Article Cite This: Acc. Chem. Res. 2018, 51, 427−435
pubs.acs.org/accounts
The Nature and Reactivity of Ferryl Heme in Compounds I and II Peter C. E. Moody‡ and Emma L. Raven*,† ‡
Department of Molecular and Cell Biology and Leicester Institute of Structural and Chemical Biology, University of Leicester, Lancaster Road, Leicester LE1 9HN, England † Department of Chemistry and Leicester Institute of Structural and Chemical Biology, University of Leicester, University Road, Leicester LE1 7RH, U.K. CONSPECTUS: Aerobic organisms have evolved to activate oxygen from the atmosphere, which allows them to catalyze the oxidation of different kinds of substrates. This activation of oxygen is achieved by a metal center (usually iron or copper) buried within a metalloprotein. In the case of iron-containing heme enzymes, the activation of oxygen is achieved by formation of transient iron-oxo (ferryl) intermediates; these intermediates are called Compound I and Compound II. The Compound I and II intermediates were first discovered in the 1930s in horseradish peroxidase, and it is now known that these same species are used across the family of heme enzymes, which include all of the peroxidases, the heme catalases, the P450s, cytochrome c oxidase, and NO synthase. Many years have passed since the first observations, but establishing the chemical nature of these transient ferryl species remains a fundamental question that is relevant to the reactivity, and therefore the usefulness, of these species in biology. This Account summarizes experiments that were conceived and conducted at Leicester and presents our ideas on the chemical nature, stability, and reactivity of these ferryl heme species. We begin by briefly summarizing the early milestones in the field, from the 1940s and 1950s. We present comparisons between the nature and reactivity of the ferryl species in horseradish peroxidase, cytochrome c peroxidase, and ascorbate peroxidase; and we consider different modes of electron delivery to ferryl heme, from different substrates in different peroxidases. We address the question of whether the ferryl heme is best formulated as an (unprotonated) FeIVO or as a (protonated) FeIV− OH species. A range of spectroscopic approaches (EXAFS, resonance Raman, Mossbauer, and EPR) have been used over many decades to examine this question, and in the last ten years, X-ray crystallography has also been employed. We describe how information from all of these studies has blended together to create an overall picture, and how the recent application of neutron crystallography has directly identified protonation states and has helped to clarify the precise nature of the ferryl heme in cytochrome c peroxidase and ascorbate peroxidase. We draw comparisons between the Compound I and Compound II species that we have observed in peroxidases with those found in other heme systems, notably the P450s, highlighting possible commonality across these heme ferryl systems. The identification of proton locations from neutron structures of these ferryl species opens the door for understanding the proton translocations that need to occur during O−O bond cleavage.
1. INTRODUCTION A widespread role for iron in biological systems is its ability to form high-valent iron(IV)−oxo intermediates, in which the iron is oxidized above the ferric (FeIII) resting oxidation state. These iron(IV)−oxo species are implicated as key intermediates in the mechanisms of both heme and non-heme iron enzymes. Following conventions for naming of oxo complexes of other metal ions, which commonly use a “-yl” suffix (e.g., chromyl), the correct terminology1 for an iron(IV)−oxo complex is “ferryl” (not oxyferryl, which is sometimes used). Thus, across the family of catalytic heme-containing enzymes these iron(IV)−oxo species tend also to be referred to as “ferryl” heme species. A large number of heme enzymes use ferryl heme as part of their catalytic cycle: these include all of the cytochrome P450s, the heme peroxidases (including diheme enzymes), heme catalases, the NO synthases, cytochrome c oxidase, the heme © 2018 American Chemical Society
tryptophan dioxygenases, and very likely other as yet unidentified catalytic heme enzymes.2−8 Ferryl iron species are also used in non-heme iron systems.9−15 With such widespread occurrence and versatility, it is no wonder that there is interest in the properties and reactivity of ferryl heme. In the case of the heme enzymes, the focus of this Account, there are two ferryl intermediates, known as Compound I and Compound II, which differ in their overall level of oxidation and which form sequentially during catalysis. Compound I forms initially in the catalytic cycle; it is a two-equivalent oxidized intermediate, is usually green in color, and was first discovered in horseradish peroxidase (HRP) by Theorell.16 Compound II forms by one-electron reduction of Compound I; it is a brown intermediate and was discovered by Keilin and Received: September 20, 2017 Published: January 12, 2018 427
DOI: 10.1021/acs.accounts.7b00463 Acc. Chem. Res. 2018, 51, 427−435
Article
Accounts of Chemical Research
Figure 1. (A) Overlay of the structures of APX (PDB 1OAF,47 in gray) and CcP (PDB 2PCC,100 in cyan). The heme group (in CcP only) and the distal and proximal His residues (in both) are indicated in red. (B) Schematic illustration of the Compound I species as formed in HRP (left), CcP (center), and APX (right). For HRP, APX, and most other peroxidases, Compound I is green in color and contains a ferryl heme (usually written as FeIVO but sometimes as [FeIVO]2+ to show the formal charge) and a porphyrin π-cation radical. In CcP, a tryptophan radical is used instead, which makes Compound I a red/brown color. See text for details. (C) The UV−visible spectrum of (the green) Compound I in APX shows a lowering of the Soret band, as is also seen in HRP, and is the hallmark of a porphyrin π-cation radical; the spectrum of Compound II in APX is also shown. (D) The structure of the heme group, showing the nomenclature used in this Account. The two main substrate binding locations for peroxidase substrates, at the γ- and δ-heme edge, are indicated. (E) Overlay of the APX/ascorbate (PDB 1OAF,47 in gray) and CcP/cytochrome c (PDB 2PCC,100 in cyan) substrate-bound complexes. In the CcP/c complex, the heme group of cytochrome c is located on an electron transfer pathway that includes the Trp191 radical used in CcP Compound I; in the APX/ascorbate complex, the substrate binds at a different location and is hydrogen bonded (gray dotted lines) to the heme propionate, which is consistent with the formation of a porphyrin π-cation radical in Compound I of APX and electron transfer directly to the heme via the propionate. The proximal His/Asp/Trp triad is shown; amino acids are labeled in the format CcP/APX. It was later shown101 that it is possible to engineer an ascorbate binding site, analogous to that found in APX, into CcP.
Mann.17 Both species were initially, and incorrectly, assigned as enzyme−substrate (Michaelis) complexes, and only later were given the names Compound I and Compound II.18−20 Many decades have passed since the first observations, but the nature of the heme ferryl species remains both highly topical and often controversial.21−24 It is important because the oxidized ferryl species can control the reactivity of an enzyme, directing the catalysis, for example, toward hydroxylation of unactivated C− H bonds (in the P450s) or electron transfer (in the peroxidases).21 This Account will discuss the properties and reactivities of the ferryl species in heme peroxidases, making comparisons with what is known in other heme species.
peroxidases) and the fungal Coprinus cinereus peroxidase30 also use a porphyrin π-cation radical in their Compound I species. These porphyrin π-cation radical intermediates are typically stable for tens of minutes30−32 (although Compound I in lignin peroxidase33 is much less stable and decays with a half-life of ≈1 min). No such green Compound I intermediate is observed in CcP. Unlike all the other peroxidases, Compound I of yeast CcP contains a tryptophan radical34 (Figure 1B), which is also stable for minutes at room temperature35,36 and indefinitely at 77 K.37,38 The same Trp radical (Trp208) also forms in the cytochrome c peroxidase from Leishmania major.39 CcP thus sits as an outlier to other peroxidases that use a porphyrin π-cation radical instead.
2. FORMATION AND STABILITY OF COMPOUND I IN HEME PEROXIDASES The majority of studies on ferryl heme species have been carried out on one of several heme peroxidases. The justification for doing so is that the intermediates are more amenable to study in these peroxidases and because, as we explain below, it is highly likely that the structure of the ferryl heme is common across all heme proteins so that information on peroxidases informs the debate across the whole family. The two most well-known peroxidases, horseradish peroxidase (HRP) and cytochrome c peroxidase (CcP), have different Compound I species (Figure 1B). Compound I in HRP contains a ferryl heme together with a porphyrin π-cation radical. Other peroxidases such as manganese peroxidase,25,26 lignin peroxidase,27 chloroperoxidase28,29 (which is thiolateligated instead of the more typical histidine ligation as in other
3. COMPARISON OF CYTOCHROME c PEROXIDASE WITH ASCORBATE PEROXIDASE It had long been presumed that the reason that CcP used a protein radical instead of a porphyrin π-cation radical was because CcP contained an easily oxidizable Trp residue close to the heme active site, whereas all other peroxidases contained a (nonoxidizable) Phe in the equivalent location. This hypothesis was proved incorrect when it was demonstrated by EPR40 and stopped-flow41−44 that Compound I in ascorbate peroxidase (APX), which has >30% sequence identity to CcP, is structurally very similar (Figure 1A) and contains the same Trp residue (Trp179) in an almost identical structural environment, uses a porphyrin π-cation radical instead of Trp179 (Figure 1B,C). The porphyrin π-cation radical in 428
DOI: 10.1021/acs.accounts.7b00463 Acc. Chem. Res. 2018, 51, 427−435
Article
Accounts of Chemical Research Compound I of APX is particularly unstable45 compared to other peroxidases, and the reasons for that are as yet unclear. On the whole, the factors controlling stability of Compounds I are incompletely understood (see, for example, ref 46). The involvement of a porphyrin π-cation radical in APX was to some extent rationalized when the substrate (ascorbate) binding site was found to involve hydrogen bonds to the heme propionates, a location referred to as the γ-heme edge, Figure 1D.47 This location for the ascorbate, Figure 1E, is different from other peroxidases, which bind their substrates close to the δ-heme edge instead (Figure 2, reviewed in ref 48). It is also
Figure 3. (A) Neutron structure of Compound I in CcP (PDB 4CVJ).73 Active site residues in the distal and proximal pockets are indicated; the neutron structure identifies hydrogen atoms on each residue (shown in magenta) and exchangeable hydrogen atoms (deuterium in the structure, shown in white). The ferryl oxygen (red sphere) is not protonated but instead is hydrogen bonded (dotted lines) to the Nε of Trp51 and to the Nε of Arg48, both of which are deuterated. His52 is protonated on the Nε, and there is a hydrogen bond to a distal water molecule (W2). Trp191 is deuterated, as predicted by spectroscopy.104 A protonated distal histidine residue has sometimes been implicated in HRP105,106 to account for pHdependent spectroscopic properties but has been assumed to hydrogen bond directly to an unprotonated ferryl species, which is not consistent with the hydrogen bonding pattern observed in CcP here, as the Nε is hydrogen bonded to an active site water molecule instead (W2). (B) Neutron structure of Compound II in APX (PDB 5JPR).75 Nuclear scattering density is shown in cyan. Notice the breaks in the density at the methylene (CH2) positions, which are caused by cancellation from the negative scattering lengths of 1H nuclei; the positions of these atoms were refined by also including X-ray data. The neutron Fo − Fc difference density calculated by omitting the distal ligand is shown in black. The O atom of the OD is at 1.88 Å from the iron. Hydrogen atoms are shown in green and deuterium atoms in white. Note that His42 is deuterated in this structure, as is His52 in Compound I of CcP shown in panel A.
Figure 2. A comparison of the binding locations of different substrates in different peroxidases. (A) In APX, the ascorbate binds at the γ-heme edge,47 which is the same location as is used in the manganese peroxidase/Mn2+ complex.102 (B) Most other peroxidases bind small aromatic molecules at the δ-heme edge, as exemplified by the structure of horseradish peroxidase in complex with benzhydroxamic acid (BHA).103
different from CcP, which binds its substrate (cytochrome c) on the end of an electron pathway that includes Trp191, Figure 1E. Electron delivery from the substrate to the heme in Compounds I/II in APX is presumably directly through the heme propionates (Figure 1E), which would explain the fact that Trp179 in APX is not essential for activity49 but Trp191 in CcP is.50 But there are other factors, aside from substrate binding, that need to be taken into account when considering the balance of Trp radical versus porphyrin π-cation radical in CcP and APX.51,52 The presence of positively charged metal ions in the proximal pocket within ≈10 Å of the heme in APX but not in CcP is certainly influential.53 But disruption of the hydrogen bonding networks (shown in Figure 3A) to the proximal Trp evidently is not, as methylation of the Nδ of the proximal histidine in APX does not affect formation of the porphyrin π-cation radical.54
indirect reporters of protonation state (because FeIVO is expected to have a shorter and stronger bond than FeIV−OH). Most, but not all, of these studies showed a short Fe−O distance (reviewed in ref 55) and were interpreted as being consistent with an FeIVO ferryl. But there are difficult experimental challenges associated with the preparation of pure, concentrated amounts of Compound I or Compound II in some of these enzymes, especially when considering the often rapid (millisecond) time scales for the formation or decay of these species, and this left a question mark over the reliability of some of the interpretations. An added problem was that laserinduced photoreduction complicated the interpretation of resonance Raman spectra, so that Compound I spectra in some cases appeared as Compound II or other reduced forms of heme (reviewed in ref 56). From about 2000 onward, X-ray crystallography began to be used alongside spectroscopic approaches (reviewed in refs 53 and 57). The idea was to measure bond lengths directly, and thus to cleanly differentiate between FeIVO and FeIV−OH. Would that it were so simple. What was discovered was that the Fe−O bond lengths in the crystal structures were longer and did not agree with most of the earlier spectroscopy, and this was interpreted as evidence of FeIV−OH species. These crystal structures of Compounds I and II would surely have carried the day had it not later been discovered that they were affected by photoreduction of the heme iron in the X-ray beam and were
4. THE TECHNICAL APPROACH AND SOME DIFFICULTIES There has been particular focus over a number of years as to the precise nature of the ferryl species and in particular whether the ferryl species is protonated (FeIV−OH) or not (FeIVO). This is of fundamental importance, as the chemical nature of the ferryl heme group controls its reactivity, and hence its usefulness, in biology. But a survey of the literature aimed at extracting an answer to this apparently simple question is not for the fainthearted, as there is at least three decades worth of sometimes contradictory data to untangle. The majority of studies in the 1980s and 1990s were focused on measurement of the Fe−O bond length (using EXAFS) or bond strength (using resonance Raman), which were taken as 429
DOI: 10.1021/acs.accounts.7b00463 Acc. Chem. Res. 2018, 51, 427−435
Article
Accounts of Chemical Research
Figure 4. A summary of the most recent information on Compound I and Compound II species in various heme systems. The nature of the ferryl species in each case is indicated.
room temperature was obtained by us as long ago as 2008 but was rejected for publication. Obtaining a neutron structure of Compound I of CcP was much more difficult and required cooling of large crystals during lengthy data collection, which at the time was only possible at BIODIFF in Munich. Using the neutron approach, we eventually identified the locations of all the hydrogen atoms in CcP.73 This established that the ferryl heme of Compound I in CcP is an FeIVO species and is not protonated (Figure 3A).
therefore invalidated. (X-ray induced reduction of metal ions is also a potential concern for X-ray absorption spectroscopy studies on heme systems.58) The use of multicrystal analyses59−62 minimizes but does not eliminate photoreduction of the FeIV species, although the new generation of X-ray free electron lasers are capable of data collection on time scales that are faster than X-ray induced photoreduction (as recently applied to Compound I of CcP63). A clear picture of the precise chemical nature of the ferryl did not, therefore, emerge because of the inconsistencies between the early spectroscopy and the later X-ray data, and because studies across different heme proteins, mainly peroxidases but also myoglobin and P450, gave different answers.
6. COMPOUND II IN PEROXIDASES In the absence of substrate, Compound I in all peroxidases decays spontaneously to Compound II. The source of the electron for this reduction is not clear in most cases but probably comes from one or more aromatic residues in the protein. As for Compound I above, spectroscopic data on Compound II of HRP (and also for myoglobin (Mb) and catalase) did not agree with crystallographic information on the same proteins.55 A particular problem is the preparation of pure samples of Compound II that are not contaminated by Compound I. There are reported X-ray crystal structures for Compound II in HRP, Mb, CcP, APX,59 and catalase,53,57 but the crystallographic interpretations are based on small (
