Investigation of the Structural Basis for Thermostability of DNA-binding Protein HU from Bacillus stearothermophilus *

Site-directed mutagenesis was used to identify amino acid residues essential for the thermostability of the DNA-binding protein HU from the thermophile Bacillus stearothermophilus (BstHU). Two mutants,BstHU-A27S and BstHU-V42I, in which Ala27 and Val42 in BstHU were replaced by the corresponding amino acids Ser27 and Ile42, respectively, in the homologue from a mesophileB. subtilis (BsuHU), were less stable than the wild-type BstHU (63.9 °C), showing T m values of 58.4 °C and 60.1 °C, respectively, as estimated by circular dichroism (CD) analysis at pH 7.0. The denaturation of two mutants was further characterized using differential scanning calorimetry; the T m values obtained by calorimetric analysis were in good agreement with those estimated by CD analysis. The results suggest that Ala27 and Val42 are partly responsible for enhancing the thermostability of BstHU. When considered together with previous results, it is revealed that Gly15, Ala27, Glu34, Lys38, and Val42 are essential for the thermostability of thermophilic protein BstHU. Moreover, five thermostabilizing mutations were simultaneously introduced into BsuHU, which resulted in a quintuple mutant with a T m value of 71.3 °C, which is higher than that of BstHU, and also resulted in insusceptibility to proteinase digestion.

A thorough understanding of molecular rules governing thermostability of proteins should provide invaluable insight into basic principles for a design of novel proteins by protein engineering. Both theoretical and experimental approaches have been undertaken to examine the thermostability of proteins and many different structural principles have been postulated for increased thermostability (for reviews, see Refs. 1-3). One convincing approach is a comparative study on proteins that are available from organisms living under different temperature conditions. Recently, Vogt et al. statistically compared the amino acid sequences and tertiary structures of mesophilic and thermophilic organisms and suggested that the increased hydrogen bonds and ion pairs may provide the most general explanation for thermostability in proteins (4). We have pursued the investigation of the structural basis of thermostability of proteins using a histone-like bacterial DNA-binding protein HU (HU) 1 as a model protein.
The HU is a small basic polypeptide chain composed of 90 -92 amino acids and occurs as a homotypic dimer in solution (5). The HU binds to DNA in a sequence-independent manner and has been thought to play an important role in the structure of the bacterial nucleoid, being involved in replication (6), inversion (7), transposition (8), and repair (9) as a DNA chaperon (10). Bacillus stearothermophilus HU (BstHU) is composed of 90 amino acids, and its tertiary structure has been extensively studied by both x-ray crystallographic (11,12) and NMR spectroscopic (13,14) methods. The BstHU has two distinct halves; the N-terminal half consists of two ␣-helices (␣1 and ␣2), which are connected by a broad turn to create a V-shaped supersecondary structure (HTH motif), while the C-terminal half consists mainly of a three-stranded antiparallel ␤-pleated sheet. In a dimeric form, a pair of HTH motifs are entangled with each other to form a tightly packed hydrophobic core domain. In addition to BstHU, we have isolated three homologous HUs from mesophilic (Bacillus globigii and Bacillus subtilis) and thermophilic (Bacillus caldolyticus) bacilli, and their amino acids have been sequenced (15,16). On the basis of sequence comparison of the four HUs and the known three-dimensional structure of BstHU, the relative thermostability with respect to amino acid differences between the four proteins was discussed (16). This study revealed 11 amino acid substitutions between the thermophilic and mesophilic proteins, which are almost all restricted to the molecular surface not implicated in the mode of DNA binding. Thus, it could be expected that these amino acid substitutions might give rise to additional hydrogen bonds and/or salt bridges that would contribute to the thermostability of the thermophilic HU. To assess the contribution of the individual amino acids to the thermostability of BstHU, we constructed BstHU mutants, in which the amino acids were individually replaced with the corresponding amino acids in B. subtilis HU (BsuHU) and evaluated their thermostability. Previously, it has been shown that Gly 15 in the bend between two ␣-helices (␣1 and ␣2) and Glu 34 and Lys 38 occurred at the molecular surface, significantly contributing to the thermostability of BstHU (17,18).
To this end, we extend this comparative study to the remaining six amino acid replacements: BstHU Ala 27 to BsuHU Ser 27 , Ser 31 to Thr 31 , Val 42 to Ile 42 , Ala 56 to Ser 56 , Met 69 to Ile 69 , and Lys 90 to Ala 90 -Gly 91 -Lys 92 . In the crystal structure of BstHU ( Fig. 1), Ala 27 resides on the outside of the second ␣-helix (␣2) and is exposed to solvent. Since the Ala residue is a strong helix-forming amino acid, it could be suggested that Ala 27 in BstHU would contribute to thermostability by stabilizing the second ␣-helix (␣2). Ser 31 in BstHU is located on the solventfacing surface of the second ␣-helix, and its side chain forms a hydrogen bond to the main chain carbonyl of Ala 27 . Since an equivalent hydrogen bond is probably formed through the side chain of the Thr 27 residue in the mesophile BsuHU, the replacement from Ser 27 to Thr was thought to be neutral. The Val 42 in BstHU is in the ␤1-strand and is buried in the interior of the protein. It was therefore suggested that the larger side chain Ile in BsuHU would produce structural strain in the tightly packed core, resulting in destabilization of the mesophilic protein. Ala 56 and Met 69 in BstHU are substituted by Ser and Ile, respectively, in BsuHU. The former residue is on the outgoing ␤-ribbon closed to the disordered arms of the BstHU, and the latter is in the returning part of the disordered arm in BstHU. These residues are expected to be accessible to solvent and, therefore, were thought to have little effect on the thermostability of the protein. The C-terminal Lys 90 in BstHU is replaced by the tripeptide Ala 90 -Gly 91 -Lys 92 in BsuHU. It is shown that the C terminus is disposed on the molecular surface, and thus the significance of this replacement was unclear.
In the present study, we reveal that Ala 27 and Val 42 are essential for the thermostability of the BstHU. Furthermore, on the basis of the present and previous results, we construct a quintuple mutant HU using the BsuHU gene as a prototype and discuss its hyperthermostability and resistance to proteolytic degradations.
Purification and Characterization of Proteins-All procedures used for production, purification, and SDS-PAGE analysis of the recombinant proteins BstHU and BsuHU and the mutants thereof have been described previously (17,18). Thermostability of the protein was determined by monitoring the change in circular dichroism (CD) at 222 nm as a function of temperature, and thermodynamic parameters were obtained as described in Refs. 17 and 18. Since it was reported that the protein BsuHU showed the strong dependence of the CD spectral prop-erties and stability under a variety of conditions (20,21), we measured all proteins using the same procedure and the same type of spectrophotometer, as described in previous papers (17,18).
Differential Scanning Calorimetry (DSC)-Calorimetric measurements were carried out with a VP-DSC (MicroCal Inc., Northampton, MA) microcalorimeter with an personal computer. The scan rate was 1.0 K/min. Sample solutions for DSC measurements were prepared by dialyzing of HU proteins dissolved in water against 0.05 M phosphate buffer at pH 7.0 exhaustively. The protein concentrations were 22-33 M. The concentrations of the protein solutions were determined using amino acid analysis of proteins after acid hydrolysis. Data analysis was done using the Origin software (MicroCal).
NMR Spectra-1 H NMR spectra were recorded at 600 MHz with a Varian Unity Plus spectrometer. All NMR measurements were carried out at pD 7 and 25°C. Dioxane was employed as the internal standard (3.743 ppm). The pD values were the pH meter readings without adjustment for isotope effects.
Proteolysis-Tryptic and chymotryptic digestions of the proteins (1 mg/ml) were carried out at 37°C in 0.1 M Tris-HCl, pH 8.0, with an enzyme:substrate ratio of 1:1000 and 1:5000 (w/w), respectively. Proteolytic digestion was assayed by measuring the change in the amount of the uncleaved protein, which was separated by reverse-phase HPLC on a YMC-gel C4 column (4.6 ϫ 250 mm) equilibrated with 0.1% trifluoroacetic acid. The protein was eluted with a linear gradient of 0 -56% acetonitrile in 0.1% trifluoroacetic acid for 30 min. The effluents were monitored by absorption at 220 nm.

Overproduction and Characterization of Mutant Proteins-
Six BstHU mutant proteins, designated as BstHU-A27S, BstHU-S31T, BstHU-V42I, BstHU-A56S, BstHU-M69I, and BstHU-K90AGK, were engineered, in which Ala 27 , Ser 31 , Val 42 , Ala 56 , Met 69 , and Lys 90 were changed to the corresponding amino acid residues in BsuHU. Expression of the mutated cDNAs was performed in Escherichia coli BL21 (DE3) cells, using the T7 system as described previously (17,18). All mutant proteins were purified from the soluble fractions of cells so as to give a single band on SDS-PAGE. Their behaviors during the purification steps were almost identical with that of the wild type protein. The yields of protein from a 1-liter culture were 15-20 mg. The integrity of the mutant protein was confirmed by measurements of far-ultraviolet CD as described previously. The CD spectrum of each mutant was almost indistinguishable from that of the wild type, indicating that none of these mutations affected the backbone conformation (data not shown).
Thermostabilities of Mutant Proteins-To define the amino acid replacements responsible for the thermostability of BstHU, the thermostabilities of the mutant proteins were analyzed by monitoring the change in the CD value at 220 nm as a function of temperature. Fig. 2A shows the thermal denaturation curves of the wild type and the mutant proteins. In all cases, the unfolding transitions appeared to be monophasic, suggesting the absence of a folding intermediate in thermal denaturing process of global structure. Thus, on the basis of the assumption that the wild type BstHU and its mutants are denatured with a two-state model, thermodynamic parameters were calculated from the thermal denaturation curves, as summarized in Table I. Of five mutant proteins, the mutants BstHU-A27S and BstHU-V42I were less stable by Ϫ5.5°C (⌬⌬G ϭ Ϫ4.97 kJ/mol) and Ϫ3.8°C (⌬⌬G ϭ Ϫ3.48 kJ/mol), respectively, in T m than the wild type BstHU. This result suggests that Ala 27 and Val 42 are the key residues to provide the extra thermostability to BstHU.
The contributions of Ala 27 and Val 42 to the thermal stabilization of BstHU were corroborated by constructing the mesophilic mutants BsuHU-S27A and BsuHU-I42V, in which Ser 27 and Ile 42 in BsuHU were conversely replaced by Ala and Val, respectively, and the resulting mutants were characterized in terms of their thermostabilities. As shown in Fig. 2B and Table  I (11). Side chains are shown only for residues where the contribution to the thermostability of BstHU was examined in this study. It should be noted that the residues in the distal region of the arm between strands 2 and 3 (59 -70) were tentatively included in this figure. creased by 5.6°C (⌬⌬G ϭ 3.07 kJ/mol) and 4.0°C (⌬⌬G ϭ 2.19 kJ/mol), respectively, compared with that of the wild-type BsuHU, demonstrating that these two amino acids, Ala 27 and Val 42 , are both responsible for enhancing the thermostability of the BstHU.
In contrast, the stabilities of BstHU-A56S and BstHU-M69I were almost identical to that of the wild type, and the mutants BstHU-S31T and BstHU-K90AGK were slightly more stable than the wild type. Two residues Ala 56 and Met 69 are located in the solvent-exposed arm region. In the crystal structure of BstHU, the top part of the arm is not visible because of its flexible nature. It was therefore concluded that these substitutions in a mobile, solvent-exposed environment do not affect the conformational stability. The T m values of the mutants BstHU-S31T and BstHU-K90AGK were higher by 1.9°C and 1.8°C than that of the wild type BstHU. The crystal structure of the BstHU shows that Ser 31 resides on the solvent-facing surface, clustering together with Thr 13 and Ala 27 , and forms a hydrogen bond to the main chain carbonyl of Ala 27 . Therefore, the increased stability of BstHU-S31T might be due to the rearrangement of a hydrogen bond formed between the side chain oxygen of the substituted Thr 31 and the main chain carbonyl of Ala 27 . Another plausible interpretation is that the introduction of the larger side chain might give some preferential van der Waals contacts with Thr 13 and/or Ala 27 . The mutant BstHU-K90AGK was also stable as compared with the wild type. At present, no explanation for thermostable property of this mutant has been obtained.
In order to understand the molecular basis for destabilization of the mutant proteins, BstHU-A27S and BstHU-V42I, their structural features were examined by 1 H NMR spectroscopic analysis. As a result, the chemical shift change of some resonances were observed around aromatic regions at 6.2, 6.26, and 7.0 ppm in BstHU-A27S and 6.2, 6.28, and 6.49, and 7.0 ppm in BstHU-V42I, as compared with that of the wild-type HU (data not shown). Although the assignment of these resonances in the BstHU has still not been reported, Kakuta classified these to be derived from Phe residues (Phe 29 , Phe 47 , Phe 50 , and Phe 79 ) in BstHU by DQF-COSY. 2 It is therefore likely that the replacement of Ala 27 or Val 42 by Ser or Ile, respectively, may cause rearrangement of some Phe residues occupying the interior of the molecule.

DSC Measurements of BstHU and Its Mutant Proteins-
The present study, together with previous studies (17,18), suggests that five amino acid residues, Gly 15 , Ala 27 , Glu 34 , Lys 38 , and Val 42 , are essential for thermostability of BstHU. To corroborate their involvement in thermostability, the thermal denaturations of five mutants (BstHU-G15E, BstHU-A27S, BstHU-E34D, BstHU-K38N, and BstHU-V42I) as well as the wild type were further characterized by DSC measurements. A typical excess heat capacity curve of the wild type BstHU is shown in Fig. 3. The examined proteins, as is the case for the BstHU, gave single peaks in calorimetric measurements; the T m values of the wild type and its five mutants could be calculated from these curves, as given in Table II  ment of five amino acid residues in the thermostability of BstHU. Interestingly, the T m values obtained by DSC measurement for the wild type and its mutant proteins were well coincident with those estimated by CD analysis. This observation, together with a single peak in calorimetric measurements, strongly suggests that BstHU as well as its mutants are denatured with a two-state model involving the native dimer and unfolded monomers.
Contribution of Thermostabilizing Mutations-In the foregoing work, it was shown that the combination of the three thermostabilizing substitutions (Glu 15 in BsuHU to Gly, Asp 34 to Glu, and Asn 38 to Lys) significantly increased the thermostability of the mesophilic protein BsuHU by 15.9°C in T m (18). Taking the present result, the five thermostabilizing mutations (Glu 15 to Gly, Ser 27 to Ala, Asp 34 to Glu, Asn 38 to Lys, and Ile 42 to Val) were simultaneously introduced into the mesophilic protein BsuHU, and the thermostability of the resultant mutant (quintuple mutant) was examined, exactly in the same manner as described above. The thermal denaturation curves of the wild type BsuHU and the quintuple mutant as well as those of the five single mutant proteins with the constituent amino acid substitutions are shown in Fig. 4. The parameters characterizing the thermal denaturation are summarized in Table III. The simultaneous introduction of the five mutations greatly increased the thermostability of the protein with the T m value of 71.3°C compared with 48.6°C for the wild-type BsuHU. The stabilization energy arising from the quintuple mutations seems to be somewhat greater than the sum of the constituent single substitutions but only by about 8.9% of the calculated value. This result indicated that the effects of the mutations on the thermostability are roughly independent of each other and nearly additive. This result suggests that the individual local reinforcements derived from the amino acid residues that reside far from each other contribute to the greater thermostability of the thermophilic protein BstHU. Interestingly, the T m value of the quintuple mutant of BsuHU was higher than that of BstHU. As shown in Table I, the mutation of amino acids in BstHU to those of BsuHU such as BstHU-S31T was not always less stable, indicating that the amino acid sequence of BstHU was not optimized for thermal stability. Considering this idea, the higher stability of the quintuple mutant of BsuHU than BstHU may be reasonable.
Susceptibility to Proteolytic Degradation-As reported on the tryptophan synthase ␣-subunit (22), kanamycin nucleotidyltransferase (23), and RNase HI (24), the correlation was observed between the thermostability and the resistance to proteolytic degradation. To examine whether or not this rule is applicable to the protein HU, we analyzed the susceptibility of the quintuple mutant and wild-type BsuHU to chymotrypsin and trypsin, as described under "Experimental Procedures." As shown in Fig. 5, the rates of chymotryptic and tryptic degra-dations of the quintuple mutant were 200 and 11 times slower than that of the wild type BsuHU, respectively. Taking into account the substrate specificity of chymotrypsin and trypsin, the resistance of the quintuple mutant to proteolytic degradation seems to be independent of the substrate specificity of the protease. Imoto et al. (25) reported that digestion by proteases proceeds mainly via the unfolded state of proteins. The stabilization of the quintuple mutant of BsuHU may depend on the stabilization of the folded state rather than the destabilization of the unfolded state, because mutations were involved in the formation of salt bridge or the introduction of favorable intermolecular interaction. Therefore, it is likely that the thermostabilizing mutations contribute to the decrease in the rate constant of unfolding.
Conclusion-This series of studies on the thermostability of BstHU revealed that its extra thermostability relative to the mesophilic protein BsuHU seems to be achieved mainly by stabilization of two ␣-helices (␣1 and ␣2) with Gly 15 , Ala 27 , and Glu 34 and by improvement of the intermolecularly close packing in the hydrophobic core with Ala 27 and Val 42 (Table IV). Moreover, the additional salt bridge between Glu 34 and Lys 38 on the hydrophobic surface is found to be also responsible for enhancing the thermostability of BstHU (Table IV). The amino acid residues found to be responsible for thermostability of BstHU are mapped in the crystal structure of BstHU (Fig. 6).
The Gly 15 residue locates on the bend between ␣1and ␣2helices and is found to stabilize the conformation of the HTH motif. The replacement of Gly 15 by the corresponding residue Glu in BsuHU caused a remarkable decrease in the thermostability of the BstHU-G15E: ⌬T m ϭ Ϫ9.9°C by CD analysis (17) and ⌬T m ϭ Ϫ10.1°C by DSC measurement (Table II). Further, Ala 27 and Glu 34 locate at the second ␣-helix (␣2) and may be involved in stabilization of its conformation because of their enhanced ␣-helix structural propensity. Two ␣-helices (␣1 and   ␣2) in BstHU adopt the HTH motif, which is structurally similar to that found in the operator/repressor family of DNAbinding proteins, such as CAP and -cro, where the motif is directly involved in DNA binding. Thus, the study on the thermostability of BstHU indicates the unique role of the HTH motif in the BstHU. The topology found in BstHU, where the helical core domain consists of a tightly entangled pair of the HTH motif from two protomers, is unique, and no similar structure has been found. However, the recent tertiary structural study of the B. stearothermophilus ribosomal protein S7 (a primary 16 S rRNA-binding protein) revealed a similar motif in its N-terminal half region (26). Although a function of the HTH motif found in S7 is still not known, it is likely that the HTH motif may have been selected for stabilization of some nucleic acid-binding proteins as a scaffold. Glu 34 , in addition to involvement in the stabilization of the second ␣-helix, is also found to be responsible for enhancing the thermostability of BstHU by forming an extra salt bridge with Lys 38 . The replacement of Glu 34 by the corresponding residue Asp 34 in BsuHU caused a decrease in the thermostability of the BstHU-E34D: ⌬T m ϭ Ϫ2.3°C by both CD analysis (18) and DSC measurement. The contribution of a salt bridge to the thermostability of proteins is still controversial. It has been reported that the engineered electrostatic interaction between pairs of mobile, solvent-exposed charged residues on the molecular surfaces of proteins contributes little to protein stability (27)(28)(29)(30). In contrast, Vogt et al. (4) reported that the salt bridge, together with the hydrogen bond, is the main explanation for the thermostability of proteins. In the crystal structure of the BstHU (Fig. 1), the side chains of these two residues are also exposed to the solvent and are mobile, the average B values for atoms within the side chains being about 100 A 2 (11,12). However, as described above, Glu 34 may stabilize the second ␣-helix by its intrinsic ␣-helix forming property. Since Glu 34 is the salt bridge partner of the Lys 38 residue, the salt bridge formed between these two residues may somewhat stabilize the conformation of the second ␣-helix (␣2), thereby contributing to the thermostability of the BstHU. This series of studies has shown that some replacements of the amino acid residues in the BstHU by the corresponding amino acids in the BsuHU, such as Thr 13 to Ala, Ser 31 to Ala, and Thr 33 to Leu, resulted in thermostabilized proteins; the resulting mutants were more stable than the thermophilic parent protein BstHU (17). In this regard, it is suggested that the thermophilic protein BstHU has not evolved to optimize the protein structure in terms of thermostability. As a result, the quintuple mesophilic mutant containing five thermostable mutations has a T m that is 7.4°C higher than the thermophilic protein BstHU. It is generally known that many proteins have been selected in evolution to be marginally stable. It is thus assumed that B. stearothermophilus, as is the case for other organisms, has selected the present form of the BstHU to coordinate the normal living temperature. This finding strongly supports an idea that a novel protein with hyperthermostability can be constructed by protein engineering using existing proteins in nature as prototypes. FIG. 5. Susceptibility of the quintuple mutant and the wild type to proteolysis by chymotrypsin and trypsin. The fraction of uncleaved protein is shown as a function of incubation time. f, BsuHU; ࡗ, quintuple. The amount of uncleaved protein was determined by subjecting the digests to reverse phase HPLC, by which the uncleaved protein could be separated from the chymotryptic and tryptic peptides. Rate constants were determined from the slope of semilog plots of the fraction uncleaved versus the incubation time, and then relative k values (0.005 and 0.091) were calculated for chymotrypsin and trypsin, respectively, by dividing the k values of the quintuple mutant by that of the wild type BsuHU.  (11), and side chains are shown only for residues that are found to be essential for the thermostability of BstHU.