10750
J. Phys. Chem. B 2000, 104, 10750-10756
Distal Interactions in the Cyanide Complex of Ferric Chlamydomonas Hemoglobin† Tapan Kanti Das,‡ Manon Couture,‡ Michel Guertin,§ and Denis L. Rousseau*,‡ Department of Physiology and Biophysics, Albert Einstein College of Medicine, 1300 Morris Park AVenue, Bronx, New York 10461, and Department of Biochemistry and Microbiology, Faculty of Sciences and Engineering, LaVal UniVersity, Quebec, G1K 7P4, Canada ReceiVed: February 4, 2000; In Final Form: April 21, 2000
Chlamydomonas hemoglobin is expressed in chloroplast during active photosynthesis. Its heme pocket has an unusual structure that undergoes substantial changes when exogenous ligands bind the heme iron. In the ferrous state of the heme, oxygen binds with high affinity and is stabilized by interactions from E7-glutamine and B10-tyrosine. In the present study, we have examined cyanide binding to the ferric heme by resonance Raman spectroscopy. The frequency of the Fe-CN stretching mode in the wild-type protein was assigned at 440 cm-1, which is significantly lower than that observed in other globins. Another cyanide isotope sensitive mode located at 315 cm-1 is tentatively assigned as an Fe(III)-His stretching mode in the six-coordinate CN adduct. To determine the sensitivity of the Fe-CN stretching mode to the interactions in the distal pocket, we also studied three distal pocket mutants. The frequency of the Fe-CN stretching mode in the cyanide complex of the Lys87Ala (E10) mutant was nearly identical to that of the wild-type protein but it increased to 452 cm-1 for the Gln84Gly (E7) mutant. In the Tyr63Leu (B10) mutant, the frequency of the Fe-CN stretching mode decreased by 5 cm-1 relative to that in the wild type. On the basis of the low frequency for the Fe-CN stretching mode in the wild-type protein and its smaller than expected cyanide isotope shift (4 cm-1 compared to 11 cm-1 expected for a two-body Fe-CN oscillator) we propose a highly bent Fe-C-N structure in Chlamydomonas hemoglobin, in sharp contrast to the widely accepted linear cyanide structure in most vertebrate globins. The occurrence of a bent cyanide structure in Chlamydomonas hemoglobin is likely caused by a congested distal cavity resulting in a strong steric interaction between the E7-glutamine residue and the heme-bound cyanide.
Hemeproteins perform innumerable biological functions necessary to sustain life in the biosphere ranging from the plant kingdom to the microbial world to mammals. The wide range of functions that hemeproteins carry out in various cellular compartments includes ligand transport, electron transfer, and an array of enzymatic activities. Hemeproteins display such functional diversity despite sharing a common redox cofactor, owing to the arrangement of the amino acid residues which coordinate to the heme and form its surrounding pocket. The understanding of the interactions between heme-bound ligands in hemeproteins and the residues in their distal environments has received a great deal of attention in recent years as it allows for the clarification of catalytic mechanisms in some proteins and the rationalization of ligand kinetic behaviors in others. This is especially important in hemoglobins and myoglobins, which have now been found to be expressed in many species ranging from bacteria to plants to animals (see ref 1 for a review), with unknown functions in many cases. Resonance Raman spectroscopy is a powerful probe for studying the heme-ligand vibrational characteristics to determine the properties of the iron-ligand coordination and the nature of the interactions between the heme-bound ligand and its distal pocket residues. In a great number of resonance Raman studies of hemeproteins, CO has been used as a probe of distal interactions (see †
Part of the special issue “Thomas Spiro Festschrift”. * Corresponding author. Phone: 718-430-4264. Fax: 718-430-8808. E-mail:
[email protected]. ‡ Albert Einstein College of Medicine. § Laval University.
for example refs 2-11) since the stretching and bending modes of the Fe-C-O moiety respond to the steric, electrostatic, and hydrogen-bonding interactions between the ligand and the residues in its local environment. Unlike CO, there have been only a limited number of resonance Raman studies on CN-bound hemeproteins in which the cyanide-sensitive modes were examined. In addition, most of these studies on such proteins as peroxidases, globins, catalases and oxidases have been done on native or wild-type forms. An interpretation of the differences in the vibrational properties of the Fe-C-N modes between these different classes of hemeproteins is not straightforward. Moreover, in general it is not well understood what factors are responsible for the modulation of the vibrational properties of the FeCN moiety. To gain a quantitative understanding of the role of the distal residues in controlling the ligand binding properties in hemeproteins, it is useful to construct single-point mutants and compare their vibrational properties to those of the native (wild type) protein. In the present report, we have studied the cyanide complex of the recently discovered chloroplast hemoglobin from Chlamydomonas eugametos by Soret-enhanced resonance Raman spectroscopy. Expression of the Chlamydomonas hemoglobin gene is induced by light and requires active photosynthesis.12,13 Chlamydomonas hemoglobin contains a b-type protoheme, which is anchored to the polypeptide by the proximal histidine (at the F8 helical position).14 In the oxidized (ferric) form,15 the distal tyrosine residue (B10) coordinates to the heme in a pH-dependent manner to form a 6-coordinate low-spin complex with an unusually low pKa of 6.3. A lysine residue located at
10.1021/jp000452y CCC: $19.00 © 2000 American Chemical Society Published on Web 06/06/2000
Cyanide Complex of Ferric Chlamydomonas Hemoglobin the E10 position influences the tyrosine coordination to the heme presumably by imparting strong hydrogen bonding that drastically lowers the pKa of the phenolic oxygen of the tyrosine.15 The contribution of the E7 glutamine to the tyrosine coordination chemistry is minimal. Binding properties of the endogenous ligand in the ferrous protein14 is different from that in the ferric form and displays a pKa of 8.5. A variety of exogenous ligands, however, can replace the endogenous sixth ligand to the heme and form stable adducts. For example, oxygen binds to the ferrous heme with high affinity and the bound oxygen is stabilized by both the E7 glutamine and B10 tyrosine residues, but not by the E10 lysine.14 Thus, it appears that in the presence of exogenous ligands, a substantial structural rearrangement occurs in the heme pocket. Given the unusual nature of the reactivity of the heme and the disposition of the distal pocket ligands in Chlamydomonas hemoglobin, it is important to determine the factors that control ligand binding. We have studied cyanide binding to wild-type ferric Chlamydomonas hemoglobin and various distal pocket single mutant forms. The vibrational properties of the FeCN group in the wild-type protein in comparison to those in the mutants give insights into the structural and electronic factors that modulate the ligand binding in the ferric protein. Methods and Materials Purification of Chlamydomonas Hb and Its Mutants. Recombinant monomeric Hb H21 was prepared as described elsewhere16 by removing the first 24 amino acids, and by substituting a lysine for the unique cysteine found at position 41 of the parent protein (LI637). This recombinant H21 Hb is referred in the present work as wild-type Chlamydomonas Hb. The single residue mutants of H21 Hb (Gln84Gly, Tyr63Leu, and Lys87Ala at putative helix positions E7, B10, and E10, respectively) were prepared as described previously.14 Resonance Raman Spectroscopy. The Raman experiments were carried out with 413.1 nm excitation from a CW Kr ion laser (Spectra Physics, Mountain View, CA). The sample cell (quartz, 2 mm path length, sample volume ∼150 µL) into which a laser beam was focused was spun at 3000 rpm to minimize local heating. The sample cells are custom designed and can be used for recording both the resonance Raman spectra and the optical absorption spectra (UV-2100U spectrophotometer, Shimadzu, Kyoto, Japan). The Raman scattered light was focused onto the entrance slit (100 µm) of a polychromator (Spex, Metuchen, NJ), dispersed by a 1200 grooves/mm grating and detected by a liquid nitrogen cooled charged couple device (CCD) (Princeton Instruments, Trenton, NJ). A holographic notch filter (Kaiser, Ann Arbor, MI) was used to eliminate Rayleigh scattering. Typically, several 30 s spectra were recorded and averaged. Frequency shifts in the Raman spectra were calibrated using acetone-CCl4 (for the 100-1000 cm-1 region) and indene (for the 100-1700 cm-1 region) as references. The accuracy of the Raman shifts is about (2 cm-1 for absolute shifts and about (0.5 cm-1 for relative shifts. The Raman data were processed by using GRAMS (Galactic Industries Corporation) software. The cosmic ray spikes were removed by using CSMA subroutines (Princeton Instruments). Cyanide complexes of the wild-type and the mutant hemoglobins were prepared by the addition of buffered KCN solution (12C14N- or 13C15N- to make the final CN- concentration of ∼2 mM) to the ferric protein (∼40 µM). Cyanide isotopes are products of ICON (Mt. Marion, NY). The laser power at the sample was ∼5 mW. Absorption spectra were recorded before and after the Raman measurements to ensure the stability of
J. Phys. Chem. B, Vol. 104, No. 46, 2000 10751
Figure 1. Resonance Raman spectra in the high-frequency region of the cyanide complex (a) of ferric Chlamydomonas hemoglobin at pH 7.5. For comparison, the spectrum of the ferric protein (at the same pH) in the presence of imidazole (b) and in the absence of any exogenous ligand (c) are also shown. The spectra are normalized to the intensity of the ν4 band (located at 1372-1374 cm-1).
the species studied. For H/D isotope (H2O/D2O) studies, solutions of the cyanide complex of the wild-type protein (40 µM) were prepared in either 100% H2O or in 80% D2O + 20% H2O (using 99.9% D2O, Aldrich Chemical Co. Inc., Milwaukee, WI) buffered with 40 mM sodium phosphate at pH 7.4. Results Resonance Raman Spectra in the High-Frequency Region. Figure 1 shows the resonance Raman spectra in the highfrequency region for wild-type ferric Chlamydomonas hemoglobin in the presence of various exogenous ligands. This region (1300-1700 cm-1) of the resonance Raman spectra of hemeproteins and model heme complexes is comprised of porphyrin in-plane vibrational modes which are markers of the oxidation state, coordination state, and spin state of the central iron atom.17-25 In Figure 1, the lines of particular interest are located at ∼1374 (assigned as ν4), ∼1500 (assigned as ν3), and ∼1637 (assigned as ν10) cm-1.24,25 These frequencies are typical of 6-coordinate ferric low-spin heme complexes confirming that all three adducts in Figure 1 have a 6-coordinate low-spin configuration. The wild-type protein in the absence of any exogenous ligand (spectrum c) is a 6-coordinate complex in which the proximal histidine provides the fifth ligand and the B10 tyrosine provides the sixth ligand.15 Upon addition of cyanide (spectrum a) or imidazole (spectrum b), the sixth ligand is replaced by the exogenous ligand to form a different 6-coordinate low-spin complex. Unlike in ferric vertebrate myoglobins and hemoglobins in which the distal axial ligand is a weakly bound water molecule, in Chlamydomonas hemoglobin the incoming ligand has to replace the intrinsic axial ligand. Thus, the deligation of the B10-tyrosine not only is a rate-limiting step for association of external ligands but also causes a significant structural rearrangement in the distal cavity. The resonance Raman spectrum in the high-frequency region is, however, not sensitive to these conformational changes that occur upon ligand binding. To probe the structural features in the cyanide-bound protein, Raman spectra in the low-frequency region were recorded.
10752 J. Phys. Chem. B, Vol. 104, No. 46, 2000
Das et al.
Figure 3. Resonance Raman difference spectra in the low-frequency region of the cyanide complex of ferric Chlamydomonas hemoglobin mutants at pH 9.5. The difference spectra (12C14N--13C15N-) shown are for Lys87Ala, Tyr63Leu, and Gln 84Gly. Figure 2. Resonance Raman spectra in the low-frequency region of the cyanide complex of ferric Chlamydomonas hemoglobin at pH 7.5. The spectra shown are (a) with 12C14N-, (b) with 13C15N-, (c) difference spectrum 12C14N--13C15N-, and (d) difference spectrum H2O-D2O with 12C14N-.
Resonance Raman Spectra of the Wild-Type Protein in the Low-Frequency Region. The low-frequency region of the resonance Raman spectra of hemeproteins is composed of both in-plane and out-of-plane vibrational modes of the heme including heme propionate modes and ligand vibrational modes.2,17-27 Very often the axial-ligand vibrational modes can be detected as they are enhanced by electronic coupling of the orbitals of the ligand to the heme electronic orbitals. Assignment of a ligand vibrational mode is extremely useful as it directly identifies the particular ligand and allows for the determination of the nature of its interactions with the amino acid residues in the heme pocket.2,28 Figure 2 shows the spectra of cyanidebound Chlamydomonas hemoglobin with two different isotopic compositions. Only the Raman lines at ∼317 and ∼440 cm-1 with 12C14N- (spectrum a) shift upon isotopic substitution with 13C15N- (spectrum b). Although the CN--sensitive lines were partially obscured by overlapping porphyrin internal modes, the difference spectrum (12C14N--13C15N-) clearly shows the presence of prominent features at 444/433 and 319/310 cm-1. The apparent isotope shifts observed in the difference spectrum are greater than the actual value in the individual spectra because of partial cancellation of intensity of the close-lying CN-sensitive bands.29 The actual isotope shift was obtained from the relative intensity of the difference spectra to that of the absolute spectra by a previously reported method.29 In the calculations and associated simulations, the widths (fwhm) of the 315 and 440 cm-1 modes are 8.2 and 11.5 cm-1, respectively, and the isotope frequency shifts were calculated to be ∼2 and ∼4 cm-1, respectively. To determine if the bound cyanide is in the ionic (CN-) or the protonated (HCN) form, we measured the Raman spectra of the cyanide complex in H2O and D2O. The difference spectrum (Figure 2, spectrum d) indicates that no significant perturbation occurs in the frequencies of the cyanide-sensitive modes. Placing a proton on the negatively charged cyanide group is expected to affect the vibrational frequencies of modes involving the motion of the CN group. Thus, these data suggest
that the bound cyanide is unprotonated in Chlamydomonas hemoglobin. The absence of an H/D shift in the CN--complex, however, does not exclude the possibility of hydrogen bonding between the cyanide and one or more of the distal residues. The distal residues in Chlamydomonas hemoglobin (E7Gln and B10Tyr) might facilitate deprotonation of HCN just as in myoglobin in which the distal histidine (E7) has been proposed to have a similar function.30 The deprotonation of HCN has been postulated to be the major kinetic barrier to cyanide binding at neutral pH.30 Resonance Raman Spectra of the Distal Pocket Mutants in the Low-Frequency Region. To investigate how the distal amino acid residues interact with the heme-bound ligand, we measured resonance Raman spectra of the cyanide complex of three distal cavity mutants. The isotope difference spectra (12C14N--13C15N-) are shown in Figure 3 for the Lys87Ala (E10), Tyr63Leu (B10), and Gln84Gly (E7) mutants. First, we compare the frequencies of the modes of the mutants in the 435-452 cm-1 region to that in the wild-type protein. The effect of the E10Lys mutation is minimal as the mode decreases by only ∼1 cm-1. In the B10Tyr mutant, despite a poorer signal quality, a significant shift to lower frequency (∼5 cm-1) of the mode could be readily detected. The E7Gln mutant shows the largest effect in which the frequency of the mode increases by ∼12 cm-1. Notably, this frequency is identical to those observed for the Fe-CN stretching mode in the cyanide complexes of vertebrate hemoglobins31,32 and myoglobins32 (452 cm-1). The isotope difference for this mutant is also larger having a value of ∼11 cm-1. The difference band observed in the 315 cm-1 region does not show any significant frequency change in response to the mutations. The only noticeable feature is that the relative intensity of this band in the E10Lys mutant is higher than those in the wild-type protein and the other two mutants. Assignment of the Isotope-Sensitive Lines. As may be seen in Table 1, Fe-CN stretching modes that have been reported for hemoglobins and myoglobins have frequencies in the 450460 cm-1 region.31-34 On this basis, we assign the line at 440 cm-1 in Chlamydomonas hemoglobin as the Fe-CN stretching mode. The origin of its low frequency as compared to the other globins will be discussed below. The assignment of the line at 315 cm-1 is not as straightforward. Lines in the same region with CN- isotope sensitivity have been detected previously35,36
Cyanide Complex of Ferric Chlamydomonas Hemoglobin
J. Phys. Chem. B, Vol. 104, No. 46, 2000 10753
TABLE 1: Stretching (νFe-CN) and Bending (δFe-CN) Mode Assignments of the FeIIICN Moiety in Cyanide Complex of Various Ferric Hemeproteins type globin
peroxidase
catalase
reductase (siroheme) heme-Cu oxidase model heme P450
protein
νFe-CN
δFe-CN
ref
Chlamydomonas Hba Chlamydomonas Hba E10Lys f Ala Chlamydomonas Hba B10Tyr fLeu Chlamydomonas Hba E7Glnf Gly human adult HbA Chironomus Hbe carp Hb Scapharca Hbf horse Mb HRP, Ig HRP, IIg alkaline HRP, Ih alkaline HRP, IIh HRP Asn70 f Val/Asp HRP His42 f Gln/Glu MPO (pH 7.4) LPO (pH 7.0) beef liver catalase, Ij beef liver catalase, IIj Aspergillus catalase, Ij Aspergillus catalase, IIj sulfite reductase, Ik sulfite reductase, IIk bovinel aa3 T. thermophilusm ba3 unhindered hemen 15-atom strapped hemen 13-atom strapped hemen cytochrome P450, I cytochrome P450, IIo
440 439 435 452 452 453 455 458 452 453-456 460 444 355 ∼450i ∼443i 361 360 434 349 435 412 451 451 478 512 451 447 445 413 434
ndb ndb ndb ndb ndd 412 nr nd ndd 405 422 405 420 ∼403/418i nr 453 453 413 445 412 456 390 352 440 485 nd nd nd 387 343
c c c c 31, 32 34 53 33 32 46, 49 46 46 46 51 52 49 36 47 47 47 47 48 48 32 45 44 44 44 50 50
a Hb from the alga Chlamydomonas eugametos. Abbreviations: ref, reference; Hb, hemoglobin; Mb, myoglobin; HRP, horseradish peroxidase; I, conformer I; II, conformer II; MPO, myeloperoxidase; LPO, lactoperoxidase; nd, not detected; nr, not reported. b The second CN--sensitive band observed at 315, 313, 312, and 314 cm-1, respectively for the wild-type, Lys87Ala, Tyr63Leu, and Gln84Gly is assigned as Fe(III)-N(His) stretching mode rather than Fe-C-N bending mode. c This work. d Although not detected, based on proposed coupling with other porphyrin bands the presence of a bending mode near ∼380 cm-1 has been suggested. e Chironomus thumni thumni hemoglobin III. f The mollusk Hb from Scapharca inaequiValVis. g HRP studied at pH 7.4 and 10.6 in ref 49, at pH 5.5 in ref 46. h At pH 12.5. i Absolute frequencies are not reported; values taken from the difference spectra. j Beef liver catalase at pH 7.0 and Aspergillus niger catalase at pH 4.9. k Escherichia coli sulfite reductase. l Bovine aa3 cytochrome c oxidase. m Thermus thermophilus ba3 cytochrome c oxidase. n In the model hemes, straps refer to covalently linked hydrocarbon chains across the distal side of the heme, unhindered heme has no such strap. o Although the modes at 434 and 343 cm-1 were assigned for the second conformer, it was not reported which one is stretching and which one is bending mode.
and assigned as a six-coordinate Fe-histidine (proximal) stretching mode.35 The Fe-N(His) bond is expected to have significant mixing with the Fe-CN bond at the trans axial position. Assignment of the Fe(III)-N(His) stretching mode in the cyanide complex of Chironomus hemoglobin35 has been proposed at ∼309 cm-1 based on its sensitivity to 54Fe/57Fe isotope substitution. In that case the line was also found to be sensitive (2 cm-1) to cyanide isotopes bound at the trans axial position. A similar cyanide isotope sensitivity of a line at 311 cm-1 was observed in the cyanide complex of lactoperoxidase,36 although the authors suggested that this mode is an out-of-plane vibration which couples to out-of-plane Fe-CN vibrations. We also note that in both ferrous and ferric six-coordinate complexes of various hemeproteins, assignment of the Fe-N(His) stretching mode has been proposed in the 245-320 cm-1 range based on 54Fe/57Fe isotope substitution (∼317 cm-1 in the Chironomus hemoglobin CO complex,37 263 cm-1 in oxymyoglobin,38 274 cm-1 in ferric horseradish peroxidase,39 267 cm-1 in the fluoride complex of horseradish peroxidase,39 and 248 cm-1 in aquometmyoglobin39). On the other hand, the Fe-His mode is not observed in the Raman spectrum of most six-coordinate hemes probably because the mode loses enhancement due to movement of the heme iron into the heme plane. For a bent Fe-C-N structure the Raman intensity of the Fe-C-N bending mode can be enhanced making this a possible assignment for the 315 cm-1 mode as well. However, as may
be seen in Table 1, no line at such a low frequency has been assigned as a bending mode. Furthermore, for the mutants in which large changes are seen in the Fe-CN stretching mode, no significant changes occurred in the mode at 315 cm-1. Large changes in the mutants would be expected for the bending mode but not for a Fe-His stretching mode. Another consideration is the magnitude of the isotope shift. We detect a ∼2 cm-1 shift of the 315 cm-1 mode in the comparison between 12C14N and 13C15N. The predicted shift for a stretching mode consisting of a diatomic oscillator composed of imidazole and the FeCN moiety is ∼2 cm-1 whereas the observed shifts for Fe-C-N bending modes are much larger, typically lying in the 10-15 cm-1 range.32 In the absence of isotopic substitution of the histidine and the iron and a complete normal-mode analysis, it is difficult to make an unequivocal assignment of the 315 cm-1 line in Chlamydomonas hemoglobin; but, based on the above arguments, we tentatively assign it as a mode that involves stretching of the Fe-His bond rather than an Fe-C-N bending mode. The difference bands at 378/370 and 376/369 cm-1 in the E10Lys and E7Gln mutants are in the same region where Hirota et al.32 proposed that the Fe-C-N bending mode lies. It was concluded that its presence is complicated by coupling to several porphyrin modes. We postulate that in the Lys and Gln mutants small structural changes in the distal pocket facilitate coupling of the CN-sensitive mode(s) to the porphyrin mode(s) giving
10754 J. Phys. Chem. B, Vol. 104, No. 46, 2000 rise to the difference feature in the 370 cm-1 region. Thus, it is likely that the Fe-C-N bending mode is in this region rather than at 315 cm-1. Discussion Structure of the Fe-CN Complex. Although CN- and CO are isoelectronic, the Fe(III)-CN- bond in the cyanide complex of ferric hemeproteins behaves differently from the corresponding Fe(II)-CO complexes. Unlike the Fe(II)-CO complex in which π-bonding is very important in determining the bond orders of the Fe-C and C-O bonds, in the Fe(III)-CNcomplex the Fe-C bond is formed mainly by σ donation from 4σ* and 5σ molecular orbitals of CN-; the π-bonding contributions are much weaker than the σ donation.40 Thus, the effects of steric and electrostatic factors on the structure and vibrational properties of FeCO and FeCN- are likely to be very different. The Fe-CO stretching modes appear in the 470-550 cm-1 region while the Fe-CN stretching frequency in various hemeproteins (except heme-copper oxidases) typically fall in the 400-460 cm-1 range, although in a few cases the Fe-CN stretching mode has been assigned at a much lower frequency (see Table 1). Charged and polar groups strongly modulate the Fe-CO stretching mode frequency via back-bonding from the C-O bond. Because of the dominant σ-bonding in the CNcomplex, the occurrence of such an effect is unlikely, consistent with the absence of a back-bonding correlation in the reported vibrational frequencies of Fe-CN and C-N (for a summary of the vibrational frequencies, see ref 32). On the other hand, the Fe-C-N moiety is expected to be more flexible than the Fe-C-O group since the bending vibration of the former is found at much lower energy (it has been assigned in the 350460 cm-1 region in various heme proteins; Table 1) than the bending mode of the Fe-C-O group, which is in the 560590 cm-1 range.2-11,41 This indicates that the Fe-C-N group is more prone to distortion than the Fe-C-O group. Strong support for a variation in the Fe-C-N geometry comes from the crystal structure of various myoglobins and hemoglobins. The Fe-C-N bend angle is observed to vary from 102° to 178° in different structures (for a summary of structural parameters, see ref 42). The cyanide complex of sperm whale myoglobin shows a nearly upright (166°) geometry with a ∼10° tilt of the Fe-C bond from the heme normal.42 The Fe-CN stretching frequency at 452 cm-1 in myoglobin thus can be used as a model representing a linear FeCN geometry. In all the hemoglobins and myoglobins studied to date for which the CN-sensitive modes have been established by isotope substitution, the νFe-CN frequency appears in the narrow range of 452-458 cm-1. Thus, the location of νFe-CN at 440 cm-1 in the wildtype Chlamydomonas hemoglobin represents a significant deviation from the hemoglobin-type behavior and suggests a difference in the structure of its heme-bound cyanide. It should also be noted that this low frequency for the Fe-CN stretching mode does not result from the presence of a proton on the bound cyanide in Chlamydomonas hemoglobin as evidenced from the absence of H/D effect on the CN--sensitive modes. On the basis of these results, we postulated that in Chlamydomonas hemoglobin the Fe-C-N group adopts a bent conformation and tested this postulate by examination of the isotope shift. The magnitude of isotope shift observed in the resonance Raman spectra of the heme-ligand complex gives additional information on its structure. If the Fe-ligand moiety behaves as an ideal two-body harmonic oscillator, the observed isotope shift of the stretching mode can be calculated from the relation ν1/ν2 ) (µ2/µ1)1/2, where µ ) m1m2/(m1 + m2), the reduced mass.
Das et al. This treatment yields a good estimation of the isotope shift only when the Fe-ligand moiety is linear so that the normal modes may be defined as a pure Fe-CN stretch and a pure Fe-C-N bend. However, with a bent structure of the Fe-diatomic ligand complex, the isotope shift deviates from the ideal value as the normal modes can no longer be defined as a pure stretch and bend. Upon substitution of 12C14N- by 13C15N-, an isotope shift of ∼11 cm-1 for Fe-CN stretching vibration is expected assuming a linear structure. Observation of 8-10 cm-1 isotope shift in vertebrate hemoglobin and myoglobin is consistent with the nearly linear structure of Fe-C-N that has been seen in the high-resolution crystal structure of the myoglobin-cyanide complex.42 Thus, a bent structure of the Fe-C-N moiety in Chlamydomonas hemoglobin is supported by the observed ∼4 cm-1 isotope shift for the stretching mode. Interaction between the Cyanide and Residues in the Distal Pocket. The heme distal pocket of Chlamydomonas hemoglobin is congested with the bulky side chain of the B10Tyr along with a glutamine residue at the E7 position. The E7 position in most vertebrate globins is a histidine that provides hydrogen-bonding stabilization to the bound ligands such as O2, CO and CN-. In Chlamydomonas hemoglobin it has been shown that both the B10 and E7 residues interact with the bound oxygen in the oxy complex.14,43 Thus, the bound oxygen can be stabilized by hydrogen bonding and/or steric “trapping,” which is consistent with a very low oxygen dissociation rate. Similarly, the iron-bound cyanide is likely to be influenced by the B10 and E7 residues although the electronic nature of their interactions with cyanide would be expected to be different from that with O2 or CO as discussed earlier. A comparison of the frequencies of the Fe-CN stretching mode in the mutants reveals the nature of distal interactions on the heme-bound cyanide (see Figure 4). First, the E10Lys does not seem to interact with cyanide just as in the oxy complex,14 although it serves as a hydrogen-bond donor to the heme-bound axial tyrosine in the absence of any exogenous ligand.15 A very sizable effect is observed upon mutation of E7 glutamine in which the frequency of the Fe-CN stretching mode shifts to the same value as that in native myoglobin. Given that in myoglobin the Fe-C-N moiety is linear and still hydrogenbonded to the distal histidine,42 the major effect of the distal glutamine on the modulation of the Fe-CN stretching mode in wild-type Chlamydomonas hemoglobin is attributed to a steric effect of the glutamine side chain (-NH2) on the cyanide. It may be noted that for glutamine to exert a steric influence on the cyanide, they must lie close to each other. Thus, the presence of a hydrogen bond between them is likely, although it, per se, is not the major determinant of the reduction in the frequency of the Fe-CN stretching mode. This is consistent with the quantum chemical calculations that predict a very small effect of hydrogen bonding on the Fe-CN vibration but much larger effects from bending of the Fe-C-N moiety.40 Surprisingly, in the B10Tyr f Leu mutant, the frequency of the Fe-CN stretching mode decreased further relative to the wild type. The magnitude of the frequency shift (5 cm-1) is attributed to a change in the FeCN structure in the mutant rather than simply a loss of hydrogen bonding. As discussed above for glutamine, tyrosine is also very likely to be in hydrogenbonding contact with the cyanide in the wild-type protein. We propose that in the absence of the tyrosine in the heme pocket, glutamine moves closer to the axial ligand increasing the steric interaction with the cyanide (Figure 4). Comparison to Other Heme Proteins. Table 1 lists the FeCN vibrational frequencies in various ferric hemeproteins and
Cyanide Complex of Ferric Chlamydomonas Hemoglobin
Figure 4. Proposed structural model of the heme pocket in the cyanide complex of ferric Chlamydomonas hemoglobin and its mutants. The heme, the proximal ligand (His111 at F8), and the distal side amino acid residues (Tyr63 at B10, Lys87 at E10 and Gln84 at E7) are shown for the wild-type protein. As shown here, the Fe-C-N group assumes a bent structure in the wild-type protein and remains unchanged in the E10Lys mutant. Consistent with the Raman data, in the B10Tyr mutant the Fe-C-N moiety is depicted as being more bent than in the wild type, while a linear Fe-C-N structure is postulated to prevail in the E7Gln mutant.
model hemes. Observation of a similar frequency for the FeCN stretching mode in model hemes and globins reflects a nearly linear Fe-C-N linkage. The steric effect on the FeC-N moiety was demonstrated by placing hydrocarbon straps of different lengths across the CN- binding site, and a decrease in the frequency of the Fe-CN stretching mode was observed when the strap length was shortened.44 The frequency of the Fe-CN stretching mode in oxidases32,45 appears at much higher values (478-512 cm-1) presumably due to the formation of bridging cyanide structure between heme a3 and CuB.40 In peroxidases, catalases, and siroheme, the CN--sensitive bands are more complex. Existence of multiple conformers, linear and bent, was invoked to explain the observed isotope shift in horseradish peroxidase,46 beef liver catalase,47 Aspergillus catalase,47 and siroheme sulfite reductase.48 On the basis of normal-mode calculations,46,47 it has been proposed that the linear conformer has a frequency for the Fe-CN stretching mode higher than its bending mode while the bent conformer exhibits a reverse pattern in which the frequency of the bending mode is higher than that of the stretching mode. Differences exist, however, in the assignment of the bending and stretching modes in the bent form, as in siroheme the higher frequency mode was attributed to the stretching vibration based on a normal-mode analysis.48 Frequency reversal of the stretching and bending modes was also proposed for both lactoperoxidase36 and myeloperoxidase,49 although Hirota et al.32 suggested that the higher frequency line (∼450 cm-1) could be the stretching mode. In the cyanide complex of cytochrome P450, multiple conformers were also detected.50 While the CN--sensitive modes shifted in deuterated buffer, indicating that the cyanide was hydrogen-bonded, in the substrate-free form of cytochrome P-450, the frequency of the Fe-CN stretching and bending modes changed very significantly in the substrate-bound forms,
J. Phys. Chem. B, Vol. 104, No. 46, 2000 10755 suggesting steric sensitivity of the bound cyanide.50 Given the complexity of cyanide binding in these systems, it will be useful to study their systematic mutants to obtain additional insights into the nature of the multiple conformers and identify the residues that give rise to the conformers. In horseradish peroxidase, cyanide-binding studies on some of its mutants have been reported.51,52 Mutation of Asn70, the residue that provides a hydrogen-bonding interaction with the distal histidine (His42), causes the Fe-CN stretching mode to shift to lower frequency by (∼5 cm-1) (Table 1), likely due to a repositioning of the distal histidine51 that slightly increases the steric interaction with the bound cyanide relative to that in the wild-type protein. Interestingly, mutation of the distal His42 residue by Gln or Glu caused very significant lowering (∼13 cm-1) of the frequency of the Fe-CN stretching mode (Table 1), which was attributed to disruption of hydrogen bond between His42 and the cyanide.52 Such a large decrease in frequency of the Fe-CN stretching mode is, however, unexpected from a simple disruption of the hydrogen bond. It should be pointed out that substitution of the E7 glutamine by glycine in Chlamydomonas hemoglobin causes exactly the reverse effect on the frequency of the Fe-CN stretching mode that is observed upon mutation of the distal histidine by glutamine in horseradish peroxidase. Both proteins have similar Fe-CN stretching frequencies when glutamine occupies the distal residue position. It is also known that in Chlamydomonas hemoglobin the distal glutamine indeed interacts with the heme-bound exogenous ligands.14,43 Thus, it appears that the steric interaction of distal glutamine plays a major role in modulation of the frequency of the Fe-CN stretching mode in the His42Gln mutant of horseradish peroxidase just as in the wild-type Chlamydomonas hemoglobin. Similarly, it may be argued that in alkaline HRP,46 at a very high pH of 12.5, structural factors in addition to disruption of the hydrogen bond might contribute to the 9 cm-1 decrease in frequency of the Fe-CN stretching mode (Table 1). Summary The present study indicates that the Fe-CN moiety is highly bent in the cyanide adduct of Chlamydomonas hemoglobin. The site-directed mutagenesis studies also demonstrate that the hemebound cyanide interacts with both E7Gln and B10Tyr in Chlamydomonas hemoglobin. The bent structure of the FeC-N linkage is caused by a congested heme pocket, resulting in a steric interaction between the glutamine side chain and the iron-bound cyanide. The cyanide binding in Chlamydomonas hemoglobin clearly reflects a different heme coordination geometry as compared to the linear structure in vertebrate globins. These data substantiate that the heme pocket of Chlamydomonas hemoglobin is very different from that of vertebrate globins, a further indication that their functions are likely very different. Acknowledgment. We thank Martino Bolognesi and Alessandra Pesce (Genove, Italy) for useful discussions. This work was supported by Natural Sciences and Engineering Research Council of Canada Grant 06P0046306 (to M.G.) and by National Institutes of Health Grants GM54806, GM54812 (to D.L.R.). References and Notes (1) Hardison, R. J. Exp. Biol. 1998, 201, 1099-1117. (2) Yu, N.-T.; Kerr, E. A. In Biological Application of Raman Spectroscopy; Spiro, T. G., Ed.; John Wiley & Sons: New York, 1988; Vol. 3, pp 39-95.
10756 J. Phys. Chem. B, Vol. 104, No. 46, 2000 (3) Das, T. K.; Friedman, J. M.; Kloek, A. P.; Goldberg, D. E.; Rousseau, D. L. Biochemistry 2000, 39, 837-842. (4) Ray, G. B.; Li, X.-Y.; Ibers, J. A.; Sessler, J. L.; Spiro, T. G. J. Am. Chem. Soc. 1994, 116, 162-176. (5) Das, T. K.; Lee, H. C.; Duff, S. M.; Hill, R. D.; Peisach, J.; Rousseau, D. L.; Wittenberg, B. A.; Wittenberg, J. B. J. Biol. Chem. 1999, 274, 4207-4212. (6) Morikis, D.; Champion, P. M.; Springer, B. A.; Sligar, S. G. Biochemistry 1989, 28, 4791-4800. (7) Lin, S. H.; Yu, N. T.; Tame, J.; Shih, D.; Renaud, J. P.; Pagnier, J.; Nagai, K. Biochemistry 1990, 29, 5562-5566. (8) Ling, J.; Li, T.; Olson, J. S.; Bocian, D. F. Biochim. Biophys. Acta 1994, 1188, 417-421. (9) Phillips, G. N., Jr.; Teodoro, M. L.; Li, T.; Smith, B.; Olson, J. S. J. Phys. Chem. B 1999, 103, 8817-8829. (10) Sakan, Y.; Ogura, T.; Kitagawa, T.; Fraunfelter, F. A.; Mattera, R.; Ikeda-Saito, M. Biochemistry 1993, 32, 5815-5824. (11) Anderton, C. L.; Hester, R. E.; Moore, J. N. Biochim. Biophys. Acta 1997, 1338, 107-120. (12) Gagne, G.; Guertin, M. Plant Mol. Biol. 1992, 18, 429-445. (13) Couture, M.; Chamberland, H.; St-Pierre, B.; Lafontaine, J.; Guertin, M. Mol. Gen. Genet. 1994, 243, 185-197. (14) Couture, M.; Das, T. K.; Lee, H. C.; Peisach, J.; Rousseau, D. L.; Wittenberg, B. A.; Wittenberg, J. B.; Guertin, M. J. Biol. Chem. 1999, 274, 6898-6910. (15) Das, T. K.; Couture, M.; Lee, H. C.; Peisach, J.; Rousseau, D. L.; Wittenberg, B. A.; Wittenberg, J. B.; Guertin, M. Biochemistry 1999, 38, 15360-15368. (16) Couture, M.; Guertin, M. Eur. J. Biochem. 1996, 242, 779-787. (17) Spiro, T. G.; Li, X.-Y. In Biological Application of Raman Spectroscopy; Spiro, T. G., Ed.; John Wiley & Sons: New York, 1988; Vol. 3, pp 1-37. (18) Li, X.-Y.; Czernuszewicz, R. S.; Kincaid, J. R.; Su, Y. O.; Spiro, T. G. J. Phys. Chem. 1990, 94, 31-47. (19) Abe, M.; Kitagawa, T.; Kyogoku, Y. J. Chem. Phys. 1978, 69, 4526-4534. (20) Kitagawa, T.; Abe, M.; Ogoshi, H. J. Chem. Phys. 1978, 69, 45164525. (21) Li, X.-Y.; Czernuszewicz, R. S.; Kincaid, J. R.; Stein, P.; Spiro, T. G. J. Phys. Chem. 1990, 94, 47-61. (22) Li, X.-Y.; Czernuszewicz, R. S.; Kincaid, J. R.; Spiro, T. G. J. Am. Chem. Soc. 1989, 111, 7012-7023. (23) Hu, S.; Mukherjee, A.; Piffat, C.; Mak, R. S. W.; Li, X.-Y.; Spiro, T. G. Biospectroscopy 1995, 1, 395-412. (24) Hu, S.; Morris, I. K.; Singh, J. P.; Smith, K. M.; Spiro, T. G. J. Am. Chem. Soc. 1993, 115, 12446-12458. (25) Hu, S.; Smith, K. M.; Spiro, T. G. J. Am. Chem. Soc. 1996, 118, 12638-12646. (26) Kitagawa, T. In Biological Application of Raman Spectroscopy; Spiro, T. G., Ed.; John Wiley & Sons: New York, 1988; Vol. 3, pp 97131. (27) Rousseau, D. L.; Friedman, J. M. In Biological Application of Raman Spectroscopy; Spiro, T. G., Ed.; John Wiley & Sons: New York, 1988; Vol. 3, pp 133-215.
Das et al. (28) Wang, J.; Caughey, W. S.; Rousseau, D. L. In Methods in Nitric Oxide Research; Feelisch, M., Stamler, J. S., Eds.; John Wiley & Sons Ltd.: New York; 1996; pp 427-454. (29) Rousseau, D. L. J. Raman Spectrosc. 1981, 10, 94-99. (30) Brancaccio, A.; Cutruzzola´, F.; Allocatelli, C. T.; Brunori, M.; Smerdon, S. J.; Wilkinson, A. J.; Dou, Y.; Keenan, D.; Ikeda-Saito, M.; Brantley, R. E., Jr.; Olson, J. S. J. Biol. Chem. 1994, 269, 13843-13853. (31) Henry, E. R.; Rousseau, D. L.; Hopfield, J. J.; Noble, R. W.; Simon, S. R. Biochemistry 1985, 24, 5907-5918. (32) Hirota, S.; Ogura, T.; Shinzawa-Itoh, K.; Yoshikawa, S.; Kitagawa, T. J. Phys. Chem. 1996, 100, 15274-15279. (33) Boffi, A.; Chiancone, E.; Takahashi, S.; Rousseau, D. L. Biochemistry 1997, 36, 4505-4509. (34) Yu, N. T.; Benko, B.; Kerr, E. A.; Gersonde, K. Proc. Natl. Acad. Sci. U.S.A. 1984, 81, 5106-5110. (35) Kerr, E. A.; Yu, N. T.; Gersonde, K. FEBS Lett. 1984, 178, 3133. (36) Hu, S.; Treat, R. W.; Kincaid, J. R. Biochemistry 1993, 32, 1012510130. (37) Gersonde, K.; Kerr, E.; Yu, N. T.; Parish, D. W.; Smith, K. M. J. Biol. Chem. 1986, 261, 8678-8685. (38) Walters, M. A.; Spiro, T. G. Biochemistry 1982, 21, 6989-6995. (39) Teraoka, J.; Kitagawa, T. J. Biol. Chem. 1981, 256, 3969-3977. (40) Kushkuley, B.; Stavrov, S. S. Biochim. Biophys. Acta 1997, 1341, 238-250. (41) Rajani, C.; Kincaid, J. R. J. Am. Chem. Soc. 1998, 120, 72787285. (42) Bolognesi, M.; Rosano, C.; Losso, R.; Borassi, A.; Rizzi, M.; Wittenberg, J. B.; Boffi, A.; Ascenzi, P. Biophys. J. 1999, 77, 1093-1099. (43) Das, T. K.; Couture, M.; Guertin, M.; Rousseau, D. L., unpublished data. (44) Tanaka, T.; Yu, N. T.; Chang, C. K. Biophys. J. 1987, 52, 801805. (45) Surerus, K. K.; Oertling, W. A.; Fan, C.; Gurbiel, R. J.; Einarsdo´ttir, O.; Antholine, W. E.; Dyer, R. B.; Hoffman, B. M.; Woodruff, W. H.; Fee, J. A. Proc. Natl. Acad. Sci. U.S.A. 1992, 89, 3195-3199. (46) Al-Mustafa, J.; Kincaid, J. R. Biochemistry 1994, 33, 2191-2197. (47) Al-Mustafa, J.; Sykora, M.; Kincaid, J. R. J. Biol. Chem. 1995, 270, 10449-10460. (48) Han, S. H.; Madden, J. F.; Siegel, L. M.; Spiro, T. G. Biochemistry 1989, 28, 5477-5485. (49) Lopez-Garriga, J. J.; Oertling; W. A.; Kean, R. T.; Hoogland, H.; Wever, R.; Babcock, G. T. Biochemistry 1990, 29, 9387-9395. (50) Simianu, M. C.; Kincaid, J. R. J. Am. Chem. Soc. 1995, 117, 46284636. (51) Mukai, M.; Nagano, S.; Tanaka, M.; Ishimori, K.; Morishima, I.; Ogura, T.; Watanabe, Y.; Kitagawa, T. J. Am. Chem. Soc. 1997, 119, 17581766. (52) Tanaka, M.; Ishimori, K.; Mukai, M.; Kitagawa, T.; Morishima, I. Biochemistry 1997, 36, 9889-9898. (53) Tsubaki, M.; Srivastava, R. B.; Yu, N. T. Biochemistry 1982, 21, 1132-1140.