Direct visualization of a Fe(IV)–OH intermediate in a heme enzyme

Hanna Kwon1, Jaswir Basran1, Cecilia M. Casadei1,2, Alistair J. Fielding4, Tobias E. Schrader5, Andreas Ostermann6, Matthew P. Blakeley2, Peter C. E. Moody1,§ and Emma L. Raven1§

1 Department of Biochemistry and Henry Wellcome Laboratories for Structural Biology, University of Leicester, Lancaster Road, Leicester, LE1 9HN, UK 2 Institut Laue-Langevin, 71 Avenue des Martyrs, 38000, Grenoble, France3 Department of Chemistry, University of Leicester, University Road, Leicester, LE1 9HN, UK4 The Photon Science Institute, The University of Manchester, Manchester, M13 9PL5 Jülich Centre for Neutron Science JCNS, Forschungszentrum Jülich GmbH, Outstation at MLZ, Lichtenbergstr.1, 85747 Garching, Germany6 Heinz Maier-Leibnitz Zentrum (MLZ), Technische Universitat München, Lichtenbergstr. 1, D-85748 Garching, Germany

§ To whom correspondence should be addressed: peter.moody@le.ac.uk; emma.raven@le.ac.uk.

Abstract. Catalytic heme enzymes carry out a wide range of oxidations in biology. They have incommon a mechanism that requires formation of highly oxidized ferryl intermediates. It is these ferryl intermediates that provide the catalytic engine to drive the biological activity. Unravelling the nature of the ferryl species is of fundamental and widespread importance. The essential question is whether the ferryl is best described as a Fe(IV)=O or a Fe(IV)–OH species, but previous spectroscopic and X-ray crystallographic studies have not been able to unambiguously differentiate between the two species. Here we use a different approach. We report a neutron crystal structure of the ferryl intermediate in Compound II of a heme peroxidase; the structure allows the protonation states of the ferryl heme to be directly observed. This, together with pre-steady state kinetic analyses, electron paramagnetic resonance spectroscopy and single crystal X-ray fluorescence, identifies a Fe(IV)–OH species as the reactive intermediate. The structure establishes a precedent for the formation of Fe(IV)–OH in a peroxidase.

The family of heme-containing peroxidase enzymes is widespread in biology. They catalyse the H2O2-dependent oxidation of a range of different substrates, and in doing so underpin a number of important biological processes in bacterial, yeast, plant, fungal and mammalian systems (1). The key to their catalytic power is the formation of two transient, oxidised heme intermediates. These intermediates – which are both FeIV (ferryl) species but differ in the oxidation state of the porphyrin ring – form sequentially during catalysis. Both intermediates were originally observed in horseradish peroxidase but were mistakenly interpreted as enzyme-substrate complexes: the first (green) intermediate was discovered by Theorell, and the second (red) intermediate by Keilin and Mann (2, 3). They were eventually given the names Compound I and Compound II, to differentiate them from the enzyme-substrate complex (4-6). These same two intermediates are used widely in numerous other O2-dependent catalytic heme enzymes, most notably the cytochrome P450s,

the nitric oxide synthases, the terminal oxidases, plus the heme dioxygenases. Many years have passed since the first observations, but establishing the nature of these transient ferryl species remains a fundamental question (7-11). A particular focus has been the protonation state of the ferryl group in Compounds I and II, and whether the ferryl group is best described as an Fe(IV)=O or a Fe(IV)-OH species. This is important because the protonation state controls the reactivity – and hence the biological usefulness – of each intermediate (7). But this conceptually simple question has proved fiendishly difficult to answer. Part of the problem is that spectroscopic approaches – which have mainly used EXAFS, resonance Raman, and Mossbauer – are only indirect reporters of the protonation state; none can directly visualise individual protons. The problem is compounded by the fact that photoreduction of the heme – which emerged relatively recently as a serious complication in X-ray crystallographic work on heme and other

1

metalloenzymes enzymes – affects the spectroscopic experiments as well (12, 13). We have argued (14) that the question needs to be approached in a completely different way. Using neutron diffraction, hydrogen and deuterium atoms are directly visible (15, 16), and photoreduction does not occur at all. Thus, if the considerable difficulties of a neutron crystallographic experiment on a reactive enzyme intermediate trapped at low temperature can be overcome, then the approach is potentially transformative because it can identify proton locations unambiguously. In this work, we have used the approach to directly visualize the positions of hydrogen and deuterium atoms in Compound II of a heme peroxidase. Ascorbate peroxidase (APX) catalyses the H2O2-dependent oxidation of ascorbate (17). APX has high sequence identity to cytochrome c peroxidase (CcP), which has served as a benchmark for heme enzyme catalysis over many years. But the experimental difficulty of isolation a Compound II is simplified considerably by working with APX because, in contrast to CcP, the Compound I and Compound II species are easily distinguished from one another by uv-visible spectroscopy. Compound I in APXs (APX-I) exists as a ferryl heme and a porphyrin π-cation radical (18-20), and is distinct from its Compound II (APX-II) which contains only a ferryl species. This is not the case in CcP, as CcP-I contains a ferryl heme and a tryptophan radical (21), which cannot be reliably differentiated from CcP-II (ferryl heme only) in uv-visible spectra. Single- and double-mixing stopped-flow experiments were used in the first instance to demonstrate formation of the requisite intermediates. Reaction of ferric APX (λmax = 407, 525, ≈630 nm (22)) with 10 eq of m-chloroperbenzoic acid (m-CPBA) initially yields a Compound I intermediate (APX-I) with wavelength maxima at 409, 527, 575sh, 649 nm, Fig. 1. This species is identical to that which is obtained on reaction of APX with 20 eq of H2O2

(Fig. S1A). APX-I is established (18, 23) as a ferryl species with a porphyrin π-cation radical. From stopped-flow pH-jump data, we find that the spectrum of this APX-I species is pH-independent in the pH range 4.5 – 11.5, Fig. S2A. The APX-I species formed on reaction with m-CPBA (Fig. 1) decays rapidly to a Compound II species (APX-II, λmax = 415, 528, 558 nm, Fig. 1) which is fully formed within 60 s and which is stable over long (500 s) timescales and eventually decays back to a ferric species. Exactly the same behaviour is observed when

APX-I is formed using H2O2 instead of m-CPBA (Fig. S1A). In the presence of higher concentrations H2O2, Compound II intermediates of peroxidases react with excess H2O2 to yield Compound III which is equivalent to a ferrous-oxy intermediate (Fig. S1B). Formation of Compound III is avoided on reaction with m-CPBA and was thus used in the crystallographic work below. The APX-II species as above is formed from the initial APX-I species (in the same way as has been used (24) to capture Compound II of cytochrome P450). We demonstrated that the spectrum of this APX-II species is identical to that of APX-II formed in a different way - either by reaction of ferrous APX with 10 eq of H2O2, to form APX-II directly, or by reaction of ferric APX in the presence of 1.5 eq of both ascorbate and H2O2 (Figs. S1C-E). As for APX-I, pH-jump experiments demonstrate that the spectrum of APX-II also does change in the pH range 4.5 – 11.5, Fig. S2B. In looking at all of the stopped flow data above and comparing with CcP, we note that the spectrum of CcP-I (ferryl/tryptophan radical) and CcP-II (ferryl) are not identical to those of APX-II, as would be expected in a simplistic interpretation where the spectrum of a “pure” ferryl (FeIV=O) species was always the same, because the ratio of the and -bands differs (Fig. S1F). The ratio of the /bands has been suggested to report on protonation state in P. mirabilis catalase, with the longer wavelength -band decreasing in intensity on formation of a Fe(IV)-OH species (25). Our stopped-flow spectra are consistent with this interpretation. Electron paramagnetic resonance (EPR) spectroscopy confirmed the observations from the stopped-flow experiments. The EPR spectrum at 7.5 K of ferric APX shows the expected mixture of high-spin (g = 6 and g‖ = 2) and low-spin (g1 = 2.69, g2 = 2.22, g3 = 1.79) resonances, Figure 2A. Formation of APX-I as the first intermediate on reaction of the ferric enzyme with 1 eq of m-CPBA (Figure SXB) or 1 eq of H2O2 (Figure SXC) is evident from the new axial resonance at g = 3.54 and the decrease in intensity of the high- and low-spin ferric signals in both spectra. Reaction of ferric APX with m-CPBA over longer timescales, Figure 2B, shows no new resonances (except for a very minor component from a radical signal (<5 %)); there is no evidence for a low-spin Fe(III)-OH species in these spectra. These changes are indicative of the formation of an EPR-silent species, APX-II. Similar spectra for APX-II are observed on reaction of ferric APX with H2O2, Figure 2D. Experiments with a greater excess of H2O2

2

showed no new resonances, consistent with the formation of EPR-silent, Compound III. The APX-I (Figures SXC,E) and APX-II (Figures S2C,E) species were stable stable over long periods, with no changes in the spectrum when stored at 77 K over 20 days. Our spectra are in close agreement with early EPR analyses on APX-I and APX-II (18). EPR spectroscopy therefore demonstrates that the APX-II species is EPR-silent, confirming the presence of a ferryl heme. There is no evidence for formation of a low-spin, paramagnetic Fe(III)-OH species. Having characterised the reactivity of APX in solution (as above), we then produced APX-II in crystallo by reacting crystals of D2O-exchanged ferric APX with m-CPBA (see Methods). Single crystal microspectrophotometry at 100 K showed absorption peaks for APX-II in the visible region (λmax = 530, 560 nm, Fig. 1, inset), in agreement with those in solution (Fig. 1). Parallel experiments with single crystals of ferric APX reacted with H2O2 gave the same absorption maxima, Fig. 1, inset. These maxima for APX-II are different from those obtained for single crystal spectra at 100 K for ferric APX (λmax = 540, 575 nm) Fig. S1G; we assign the spectrum of ferric APX at 100 K to formation of a low-spin Fe(III)-OH species, by analogy with known low-spin heme-hydroxide species (26). The neutron structure of APX-CII, prepared as above by reaction with m-CPBA, was solved at 100 K using data to 2.2 Å. Data and joint X-ray/neutron refinement statistics are shown in Table S1. Nuclear and electron densities for APX-II in the region of the heme are shown in Figs. 3A,B respectively. Nuclear and electron density maps for individual active site residues are shown in Fig S4. On the distal side of the heme, Trp41 is deuterated on Nε, and Arg38 is fully deuterated and is in orientated away from the ferryl heme (previously referred to as the “out” position (27)). The distal histidine residue, His42, is doubly protonated (on Nε and Nδ). An Fo –Fc neutron map calculated in the absence of a ligand on the distal side of the heme iron shows a peak (3.0 σ) with a positive difference density that was interpreted as arising from OD (Fig. 3C). The identity of this ligand was verified by calculations using an oxygen atom alone (i.e. as in Compound I (14)), which resulted in a positive difference peak at the D position (Fig S5). If the distal site was modelled as D2O, the other D atom refined to zero occupancy. No residual difference density was observed after refinement as OD. The measured Fe-O bond length is 1.88 Å. This is longer than would be expected for an unprotonated Fe(IV)=O double

bond species (28) and is distance is consistent with a single bonded Fe(IV)-OH species. The hydrogen bonding pattern in the active site is shown in Fig. S6. It has been previously established using multi-crystal X-ray methods (that minimize photoreduction), that the corresponding bond length in CcP-I is 1.63 Å (27), which is in close agreement with the distance observed in neutron diffraction studies on CcP-I (14). In this work, independent verification of the Fe-O bond length in APX-II using the same multi-crystal X-ray methods (see Methods) yielded a bond length of 1.87 Å (Fig. S6, which agrees with the neutron data above. The 1.87 Å bond length is also in close agreement with previous multi-crystal X-ray data for APX-II (1.84 Å (27)) formed by reaction with H2O2 instead of m-CPBA, which we had previously interpreted as being consistent with protonation of the ferryl heme. Putting aside for one moment the caveats that bond lengths in X-ray experiments are not determined precisely at this resolution, nor indeed that they report directly on protonation state, the salient point is that the Fe-O bond length measured here for APX-II is longer than that of CcP-I determined by both neutron diffraction (14) and by several reliable (i.e. where photoreduction has been minimised) X-ray determinations (27, 29, 30). Together, the spectroscopic and crystallographic data are all consistent with protonation of the ferryl heme in APX-II. 

DiscussionThe results presented here sit within a framework of other information that has recently emerged on in other P450 and peroxidase systems. Compound I of cytochrome P450 also exists as a ferryl species with a radical on the porphyrin or the thiolate ligand (or delocalised between the two), and the ferryl heme is unprotonated (31). The Compound I species in two closely related thiolate-ligated enzymes – aromatic peroxygenase (APO) and chloroperoxidase (CPO) – have also been assigned as unprotonated Fe(IV)=O/porphyrin π-cation radical intermediates (7, 32). There is no information yet on a histidine-ligated peroxidase with a Compound I carrying a porphyrin π-cation radical. The best information comes from Compound I of the somewhat atypical cytochrome c peroxidase (containing a Trp radical instead (21)), which also exists as an unprotonated ferryl heme (14). So, the most up-to-date information on the Compound I intermediates that have been examined so far are consistent with one another.

3

Information on Compound II has been slower to emerge, but the most recent information comes from the P450s. Compound II in the bacterial CYP-158 enzyme is protonated, with a pKa ≈ 12 (33). This pattern of unprotonated/protonated ferryl in Compound I/II in P450 is mirrored in both APO and CPO, as both APO-II and CPO-II are also protonated (34, 35). This means that there is a large “uplift” in the ferryl pKa from Compound I to Compound II, which leads to the protonation event. What causes this shift in pKa

of the ferryl heme in these thiolate-ligated systems is not clear, although bond length changes to the axial ligand (36), other hydrogen bond changes in the active site, and even distortions of the heme (which are well documented in other systems (37, 38)) may all play a role as part of a complex mixture of variables that, together, dictate the nature and reactivity of the ferryl heme. In the case of the peroxidases, our neutron data show that APX-II also contains a protonated ferryl group at pH 8, and we found no evidence from kinetic data that this species is pH-dependent within the range of stability of the enzyme (up to pH 11.5). It remains to be seen whether APX is typical of the peroxidase family, but there is evidence that the ferryl heme could be protonated under certain conditions in histidine-ligated catalytic enzymes (39, 40). There is an on-going debate on the mechanism of formation of Compound I and its subsequent reduction in the peroxidases. Computational analyses have played an important role: different mechanistic suggestions have been put forward, with some data favouring the involvement of active site water molecules (41) and some not (42). The data presented here for Compound II, together with that for Compound I (14), define the positions of hydrogen atoms much more precisely than has been previously possible. This is significant and lays a foundation that can be used to understand the complicated delivery mechanisms that feed protons to the O-O bond cleavage machine. This, in turn, will aid the design of artificial catalysts designed specifically for that purpose.

References and notes1. H. B. Dunford, Peroxidases and

Catalases: Biochemistry, Biophysics, Biotechnology and Physiology. (John Wiley, Chichester, ed. 2nd edition, 2010).

2. H. Theorell, Enzymologia 10, 250 (1941).

3. D. Keilin, T. Mann, On the haematin compound of peroxidase. Proc. Roy. Soc. (London) 122B, 119 (1937).

4. P. George, Chemical nature of the secondary hydrogen peroxide compound formed by cytochrome-c peroxidase and horseradish peroxidase. Nature 169, 612 (Apr 12, 1952).

5. D. Keilin, E. F. Hartree, Purification of horse-radish peroxidase and comparison of its properties with those of catalase and methaemoglobin. Biochem J 49, 88 (Jun, 1951).

6. P. George, The third intermediate compound of horseradish peroxidase and hydrogen peroxide. J Biol Chem 201, 427 (Mar, 1953).

7. J. T. Groves, Enzymatic C-H bond activation: Using push to get pull. Nature chemistry 6, 89 (Jan 23, 2014).

8. J. T. Groves, N. C. Boaz, Biochemistry. Fishing for peroxidase protons. Science 345, 142 (Jul 11, 2014).

9. S. G. Sligar, Glimpsing the Critical Intermediate in Cytochrome P450 Oxidations. Science 330, 924 (Nov 12, 2010).

10. K. D. Karlin, Bioinorganic chemistry: Model offers intermediate insight. Nature 463, 168 (Jan 14, 2010).

11. J. H. Dawson, Probing structure-function relations in heme-containing oxygenases and peroxidases. Science 240, 433 (Apr 22, 1988).

12. J. Terner et al., Resonance Raman spectroscopy of oxoiron(IV) porphyrin pi-cation radical and oxoiron(IV) hemes in peroxidase intermediates. J Inorg Biochem 100, 480 (Apr, 2006).

13. G. N. George et al., X-ray-induced photo-chemistry and X-ray absorption spectroscopy of biological samples. Journal of synchrotron radiation 19, 875 (Nov, 2012).

14. C. M. Casadei et al., Neutron cryo-crystallography captures the protonation state of ferryl heme in a peroxidase. Science 345, 193 (Jul 11, 2014).

15. M. P. Blakeley, Neutron macromolecular crystallography. Crystallogr Rev 15, 157 (2009).

16. V. L. Davidson, The many faces of a proton. Nature chemistry 3, 662 (Sep, 2011).

17. E. L. Raven, Understanding functional diversity and substrate specificity in haem peroxidases: what can we learn from

4

ascorbate peroxidase? Nat Prod Rep 20, 367 (Aug, 2003).

18. W. R. Patterson, T. L. Poulos, D. B. Goodin, Identification of a porphyrin pi cation radical in ascorbate peroxidase compound I. Biochemistry 34, 4342 (Apr 4, 1995).

19. L. Lad, M. Mewies, E. L. Raven, Substrate binding and catalytic mechanism in ascorbate peroxidase: evidence for two ascorbate binding sites. Biochemistry 41, 13774 (Nov 19, 2002).

20. M. Kvaratskhelia, C. Winkel, R. N. F. Thorneley, Purification and characterization of a novel class III peroxidase isoenzyme from tea leaves. Plant Physiol 114, 1237 (Aug, 1997).

21. M. Sivaraja, D. B. Goodin, M. Smith, B. M. Hoffman, Identification by ENDOR of Trp191 as the free-radical site in cytochrome c peroxidase compound ES. Science 245, 738 (Aug 18, 1989).

22. S. K. Badyal et al., Conformational mobility in the active site of a heme peroxidase. J Biol Chem 281, 24512 (Aug 25, 2006).

23. A. N. P. Hiner et al., Detection of a radical intermediate in the reaction of ascorbate peroxidase with hydrogen peroxide. European Journal of Biochemistry 268, 3091 (2001).

24. T. H. Yosca et al., Setting an upper limit on the myoglobin iron(IV)hydroxide pK(a): insight into axial ligand tuning in heme protein catalysis. J Am Chem Soc 136, 9124 (Jun 25, 2014).

25. O. Horner et al., Spectroscopic description of an unusual protonated ferryl species in the catalase from Proteus mirabilis and density functional theory calculations on related models. Consequences for the ferryl protonation state in catalase, peroxidase and chloroperoxidase. J Biol Inorg Chem 12, 509 (May, 2007).

26. M. Antonini, E. Brunori, Hemoglobin and Myoglobin and their Reactions with Ligands. A. Neuberger, E. L. Tatum, Eds., (North Holland Publishers, Amsterdam, 1971), pp. 10-12.

27. A. Gumiero, C. L. Metcalfe, A. R. Pearson, E. L. Raven, P. C. Moody, Nature of the Ferryl Heme in Compounds I and II. J. Biol. Chem. 286, 1260 (Jan 14, 2011).

28. R. K. Behan, M. T. Green, On the status of ferryl protonation. J Inorg Biochem 100, 448 (Apr, 2006).

29. Y. T. Meharenna, T. Doukov, H. Li, S. M. Soltis, T. L. Poulos, Crystallographic and single-crystal spectral analysis of the peroxidase ferryl intermediate. Biochemistry 49, 2984 (Apr 13, 2010).

30. G. Chreifi et al., Crystal structure of the pristine peroxidase ferryl center and its relevance to proton-coupled electron transfer. Proc Natl Acad Sci U S A 113, 1226 (Feb 2, 2016).

31. J. Rittle, M. T. Green, Cytochrome P450 compound I: capture, characterization, and C-H bond activation kinetics. Science 330, 933 (Nov 12, 2010).

32. X. Wang, S. Peter, M. Kinne, M. Hofrichter, J. T. Groves, Detection and kinetic characterization of a highly reactive heme-thiolate peroxygenase compound I. J Am Chem Soc 134, 12897 (Aug 8, 2012).

33. T. H. Yosca et al., Iron(IV)hydroxide pK(a) and the role of thiolate ligation in C-H bond activation by cytochrome P450. Science 342, 825 (Nov 15, 2013).

34. X. Wang, R. Ullrich, M. Hofrichter, J. T. Groves, Heme-thiolate ferryl of aromatic peroxygenase is basic and reactive. Proc Natl Acad Sci U S A 112, 3686 (Mar 24, 2015).

35. M. T. Green, J. H. Dawson, H. B. Gray, Oxoiron(IV) in chloroperoxidase Compound II is basic: implications for P450 chemistry. Science 304, 1653 (2004).

36. C. M. Krest et al., Significantly shorter Fe-S bond in cytochrome P450-I is consistent with greater reactivity relative to chloroperoxidase. Nature chemistry 7, 696 (Sep, 2015).

37. J. G. Kleingardner, K. L. Bren, Biological significance and applications of heme c proteins and peptides. Acc Chem Res 48, 1845 (Jul 21, 2015).

38. J. A. Shelnutt et al., Nonplanar porphyrins and their significance in proteins. Chem Soc Rev 27, 31 (1998).

39. N. C. Boaz, S. R. Bell, J. T. Groves, Ferryl protonation in oxoiron(IV) porphyrins and its role in oxygen transfer. J Am Chem Soc 137, 2875 (Mar 4, 2015).

40. Z. Ma, H. R. Williamson, V. L. Davidson, Roles of multiple-proton transfer pathways and proton-coupled electron transfer in the reactivity of the bis-FeIV state of MauG. Proc Natl Acad Sci U S A 112, 10896 (Sep 1, 2015).

41. P. Vidossich et al., On the role of water in peroxidase catalysis: a theoretical investigation of HRP compound I

5

formation. J. Phys. Chem. B 114, 5161 (Apr 22, 2010).

42. C. Zazza et al., Theoretical modeling of enzyme reactions: the thermodynamics of formation of compound 0 in horseradish peroxidase. J Phys Chem B 112, 3184 (Mar 13, 2008).

Supplementary materials Materials and methods. Table S1.Figs. S1-SX.

Acknowledgements. We thank Adrian Mulholland (University of Bristol) and Jeremy Harvey (KU Leuven) for helpful discussions. This work was supported by BBSRC (grant BB/K015656/1 to PM/ER), The Wellcome Trust (grant WT094104MA to PM/ER), Bruker (Sponsorship of A.J.F), an ILL studentship (to CC), and beam time at LADI-III and BIODIFF.

Fig. 1. Formation of Compound II. (A) Stopped flow UV-visible spectra obtained on reaction of ferric APX (solid line, at t = 0) with m-CPBA (10 eq), monitored over 100s, showing formation of Compound I (dot-dashed line) and then Compound II (dashed line). Absorbance values in the visible region have been multiplied by a factor of four. Conditions: 10 mM sodium phosphate, 150 mM KCl pH 7.0, 10.0 °C. Inset: Single crystal UV-visible spectra (100 K) of APX-II formed by reaction with m-CPBA (top spectrum) and H2O2 (bottom spectrum). Both spectra show the characteristic Compound II peaks (~ 530 and 560 nm) in the visible region. The top spectrum has been offset on the y-axis, for clarity.

Fig. 2. Analysis of APX-II by electron paramagnetic resonance. 9 GHz EPR spectra of a solution of (A) ferric APX, (B) APX-II prepared by reaction of ferric APX with 20-equivalents of m-CPBA and flash frozen after 40s, (C) the same sample as (B) but recorded after 20 days, (D) APX-II prepared by reaction of ferric APX with 20-equivalents of H2O2 and flash frozen after 40s, (E) the same sample as (D) but recorded after 20 days. Spectra were recorded at 7.5 K, 0.4 mT modulation amplitude, 1 mW power, 4 scans, 2048 points.

Fig. 3. Neutron structure of APX-II. (A) Nuclear scattering density is shown in cyan (contoured at 1.5 s). (B) Electron scattering density shown in magenta (contoured at 1.5 s). (C) The neutron difference density calculated in the absence of the ligand is shown in green (contoured at 3.0 s). The OD is positioned at 1.88 Å from the heme iron. Colour scheme: hydrogen - green; deuterium - white; carbon - yellow; oxygen - red; nitrogen - blue; iron - brown sphere.

6

Methods

Protein expression, purification.Soybean cytosolic ascorbate peroxidase (APX), in a pLEICS-03 vector carrying kanamycin resistance and a TEV cleavable N-terminal His tag, was expressed in BL21 (DE3). Cells were expressed in 2-YT media for 16 h at 37 °C without induction. Cells were harvested by centrifugation (4,000 x g at 277 K for 20 min) then purified as previously described (22, 43, 44). Stopped-flow kinetics. Pre-steady state stopped flow experiments were carried out using an Applied Photophysics SX.18MV stopped-flow spectrometer. All experiments were carried out at 10 °C unless otherwise stated, using 10 mM sodium phosphate buffer, 150 mM KCl pH 7.0 (for APX) or pH 6.5 (for CcP). Spectral deconvolution was performed by global analysis and numerical integration methods using Pro-Kineticist software (Applied Photophysics Ltd).Formation of Compounds I and II: Formation of Compound I and Compound II was followed in single mixing mode by mixing ferric enzyme (typically 2-3 M) with 5-20 equivalents of H2O2 or m-chloroperbenzoic acid (m-CPBA) and time dependent spectral changes monitored using a photodiode array detector. Data were fitted to a one-step model A → B, where A = Compound I and B = Compound II.Formation of Compound II from ferrous enzyme: Compound II was also generated directly, under anaerobic conditions, by reaction of peroxide with the ferrous (Fe2+) form of the enzyme. For these experiments the sample handling unit of the stopped-flow instrument was housed in an anaerobic glove box (Belle Technology Ltd., [O2] < 5 ppm) and was used in the single mix mode. Ferrous enzyme, produced by titration of ferric APX (5-10 M) with 2-5 equivalents of sodium dithionite, was mixed with 10 equivalents of H2O2. Time dependent spectral changes accompanying Compound II formation were followed and data analysis carried out as outlined above.pH jump experiments. In order to avoid enzyme instability problems (below pH 4.5 and above pH 11.5) the pH-jump method was used to investigate the pH-dependence of the spectra of APX-I and APX-II. Enzyme samples were prepared in water, adjusted to pH 7 with trace amounts of phosphate buffer (5 mM, pH 8.0). For Compound I formation, the enzyme was mixed with a stoichiometric amount of H2O2 which was made up in a buffer of twice the desired final concentration. The buffers used were citrate-phosphate in the pH range 4.0–6.0 (0.20 M), sodium carbonate-bicarbonate buffer in the range 8.0–10.5 (0.20 M) and sodium hydrogen phosphate (0.2 M) at pH 11.5. For Compound II formation, the sequential mix method was used: the enzyme was first mixed with H 2O2 (also prepared in water at pH 7.0 as outlined above), the reaction allowed to age for 80 s (for APX) to enable complete conversion to Compound II prior to a second mix with 0.20 M buffer. In all experiments, the pH of the solution was measured after mixing to ensure consistency.EPR Spectroscopy. All EPR experiments were carried out at the EPSRC National EPR Facility and Service. Continuous-wave EPR spectra were recorded at 9.4 GHz on a Bruker EMX spectrometer with a Super-high-Q rectangular cavity and an Oxford ESR-900 liquid helium cryostat. The operating conditions are stated on the Figure legends. APX crystals are not large enough for single crystal EPR experiments. Samples of APX in solution were prepared in 10 mM potassium phosphate buffer, 150 mM KCl, pH 7.0. APX-I and APX-II were prepared by manually mixing ferric APX (370 M) with an equivalent volume of m-CPBA or H2O2 directly in 4 mm quartz EPR tubes, followed by flash-freezing in liquid nitrogen.Crystallisation. Crystallisation of APX was a modification of previous procedures and gave larger crystals (some as large as 1 x 0.6 x 0.4 mm) than previously (typically 0.15 x 0.075 x 0.075 mm (45, 46)). Crystals were grown by vapour diffusion hanging drops made up of 2 μl protein (20 mg/ml in 10 mM potassium phosphate pH 7.0, 150 mM KCl) and 2 μl precipitant (2.25 M Li2SO4, 0.1 M HEPES pH 8.3 – 8.9). The drop was allowed to equilibrate with 700 μl of precipitant. The crystals appeared in 2 -14 days. Deuteration of APX was carried out by crystallising the protein with the mother liquor made up with D2O. Once the crystals were fully grown, the crystals were transferred and kept in the mother liquor with D2O until needed. Formation of Compound II in APX crystals for X-ray and neutron data collection was achieved by soaking the crystals in 3-chloroperbenzoic acid (m-CPBA, 0.2 mM) for 5 ~ 40 seconds at 4 °C and then flash freezing crystals in liquid nitrogen. Single crystal microspectrophotometry was carried out as previously described (14) on single crystals of APX that had been reacted with m-CPBA or H2O2 as above at 4 C followed by flash freezing at 77 K.X-ray data collection. The APX-II X-ray structure was solved by merging the first 10° of data from ten different crystals. The datasets were collected in-house at 100 K using CuKα radiation (λ = 1.5418

7

Å) from a Rigaku MicroMax 007HF generator. A 20 images of 0.5° oscillation with 1 s exposure per image were recorded on a Rigaku Saturn 944+ detector to a resolution of 1.8 Å from each crystal. Data were indexed using iMOSFLM (47) then scaled and merged using AIMLESS as part of the CCP4 suite (48). Atomic coordinates have been deposited in the Protein Data Bank.Neutron data collection. A large single crystal (0.7 x 0.5 x 0.4 mm) of APX was reacted with m-CPBA to form Compound II. The crystal was flash frozen in liquid nitrogen.  Preliminary tests at BIODIFF (TUM-MLZ Munich) were used to develop the experimental protocols . Quasi-Laue neutron diffraction data to 2.4 Å resolution were collected at 100 K on the LADI-III beamline at the Institut Laue-Langevin (ILL), Grenoble, France. In total ?? images were collected with an exposure time of 24 h per image. The neutron data were processed using the program LAUEGEN. Atomic coordinates have been deposited in the Protein Data Bank.Structure refinement. Phase determination, density modification and model building used PHENIX software (49). The crystal used for the data collections were isomorphous with previously published sAPX CII x-ray structure (PDB ID: 2XIF). All solvent molecules and ligand was removed from 2XIF then used for as the starting model. X-ray structure was solved and refined first then the joint x-ray and neutron refinement was carried out with PHENIX (50). H- and D- atoms were added with the program ReadySet (49) and D2O molecules were added based on neutron Fo – Fc map. The model building was completed with Crystallographic Object-Oriented Toolkit (COOT) software (51). Determination of the identity and the position of heme ligand were based on neutron data only. Data collection and refinement statistics are in Table S1. Single crystal spectrophotometry. Absorption spectra of single crystals of Compound II of APX, obtained as outlined above, were collected at 100 K using an Ocean Optics Maya 2000 PRO spectrometer, with an Ocean Optics DH-2000-BAL UV-VIS-NIR light source and a Humamatsu S10420 FFT-CCD back thinned detector with fibre optic coupled to 80 mm diameter 4X reflective lenses (Optique Peter, Lentilly, France) and mounted with a custom mount on Rigaku Raxis IV φ drive. The temperature was maintained at 100 K with an Oxford Cryosystems cryostream. Absorption spectra were acquired by means of the Ocean Optics SpectraSuite software.

References for methods

14. C. M. Casadei et al., Neutron cryo-crystallography captures the protonation state of ferryl heme in a peroxidase. Science 345, 193 (Jul 11, 2014).

22. S. K. Badyal et al., Conformational mobility in the active site of a heme peroxidase. J Biol Chem 281, 24512 (Aug 25, 2006).

43. D. K. Jones, D. A. Dalton, F. I. Rosell, E. L. Raven, Class I heme peroxidases: characterization of soybean ascorbate peroxidase. Arch Biochem Biophys 360, 173 (Dec 15, 1998).

44. I. K. Macdonald, S. K. Badyal, L. Ghamsari, P. C. Moody, E. L. Raven, Interaction of ascorbate peroxidase with substrates: a mechanistic and structural analysis. Biochemistry 45, 7808 (Jun 27, 2006).

45. K. H. Sharp, M. Mewies, P. C. Moody, E. L. Raven, Crystal structure of the ascorbate peroxidase-ascorbate complex. Nature structural biology 10, 303 (Apr, 2003).

46. K. H. Sharp, P. C. Moody, K. A. Brown, E. L. Raven, Crystal structure of the ascorbate peroxidase-salicylhydroxamic acid complex. Biochemistry 43, 8644 (Jul 13, 2004).

47. T. G. Battye, L. Kontogiannis, O. Johnson, H. R. Powell, A. G. Leslie, iMOSFLM: a new graphical interface for diffraction-image processing with MOSFLM. Acta Crystallogr D Biol Crystallogr 67, 271 (Apr, 2011).

48. M. D. Winn et al., Overview of the CCP4 suite and current developments. Acta Crystallographica Section D 67, 235 (2011).

49. P. D. Adams et al., PHENIX: a comprehensive Python-based system for macromolecular structure solution. Acta Crystallographica Section D 66, 213 (2010).

50. P. V. Afonine et al., Joint X-ray and neutron refinement with phenix.refine. Acta Crystallogr D Biol Crystallogr 66, 1153 (Nov, 2010).

51. P. Emsley, B. Lohkamp, W. G. Scott, K. Cowtan, Features and development of Coot. Acta Crystallographica Section D 66, 486 (2010).

8