Static Normal Coordinate Deformations of the Heme Group in Mutants

May 18, 2007 - Department of Chemistry, Drexel University, 3141 Chestnut Street, Philadelphia, ... Department of Chemistry, University of Puerto Rico,...
0 downloads 0 Views 200KB Size

J. Phys. Chem. B 2007, 111, 6527-6533


Static Normal Coordinate Deformations of the Heme Group in Mutants of Ferrocytochrome c from Saccharomyces cereWisiae Probed by Resonance Raman Spectroscopy Reinhard Schweitzer-Stenner,*,† Qing Huang,†,‡ Andrew Hagarman,† Monique Laberge,§ and Carmichael J.A. Wallace| Department of Chemistry, Drexel UniVersity, 3141 Chestnut Street, Philadelphia, PennsylVania 19104, Department of Chemistry, UniVersity of Puerto Rico, Rı´o Piedras Campus, San Juan, Puerto Rico 00934, Department of Biochemistry and Biophysics and Johnson Research Foundation, UniVersity of PennsylVania Medical Center, Philadelphia, PennsylVania 19104-6059, Department of Biochemistry and Molecular Biology, Dalhousie UniVersity, NoVa Scotia, Canada B3H 4H7 ReceiVed: January 18, 2007; In Final Form: March 27, 2007

The function of heme proteins is, to a significant extent, influenced by the ligand field probed by the heme iron, which itself can be affected by deformations of the heme macrocycle. The exploration of this field is difficult because the heme structure obtained from X-ray crystallography is not resolved enough to unambiguously identify structural changes on the scale of 10-2 Å. However, asymmetric deformations in this order of magnitude affect the depolarization ratio of the resonance Raman lines assignable to normal vibrations of the heme group. We have measured the dispersion of the depolarization ratios of four structure sensitive Raman bands (i.e., ν4, ν11, ν21, and ν28) in yeast iso-1-ferrocytochrome c and its mutants N52V, Y67F, and N52VY67F with B- and Q-band excitation. The DPR dispersion of all bands indicates the presence of asymmetric in-plane and out-of-plane deformations. The replacement of the polar tyrosine residue at position 67 by phenylalanine significantly increases the triclinic B2g deformation, which involves a distortion of the pyrrole symmetry. We relate this deformation to changes of the electronic structure of pyrrole A, which modulates the interaction between its propionate substituents and the protein environment. This specific heme deformation is eliminated in the double mutant N52VY67F. The additional substitution of N52 by valine induces a tetragonal B1g deformation which involves asymmetric changes of the Fe-N distances and increases the rhombicity of the ligand field probed by the heme iron. This heme deformation might be caused by the elimination of the water-protein hydrogen-bonding network in the heme cavity. The single mutation N52V does not significantly perturb the heme symmetry, but a small B1g deformation is consistent with our data and the heme structure obtained from a 1 ns molecular dynamics simulation of the protein.

Introduction It is well-established that the active site(s) of heme proteins, generally an iron protopophyrin IX derivative, is subject to a variety of perturbations, which lower its symmetry from ideal D4h.1-3 This particularly applies to cytochrome c, which belongs to an important class of redox active proteins. More than 30 years ago, Sutherland and Klein, by means of magnetic dichroism measurements,4 observed a ∼120 cm-1 splitting of the Q band of ferrocytochrome c at low cryogenic temperatures, which was later found to be mirrored by the resonance Raman excitation profile of the ν15 mode.5 In their investigation of band c-type cytochromes, Wagner and Kassner suggested that the proximal histidine was primarily responsible for lifting the degeneracy of the Q band.6 The depolarization ratios (DPRs) of various resonance Raman lines exhibit a very pronounced dispersion particularly in the Q-band region, which reflect inplane as well as out-of-plane deformations of the heme * Corresponding author. Phone: 215-895-2268. Fax: 215-895-1265. E-mail: [email protected] † Drexel University. ‡ University of Puerto Rico. § University of Pennsylvania Medical Center. | Dalhousie University.

macrocycle.7-10 The relevance of this finding was demonstrated by Shelnutt and co-workers, who described comparatively large out-of-plane deformations (ruffling and waving), found to be mostly conserved for cytochrome c species in which the amino acid residues between the cysteine linkages to the heme group were homologous.11,3 Vanderkooi and associates subsequently used semiempirical calculations to rationalize the Q-band splitting in low temperature absorption spectra in terms of the influence of the internal electric field in the heme pocket.13 More recently, Levantino et al. measured the Q- and Qv-band region of the absorption spectra of horse heart (hhc) and yeast cytochrome c (yc) at 10 K and analyzed the spectra in terms of vibronic coupling theory, which takes into account electronic as well as vibronic perturbations of the heme macrocycle.12 Evidence was provided for the former to reflect the gradient inhomogeneity of the internal electric field at the heme (i.e., different gradients along the two perpendicular N-Fe-N heme axes), giving rise to an additional electronic coupling of B1g symmetry.13 The electronic perturbation (and thus the field gradient difference) was found to be substantially larger for hhc (95 cm-1) than for yc (50 cm-1), in agreement with the theoretical calculations of Laberge et al.14

10.1021/jp070445a CCC: $37.00 © 2007 American Chemical Society Published on Web 05/18/2007

6528 J. Phys. Chem. B, Vol. 111, No. 23, 2007

Figure 1. Heme c group of wild-type iso-1-cytochrome c and its protein environment represented by the two axial heme ligands H18 (green), M80 (yellow), and the functionally important residues N52, Y67, and G41. G41 donates a hydrogen bond to the propionate group of pyrrole A. The figure was generated using Protein Data Bank file 1ycc, the X-ray structure of yeast iso-1-ferrocytochrome c.17

The internal electric field, the heme cavity structure, and the redox potential of (Saccharomyces cereVisiae) cytochrome c can be modified by point mutations in the interior of the protein.15,16,18 We selected two residues for replacement, namely N52 and Y67, as shown in Figure 1.17 Both residues, via their respective polar CO and OH groups, are incorporated in a hydrogen-bonding network of water molecules which shields the heme group from the dipoles in its environment, thus stabilizing the ferrous state of the heme iron.18 Substitutions of N52 by alanine, valine, and isoleucine substantially modify the hydrogen-bonding network in the heme cavity, whereas it is completely eliminated by the double mutation N52VY67F.15 It has been shown that all of these mutations lower the redox potential, thus favoring the oxidized state. Interestingly, this is predominantly associated with a positive entropic contribution to the respective Gibbs energy rather than with enthalpic changes, which rather favor the reduced state.18 It is reasonable to assume that the positive ∆S reflects a larger conformational flexibility. Thus, all of these mutations make yc more similar to hhc, which has a much lower redox potential and exhibits more conformational heterogeneity. We have investigated the modification of the electronic perturbation of the heme group caused by the above mutations of yc by low temperature absorption spectroscopy.19 All mutations resulted in an increase of the quadrupole moment of the internal electric field and thus of the Q-band splitting asymmetry. We showed that the mutation-induced variation of the electronic perturbation correlates nonlinearly with the enthalpic contribution to the redox potential. The present study is aimed at identifying and comparing the respective in-plane static normal coordinate deformations (SNCD) of the mutant’s heme group to investigate how the previously identified structural changes of the heme pocket affect the heme conformation in the ferro state. To this end, we measured and analyzed the DPR dispersion of some prominent resonance Raman lines at different excitation wavelengths between 442 and 568 nm. Thus, our data covers the region of Q-band scattering for which the DPR is very sensitive to changes of heme deformations.2,20

Schweitzer-Stenner et al.

Figure 2. Polarized resonance Raman spectra of wild-type iso-1ferrocytochrome c and the indicated mutants in the region between 1100 and 1700 cm-1 measured with 514 nm (Qv) excitation. Red line: spectra taken with j ) x polarization (parallel to the laser polarization). Blue line: spectra taken with j ) y polarization (perpendicular to the laser polarization). The band assignment is based on the study of Hu et al.25 A spectral decomposition (black solid line) is shown for the two wild-type spectra.

Materials and Methods Materials. Proteins were prepared by expression of plasmidborne mutant yeast iso-1-cytochrome c (cyc-1) genes in the host S. cereVisae GM3C2,21 which lacks the wild-type gene. Cytochromes were purified, after yeast cell lysis and extraction or precipitation of most other soluble components, by cationexchange chromatography.22 In addition to the indicated mutations, all cytochromes thus prepared contain a C102T mutation to ensure that dimerization via disulfide formation does not occur. For absorption and resonance Raman experiments, the protein was dissolved in 0.01 M tris buffer at pH 8 and reduced with a small amount of sodium dithionite. The final concentrations for the resonance measurements were 0.5 mM. Spectroscopy. The set up and the performance of polarized resonance Raman measurements23 and the spectral analysis with MULTIFIT are described in detail in an earlier publication.24 The terms x and y polarization are used for describing the Raman scattering parallel (Ix) and perpendicular polarized (Iy) to the polarization of the exciting laser beam. The DPR was calculated as


Ix Iy


Results We have measured the x- and y-polarized resonance Raman spectra of yc and its mutants N52V, Y67F, and N52VY67F with 442, 514, 521, 531, and 568 nm excitation. Figure 2 shows the respective spectra taken at 514 nm (Qv band) excitation in the wavenumber region between 1100 and 1700 cm-1. The x-polarized spectra are dominated by bands assignable to B1gand B2g-type modes, whereas bands from A2g modes dominate

Deformations of the Heme Group

J. Phys. Chem. B, Vol. 111, No. 23, 2007 6529

Figure 3. Visible and near UV absorption spectra of wild-type (black line) and Y67F (red line) yeast ferrocytochrome c. The wavenumbers are expressed in units of cm-1 × 103.

the y-polarized spectra. Differences between the wavenumber positions are negligibly small (e (1 cm-1), but the intensity distributions are significantly different. This reflects the different Q0- and Qv-resonance positions of the different mutants, as demonstrated by the absorption spectra of wild-type and Y67F ferrocytochromes in Figure 3. Generally, all of the investigated mutations cause a spectral redshift of the Q-band profile, which reflects the reduced polarity of the heme environment.19 The 514 nm excitation is very close to the Qv resonance of the depicted Raman bands, so that small shifts cause a significant increase or decrease of the band intensities.9,10 In the following section, we mostly confine ourselves to analyzing the DPR dispersion of ν4 (A1g, 1360 cm-1), ν11 (B1g, 1546 cm-1), ν28 (B2g, 1171 cm-1), and to a minor extent ν21 (A2g, 1313 cm-1),25 since their DPR values could be determined with the highest accuracy and are easiest to interpret. The DPR value of another prominent marker band, that is, ν10 at 1621 cm-1, is too difficult to determine for 531 nm excitation, owing to this band’s very low intensity in the y-polarized spectrum. Figure 4 illustrates the self-consistent spectral decomposition of the polarized Raman spectra for the wild-type spectra recorded with 442 and 521 nm excitations. All spectra of a specific protein were reproduced in terms of Voigtian band profiles with identical wavenumbers and half widths.24,28 As shown earlier, this allows the determination of the integrated intensities and thus also DPRs of even heavily overlapping bands.20,24 The bands selected for further analysis are indicated. Figure 5 shows the DPR dispersion of ν4 (A1g, 1360 cm-1), ν11 (B1g, 1546 cm-1), ν28 (B2g, 1171 cm-1), and ν21 (A2g, 1313 cm-1) of all cytochrome c derivatives investigated. We focus first on ν4 and ν11. At 442 nm excitation (preresonance B0), all DPRs of ν11 are close to the value of 0.75, the expected value for an ideal D4h symmetry. The double mutant N52VY67F is an exception in that its ν11 DPR value of 0.57 is substantially lower. For ν4, the respective DPR is 0.14 for N52V and clusters for the remaining mutants between 0.16 and 0.17, which is within the uncertainty of (0.01. The DPRs of both modes increase with decreasing wavenumber toward a maximum between the respective Q0 and Qv wavenumbers. The statistical error of the respective values increases concomitantly and reaches (0.5 for ν4 at 521 nm and (0.3 for ν11 at 531 nm, owing to the relatively low intensities of these bands and a substantial overlap with rather intense neighboring bands. The

Figure 4. Spectral decomposition of the polarized resonance Raman spectra of wild-type yeast iso-1-ferrocytochrome c taken with 442 and 521 nm excitation. All spectra were fitted by using identical halfwidths, band profiles, and wavenumber positions for corresponding bands in the different spectra.

experimental uncertainty of the ν4 DPR values specifically results from the overlap with a band at 1367 cm-1 25,26 assignable to a CH3 bending mode. It most likely gains resonance enhancement due to vibrational mixing with local vibrations contributing to the eigenvector of the ν4 mode. A similar mixing effect has earlier been observed for Ni(II)-octethylporphyrins with NO2 substituents at the meso-carbons.27 As reported earlier,9,10 ν4 becomes inverse polarized at excitation wavelengths between Q0 and Qv. At 514 and 521 nm excitation, the DPR values of wild type, N52V, and Y67F are practically identical (within their statistical uncertainty), whereas the respective values of N52VY67F are somewhat lower. The DPR values of ν4 become rather large and unreliable at 531 nm; hence, they have been omitted in Figure 5. For ν11, wild type and N52V cytochrome c exhibit very similar DPR values in the Qv-band region, whereas the Y67F mutant exhibits a significantly larger DPR value at 531 nm. The DPR values of the double mutant N52VY67F are systematically lower than those of the other cytochrome c derivatives at all excitation wavelengths investigated. The DPRs of ν29 are particularly interesting because a significant dispersion was only observed for wild-type cytochrome c, that is, an increase of the DPR value with decreasing wavenumber. The DPR value at 531 nm is 1.5. For the mutants N52V and N5VF67F, the DPR values are close to the D4h values of 0.75 at all excitation wavelengths. The respective DPR of Y67F cytochrome c shows some dispersion in that it slightly

6530 J. Phys. Chem. B, Vol. 111, No. 23, 2007

Schweitzer-Stenner et al. wavenumber. The tensor elements are calculated as the coherent superposition of all scattering amplitudes brought about by Herzberg-Teller and Jahn-Teller coupling within and between Q and B states.29 Resonance Raman scattering due to eq 2 yields a DPR value of 0.75 independent of the excitation wavelength. Deviations from this value thus reflect admixtures of tensor elements Γ(ν˜ L) related to other symmetries Γ into the tensor because of symmetry lowering perturbations of the environment.28 Since different mixtures of scattering processes (Franck-Condon, Jahn-Teller, and Herzberg-Teller) are associated with tensor elements of different symmetries, the DPR becomes dependent on the excitation wavenumber.2,29 The case of lowest symmetry is reflected by

R(ν11) )


b1g(ν˜ L) + a1g(ν˜ L) b2g(ν˜ L) + a2g(ν˜ L) b2g(ν˜ L) - a2g(ν˜ L) -b1g(ν˜ L) + a1g(ν˜ L)



The corresponding DPR can be expressed as28 2 2 2 2 3 a1g (ν˜ L) + 5a2g (ν˜ L) + 2[b1g (ν˜ L) + b2g (ν˜ L)] (4) F) ‚ 4 6a1g2(ν˜ L) + 2[b1g2(ν˜ L) + b2g2(ν˜ L)]

Figure 5. DPR dispersion of the resonance Raman bands assignable to the modes ν4 (A1g), ν11 (B1g), ν28 (B2g), and ν21 (A2g) for the ferro state of wild-type yeast iso-cytochrome c (black), N52V (blue), Y67F (red), and N52VY67F (green). Note that the plot for ν4 is semilogarithmic.

decreases toward lower wavenumbers. At 531 nm, the DPR value is 0.62. The DPR of ν21 shows the usual behavior for all cytochrome c derivatives, namely, a nearly inverse polarization, which substantially decreases toward higher wavenumbers.9,10 We consider all DPR values of wild type, N52V, and Y67F as practically identical within their respective experimental uncertainty ((0.3 at 442 nm and (5 at 531 nm). The 531 nm DPR value of N52VY67F is significantly smaller than those of the other cytochrome c derivatives. Discussion To interpret our result, we first focus on ν11. In D4h, the corresponding Raman tensor can be written as

R(ν11) )


b1g(ν˜ L) 0 -b1g(ν˜ L) 0



where the tensor element b1g(ν˜ L) depends on transition dipole moments, the vibronic coupling operators of the respective Raman mode, and specific frequency functions related to the contributing vibronic coupling processes.28,29 ν˜ L is the excitation

Equation 4 implies that admixture of b2g(ν˜ L) alone does not change the DPR of 0.75. The lower values observed with 442 nm excitation result from symmetric a1g(ν˜ L) contributions, whereas antisymmetric a2g(ν˜ L) scattering causes DPRs larger than 0.75.2 It should be noted that a2g(ν˜ L) contributions are generally detectable only in the Q-band region (514, 521, and 531 nm) for reasons provided in earlier studies.2,9,10 In principle, a1g(ν˜ L) contributes at all excitation wavelengths, but predominantly in the B-band region, because the strong dipole moment, associated with the transition into the B state, gives rise to a relatively large scattering amplitude, caused by intrastate Franck-Condon coupling.20,29 Hence, a1g(ν˜ L) admixtures to the Raman tensor can best be inferred from DPR values observed with B-band excitation (442 nm) and generally causes DPR values to be smaller than 0.75. Apparently, the DPR dispersion of ν11 predominantly reflects a2g(ν˜ L) contributions to the Raman tensor, though some a1g(ν˜ L) admixture is discernible for N52VY67F. As shown in numerous earlier studies, the composition of the Raman tensor, discussed above, predominantly reflects vibronic perturbations that can be written as12,20,30

VΓi )

∂2H ˆ el ∂qΓr r∂qΓi i

δqΓi r +

∂ 3H ˆ el ∂qΓr r∂qΓi i∂qΓj j

δqΓi iδqΓj j


where H ˆ el is the pure electronic Hamiltonian, qΓr r denotes the normal coordinate of the Raman active mode under consideration, Γr is its symmetry in D4h, and δqΓi i and δqΓj j are SNCDs along the ith and jth normal coordinate of symmetry Γi and Γj in D4h. The symmetry Γ of the perturbation is given by the product representations Γr X Γi and Γr X Γi X Γj. The first term accounts for in-plane deformations, and the second term accounts for out-of-plane deformations. Hence, in the case of the B1g mode ν11, the tensor elements a2g(ν˜ L) and a1g(ν˜ L) reflect SNCDs exhibiting of B2g and B1g symmetry or the simultaneous occurrence of A2u and B1u or A2u and B2u, respectively. The group theoretical possible combinations containing A1u deformations can be neglected because deformations of this symmetry are generally negligible.11

Deformations of the Heme Group

J. Phys. Chem. B, Vol. 111, No. 23, 2007 6531 deformations (a1g(νˆ L) increases) or reduced A1g deformations (b1g(νˆ L) decreases). In order to differentiate between all of these options, we utilize the DPRs of the B2g-type mode ν28 (1172 cm-1) observed with 531 nm excitation (Figure 5). The comparatively low DPR value of Y67F reflects an admixture of a1g(ν˜ L) to the Raman tensor. The corresponding vibronic coupling term is written as



) Qx

∂2H ˆ el

B B y ∂qV 2g∂qi 2g 28


δqBi 2g By qB2g



and reflects a triclinic B2g deformation (Figure 6). Hence, the respective DPR values of ν11 and ν28 yield an unambiguous picture in that they both indicate that the Y67F mutation induces an additional B2g-type triclinic deformation of the heme group (compared with the wild type). For the double mutant N52VY67F, we observed DPR values slightly above 0.75 at all Qv-band excitation. It reflects vibronic coupling of A2g symmetry:


Figure 6. Planar deformations of the heme macrocycle along the normal coordinates of the lowest energy B1g (upper figure) and B2g mode (lower figure) as used in the NSD analysis of heme groups and porphyrins as obtained from J.A. Shelnutt,

As shown by Shelnutt and co-workers, the pattern of SNCDs are generally dominated by displacements along normal coordinates of the lowest frequency modes of a distinct symmetry.11 This results from the fact that the energy to be invested increases with the square of the frequency or wavenumber.11,20 Figure 6 shows the pattern of a rhombic B1g and a triclinic B2g deformation inferred from the eigenvector of the lowest frequency modes of these symmetries.31,33 In the following discussion, we focus on these two types of in-plane deformations. The DPR values of ν11 at 531 nm reflect a hierarchy Y67F > wild type, N52V, and N52VY67F with respect to the ratio RA2g/RB2g. As mentioned above, DPR values larger than 0.75 reflect an admixture of A2g-type vibronic coupling, which per se is assignable to in-plane triclinic B2g deformations (Figure 6) due to


cAes2g ) Qx y

∂ 2H ˆ el

∂qBV111g∂qBi 2g


δqBi 2g By qB2g



where qB1g denotes the vibrational matrix element in the electronic ground state. In principle, the larger DPR value of Y67F cytochrome c can reflect larger B2g deformations or result from decreasing B1g, A2g, and A1g deformations, which add a1g(νˆ L), b2g(νˆ L), and b1g(νˆ L) contributions to the total Raman tensor. However, in view of the very similar DPR values of the wildtype, N52V, and Y67F cytochromes c at 442 nm excitation, the B1g deformations should be similar for the respective heme groups. For the N52VY67F, the DPR values are consistently lower than those of the wild type, indicative of larger B1g


) Qx y

∂2H ˆ el

∂qBV282g∂qBi 1g


δqBi 1g By qB1g



and is due to rhombic B1g-type deformations. Thus, a consistent picture is again obtained, which suggest that the double mutation induces some (additional) deformation of this symmetry. With respect to the single mutant Y67F, this reduces the DPR at 531 nm. A more quantitative analysis requires the modeling of the DPR dispersion in terms of vibronic coupling parameters cΓes. This normally requires the utilization of a complex formalism which describes Q-band scattering in terms of multimode mixing and a nonadiabatic perturbation approach.32 However, the multimode contributions to the scattering amplitude mostly affect the DPR values of A2g and A1g modes.20,30,32 We therefore used the single mode model of Huang et al.33 to simulate the DPR dispersion of the ν11 mode for the wild type, Y67F, and N52VY67F. To this end, the tensor elements in eq 4 were expressed in terms of the respective resonance excitation profiles of Franck-Condon, Jahn-Teller, and Herzberg-Teller coupling to describe resonance enhancement for Q0 and Qv excitation.33 The respective theory is based on third-order time dependent perturbation theory approach.34 For the sake of simplicity, we assumed that all three derivatives exhibit the same HerzbergTeller coupling for B1g-symmetry; that is, the matrix elements cBQB1g were assumed to be identical and were arbitrarily assumed to be 1. In the framework of the four-orbital model, the 1g respective Jahn-Teller coupling is cBQQ ) -sin V‚cBQB1g, where ν can be obtained from the intensity ratio of the Q0 and B band.32 The simulated DPR dispersion for the above derivatives, for which we also considered the different Q0- and Qv-resonance positions of Lorentzian excitation profiles inferred from the respective absorption spectra,19 are shown in Figure 7. For the wild type, we had to assume cAQB2g ) 0.12‚cBQB1g, cAQB1g ) -0.12‚ 1g ) 0.012‚cBQB1g. These parameters reflect B2g cBQB1g, and cABB1g ) cAQQ and B1g deformations, respectively. The Y67F mutation brings about an increase in antisymmetric coupling; that is, cAQB2g ) 0.18‚cBQB1g, which indicates that a 1.5 increase of δqB2g is occurring. The corresponding parameters of symmetric A1g-type coupling are all reduced by 40%. For the double mutant cAQB2g is the same as for the wild type, but all A1g parameters had to be increased by a factor of 1.6, indicating a corresponding increase of δqB1g.

6532 J. Phys. Chem. B, Vol. 111, No. 23, 2007

Schweitzer-Stenner et al.

Figure 7. Simulation of the DPR dispersion of the ν11 mode of wildtype iso-1-ferrocytochrome c (black) and its mutants Y67F (red) and N52VY67F (green).

TABLE 1: Total In-Plane B1g and B2g Deformations in Angstroms of the Heme Group of Wild-Type Iso-1-ferrocytochrome c and Its Y67F and N52VY67F Mutants Obtained from the Indicated PDB Files B1g wild type (1ycc) Y67F (1ctz) N52VY67F (1crj)


-0.07 -0.02

-0.09 0.11

Two observations are noteworthy with respect to the DPR values of ν4. First, we address the surprisingly large DPR observed with Qv-resonance excitation (521 nm). This phenomenon has been observed earlier by Bobinger et al.9 and Schweitzer-Stenner et al.,10 who attributed it to a very large antisymmetric A2g deformation of the heme. Later, Hu et al. argued that the observed DPR dispersion is an artifact in that it results from an overlap with a band assignable to a CH3 bending mode.25 Schweitzer-Stenner, however, showed that the two bands could be disentangled and confirmed the anomalous behavior of the DPR, which he now explained as resulting from the combined presence of ruffling and saddling out-of-plane deformations.26 More recently, the influence of these out-ofplane deformations on the DPR of A1g-type modes was demonstrated for Ni(II)-octaethyltetraphenylporphyrin20 and for Ni(II)-octaethylporphyrin with NO2-substitution at meso positions.27,30 It is difficult to derive any specific structural information from the DPR dispersion of the ν21 mode without a detailed analysis of its scattering amplitudes. However, in general terms, the observed deviations from the D4h value of the DPR (i.e., a theoretically infinite DPR value, which in fact can be as low as 20 depending on the acceptance angle of the collimator optic) can be caused by all of the aforementioned heme deformations. The slightly lower DPRs at 531 nm observed for ν4 and ν21 N52VY67F cytochrome c can be attributed to the B1g deformation derived above which admix b2g(νˆ L) to the Raman tensor. We inferred the in-plane deformations of the wild type (1ycc), Y67F (1ctz) and the double mutant N52VY67F (1crj) by using the web-based decomposition program on Shelnutt’s home page.3,11,31 It yields the total deformation along a given normal coordinate of a distinct (D4h) symmetry in units of Å. We focused on the deformations along the coordinates associated with the lowest frequency modes of the heme macrocycle. The respective values are listed in Table 1. It should be noted in this context once again that (a) the in-plane deformations should be used with caution owing to their large uncertainty and (b) the crystal structure might not represent the dynamical average

of the heme structure.11 Shelnutt’s analysis reveals a significant B2g-type deformation for Y67F, whereas this deformation type is negligible for the wild-type heme. This is in qualitative accordance with the conclusions drawn from our data. Moreover, this analysis did not reveal any B2g deformation for the wildtype heme and is not in agreement with our data. Recently, we analyzed the structure of the wild-type heme obtained from a 1 ns molecular dynamics (MD) simulation.12,19 The corresponding total B1g and B2g deformations are 0.03 and 0.07 Å, respectively. If we take these values as reference, we derive a total B2g deformation of 0.1 Å for Y67F and a total B1g deformation of 0.05 Å for the double mutant. One should not take the absolute values of these deformations too seriously, but they show that changes on the scale of 10-2 Å lead to substantial changes of the DPR dispersion. These changes show that our spectroscopic method is much more sensitive to structural changes than X-ray crystallography. As known from the crystal structures of wild-type and mutant Y67F, the mutation substitutes the S(M80)-Y(67) hydrogen bonding in the former by a S(M80)-W300 bond, where W300 is a newly added water molecule. This substitution might cause the change of the electric field gradient discussed in our recent paper.19 In the context of the present study, changes of the interaction between the propionate group of pyrrole A and the protein are likely to be more relevant, since the corresponding hydrogen-bonding network is substantially affected by the Y67F mutation.16 Berghuis et al. concluded from their crystallographic data that the mutation alters the charge distribution on the propionate carboxyl group by modifying the π-electron distribution of pyrrole A,16 thus adopting a mechanism proposed by Moore.35 Such an electronic perturbation of the heme’s ground state, which should be distinguished from perturbations of the excited state, would give rise to an overall B2g deformation,2 since it increases the nonequivalence of the pyrrole rings.2 We therefore suggest that the observed increase of the B2g deformation induced by the Y67F mutations reflects the electronic changes that modify the pyrrole A interaction with its heme environment. The crystal structure of the yeast cytochrome c mutant N52IY67F might be used as a substitute for our double mutant, N52VY67F.15 In the former, a nearly complete elimination of the water-protein hydrogen network has been observed, which certainly creates a much more hydrophobic heme environment. Boulin and Wallace argued that this might induce Coulomb interactions between negatively charged propionate and the positive iron charge, thus stabilizing the oxidized state.18 A direct correlation between this perturbation and the rhombic B1g deformation obtained from our data cannot be derived. However, the crystallographic data for N52IY67F are not indicative of the electronic perturbation of pyrrole A, which is consistent with the fact that the additional B2g deformation induced by the Y67F mutation is absent in the double mutant. The DPR values of the wild-type and the N52V mutant are very similar, but the observed small differences might indicate that the mutation reduces the B1g-type deformation of the heme macrocycle. An NSD analysis of the heme conformation obtained from the 1ns MD simulation of this mutant qualitatively agree with this notion, but predicts also a concomitant decrease of the B2g-type deformation. Indeed, if one performs a simulation which considers the fact that the Q-band resonance energies of N52V are downshifted with respect to the wild-type values, one also obtains a decrease of cAQB2g which reflect B2g-type deformation (data not shown). Substitution of N52 by a bulky residue such as valine is likely to expel all water molecules from the

Deformations of the Heme Group heme cavity, thus inducing a weak net dipole moment. Apparently, this significant modification of the heme cavity has only a limited influence on the ground state of the heme group. We have recently performed a vibronic analysis of the lowtemperature Q- and Qv-absorption bands and found that the investigated mutations increase the electronic B1g perturbation of the heme, which we related to the inhomogeneity of the electric field at the heme.12 Furthermore, we showed that the variations of this perturbation correlate well with changes of redox potential and tried to rationalize this finding in terms of induced changes of the heme iron’s electronic structure. A complete picture of how heme deformations can affect the protein’s functional properties requires a comparison of in-plane and out-of-plane deformations of the respective ferro and ferri states. The B1g and B2g deformations can both increase the splitting between the dπ, dxz, and dyz orbitals. If this splitting is large enough, then one of the dπ orbitals substitutes dxy as the highest occupied orbital, which destabilizes the reduced state. To summarize, this Raman dispersion study investigated mutations of yeast iso-1-cytochrome c that replace polar by nonpolar residues in the heme vicinity. The most pronounced effect was observed for Y67F, which increases the triclinic B2g deformation of the heme group. Crystallographic data for this mutant suggest that this might involve predominantly pyrrole A and its propionate β substituent. The double mutation N52VY67F eliminates this deformation and causes an additional rhombic B1g deformation. These structural changes occur on a scale of 10-2 Å for the total displacement along a given normal coordinate. Acknowledgment. We thank Bruce Stewart (Dalhousie University) for the expert expression and purification of the mutant cytochromes. Financial support was provided from a grant from the National Science Foundation (MCB-0318749) to R.S.S. and from the Natural Sciences and Engineering Research Council of Canada to C.W. References and Notes (1) TenEyck, L. F. In The Porphyrins; Dolphin, D., Ed.; Academic Press: New York, 1979; Vol. III. (2) Schweitzer-Stenner, R. Q. ReV. Biophys. 1989, 22, 381. (3) Jentzen, W.; Ma, J.-G.; Shelnutt, J. A. Biophys. J. 1998, 74, 753. (4) Sutherland, J. C.; Klein, M. P. J. Chem. Phys. 1972, 51, 76. (5) Friedman, J. M.; Rousseau, D. L.; Adar, F. Proc. Natl. Acad. Sci. U.S.A. 1977, 74, 2607.

J. Phys. Chem. B, Vol. 111, No. 23, 2007 6533 (6) Wagner, G. C.; Kassner, R. J. Biochem. Biophys. Res. Commun. 1975, 63, 385. (7) Collins, D. W.; Fitchen, D. B.; Lewis. A. J. Chem. Phys. 1973, 59, 5741. (8) Zgierski, M. Z.; Pawlikowski, M. Chem. Phys. 1982, 65, 335. (9) Bobinger, U.; Schweitzer-Stenner, R.; Dreybrodt, W. J. Raman Spectrosc. 1988, 20, 191. (10) Schweitzer-Stenner, R.; Bobinger, U.; Dreybrodt, W. J. Raman Spectrosc. 1991, 92, 65. (11) Jentzen, W.; Song, X.-Z.; Shelnutt, J. A. J. Phys. Chem. B 1997, 101, 1684. (12) Levantino, M.; Huang, Q.; Cupane, A.; Laberge, M.; Hagarman, A.; Schweitzer-Stenner, R. J. Chem. Phys. 2005, 123, 054508. (13) Manas, E. S.; Vanderkooi, J. M.; Sharp, K. A. J. Phys. Chem. B 1999, 103, 6344. (14) Laberge, M.; Ko¨hler, M.; Vanderkooi, J. M.; Friedrich, J. Biophys. J. 1999, 77, 3293. (15) Berghuis, A. M.; Guillemette, J. G.; McLendon, G.; Sherman, F.; Smith, M.; Brayer, G. D. J. Mol. Biol. 1994, 236, 786. (16) Berghuis, A. M.; Guillemette, J. G.; Smith, M.; Brayer, G. D. J. Mol. Biol. 1994, 235, 1326. (17) Berghuis, A. M.; Brayer, G. D. J. Mol. Biol. 1992, 223, 959. (18) Blouin, C.; Wallace, C. J. A. J. Biol. Chem. 2001, 276, 28814. (19) Schweitzer-Stenner, R.; Levantino, M.; Cupane, A.; Wallace, C.; Laberge, M.; Huang, Q. J. Phys. Chem. B 2006, 110, 12155. (20) Schweitzer-Stenner, R.; Stichternath, A.; Dreybrodt, W.; Jentzen, W.; Song, X.-Z.; Shelnutt, J. A.; Faurskov-Nielsen, O.; Medforth, C. J.; Smith, K. M. J. Chem. Phys. 1997, 107, 1794. (21) Parrish, J. C.; Guillemette, J. G.; Wallace, C. J. A. Biochem. Cell. Biol. 2001, 79, 83. (22) Wallace, C. J. A.; Clark-Lewis, I. Biochem. Cell. Biol. 2000, 78, 79. (23) Huang, Q.; Medforth, C.; Schweitzer-Stenner, R. J. Phys. Chem. A 2005, 109, 10493. (24) Jentzen, W.; Unger, E.; Karvounis, G.; Shelnutt, J. A.; Dreybrodt, W.; Schweitzer-Stenner, R. J. Phys. Chem. 1996, 100, 14184. (25) Hu, S.; Smith, K. S.; Spiro, T. G. J. Am. Chem. Soc. 1996, 118, 12638. (26) Schweitzer-Stenner, R. J. Phys. Chem. 1994, 98, 9374. (27) Lemke, C.; Schweitzer-Stenner, R.; Shelnutt, J. A.; Quirke, J. M.; Dreybrodt, W. J. Phys. Chem. A 2001, 105, 6668. (28) Lemke, C.; Dreybrodt, W.; Shelnutt, J. A.; Quirke, J. M. E.; Schweitzer-Stenner, R. J. Raman Spectrosc. 1998, 29, 945-953. (29) Schweitzer-Stenner, R. J. Porphyrins Phthalocyanines 2001, 5, 198-224. (30) Schweitzer-Stenner, R.; Lemke, C.; Shelnutt, J. A.; Quirke, J. M. E.; Dreybrodt, W. J. Phys. Chem. A 2001, 105, 6680. (31) Shelnutt, J. A., (32) Unger, E.; Bobinger, U.; Dreybrodt, W.; Schweitzer-Stenner, R. J. Phys. Chem. 1993, 97, 9956-9968. (33) Huang, Q.; Szigeti, K.; Fidy, J.; Schweitzer-Stenner, R. J. Phys. Chem B 2003, 107, 2822. (34) Schweitzer-Stenner, R.; Dreybrodt, W.; el Naggar, S. Biophys. Struct. Mech. 1984, 10, 241. (35) Moore, G. R. FEBS Lett. 1983, 161, 171.