Identification of Trp106 as the tryptophanyl radical intermediate in Synechocystis PCC6803 catalase-peroxidase by multifrequency Electron Paramagnetic Resonance spectroscopy

JOURNAL OF

Inorganic Biochemistry Journal of Inorganic Biochemistry 100 (2006) 1091–1099 www.elsevier.com/locate/jinorgbio

Identification of Trp106 as the tryptophanyl radical intermediate in Synechocystis PCC6803 catalase-peroxidase by multifrequency Electron Paramagnetic Resonance spectroscopy Christa Jakopitsch b, Christian Obinger b, Sun Un a, Anabella Ivancich a

a,*

Service de Bioe´nerge´tique, URA 2096 CNRS, De´partement de Biologie Joliot-Curie, CEA Saclay, Bat. 532, 91191 Gif-sur-Yvette, France b Department of Chemistry, BOKU-University of Natural Resources and Applied Life Sciences, A-1190 Vienna, Austria Received 11 November 2005; received in revised form 14 February 2006; accepted 14 February 2006 Available online 6 March 2006

Abstract The reactive intermediates formed in the catalase-peroxidase from Synechocystis PCC6803 upon reaction with peroxyacetic acid, and in the absence of peroxidase substrates, are the oxoferryl-porphyrin radical and two subsequent protein-based radicals that we have previously assigned to a tyrosyl (Tyr) and tryptophanyl (Trp) radicals by using multifrequency Electron Paramagnetic Resonance (EPR) spectroscopy combined with deuterium labeling and site-directed mutagenesis. In this work, we have further investigated the Trp in order to identify the site for the tryptophanyl radical formation, among the 26 Trp residues of the enzyme and to possibly understand the protein constraints that determine the selective formation of this radical. Based on our previous findings about the absence of the Trp intermediate in four of the Synechocystis catalase-peroxidase variants on the heme distal side (W122F, W106A, H123Q, and R119A) we constructed new variants on Trp122 and Trp106 positions. Trp122 is very close to the iron on the heme distal side while Trp106 belongs to a short stretch (11 amino acid residues on the enzyme surface) that is highly conserved in catalase-peroxidases. We have used EPR spectroscopy to characterize the changes on the heme microenvironment induced by these mutations as well as the chemical nature of the radicals formed in each variant. Our findings identify Trp106 as the tryptophanyl radical site in Synechocystis catalase-peroxidase. The W122H and W106Y variants were specially designed to mimic the hydrogen-bond interactions of the naturally occurring Trp residues. These variants clearly demonstrated the important role of the extensive hydrogen-bonding network of the heme distal side, in the formation of the tryptophanyl radical. Moreover, the fact that W106Y is the only Synechocystis catalase-peroxidase variant of the distal heme side that recovers a catalase activity comparable to the WT enzyme, strongly indicates that the integrity of the extensive hydrogen-bonding network is also essential for the catalatic activity of the enzyme.  2006 Elsevier Inc. All rights reserved. Keywords: Protein-based radical; High-field EPR; Compound I; Electron transfer; Protein cofactors; Tryptophan radical; Multifrequency EPR spectroscopy; Catalase-peroxidase

1. Introduction The bi-functional peroxidases, also known as KatGs after the encoding gene, have an iron-protoporphyrin IX prosthetic group as catalytic site, the iron being pentacoordinated [1–4] and in the high-spin Fe(III) oxidation state

*

Corresponding author. Tel.: +33 1 69 08 28 42; fax: +33 1 69 08 87 17. E-mail address: [email protected] (A. Ivancich).

0162-0134/$ - see front matter  2006 Elsevier Inc. All rights reserved. doi:10.1016/j.jinorgbio.2006.02.009

for the resting enzyme [5–9]. The heme-iron active site is able to carry out the one-electron oxidation of selected substrates in a similar way than mono-functional peroxidases. The further ability to disproportionate hydrogen peroxide as efficiently as catalases is at the origin of the other name by which these enzymes are known, that is catalase-peroxidases. Thus, it is evident that these bi-functional enzymes have specific structural–functional features that were possibly lost in the evolution to the more specialized mono-functional peroxidases [10]. Such special features that enable

1092

C. Jakopitsch et al. / Journal of Inorganic Biochemistry 100 (2006) 1091–1099

KatGs to undergo a catalase-type reaction are the focus of intense studies by different groups working on the Synechocystis PCC6803, Mycobacterium tuberculosis and Burkholderia pseudomallei enzymes. The intriguing complexity of these bi-functional peroxidases is reinforced by the fact that low levels of NADH oxidase, INH lyase and isonicotinoyl-NAD synthase activities have been also reported for the B. pseudomallei ferric enzyme [11]. Three-dimensional crystallographic structures of the Haloarcula marismortui [1], B. pseudomallei [2], M. tuberculosis [4] and Synechococcus PCC7942 [3] enzymes have became available only recently, despite the long term efforts in this direction. The structural data showed that the enzyme is organized as a dimer, each monomer containing two sequence-related domains but, only the N-terminal domain binds the heme. Although the overall structure of the N-terminal domain has a similar topology [1] to that of cytochrome c peroxidase, the bi-functional enzyme has three long loops and a short stretch of eleven amino acid residues (103–113 in Synechocystis numbering). All four insertions are highly conserved among KatGs but non-existent in mono-functional peroxidases, with the exception of the short stretch, that is partially present in cytochrome c peroxidase. Although the KatG active sites globally resembles that of mono-functional peroxidases, in the detail there are significant differences on the heme environment. Recent investigations based on structural aspects [2,3,12,13] as well as kinetic characterization of selected variants of the heme environment [14–17,19] have given the basis to better understand the contribution of such differences in the remarkable reactivity of KatGs, as compared to mono-functional peroxidases. For example, it has been shown that the very unusual adduct resulting from the crosslink of three residues of the heme distal side, namely Trp122, Tyr249 and Met275 (Synechocystis numbering) is crucial for the catalatic activity of KatGs. The 3D-crystal structure of BpKatG clearly showed that the main access channel and a putative binding site for substrate(s) are different from those of mono-functional peroxidases [2]. Another substantial difference resides on the hydrogen-bond network of the heme microenvironment. In mono-functional peroxidases, the well characterized hydrogen-bonding network of the heme distal side involves the catalytically active His and Arg residues and three water molecules (for a recent review, see [20]). As it will be demonstrated in this work, and confirming our previous proposal [9], KatGs have an extensive hydrogen-bonding network involving the distal-side Trp, His and Arg residues (Trp122, His123, Arg119 in Synechocystis numbering), the heme 6-propionate group, a highly conserved Trp residue belonging to the short stretch (Trp106 in the Synechocystis numbering) and seven structural waters. The oxidizing intermediate of peroxidases and catalases, the so-called Compound I, is formed upon reaction of the resting enzyme with a molecule of hydrogen peroxide. In peroxidases, this intermediate catalyzes the one-electron oxidation of various substrates while in

the case of catalases, it reacts with another molecule of hydrogen peroxide (for a review, see [21]). Compound I originates from the two-electron oxidation of the ferric enzyme by hydrogen peroxide that forms concomitantly the oxoferryl iron (Fe(IV)@O) and the porphyrin p-cation radical (por+). A tryptophanyl radical, formed on the proximal side tryptophan residue (Trp 191) was identified as the relevant intermediate [Fe(IV)@O Trp+] for the substrate oxidation in cytochrome c peroxidase [22]. A tryptophanyl radical, located close to the enzyme surface, has been identified as the oxidation site for veratryl alcohol in lignin peroxidase [23,24]. Moreover, the formation of an oxoferryl-tyrosyl radical intermediate [Fe(IV)@O Tyr] by means of an intramolecular electron transfer between the porphyrin cofactor and a tyrosine residue has been observed in catalase [25], peroxidases [26,27] and prosthaglandin H synthase [28–30]. Not surprisingly, protein-based radical intermediates have been detected in the bi-functional peroxidases as well. By using a combined approach of multifrequency (9–285 GHz) EPR spectroscopy, isotopic labeling and site-directed mutagenesis, we have shown that the Synechocystis KatG forms the oxoferryl-porphyrin radical intermediate as well as a tryptophanyl radical and a tyrosyl radical, as a result of intramolecular electron transfer in the absence of peroxidase substrate(s) [9]. We have also shown that, unlike the case of CcP, the Trp radical in Synechocystis KatG is not formed on the proximal Trp residue (Trp341) and also it is not an exchange-coupled species [9]. Similar studies, except for the use of conventional (9 GHz) EPR spectroscopy, led Magliozzo and coworkers to the conclusion that the M. tuberculosis enzyme forms only a tyrosyl radical [31], the radical site (Tyr 353) being a non-conserved Tyr residue among KatGs [32]. Our previous deuterium-labeling multifrequency (9– 285 GHz) EPR experiments on selected variants of the Synechocystis catalase-peroxidase showed that the tryptophanyl radical intermediate was not formed in any of the four distal-side single mutations, W122F, H123Q, R119A and W106A [9]. Accordingly, two related questions need to be addressed: which Trp residue is the radical site (Trp106 or Trp122) and why the mutations on His123 and Arg119 hindered the formation of the Trp radical. In this work, we have further investigated the Trp radical intermediate in Synechocystis catalase-peroxidase in order to identify the radical site and, at the same time, to provide evidence for our previous proposal about the essential role in the radical formation of the extensive hydrogen-bonding network. Specific mutations on Trp106 (W106Y) and Trp122 (W122H, W122Y) were constructed and characterized by multifrequency EPR spectroscopy and activity measurements. Three different aspects were investigated: (a) the changes induced on the heme microenvironment as monitored by the pH-dependent ferric EPR spectrum; (b) the type of radical formed; (c) the catalase and the peroxidase activities. Our findings not only show that Trp106 is the site for the formation of the tryptophanyl radical

C. Jakopitsch et al. / Journal of Inorganic Biochemistry 100 (2006) 1091–1099

intermediate in Synechocystis KatG but also demonstrated the crucial role of the extensive hydrogen-bonding network in the formation of this radical. Moreover, kinetic characterization of these mutant enzymes showed that the integrity of such an H-bonding network is one of the special features allowing bi-functional peroxidases to perform a catalase-type reaction. 2. Materials and methods 2.1. Materials The perdeuterated tryptophan (DL-Trp d8) and perdeuterated tyrosine (DL-4-hydroxyphenyl-d4-alanine-2,3,3-d3) were purchased from CDN isotopes. The isotopic purity was 98% in all cases. 2.2. Sample preparation 2.2.1. Mutagenesis Oligonucleotide site-directed mutagenesis was performed using polymerase chain reaction (PCR)-mediated introduction of silent mutations as described previously [33]. A pET-3a expression vector that contained the cloned catalase-peroxidase gene from the cyanobacterium Synechocystis PCC6803 [34] was used as the template for PCR. The deletion of the short stretch (residues 103–113 in Synechocystis numbering) was done via PCR with overlapping primers. All constructs were sequenced to verify DNA changes using thermal cycle sequencing. Expression and purification of non-deuterated Synechocystis wild-type KatG and variants were described previously [33,34]. 2.2.2. Isotope labeling Escherichia coli DL39W cells, auxotroph for tyrosine and tryptophan, were obtained from the E. coli Genetic Stock Center at Yale University. The E. coli cells were treated with the kDE3 Lysogenization Kit (Novagen) to obtain DL39W(DE3) cells and a pET-3a vector containing the sequences of wild-type or mutated katGs was transformed into those cells. To produce specifically labeled KatG on tyrosines and/or tryptophans, the E. coli DL39W(DE3) cells were grown in a defined M9 medium as previously described [35]. Cells were grown to an OD600 = 1 and expression was induced by isopropyl b-thiogalactoside (IPTG). At the time of induction hemin (40 mg/L) was added to the media. Twenty hours after induction the cells were harvested. Isolation and purification of the specifically deuterated KatGs was done as described previously for the non-deuterated samples. 2.3. EPR samples Native enzyme (wild-type and mutants) in 50 mM Tris(hydroxymethyl)aminomethane–maleate (TRIS–maleate) buffer was used for the pH titration experiments on

1093

the resting (ferric) state. Typically, the Compound I samples were prepared by mixing manually 1.0 mM ferric enzyme (50 mM TRIS–maleate buffer, pH 7.0) with a 10fold excess of buffered peroxyacetic acid solution (final pH 4.5), directly in the 4 mm-EPR tubes kept at 0 C. The reaction was done at 0 C, the mixing time being 10 s for the wild-type enzyme and 5 s for the variants. The reaction was stopped by rapid immersion of the EPR tube in liquid nitrogen. The peroxyacetic acid concentration and the mixing time used in these experiments were those providing the higher yield on the protein-based radical(s) EPR signal but did not have an influence on the type of radical formed. Accordingly, no differences in the 9 GHz EPR spectrum of the radical(s) were observed when using lower excess of peroxyacetic acid except for the lower yield of the radical signal, which scaled inversely with the conversion of the ferric iron signal. No detectable change in the relative contribution of the tyrosyl and tryptophanyl radical signals to the high-field EPR spectrum were observed for the wild-type enzyme in samples prepared with mixing times of 5 or 10 s, nor for different excess of peroxyacetic acid. 2.4. EPR spectroscopy Conventional 9-GHz EPR measurements were performed using a Bruker ER 300 spectrometer with a standard TE102 cavity equipped with a liquid helium cryostat (Oxford Instrument) and a microwave frequency counter (Hewlett–Packard 5350B). The home-built high-field EPR spectrometer (95–285 GHz) has been described elsewhere [36]. The absolute error in g-values was 1 · 104. The relative error in g-values between any two points of a given spectrum was 5 · 105. 2.5. Steady-state kinetics Catalase activity was determined polarographically in 50 mM phosphate buffer using a Clark-type electrode (YSI 5331 Oxygen Probe) inserted into a stirred water bath (YSI 5301B) at 30 C. All reactions were performed at 30 C and started by the addition of KatG. One unit of catalase is defined as the amount that decomposes 1 lmol of H2O2/min at pH 7 and 30 C. Peroxidase activity was monitored spectrophotometrically using 1 mM H2O2 and 5 mM guaiacol (e470 = 26.6 mM1 cm1) or 1 mM o-dianisidine (e460 = 11.3 mM1 cm1). One unit of peroxidase is defined as the amount that decomposes 1 lmol of electron donor/ min at pH 7 and 30 C. 3. Results We have previously shown that upon treatment of the ferric enzyme with peroxyacetic acid, three distinct organic radical signals were observed in the EPR spectrum of Synechocystis catalase-peroxidase [9]. By combining multifrequency EPR spectroscopy and deuterium labeling of

1094

C. Jakopitsch et al. / Journal of Inorganic Biochemistry 100 (2006) 1091–1099

tyrosine and tryptophan residues, such signals could be assigned to the oxoferryl-porphyrin radical [Fe(IV)@O por+] and two other protein-based radical intermediates formed by intramolecular electron transfer and identified as a Tyr and a Trp species [9]. In order to identify the tryprophanyl radical site and better understand the factors that determine the formation of such radical in Synechocystis catalase-peroxidase, we have constructed and characterized specific variants of the Trp residues at positions 106 and 122. Fig. 1 shows the comparison of the 9-GHz EPR spectra of the protein-based radical intermediates formed after reaction of the ferric Synechocystis catalase-peroxidase (wild-type (WT) and the Trp106 and Trp122 variants) with peroxyacetic acid. The experimental parameters used to record the spectra in Fig. 1 (60 K and 0.05 mW microwave power) are non-saturating conditions for the Tyr and Trp signals but also avoid the contribution of the exchange-coupled oxoferryl-porphyrin radical, due to relaxation properties of this species [9]. Deuterium-labeling experiments were required to determine whether the

Fig. 1. The 9 GHz EPR spectra of the protein-based radicals (Trp and/or Tyr) formed in wild-type and the Trp122 and Trp106 variant of Synechocystis catalase-peroxidase upon reaction with peroxyacetic acid. In each case, the EPR spectra of the perdeuterated-Trp samples (dotted trace) is superimposed to the non-deuterated case (solid trace). The peakto-trough width of the control (non-deuterated) and perdeuterated-Trp samples is shown in solid and dashed vertical lines, respectively (see Section 3). Spectra were recorded at 60 K, 0.05 mW microwave power, 0.5 G modulation amplitude, 100 kHz modulation frequency. Spectra were arbitrary scaled to facilitate comparisons.

Trp radical, observed in the WT enzyme, was also formed in the variants. We have previously demonstrated that in the WT enzyme (Fig. 1, top) both Tyr and Trp signals contribute to the 9-GHz EPR spectrum, the Trp signal (peak-to-trough width of 20 G) being broader than the Tyr signal (see Fig. 3 in [9]). We have also demonstrated that in the Trp-perdeuterated samples, the observed narrower EPR spectrum (peak-to-trough width of 14 G) is predominantly that of the tyrosyl radical (see Fig. 1 top, dotted trace) with the contribution of a narrower signal of the perdeuterated Trp in the central part of the spectrum. In the case of the W122H variant, the control EPR spectrum was virtually identical to the WT enzyme. Also, the Trp-perdeuterated W122H spectrum was narrower than the control sample (Fig. 1, bottom). These results indicate that, as in the case of the wild-type sample, the Trp species was formed in the W122H variant. Interestingly, when replacing Trp122 by a Phe (or a Tyr) residue(s) a narrower spectrum was observed with no change in the overall line width of the signal after Trpperdeuteration (Fig. 1). Taken together, these results showed that Trp122 is not the radical site and that the type of residue used to replace the naturally occurring Trp can affect the formation of the tryptophanyl radical, presumably via a long range interaction. In the case of the Trp106 variants (W106A and W106Y) the EPR spectra were narrower than the WT spectrum and, no spectral differences were observed between the control (Fig. 1, solid trace) and the Trp-perdeuterated (Fig. 1, dotted trace) samples. Accordingly, these results showed that no Trp signal contributed to the EPR spectrum of the Trp106 variants. The high-field EPR spectra of the Trp106 variants (Fig. 2) further substantiate the assignment of these spectra to a Tyr-only signal. We have previously shown that the 285 GHz (10 T) EPR spectrum of the radicals formed in wild-type Synechocystis catalase-peroxidase (Fig. 2, inset) has the contribution of a tyrosyl and a tryptophanyl radical [9]. The 285-GHz EPR spectrum of the W122H variant (Fig. 2) readily showed the contribution of both radical signals as in the case of the wild-type spectrum (Fig. 2, dashed trace), although the intensity of the Trp appeared to be lower. A similar lower intensity of the Trp was previously observed for a Synechocystis variant on the heme proximal side [9] and was attributed to mild structural changes on the heme environment affecting the kinetics of the radical formation. In contrast, the 285 GHz EPR spectra of the Trp106 variants (represented by the W106Y in Fig. 2) as well as the W122F and W122Y variants (represented by the W122F spectrum in Fig. 2) well agreed with a Tyr-only signal (Fig. 2, dotted trace) as expected from the deuterium-labeling experiments. The observed differences in the gx component of the tyrosyl radical spectrum of the W122F variant as compared to that of the wild-type sample (Fig. 2, dashed trace) could be due to a change on the tyrosyl radical

C. Jakopitsch et al. / Journal of Inorganic Biochemistry 100 (2006) 1091–1099

. Tyr

Tyr

.

.

1095

Trp

.

wild type KatG

Trp

2.007

W122F

2.004 g

2.001

W106Y

W122H 10.15

10.16

10.17

10.18

10.19

Magnetic Field (T) Fig. 2. The high-field (10 T, 285 GHz) EPR spectra (plotted in the absorption mode) of the wild-type (dashed trace) and Trp122 and Trp106 variants of Synechocystis catalase-peroxidase radical intermediates. The spectrum of the tyrosyl radical (dotted trace) of the cytochrome c peroxidase W191G variant is plotted for comparisons. All spectra were recorded at 4 K and with 10 G field modulation. Each spectrum was recorded at slightly different microwave frequencies; for comparison, all the spectra were aligned to a nominal field and normalized so that the double integral was unity. Inset: The high-field (10 T, 285 GHz) EPR spectra (plotted in g scale) of the protein-based radical intermediates (Tyr and Trp) in wild-type Synechocystis catalase-peroxidase (KatG, solid trace) and the tyrosyl radical (dotted trace) in cytochrome c peroxidase W191G variant with g values of 2.0064(4), 2.0040(5) and 2.0020(8). The contribution of the more isotropic Trp signal (Dg = gz  gx  0.0012) in Synechocystis KatG is highlighted by an ellipse.

(electropositive) environment or to a different Tyr site [17]. 3.1. The effect of the heme distal-side mutations on the ferric EPR spectrum of Synechocystis catalase-peroxidase The 9 GHz EPR spectrum of the native (ferric) Synechocystis catalase-peroxidase and its pH-dependence are highly sensitive to changes on the heme microenvironment [9,17]. Accordingly, we have shown that the pH-dependence of the ferric EPR spectrum can be used to monitor changes on the heme microenvironment, possibly induced by the mutations [9]. The 9-GHz EPR spectrum of the wild-type enzyme (Fig. 3, top) showed the contribution of mainly two S = 5/2 Fe(III) high-spin species, one axial signal (labeled A) with effective g-values of gA^ = 5.93 and gAi = 1.99 and another rhombically distorted signal (labeled B) with gBx = 6.57, gBy = 5.10 and gBz = 1.97. The Trp variants studied in this work also showed the characteristic spectrum of ferric heme-iron enzymes (S = 5/2) in the high-spin state, with two main resonances at eff geff ?  6 and gk  2. Fig. 4 shows an expansion of the geff ?  6 region of the 9 GHz EPR spectra of the wild-type and Trp variants, recorded at 4 K and at two different pH values. We have previously shown [9] that in wild-type Synechocystis KatG, the degree of rhombicity of signal B is pH dependent, as it also is the case for the relative intensity of the two signals

(Fig. 4A). In the case of the Trp variants studied in this work, the Trp106 to Ala (W106A) and the Trp122 to Phe/Tyr mutations (W122F and W122Y) showed a predominant axial signal and no pH-induced changes on the EPR spectrum (Fig. 4C and D). These results are indicative of small but significant modifications on the heme microenvironment induced by these Trp variants, even though the position of Trp106 and Trp122 relative to the heme is quite different (see Fig. 5). A different situation was observed when Trp122 and Trp106 were replaced by residues that could be involved in H-bond interactions, such as Tyr or His. Specifically, the ferric EPR spectrum of the W106Y variant was indistinguishable from that of the WT enzyme, including the pH dependence (Fig. 4B). This result strongly suggests that the Tyr residue, introduced at position 106, restored the (H-bonding) interactions of the naturally occurring Trp106. In the case of the W122H variant, the EPR spectrum (Fig. 4E) showed a predominant axial signal that changed with pH (Fig. 4E). Interestingly, the effect of changing the pH directly affected the axial W122H signal while in the case of the wild-type enzyme, the pH change mostly affected the rhombically distorted component (signal B). Such a dissimilarity can be explained by the expected different positioning of the introduced His residue in relation to the neighboring residues and water molecule(s), or the different relative position of the His ring and the heme plane (see Fig. 5). Nevertheless, this pH

1096

C. Jakopitsch et al. / Journal of Inorganic Biochemistry 100 (2006) 1091–1099

Fig. 3. The 9 GHz EPR spectra of wild-type (top) Synechocystis catalaseperoxidase and the (D103  113) variant (bottom) both in the resting (ferric) state. An expansion of the g  2 region (highlighted by a dashed ellipse) of the wild-type enzyme spectrum (top) is shown in the inset. Spectra were recorded at 4 K, 1 mW microwave power, 3 G modulation amplitude, 100 kHz modulation frequency.

dependence indicates that the His (but not the Phe) could mimic to some extent the interactions of the naturally occurring Trp122. Moreover, the pH independent W122Y (axial) signal indicated that replacing Trp 122 by a Tyr did not result in restoring the integrity of the H-bonding interactions. This is not surprising if considering that the environment of Trp 122 is rather crowded in order to accommodate a Tyr residue pointing towards the iron, as suggested by the available 3D-crystal structures of the enzyme [1–4]. The highly conserved short stretch of 11 amino acid residues in catalase-peroxidases (Fig. 5, dotted spheres) includes the Trp106 in Synechocystis KatG. Based on the fact that the W106A variant had a remarkable effect on the environment of the heme distal side and that the short stretch makes two other possible long range contacts with the heme (on the heme plane and proximal side), we designed a mutation lacking residues 103–113 (D(103–113) variant). The crystal structure of H. marismortui KatG [1] shows that the short stretch is on the enzyme surface, rather isolated and protruding into a cavity. Only three residues of this short stretch (Trp106, Trp 110 and Asp 109, using Synechocystis numbering) have relatively close contacts with the heme, the closest being ˚ ). AccordAsp 109 (OAsp  Opropionate distance of 4.3 A

Fig. 4. Comparison of the ferric high-spin signals of the wild-type (A) and the Trp106 (B and C) and Trp122 (D and E) variants of Synechocystis catalase-peroxidase at two different pH values (6.0 and 8.0). Only the g  6 region of the spectra is shown. Experimental conditions are the same as those in Fig. 3.

Fig. 5. Structure of the heme environment of H. marismortui catalaseperoxidase (Protein data bank, accession pdb code 1ITK). The short stretch of 11 amino acid residues, containing the identified site (Trp 106) of the Trp intermediate in Synechocystis catalase-peroxidase and the positions of the Trp variants (Synechocystis numbering) characterized in this work are also shown.

ingly, and based on the previous single variants observations [9] a repositioning of distal-side residues and/or the water matrix induced by removing Asp 109, Trp 110

C. Jakopitsch et al. / Journal of Inorganic Biochemistry 100 (2006) 1091–1099

and Trp 106 would be expected. The single mutations on ˚) Trp 110 and Trp 106 (NTrp  Opropionate distance of 6.9 A showed a high-spin ferric EPR spectrum, with the W106A impairing the pH-dependence (Fig. 4B). At variance, the EPR spectrum of the D(103–113) variant (Fig. 3, bottom) with effective g values of g1 = 2.51, g2 = 2.28 and g3 = 1.86 well agreed with a low spin hexa-coordinated ferric heme (the absorption spectrum having a distinct Soret band at 420 nm). This EPR signal is reminiscent of those previously reported for other peroxidases [37,27] and cytochrome P450cam [38] and that were attributed to a water (or hydroxy) molecule acting as a sixth ligand to the heme. A low proportion (20%) of the same low-spin ferric signal contributed to the EPR spectrum of the heme distal site variants (see Fig. 1B in [9]). A rather low yield of radical signal was observed upon reaction of the D(103–113) variant with peroxyacetic acid, unlike the case of the single-mutants. This result indicates that the surface short stretch of catalase-peroxidases may have a crucial role in maintaining a defined geometry of the heme environment, that is required for the access of hydrogen peroxide (or peroxyacetic acid) to the heme distal side through the proposed channel in the B. pseudomallei crystal structure (see Fig. 5B in [2]). It is of note that the possibility of a misfolding of the enzyme induced by the deletion cannot be ruled out in the absence of the crystal structure of this variant. 3.2. Catalase and peroxidase activity measurements When replacing the naturally occurring Trp106 with an Ala residue the enzyme showed only 20% of the wild-type enzyme activity (kcat = 3500 s1) (the error on the measurements is 10%). In contrast, when introducing the Tyr residue (W106Y), the enzyme recovered high levels of catalase activity (75% of the WT enzyme). The situation was very different for the Trp 122 variants since the replacement of Trp 122 by a Tyr or His resulted in an enzyme with virtually no catalase activity, as in the case of the W122F variant (less than 1% of the WT enzyme). The peroxidase activity was determined with two different substrates, guaicol and o-dianisidine, both frequently used in activity tests of mono-functional peroxidases. The peroxidase activity of the W122H variant was three times higher that the WT enzyme for both substrates while that of the W106Y and W122Y was virtually the same as in the WT enzyme, within the experimental error. 4. Discussion 4.1. Integrity of the extensive H-bonding network of the heme distal side and catalase activity Our previous characterization of the Synechocystis catalase-peroxidase reactive intermediates formed upon reaction with peroxyacetic acid and in the absence of

1097

peroxidase substrates showed an unexpected, but very important indication about the purported site for the tryptophanyl radical formation. The Trp radical was not formed in any of the H123Q, R119A, W122F and W106A variants [9]. Based on a careful inspection of the two available 3D-crystal structures of KatGs at the time [1,2] we proposed that an extensive H-bonding network connecting Trp122, His123, Arg119, the heme 6-propionate and Trp106 through seven structural water molecules, was essential for the formation of the Trp radical. In this work and in order to provide evidence for our previous proposal, we designed new variants in which Trp122 and Trp106 were replaced by residues that are capable to make H-bond interactions, possibly mimicking the naturally occurring situation in the wild-type enzyme (see Fig. 5). The W106Y, W122H and W122Y variants were constructed and characterized by EPR spectroscopy. When replacing Trp106 by a Tyr the ferric EPR signal and its pH dependence was indistinguishable from that of the wild-type enzyme (see Fig. 4). Clearly, the Tyr residue could form an H-bond mimicking that of the naturally occurring Trp106. In contrast, the changes on the ferric EPR signal observed for the W106A variant were the same as those of the H123Q and R119A, although Trp106 is not close enough to the heme propionates to induce direct geometrical changes as in the case of His123 or Arg119. The distance between the indole nitrogen of Trp 106 and ˚ (or 14.8 A ˚ if considering the heme 6-propionate is 6.9 A the distance of the indole N to the heme iron). Taken together, these findings strongly indicate that the relative interactions connecting Trp106 with the other distal-side residues, the seven structural water molecules and the heme propionate (see Fig. 5) are conserved if Trp106 is replaced with a Tyr residue. In addition, the catalase activity lost in the W106A variant is restored by the W106Y variant. If replacing Trp122 with a His residue, the pH dependence of the ferric signal was recovered indicating that the naturally occurring interactions on the heme distal side were also restored to a significant extent, although the catalase activity was not recovered. The greater contribution of the axial EPR signal in the W122H variant (see Fig. 4E) could be rationalized by a different positioning of the imidazole ring and/or distance to the heme iron even though the overall long range connections on the heme distal side were restored. The fact that no catalase activity was observed for the W122H variant may be explained by the crucial role assigned to the crosslinked Trp residue in the catalatic activity of the enzyme (see [18] and references therein). Evidence for whether this is an indication of the absence of a crosslink between the introduced His and Tyr249 requires crystallographic studies. Our findings in Synechocystis catalase-peroxidase indicate that there are several related features that determine the catalase-like activity of these bi-functional enzymes. The proposed role of the covalent adduct (resulting from the

1098

C. Jakopitsch et al. / Journal of Inorganic Biochemistry 100 (2006) 1091–1099

crosslink between Trp122, Tyr249 and Met275) in relation to the catalase-like activity in KatGs [14–17] could be partially related to the integrity of the extensive H-bonding network. There could also be an influence of this heme distal-side network on the recently proposed mechanism in which the association of Arg to the crosslinked adduct modulates the heme reactivity [18]. Whether the extended H-bonding network is directly related to specific binding sites for hydrogen peroxide requires further studies, including the crystal structure. In this work, we have also demonstrated that the peroxidase activity towards guiacol and o-dianisidine is not related to the integrity of the extensive H-bonding network in the Synechocystis enzyme or to the formation of the Trp radical intermediate. Accordingly, the binding site(s) for these substrates may not be connected to or affected by this network. Implication of this observation is that the binding site for substrates in monofunctional peroxidases such as benzhydroxamic acid [43] may not be the same in catalase-peroxidases. In conclusion, we have demonstrated that there is a long range connection between Trp111, His123, Arg119, the heme 6-propionate and Trp106 through an extensive Hbonding network involving most possibly seven structural water molecules [1–4]. The integrity of this network proved to be essential for the enzyme catalatic activity. A similar crucial role of a H-bonding network in the oxygen activation mechanism has been reported for the human heme oxygenase [39]. In this case, it was shown that the distalside Asp140 is essential to activate iron-bound dioxygen and hydroperoxide [40] and that this catalytically essential aspartic acid residue is locked in place by a H-bonding network, involving Asn210, Arg136, Tyr58, Tyr114 and structural water molecules [41,42]. 4.2. The tryptophanyl radical site in Synechocystis catalase-peroxidase The changes on the 9 GHz EPR spectrum of the Trpdeuterated W122H as compared to the WT enzyme, clearly showed that the Trp radical was formed (see Fig. 1). The same situation was previously observed for the mutation on the proximal-side Trp (W341F) [9]. Accordingly, we can rule out both Trp122 and Trp341 as being the radical site. The only difference of the W122H and WT enzymes was the relative ratio of the Trp to Tyr signals, readily observed in the HF-EPR spectrum (see Fig. 2). In contrast, the Trp was not formed when Trp106 was replaced by an Ala or a Tyr residue. Even though Synechocystis catalase-peroxidase has 26 Trp residues, our studies showed that no other Trp is formed when Trp106 is removed. Taken together, these results demonstrate that Trp106 is the site for the formation of the Trp radical in this enzyme. It is tempting to suggest that the tryptophanyl radical intermediate in Synechocystis KatG may be the oxidizing species for a yet unidentified, but possibly well defined substrate, as in the case of cytochrome c peroxidase. Further work to

address this question is in progress. Tryptophanyl radicals having a role in enzyme catalysis are less well characterized than tyrosyl radicals, in particular the influence on the physico-chemical properties of the radicals exerted by the protein via electrostatic or hydrogen-bonding interactions. The best characterized example is the exchangecoupled Trp+ in cytochrome c peroxidase [22] for which it was shown that removing the Asp residue (D235A) that is in H-bonding distance to the radical site (Trp191), resulted in a reorientation of Trp191 and the non-formation of the radical [43]. A tryptophanyl radical intermediate in a very electropositive environment [44] has been proposed as the oxidation site for veratryl alcohol in lignin peroxidase [23,24]. The photogenerated tryptophan radical engineered in modified P. aeruginosa azurin showed g-values that well agreed with a neutral species in a polar environment [45], the crystal structure indicating that the tryptophan residue interacts with a Glu or a water molecule. Thus, the common feature in all these tryptophanyl radicals formed during enzyme turnover, is the presence of a H-bond interaction from the protein that appears to be essential for the radical formation. In the case of the tryptophanyl radical intermediate (Trp106) in Synechocytis catalase-peroxidase, we have demonstrated that the formation of the radical depends on the integrity of the long range connections on the heme distal side (Fig. 5). 5. Conclusion An oxoferryl-porphyrin cation radical [Fe(IV)@O por+] and two subsequent protein-based radical intermediates, Trp and Tyr, are formed in Synechocystis PCC6803 catalase-peroxidase, in the absence of peroxidase substrate(s) and resulting from intramolecular electron transfer. Trp 106 was identified as the site for the Trp formation. This apparently unique tryptophan radical site belongs to a KatG-specific short stretch on the enzyme surface (Fig. 5) that appears to maintain a defined geometry of the heme environment required for the access of peroxides. An important finding is that in the W122H variant (but not in W122Y) the Trp radical was formed, since not only allows us to ruled out Trp122 as a possible radical site but also to demonstrate the importance of the long range interconnection between Trp106, seven structural waters, the heme 6-propionate, Arg119, His123 and Trp122 (which is crosslinked to Tyr249 and Met275) for the radical formation. We have also demonstrated that the integrity of such an extensive network is crucial for the catalase-like reactivity of the Synechocystis bi-functional enzyme. Acknowledgements This work was supported in part by the Austrian Science Fund (FWF Project P15417) to C.J. and C.O. The COST Action D21 is acknowledged for supporting exchange

C. Jakopitsch et al. / Journal of Inorganic Biochemistry 100 (2006) 1091–1099

among the laboratories. A.I. thanks Prof. Peter Loewen (University of Canada) for stimulating discussions. References [1] Y. Yamada, T. Fujiwara, T. Sato, N. Igarashi, N. Tanaka, Nat. Struct. Biol. 9 (2002) 691–695. Accession pdb code: 1ITK. [2] X. Carpena, S. Lopraset, S. Mongkolsuk, J. Switala, P. Loewen, I. Fita, J. Mol. Biol. 327 (2003) 475–489. Accession pdb code: 1MWV. [3] K. Wada, T. Tada, Y. Nakamura, T. Kinoshita, M. Tamoi, S. Sigoeka, K. Nishimura, Acta Crystallogr. D 57 (2002) 157–159. Accession pdb code: 1UB2. [4] T. Bertrand, N.A.J. Eady, J.N. Jones, B. Jamart-Gre´goire, E.L. Raven, K.A. Brown, J. Biol. Chem. 279 (2004) 38991–38999. Accession pdb code: 1SJ2. [5] K. Johnsson, W.A. Froland, P.G. Schultz, J. Biol. Chem. 272 (1997) 2834–2840. [6] N.L. Wengenack, S. Todorovic, L. Yu, F. Rusnak, Biochemistry 37 (1998) 15825–15835. [7] G.S. Lukat-Rodgers, N.L. Wengenack, F. Rusnak, K.R. Rodgers, Biochemistry 39 (2000) 9984–9993. [8] H.A. Heering, C. Indiani, G. Regelsberger, C. Jakopitsch, C. Obinger, G. Smulevich, Biochemistry 41 (2002) 9237–9247. [9] A. Ivancich, C. Jakopistch, M. Auer, S. Un, C. Obinger, J. Am. Chem. Soc. 125 (2003) 14093–14102. [10] M.G. Klotz, P.C. Loewen, Mol. Biol. Evol. 20 (2003) 1098–1112. [11] R. Singh, B. Wiserman, T. Deemagam, L.J. Donald, H.W. Duckworth, X. Carpena, I. Fita, P.C. Loewen, J. Biol. Chem. 279 (2004) 43098–43106. [12] L. Mo, W. Zhang, J. Wang, X.H. Wenig, S. Chen, L.Y. Shao, M.Y. Pang, Z. Chen, Microbial Drug Resist. 10 (2004) 269–279. [13] T. Deegmagadan, X. Carpena, R. Singh, B. Wiserman, I. Fita, P.C. Loewen, J. Mol. Biol. 345 (2005) 21–28. [14] C. Jakopitsch, M. Auer, A. Ivancich, F. Ruker, P.G. Furtmuller, C. Obinger, J. Biol. Chem. 278 (2003) 20185–20191. [15] R.A. Ghiladi, K.F. Medzihradszky, F.M. Rusnak, P.R. Ortiz de Montellano, J. Am. Chem. Soc. 127 (2005) 13428–13442. [16] R.A. Ghiladi, K.F. Medzihradszky, P.R. Ortiz de Montellano, Biochemistry 44 (2005) 15093–15105. [17] C. Jakopitsch, A. Ivancich, F. Schmuckenschlager, A. Wanasinghe, G. Poltl, P.G. Furtmuller, F. Ruker, C. Obinger, J. Biol. Chem. 279 (2004) 46082–46095. [18] X. Carpena, B. Wiserman, T. Deemagan, R. Sungh, J. Switala, A. Ivancich, I. Fita, P.C. Loewen, EMBO Rep. 6 (2005) 1156–1162. [19] S. Yu, S. Girotto, X. Zhao, R.S. Magliozzo, J. Biol. Chem. 278 (2003) 44121–44127.

1099

[20] G. Smulevich, A. Feis, B.D. Howes, Acc. Chem. Res. 38 (2005) 433– 440. [21] B.H. Dunford, Heme Peroxidases, Wiley, New York, 1999. [22] M. Sivaraja, D.B. Goodin, M. Smith, B.M. Hoffman, Science 245 (1989) 738–740. [23] A.T. Smith, N. Veitch, Curr. Opin. Chem. Biol. 2 (1998) 269–278. [24] M.D.S. Sollewiyn Gelpke, J. Lee, M.H. Gold, Biochemistry 41 (2002) 3498–3506. [25] A. Ivancich, H.M. Jouve, B. Sartor, J. Gaillard, Biochemistry 36 (1997) 9356–9364. [26] A. Ivancich, P. Dorlet, D.B. Goodin, S. Un, J. Am. Chem. Soc. 123 (2001) 5050–5058. [27] A. Ivancich, G. Mazza, A. Desbois, Biochemistry 40 (2001) 6860–6866. [28] A. Tsai, G. Wu, G. Palmer, B. Bambai, J.A. Koehn, P.J. Marshall, R. Kulmacz, J. Biol. Chem. 274 (1999) 21695–21700. [29] W. Shi, C.W. Hoganson, M. Espe, C.J. Bender, G.T. Babcock, Biochemistry 39 (2000) 4112–4121. [30] P. Dorlet, S.A. Seibold, G.T. Babcock, G.J. Gerfen, W.L. Smith, A. Tsai, S. Un, Biochemistry 41 (2002) 6107–6114. [31] S. Chouchane, I. Lippai, R.S. Magliozzo, Biochemistry 39 (2000) 9975–9983. [32] S. Yu, S. Girotto, X. Zhao, R.S. Magliozzo, J. Biol. Chem. 278 (2003) 44121–44127. [33] G. Regelsberger, C. Jakopitsch, M. Engleder, F. Ru¨ker, G.A. Peschek, C. Obinger, Biochemistry 38 (1999) 10480–10488. [34] C. Jakopitsch, F. Ruker, G. Regelsberger, M. Dockal, G.A. Peschek, C. Obinger, Biol. Chem. 380 (1999) 1087–1096. [35] X. Hu, T.G. Spiro, Biochemistry 36 (1997) 15701–15712. [36] S. Un, P. Dorlet, A.W. Rutherford, Appl. Mag. Res. 21 (2001) 341– 361. [37] W.R. Patterson, T. Poulos, D.B. Goodin, Biochemistry 34 (1995) 4342–4345. [38] V. Schu¨neman, C. Jung, A.X. Trautwein, D. Mandon, R. Weiss, FEBS Lett. 179 (2000) 149–154. [39] H. Fuji, X. Zhang, T. Tomita, M. Ikeda-Saito, T. Yoshida, J. Am. Chem. Soc. 123 (2001) 6475–6484. [40] L. Lad, D.J. Schuller, H. Shimizu, J. Friedman, H. Li, P.R. Ortiz de Montellano, T.L. Poulos, J. Biol. Chem. 278 (2003) 7834–7843. [41] R. Davydov, T. Matsui, H. Fujii, T. Yoshida, M. Ikeda-Saito, B.M. Hoffman, J. Am. Chem. Soc. 123 (2003) 16208–16209. [42] A. Henriksen, D.J. Schuller, K. Meno, K.G. Welinder, A.T. Smith, M. Gajhede, Biochemistry 37 (1998) 8054–8060. [43] D.B. Goodin, D. McRee, Biochemistry 32 (1993) 3313–3324. [44] W. Blodig, W.A. Doyle, A.T. Smith, K. Winterhalter, T. Choinowski, K. Piontek, Biochemistry 37 (2002) 8832–8838. [45] J.E. Miller, C. Gradinaru, B.R. Crane, A. Di Biblio, W.A. Wehbi, S. Un, J.R. Winkler, H.B. Gray, J. Am. Chem. Soc. 125 (2003) 14220–14221.