a simple monomer-dimer-tetramer stoichiometric model (17), clearly indicates the presence of two main fractions corresponding to monomers and tetramers. It can be deduced, however, that a finite amount of dimers is also formed even though this state is scarcely populated at equilibrium.
Effect of Monovalent Cations-Because experiments shown in Fig. 4A were performed in the presence of 200 mM NaCl, the large-zone gel permeation experiments were repeated in the presence of different monovalent cations, to analyze the nature of the interactions involved in the oligomerization of H-NS. The dissociation curves (i.e. the centroid volumes measured at 10 protein concentrations in the presence of three different ions) presented in Fig. 4B clearly indicate that the nature of the monovalent cation has a significant influence on the self-association of H-NS. The effect of the monovalent cations on the equilibrium constants of oligomerization (see Table I) were then estimated from the analysis (see Data Analysis section) of the data presented in Fig. 4B. Taken together, these results indicate a dependence of the equilibrium oligomerization constants on the nature of the cation utilized, the ion effects being correlated to their hydration shell (Na ϩ Ͼ K ϩ Ͼ NH 4 ϩ ) following the Hofmeister cation series and suggesting the involvement of hydrophobic interactions in the oligomerization process. Because it is known that cations with increasing ionic radius enhance the entropy of the water molecules in the ion hydration shell (21), we further analyzed the effect of this parameter on the free energy difference associated with dimer and tetramer formation. The data presented in Fig. 5 indicate that the absolute value of the free energy change relative to the dimerization process (⌬G 2 ) increases with increase in the crystal radius (corresponding also to an entropy enhancement of the water molecules in the ion hydration shell). Furthermore, the absolute value of the free energy change associated with the stepwise formation of the tetramer (from the association of two dimers) ⌬G 4 Ϫ 2 ⌬G 2 appears to be negatively modulated by the ions in a nonmonotonic manner, suggesting that the interactions involved in tetramerization are different from, and less homogeneous, than those involved in dimerization. These results can be clearly deduced from the equilibrium distribution of the species in the presence of the three different cations which influence to different extents the formation of the dimers as well as of the tetramers (see Fig. 6). The amount of dimers formed increases regularly as the crystal radius of the cation increases from Na ϩ to K ϩ to NH 4 ϩ , whereas the formation of tetramers is fairly similar in the presence of Na ϩ and K ϩ and requires higher protein concentrations in the presence of NH 4 ϩ . Temperature Effect-To evaluate the influence of temperature on the equilibrium constants of H-NS oligomerization, the large zone gel permeation experiments were performed at different temperatures. As seen in Fig. 7, the centroid volumes show a strong temperature-dependence, decreasing with increasing temperature in the range 285-300 K. The values of the thermodynamic parameters ⌬H and ⌬Cp for the oligomerization processes obtained by non-linear regression analysis of the experimental data, according to standard van't Hoff formalism are presented in Table II. Because of the small set of temperatures investigated, however, the standard deviations associated with the best-fit values are quite high, and in particular the standard deviation associated with ⌬Cp is relatively large; therefore, no relevant significance can be assigned to the ⌬Cp values associated with the oligomerization phenomena. However, these data, as with those concerning the effect of

TABLE I Effect of monovalent cations on the oligomeric equilibrium constants of H-NS at 293 K
The free energies (expressed in kJ/mol of monomer) were calculated from the best-fit values for the equilibrium constants as deduced from the analysis of the experimental data.

Cations
Ϫ⌬ monovalent cations, can be used to estimate the distribution of the species present at the three different temperatures studied (Fig. 8, top, middle, and bottom).
As seen in the figure, while the formation of the dimeric species shows a maximum around room temperature (293 K), the formation of tetramers is linearly promoted by a temperature increase (Fig. 8). DISCUSSION Self-association equilibria among H-NS monomers have been observed by several authors and shown to be of crucial importance for the functional properties of this molecule (22,27). The present use of "large-zone" gel-permeation chromatography allowed us to measure quantitatively the multiple self-association equilibria characterizing H-NS. Our data indicate that at Ͼ10 Ϫ7 M H-NS, these equilibria are shifted toward the tetrameric form of the protein. Analysis of the oligomer dissociation curve reported in Fig. 4A seems to indicate that the H-NS multiple self-association equilibria is compatible with the presence of three stable molecular populations (i.e. monomers, dimers, and tetramers). However, the concentration of the dimeric species, though not vanishable, turns out much smaller with respect than that of the other two species. The disagreement between this finding and previous reports, which identified a substantial amount of H-NS dimers (14,28), is only apparent because neither one of the experimental approaches used in those studies (i.e. chemically induced cross-linking (1) and small-zone gel-permeation chromatography (14)) is suitable to give quantitative information on the equilibria. In particular, small zone gel-permeation chromatography is a nonequilibrium method in which dissociation of the protein oligomers is expected to take place as a consequence of the dilution of the sample during the chromatographic elution (17).
The study of the effect exerted by monovalent cations on the oligomerization equilibria seems to suggest a non-symmetrical quaternary organization of the H-NS tetramer that could stem from the different nature of the interactions occurring at the monomer-monomer interface with respect to those occurring at the dimer-dimer interface. In fact, it has been found that the increase in the ionic size of cations has a negative effect on the stepwise tetramerization equilibrium constant (⌬G 4 Ϫ 2⌬G 2 ) while having a positive influence on the formation of dimers (Fig. 5, and Table II).
The dependence of the dimerization equilibrium constant ⌬G 2 on the crystal radius (at the present experimental noise level) cannot be ruled out quantitatively, whereas data relative to the tetramerization constants (together with their standard deviation) suggest an inverse relationship between the (⌬G 4 Ϫ 2⌬G 2 ) values and ion size. It is known that cations not only affect water organization, but they also mask protein-ionizable groups in ion-ion pairs. In this respect, a perturbation of ion-ion pairing could be responsible for the tetramer dissociation under chaotropic conditions, whereas the presence of chaotropic cations should stabilize the monomer-monomer interactions.
These results appear to be in good agreement with a model based on the assumption that the occurrence of polar interactions contributes to the formation of a tetramer (from two dimers), whereas the dimer formation (from two monomers) is dominated by stronger hydrophobic effects.
On the other hand, the global effect of temperature on the quaternary structure of H-NS suggests a strong involvement of hydrophobic interactions in the overall oligomer assembly (23) Here we attempt to rationalize these data in light of the primary structure of the molecule. H-NS secondary structure predicted using PROSITE (25) is mainly characterized by ␣-helix stretches connected by non-structured regions around positions 38 and 70, and a ␤-strand stretch between the residues 93-108 (dark boxes in Fig. 9A).
Furthermore, the hydropathic pattern of H-NS (reported in Fig. 9A), calculated using one amino acid frame (a modification of the Kyte and Doolittle method (24)) shows an apparent hydrophobic-hydrophilic intercalating motif (amphipathic) which mainly and most clearly shows up in the first and in the last portions of the molecule. In addition, the hydropathic pattern predicts a fairly hydrophilic region in positions 100 -115, whereas the carboxyl end seems to be characterized by another amphipathic helix region. An ␣-helical wheel representation allows further insight on the distribution of hydrophobic and hydrophilic residues along the polypeptide chain. In fact, some sequence stretches predicted as ␣-helices (1-10, 19 -31, 50 -60,  115-125, 128 -133) show the typical amphipathic distribution of hydrophobic/hydrophilic residues, as already known for other proteins (23). As an example of this topological distribution, we show an ␣-helical wheel representation of residues 1-10 ( Fig. 9B1) and 19 -31 (Fig. 9B2) of H-NS. This findings are in good agreement with the structure of the C-terminal portion of the molecule as deduced by NMR data (13) which is reported in Fig. 10, highlighting with darker color the hydrophobic amino acids. As a whole, we suggest a role for the helix-helix pairing in the oligomer stabilization and, in particular, that dimer formation could depend on a homo-interaction between the regions (e.g. the N-terminal 40 residues) where the hydrophobic-hydrophilic intercalating motif appears to be more homogeneous. This hypothesis is also supported by mutagenesis studies showing that N-terminal-truncated forms of H-NS are not able to form dimers and lose the ability to function as transcriptional repressor (22).
The non-linear negative dependence of the tetramerization constant on chaotropic agents, which suggests an involvement of polar interaction in the formation and stabilization of tetramers could be tentatively explained assuming this interaction to be mediated by the region included between residues 50 and 90, which displays a less organized amphipathic pattern (see Fig. 9A) (24,26). The interplay between oligomerization and DNA-binding properties of H-NS has been pointed out by several authors (14,22), and the control exerted by different ions on the H-NS oligomerization process could be seen as a general heterotropic effect on the DNA recognition process, as already described for other DNA-binding proteins (25,27).
In any case, the concentration-dependent binding properties of H-NS could play a regulative feedback mechanism to maintain the protein concentration at sub-toxic levels (11), even under stress conditions.
In conclusion, the results presented here clearly indicate that H-NS undergoes self-association and that the oligomerization process is affected by the presence of different ions and by temperature. The last finding appears to be of relevant significance to rationalize the in vivo properties and behavior of H-NS, which is known to modulate gene expression in response to changes in environmental parameters. In particular, the increased synthesis of H-NS upon induction of cold-shock could fulfill the need of maintaining a constant supply of H-NS tetramers at low temperatures which would otherwise favor a shift in the oligomerization equilibria toward their dissociation. Another relevant aspect of temperature concerns the known H-NS temperature-dependent repression of virulence genes, such as Shigella virF. The repression of this gene has been attributed to a conformational transition of the DNA target of H-NS and not to modifications of the protein structure within the narrow temperature range (30 -35°C) in which virF transcription goes from being in an H-NS-repressed state to being fully de-repressed (29). In fact, most of the temperature effects on the oligomerization equilibria of H-NS (Fig. 7) occur below the critical temperature of 25-30°C, while above this temperature there are only marginal effects on the equilibria governing the quaternary structure of the protein.