Conformational Stabilities of Escherichia coli RNase HI Variants with a Series of Amino Acid Substitutions at a Cavity within the Hydrophobic Core*

Escherichia coli ribonuclease HI has a cavity within the hydrophobic core. Two core residues, Ala52 and Val74, resided at both ends of this cavity. We have constructed a series of single mutant proteins at Ala52, and double mutant proteins, in which Ala52 was replaced by Gly, Val, Ile, Leu, or Phe, and Val74 was replaced by Ala or Leu. All of these mutant proteins, except for A52W, A52R, and A52G/V74A, were overproduced and purified. Measurement of the thermal denaturations of the proteins at pH 3.2 by CD suggests that the cavity is large enough to accommodate three methyl or methylene groups without creating serious strains. A correlation was observed between the protein stability and the hydrophobicity of the substituted residue. As a result, a number of the mutant proteins were more stable than the wild-type protein. The stabilities of the mutant proteins with charged or extremely bulky residues at the cavity were lower than those expected from the hydrophobicities of the substituted residues, suggesting that considerable strains are created at the mutation sites in these mutant proteins. However, examination of the far- and near-UV CD spectra and the enzymatic activities suggest that all of the mutant proteins have structures similar to that of the wild-type protein. These results suggest that the cavity in the hydrophobic core of E. coliRNase HI is conformationally fairly stable. This may be the reason why the cavity-filling mutations effectively increase the thermal stability of this protein.

In the interiors of globular proteins, and especially in the protein cores, hydrophobic side chains are generally well packed. Since the formation of a protein core governs the entire folding process of a protein, and because general methods for designing amino acid sequences that fold into the desired core structures have not been established, efforts have been made to evaluate the role of core residues in the stability and function of natural proteins by site-directed mutagenesis , random mutagenesis (22)(23)(24), and theoretical approaches (25)(26)(27)(28)(29). The data thus far accumulated suggest that the following phenomena are characteristic to the core mutations. 1) Mutations do not completely abolish the protein functions, unless the hydrophobicity at the core is considerably changed. This means that the hydrophobicity is a sufficient criterion for the formation of a functional core. 2) Mutations to smaller hydrophobic residues usually create cavities, which are either vacant (9,19,21) or occupied by a water molecule (15). The protein stability decreases as the volume of such a cavity increases. In this case, the cost for the loss of a single methylene group has been reported to be 1.3 Ϯ 0.5 kcal/mol (30). 3) Mutations to much larger hydrophobic residues or those to polar residues seriously affect the protein structures due to unfavorable interactions in the cores, and thereby almost completely abolish the protein functions. These mutations also dramatically destabilize the proteins.
Although the packing of the hydrophobic side chains is highly efficient in the interiors of globular proteins, it is not perfect, and cavities exist in virtually all large proteins, even within the protein cores (31)(32)(33)(34). These cavities may be required to make the proteins highly functional, probably at the cost of reduced stability, because they serve to reduce the strains caused by dense packing of the hydrophobic side chains, and to provide conformational flexibility to proteins. Therefore, it seems likely that the tolerance of proteins to core mutations increases as the volume of the naturally existing cavity at the core increases.
Escherichia coli RNase HI, 1 which hydrolyzes only the RNA strand of RNA/DNA hybrids, is a small globular protein with 155 amino acid residues. We have used this protein for studies of protein stability (14,(35)(36)(37)(38)(39)(40)(41)(42)(43) and folding (44,45). This protein is suitable for these studies for the following reasons: (i) highly refined coordinates of this protein are available (46), (ii) an overproduction system for this protein is available (36), and (iii) this protein reversibly unfolds in a single cooperative fashion with thermal and chemical denaturations (38).
In the hydrophobic core of this protein, a cavity exists, which is not occupied by water molecules. We have previously shown that cavity-filling mutations, such as Val 74 3 Leu and Ile, enhanced the protein stability by roughly 1 kcal/mol in ⌬G (14). We also found by random mutagenesis experiments that the mutation of Ala 52 3 Val also enhanced the protein stability by 1.7 kcal/mol, probably due to the reduction in the cavity volume around the mutation site (42). These results allowed us to propose that the replacement of a hydrophobic residue facing the cavity with a bulkier and more hydrophobic residue is a general method to increase protein stability. However, proteins are not always stabilized by this method. Karpusas et al. (1) found two large cavities in T4 lysozyme and designed two mutations (Leu 133 3 Phe and Ala 129 3 Val) to fill them. Eijsink et al. (8) introduced various mutations to fill the cavities in the neutral protease from Bacillus stearothermophilus. None of these mutations resulted in a significant increase in the protein stability. It was shown for the T4 lysozyme mutants that the hydrophobic effects gained by the mutation were canceled by the strains and van der Waals contacts. Determination of the crystal structures of the RNase HI mutants revealed that the conformation of the cavity was not seriously altered by the mutation, except for the mutation site, and the cavity volume around the mutation site was remarkably reduced (14). Therefore, it seems likely that the conformation of the cavity in the hydrophobic core of E. coli RNase HI is rather rigid and stable. The difference in the conformational stability of the cavity may affect the dependence of the protein stability on the cavityfilling mutation. To examine whether E. coli RNase HI is tolerant to mutations in which polar or extremely bulky side chains are introduced into the cavity, we have constructed a series of single and double mutant proteins at Ala 52 and Val 74 , and have analyzed their conformational stabilities and enzymatic activities.
Here we report that the size of the cavity in the hydrophobic core of E. coli RNase HI is large enough to introduce three methyl or methylene groups without causing significant steric hindrance. The thermal stabilities of a series of the mutant proteins suggest that this cavity is more effectively filled by the double mutations (Ala 52 3 Val and Val 74 3 Leu) than by the single mutation (Ala 52 3 Ile). We also report that this cavity seems to be conformationally fairly stable and the protein is tolerant to mutations that seriously affect the hydrophobicity and the packing of the cavity residues. Thus, the structure of this protein was not severely damaged when the cavity residues were replaced by either aromatic or charged residues, although these mutations greatly reduced both the protein stability and the enzymatic activity.

EXPERIMENTAL PROCEDURES
Materials-The wild-type E. coli RNase HI protein (36) and the mutant protein A52V (42), in which Ala 52 was replaced by Val, were prepared previously. Restriction enzymes and modifying enzymes for recombinant DNA technology were from Takara Shuzo Co., Ltd. Other chemicals were of reagent grade.
Mutagenesis-Alteration of the rnhA gene was carried out by sitedirected mutagenesis using polymerase chain reaction, as described previously (36). The DNA oligomers used as the 5Ј-and 3Ј-mutagenic primers (30 -40 bases long) were synthesized by Sawady Technology Co., Ltd. These DNA oligomers were designed such that the codon for Ala 52 was changed from GCT to ATC for Ile, CTG for Leu, TGT for Cys, ATG for Met, TTC for Phe, ACT for Thr, CAG for Gln, GAA for Glu, CCG for Pro, TCT for Ser, AAC for Asn, GAT for Asp, TAC for Tyr, GGT for Gly, CAT for His, AAA for Lys, TGG for Trp, and CGT for Arg. For the construction of the single mutant proteins at position 52, the wildtype rnhA gene in the plasmid pJAL600 was used as a template. For the construction of the double mutant proteins at positions 52 and 74, the mutant rnhA gene in either plasmid pJAL74A or pJAL74L was used as a template. The resultant plasmids for the overproduction of the mutant proteins were designated as either pJAL52X or pJAL52X74XЈ, in which X and XЈ represent the amino acid residues (one-letter notation) substituted for Ala 52 and Val 74 , respectively. Likewise, the mutant proteins were designated as either A52X or A52X/V74XЈ. All of the nucleotide sequences of the mutant rnhA genes were confirmed by the dideoxy chain termination method (48).
Overproduction and Purification-The mutant proteins were overproduced in E. coli HB101 cells harboring the pJAL600 plasmid deriv-atives by raising the cultivation temperature from 30 to 42°C, and were purified to homogeneity, as described previously (36). The protein concentration was determined from the UV absorption, assuming that the mutant proteins obtained in this experiment have the same A 280 0.1% value of 2.0 as that of the wild-type protein (49). The production levels of the mutant proteins in the cells were estimated by subjecting whole cell lysates to SDS-polyacrylamide gel electrophoresis on a 15% polyacrylamide gel (50), followed by staining with Coomassie Brilliant Blue.
RNase H Activity-The enzymatic activity was determined at 30°C and pH 8.0 in the presence of 10 mM MgCl 2 , 50 mM NaCl, 1 mM 2-mercaptoethanol, and 100 g/ml bovine serum albumin, by measuring the radioactivity of the acid-soluble digestion product from a 3 H-labeled M13 RNA/DNA hybrid, as described previously (35). One unit of enzymatic activity is defined as the amount of enzyme producing 1 mol of acid-soluble material/min. For the kinetic analyses, the substrate concentration was varied from 0.1 to 1.6 M (nucleotide phosphate concentration). The hydrolysis of the substrate with the enzyme follows Michaelis-Menten kinetics, and the kinetic parameters, K m and V max , were determined from the Lineweaver-Burk plot.
Circular Dichroism Spectra-The CD spectra were measured on a J-720 automatic spectropolarimeter (Japan Spectroscopic Co., Ltd.) at 25°C in 10 mM glycine-HCl (pH 3.2). For the measurement of the far-ultraviolet (UV) CD spectra (200 -260 nm), the protein concentration was approximately 0.2 mg/ml, and a cell with an optical path length of 2 mm was used. For the measurement of the near-UV CD spectra (250 -320 nm), the protein concentration was 0.5-1.0 mg/ml and a cell with an optical path length of 10 mm was used. The mean residue ellipticity (, deg⅐cm 2 ⅐dmol Ϫ1 ) was calculated by using an average amino acid molecular weight of 110.
Thermal Denaturation-Thermal denaturation curves and the temperature of the midpoint of the transition (T m ) were determined at pH 3.2 by monitoring the change in the CD value at 220 nm, as described previously (38). Proteins were dissolved in 10 mM glycine-HCl (pH 3.2). The enthalpy change of unfolding at the T m (⌬H m ) was calculated by van't Hoff analysis. The difference in the free energy change of unfolding between the mutant and wild-type proteins at the T m of the wildtype protein (⌬⌬G m ) was estimated by the relationship given by Becktel and Schellman (51), ⌬⌬G m ϭ ⌬T m ⅐⌬S m . ⌬T m is the change in T m of a mutant protein relative to that of the wild-type protein. ⌬S m is the entropy change of the wild-type protein at the T m . We used the ⌬S m value of 0.304 kcal/(mol⅐K), which was previously determined at pH 3.0 (38), for the calculation of the ⌬⌬G m values at pH 3.2. 52 and Val 74 , which face the hydrophobic cavity of E. coli RNase HI, are located in the ␣I and ␣II helices, respectively (Fig. 1). In addition to these residues, Thr 9 , Leu 49 , Leu 56 , Leu 67 , Leu 107 , and Trp 118 form the cavity. Of them, Thr 9 is located in the ␤A strand, Leu 49 and Leu 56 are located in the ␣I helix, Leu 67 is located in the ␤D strand, Leu 107 is located in the ␣IV helix, and Trp 118 is located in the ␤E strand. All of these residues are almost fully buried inside the protein molecule. The hydroxyl group of Thr 9 does not face the cavity but forms the hydrogen bond with the hydroxyl group of Thr 69 (46). For a comprehensive analysis of the effect of the mutation at the cavity on the protein stability, we have constructed a series of single mutant proteins, in which Ala 52 is replaced by the 19 other amino acid residues. Position 52 was chosen as the site for the introduction of the series of mutations, because Ala 52 is the smallest residue among those forming the cavity and therefore it is possible to introduce into the cavity a variety of side chains, that greatly differ in size and hydrophobicity. In addition, we have constructed 10 double mutant proteins, in which Ala 52 is replaced by Gly, Val, Leu, Ile, or Phe and Val 74 is replaced by Ala or Leu. These mutant proteins were constructed to estimate the number of methylene groups that could be introduced into the cavity without creating serious strains, and to examine whether effects other than hydrophobicity, such as packing, at the cavity contribute to the protein stability.

Mutant Constructions-Ala
Overproduction and Purification-All of the single mutant proteins, except for A52W and A52R, were overproduced and purified in an amount sufficient for biochemical characterizations. The mutant proteins A52W and A52R could not be purified, because of the extremely low production levels in cells (data not shown). The mutations of Ala 52 3 Trp and Arg may dramatically destabilize the protein and thereby increase the susceptibility to proteolytic degradation, probably because these residues are too large to fill the cavity without causing significant steric hindrance. The ionized group in the Arg side chain may also contribute to altering the protein conformation, because it is unlikely that an ionized group remains alone without a hydrogen-bonding or ion-pair partner within the hydrophobic core of a protein. In fact, the cellular production levels of the mutant proteins A52D, A52E, A52K, A52Y, and A52F, in which Ala 52 is replaced by ionic or aromatic residues, were considerably lower than that of the wild-type protein (data not shown). In contrast, the cellular production levels of the other mutant proteins were similar to that of the wild-type protein. Consequently, the yields of the mutant proteins A52D, A52E, A52K, A52Y, and A52F from 1-liter cultures were 3-10 mg, and those of the other mutant proteins were 27-62 mg (Table I). Likewise, of the 10 double mutant proteins, only the mutant protein A52G/V74A could not be purified, because of the extremely low production level in cells (data not shown). The simultaneous introduction of the two cavity-creating mutations probably dramatically destabilizes the protein.
Stabilities and Activities of Single Mutant Proteins-The far-UV CD spectra of all of the single mutant proteins are basically the same as that of the wild-type protein (Fig. 2a). In contrast, the near-UV CD spectra of these mutant proteins differed from one another. They were roughly classified into three groups (types A-C), based on the CD values at 260 -280 nm (Fig. 2b). The wild-type protein and the mutant proteins A52C, A52P, and A52G gave the type A spectrum. In the type B and type C spectra, the CD values at 260 -280 nm increased as compared with those in the type A spectrum. The extent of this increase is relatively small for the type B spectrum and is relatively large for the type C spectrum. The mutant proteins A52I, A52L, A52V, A52N, A52D, A52Q, A52E, A52S, and A52T gave the type B spectrum. The mutant proteins A52F, A52Y, A52M, A52H, and A52K gave the type C spectrum. Thus, the extent of the increase in the CD values at 260 -280 nm seems to be correlated with the volume of the side chain of the replaced residue. Nevertheless, the shape of the near-UV CD spectrum of the protein, especially that around 290 nm, is not markedly changed by the mutations. These results suggest that the mutations of Ala 52 cause a local conformational change, but only to a small extent, even when an aromatic or ionized group is introduced into the cavity by the mutation.
The thermal denaturation curves of the wild-type and mutant proteins are shown in Fig. 3. All of the mutant proteins reversibly unfolded in a single cooperative manner with thermal denaturation. The parameters characterizing the thermal denaturation of the mutant proteins, which were determined based on the assumption that these proteins unfold in a twostate mechanism, are summarized in Table I  single mutants Yield represents the amount of the protein purified from 1-liter culture. The enzymatic activity was determined at 30°C for 15 min in 10 mM Tris-HCl (pH 8.0) containing 10 mM MgCl 2 , 50 mM NaCl, 1 mM 2-mercaptoethanol, and 10 g/ml bovine serum albumin, by using an M13-RNA/DNA hybrid as a substrate. Errors, which represent the 67% confidence limits, are within 30% of the values reported. Thermal denaturation curves of the mutant proteins were measured at pH 3.2, as described under "Experimental Procedures." ⌬T m is the change in the melting temperature, T m , relative to that of the wild-type protein, which is 53.0°C. ⌬H m is the enthalpy change of unfolding at T m calculated by van't Hoff analysis. The change in the free energy of unfolding of the mutant protein relative to that of the wild-type protein (⌬⌬G m ) was estimated by the relationship given by Becktel and Shellman (51): ⌬⌬G m ϭ ⌬T m ⅐ ⌬S m (wild-type). The ⌬S m (wild-type) value of 0.304 kcal/(mol ⅐ K), which was previously determined at pH 3.0 (38) proteins, those in which Ala 52 is replaced by more hydrophobic aliphatic or sulfur-containing residues were more stable than the wild-type protein by 0.5-1.9 kcal/mol in ⌬G. In contrast, the other mutant proteins were less stable than the wild-type protein by 0.5-5.9 kcal/mol. We have analyzed the thermal stabilities of these mutant proteins at pH 5.5 in the presence of 1 M guanidine hydrochloride, conditions under which the wild-type protein reversibly unfolds (38). However, except for the apparent T m values, the thermodynamic parameters could not be determined, because the thermal denaturations of all mutant proteins were not reversible under these conditions. The T m values of the mutant proteins relative to that of the wild-type protein (⌬T m ) at pH 5.5 were comparable with those determined at pH 3.2, except for those of the mutant proteins A52D and A52E (data not shown). The stabilities of these mutant proteins relative to that of the wild-type protein at pH 5.5 were much lower than those at pH 3.2, probably because Asp and Glu are not ionized at pH 3.2, but are at least partially ionized at pH 5.5. Because the mutant proteins unfold irreversibly at pH 5.5, only the thermodynamic values at pH 3.2, at which all the mutant proteins unfold reversibly, are discussed in this report. The enzymatic activities of the mutant proteins varied from 0.1 to 112% of that of the wild-type protein (Table I). To determine whether the mutation affects the catalytic efficiency or substrate binding, the kinetic parameters were determined for some of the mutant proteins with poor enzymatic activities. The results are summarized in Table II. The K m values of these mutant proteins increased only by at most 2.5-fold, whereas their V max values decreased by 12-44-fold, as compared with those of the wild-type protein, suggesting that the mutation does not seriously affect substrate binding, but affects the catalytic efficiency. The conformations of the active-site residues may be altered by the mutations, so that the catalytic efficiencies of the enzymes are considerably decreased.
Stabilities of Double Mutant Proteins-The cavity volumes of the mutant proteins V74A and V74L are larger and smaller than that of the wild-type protein, respectively (14). Accordingly, the mutant protein V74A must be much more tolerant of mutations at Ala 52 to bulkier hydrophobic residues than the V74L protein. The conformational changes of the proteins caused by the mutations at Ala 52 were analyzed by CD. Comparison of the near-UV CD spectra of the double mutant proteins with that of the parent single mutant protein (V74A or V74L) revealed that the changes in the CD spectra by the mutations at position 52 are much smaller for V74A than for V74L (Fig. 4). These results suggest that the tolerance of the protein conformation to the cavity mutations increases as the volume of the cavity increases. The enzymatic activities are summarized in Table III. The enzymatic activities of the double mutant proteins A52X/V74L were always lower than those of the corresponding mutants A52X/V74A. In addition, some of the double mutant proteins A52X/V74A, such as A52I/V74A and A52F/V74A, were more active than the corresponding single mutant proteins at position 52. Therefore, the increase in the cavity volume apparently contributes to reducing the strains caused by the cavity filling mutations that are unfavorable for the enzymatic activity.
The parameters characterizing the thermal denaturation of the double mutant proteins are also summarized in Table III. Among these parameters, ⌬⌬G m Ј, instead of ⌬⌬G m , reflects the effect of the mutation at Ala 52 on the stability of the mutant protein V74XЈ, and ⌬⌬⌬G m , which is calculated as ⌬⌬G m Ј(A52X/V74XЈ) Ϫ ⌬⌬G m (A52X), reflects the difference between the effect of the mutation at Ala 52 on the stability of the wild-type protein (⌬⌬G m (A52X)) and that on the stability of the mutant protein V74XЈ (⌬⌬G m Ј(A52X/V74XЈ)). The double mutant proteins A52X/V74XЈ should give positive ⌬⌬⌬G m values, if the mutation at Ala 52 creates strains at the cavity of the wild-type protein, that are unfavorable for the protein stability, and if the mutation at Val 74 contributes to eliminating these strains. In contrast, they should give negative ⌬⌬⌬G m values, if the mutation at Val 74 contributes to creating additional strains. As shown in Table III, all of the double mutant proteins A52X/V74A gave positive ⌬⌬⌬G m values. Whereas, all of the double mutant proteins A52X/V74L, in which Ala 52 is replaced by bulkier hydrophobic residues, gave negative ⌬⌬⌬G m values. These results suggest that the mutation of Ala 52 to Val, Leu, Ile, or Phe creates strains within the cavity of the wild-type protein due to a collision between the substituted residue and the surrounding residues. These strains must be at least partially eliminated when Val 74 is replaced by Ala, because of the   increase in the cavity volume. In contrast, additional strains must be created when Val 74 is replaced by Leu, because of the decrease in the cavity volume. It should be noted that the A52V/V74A mutant gave a positive ⌬⌬⌬G m value, but of only Ϫ0.1 kcal/mol. Likewise, the A52V/V74L mutant gave a negative ⌬⌬⌬G m value, but of only Ϫ0.4 kcal/mol. Therefore, the mutation of Ala 52 3 Val probably does not create a significant strain at the cavity of the wild-type protein, even when its volume is reduced by the mutation of Val 74 3 Leu. Consequently, the double mutant protein A52V/V74L was the most stable mutant protein (⌬⌬G m of 2.43 kcal/mol) among those constructed in this experiment, and it was more stable than the most stable single mutant protein, A52I, by 0.6 kcal/mol. DISCUSSION As previously reported for a series of tryptophan synthase ␣ subunit mutants (52) and T4 lysozyme mutants (53) with multiple amino acid substitutions at a unique position in the protein interior, a correlation was observed between the changes in the thermal stability (⌬⌬G m ) and the hydrophobicities of the substituted residues for the RNase HI variants with a series of mutations at Ala 52 (Fig. 5). However, the data for the RNase HI variants substituted with Tyr, Phe, His, Lys, Pro, and Gly do not fall on the straight line that was obtained from a leastsquare fit of the data for 12 proteins with Ile, Leu, Cys, Val, Met, Ala, Thr, Ser, Gln, Asn, Glu, and Asp at position 52. The stabilities of the mutant proteins A52Y, A52F, A52H, A52K, A52P, and A52G are much lower than those expected from the hydrophobicities of the substituted residues.
The mutant proteins A52Y, A52F, A52H, and A52K are unexpectedly unstable, probably because the introduction of an aromatic or ionized group into the cavity alters the protein conformation. In fact, the near-UV CD spectra of these mutant proteins are significantly different from that of the wild-type protein, whereas those of the other mutant proteins are similar to or only slightly different from that of the wild-type protein (Fig. 2). Of the six Trp residues in the protein molecule, Trp 118 faces the cavity. Computer modeling suggests that this Trp residue is forced to change its orientation to avoid a collision with the bulky residues substituted for Ala 52 . 2 It is therefore likely that the changes in the near-UV CD spectra mainly reflect a conformational change of Trp 118 . This Trp residue, as well as other residues that are located around the cavity, may change their conformations, so that the strains caused by the introduction of an aromatic or ionized side chain into the cavity can be eliminated. However, these conformational changes must create additional interactions that are unfavorable for protein stability. Nevertheless, the mutant proteins A52Y, A52F, A52H, and A52K accumulated in the cells in an amount sufficient for purification. In contrast, the mutant proteins A52W and A52R did not accumulate in the cells. Since Trp and Arg are the largest aromatic and ionized residues, respectively, the mutations to these residues may create a serious conformational change that makes the protein extremely unstable. Considerable destabilization due to the introduction of aromatic residues or Arg into a space in the interior of the protein molecule has also been reported for the tryptophan synthase ␣ subunit mutants (52).
The stabilities of the RNase HI variants substituted with Glu and Asp were unexpectedly low, only when these residues were ionized. The poor cellular production levels of the mutant proteins A52E and A52D, and the normal cellular production level of A52H (data not shown), also suggest that these mutant proteins are considerably unstable only when the substituted residues are ionized. At around pH 7.0, at which the cells grow, Asp and Glu may be ionized, whereas His may not. Since the introduction of a polar group into the cavity neither unexpectedly destabilizes the protein nor seriously affects the near-UV CD spectrum of the protein, unless it is ionized, it seems likely that a polar group can be introduced into the cavity without creating significant strain. In addition to the hydrophobic residues that surround the cavity (Leu 49 , Ala 52 , Leu 56 , Leu 67 , Val 74 , Leu 107 , and Trp 118 ), many hydrophobic residues, such as Val 5 , Ile 7 , Tyr 22 , Leu 26 , Tyr 28 , Phe 35 , Ile 53 , Leu 59 , Val 65 , Ile 78 , Leu 103 , Leu 111 , Ile 116 , and Trp 120 , form the hydrophobic core of the protein. The extensive hydrophobic interactions among these residues probably make the conformation of the cavity fairly stable and thereby make it tolerant of the mutations. In the wild-type protein, the cavity is vacant and is not occupied by water molecules. However, it remains to be determined whether the introduction of a polar group into the cavity is accompanied by the introduction of a water molecule.
In contrast to the mutations to Tyr, Phe, His, and Lys, the mutations to Pro and Gly did not seriously affect the protein conformation, suggesting that these mutations do not create unfavorable van der Waals contacts within the cavity. Ala 52 is located in the ␣I helix. Therefore, the instabilities of the mu-  5. Correlation between the change in the free energy of unfolding of the mutant proteins relative to that of the wildtype protein (⌬⌬Gm) and the hydrophobicity of the substituted residues at position 52 in E. coli RNase HI. The data for the A52X single mutant proteins are indicated by X ("A" for the wild-type protein). The hydrophobic parameter represents the free energy of transfer of individual amino acids from octanol to water (60). The straight line was obtained from a least-squares fit of the 12 points shown by solid circles.
tant proteins A52P and A52G must reflect the intrinsic helixdestabilization associated with the mutation of Ala 52 3 Pro or Gly. Statistical analyses in natural proteins suggest that Pro is an ␣ helix breaker, and the introduction of a Pro residue into an ␣ helix kinks it by an average of 26 Ϯ 5° (54). However, the similarities in the near-UV CD spectra and the enzymatic activity between the mutant protein A52P and the wild-type protein suggest that the conformation of the ␣I helix is not seriously affected by the mutation of Ala 52 3 Pro. Energy minimization was carried out to see the effect of the mutation of Ala 52 3 Pro on the protein structure by the molecular mechanics program PRESTO (55). The result supports the hypothesis that the ␣I helix is not appreciably kinked at the position where Pro is introduced (data not shown). The crystallographic (56) and computer (57) analyses for the structures of the mutant proteins of T4 lysozyme, in which Pro is introduced into ␣ helices, also support this hypothesis. The long interdomain ␣ helix in T4 lysozyme, which is originally kinked by 8.5°, was shown to be additionally kinked by only 5.5°by the mutation of Asp 72 3 Pro (56). All of the mutant proteins of T4 lysozyme with Pro in the helix are dramatically less stable than the wild-type protein, by 2.7-8.2 kcal/mol. In contrast, the RNase HI mutant A52P is less stable than the wild-type protein by only 1.6 kcal/mol. In addition, this mutant protein is enzymatically fully active. The ␣I helix of E. coli RNase HI seems to be more tolerant to Pro substitutions than the ␣ helices of T4 lysozyme, probably because the ␣I helix of E. coli RNase HI is located in the interior of the protein molecule and is highly stabilized through extensive hydrophobic interactions with the rest of the protein. In fact, the mutation of either Ala 51 or Ala 55 3 Pro within the ␣I helix neither seriously destabilizes the protein nor dramatically affects the enzymatic activity. 3 The cavity-filling mutations effectively increase the stability of E. coli RNase HI, probably because the conformation of the cavity is fairly stable. Some of the stabilized mutant proteins, such as A52C and A52M, are enzymatically as active as the wild-type protein. Therefore, this cavity may not provide flexibility to the protein, which is required to make it enzymatically active. However, the enzymatic activity of the protein is rather sensitive to mutations within the cavity, probably because the cavity is located close to the active site. The determination of the kinetic parameters of the mutant proteins with poor RNase H activities suggests that the mutations within the cavity affect the catalytic efficiency, rather than the substrate binding (Table II). Some of the catalytic residues, such as Asp 10 and Asp 70 , may change their conformations when the conformation of the cavity is altered. Comparison of the effects of the mutations on the enzymatic activity, the protein stability, and the near-UV CD spectrum suggests that the changes in the enzymatic activity roughly correlate with the changes in the CD spectrum. However, some of the mutant proteins with small changes in the CD spectrum, such as A52L and A52Q, are considerably less active than the wild-type protein.
These results indicate that conformational changes around the cavity that affect the enzymatic activity are not always detected by measuring the near-UV CD spectrum. The CD spectra of mutant proteins A52L and A52Q were measured at pH 3.2, instead of pH 8.0, at which the enzymatic activity was determined. However, it is unlikely that the CD spectra of these mutant proteins are considerably changed at pH 8.0. In contrast to the correlation between the changes in the enzymatic activity and the near-UV CD spectrum, the changes in the enzymatic activity do not correlate with the changes in the protein stability. Analyses of the stabilities of a series of double mutant proteins, in which Ala 52 and Val 74 are simultaneously replaced by other hydrophobic residues, allow us to discuss the effects of the mutations within the cavity on the protein stability in more detail. The changes in stability and conformation induced by 3 M. Haruki, personal communication.
FIG. 6. Schematic illustration of the conformational changes around the cavity and the stability changes induced by the mutations. A gray oval represents the structure formed by the cavity residues and a dark circle represents a vacant space (cavity). Larger the size of the dark circle is, larger the cavity volume is. The size of the structure formed by the cavity residues of the wild-type protein is shown by broken line(s) in the structures of the mutant proteins A52I and A52I/V74L to indicate that it increases by the mutations. the single and double mutations at positions 52 (Ala 52 3 Ile) and 74 (Val 74 3 Ala and Leu) are schematically illustrated in Fig. 6. The mutation of Val 74 3 Ala increases the cavity volume, whereas the mutation of Val 74 3 Leu decreases it. The former destabilizes the protein by 2.3 kcal/mol and the latter stabilizes it by 1.1 kcal/mol. The mutation of Ala 52 3 Ile increases the stability of the V74A mutant by 3.5 kcal/mol. These results indicate that a single methyl or methylene group contributes to increase the protein stability by 1.1 kcal/mol. This value is comparable to the empirically derived value of 1.3 Ϯ 0.5 kcal/mol reported by Pace (30). The Ile side chain within the cavity of the V74A mutant probably does not create a strain. In contrast, the mutation of Ala 52 3 Ile increases the stabilities of the wild-type protein and the V74L mutant by only 1.9 and 0.1 kcal/mol, respectively. Therefore, it seems likely that the stabilizing effect caused by the burial of the additional methyl or methylene groups is partially or almost fully canceled by the destabilizing effect caused by the unfavorable van der Waals contacts in these proteins.
Dependence of the protein stability on the number of the methyl or methylene group(s) that are introduced into or removed from the cavity is shown in Fig. 7. It suggests that the volume of the cavity in the wild-type protein is large enough to allow the introduction of three methyl or methylene groups without creating considerable strain. However, the ⌬⌬G m value varied from 1.3 to 2.4 kcal/mol for the mutant proteins A52V/ V74L, A52I, and A52L, in which three methyl or methylene groups are introduced into the cavity. This suggests that the shape of the substituted residue is important for the cavityfilling mutations. The A52I and A52L mutants are less stable than the A52V/V74L mutant by 0.6 and 1.1 kcal/mol, respectively, probably due to unfavorable packing effects. Similar effects have been observed for the core mutations in T4 lysozyme (6) and N-terminal domain of repressor (7). Thus, the cavity must be most effectively filled in the A52V/V74L mutant. However, the strains still contribute to destabilizing this mutant protein. If the introduction of the double mutation of Ala 52 3 Val and Val 74 3 Leu into the wild-type protein did not create strain, it would increase the protein stability by ϳ3.3 kcal/mol. However, the A52V/V74L mutant is more stable than the wild-type protein by only 2.4 kcal/mol. It has been reported that "swapped" mutant proteins, such as V35I/I47V of the gene V protein (3,21) and L121A/A129L of T4 lysozyme (17), in which the core residues are reversed, are considerably less stable than the parent wild-type protein, by 2.9 and 1.1 kcal/mol, respectively. These results indicate that the packing effects are the major determinants of the stabilities of the protein variants with core mutations. However, the swapped RNase HI mutant, A52V/V74A, is less stable than the wild-type protein by only 0.5 kcal/mol. The packing effects may not be major determinants of the stabilities of the proteins with mutations at the cavity. A space in the cavity may serve to reduce the unfavorable contacts caused by a swapped mutation.
In this report, the ⌬⌬G m values estimated from the equation of Becktel and Schellman (51) were used to evaluate the effects of the mutations on the protein stability. These values are almost identical with those calculated by using the enthalpy change of unfolding (⌬H m ) and the change in the heat capacity (⌬C p ) as reported previously (58), except for that of the mutant protein A52K (data not shown). The estimated ⌬⌬G m value of the mutant protein A52K (Ϫ5.93 kcal/mol) is different from the calculated one (Ϫ5.35 kcal/mol), but only by 10%. These results indicate that the equation of Becktel and Schellman (51) Tables  I and III. The straight line represents the correlation between the ⌬⌬G m and the change in the number of methyl or methylene group, which is drawn based on the assumption that the burial of a single methyl or methylene group contributes to increase the protein stability by 1.1 kcal/mol.