The Effects of Heme Pocket Hydrophobicity on the Ligand Binding Dynamics in Myoglobin as Studied with Leucine 29 Mutants*

To examine the effects of heme pocket hydrophobicity on the ligand binding in myoglobin, some artificial mutants of human myoglobin have been prepared, in which less hydrophobic amino acid residue (Ala, Gly, Ser) is located at the Leu29 (10th residue of the B helix) position. CO rebinding rates for the mutants were markedly decelerated, while the1H, and 15N NMR spectra of the mutants show that the structural changes around the heme iron for these mutants are rather small. The kinetic and structural properties of the mutants indicate that the ligand binding rate depends on the hydrophobicity inside the heme cavity for these mutants in addition to the volume of the side chain at the 29-position. On the basis of the IR stretching frequency of liganded CO, invasion of water molecules into the heme pocket in the mutants is suggested, which would be induced by the decrease in the hydrophobicity due to the amino acid substitution. A slight red shift of the position of the Soret peak for the serine mutant L29S also supports the reduced hydrophobicity inside the heme cavity. We can concluded that, together with the kinetic properties of the mutants, the hydrophobicity of the heme pocket is one of the key factors in regulating the ligand binding to the heme iron.

To examine the effects of heme pocket hydrophobicity on the ligand binding in myoglobin, some artificial mutants of human myoglobin have been prepared, in which less hydrophobic amino acid residue (Ala, Gly, Ser) is located at the Leu 29 (10th residue of the B helix) position. CO rebinding rates for the mutants were markedly decelerated, while the 1 H, and 15 N NMR spectra of the mutants show that the structural changes around the heme iron for these mutants are rather small. The kinetic and structural properties of the mutants indicate that the ligand binding rate depends on the hydrophobicity inside the heme cavity for these mutants in addition to the volume of the side chain at the 29-position. On the basis of the IR stretching frequency of liganded CO, invasion of water molecules into the heme pocket in the mutants is suggested, which would be induced by the decrease in the hydrophobicity due to the amino acid substitution. A slight red shift of the position of the Soret peak for the serine mutant L29S also supports the reduced hydrophobicity inside the heme cavity. We can concluded that, together with the kinetic properties of the mutants, the hydrophobicity of the heme pocket is one of the key factors in regulating the ligand binding to the heme iron.
Various elegant techniques to elucidate the molecular mechanism of the ligand binding process in hemoproteins have provided us with abundant information about the structural and functional factors that regulate the ligand binding dynamics (1)(2)(3). Site-directed mutagenesis, which is one of the potent techniques recently developed, has been applied to investigate functional roles of key residues in the active site of myoglobin (3)(4)(5)(6)(7), since it is a small molecular weight protein and has been used as a model for hemoproteins.
In myoglobin, some amino acid residues essential for the ligand binding have already been suggested (3, 4, 6 -14). Among these amino acid residues, Leu 29 at B10 1 has been one of the crucial amino acid residues that controls the ligand binding process and geometry of ligand (5,(15)(16)(17). The x-ray structure of myoglobin has shown that leucine 29 forms a hydrophobic cluster with other distal hydrophobic residues to restrict the movement of the distal side chains (Fig. 1) (18). A simulation study (10) has revealed that most CO molecules undergo many collisions with the residues forming walls of the heme pocket including the hydrophobic cluster. The number of collisions of Leu 29 was second to that of Val 68 in myoglobin (10), which has been also supported by picosecond and nanosecond geminate recombination studies (16). On the basis of a detailed analysis of kinetics, Gibson and co-workers (16,17) concluded that the initial movements of ligand after dissociation are toward the back of the distal pocket with the side chain of Leu 29 acting as a part of the physical barrier that restricts the ligand movement away from and back toward the heme iron atom.
In our previous study (5), we prepared two mutants that replaced Leu 29 with alanine (L29A) or isoleucine (L29I). These substitutions caused a 3-5-fold decrease in the rate constants for CO and O 2 association. Based on the remarkable decrease in the association constants, we have concluded that the leucine residue is an important constituent of the hydrophobic cluster for maintaining myoglobin's ligand binding properties. Since the mutation of the amino acids forming other hydrophobic clusters in the distal pocket affects a slight alteration of the ligand binding process (5), the hydrophobicity of the leucine residue seems to be essential for the large alteration of the ligand binding rates. To gain further insights into the functional and structural roles of the hydrophobicity at the position of 29 in the ligand binding dynamics, we prepared some more Leu 29 mutants in which the hydrophobicity of the amino acid substituted for leucine is decreased.
One of the mutants we have prepared here has a glycine residue at position 29 (L29G). The hydrophobic index of glycine is 0, which is lower than that of leucine (ϩ2.31) (19). The other amino acid we substituted for the leucine is serine (L29S). Serine is less hydrophobic due to its hydroxy group (hydrophobic index is Ϫ0.05), while its steric hindrance is similar to that of alanine. We also tried to introduce some other hydrophilic amino acid residue such as threonine and asparagine. Unfortunately, however, the mutants having a hydrophilic amino acid residue at the 29-position are highly unstable and fail to keep the heme inside the heme pocket in the cyano-met form. To discriminate effects of the hydrophobicity on the ligand binding from those of the steric difference in the side chain of the substituted amino acid residue, we prepared an additional mutant in which a phenylalanine residue is introduced into position 29 (L29F). The hydrophobicity of phenylalanine (its hydrophobic index is ϩ2.43) is similar to that of leucine, whereas the steric hindrance is quite different (15).
In this study, we utilized the laser photolysis technique to characterize the ligand binding properties of the mutants. Also, we examined the structural changes around the active site by using 1 H and 15 N NMR, IR, and electronic absorption spectroscopies to elucidate the relationship between ligand binding properties and the hydrophobicity of heme pocket.
Preparation of Mutant Myoglobins-The original expression vector of human myoglobin, 2 pMb3 (pLcIIFXMb), is a gift from Varadarajan and Boxer (20). The procedures for site-directed mutagenesis are described in previous papers (5,20,21). DNA sequencing for all mutants was performed by the DyeDeoxy Terminator method using ABI 373S DNA sequencer, and mutants were analyzed by the 373S DNA sequencing system. Protein preparation and purification were performed following the method described previously (5,22). Wild type myoglobin and the L29F mutant were purified as an aquo-met form. Since L29G, L29A, and L29S mutants are unstable in the aquo-met form, we purified them in the cyano-met form.
Electronic Absorption and IR Spectra-Electronic absorption spectra of purified proteins in 100 mM sodium phosphate buffer, pH 7.0, were recorded on a Shimadzu UV 2200 UV/visible spectrophotometer. The sample concentration was 5 M.
Infrared spectra were measured at 1-cm Ϫ1 resolution on a Bio-Rad FTS-30 spectrophotometer. The protein solution was loaded into a CaF 2 cell with 0.1-mm path length. The spectral ranges were 1900 -2000 cm Ϫ1 for carbonmonoxy forms and 2000 -2200 cm Ϫ1 for cyano-met forms. We used the equimolar aquo-met form of wild type myoglobin as a reference. In the measurements of the cyano-met forms, we used a D 2 O buffer solution, since the signal:noise ratio was significantly higher in D 2 O than in H 2 O, and no isotope effects have been observed (23). An average of 512 scans was used for each spectrum. The sample concentration was about 1 mM for the carbonmonoxy forms, pH 7.0, and 3 mM for cyano-met forms in 100 mM sodium phosphate, pD 7.0.
Kinetic Measurements of CO and O 2 Binding-The association rate constants of CO and O 2 were obtained by a laser photolysis apparatus, which has been previously described in detail (5,24). For milli-to microsecond experiments, a flashlamp-pumped dye laser with a halfpeak duration of 300 ns (Unisoku LA-501) was used. Rhodamine 6G (Eastman Kodak Co.) in methanol was used to produce an excitation pulse at 590 nm, the wavelength of maximum intensity. The probe light at 440 nm was focused onto the slit of a monochromator (Unisoku USP-501) and detected by a photomultiplier. A transient memory (Graphtec TMR-80) was used to digitize the signal (50 ns/point, 4096 points), and data were transferred to a NEC PC-9801VX computer for further data analysis. The sample concentration was about 5 M. The buffer conditions were 100 mM sodium phosphate, pH 7.0, at 20°C.
Ligand rebinding to the mutant and wild type myoglobins were analyzed by fitting to the equation, where ⌬A t is the absorbance change at any time t and ⌬A 0 is the total absorbance change (absorbance at t ϭ 0 minus absorbance at t ϭ ϱ). k app is the observed first order rate constant and satisfies the equation (26), where k on is the bimolecular ligand binding rate constant and [L] is ligand concentration. The kinetics measurements of the CO dissociation rates were carried out with a UV/visible spectrometer (Shimadzu UV-2200). k off was determined by analyzing the replacement reaction in which ligated CO was replaced by NO as described in detail by Lambright et al. (25). The concentrated myoglobin stocks were converted to the carbonmonoxy form by stirring under CO followed by reduction with sodium dithionite. NO-saturated buffer (100 mM sodium phosphate, pH 7.0) was prepared by bubbling NO through 3 ml of buffer in a sealed 1-cm path length UV cell. About 20 l of the concentrated MbCO solution was then injected into the cell at 20°C, and the reaction was followed by monitoring the absorbance at 424 nm. Association equilibrium constants were calculated as the ratio of the rate constants.
NMR Spectroscopy-1 H NMR spectra were recorded by using a GE Omega 500 spectrometer equipped with a SUN 3 workstation. Hyperfine-shifted NMR spectra were obtained with an 8-kilobyte data transform of 125 kHz and 7.0-s 90°pulse by using a conventional watereliminated Fourier transform pulse sequence (180--90 degrees) to minimize the strong solvent resonances in H 2 O solution. An appropriate setting of the value (typically 120 -130 ms) can eliminate H 2 O signal under rapid repetition of the sequence. The probe temperature was 23 Ϯ 0.5°C. The concentration was about 1 mM in 100 mM sodium phosphate, pH 7.0, and the sample volume was about 500 l. Proton shifts were referenced with respect to the signal from the proton resonance of tetramethylsilane. 15 N NMR spectra of the C 15 N derivatives were taken at 50.67 MHz. Chemical shifts are given in ppm with reference to the resonance of external 15 NH 3 . Sample conditions for the 15 N NMR spectra were the same as those for the 1 H NMR spectra (5).
Near-infrared Absorption Spectra of Deoxymyoglobins-Deoxymyoglobin was prepared by reduction of metmyoglobin with a small amount of solid sodium dithionite under an argon atmosphere. The metmyoglobin solution was calmly stirred under the argon atmosphere on ice more than 30 min before reducing. The sample concentration was 1 mM, and we used 100 mM sodium phosphate buffer, pH 7.0, at 20°C. Nearinfrared absorption spectra were recorded on a Shimadzu UV 2200 UV/visible spectrophotometer over the range of 700 -800 nm.

Electronic Absorption Spectra in Carbonmonoxy Form-
The peak positions of the Soret bands for the wild type and mutant myoglobins are listed in Table I. The carbonmonoxy form of wild type myoglobin has the Soret band at 423.0 nm. In the absorption spectra of the mutants, subtle but significant deviations of the absorption maxima were observed. The L29A and L29S mutants exhibited the Soret peak at 423.2 and 423.8 nm, whereas the peaks for L29G and L29F mutants were blueshifted to 422.8 and 422.4 nm, respectively.
Laser Photolysis of Bimolecular Binding for Carbonmonoxyand Oxymyoglobins- Table II shows the bimolecular association rate constants for CO and O 2 rebinding to wild type and mutant myoglobins. The second-order rate constants (k on ) were determined by milli-and microsecond laser photolysis measurements. Fig. 2 shows the time courses for CO rebinding to wild type and mutant myoglobins. The time courses were fitted  to a simple exponential curve, and the bimolecular association rates were calculated. Compared with wild type, the CO association rates of the mutants were largely decreased; k on (CO) for the L29G mutant was 0.11 M Ϫ1 s Ϫ1 , and k on (CO) for the L29F mutant was 0.072 M Ϫ1 s Ϫ1 at pH 7.0, 20°C. The rate constant for the L29S mutant is extremely small (0.069 M Ϫ1 s Ϫ1 ). 3 In our previous paper (5), we speculated that increasing the space in the immediate vicinity of the heme in the L29A mutant allows more water molecules to enter the heme pocket, resulting in decrease of the CO binding rate. In the present results, L29G, in which the space in the vicinity of heme pocket would be larger than that in L29A, also exhibited a smaller CO rebinding rate (0.11 M Ϫ1 s Ϫ1 ) than L29A (0.15 M Ϫ1 s Ϫ1 ) (Table II). Furthermore, it is rather surprising that an extremely slow ligand binding (0.069 M Ϫ1 s Ϫ1 ) was observed for the serine mutant (L29S), since the replacement of alanine by serine would not increase the space of the heme pocket. The comparison of the ligand binding rates for the L29S and L29G mutants clearly indicates that the decrease in the CO binding rate in Leu 29 mutants (5) is not simply due to the increasing space in the vicinity of the heme (see Fig. 5A). It is quite interesting that the CO association rate constants depend on the hydrophobicity for the amino acid at the 29-position as plotted in Fig. 5B. 4 Fig. 5B also includes the rebinding data of the sperm whale mutant myoglobins previously reported (7,15). The mutant having the most hydrophilic residue at the 29-position, L29S, shows the slowest CO association rate constant, followed by the L29G and L29A mutants. Wild type myoglobin, in which the most hydrophobic residue, leucine, is occupied at the 29-position, exhibits a much faster rebinding rate. These observations strongly suggest that the hydrophobicity at position 29 is one of the crucial factors for the ligand binding process. However, the ligand binding rates of the L29F, L29W (sperm whale), and L29I mutants are not along this line. In these mutants, the effects of the steric hindrance of the side chains would be a dominant factor for regulating CO rebinding and overwhelm the hydrophobic effect (7,11). 5 The O 2 association rate constants for the mutants were also largely decreased. k on (O 2 ) for the L29F mutant was 5.1 M Ϫ1 s Ϫ1 . The rate constants for the L29G and L29S mutants are smallest (2.1 M Ϫ1 s Ϫ1 and 2.2 M Ϫ1 s Ϫ1 ). Although the oxygen binding in human myoglobin is sensitive to volume and/or hydrophobicity at position B10, the oxygen binding is little affected by decreasing the size of the B10 residue in sperm whale myoglobin (15), suggestive of the different regulation mechanism between the CO and O 2 rebinding. 6 Dissociation Rates of Carbon Monoxide-The dissociation rates, k off , of the carbon monoxide from wild type and mutant 3 By freeze-thawing, a minor component appeared in the rebinding reaction for L29S mutant, and the time course was biphasic. However, the fraction amplitude of the minor process is less than 10%, and the rate constant of the major component estimated by a single exponential fitting does not seriously deviate from a double exponential fitting. 4 The hydrophobicity index we used here was proposed by Fauchère and Pliska (19). This value was defined as the partitioning of N-acetylamino acid amides between water and octanol, which are probably the most commonly used for estimating the contribution of the hydrophobic effect. We confirmed that the similar relationship between the hydrophobicity and the rebinding rate was observed by the use of other hydrophobic indexes (57)(58)(59). 5 For L29I mutant, the association rate constant (k on ϭ 0.19 M Ϫ1 s Ϫ1 ) is smaller than that of wild type (k on ϭ 1.1 M Ϫ1 s Ϫ1 ), whereas the hydrophobicity index and molecular volume of isoleucine are almost the same as those of leucine. By inspection of the x-ray structure, Ile 29 (B10) C ␥ H 3 branching at C ␤ H would interfere with the motion of the E heliix, which perturbs the dynamic interactions for the packing between the B and E helices and induces the steric effects as suggested in the L29F and L29W (sperm whale) mutants (15). 6 All of the mutant myoglobins showed no CO geminate binding (geminate recombination yields are less than 2%) under these conditions (data not shown). This result for the L29S mutant of human myoglobin is significantly different from that of sperm whale myoglobin reported by Huang and Boxer (13), although the time course for O 2 binding to the human L29S mutant is quite similar to that of the sperm whale L29S mutant (13). They showed that about 20% of photodissociated carbon monoxide geminately rebinds to the heme, and there were two phases in bimolecular recombination ranging from 10 s to 10 ms for the sperm whale L29S mutant. On the other hand, the mutant human myoglobins we prepared here exhibit no CO geminate rebinding (geminate recombination yields are less than 2%) under these conditions (data not shown). The different measurement conditions might be responsible for the different geminate rebinding. Our conditions were 100 mM sodium phosphate, pH 7.0, at 20°C, whereas 50 mM Tris/Cl and 1 mM EDTA, pH 8.0, at 23°C were used for the measurements by Huang and Boxer. In addition to the experiment conditions, we measured the rebinding kinetics for the purified human L29S mutant protein, while they used sperm whale L29S mutant in the crude lysate of Escherichia coli (13).  myoglobins were obtained by the NO replacement method (2,25,26) and are compiled in Table II. k off of human wild type myoglobin (0.018 s Ϫ1 ) was virtually identical to that reported previously (27). As listed in Table II, the dissociation rates for the mutants bearing a less hydrophobic amino acid residue at B10, L29A, L29G, and L29S mutants, are almost identical to that of wild type myoglobin. However, the dissociation rate of the L29F mutant was markedly decreased to 0.0014 s Ϫ1 , which is a more than 10-fold decrease as compared with that of wild type myoglobin. We also calculated the carbon monoxide affinity (K CO ϭ k on /k off ) of the wild type and mutant myoglobins (Table II). The affinity to carbon monoxide is markedly reduced in the L29A, L29G, and L29S mutants, whereas that for the L29F mutant is almost equal to that for wild type myoglobin.

Infrared and 1 H NMR Spectra of Carbonmonoxymyoglobins-
The IR absorption spectra in the iron-ligated C-O stretching region, 1900 -2000 cm Ϫ1 , of wild type and mutant myoglobins are illustrated in Fig. 3. Table I summarizes parameters of the C-O stretching mode infrared band. The spectral patterns of the L29A and L29F mutants correspond to those of Leu 29 mutants of sperm whale myoglobin (28). The two or three distinguishable bands were observed at around 1932, 1945, and 1965 cm Ϫ1 for all mutants except for the L29F mutant. The intensity of the peak around 1945 cm Ϫ1 in the Leu 29 mutants decreased, and the peak around 1932 cm Ϫ1 gained its intensity. A single broad IR band at 1933 cm Ϫ1 for the L29F mutant implies the complete conversion from the conformer A 1 of wild type (29,30) to the other conformer by the substitution for leucine by phenylalanine. The IR band for the L29S mutant is composed of two distinct bands centered at 1950 and 1936 cm Ϫ1 . Although they correspond to A 1 and A 3 conformers of wild type, the positions of these peaks show a shift to high frequency by 4 -5 cm Ϫ1 .
By using 1 H NMR spectra, we also investigated the structural changes around the heme. A ring current-shifted signal observed at Ϫ2.5 ppm has been assigned to C ␥ H 3 in Val 68 , which has served as a sensitive marker for the structural changes near the basic heme (28, 29, 31-34). As shown in Table  I, the mutation at the position of Leu 29 has little effect on this marker, indicating that the heme environmental structures in CO-bound forms are essentially the same as those found in wild type myoglobin.
Infrared, 1 H NMR, and 15 N NMR Spectra of Met-cyanomyoglobins-The prominent resonances in 1 H NMR spectra of cyanide adducts of wild type and mutant myoglobins are summarized in Table III. As listed in Table III, the resonance positions of 1-, 5-, and 8-heme methyl protons (a-c) were almost insensitive to the amino acid substitution. This indicates that the orientation of the axial histidine is little affected by this mutation (35,36). In the L29F mutant, however, the signal of the distal histidine N ⑀ H (peak d) appeared at 28.9 ppm, which was about 5.8 ppm downfield shifted from that for wild type myoglobin, implying a perturbation in the orientation of the distal histidine by the amino acid substitution.
The signals of the proximal histidine C ⑀ H (peak f) in L29A and L29G were downfield shifted by 2.7 and 2.9 ppm, respectively, and such shifts were also encountered in the signal position of the proximal histidine C ␦ H (peak g). Since the hyperfine shifts of these two protons depend on the tilting angle of the magnetic z axis (37), the tilting angle of bound cyanide from the heme normal would be increased in these mutants. The resonance from C ␥ H (peak h) in Ile 99 (FG5) is also one of the useful proximal side structural markers (38). As shown in Table III, the resonance positions of C ␥ H (peak h) in Ile 99 (FG5) are almost independent of the mutation at B10. The 15 N NMR spectra of met-cyano-myoglobins were measured in the presence of excess cyanide ion (C 15 N Ϫ ), which are illustrated in Fig. 4, and the resonance positions are summarized in Table IV. The resonance from the liganded C 15 N Ϫ in wild type human myoglobin appeared at 1342 ppm from 15 NH 3 , which corresponds to that of wild type sperm whale and horse myoglobin (39,40). The L29F mutants exhibited resonance at 1306 ppm, which is relatively close to that of human wild type myoglobin. However, the substitution of leucine for the smaller and less hydrophobic amino acid residues, glycine or alanine, induced large upfield shifts, 1227 and 1236 ppm, respectively. The L29S mutant showed the largest upfield shifts, 1202 ppm as compiled in Table IV. The C-N stretching band of wild type myoglobin appeared at 2125 cm Ϫ1 (23,41), whereas leucine 29 mutants, L29G, L29A, and L29S, exhibited the corresponding band at 2126 -2128 cm Ϫ1 , which is shifted by 2-3 cm Ϫ1 to slightly higher frequencies, as listed in Table IV. Since the C-N stretching is one of the indicators for the configuration of the liganded cyanide in hemoproteins (23,41), the slight shifts of the C-N stretching mode in these mutants imply that the bending in the C-N bond for the mutants is not so seriously affected by the amino acid substitution. On the other hand, in the IR spectra of the cyanomet phenylalanine mutant, L29F, the C-N stretching mode shifts 10 cm Ϫ1 to higher frequencies.
1 H NMR and Near-infrared Absorption Spectra of Deoxymyoglobins-A resonance of exchangeable proton in the far down-  field region observed at 80.6 ppm in wild type deoxymyoglobin has been assigned to the N ␦ H proton of the proximal histidine (42,43). These signals of all the mutant myoglobins showed a downfield shift. The signals for the L29G and L29A mutants appeared at 82.2 ppm, and that of the L29F mutant was at 84.2 ppm. The L29S mutant exhibited the largest downfield shift at 85.1 ppm (Table V). The near-infrared absorption spectra (band III) for wild type and Leu 29 mutant myoglobins in the deoxy form are examined to investigate the structural change around the heme proximal site. These spectra are well described by a single Gaussian function on top of a cubic polynomial background (45,46). The center frequency, 763.9 nm, for wild type myoglobin is identical with the previous data (46). The range of the center frequencies for the Leu 29 mutants was from 763.0 -765.8 nm (Table V). The band III has been assigned to charge transfer transition of a 2u 3 d xy , porphyrin to iron (47,48), and the center frequency of band III reflects the position of the iron relative to the plane of the porphyrin (45). The red shift for the L29G, L29A, and L29S mutants suggests that the heme iron in these mutants is less out of the heme plane than that of wild type, whereas, in the L29F mutant, the deviation of the heme iron from the heme plane is larger.

DISCUSSION
The Ligand Environment Structure of the Less Hydrophobic Leucine 29 (L29A, L29G, and L29S) Mutants-Although the structural deviations in the less hydrophobic mutants, L29S, L29A, and L29G, from wild type myoglobin would be subtle as shown by their spectroscopic data, the environmental changes around heme ligand are suggested by IR, electronic, and NMR spectra. In the IR spectra of L29G, L29A, and L29S mutants (Fig. 3), the stretching mode at 1933 cm Ϫ1 was enhanced, while the intensity of the band at 1945 cm Ϫ1 was decreased. These changes in the intensity of the stretching mode were also detected in various myoglobin mutants (28), and the appearance of the stretching mode at the lower frequency region has been ascribed to the destabilization of the partial negative charge at the oxygen atom by positive electrostatic environment near the CO ligand (28).
Electronic absorption spectra of the carbon monoxide complexes of the L29S mutant myoglobin can be another marker for the polarity of the heme vicinity (49,50). In the polar environment, the excited state is more effectively polarized than the ground state and stabilized, which lowers the heme -* transition energy and shifts the Soret maximum to red. The peaks of the Soret band maximum for less hydrophobic mutants, L29S and L29A, were observed at 423.8 and 423.2 nm, respectively, while that of wild type myoglobin appears at 423.0 nm. These red shifts in the Soret peak of the mutant would correspond to the increase of the polarity in the heme pocket, implying that the replacement of Leu 29 by a less hydrophobic residue reduces the hydrophobicity of the heme pocket. Although the L29G mutant has a less hydrophobic residue at the 29-position, the Soret band is slightly blueshifted, suggesting that some other electronic factors affect the peak position of the Soret band in this mutant. We are now trying to identify the fine structural differences in these mutants by multidimensional NMR measurements.
The environmental changes around the ligand in the less hydrophobic mutants are also reflected in 15 N NMR spectra of the C 15 N-bound form. The replacement of the less hydrophobic amino acid residue at position 29 induced a large upfield shift as illustrated in Fig. 4 (1342, 1236, 1227, and 1202 ppm for wild type, L29A, L29G, and L29S, respectively). One of the major factors to determine the resonance position of 15 N in liganded cyanide is the steric hindrance around the ligand. Using the strapped heme model, Avilés and Chang (51) have indicated that the increase in the steric hindrance around liganded cyanide causes tilting and/or bending of the cyanide ligand, resulting in an upfield shift. In contrast to the large upfield bias of the resonance position of the liganded C 15 N in the mutants from that of wild type myoglobin, the deviations of the stretching mode of cyanide in the less hydrophobic mutants are within 3 cm Ϫ1 . Since the C-N stretching mode is affected by the perturbation of the bending of the liganded CN Ϫ , but not so much by the tilting (23), the mutation at the 29-position to glycine, alanine, or serine might induce the tilting of the liganded cyanide rather than the bending. Tilting of the cyanide ligand is also supported by 1 H NMR. The resonance positions of the proximal histidine ring protons, C ⑀ H (peak f) and C ␦ H (peak g), have been considered to be sensitive to the local structure of the liganded cyanide (37). Table III    and L29S (38 ppm) mutants is significantly larger than that for wild type myoglobin (27 ppm), which indicates that tilting of the liganded cyanide increased (37). The Effect of Heme Pocket Hydrophobicity on CO Bimolecular Binding Rate-Although the heme environmental structure was affected by the replacement of leucine 29 by alanine, glycine, and serine, the structural changes are subtle, and it is unlikely that the remarkably reduced CO binding rate in these less hydrophobic mutants can be ascribed solely to the conformational changes in the heme cavity induced by the mutation. One of the key features in these mutants is the dependence of their rebinding rate on the hydrophobicity index of the amino acid residue at position 29 as depicted in Fig. 5B. The reduced hydrophobicity at position 29 decelerates the rebinding rate.
The decreased hydrophobicity of the heme cavity allows us to speculate that the contribution of water molecules inside the heme pocket to the ligand binding process is affected by the mutation. As previous studies have shown, the replacement of noncovalently bound water molecule, which is hydrogenbonded to N ⑀ of His 64 (E7) in deoxymyoglobin, would be the major factor of regulating the carbon monoxide binding to the heme iron (2,3,52,53). In the CO-bound form, the possibility of water in the heme pocket might effectively increase the local water concentration, which is possibly an additional factor that contributes to decreasing the ligand rebinding rate. It is, therefore, most plausible that the decreased hydrophobicity in the heme pocket introduces the extra water molecules into the heme cavity or increases the density of water molecules around the heme iron in the L29A, L29G, and L29S mutants.
Although it is quite difficult to estimate the position of the labile water molecules inside the heme pocket, even by crystallography, the present spectroscopic data suggest that the position and/or number of the water molecules inside the heme pocket is changed by the mutation at Leu 29 . As discussed above, the infrared spectra of carbon monoxide myoglobins (Fig. 3) indicates the positive charge near the heme ligand in the L29G, L29A, and L29S mutants. Since the conformational changes around the heme iron are small, the presence of a positive charge can be derived from water molecules near the heme iron that are not observed for native myoglobin, as dis-cussed by Li et al. (28). The presence of water molecules near the heme iron also corresponds to the upfield shift of the 15 N resonance in the mutants (Fig. 4). The interaction between the nitrogen atom of the ligated cyanide and water molecule, possibly a hydrogen bond between the ligated cyanide and water molecule, would cause upfield shift in the resonance position of 15 N (54). The red shift in the Soret peak might also imply the invasion of water molecules into the heme pocket by the replacement of leucine 29 with the less hydrophobic amino acid residue. On the basis of the microscopic theory of solvent-solute interactions (49), Jung et al. (55) suggested that the red shift and broadening of the Soret band observed for cytochrome P450 by pressurization indicates a higher water content in the heme environment. In myoglobin, however, the red shift of the Soret transition was not accompanied by a broadening, which is in sharp contrast to that in cytochrome P450 (55). Since the heme environmental structure of myoglobin is significantly different from that of cytochrome P450, the higher water content in the heme pocket would not be the only factor for the reduced hydrophobicity in the heme pocket of the mutants.
Thus, we can suggest that the hydrophobicity at position B10 is one of the factors to control the ligand binding process, and the reduced hydrophobicity could favor the easy access of the water molecules into the heme pocket, which raises the barrier to CO binding and decreases the CO binding rate. The effects of hydrophobicity in the heme pocket on the ligand binding process were also manifested in the sperm whale mutant myoglobin (H64G), bearing a glycine residue at the position of the distal histidine (2,53). The x-ray structure of the sperm whale myoglobin mutant clearly showed that approximately two water molecules are located adjacent to the coordinated ligand in the CO form in its highly polar heme pocket, and the ligand binding rate for the mutant was markedly retarded (53).
Ligand Binding and Heme Environmental Structure in the L29F Mutant-As clearly shown by the spectroscopy used here, the structural changes around the heme active site for the L29F mutant are more prominent than those for the other mutants. The IR stretching mode of the carbonmonoxy L29F mutant consists of almost one component, while two or three components in the stretching modes were found for other mu-  (Table I). The 1 H NMR signal of the distal histidine N ⑀ H in the mutant was detected in the far downfield region (Table  III), whereas the deviation of the 15 N resonance is much smaller than other mutants (Table IV). These spectral changes show that the effects of the substitution of phenylalanine for leucine at the 29-position are quite different from those of alanine, glycine, and serine. In the crystallographic structures of the sperm whale mutant myoglobin, the large benzene ring of phenylalanine occupies the CO binding site (15). Since the edge of the benzene ring has been proposed to be a partial positive charge and favors interactions with the negative charged face (56), this electrostatic interaction would affect the C-O stretching mode of the carbonmonoxy form (Fig. 3) and the C-N stretching mode of the cyanide adduct (Table IV). However, the NMR signal from the methyl group of Val 68 was insensitive to the L29F mutation, implying that the structural changes in the L29F mutant might be localized near the ligand. Although the detailed conformational changes are not yet clear, it can be safely said that these conformational changes around the ligand, which are not observed for the other mutants in this study, would lead to the extremely slow CO rebinding rate of the L29F mutant.
In summary, the mutants in which Leu 29 is substituted for a less hydrophobic amino acid residue exhibited a substantial decrease in the carbon monoxide binding rates, and their binding rates were dependent on the hydrophobicity of the amino acid residue at position 29. Although the volume of amino acid at that position is one of the factors that affects the ligand binding kinetics of myoglobin, we propose a contribution of the hydrophobicity at the 29-position to the control mechanism of the ligand binding process. Despite the limited structural information, it can be concluded that the amino acid substitutions would introduce water molecules inside or near the heme cavity, and these noncovalently bound water molecules would perturb the barrier of CO binding to heme iron.