A Nuclear Overhauser Effect Study of the Active Site of Myeloperoxidase STRUCTURAL SIMILARITY OF THE PROSTHETIC GROUP TO THAT ON LACTOPEROXIDASE*

The low-spin, cyanide-ligated ferric complex of the intact bovine granulocyte myeloperoxidase (MPO-CN) has been studied by proton nuclear magnetic resonance utilizing the nuclear Overhauser effect (NOE). This is the largest globular protein (-1.5 X lo6 for the intact (Ye& tetrameric species) for which successful NOES have been observed without serious interference of spin diffusion, and demonstrably confirms the utility of such studies on large paramagnetic as compared to diamagnetic proteins. The lH NMR spectrum of MPO- CN is found to have a remarkable similarity in the number, resonance pattern, and metal ion-induced re- laxation properties of the resolved, hyperfine-shifted resonances to those reported earlier for the analogous complex of bovine lactoperoxidase (LPO-CN); more-over,

The low-spin, cyanide-ligated ferric complex of the intact bovine granulocyte myeloperoxidase (MPO-CN) has been studied by proton nuclear magnetic resonance utilizing the nuclear Overhauser effect (NOE). This is the largest globular protein (-1.5 X lo6 for the intact (Ye& tetrameric species) for which successful NOES have been observed without serious interference of spin diffusion, and demonstrably confirms the utility of such studies on large paramagnetic as compared to diamagnetic proteins.
The lH NMR spectrum of MPO-CN is found to have a remarkable similarity in the number, resonance pattern, and metal ion-induced relaxation properties of the resolved, hyperfine-shifted resonances to those reported earlier for the analogous complex of bovine lactoperoxidase (LPO-CN); moreover, the interproton connectivities between pairs of hyperfine-shifted proton sets, as reflected by the NOES, are also essentially the same (Thanabal, V., and La Mar, G. N. (1989) Biochemistry 28, 7038-7044). Since the extracted prosthetic group of lactoperoxidase is a porphyrin with proposed functionalization of the 8-methylene group (Nichol, A. W., Angel, L. A., Moon, T., andclezy, P. S. (1987)  Myeloperoxidase, MPO,' is a hemoprotein found in neutrophils which, upon reacting with hydrogen peroxide to yield compound I two oxidizing equivalents above the resting state, is capable of oxidizing halogen ions to hypohalous acid (Harrison and Schultz, 1976), which is thought to act as the effective in uivo bactericide (Klebanoff and Clark, 1978). The enzyme is a a& tetramer with two -60,000 and two 15,000 subunits, with each of the larger subunits containing one prosthetic group (Andrews and Krinsky, 1981;Ikeda-Saito et al., 1989). The intact enzyme also contains two calcium ions (Booth et al., 1989). While this enzyme retains most of the characteristic properties of heme peroxidases, it possesses an unusual optical spectrum that is red shifted as compared to conventional hemoproteins in the same oxidation/spin state, and exhibits a low energy band at -680 nm that is responsible for its green color (Newton et al., 1965a(Newton et al., , 1965bWever and Plat, 1981). This latter observation has led to the proposal that the chromophore is an iron chlorin, 1 in Fig. 1, rather than the ubiquitous iron porphyrin, 2 in Fig. 1 with X = H, and resonance Raman and MCD properties have been cited to provide at least consistency with, if not proof for, this proposal (Sibbett and Hurst, 1984;Babcock et al., 1985;Sono et al., 1986). Others have suggested that the chromophore is a porphyrin with electron-withdrawing substituents based on the strong similarity of the spectrum of the pyridine hemochromogen with that of heme a (Schultz and Schmukler, 1964;Newton et al., 1965b). The lack of definitive characterization of the prosthetic group is that, like in the case of mammalian peroxidases, and unlike in the case of the known plant or bacterial peroxidases, it is not readily extractable (Carlstrom, 1969;Sievers, 1981). Recent work on lactoperoxidase, LPO, has yielded a free prosthetic group proposed to possess a 8thiol functionalization of the protoheme skeleton, 2 in Fig. 1 with X = SH (Nichol et al., 1987), which was suggested to form a disulfide link to the protein (Nichol et al., 1987).
'H NMR can provide a variety of useful electronic and molecular structural information on paramagnetic enzymes (La Mar, 1979;Satterlee, 1985;Bertini and Luchinat, 1986), and has been successfully utilized to provide detailed information on the component amino acids and their geometrical disposition in the heme pocket of horseradish peroxidase (Thanabal et al., 1987(Thanabal et al., ,1988, for which there exists no crystal structure. Thus it was possible to unequivocally identify a distal His and Arg in positions similar to those found in the x-ray structure for cytochrome c peroxidase (Poulos et al., 1980), and to demonstrate that the protonated distal His side chain is involved in a strong hydrogen bond to the bound ligand in the low spin cyanide complex (Yoshikawa et al., 1985;Thanabal et al., 1988). The key NMR spectroscopic methodologies for these studies are the nuclear Overhauser effect (NOE) which probes spatial proximity and identifies interacting spin systems (Noggle and Shirmer, 1971), in conjunction with isotope labeling of the removable heme group (La Mar, 1979).
In an enzyme containing a nonremovabale prosthetic group of unknown structure, a NMR study may be less definitive, but can be pursued on several levels, by either uniquely identifying the various functional groups and determining their spatial disposition (the ultimate solution structure now possible for small proteins), or by comparing the hyperfine shift pattern and spatial dispositions of the protons yielding resolved signals with those of a better understood and related reference protein. A strong similarity in the pattern of the hyperfine shifts to a structurally characterized protein would support a very similar electronic and molecular structure if the geometric relationship among the proton sets giving rise to the hyperfine shifted peaks can be demonstrated to be the same. In such a comparison, it would be desirable, but not an absolute necessity, to have unambiguously identified the hyperfine shifted signals in the reference protein, as long as the molecular structure of the entity giving rise to the hyperfine shifts is known.
We have recently succeeded in obtaining high quality and informative NOE spectra of the cyanide complex of lactoperoxidase (LPO-CN) (Thanabal and La Mar, 1989), a -78,000 molecular weight single chain glycoprotein which has been proposed to possess the unusually functionalized heme, 2 in Fig. 1 with X = SH (Nichol et al., 1987). While unambiguous assignments were not possible, the established dipolar connectivities were not only shown to be consistent with the structure of 2 with an 8-methylene rather than methyl group, but two sets of non-heme resonances were found to have similar hyperfine shifts and dipolar connectivities as previously identified for the distal His and Arg in both cytochrome c peroxidase in single crystal (Poulos and Kraut, 1980) and horseradish peroxidase in solution (Thanabal et al., 1987). These two distal residues have been proposed as key components of the catalytic mechanism of heme peroxidases (Poulos and Kraut, 1980;Finzel et al., 1984). Apparent sequence homology has already suggested (Kimura and Ikeda-Saito, 1988) that MPO possess the His-X-Y-Arg sequence analogous with the distal fragment in cytochrome c peroxidase (Poulos et al., 1980). Some similarities in the 'H NMR spectra of the nitrite complex of the spleen green hemeprotein (Ikeda-Saito and Inubushi, 1987) to the LPO-CN trace has already been noted, and it has since been shown that this green hemeprotein is identical to MPO (Ikeda-Saito et al., 1989). NOE studies of MPO-CN present an intriguing challenge from the point of NMR strategy, since this is by far the largest protein (-150 x 103) to be subjected to such NMR studies. An analogous diamagnetic system would be many factors too large so as to preclude obtaining any useful data in an NOE experiment due to the detrimental effects of spin diffusion (Kalk and Berendsen, 1976;Sykes et al., 1978). However, we have shown that, not only does paramagnetism undermine spin diffusion and allow useful NOE studies for much larger paramagnetic than diamagnetic systems, but the magnitude, and hence detectability, of NOES actually improve with increasing molecular weight (Thanabal et al., 1987;Thanabal and La Mar, 1989;Dugad et al., 1990). Hence large size may be an advantage rather than a disadvantage for detecting NOES in a paramagnetic enzyme. We present herein a 'H NMR relaxation and NOE study that demonstrates a remarkably high degree of similarity of both the electronic and molecular structures of the heme cavities of MPO-CN and LPO-CN that argues that they possess a similarly functionalized heme prosthetic group, 2 (with X + H), and moreover, that MPO also possesses distal Arg and His, the 2 residues proposed critical for efficient activation of peroxides (Poulos and Kraut, 1980;Finzel et al., 1984).

EXPERIMENTAL PROCEDURES
MPO was isolated from bovine granulocytes as described earlier (Ikeda-Saito et al., 1989). NMR samples (0.5 ml) of -2 mM MPO concentration in *Hz0 and 90% H,O, 10% 'H20) were prepared; each containing 0.1 M phosphate buffer and 10 mM KCN. The reported pH valuesare met& readings not corrected for isotope effect. The 'H NMR and ontical snectra gave no evidence of degradation of MPO-CN upon completion of the described NMR experiments.
'H NMR experiments were carried out on Nicolet NT-360 and NM-500 (360 and 500 MHz, respectively) spectrometers. The spectra for NOE were recorded in double precision using 16,384 data points over 100 kHz band width. The signal to noise was improved by exponential apodization of each free induction decay which introduced 50.Hz line broadening. The chemical shifts are referenced to residual water signal which was calibrated with 2,2-dimethyl-Z-sila-pentane&sulfonate (DSS). The reported chemical shifts are in parts per million with downfield shifts taken as positive. The spectra in 'H,>O solution were collected with Redfield selective excitation (Redfieid et al., 1975) by placing the carrier frequency near 21 ppm (Fig.  3C). The nonselective spin lattice relaxation times, T1, were measured using the water-eliminated Fourier transform pulse sequence (Gupta, 1976) to overcome a severe dynamic range problem from the intense diamagnetic envelope. The NOE difference traces were obtained by subtracting spectra with off resonance irradiation from spectra with on resonance irradiation. The on and off resonance NOE spectra were collected in an interleaved fashion, each with 3 to 4 X X04 transients.
Both fast recycle time and NOE experiments were carried out using    (Gupta, 1976). To achieve rapid repetition rates, the spectra were collected by varying the data size from 8,192 to 512 points. The spectra were zero filled to 16,384 data points to have appropriate digitization. Primary NOES were differentiated from off resonance effects by their independence of the decoupler power or degree of saturation. The off resonance saturation is approximately inverse quadratically dependent on decoupler power (Thanabal et al., 1987(Thanabal et al., , 1988Thanabal and La Mar, 1989). Thus, to unambiguously establish the presence of a NOE, carefully chosen control experiments were performed as discussed in detail previously (Thanabal and La Mar, 1989). For example, to show the NOE connectivity between peaks x and y, the experiments at two different decoupler power levels were carried out with two off-sets placed symmetric to both peaks (Fig. 5). It is noted that significant off resonance irradiation of the diamagnetic envelope is observed (Fig. 5). The large NOES (50-80s) observed between a pair of single proton peaks most likely reflect a pair of geminal methylene protons. 0 -20 40 PPII

RESULTS
The low spin cyanide complex of ferric myeloperoxidase exhibits a 'H NMR spectrum very similar to that of the previously reported nitrite complex (Ikeda-Saito and Inubushi, 1987). Moreover, the spectrum displays a remarkable similarity with that of LPO-CN (see Fig. 2A of Thanabal and La Mar, 1989). The 'H NMR traces under a range of pulse repetition rates provide qualitative information on the relative relaxation times and allow improved detectability of broad, fast relaxing resonances. At slow repetition rates where no hyperfine shifted signals are saturated ( Fig. 2A), we resolve a set of one-proton intensity peaks u-c, y; additional one proton peaks appear as partially resolved shoulders to the threeproton peak f (peak d), and on the high field side of the diamagnetic envelope (peak x). At faster repetition rates, shoulder d saturates first, followed by peaks x, y, and b, with the concomitant appearance of broad resonances e and z ( Fig.  2B). At more rapid pulsing rates, peak e and z display unit intensity although slightly different line widths (Fig. 2C). The T,s resulting from a nonselective inversion recovery experiment are listed in Table I, along with the chemical shifts for all resonances at 20 "C. The number of lines, their relative intensities, chemical shifts, and T, values of MPO-CN (Fig.  3B) bear a striking resemblance to those previously reported (Thanabal and La Mar, 1989) for LPO-CN (Fig. 3A) which are reproduced in Table I is that the larger line widths and more intense diamagnetic envelope preclude resolution of several peaks near the envelope for the latter protein (i.e. those labeled g, h, and w in LPO-CN) (Thanabal and La Mar, (1989). Upon dissolving MPO-CN in 'H20, two new resonances down field of 12 ppm are observed which must arise from exchangeable protons, a broad (-500 Hz) peak cz* at 26.6 and a narrow (-250 Hz) peak b* at 16.0 ppm ( Fig. 3B and C), which can be compared to the similarly shifted labile protons of LPO-CN in 'Hz0 whose trace is reproduced in Fig. 3A Thanabal and La Mar, (1989); the labile proton peaks are marked as u* and b*. B, 360 MHz 'H NMR spectrum of -2 mM MPO-CN in 90% HZO, 10% *Hz0 at 20 "C, pH 7.1. One of the labile proton peaks (marked as b*) is clearly seen. C, Redfield detected 500 MHz NMR spectral region of MPO-CN in 90% H20, 10% 'Hz0 at 20 'C, pH 6.6; the labile proton peak a* at 26.6 ppm is clearly seen. Note the similarity in chemical shift positions of exchangeable protons with those in A. D *------T---cc l"'l'"l"'l"'l"'l"'l"', FIG. 4. A, the 360 MHz 'H NMR spectrum of -2 mM MPO-CN in 'Hz0 at 20 "C, pH 7.2. The peaks are labeled as described in the legend to Fig. 2 (A). Traces B-D are the NOE difference spectra generated by subtracting a spectrum with the decoupler off resonance from a similar spectrum of the same sample in which the desired resonance was saturated. In each trace a vertical arrow indicates the saturated peak. B, irradiate peak a; note (--80%) NOE observed to peak c and smaller NOE to peak j at 8.8 ppm. C, irradiate peak c; note reciprocal NOE to peak a as well as NOE to peak j. D, saturate peak f; note NOES observed to four peaks U, u, W, and w' in 0.0 to -4.0 ppm region. In each trace A indicates the position of off resonance decoupler frequency. Note the identical NOE pattern for LPO-CN in Fig. 3 of Thanabal and La Mar (1989). field et al., 1975) because of severe dynamic range problems with the solvent resonance. The comparison of the traces in Fig. 3 (A and B) (Thanabal and La Mar, 1989); reference to the appropriate comparison trace for LPO-CN is given in each case. Saturation of peak a yields a very large (--80%) NOE to peak c, and a smaller (--20%) NOE to a single proton peak j incompletely resolved on the low field side of the diamagnetic envelope, as shown in Fig. 4B (compare to Fig. 3B in Thanabal and La Mar, 1989). Peak j is judged to have single proton peak intensity based on the line width in the difference trace and the total intensity of the incompletely resolved envelope at that chemical shift value at slow repetition rates. The results of the saturation of peak c is shown in Fig. 4C, which exhibits the reciprocal NOE to peak a as well as an NOE to j, just as found earlier for LPO-CN (see Fig. 3C in Thanabal and La Mar, 1989). Irradiation of the apparent methyl peak, f, as shown in Fig. 40, failed to yield any NOES to resolved resonance, but gave small NOES to multiple peaks just upfield of DSS (as found in LPO-CN; see Fig. 30 in Thanabal and La Mar, 1989).
Irradiation of peak b (Fig. 5B) yields a large NOE to a peak i at 12.2 ppm (--50% based on it being a single proton), just as found in LPO-CN (see Fig. 4B of Thanabal and La Mar, 1989). Peak d could not be resolved from methyl peak f at any temperature.
Saturation of peak d (not shown) failed to yield any NOE that could not be completely accounted for by the partial saturation of peak f. Hence we conclude peak d does not yield detectable NOES. The similarly labeled peak in LPO-CN also failed to exhibit detectable NOES (see Fig. 4C of Thanabal and La Mar, 1989). The two upfield single proton signals, x and y, exhibit --50% reciprocal NOES between them, as shown in Fig. 5, C and D, again as previously reported for LPO-CN (see Fig. 4, D and E, of Thanabal and La Mar, 1989). However, because of the larger intensity of the diamagnetic envelope in MPO-CN, off resonance effects precluded detection of any additional NOES in the diamagnetic envelope similar to those found in LPO-CN (Thanabal and La Mar, 1989).
The exchangeable proton peak a* was too broad to saturate effectively without causing unacceptable off resonance excitation of the intense diamagnetic envelope which obscured any potential NOES in the region 10-O ppm. In LPO-CN, it was possible to detect a very small NOE from a* to a relatively FIG. 5. A, the 360 MHz 'H NMR reference spectrum of -2 mM MPO-CN in 'HZ0 at 20 "C, pH 7.2. The NOE difference spectra in B-D show irradiated peaks with vertical arrows and position of the decoupler reference frequency by A. 3, saturate peak b; note --25% NOE to peak i at 12.2 ppm. C, irradiate peak y and observe NOE to peak n. D, irradiate peak X; note reciprocal NOE to peak y. The reference trace for D was obtained by placing decoupler off-set symmetric to peak y so as to cancel off resonance saturation effects. Instrumental artifacts are labeled by asterisk (*). The C and D traces were obtained at two decoupler power levels with two controls placed symmetric to each peak with the same magnitude NOE (trace at only one power level is shown). Note the striking similarity of traces B, C, and D of this Figure with traces B, D, and E, respectively, of Fig. 4 of Thanabal and La Mar (1989).
narrow single proton peak near 10 ppm (Thanabal and La Mar, 1989), identified as the distal His C,H by analogy to cyanide-ligated horseradish peroxidase (Thanabal et al., 1988). Thus we find that MPO-CN and LPO-CN not only exhibit the same number of peaks with comparable shifts and relative relaxation properties (Table I), but that their dipolar connectivities, and hence their relative spatial dispositions, are essentially identical. This indicates that the similarly labeled resonances in LPO-CN and MPO-CN arise from the same functional group with essentially the same relative orientations. While the present NOE data do not shed additional light on the assignment of resonances than already available from LPO-CN (Thanabal and La Mar, 1989) or provide information on the chemical nature of X in 2 of Fig. 1, it is clear that the identically labeled peaks must have the same assignments in the two proteins. Hence LPO may serve as a useful model protein for spectroscopic identification designed to develop in detail the structural features of the prosthetic group and the active site in MPO.

NOE strategies-The
present NOE studies on MPO-CN, molecular weight -1.5 x 105, demonstrate convincingly the utility of carrying out such experiments with large paramagnetic proteins; spin diffusion near the active site, which would obliterate useful NOE information in a similar diamagnetic protein, is essentially negligible, and the steady-state NOES are somewhat larger than for LPO-CN.
The larger NOES of MPO-CN relative to LPO-CN are in accord with expectations for paramagnetic proteins, as witnessed by the observation of larger NOES for a given protein upon increasing solvent viscosity (Dugad et al., 1990). Hence it can be expected that 'H NOE studies will contribute significantly towards elucidating the active site stereochemistry of particularly large paramagnetic metalloenzymes. Proximal Histidine Bonding-That MPO, like LPO, or the better characterized horseradish peroxidase and cytochrome 0 -20 -Yo PPtl c peroxidase, possesses an axial His has already been supported by a variety of spectroscopic means (Bolsher and Wever, 1984;Babcock et al., 1985;Ikeda-Saito and Inubushi, 1987). The resolution of the two broad and fast relaxing peaks e and .z with essentially the same shifts as similar peaks in both LPO-CN and HRP-CN and directly assignable to the coordinated His ring CaH and C,H in the latter protein (Thanabal et al., 1987), not only confirms the presence of a proximal histidine, but argues that the side chain ring, as in other cyanide-ligated heme peroxidases, is deprotonated in MPO-CN. Such imidazolate character of the axial His is likely to stabilize the highly oxidized intermediates for all heme peroxidases (Mincey and Traylor, 1979;Teraoka and Kitagawa, 1981;Desbois et al., 1984). The i5N chemical shift for MPO-C?"N has been reported as 508 ppm from free cyanide or 390 ppm from nitrate ion (Morishima et al., 1988). This shift is similar to that of LPO-CN (Behere et al., 1985), and probably reflects both the trans imidazolate character of the proximal His, as well as likely hydrogen bonding to the bound cyanide by the distal His as characterized in detail for cyanideligated horseradish peroxidase (Thanabal et al., 1987(Thanabal et al., , 1988. Distal Cavity Residues-It has been argued that a distal His and Arg are critical constituents of a heme cavity activated by peroxides (Schonbaum et al., 1979;Poulos and Kraut, 1980;Finzel et al., 1984), and such residues are found in cytochrome c peroxidase and horseradish peroxidase with remarkably similar stereochemistry relative to the heme (Thanabal et al., 1987(Thanabal et al., , 1988. The upfield peaks, n and y (as well as dipolar coupled peaks in the envelope), in LPO-CN have shifts similar to those of the distal Arg in cyanide-ligated horseradish peroxidase, and have been argued to have a similar origin in LPO-CN (Thanabal and La Mar, 1989). The observation of the two resolved upfield peaks with similar shifts and NOES in MPO-CN indicates that it also possesses such a residue in the pocket. The exchangeable proton peaks a* (broad) at 26.6 ppm and b* (narrow) at 16.0 ppm have the same shifts and relative line widths as labile proton peaks in cyanide-ligated horseradish peroxidase and LPO-CN, which in the former protein have been unequivocally determined to arise from a distal His hydrogen bonded to the bound cyanide. In LPO-CN, the added NOE to a nonexchangeable peak near 10 ppm confirmed the likely distal His origin in the protein (Thanabal and La Mar, 1989). While a similar NOE could not be detected in MPO-CN because of experimental limitations, the shifts and line widths argue for their origin from a protonated distal His. A ligand binding study has predicted a protonated distal His in MPO-CN (Ikeda-Saito, 1987). The peaks x, y, a*, and b* support the conclusion that both MPO and LPO possess distal His and Arg in the active site with similar stereochemistry relative to the prosthetic group. In MPO, sequence information (Kimura and Ikeda-Saito, 1988) indicates a segment His-X-Y-Arg with homology to the distal fragment of cytochrome c peroxidase and horseradish peroxidase, comparable sequence information is not available for LPO.
The Nature of the Prosthetic Group-The characteristic green color of MPO, as opposed to the red color of related peroxidases known to contain a heme, led to the proposal that MPO contains an iron chlorin, and certain features of both resonance Raman and MCD spectra (Sibbett and Hurst, 1984;Babcock et al., 1985;Sono et al., 1986) have been cited as support of this hypothesis.
The present nuclear magnetic resonance data contain three types of information: the hyperfine shift pattern which yields information on both the functional groups appended to the K skeleton and on the molecular orbital containing the unpaired spin, the NOES reflect the distance between the pairs of protons on different substituents, and the TIs qualitatively reflect the distance of the protons from the iron center and the intrinsic iron relaxivity as reflected in the electron spin relaxation time, Tie. All three of these observable sets are very similar for MPO-CN and LPO-CN, and hence provide support for the conclusion that the prosthetic groups in MPO and LPO have very similar molecular and electronic structures, particularly with respect to functionalization (i.e. 8-methylene rather than methyl group) of pyrroles B and D. The heteroatom attached to the 8-methylene group (X in 2 of Fig. 1) need not be the same in the two proteins. The inability to readily extract the prosthetic group from MPO could arise from a covalent link to the protein via the functionalized 8-methylene group. The small differences in chemical shifts between MPO-CN and LPO-CN for the homologous functional groups could arise from slight differences in the orientation of these groups, or from small differences in axial field modulation by interactions such as hydrogen bonding to the coordinated cyanide ligand (La Mar et al., 1977). The smaller g-tensor anisotropy for MPO-CN relative to LPO-CN (Ikeda-Saito et al., 1985) could also be interpreted to reflect strong hydrogen bonding to the bound cyanide in the former protein. It is of interest to note that the similarity of the spectral features for the hyperfine shifted resonances in MPO-CN and LPO-CN are stronger than for any other pair of such divergent pairs of hemoproteins (Satterlee, 1985). Thus the NMR spectral features, shifts, relaxation times, and NOES argue for a similarly functionalized prosthetic group in MPO and LPO. Since the red prosthetic group of LPO has been shown to be a porphyrin, that leads to the conclusion that MPO also contains a porphyrin.
The present IH NMR data do not support a chlorin as the prosthetic group in MPO, but also do not rule out this possibility. Generally the hyperfine shift patterns of an analogous porphyrin and chlorin, either as models, or embedded in the same heme pocket yield rather different contact shift magnitudes, as revealed in the comparison of cyanometmyoglobin and the green sulfmyoglobin derivatives (Chatfield et al., 1988a;Parker et al., 1989) or iron pheophorbide a as a model compound or reconstituted myoglobins (Licoccia et al., 1989).' Moreover, the intrinsic low spin iron(II1) relaxivity, as reflected in 'H relaxation rates, is generally significantly larger by factors of 2-5 in chlorin as compared to analogous porphyrin complexes, as observed for both the iron pheophorbide a and sulfhemin systems (Chatfield et al., 1988b;Parker et al., 1989;Licoccia et al., 1989). The similar 'H T,s in MPO-CN and LPO-CN argue for very similar electron spin relaxation times. Other NMR support of a porphyrin rather than a chlorin in MPO is derived from the proximal His ring NH chemical shift in the reduced high spin iron(I1) protein, reported earlier to resonate at 76 ppm (Ikeda-Saito and Inubushi, 1987), at essentially the same position as in deoxymyoglobin (La Mar et al., 1977). NMR experiments with deoxymyoglobin and hemoglobin reveal a strong upfield bias relative to a porphyrin by 6-12 ppm for this labile proton when they contain chlorins such as sulfhemins (Chatfield et al., 1987;Chatfield, 1987) or iron pheophorbide a.3 The present 'H NMR interpretation is contradictory to that derived from optical, resonance Raman, and MCD studies. MPO-CN exhibits an optical absorption spectrum with an a-band at 632 nm, a weak P-band near 560 nm, and a Soret band at 454 nm (Ikeda-Saito, 1985). Corresponding LPO-CN shows a more typical low spin ferric hemoprotein spectrum with a weak a-band around 575 nm as a shoulder of the P-band at 540 nm, and a Soret band at 416 nm (Sievers et al., 1983). Not only the position of the absorption bands, but also their relative intensities are different; the a-band of MPO-CN is more intense than its P-band, while the reverse is the case for LPO-CN.
The ratio of intensities of Soret to visible band are also different between LPO-CN and MPO-CN. The observed strong similarity of 'H NMR spectral parameters that leads to the conclusion for closely related prosthetic groups in MPO and LPO therefore have to rationalize the presence of either a red chlorin in LPO or a green porphyrin in MPO. It is known that several derivatized porphyrins such as N-alkylated (Lavallee, 1987) and meso-sulfursubstituted (Clezy et al., 1981) porphyrins exhibit highly perturbed optical spectra with similarly red-shifted Soret and visible bands, in particular, bands at 620-650 that impart the characteristic green color to both these porphyrins and MPO, as well as chlorins in general. A perturbation as small as adding a nitro group to the terminal carbon of a vinyl of protohemin renders it green in nitrimyoglobin, but minimally affects the 'H NMR spectral parameters (Bondoc and Timkovich, 1989). Axial ligation with phenoxide in models (Peisach and Gersonde, 1977) or tyrosinate in catalase (Holmquist and Vallee, 1973), and even solvation effects (Caughey, 1973), can impart green color to hemes. Theoretical work has suggested (Eccles and Honig, 1983) that charged groups near the heme can strongly perturb optical transition energies, although such effects were not observed in genetically engineered myoglobin mutants (Varadarajan et al., 1989). Such spectral shifts, however, are observed for chlorins embedded in photosynthetic reaction centers (Sauer, 1978). Hence the optical results are not necessarily inconsistent with a porphyrin skeleton, albeit a perturbed one. The MCD and resonance Raman spectra of MPO also differ from those of LPO (Kitagawa et al., 1983;Sievers et al., 1983;Manthey et al., 1986), although they are not the same as those of characterized chlorins.
The difficulty here may be the absence of any known systems which model the spectral characteristics of MPO. In the resonance Raman studies, a splitting of vq was observed, indicating a strong reduction of the symmetry of the macrocycle, although the splitting is less than that of known chlorins. The question that remains is the physical origin of this low symmetry perturbation.
The NMR data provide no direct information on the magnitude of a rhombic perturbation, since the nature of the orbital ground state is determined only by the location of the perturbation and is not affected by the magnitude as long as it is larger than kT (Shulman et al., 1971;Chatfield et al., 1988b). Possible origins for a strong rhombic perturbation for a porphyrin could arise from interaction of the heteroatom in the covalent link to the protein matrix if one accepts that the NMR similarities between MPO-CN and LPO-CN reflect the same porphyrin functionalization at least on the pyrroles B and D. The stronger perturbation in MPO relative to LPO could arise from a stronger interaction of the linking heteroatom (X in 2), with the heme r system.
With the available NMR, optical, resonance Raman, and MCD data on MPO, it is not possible to provide a unique and definitive rationalization of the unusual spectral features in terms of the detailed structure of the active site. Our purpose here was to show that the NMR properties of MPO-CN's active site are extraordinarily close to those of LPO-CN, and hence provide evidence that the prosthetic groups in both proteins are porphyrins.
Further spectroscopic research on these enzymes and more appropriate models, as well as efforts to extract the prosthetic group from MPO, should resolve the current paradox and provide us with a detailed structure of the prosthetic group.