How O2 Binds to Heme
2004; Elsevier BV; Volume: 279; Issue: 15 Linguagem: Inglês
10.1074/jbc.m314007200
ISSN1083-351X
Autores Tópico(s)Metal-Catalyzed Oxygenation Mechanisms
ResumoWe have used density functional methods to calculate fully relaxed potential energy curves of the seven lowest electronic states during the binding of O2 to a realistic model of ferrous deoxyheme. Beyond a Fe–O distance of ∼2.5 Å, we find a broad crossing region with five electronic states within 15 kJ/mol. The almost parallel surfaces strongly facilitate spin inversion, which is necessary in the reaction of O2 with heme (deoxyheme is a quintet and O2 a triplet, whereas oxyheme is a singlet). Thus, despite a small spin-orbit coupling in heme, the transition probability approaches unity. Using reasonable parameters, we estimate a transition probability of 0.06–1, which is at least 15 times larger than for the nonbiological Fe–O+ system. Spin crossing is anticipated between the singlet ground state of bound oxyheme, the triplet and septet dissociation states, and a quintet intermediate state. The fact that the quintet state is close in energy to the dissociation couple is of biological importance, because it explains how both spin states of O2 may bind to heme, thereby increasing the overall efficiency of oxygen binding. The activation barrier is estimated to be <15 kJ/mol based on our results and Mössbauer experiments. Our results indicate that both the activation energy and the spin-transition probability are tuned by the porphyrin as well as by the choice of the proximal heme ligand, which is a histidine in the globins. Together, they may accelerate O2 binding to iron by ∼1011 compared with the Fe–O+ system. A similar near degeneracy between spin states is observed in a ferric deoxyheme model with the histidine ligand hydrogen bonded to a carboxylate group, i.e. a model of heme peroxidases, which bind H2O2 in this oxidation state. We have used density functional methods to calculate fully relaxed potential energy curves of the seven lowest electronic states during the binding of O2 to a realistic model of ferrous deoxyheme. Beyond a Fe–O distance of ∼2.5 Å, we find a broad crossing region with five electronic states within 15 kJ/mol. The almost parallel surfaces strongly facilitate spin inversion, which is necessary in the reaction of O2 with heme (deoxyheme is a quintet and O2 a triplet, whereas oxyheme is a singlet). Thus, despite a small spin-orbit coupling in heme, the transition probability approaches unity. Using reasonable parameters, we estimate a transition probability of 0.06–1, which is at least 15 times larger than for the nonbiological Fe–O+ system. Spin crossing is anticipated between the singlet ground state of bound oxyheme, the triplet and septet dissociation states, and a quintet intermediate state. The fact that the quintet state is close in energy to the dissociation couple is of biological importance, because it explains how both spin states of O2 may bind to heme, thereby increasing the overall efficiency of oxygen binding. The activation barrier is estimated to be <15 kJ/mol based on our results and Mössbauer experiments. Our results indicate that both the activation energy and the spin-transition probability are tuned by the porphyrin as well as by the choice of the proximal heme ligand, which is a histidine in the globins. Together, they may accelerate O2 binding to iron by ∼1011 compared with the Fe–O+ system. A similar near degeneracy between spin states is observed in a ferric deoxyheme model with the histidine ligand hydrogen bonded to a carboxylate group, i.e. a model of heme peroxidases, which bind H2O2 in this oxidation state. All electrons have a spin, which is an intrinsic quantum chemical property that can take only two possible values, normally called α and β (or spin up and down). Almost all normal organic molecules contain an even number of electrons and also an equal number of α and β electrons. They are then said to have paired spin or to be singlets. Molecular oxygen (O2) is a famous exception to this rule. In its ground state, it has two more electrons of one spin state than the other. Thus, it is said to have two unpaired electrons or to be a triplet. The singlet state of O2, with all electron spins paired, is ∼90 kJ/mol higher in energy than the triplet ground state (1Kaim W. Schwederski B. Bioinorganic Chemistry: Inorganic Elements in the Chemistry of Life. John Wiley & Sons, West Sussex, UK1994: 84Google Scholar). A chemical reaction can normally not change the spin state of an electron. Therefore, reactions between singlet and triplet states are formally spin-forbidden, which means that they are slow. This is the reason why organic matter may exist in an atmosphere containing much O2. There is a strong thermodynamic drive of O2 to oxidize organic matter to H2O and CO, but because these products (as well as the organic molecules) are singlets (whereas O2 is a triplet), this reaction is spin-forbidden and therefore very slow at ambient temperatures. On the other hand, this is a problem when living organisms want to employ O2 in their metabolism; the reactions are still spin-forbidden and slow. Nature has handled this problem by using transition metals to carry, activate, and reduce O2. There are many reasons for this choice. First, most transition metals also contain unpaired electrons, allowing reactions with triplet O2. Second, transition metals are relatively heavy atoms, which increases spin-orbit coupling (SOC), 1The abbreviations used are: SOC, spin-orbit coupling; DFT, density functional theory. 1The abbreviations used are: SOC, spin-orbit coupling; DFT, density functional theory. and thereby provide a quantum mechanical mechanism to change the spin state of an electron, called spin inversion. However, the SOC of the first-row transition metals is too small alone to allow for spin transitions. Third, transition metals often have several excited states with unpaired electrons close in energy to the ground state. This can also be used to enhance the probability of spin inversion. One of the most simple biological reactions involving molecular oxygen is the binding of O2 to hemoglobin, i.e. the binding of O2 to the Fe(II) ion in a heme group. This reaction is formally spin-forbidden, because the reactant deoxyheme contains four unpaired electrons in the 3d orbitals of iron (it is a quintet), and triplet O2 has two unpaired electrons. Thus, depending on the relative direction of these two sets of unpaired electrons, the adduct would be expected to have two (4 – 2) or six (4 + 2) unpaired electrons (i.e. a triplet or a septet state). However, experimentally, the product complex is a singlet state with an equal number of α and β electrons. As discussed already by Pauling and Coryell (2Pauling L. Coryell C.D. Proc. Natl. Acad. Sci. U. S. A. 1936; 22: 210-216Crossref PubMed Google Scholar, 3Pauling L. Nature. 1964; 203: 182-183Crossref PubMed Scopus (146) Google Scholar), this problem makes the hemoglobin reactions troublesome to understand (4Kotani M. Rev. Mod. Phys. 1963; 35: 717-720Crossref Scopus (23) Google Scholar), and it is not clear how nature has coped with the spin-forbidden nature of this reaction. The importance of spin inversion is also reflected in the Perutz model of hemoglobin cooperativity (5Perutz M.F. Fermi G. Luisi B. Shaanan B. Liddington R.C. Acc. Chem. Res. 1987; 20: 309-321Crossref Scopus (414) Google Scholar, 6Perutz M.F. Wilkinson A.J. Paoli M. Dodson G.G. Annu. Rev. Biophys. Biomol. Struct. 1998; 27: 1-34Crossref PubMed Scopus (470) Google Scholar, 7Scheidt R.W. Reed C.A. Chem. Rev. 1981; 81: 543-555Crossref Scopus (674) Google Scholar). The movement of iron into the heme plane is assumed to trigger a transition from a tense state to a relaxed state after the binding of two oxygen molecules, and this trigger, in the form of the Fe–Nax pull, depends on the spin state of heme. Theoretical methods have been successfully applied to many problems in heme chemistry. Already the simple Hartree-Fock formalism correctly predicts the bent form of the O2 adduct (8Dedieu A. Rohmer M.-M. Benard M. Veillard A. J. Am. Chem. Soc. 1976; 98: 3717-3718Crossref PubMed Scopus (71) Google Scholar), whereas state-of-the-art density functional theory (DFT) provides excellent geometries of porphyrins in general (9Ghosh A. Acc. Chem. Res. 1998; 31: 189-198Crossref Scopus (180) Google Scholar, 10Rovira C. Parrinello M. Int. J. Quant. Chem. 1998; 70: 387-394Crossref Scopus (23) Google Scholar, 11Jensen K.P. Ryde U. ChemBioChem. 2003; 4: 413-424Crossref PubMed Scopus (61) Google Scholar, 12Johansson M.P. Sundholm D. Gerfen G. Wikström M. J. Am. Chem. Soc. 2002; 124: 11771-11780Crossref PubMed Scopus (59) Google Scholar, 13Vogel K.M. Kozlowski P.M. Zgierski M.Z. Spiro T.G. J. Am. Chem. Soc. 1999; 121: 9915-9921Crossref Scopus (187) Google Scholar). Among these, the B3LYP density functional predicts very close-lying quintet and triplet states in deoxyheme models, sometimes with a triplet ground state (14Rovira C. Kunc K. Hutter J. Ballone P. Parrinello M. J. Phys. Chem. A. 1997; 101: 8914-8925Crossref Scopus (365) Google Scholar). Recently, DFT was used to compute the electronic spectrum of FeII-porphine with 2-methylimidazole as the axial ligand, making the quintet state the lowest in energy, but the triplet state only 12 kJ/mol higher (15Liao M.-S. Scheiner S. J. Chem. Phys. 2002; 116: 3635-3645Crossref Scopus (83) Google Scholar), in excellent agreement with the experiment. These circumstances indicate that the treatment of spin states in porphyrins is delicate because of the closeness in energy of various spin states. The fact that spin inversion occurs in globins during oxygen binding means that all low-lying states, independent of their number of unpaired electrons, need to be considered in a proper study of the reaction. Spin-dependent mechanisms relevant for the present work have been studied in particular by Franzen (16Franzen S. Proc. Natl. Acad. Sci. U. S. A. 2002; 99: 16754-16759Crossref PubMed Scopus (98) Google Scholar) and Poli and Harvey (17Poli R. Harvey J.N. Chem. Soc. Rev. 2003; 32: 1-8Crossref PubMed Scopus (400) Google Scholar). In this work, we have optimized the ground state and several low-lying excited states at many points along the heme–O2 binding curve. Our results indicate the reason for the facilitated binding of O2 to heme is a broad crossing region of the relevant spin states, which provides significant transition probabilities. We show that porphyrin is an ideal iron ligand for the spin transition problem, because it tunes the spin states to be close in energy, giving parallel binding curves, small activation energies, and large transition probabilities. This finding explains why the porphyrin ring is designed to bring spin states close in energy and why spin inversion and reversible binding is possible in heme proteins. We also provide evidence that similar arguments apply to other heme proteins, e.g. the heme peroxidases, where near degeneracy, in this case in the ferric state, is caused by strengthening the ligand field of the proximal histidine by a hydrogen bond to a carboxylate group. Hence, we suggest a new role for the choice of an axial ligand in such systems, viz. to bring spin states close in energy and thereby facilitate spin-forbidden binding of ligands. In the present work, we studied the reversible binding process FeII(heme)+O2↔FeII(heme)(O2)Reaction 1 with particular emphasis on analyzing possible states along this reaction coordinate. Such a detailed approach seems necessary to study the nature of the reversible process. Computational Details—All geometry optimizations were performed with the Becke 1988 exchange functional together with the Perdew 1986 correlation functional (BP86) (18Perdew J.P. Phys. Rev. B. 1986; 33: 8822-8824Crossref PubMed Scopus (17014) Google Scholar, 19Becke A.D. Phys. Rev. A. 1988; 38: 3089-3097Crossref Scopus (40392) Google Scholar). Accurate energies were then estimated by single-point calculations using the Becke three-parameter hybrid method with the local spin-density approximation correlation functional of Vosko-Wilk-Nusair and the nonlocal Lee-Yang-Parr correlation functional (B3LYP) (20Becke A.D. Phys. Rev. A. 1988; 38: 3098-3100Crossref PubMed Scopus (44777) Google Scholar, 21Becke A.D. J. Chem. Phys. 1992; 96: 2155-2160Crossref Scopus (2201) Google Scholar, 22Becke A.D. J. Chem. Phys. 1993; 98: 1372-1377Crossref Scopus (13560) Google Scholar, 23Becke A.D. J. Chem. Phys. 1993; 98: 5648-5662Crossref Scopus (88793) Google Scholar, 24Lee C. Yang W. Parr R.G. Phys. Rev. B. 1988; 37: 785-789Crossref PubMed Scopus (85911) Google Scholar, 25Hertwig R.H. Koch W. Chem. Phys. Lett. 1997; 268: 345-351Crossref Scopus (799) Google Scholar). B3LYP is probably the most accurate of the generally available exchange-correlation functionals for calculating relative energies and frequencies (26Bauschlicher C.W. Chem. Phys. Lett. 1995; 246: 40-44Crossref Scopus (398) Google Scholar, 27Siegbahn P.E.M. Blomberg M.R.A. Chem. Rev. 2000; 100: 421-437Crossref PubMed Scopus (585) Google Scholar, 28Siegbahn P.E.M. Blomberg M.R.A. Annu. Rev. Phys. Chem. 1999; 50: 221-249Crossref PubMed Scopus (254) Google Scholar, 29Jensen F. Introduction to Computational Chemistry. John Wiley & Sons, Inc., New York1999: 188-189Google Scholar). However, in our experience, BP86 provides slightly better geometries for metal complexes than B3LYP and at an appreciably lower cost (30Ryde U. Nilsson K. J. Am. Chem. Soc. 2003; 125: 14232-14233Crossref PubMed Scopus (109) Google Scholar). The calculations were carried out with the Turbomole software, version 5.6 (31Alrichs R. Bär M. Häser M. Horn H. Kölmel C. Chem. Phys. Lett. 1989; 162: 165-169Crossref Scopus (7403) Google Scholar). The basis sets used for geometry optimization were 6–31G(d) for all atoms except iron, which was described by the double-ζ basis set of Schäfer et al. (32Schäfer A. Horn H. Ahlrichs R. J. Chem. Phys. 1992; 97: 2571-2577Crossref Scopus (7954) Google Scholar), augmented with two p, one d, and one f function (DZpdf, exponents: 0.141308 and 0.043402 (p); 0.1357 (d); and 1.6200 (f)) with the contraction scheme (14s11p6d1f)/((8s7p4d1f). Only the pure five d and seven f-type functions were used. Our basis set is balanced and, based on experience (34Sigfridsson E. Ryde U. J. Biol. Inorg. Chem. 1999; 4: 99-110Crossref PubMed Scopus (101) Google Scholar), is flexible enough to account for the electronic structure and polarization effects encountered in heme systems. We applied the default (m3) grid size of Turbomole, and all optimizations were carried out in redundant internal coordinates. Unrestricted calculations were performed for all open-shell systems. We made use of default convergence criteria, which imply self-consistency down to 10–6 Hartree (2.6 J/mol) for the energy and 10–3 atomic units for the maximum norm of the gradient. Model System—All calculations were performed on the FeIIPorImO2 model, where Por is porphine (heme without side chains) and Im is imidazole, a model of the proximal histidine. We calculated the structure and energetics of Fe–O bond breaking for the seven lowest states by systematically increasing the Fe–O bond distance and optimizing the structure with a fixed Fe–O distance. From the fully optimized potential energy surfaces, we obtain the crossing points of the various spin states involved in the binding mechanism. Strictly speaking, proper crossing points would require identical structures of the states (true transition states). Methods to obtain such structures have been developed, but the geometric effect is usually small and does not significantly affect the location of the crossing point along the reaction coordinate (17Poli R. Harvey J.N. Chem. Soc. Rev. 2003; 32: 1-8Crossref PubMed Scopus (400) Google Scholar, 35Harvey J.N. J. Am. Chem. Soc. 2000; 122: 12401-12402Crossref Scopus (103) Google Scholar). We found that the most stable state of the FeIIPorImO2 model had Cs symmetry. This is consistent with the most accurate crystal structure of oxymyoglobin (1.0 Å resolution) (36Vojtechovsky J. Chu K. Berendzen J. Sweet R.M. Schlichting I. Biophys. J. 1999; 77: 2153-2174Abstract Full Text Full Text PDF PubMed Scopus (502) Google Scholar), in which the imidazole ligand has a staggered conformation, with a C–Nax–Fe–Neq torsion angle of 45°.O2 adopts a 4-fold occupancy in the thermally disordered structure at staggered conformations with respect to the equatorial Fe–Neq bonds: two coplanar with imidazole and two orthogonal to it. Hence, this structure indicates that there are two binding modes of O2, one with Cs symmetry and the other unsymmetrical C1. We have optimized the structure of both states, and the symmetric one was 0.6 kJ/mol more stable than the unsymmetric one. The geometries, charges, and spin densities of the two states were identical to within the accuracy of the present work, i.e. ±0.01e and 0.001 Å. In addition, the calculation shows that π-bonding and trans electronic effects, including back-bonding to iron d-orbitals, are absent. This implies that imidazole is an innocent ligand. In fact, it has been shown that the rotation of the O2 group in a model similar to ours has a barrier of less than 8 kJ/mol (37Rovira C. Parrinello M. Int. J. Quant. Chem. 2000; 80: 1172-1180Crossref Scopus (45) Google Scholar). Likewise, another unsymmetrical conformation arising from a 45° rotation of imidazole (staggered oxygen and eclipsed imidazole with respect to the Fe–Neq bonds) was 2 kJ/mol less stable than the Cs conformation. The spin densities and charges were similar to within 0.02e of the Cs state, but the geometry showed differences of up to 0.07 Å in the Fe–O bond. Thus, we can conclude that the symmetric structure is the most stable geometry of this system and we have therefore employed Cs symmetry in all the calculations. This strongly facilitates the optimization and characterization of the various excited states. In Cs symmetry, the electronic states are labeled as symmetric A′ or antisymmetric A″, respectively, depending on whether their wave function preserves or changes sign upon reflection in the symmetry plane (xz, cf. Fig. 1). The states that we shall discuss will follow this nomenclature with a superscript in front indicating the multiplicity (the number of unpaired electrons plus one). States with the same symmetry and multiplicity are numbered (in brackets) after their optimum energy. For example, the ground state is a symmetric singlet, 1A′(1). Throughout this work, we used a coordinate system with Fe in the origin, the z-axis along the NIm–Fe–O bonds, the x-axis through two methine bridge atoms (not the nitrogens), and the y-axis through the two other methine bridges (Fig. 1). Thus, the imidazole and O2 molecules lie in the xz plane. The two unpaired π-electrons on oxygen are situated in two degenerate antibonding π orbitals, which transform as a′ and a″ in the reduced Cs symmetry when binding to deoxyhemoglobin. In our coordinate system, three of the Fe d-orbitals transform as a′, viz. xz, z2, and x2 – y2 and would therefore couple to the a′ unpaired π-electron of O2, whereas the other two orbitals, xy and yz, would interact with the a″ electrons instead. Selection of States—We have searched for low-energy states that could contribute to the process of oxygen binding. The states were obtained from a systematic permutation of the occupations of 6 electrons in an active space consisting of molecular orbitals 73–75 a′ and 45–47 a″. There are four classes of orbitals: symmetric and antisymmetric α and β orbitals. Some restrictions were introduced to minimize the search, based on the ground state. We have only examined those configurations that distinguish themselves by one occupied orbital per class from the ground-state configuration (74 45 74 45, i.e. 74 electrons in α a′ orbitals, 45 electrons in α a″ orbitals, 74 electrons in β a′ orbitals, and 45 electrons in β a″ orbitals). Some orbitals were found to be very high in energy and were subsequently avoided. For example, the state (75 44 74 45) had 212 kJ/mol higher energy than the ground state. Hence, we avoided the (45→75) excitation. Such selections reduced our number of states to 20, and the seven lowest are presented in this work. The electronic configurations and optimized energies of these states are shown in Table I. The states are unrestricted Kohn-Sham wave functions with a large degree of spin polarization in most cases.Table IRelative energies (kJ/mol) and occupation numbers of the seven lowest states of oxyheme in Cs symmetryStatea′-αa″-αa′-βa″-βErel1A′(1)744574450.05A′(1)7546734422.03A″(1)7446734524.13A″(2)aAntiferromagnetic state obtained from the septet dissociation product.7446734519.77A″(1)7547724428.51A″(1)7346744528.93A′(1)7446744424.7a Antiferromagnetic state obtained from the septet dissociation product. Open table in a new tab The Ground State of the Adduct—The lowest energy was obtained for the (74 45 74 45) open-shell singlet 1A′(1) state in Table I (the lowest closed-shell singlet with the same occupation numbers is 5 kJ/mol higher in energy). Its geometry is displayed in Table II and Fig. 1. It can be seen that it closely resembles the x-ray structure of oxymyoglobin (36Vojtechovsky J. Chu K. Berendzen J. Sweet R.M. Schlichting I. Biophys. J. 1999; 77: 2153-2174Abstract Full Text Full Text PDF PubMed Scopus (502) Google Scholar). The Fe–O bond lengths differ by only 0.001 Å. For the more soft Fe–Nax bond, the error is what can be expected with state-of-the-art DFT methods, 0.03 Å, whereas for the average equatorial Fe–Neq bonds, the difference is only 0.006 Å. This gives us confidence that this is the correct ground state and that the description of the Fe–O bond is essentially correct.Table IIOptimum bond distances (Å) of the seven lowest states of oxyheme, obtained with the BP86 functionalStateFe—OFe—NaxFe—Neq1Fe—Neq2Fe oopeDistance of the Fe ion out of the porphyrin plane (oop).1A′(1)1.8072.0962.0242.0010.0345A′(1)2.6792.2532.0121.9980.0873A″(1)1.9522.1002.0112.0050.0083A″(2)1.8922.1332.0792.0650.0187A″(1)2.5192.2002.0862.0690.1441A″(1)1.8782.0712.0122.0080.0123A′(1)2.0802.1352.0752.0750.044oxyexpaExperimental structure of oxymyoglobin.1.8062.0642.006bAverage of the four Ne—Neq distances.2.006bAverage of the four Ne—Neq distances.0.023deoxyexpcExperimental structure of deoxymyoglobin (36).2.1412.074bAverage of the four Ne—Neq distances.2.074bAverage of the four Ne—Neq distances.0.3647A″(1)dissdGeometry at largest Fe—O distance, 4.3 Å.4.3002.1582.0822.0790.257deoxycalc2.1532.0832.0790.266a Experimental structure of oxymyoglobin.b Average of the four Ne—Neq distances.c Experimental structure of deoxymyoglobin (36Vojtechovsky J. Chu K. Berendzen J. Sweet R.M. Schlichting I. Biophys. J. 1999; 77: 2153-2174Abstract Full Text Full Text PDF PubMed Scopus (502) Google Scholar).d Geometry at largest Fe—O distance, 4.3 Å.e Distance of the Fe ion out of the porphyrin plane (oop). Open table in a new tab The ground state is an open-shell singlet, in accordance with the experimental observation that the O2 adduct is silent in electron paramagnetic resonance experiments (38Momenteau M. Reed C.A. Chem. Rev. 1994; 94: 659-698Crossref Scopus (615) Google Scholar). However, the spin is unevenly distributed in the complex with a surplus of α spin on O2 (0.75 electrons) and a surplus of β spin on iron (-0.79e), as is quantified in Table III and illustrated in Fig. 1, bottom. The literature is rich in discussions about the nature of the Fe–O bond (38Momenteau M. Reed C.A. Chem. Rev. 1994; 94: 659-698Crossref Scopus (615) Google Scholar, 39Niu S. Hall M.B. Chem. Rev. 2000; 100: 353-406Crossref PubMed Scopus (836) Google Scholar). In particular, it has been argued whether the electronic structure of oxyheme is better described as singlet oxygen bound to low-spin FeII (2Pauling L. Coryell C.D. Proc. Natl. Acad. Sci. U. S. A. 1936; 22: 210-216Crossref PubMed Google Scholar) or as a superoxide radical antiferromagnetically coupled to low-spin FeIII (40Weiss J.J. Nature. 1964; 202: 83-84Crossref PubMed Scopus (284) Google Scholar). Some consensus has arisen on the point that the FeIII-O2- form agrees better with experiments, e.g. the O–O frequency of 1100 cm–1, which is close to what is expected for O2- (3Pauling L. Nature. 1964; 203: 182-183Crossref PubMed Scopus (146) Google Scholar), some aspects of the chemical reactivity (41Kaim W. Schwederski B. Bioinorganic Chemistry: Inorganic Elements in the Chemistry of Life. John Wiley & Sons, West Sussex, UK1994: 93Google Scholar), and changes in the electric field gradient studied with Mössbauer spectroscopy (42Bade D. Parak F.Z. Z. Naturforsch. C. 1978; 33: 488-494Crossref PubMed Scopus (12) Google Scholar).Table IIISpin densities on various atoms of the seven lowest states of oxyheme, obtained with the BP86 functionalStateFeO1O21A′(1)-0.790.260.495A′(1)2.250.900.923A″(1)3.35-0.73-0.843A″(2)2.98-0.68-0.457A″(1)3.890.870.911A″(1)-0.770.350.413A′(1)1.050.380.58 Open table in a new tab Our results are closest to the FeIII-O2- description in accordance with earlier DFT calculations (14Rovira C. Kunc K. Hutter J. Ballone P. Parrinello M. J. Phys. Chem. A. 1997; 101: 8914-8925Crossref Scopus (365) Google Scholar) (the FeII–O2 form would be a closed-shell singlet). However, the spin densities are far from ±1, which clearly shows that the electronic structure cannot be fully described by a single configuration (such as FeIII-O2- or FeII–O2), but rather as a mixture of both these and possibly also other configurations. Thus, our spin densities could be interpreted as a mixture of 75–80% FeIII-O2- and 20–25% FeII–O2. This is in accordance with the experimental observation that a quantum mixture of approximately two-thirds ferric and one-third ferrous states gives the best agreement with Mössbauer spectra (43Tsai T.E. Groves J.L. Wu C.S. J. Chem. Phys. 1981; 74: 4306-4314Crossref Scopus (19) Google Scholar). Thus, oxyheme is inherently multiconfigurational, with an electronic structure that is somewhat analogous to that found in ozone (44Case D.A. Huynh B.H. Karplus M. J. Am. Chem. Soc. 1977; 101: 4433-4453Crossref Scopus (134) Google Scholar). Early complete active space self-consistent field studies (on a simplified heme model with ammonia instead of imidazole) gave a mainly closed-shell 1A′ ground state (45Yamamoto S. Kashiwagi H. Chem. Phys. Lett. 1989; 161: 85-89Crossref Scopus (36) Google Scholar), as did a symmetry-adapted cluster configuration interaction (SAC-CI) study (46Nakatsuji H. Hasegawa J. Hada M. Chem. Phys. Lett. 1996; 250: 379-386Crossref Scopus (35) Google Scholar), with the lowest open-shell singlet 150 kJ/mol higher in energy. However, the present results give a better description of the ground state in terms of geometry. The Dissociated States—Isolated deoxyheme is experimentally a high-spin quintet (38Momenteau M. Reed C.A. Chem. Rev. 1994; 94: 659-698Crossref Scopus (615) Google Scholar). The optimized structure of this complex (Table II) agrees well (within 0.02 Å) with the crystal structure of deoxymyoglobin at 1.15 Å resolution (36Vojtechovsky J. Chu K. Berendzen J. Sweet R.M. Schlichting I. Biophys. J. 1999; 77: 2153-2174Abstract Full Text Full Text PDF PubMed Scopus (502) Google Scholar). It is notable that both structures show a strongly distorted porphyrin with the iron ion ∼0.3 Å out of the ring plane, illustrating that high-spin iron is too large to fit into the ring cavity. When this complex is associated with triplet O2, there are six unpaired electrons in the total system. The unpaired spin on deoxyheme and O2 may be either parallel, giving rise to a septet, 7A″(1), or antiparallel, which gives rise to a triplet state, which turns out to be 3A″(2). At long (noninteracting) Fe–O distances, these two states are degenerate, as expected. Ideally, both states should give rise to rapid O2 binding (i.e. all active sites of hemoglobin should be able to bind all O2 molecules, independent of their spin states). However, as the Fe–O distance is decreased, the degeneracy is lifted. In the optimal structure, the 7A″(1) state has a Fe–O bond length of 2.52 Å, whereas it is 1.89 Å for state 3A″(2) (cf. Table II). The potential energy surface of the 7A″(1) state is flat around the minimum, and the energy is close to the dissociation limit, which is at 27 kJ/mol when calculated from separated species. The two states have very similar energies in their optimum geometries. Interestingly, the B3LYP method gives a quite different behavior of the 3A″(2) state. The energy of this state increases steadily as the Fe–O bond length is decreased, with an energy of ∼50 kJ/mol at the BP86 minimum. The B3LYP curve shows a very shallow minimum at Fe–O = 2.39 Å, with an energy close to the dissociation limit. The lowest triplet (intermediate-spin) state of deoxyheme is close in energy to the lowest quintet state. In fact, in the present calculations (as well as in most previous DFT calculations (11Jensen K.P. Ryde U. ChemBioChem. 2003; 4: 413-424Crossref PubMed Scopus (61) Google Scholar, 34Sigfridsson E. Ryde U. J. Biol. Inorg. Chem. 1999; 4: 99-110Crossref PubMed Scopus (101) Google Scholar)), it is actually 3 kJ/mol more stable (4 kJ/mol when optimized at the B3LYP level; hence the dissociation limit of the lowest triplet state is 24 kJ/mol). Thus the states are degenerate to within the uncertainty of the method. If this triplet state is associated with triplet O2, we once again obtain two states, depending on the relative orientation of the two sets of unpaired spin, a quintet state 5A′(1) and a singlet state, which actually turns out to be the dissociation product of the singlet ground state 1A′(1). Excited Sta
Referência(s)