15274
J. Phys. Chem. 1996, 100, 15274-15279
Observation of Multiple CN-Isotope-Sensitive Raman Bands for CN- Adducts of Hemoglobin, Myoglobin, and Cytochrome c Oxidase: Evidence for Vibrational Coupling between the Fe-C-N Bending and Porphyrin In-Plane Modes Shun Hirota,†,‡ Takashi Ogura,† Kyoko Shinzawa-Itoh,§ Shinya Yoshikawa,§ and Teizo Kitagawa*,† The Graduate UniVersity for AdVanced Studies and Institute for Molecular Science, Okazaki National Research Institutes, Okazaki 444, Japan, and Department of Life Science, Faculty of Science, Himeji Institute of Technology, 1479-1 Kanaji, Kamigoricho, Akogun, Hyogo 678-12, Japan ReceiVed: October 30, 1995; In Final Form: May 8, 1996X
The CN-isotope-sensitive resonance Raman (RR) bands were investigated for CN- adducts of hemoglobin (Hb), myoglobin (Mb), and cytochrome c oxidase (CcO). All proteins gave multiple CN-isotope-sensitive bands around 450-480 and 340-440 cm-1. The CN--bound resting CcO (CcOrest‚CN) gave intense isotopesensitive bands at 478, 473, 473, and 468 cm-1 for the 12C14N-, 13C14N-, 12C15N-, and 13C15N- adducts, respectively, which were distinctly higher than those for cyanometHb (HbCN) and cyanometMb (MbCN), presumably due to interactions with the CuB ion present at the binuclear site. The monotonous feature of the frequency changes upon the increase of a total mass of CN suggests that these bands arise from the Fe-CN stretching mode (νFe-CN). Besides this main band, several weak CN-isotope-sensitive bands were observed below 440 cm-1 for all three proteins, but the pattern of the isotope-difference spectra was specific to each protein. These low-frequency difference peaks were significantly weaker in intensity for 15N isotopes compared with 13C isotopes in common. The band-fitting calculations indicated that the Raman intensities of several porphyrin vibrations were altered by CN isotopes without changing their frequencies, suggesting that the Fe-C-N bending mode (δFeCN) is present around ∼380 cm-1 and this mode is coupled with more than two porphyrin vibrations which differ among Hb, Mb, and CcO. The C-N stretching (νCN) mode of CN--bound heme proteins was observed in Raman spectra for the first time.
Introduction The FeII hemes of myoglobin (Mb), hemoglobin (Hb), and cytochrome c oxidase (CcO) yield similar O2 adducts with endon geometry, but the physiological reactions of bound O2 are distinct between the former two and the last. This is partly caused by the heme pocket structure characteristic of each protein. For better understanding of the structure-function relationship of proteins, it is highly desirable to elucidate the differences associated with a bound ligand and its environment among these proteins. The ligand-related vibrations are expected to reflect most sensitively the nature of the Fe-ligand bond and the ligand-protein interactions, and resonance Raman (RR) spectroscopy has been acknowledged as the most powerful tool for observing such vibrations.1 Hitherto we have systematically reinvestigated the RR spectra of diatomic-ligand (XY) adducts of heme proteins to reveal all of the isotope-sensitive bands assignable to the Fe-XY stretching (νFe-XY), Fe-X-Y bending (δFeXY), X-Y stretching (νXY), and their combination modes. For XY ) OO, Raman bands associated with δFeOO fundamental and 2νFe-OO were newly observed in addition to the νFe-OO fundamental.2,3 For XY ) CO, a new CO-isotope-sensitive band assignable to the δFeCO fundamental was found around 360 cm-1 for several kinds of heme proteins,4 and furthermore, a few CO-isotope-sensitive bands found above 800 cm-1 were satisfactorily interpreted on the basis of the new assignments,3 although there has been * Author to whom correspondence should be addressed [fax (81) 56455-4639]. † Okazaki National Research Institutes. ‡ Present address: Department of Chemistry, Nagoya University, Chigusa-ku, Nagoya, 464-01 Japan. § Himeji Institute of Technology. X Abstract published in AdVance ACS Abstracts, August 1, 1996.
S0022-3654(95)03190-X CCC: $12.00
argument against it.5 In this study, similar measurements are extended to XY ) CN- bound to metMb, metHb, and oxidized and reduced CcO. A CN- anion has the same number of electrons as CO has, and in the absence of any steric hindrance, it binds to a FeIII heme in an upright geometry. If the Fe-C-N unit were linear and perpendicular to the heme plane, δFeCN fundamental should be Raman inactive, similar to the case of the FeII-CO adduct. Actually, the Fe-C-N unit in MbCN is linear but tilted according to X-ray crystallographic analysis6 and NMR studies.7 Raman bands due to νFe-CN and δFeCN are reported for many heme proteins, and their frequencies are significantly scattered. The first observation by Yu et al.8 for metHb III isolated from Chironomus thummi thummi (Hb CTT III) revealed νFe-CN ) 453 cm-1and δFeCN ) 412 cm-1, and the frequencies for human metHb A and sperm whale metMb were similar to them.9,10 For peroxidases, on the other hand, the νFe-CN band was located at significantly lower frequencies than the δFeCN band except for one conformer of horseradish peroxidase (HRP); myeloperoxidase (MPO), νFe-CN ) 361 cm-1, δFeCN ) 454 cm-1;11a one of the two conformers of HRP (νFe-CN ) 360 cm-1, δFeCN ) 422 cm-1);11b and lactoperoxidase (LPO) (νFe-CN ) 360 cm-1, δFeCN ) 453 cm-1).12 The other conformer of HRP gave νFe-CN ) 453 cm-1 and δFeCN ) 405 cm-1.11b In the case of LPO, another CN-isotope-sensitive band is reported at 311 cm-1 but remains to be interpreted.12 All of these assignments are based on the principles that when the total mass of CN is increased as 12C14N, 13C14N, 12C15N, and 13C15N, the monotonic frequency decrease and zigzag pattern are expected for the νFe-CN and δFeCN modes, respectively. This is correct when the Fe-C-N adopts a nearly linear structure, but not so when it becomes bent. Recently, coexistence of two conformations, linear and bent forms of the Fe-C-N moiety, were reported © 1996 American Chemical Society
CN Adducts of Hb, Mb, and CcO
J. Phys. Chem., Vol. 100, No. 37, 1996 15275
for HRP11b and cytochrome P-450.13 In this study, we observed the νFe-CN Raman bands of oxidized and reduced CcO, and in addition found a number of new CN-isotope-sensitive Raman bands for metMbCN and metHbCN in the frequency region lower than νFe-CN for the first time. Experimental Procedures Raman Measurements. Raman scattering was excited at 441.6 nm by a He/Cd laser (Kinmon Electrics, Model CD1805B) and at 426.0 and 427.0 nm by an Ar+ laser (Spectra Physics, Model 2045)-pumped dye laser (Spectra Physics, Model 375B) with stilbene-420 dye. The latter excitation wavelength was also obtained through the second harmonic generation by a KNbO3 crystal (VIRGO OPTICS, USA) for the output of a Tisapphire laser (Spectra Physics, Model 3900) which was pumped by an Ar+ ion laser (Spectra Physics, Model 2045). Raman scattering was detected with a CCD (Astromed CCD, 3200) or a cooled (-20 °C) intensified photodiode array (Princeton Applied Research, 1421HQ) attached to a single polychromator (Ritsu Oyo Kogaku, DG-1000). This polychromator was equipped with two interchangeable blazed holographic gratings; a 500 nm blazed-1200 grooves/mm grating was used in the first-order geometry and a 900 nm blazed-1200 grooves/mm one was used in the second order. The former (∼1.0 cm-1/ channel) gives better S/N ratio, while the latter gives higher resolution (∼0.4 cm-1/channel). The slit width and slit height were 200 and 10 mm, respectively, for all measurements. The excitation laser beam was focused to about 50 µm, and its power was adjusted to 1-5 mW at the sample. A spinning cell (3000 rpm) was used for measurements in the low-frequency region to avoid photodissociation of CN-. A 0.6 × 0.6 mm flow cell was used for measurements in the high-frequency region, and the flow rate was adjusted to 40 mL/min. All measurements were carried out at room temperature. Raman shifts were calibrated with CCl4 for the low-frequency region and with indene and cyanomethane for the high-frequency region. The accuracy of the absolute frequencies of Raman bands was 1 cm-1, but that of the frequency differences between the CN isotope derivatives obtained by simulations was 0.1 cm-1. Preparation of Samples. Horse myoglobin (Sigma, type M630) was dissolved in a minimum amount of 50 mM TrisHCl buffer, pH 8.0, and was subjected to gel filtration through Sephadex G-25 after complete oxidation of the remaining deoxyMb by potassium ferricyanide. The eluted solution was diluted to 50 mM with 50 mM Tris-HCl buffer, pH 8.0. The CN- adducts were formed by adding 100 mM KCN/0.05 N NaOH solution, adjusted to pH ∼8.0 with HCl solution, to the protein solution so that the final KCN concentration could be ∼1 mM. CN isotope complexes were formed in the same way using K13CN (ICON, 13C 99 atom %), KC15N (ICON, 15N 99 atom %), and K13C15N (ICON, 13C 99 atom % and 15N 99 atom %). Human adult Hb (HbA) was prepared according to the method of Geraci et al.,14 and the CN- complexes were obtained in the same way as that for MbCN. Cytochrome c oxidase (cytochrome c:oxidoreductase, EC 1.9.3.1) was isolated from bovine heart according to Yoshikawa’s method.15 CcO was dissolved in 50 mM sodium phosphate buffer, pH 7.2. The CN-bound resting CcO (CcOrest‚CN) was obtained by adding a small amount of 100 mM KCN/0.05 N NaOH solution, adjuted to pH ∼8.0 with HCl solution, to the CcO solution, in the same way as for Mb and Hb, and leaving it for more than 3 h before measurement. The CN-bound reduced CcO (CcOred‚CN) was obtained by repeating the evacuation of gas inside the cell to ca. 0.1 mmHg followed by incorporation of N2 gas into it to ca. 1 atm by three times.
Figure 1. Raw RR spectra (a) and CN isotope difference spectra (b) in the 550 to 200 cm-1 region for CN- adducts of ferric horse Mb. (a) Raw spectra for 12C14N- (A), 13C14N- (B), 12C15N- (C), and 13C15N(D) adducts; solid lines, observed spectra; broken lines, calculated spectra (broken lines of Figure 5a). The ordinate scales of the observed spectra are normalized by the intensity of porphyrin bands. Experimental conditions: excitation, 426.0 nm, obtained with a dye laser and 5 mW (at the sample); detector, CCD; grating, 900 nm-1200 grooves/mm; sample, 50 mM (heme) in 50 mM Tris-HCl buffer, pH 8.0, with ∼1 mM KCN. (b) CN isotope difference RR spectra of MbCN (A) Mb12C14N - Mb13C14N, (B) Mb12C14N - Mb12C15N, (C) Mb12C14N Mb13C15N, and (D) Mb12C14N - Mb13C15N; solid lines, observed spectra; broken lines, calculated from Figure 5a.
Removal of remaining oxygen was performed by adding N2saturated dithionite solution to the protein solution so that the final concentration of dithionite became 10 mM. Finally, a small amount of ∼1 M KCN/0.05 N NaOH solution, adjusted to pH ∼8.0 with HCl solution, was added to the reduced CcO solution to the final KCN concentration of ∼10 mM. The formation of CN- adducts was confirmed by the disappearance of the iron-histidine stretching band at 214 cm-1.16 Results Figure 1a shows the RR spectra in the 550 to 200 cm-1 region of 12C14N- (A), 13C14N- (B), 12C15N- (C), and 13C15N- (D) adducts of horse metMb excited at 426.0 nm. Solid lines and broken lines denote the observed and calculated (described later) spectra. The band at 452 cm-1 for Mb12C14N (A) is shifted to 449 cm-1 for Mb13C14N (B) , to 448 cm-1 for Mb12C15N (C) , and to 444 cm-1 for Mb13C15N (D), in agreement with the reported results,10 and has been assigned to νFe-CN due to the monotonous character of the frequency shifts upon the increase of the total mass of CN. No other CN-isotope-sensitive band has been noted for RR spectra of MbCN so far, but the presence of such a band could be revealed by difference calculations. Figure 1b displays the difference spectra of the RR spectra shown in Figure 1a. The solid lines and broken lines denote the experimental and calculated (see below) spectra, respectively. Although the difference peaks around 450 cm-1 are most pronounced as expected, there are a number of difference peaks in a frequency range between 300 and 450 cm-1, which have never been noted. Intensities of these difference peaks are stronger in the pair in which masses of the carbon atoms are different (Figure 1b-A,b-C) than in the pair in which masses of the nitrogen atoms are different (Figure 1b-B,b-D). To understand the implications of these difference peaks, polarization measurements were carried out (data not shown). The intensity ratios of bands of perpendicular to parallel components in the raw spectra were all ∼0.25 and, therefore, they are regarded as polarized bands. This is consistent with the results that RR bands of cytochrome c in the low-frequency region are all polarized upon Soret excitation.17
15276 J. Phys. Chem., Vol. 100, No. 37, 1996
Figure 2. RR spectra in the 2200 to 1500 cm-1 region for 12C14N(A) and 12C15N- (B) adducts of ferric horse Mb and their difference spectrum (C). The ordinate scales of spectra are normalized by the intensity of porphyrin bands. Experimental conditions were the same as those for Figure 1a except for using a diode array for a detector, a 500 nm-1200 grooves/mm grating, and a flow cell.
Figure 2 shows the RR spectra of the MbCN in the higher frequency region excited at 427.0 nm. Although IR spectra locates νCN at 2126 cm-1 for MbCN [2122 cm-1 for HbCN, 2131 cm-1 for HRP.CN, and 2152 cm-1 for CN--bound resting (oxidized) CcO (CcOrest.CN)],18 there was no clear CN-isotopesensitive band in the raw spectra of Mb12C14N (A) and Mb12C15N (B). However, in their difference spectrum (C), positive and negative peaks appeared at 2127 and 2097 cm-1, respectively. Sharp difference peaks around 1550-1600 cm-1 are caused by small deviations of the laser beams at the sample point between two separate measurements of isotope species. Such effects are negligible for weak bands at 2049 cm-1 and, therefore, the presence of the difference peaks at 2127/2097 cm-1 is confirmed. These frequencies of the difference peaks are in good agreement with the νCN frequencies of Mb12C14N and Mb12C15N observed with IR spectra.18a This is the first detection of a νCN Raman band for heme proteins and may serve as a measure for the extent of mixing of the CN internal molecular orbital with porphyrin orbitals. Figure 3 shows the results for Hb. Solid lines in Figure 3a denote the RR spectra in the 550 to 200 cm-1 region for 12C14N(A), 13C14N- (B), 12C15N- (C), and 13C15N- (D) adducts of human metHbA excited at 426.0 nm, while the meaning of broken lines will be described later. The band of Hb12C14N (A) at 452 cm-1 is shifted to lower frequencies with the increase of the total mass of CN [448 cm-1 (Hb13C14N, B), 447 cm-1 (Hb12C15N, C), and 443 cm-1 (Hb13C15N, D)] and is assigned to νFe-CN, in agreement with the reported results.9 No other clear CN-isotope-sensitive band was recognized in this figure except for the left shoulder of a strong band at 345 cm-1, which seemed to be definitely influenced by CN isotopes. To reveal the buried CN-isotope-sensitive bands, difference calculations were carried out, and the results are shown in Figure 3b. In addition to the difference peaks around 450 cm-1, which represent the νFe-CN mode, a number of difference peaks appeared when the mass of the carbon atom was changed (Figure 3b-A,b-C), while such peaks were quite weak when the mass of nitrogen was changed (Figure 3b-B,b-D), similar to the case for MbCN (Figure 1b).
Hirota et al.
Figure 3. Raw RR spectra (a) and CN isotope difference spectra (b) in the 550 to 200 cm-1 region for CN- adducts of ferric human Hb. (a) Raw spectra for 12C14N- (A), 13C14N- (B), 12C15N- (C), and 13C15N(D) adducts; solid lines, observed; broken lines, calculated (broken lines of Figure 5b). The ordinate scales of spectra are normalized by the intensity of porphyrin bands. Experimental conditions were the same as those for Figure 1a. (b) CN isotope difference RR spectra of HbCN. Difference calculations were carried out in the same way as in Figure 1b. The difference combinations are specified on the left side; solid lines, observed; broken lines, calculated from Figure 5b.
Figure 4. RR spectra in the 600 to 300 cm-1 region for 12C14N- (A), 13 14 C N (B), 12C15N- (C), and 13C15N- (D) adducts of resting bovine CcO and their difference spectra, CcO‚12C14N - CcO‚13C14N (E), CcO‚12C14N - CcO‚12C15N (F), and CcO‚12C14N - CcO‚13C15N (G). The ordinate scales of spectra A-D are normalized by the intensity of porphyrin bands. Experimental conditions: excitation, 427.0 nm and 1 mW (at the sample); detector, diode array; grating, 900 nm-1200 grooves/mm; sample, 50 mM (heme) in 50 mM sodium phosphate buffer, pH 7.2, with ∼1 mM KCN.
CN--bound cytochrome c oxidase (CcO‚CN) is of great interest since it has the CuB ion at 4.5 Å apart from the Fe ion and the bound CN- might strongly interact with the CuB ion.19 Figure 4 shows the RR spectra in the region from 600 to 300 cm-1 for 12C14N- (A), 13C14N- (B), 12C15N- (C), and 13C15N(D) adducts of resting CcO (CcOrest‚CN) excited at 427.0 nm, and their difference spectra, CcOrest‚12C14N - CcOrest‚13C14N (E), CcOrest‚12C14N - CcOrest‚12C15N (F), and CcOrest‚12C14N - CcOrest‚13C15N (G), in which the ordinate scales of the difference spectra are expanded by a factor of 2. The presence of CN-isotope-sensitive bands is not clear in the raw spectra, but it becomes clear in the difference spectra, where the band at 498 cm-1 was completely canceled in all of the difference
CN Adducts of Hb, Mb, and CcO spectra (E-G). The largest difference intensity and the largest frequency shift are seen in spectrum G, the difference between 12C14N- and 13C15N- adducts. The band at 478 cm-1 for CcOrest‚12C14N (A) is shifted to lower frequencies upon the increase of the total mass of CN; 473 cm-1 (CcOrest‚13C14N, B), 473 cm-1 (CcOrest‚12C15N, C), and 468 cm-1 (CcOrest‚13C15N, estimated from G). Therefore, this band is assigned to νFe-CN. We note that the νFe-CN frequency of CcOrest‚CN at 478 cm-1 is significantly higher than those for MbCN and HbCN (452 cm-1). The band of the 12C14N- adduct at 440 cm-1 is shifted to 426, 440, and 426 cm-1, with the 13C14N-, 12C15N-, and 13C15N- adducts, respectively, and accordingly is assigned to δFeCN. In the corresponding RR spectra for reduced CcO (CcOred‚CN) excited at 441.6 nm, which is close to the absorption maximum of CcOred‚CN (data not shown), there was a CN-isotope-sensitive band at 475 cm-1, which was shifted to lower frequencies upon the increase of the total mass of CN: 472 cm-1 (CcOred‚13C14N, B), 472 cm-1 (CcOred‚12C15N, C), and 465 cm-1 (CcOred‚13C15N, D). Although those bands were overlapped with porphyrin modes, the difference spectrum between the 12C14N- and 13C15N- adducts indicated the presence of difference peaks around 485-456 cm-1 under the complete cancellation of a band at 583 cm-1. However, CcOred‚CN is extremely liable to be oxidized by a trace amount of oxygen, yielding a mixed valence state; the similarities of these frequencies to those in Figure 4G might be caused by the mixing of partially oxidized species. Accordingly, we refrain from detailed discussion on this species at the present stage. Discussion νFe-CN Mode of MbCN and HbCN. The νFe-CN RR bands of MbCN and HbCN were observed around 452 cm-1 (Figures 1a and 3a) which were lower than the νFe-CO frequencies of MbCO and HbCO. For “strapped hemes”, in which the ligand binding is sterically hindered by a (CH2)n-chain strap over the iron at the CN- binding site, the νFe-CN bands are observed at 447 (n ) 15), 447 (n ) 14), and 445 cm-1 (n ) 13), while an unhindered one gives it at 451 cm-1.20 Thus, the steric hindrance slihgtly lowers the νFe-CN frequency. This frequency is also slightly altered by the trans ligand, since the νFe-CN RR band of Fe(OEP)(pyridine)CN (OEP, octaethylporphyrin) is observed at 449 cm-1, but its frequency varies in a range of 442-452 cm-1 depending on substituents of the pyridine ligand at the trans position.21 The νFe-CN frequencies of MbCN and HbCN are comparable to those of model compounds,22 but so far δFeCN RR bands have never been detected for the model compounds. The νFe-CN frequencies of MbCN and HbCN are very close to that of cyanomet Hb CTT III (453 cm-1), while their νFe-CO frequencies are appreciably different: 512 cm-1 for MbCO, 507 cm-1 for HbCO, and 500 cm-1 for CO-bound CTT III.3,8,23 Although both Fe-C-O and Fe-C-N have been thought to adopt a linear upright structure to a heme, these observations suggest that the Fe-C-N geometry is much more fixed than the Fe-C-O geometry in the proteins. In fact, it is reported from X-ray24 and time-resolved IR studies25,26 that CO is more bent in MbCO than in HbCO. Multiple CN-Isotope-Sensitive Bands. To understand the implications of the appearance of many CN-isotope-difference peaks in Figure 1b, we carried out simulation calculations. Individual bands were assumed to be Gaussian, and their positions, peak intensities, and widths were adjusted so as to reproduce the observed raw spectra and difference spectra. So far as possible, an identical bandwidth was used for corresponding bands of isotope species. The spectra finally assumed for
J. Phys. Chem., Vol. 100, No. 37, 1996 15277
Figure 5. Simulation of RR spectra of MbCN (a) and HbCN (b): solid lines, assumed bands with Gaussian band shapes; broken lines, envelopes for the sum of assumed bands.
the 12C14N- (A), 13C14N- (B), 12C15N- (C), and 13C15N- (D) adducts of Mb are depicted in Figure 5a, where expected band envelopes, that is, the sum of assumed bands, are designated by broken lines. These envelopes are also drawn in Figure 1a as broken lines. Since the base lines of the calculated spectra are not fitted to the observed ones, there are some differences between the observed and calculated spectra, but such differences do not affect the peak positions. The results from the difference calculations between the assumed spectra are depicted by broken lines in Figure 1b. Apart from the deviation of zero lines, the peak positions and patterns of the experimental curves are approximately reproduced by the calculated curves. Therefore, the appearance of a number of CN-isotope-difference peaks in Figure 1b is interpreted on the basis of the spectra shown in Figure 5a. This procedure enabled accurate estimation of many isotope-sensitive bands overlapped with each other. The peak frequencies of the bands that exhibit appreciable changes with CN isotopes in Figure 5a are listed in Table 1, in which all of the CN-isotope-sensitive RR bands of heme proteins reported hitherto are also contained. Table 1 indicates that the 12C14N- and 12C15N- adducts of Mb give rise to weak side bands at 425, 404, 385, 302, and 257 cm-1, but the intensities of the corresponding bands of 13C14N- and 13C15N- adducts are too weak to identify. The 374 cm-1 band exhibited a weak zigzag pattern in frequencies. In addition, the relative intensity of bands at 349 cm-1/342 cm-1 is monotonously increasing from 1/3 for Mb 12C14N to 3/7 to Mb13C15N. It was quite unexpected that the CN-isotopedifference peaks in the 400-250 cm-1 region in Figure 1b arise from simple intensity changes with little frequency shifts. If one of these bands arose from δFeCN, it should exhibit an appreciable frequency shift upon isotope substitution. Therefore, we are forced to assign them to porphyrin vibrations. This is distinct from the case of cytochrome P-450,13 for which four bands in the 380-430 cm-1 region exhibited frequency shifts with isotopes of CN-. The pattern of isotopic frequency shifts of the 452 cm-1 band of Mb12C14N suggests that it binds to the heme in the linear geometry. Then δFeCN, whose frequency would be more sensitive to the mass of carbon than to the mass of nitrogen, should be Raman inactive but would interact with in-plane Eu (for D4h but E for C4V) vibrations of porphyrin, since δFeCN is accompanied by in-plane displacements of the Fe ion. Several Eu skeletal modes are expected below 400 cm-1 (ν50-ν53),27,28 and some of them have been observed in the Raman spectrum of reduced cytochrome c.17 Among them, the Fe-N stretching mode (ν50), which has been observed at 360 cm-1 for reduced
15278 J. Phys. Chem., Vol. 100, No. 37, 1996
Hirota et al.
TABLE 1: νFe-CN, δFeCN, and νCN Frequencies for Cyanoferric Heme Proteinsa 12C14N 13C14N 12C15N 13C15N
Mbb
452
449
448
444
Hbb
452
448
447
443
CcO(aa3)b
478
473
473
468
Hb CCTIIIc HRPd (pH 5.5) HRPd (pH 11.6) MPOd LPOe (pH 7.0) LPOe (pH 10.5) SiRHPf
452 453 422 444 420 453 453
449 416 439 413 448 448
449 422 440 419 453 452
445 416 435 413 448 445
452
446
451
444
P-450g (substrate free)
451 451 413 434
445 434 409 426
449 449 409 434
443 434 405 426
P-450g (+camphor) P-450g (+adamantanone)
416 424 423 437
411 416 417 429
412 424 419 437
406 416 413 429
modeh
12C14N
νFe-CN 441 425 404 385 (s)j 374.5 349 [25%] 342.1 [75%] 302 257 νFe-CN 434 (w)j 429 (w)j 392 (s)j 379 [55%] 373 [45%] 351 (9.4)k 313 νFe-CN 440 400.4 369 (vs)j νFe-CN 412 νFe-CN 405 δFeCN 360 νFe-CN 405 δFeCN 355 δFeCN 361 δFeCN 360 (311) δFeCN 355 (311) νFe-CN 390 δFeCN 352 νFe-CN 387 328 343 350 νFe-CN 392 359 νFe-CN 387 357
13C14N
439
12C15N
13C15N
δFeCN
2126
434 (s)j 429 (s)j 392 (w)j 379 [30%] 374 [70%] 351 (8.4)k
δFeCN
2122
426 400.0 369 (s)j
δFeCN
2150
426 400.0 369 (s)j
440 425 404 385 (s)j 374.4 349 [27%] 341.9 [73%] 302 257 434 (w)j 429 (w)j 392 (s)j 379 [55%] 373 [45%] 351 (9.0)k 313 440 400.4 369 (vs)j
393
405
393
393
405
393
359 358 (310) 353 (310) 376 350 371 327 340 349 381 354 377 350
359 357 (310) 353 (310) 388 348 387 327 340 349 391 356 384 353
356 356 (309) 352 (309) 374 346 371 325 335 348 377 350 372 348
385 (w)j 374.0 349 [27%] 341.9 [73%] 434 (s)j 429 (s)j 392 (w)j 379 [30%] 374 [70%] 351 (8.5)k
modeh νCNi
438 385 (w)j 373.9 349 [30%] 341.7 [70%]
δFeCN δFeCN 2131 νFe-CN δFeCN 2131 νFe-CN νFe-CN νFe-CN νFe-CN δFeCN νFe-CN δFeCN
δFeCN δFeCN
a In cm-1 unit. b This study. c Reference 8. d Reference 11b. e Reference 12. f Siroheme protein, ref 33. g Reference 13. h Assignment reported. Reference 18. j vs, s, and w represent the intensities of the bands: very strong, strong, and weak, respectively. k Bandwidth of the simulated Gaussian bands. i
cytochrome c,17 should involve in-plane displacements of the Fe ion and, accordingly, is most likely to couple with δFeCN. Since such vibrational coupling is very sensitive to intrinsic frequencies, it is understandable that when the intrinsic frequency of δFeCN is shifted from the intrinsic frequency of ν50 by the isotope substitution, the vibrational coupling, and thus the appearance of CN-isotope-sensitive RR bands around 385 cm-1, would be varied. It is noticed in Table 1 that there is another CN-isotopesensitive band: 441 cm-1 for Mb12C14N, 439 cm-1 for Mb13C14N, 440 cm-1 for Mb12C15N, and 438 cm-1 for Mb13C15N. The shift pattern is somewhat zigzag, but the size of frequency shift is noticeably small. This might be a porphyrin band which couples with νFe-CN. However, it is also likely that the FeCN group with a bent structure is coexistent as a minor population and its νFe-CN gives rise to this band. When the FeCN adopts a bent structure, the νFe-CN and δFeCN are significantly mixed with each other and, accordingly, both exhibit partially zigzag pattern depending on the extent of vibrational mixing. The δFeCN mode of the bent form would be located at a higher frequency than the δFeCN of the linear form. If the intrinsic frequency of δFeCN of the bent form is located near 400 cm-1, the idea of vibrational coupling mentioned above could be applied to the appearance of the 404 cm-1 band. In this way, the present observations demonstrated that the hidden mode of the FeCN group appreciably couples with the porphyrin skeletal modes and perturbs Raman intensity of forbidden bands.
Similar analysis was carried out for HbCN. Assumed spectra are depicted in Figure 5b and the envelope of the sum of assumed bands is drawn by broken lines in Figure 3a, where base lines are not fitted to the observed spectra. The isotopedifference spectra calculated from Figure 5b are drawn by broken lines in Figure 3b. Frequencies of assumed bands, which may bring about some differences among CN isotopes, are summarized in Table 1. Similar to MbCN, the frequency shifts are not seen with CN isotopes except for the νFe-CN band at 452 cm-1, and only intensity changes seem to cause the observed isotopic difference spectra. Accordingly, interpretation of Figure 3b is similar to that of MbCN. The appearance of isotopedifference peaks due to coupling with Raman inactive porphyrin vibrations was recently reported for O2-bound heme oxygenase, for which multiple O2-sensitive RR bands are observed around 340-420 cm-1.29 Similar phenomena are noted for CO-bound CcO.4 νFe-CN Mode of CN--Bound Resting CcO. The νFe-CN band for CcOrest‚CN was observed at 478 cm-1 in Figure 4, while νFe-CN frequencies of other heme proteins are below 450 cm-1. The νCN frequency observed by IR spectroscopy is also higher than those of other cyanoferric heme proteins (see Table 1).18 This was unexpected since usually νFe-XY becomes lower as νXY becomes higher. A similar trend was noted for the CObound CcO, whose νFe-CO frequency was higher than that expected from the linear correlation between νFe-CO and νCO of histidine-coordinated heme proteins.30 This might be caused by the CuB ion located 4.5 Å from the heme iron (Figure 6).19 When CuB is removed from cytochrome bo, a heme-copper
CN Adducts of Hb, Mb, and CcO
J. Phys. Chem., Vol. 100, No. 37, 1996 15279 References and Notes
Figure 6. Schematic illustration of the binding site structures and the νFe-CN and νCN frequencies of CN--bound forms.
oxidase from Escherichia coli, by molecular engineering, the νFe-CO band shifts to a lower frequency31 with a slight highfrequency shift of the iron-histidine stretching mode.32 Therefore, CuB is more likely to cause an upshift in the Fe-XY stretching frequencies for XY-bound forms. The difference spectra of CcOrest‚CN shown in Figure 4E-G can be interpreted by similar simulation calculations mentioned above. The difference peaks at 405-395 cm-1 arise from small frequency shifts (0.4 cm-1) of the 400 cm-1 band, while the strong difference peak around 369 cm-1 is ascribed to a simple intensity change of the 369 cm-1 band; the 13C14N- and 13C15Nadducts have weaker intensity by 20% than the 12C14N- and 12C15N- adducts. The difference peaks at 440/426 cm-1 mean lower frequencies for the 13C14N- and 13C15N- adducts than for the 12C14N- and 12C15N- adducts, and therefore this band is assignable to δFeCN. Structures of the CN-binding sites of MbCN, HbCN, and CcOrest‚CN are illustrated in Figure 6 together with their νFe-CN and νCN frequencies. CN-Isotope-Sensitive Bands of Other Heme Proteins. The νFe-CN, δFeCN, and νCN frequencies reported so far are summarized in Table 1. Most of them give one CN-isotope-sensitive band around 450 cm-1, but the shift patterns upon the increase of the total mass of CN (in the order 12C14N, 13C14N, 12C15N, and 13C15N) are dissimilar. Mb, Hb, CcO, and one conformer of HRP and cytochrome P-450 yield monotonous shifts, and the sizes of shifts are as large as 8-10 cm-1 between 12C14N and 13C15N. The heme pockets of these proteins probably allow the linear FeCN structure and, accordingly, δFeCN is theoretically Raman inactive. On the other hand, MPO, LPO, SiRHP, and the other conformer of HRP and cytochrome P-450 exhibit zigzag patterns. In these proteins the heme pocket provides steric hindrance to the bound CN- and, accordingly, the FeCN adopts a bent structure, for which the νFe-CN and δFeCN are mixed significantly. Simple normal-mode calculations of a triatomic molecule suggest that when the Fe-C-N bond angle becomes