Role of histidine 42 in ascorbate peroxidase

Kinetic analysis of the H42A and H42E variants


  • Enzymes: ascorbate peroxidase (EC; horseradish peroxidase (EC; cytochrome c peroxidase (EC

E. L. Raven, Department of Chemistry, Universityof Leicester, University Road, Leicester LE1 7RH, UK. Fax: + 44 (0)116 2523789, Tel.: + 44 (0)116 2522099, E-mail:


To examine the role of the distal His42 residue in the catalytic mechanism of pea cytosolic ascorbate peroxidase, two site-directed variants were prepared in which His42 was replaced with alanine (H42A) or glutamic acid (H42E). Electronic spectra of the ferric derivatives of H42A and H42E (pH 7.0, µ = 0.10 m, 25.0 °C) revealed wavelength maxima [λmax (nm): 397, 509, ≈ 540sh, 644 (H42A); 404, 516, ≈ 538sh, 639 (H42E)] consistent with a predominantly five-co-ordinate high-spin iron. The specific activity of H42E for oxidation of l-ascorbate (8.2 ± 0.3 U·mg−1) was ≈ 30-fold lower than that of the recombinant wild-type enzyme (rAPX); the H42A variant was essentially inactive but activity could be partially recovered by addition of exogenous imidazoles. The spectra of the Compound I intermediates of H42A [λmax (nm) = 403, 534, 575sh, 645] and H42E [λmax (nm) = 404, 530, 573sh, 654] were similar to those of rAPX. Pre-steady-state data for formation of Compound I for H42A and H42E were consistent with a mechanism involving accumulation of a transient enzyme intermediate (Kd) followed by conversion of this intermediate into Compound I (k1). Values for k1 and Kd were, respectively, 4.3 ± 0.2 s−1 and 30 ± 2.0 mm (H42A) and 28 ± 1.0 s−1 and 0.09 ± 0.01 mm (H42E). Photodiode array experiments for H42A revealed wavelength maxima for this intermediate at 401 nm, 522 nm and 643 nm, consistent with the formation of a transient [H42A–H2O2] species. Rate constants for Compound I formation for H42A were independent of pH, but for rAPX and H42E were pH-dependent [pKa = 4.9 ± 0.1 (rAPX) and pKa = 6.7 ± 0.2 (H42E)]. The results provide: (a) evidence that His42 is critical for Compound I formation in APX; (b) confirmation that titration of His42 controls Compound I formation and an assignment of the pKa for this group; (c) mechanistic and spectroscopic evidence for an intermediate before Compound I formation; (d) evidence that a glutamic acid residue at position 42 can act as the acid–base catalyst in ascorbate peroxidase.


ascorbate peroxidase


wild-type pea cytosolic APX


recombinant wild-type pea cytosolic APX


a variant of rAPX in which His42 has been replaced withalanine


a variant of rAPX in which His42 has been replaced with glutamic acid


cytochrome c peroxidase


horseradish peroxidase



The plant peroxidase superfamily has been classified [1] into three major categories: class I contains the enzymes of prokaryotic origin, class II contains the fungal enzymes (e.g. manganese peroxidase, lignin peroxidase) and class III contains the classical secretory peroxidases [e.g. horseradish peroxidase (HRP)]. The most notable member of the class I peroxidase subgroup is cytochrome c peroxidase (CcP), which was first identified in 1940 [2]. In spite of the fact that CcP has some rather unusual features, most notably the existence of a stable tryptophan radical during catalysis [3–6] and the utilization of a large macromolecular substrate (cytochrome c), it has been the subject of such intense mechanistic, structural and spectroscopic scrutiny that it has become the benchmark against which all other peroxidases are measured.

More recently, it has been possible to isolate and purify in good yields a second member of the class I peroxidase subgroup, ascorbate peroxidase (APX) [7,8]. Ascorbate-dependent peroxidase activity was first reported in 1979 [9,10] and the enzyme catalyses the reduction of potentially damaging H2O2 in plants and algae using ascorbate as a source of reducing equivalents [11,12]. APX was known from sequence comparisons [13] to contain the same active-site Trp residue (Trp179) as is used by CcP (Trp191) during catalysis. With high-resolution structural information available for the recombinant pea cytosolic enzyme (rAPX) [14] (Fig. 1), APX has provided a new opportunity to reassess the functional properties of CcP and to determine whether it is indeed representative of class I peroxidases. As detailed functional information has emerged, however, it seems that APX has several rather curious features of its own, and, in some ways, more questions have been raised than answered. (In fact, even the current classification of APX as a class I enzyme has been recently questioned [15].) For example, Trp179 in APX is not a necessary requirement for oxidation of ascorbate [16] and there is general agreement from kinetic [17–19] and EPR data [20] that the initial product (Compound I) of the reaction of APX with H2O2 is a porphyrin π-cation intermediate and not a protein-based trytophan radical. Equally intriguing is the existence of a potassium-binding site (not present in CcP), located ≈ 8 Å from the α-carbon of Trp179; the functional role of this site (if, indeed, there is one) is not yet fully understood.

Figure 1.

Active site of ascorbate peroxidase [14]. Hydrogen bonds (dotted lines) are indicated.

Although steady-state kinetic analyses have been a fairly prominent feature of most of the early literature on APX (reviewed in [12]), pre-steady-state kinetic data have been very much more limited, largely as a result of insufficient quantities of enzyme, and only preliminary mechanistic information is available [16–19,21,22]. The enzyme operates through a classical peroxidase mechanism in which the ferric enzyme is oxidized by two electrons to a so-called Compound I intermediate with concomitant release of one molecule of water, followed by two successive single-electron reductions of the intermediate by ascorbate (HS) to regenerate ferric enzyme.


Although Eqn (1) is commonly written as a single step, experimental [23–26] and theoretical [27–30] evidence suggests that this is an over-simplification. A more complex mechanism, as first suggested by Poulos and Kraut [31]– involving binding of neutral peroxide to the enzyme, concomitant proton transfer from the bound peroxide to the distal histidine residue, followed by O–O bond cleavage and release of H2O – has been suggested (Scheme 1). Of particular interest is the role of the distal histidine residue (His42 in APX), which has been proposed [31,32] to act as an acid–base catalyst, by accepting a proton from H2O2 and releasing it subsequently as water. Site-directed mutagenesis studies on CcP and HRP have provided evidence to support these predictions (reviewed in [33–37]). In the work reported here, we replaced the distal histidine residue of APX (Fig. 1) with alanine and glutamic acid (H42A and H42E variants, respectively). The aims of the work were severalfold. First, to establish a definitive role for His42 in Compound I formation by replacing it with a residue that is not capable of hydrogen bonding (H42A) and to examine whether other residues at this position are able to act as alternative acid–base catalysts (H42E). Secondly, to use these variants to provide information on the origin of the pH-dependent kinetic rate profile for Compound I formation [17]. Finally, as we anticipated that the replacement of His42 would probably generate variant enzymes that may well have altered kinetic properties, we sought to utilize these alterations in intimate mechanism to probe in more detail the formation of Compound I in APX. As such, we present the first spectroscopic evidence for the nature of the intermediate formed during the reaction of APX with H2O2.

                Scheme 1. .

Figure  Scheme 1. .

Proposed steps in the formation of Compound I. The mechanism depicts the neutral peroxide-bound and anionic peroxide-bound intermediates. The distal histidine residue that acts as the acid–base catalyst is indicated (B).

Materials and methods


l-Ascorbic acid (Aldrich Chemical Co.), guaiacol, imidazole, 1,2-dimethylimidazole (Sigma Chemical Co.) and the chemicals used for buffers (Fisher) were of the highest analytical grade (more than 99% purity) and used without further purification. H2O2 solutions were freshly prepared by dilution of a 30% (v/v) solution (BDH): exact concentrations were determined using the published absorption coefficient (ε240 = 39.4 m−1·cm−1[38]). Aqueous solutions were prepared using water purified through an Elgastat Option 2 water purifier, which itself was fed with deionized water. All pH measurements were made using a Russell pH-electrode attached to a digital pH-meter (Radiometer Copenhagen, model PHM 93).

Mutagenesis and protein purification

Site-directed mutagenesis was performed according to the Quikchange™ protocol (Stratagene Ltd, Cambridge, UK). Two complementary oligonucleotides encoding the desired mutation were synthesized and purified (PerkinElmer). For H42A, the primers were: 5′-CGTTTGGCATGGGCTTCTGCTGGTAC-3′ (forward primer) and 3′-GCAAACCGTACCCGAAGACGACCATG-5′ (reverse primer). For H42E the primers were: 5′-CGTTTGGCATGGGAATCTGCTGGTAC-3′ (forward primer) and 3′-GCAAACCGTACCCTTAGACGACCATG-5′ (reverse primer). To confirm the identity of the transformants, overnight cultures containing 100 µg·mL−1 ampicillin were incubated at 37 °C with vigorous shaking (250 r.p.m.). The plasmid DNA was isolated using the Hybaid mini-plasmid system and sequenced to confirm the desired mutation. Automated fluorescent sequencing, using New England Biolabs pUC and malE primers, was performed by the Protein and Nucleic Acid Chemistry Laboratory, University of Leicester, on an Applied Biosystems 373-Stretch machine, and sequence data were analysed using the program SeqED (Applied Biosystems). Individual mutations were confirmed by sequencing across the whole rAPX-coding gene.

Bacterial fermentation of cells and purification of rAPX were carried out according to published procedures [7]. Enzyme purity was assessed by examination of the ASoret/A280 value; in all cases an ASoret/A280 value > 1.9 for rAPX, H42A and H42E was considered pure. Enzyme purity was additionally assessed using SDS/PAGE, and the preparations were judged to be homogeneous by the observation of a single band on a Coomassie Blue-stained reducing SDS/polyacrylamide gel. Enzyme concentrations (pH 7.0, µ = 0.10 m, 25.0 °C) were determined using the pyridine haemochromagen method [39]: absorption coefficients were ε403 = 88 mm−1·cm−1 for rAPX [40], ε397 = 83 mm−1·cm−1 for H42A and ε404 = 95 mm−1·cm−1 for H42E.

UV/visible spectroscopy

Spectra were recorded using a variable-slit Perkin-Elmer Lambda 14 UV/visible spectrometer, linked to an Exacta 486D computer, and an Epsom-LQ-1060 printer. Temperature was controlled (± 0.1 °C) using a thermally jacketed cell holder connected to a circulating water bath (Julabo U3) and a water cooler (MK Refrigeration Limited), which was operated in tandem.

Mass spectrometry

Samples were analysed using a Micromass Quattro BQ (Tandem Quadrupole) electrospray mass spectrometer. Horse heart myoglobin (Sigma) was prepared as described for rAPX below, and used to calibrate the spectrometer in the range 600–1400 m/z. Protein samples were introduced into the instrument at a flow rate of 5 µL·min−1. Trace salt was removed using a Centricon-10 concentrator (Amicon) and successive centrifugation and dilution with highly purified water (Elgastat). Samples (≈ 2 mg·mL−1, 20 µL) were then diluted 10-fold with a solution of 50 : 50 (v/v) acetonitrile/water containing 0.1% acetic acid.

Steady-state measurements

Stock solutions of l-ascorbic acid, guaiacol, H2O2 and enzyme were prepared in sodium phosphate (µ = 0.10 m, pH 7.0, 25.0 °C). Enzyme assays were performed in a 1-mL quartz cuvette: various concentrations of substrate and 25 nm enzyme were preincubated for 3 min in buffer and the reaction was initiated by the addition of H2O2 (≈ 2.5 µL, ≈ 30 mm) to a final concentration of 0.1 mm. The wavelengths and absorption coefficients used for various substrates were as follows: l-ascorbic acid, ε290 = 2.8 mm−1·cm−1[41]; guaiacol, ε470 = 22.6 mm−1·cm−1[42]. Activities were determined by dividing the change in absorbance by the absorption coefficient of the substrate. Values for kcat were calculated by dividing the maximum rate of activity (µm−1·s−1) by the micromolar concentration of enzyme; values for Km were determined by a fit of the data to the Michaelis–Menten equation using a nonlinear regression analysis program (Grafit32 version 3.09b; Erithacus Software Ltd). All reported values are the meanof three independent assays. Errors on kcat and Km are estimated to ± 5% and ± 10%, respectively. For pH-dependent assays, a mixed sulfonic acid buffer system (µ = 95–110 mm depending on the exact pH) that buffered over the entire pH range was used; reactions were initiated by the addition of H2O2 (to 0.10 mm). In these cases, [l-ascorbic acid] = 0.70 mm (rAPX) and 0.50 mm (H42E), [guaiacol] = 30 mm (rAPX) and 11 mm (H42E), and [enzyme] = 25 nm. Specific activities ([enzyme] = 25 nm, sodium phosphate, pH 7.0, µ = 0.10 m, 25.0 °C) were calculated from initial slopes of activity measurements; 1 unit of activity is defined as the amount of enzyme that oxidizes 1 µmol substrate per minute (µmol·min−1·mg−1).

Transient-state kinetics

Transient-state kinetics were performed using a SX.18 MV microvolume stopped-flow spectrophotometer (Applied Photophysics) fitted with a Neslab RTE200 circulating water bath (± 0.1 °C). Reported values of kobs are an average of at least three measurements. All curve fitting was performed using the Grafit software package. All data were analysed using nonlinear least-squares regression analysis on an Archimedes 410–1 microcomputer using Spectrakinetics software (Applied Photophysics). Pseudo-first-order rate constants for the formation of Compound I (k1,obs) were monitored at 403 nm (rAPX), 404 nm (H42E) and 397 nm (H42A), in single mixing mode by mixing enzyme (0.5–1.0 µm) with various concentrations of H2O2. Absorbance changes were independent of [H2O2]; observed changes in absorbance were 97–99% of the calculated values. The pH-jump method was used to examine the pH-dependence of Compound I formation, to avoidenzyme instability problems below pH 5 and above pH 8.5. Enzyme samples were prepared in water, adjusted to pH 7 with trace amounts of phosphate buffer (5 mm, pH 8.0); H2O2 solutions were made up in buffers of twice the final concentration. The buffers used were sodium phosphate in the pH range 5.5–8.5 (µ = 0.20 m), citrate-phosphate in the pH range 4.0–6.0 (µ = 0.20 m) and carbonate buffer in the range 8.0–9.0 (µ = 0.20 m). The pH of the solution was measured after mixing to ensure consistency. pH-dependent data were fitted to the Henderson–Hasselbach equation for a single-proton process (Eqn 4):


where A and B are the rate constants for Compound I formation at the extremes of acidic and basic pH, respectively, and k is the rate constant (either second-order, k1, for rAPX or limiting first-order, k1, for H42A/H42E) for Compound I formation. Formation of Compound I in the presence of exogenous imidazole was carried out using single-wavelength mode (397 nm), where one syringe contained H42A (1 µm) and the other H2O2 (0.5–35 mm) in the presence of either imidazole or 1,2-dimethylimidazole (20 mm) (relatively low concentrations of exogenous imidazole and a high buffer concentration were used to minimize the effect of fluctuating imidazole levels on the ionic strength and pH and to prevent binding of theimidazole to the haem). Time-dependent spectra of the various reactions were obtained by multiple-wavelength stopped-flow spectroscopy using a photodiode array detector and x-scan software (Applied Photophysics). Spectral deconvolution was performed by global analysis and numerical integration methods using prokin software (Applied Photophysics).


Mass spectrometry

The integrity of the variant proteins was examined to ensure that post-translational modification of the protein had not occurred. Analysis of H42A and H42E (data not shown) gave average masses for the apoproteins of 27126.9 ± 0.8 Da and 27185.1 ± 0.5 Da, respectively, in good agreement with the calculated masses of 27126.74 Da (H42A) and 27184.88 Da (H42E).

Electronic spectra

Electronic spectra of the ferric derivatives of H42A and H42E (pH 7.0, µ = 0.10 m, 25.0 °C) are shown in Fig. 2. Wavelength maxima (Table 1) for H42A and H42E were found to be slightly different from those for rAPX, but are consistent with a predominantly five-coordinate high-spin iron. In the presence of excess cyanide, spectra for H42A and H42E were consistent with the formation of a low-spin haem species, with absorption maxima [H42A–CN λmax (nm) (ε (mm−1·cm−1)) = 420 (102), 540, 574sh; H42E–CN λmax (nm) (ε (mm−1·cm−1)) = 420 (109), 540, 573sh] similar to those for the corresponding derivative of rAPX [rAPX–CN λmax (nm) (ε (mm−1·cm−1)) = 419 (104), 539, 572sh]. The spectra of both ferric H42A and H42E were, on the other hand, unaffected by the addition of either azide or fluoride, suggesting that these (weak field) ligands do not bind to the haem under these conditions.

Figure 2.

UV/visible spectra of ferric rAPX (solid line), H42A (dotted line) and H42E (dashed line). The region 450–700 nm has been multiplied by a factor of five. Sample conditions: sodium phosphate, pH 7.0, µ = 0.10 m, 25.0 °C.

Table 1.  Wavelength maxima (nm) and in parentheses absorption coefficients (mm−1·cm−1) for the ferric and Compound I derivatives of rAPX, H42A and H42E.
FeIII403(88), 506, 540sh, 636397(83), 509, ≈540, 644404(95), 516, 538sh, 639
Compound I404(59), 529, 583sh, 650403(62), 534, 575sh, 645404(66), 530, 573sh, 654

Steady-state kinetics

The specific activity of rAPX for oxidation of l-ascorbic acid (256 ± 6 U·mg−1) is comparable to published data (411 U·mg−1) for pAPX [8]. The specific activity of H42E (8.2 ± 0.3 U·mg−1) was ≈ 30-fold lower than that of rAPX. Under conditions identical with those used for rAPX and H42E, H42A exhibited no activity with l-ascorbic acid, although residual activity was detected (9.2 ± 0.3 × 10−2 U·mg−1) when a higher enzyme concentration was used ([H42A] = 200 nm).

Steady-state data (kcat, Km and the arithmetically calculated selectivity coefficient, kcat/Km) for oxidation of l-ascorbic acid and guaiacol by rAPX and H42E are shown in Table 2. (The oxidation of l-ascorbic acid by rAPX does not obey standard Michaelis kinetics [8,22,43] and, in this case, data were fitted to the Hill equation. Oxidation of guaiacol by rAPX obeys Michaelis kinetics and data were fitted to the Michaelis–Menten equation. The origin of the different concentration-dependencies for these two substrates is not known. Oxidation of both l-ascorbic acid and guaiacol by H42E was observed to obey Michaelis–Menten kinetics.) For both substrates, kcat values for H42E are ≈ 50-fold lower than for rAPX, with Km values largely unaffected (about threefold lower for H42E). The H42A variant was inactive with both l-ascorbic acid (above) and guaiacol. The dependence of the rate of substrate oxidation (µm·s−1) vs. pH yielded a pH optimum for rAPX at ≈ 7 for the oxidation of both l-ascorbic acid (pH optimum 7.0) and guaiacol (pH optimum 6.9) (data not shown) (the pH optimum for l-ascorbic acid is consistent with that reported previously for pAPX [8]). For H42E, the pH optimum is shifted by ≈ 1 pH unit for both l-ascorbic acid (pH optimum 8.1) and guaiacol (pH optimum 8.0).

Table 2.  Values for kcat, Km and kcat/Km for the oxidation of l-ascorbic acid and guaiacol by rAPX and H42E (sodium phosphate, pH 7.0, µ = 0.10 m).
Enzyme l-AscorbateGuaiacol
k cat (s−1) K m (mm) k cat/Km

k cat (s−1)
K m (mm) k cat/Km (mm−1·s−1)
rAPX248 ± 280.41 ± 0.0460566 ± 312.3 ± 0.95.4
H42E3.9 ± 0.20.17 ± 0.09 22.91.5 ± 0.1 2.9 ± 0.350.5

Pre-steady-state kinetics

Spectra of the transient Compound I intermediates of H42A and H42E, formed by reaction of the ferric derivatives with 10 equivalents of H2O2, were obtained by photodiode array experiments. These preliminary experiments showed that Compound I formation for both variants is much slower than for rAPX, reactions being complete in less than 1000 s for H42A and 100 s for H42E, compared with less than 300 ms for rAPX under identical conditions. Wavelength maxima and absorption coefficients for the Compound I intermediate of H42A and H42E were found to be similar to those of rAPX (Table 1) and to those previously published for wild-type pAPX (λmax = 404 nm [17]). For H42A and H42E, Compound I is surprisingly stable, for up to 30 s, but does not spontaneously convert into Compound II as is observed for rAPX. Instead, Compound I for both H42A and H42E slowly returns to a spectrum with a slightly red-shifted Soret band, with wavelength maxima at 405, 514, 550sh and 640 nm for H42A and 406, 516, 552sh and 638 nm for H42E.

The dependence of the observed rate constant, k1,obs (individual traces were monophasic in all cases), on the concentration of H2O2 for H42A and H42E (Fig. 3) exhibits hyperbolic behaviour. For rAPX (data not shown and [21,22]) and pAPX [17], a linear dependence on [H2O2] is observed; in this work, a second-order rate constant of (6.1 ± 0.1) × 107 m−1·s−1 was derived for reaction of rAPX with H2O2. Saturation behaviour of the kind exhibited by H42A and H42E is consistent with a mechanism involving a pre-equilibrium step which precedes Compound I formation (Eqns 5 and 6):

Figure 3.

Dependence of k1,obs on [H2O2] concentration for the reaction of H42A (A) and H42E (B) with H2O2 (sodium phosphate, pH 7.0 µ = 0.10 m, 5.0 °C, [H42A] = 0.5 µm, [H42A] = 0.5 µm). Data were fitted using a nonlinear least-squares fitting procedure to Eqn (7).


(E = H42A or H42E). This mechanism predicts a linear (first-order) dependence at low concentrations of peroxide and a zero-order dependence at high concentrations. An expression for k1,obs can be derived (Eqn 7):


where Kd is the dissociation constant of the bound complex in Eqn (5) (Kd = 1/Ka) and k1 is the limiting first-order rate constant at high peroxide concentrations. A fit of these data for H42A and H42E to Eqn (7) (Fig. 3) yields values for k1 and Kd of 4.3 ± 0.2 s−1 and 30 ± 2.0 mm, respectively (H42A) and 28 ± 1.0 s−1 and 0.09 ± 0.01 mm, respectively (H42E). These data pass through the origin, indicating that the second step of the reaction is irreversible. At low concentrations of peroxide, where a linear dependence is observed, it is possible to extract an approximate value for the second-order rate constant for reaction with H2O2[Eqn (5) where Kd is related to the microscopic second-order (ka) and first-order (kb) rate constants for this step (Kd = kb/ka)]: values for ka of 84 ± 6 m−1·s−1 and 1.1 ± 0.2 × 105 m−1·s−1 were obtained for H42A and H42E, respectively.

The mechanistic scheme implicated by the above data suggested the accumulation of a reaction intermediate, the conversion of which to product was rate-limiting at high peroxide concentrations. To examine the nature of this intermediate, photodiode array experiments were carried out for H42A (Fig. 4). Intermediate spectra were obtained from a spectrally deconvoluted model: A → B → C, where A corresponds to ferric H42A, B corresponds to the intermediate (tentatively assigned as the [H42A–H2O2] complex, vide infra) and C corresponds to Compound I. Wavelength maxima for the proposed intermediate were at 401 nm, 522 nm and 643 nm. The model yielded rate constants for each step: kA (A → B) and kB (B → C) of 147 s−1 and 1.19 s−1, respectively ([H2O2] = 35 mm), which is in approximate agreement with the limiting rate constant of 4.3 s−1 determined above. The maximum concentration of the intermediate under these conditions was 91% of the initial enzyme concentration. Under these experimental conditions and in contrast with rAPX, there was no clear isosbestic point for the conversion of ferric H42A into Compound I, indicating that there are more than two absorbing species in solution and therefore consistent with the formation of an intermediate. At 342 nm (Fig. 4, inset), the isosbestic point between ferric enzyme and the intermediate, the absorbance–time trace shows a lag phase before Compound I formation, providing further evidence to support the proposed intermediate; the filled circles in this inset are simulated data points derived from the model and show good agreement between the simulated and experimental data. For H42E, no intermediate complex could be detected under these conditions.

Figure 4.

Photodiode array spectra (sodium phosphate, pH 7.0, µ = 0.10 m, 5.0 °C) for the reaction between H42A (1 µm) and H2O2 (35 mm). The reaction was monitored over a time base of 2 s, and 100 spectra were recorded with a time interval of 20 ms between each scan. Experimental data were fitted to an A → B → C model, using prokin software. Deconvoluted spectra from this analysis are shown in the Soret (A) and visible (B) regions. The solid line shows the spectrum of ferric H42A; the dotted and dashed lines show the intermediate spectra derived from the model, and represent the spectra of the proposed [H42A–H2O2] complex and Compound I, respectively. (B) Inset: time course at 342 nm. (●) Simulated data points derived from the model calculated using the prokin software.

The pH-dependences of the observed rate constants for Compound I formation for rAPX, H42A and H42E enzymes are shown in Fig. 5. Nonlinear dependencies with zero intercepts on [H2O2] were observed at all pH values for H42A and H42E. The use of acetate and nitrate buffers has been avoided to prevent anionic effects previously observed in pH-dependent studies on HRP and CcP [44–46]. Compound I spectra for each enzyme were identical at all pHs (data not shown). Determination of second-order (k1) and limiting first-order (k1) rate constants for rAPX and H42E, respectively, revealed pH-dependent behaviour for Compound I formation in both cases (Fig. 5A,C). These data were fitted to a single-proton process (Eqn 4), and pKa values for rAPX and H42E of 4.9 ± 0.1 and 6.7 ± 0.2, respectively, were obtained. In contrast, H42A exhibits no detectable pH-dependence for the rate constant for Compound I formation (k1) from pH 5.5–8.5 (Fig. 5B).

Figure 5.

pH-dependence of Compound I formation for (A) rAPX, (B) H42A and (C) H42E. The solid line for rAPX and H42E represents a fit of the data to the Henderson–Hasselbach equation for a single proton process (Eqn 4); data for H42A were fitted to y = c. Conditions: µ = 0.10 m, 5.0 °C.

Figure 6.

Dependence of k1,obs, the pseudo-first-order rate constant for Compound I formation in H42A, on H2O2 concentration in the absence and presence of exogenous imidazoles (20 mm). Conditions: sodium phosphate, µ = 0.20 m, 5.0 °C, pH 7.0, [H42A] = 1 µm. (▵) H42A; (○) H42A + imidazole; (●) H42A + 1,2-dimethylimidazole.

Recovery of H42A activity

Steady-state and pre-steady-state analyses of H42A indicated that the enzyme was essentially inactive. Rate constants for Compound I formation for H42A in the presence of either imidazole or 1,2-dimethylimidazole (20 mm) were also determined (sodium phosphate, µ = 0.20 m, 5.0 °C, pH 7.0, [H42A] = 1 µm; Fig. 6). [Higher concentrations of imidazole were not experimentally accessible for two main reasons: (a) high concentrations of either imidazole led to large changes in both ionic strength and solution pH; (b) binding of imidazole to the iron is observed at higher concentrations, leading to enzyme inhibition.] The spectrum of Compound I formed under these conditions (λmax = 403, 534, 575sh, and 645 nm) was identical with that obtained in the absence of exogenous imidazoles. Formation of Compound I again showed saturation kinetics (Fig. 6), and a nonlinear least-squares fit of the data to Eqn (7) in the presenceofimidazole and 1,2-dimethylimidazole yields k1 = 40 ± 3 s−1 and 168 ± 30 s−1, respectively, and Kd = 25 ± 5 mm and 23 ± 8 mm, respectively. Comparison with the corresponding data obtained under similar experimental conditions but in the absence of exogenous ligand [k1 = 4.1 ± 0.1 s−1, Kd = 30 ± 2.0 mm (sodium phosphate, pH 7.0 µ = 0.20 m], indicates that imidazole and 1,2-dimethylimidazole lead to ≈ 10-fold and ≈ 40-fold increases in k1, respectively, with the values for Kd largely unaffected. [These values are slightly different from thosedetermined above (k1 = 4.3 ± 0.2 s−1 and Kd = 30 ± 2.0 mm), because the two determinations were carried out at a slightly different ionic strengths.]

In parallel steady-state experiments ([H42A] = 100 nm, [guaiacol] = 30 mm, [H2O2] = 0.10 mm, sodium phosphate, pH 7.0, µ = 0.20 m, 25.0 °C), the oxidative activity of H42A with guaiacol was observed to increase with increasing concentration of imidazole and to saturate at high concentration of imidazole (data not shown). (l-Ascorbic acid activity was not examined because both l-ascorbic acid and imidazole absorb strongly between 265 and 290 nm.) Separate UV/visible experiments (data not shown) provided independent evidence for binding of imidazole to the iron (Kd = 51 ± 6 mm for H42A; Kd = 0.3 ± 0.1 mm for rAPX) to generate a low-spin H42A–imidazole derivative (λmax = 412, 533 and 565sh nm) that inhibited activity. Guaiacol activities in the presence of 1,2-dimethylimidazole were ≈ 10-fold higher than for imidazole itself and were observed to increase linearly with increasing 1,2-dimethylimidazole concentration (1–40 mm); no evidence for saturation was observed in this case and addition of 1,2-dimethylimidazole to H42A did not generate an observable low-spin derivative.


To investigate the catalytic role of the conserved His42 residue in APX catalysis, two site-directed variants were prepared in which the distal histidine was replaced by alanine (H42A) and glutamic acid (H42E) residues. The results provide: (a) unambiguous evidence that His42 is critical for efficient Compound I formation in APX; (b) confirmation that titration of this residue controls the rate constant for Compound I formation and an assignment of the pKa for this group; (c) mechanistic evidence for an intermediate before Compound I formation; (d) spectroscopic evidence on the nature of this intermediate. The detailed implications of these data are discussed below.

Examination of the electronic spectra for the two variants indicates that large-scale structural alterations have not been induced as a direct consequence of the mutations. Hence, wavelength maxima for ferric H42A and H42E are shifted slightly compared with ferric rAPX (Table 1), but are consistent with a largely five-coordinate, high-spin haem geometry. As ferric rAPX has itself been found [47] to be a mixture of five and six-coordinate high-spin haem together with some six-coordinate low-spin haem (the relative proportions of which are themselves likely to be affected by sample and storage conditions), it is probable that the slight differences in the exact wavelength maxima reflect differences in the spin state/coordination number distribution for each variant compared with rAPX. In particular, we note the broad Soret band, distinct shoulder (≈ 375 nm) and red-shifted CT1 band (644 nm) for H42A, all of which are consistent with increased five-coordinate character for this variant compared with both rAPX and H42E. However, it is not possible to make fully quantitative assessments from these data, and more detailed spectroscopic analyses were not attempted [although resonance Raman data (unpublished work) for rAPX and H42A at neutral pH indicate that the five and six-coordinate high-spin haem represent major and minor components, respectively].

The steady-state oxidation of l-ascorbate by pAPX [8] and rAPX [43] cannot be satisfactorily fitted to a standard Michaelis–Menten treatment, and sigmoidal kinetics are observed; steady-state oxidation of guaiacol by rAPX obeys Michaelis–Menten kinetics. Although it was initially proposed that the origin of the sigmoidal dependence may arise from the dimeric structure of the enzyme, site-directed mutagenesis work [43] has indicated that this is unlikely to be the case. It is possible that nonspecific effects, perhaps involving radical chemistry associated with oxidation of ascorbate, or the existence of more than one substrate binding site, may be influential. A sigmoidal dependence on substrate concentration is not observed for H42E-catalysed oxidation of l-ascorbate, an observation that we are not, at present, able to fully rationalize. Replacement of the distal His42 by alanine renders the H42A enzyme essentially inactive, which is presumably directly linked to the very poor reactivity of this variant with H2O2.

There is now general agreement that the Compound I species formed immediately after reaction of both rAPX and pAPX with H2O2 contains a porphyrin π-cation radical and not a protein-based radical as observed in CcP [17–20]. Spectra of Compound I obtained in this work for rAPX agree with those previously published for pAPX (λmax = 404 nm [17]) and those for HRP in which a porphyrin π-cation radical is known to be generated [37]. Spectra for the Compound I intermediates of H42A and H42E were similar to those of rAPX and are also consistent with a porphyrin π-cation formulation. The decay of Compound I of rAPX in the absence of substrate through the normal Compound II/ferric route is not straightforward and generates spectra that are similar to, but not identical with, authentic samples of either Compound II or ferric rAPX [48]. It has been proposed [48] that (noncatalytic) protein radical chemistry involving a Trp amino acid occurs under these conditions, leading to a permanent alteration of the haem structure that is reflected in the absorption maxima of the final (ferric-like) decay product. For H42A and H42E in the absence of substrate, Compound I does not spontaneously convert into a Compound II species asobserved for rAPX, and the Compound I intermediates for both enzymes slowly generate species with a slightly red-shifted Soret band [λmax (H42A) = 405 nm; λmax (H42E) = 406 nm] compared with the ferric state. These maxima are analogous to those observed for the decay of rAPX (λmax = 406 nm), suggesting that similar radical chemistry may occur in the variants, although this has not been examined in detail in this work. The failure to observe Compound II for the two variants under these conditions is intriguing (particularly as H42E is clearly active against both l-ascorbate and guaiacol) and has been noted previously for the H42E [49,50] and H42A and H42V [51] variants of HRP. These data suggest that the distal histidine residue stabilizes Compound II through a hydrogen-bonding structure involving the Nε of His42 and the ferryl oxygen. However, although hydrogen-bonding interactions of this kind have been structurally confirmed for the ferrous-oxy derivative of CcP [52], no such interaction was detected in the crystal structure of Compound I [53,54]. Instead, stabilizing hydrogen-bonding interactions from the distal arginine residue have been identified for Compound I [53,54]. Indeed, Compound II of the R38A variant of HRP was also not detected [25] and the Compounds I of the R48L, R48K [55] and R48E [56] variants of CcP are similarly unstable. Hence, it is possible that these hydrogen-bonding interactions involving Arg38 in rAPX have been simultaneously disrupted as a secondary consequence of the His42 mutations and, together with His42, may be influential in defining the stability of Compound II.

The replacement of the distal histidine residue by alanine or glutamic acid clearly has a profound influence on the ability of the rAPX enzyme to react with H2O2 and is clearly demonstrated by the nonlinear dependence of the observed rate constant on peroxide concentration. A comparison of rate constants for rAPX and variant proteins is only possible by comparing the second-order rate constant derived from the linear part of Fig. 3 with the second-order rate constant for formation of Compound I in rAPX. These values indicate that a ≈ 106-fold and ≈ 102-fold decrease has occurred for H42A and H42E, respectively. These data provide convincing evidence to support a key catalytic role for His42 and indicate that the glutamic acid residue is able to replace the distal histidine residue such that reasonably efficient Compound I formation is effected. The corresponding H42A and H42E variants in HRP have been shown to lower the rate constant for Compound I formation by six and four orders of magnitude, respectively [49–51,57], although no evidence for an intermediate was found for these variants. Large effects on the rate constant of Compound I formation have also been reported for other His42 variants of HRP [58–60] and for CcP [61,62].

The pre-steady-state data (Fig. 3) are consistent with a mechanism involving formation of an enzyme–substrate intermediate, the conversion of which into product is rate-limiting at high concentrations of peroxide. The simplest mechanism that is consistent with these data is described by Eqns (5) and (6). Although mechanisms involving conformational gating of the reaction at high peroxide concentrations [55] or before Compound I formation [63] are possible, the observation of a spectroscopically distinct intermediate for H42A provides good evidence for the proposed mechanism (but leaves this interpretation slightly more open for H42E). This enzyme intermediate has wavelength maxima at 401 nm, 522 nm and 643 nm and we assign the intermediate as arising from a transient [H42A–H2O2] species (see below). Our ability to detect this species for the first time arises from the lowering of the observed rate constant compared with rAPX as a consequence of the mutation, such that high concentrations of H2O2 (above the Kd) are experimentally accessible in the stopped-flow experiment. It seems unlikely that rAPX would utilize a different mechanism and we assume that detection of the intermediate is not possible in this case because the concentrations of H2O2 required to satisfy the inequality [H2O2] ≫ Kd would generate observed rate constants outside of the experimental stopped-flow limit. As such, only the linear part of the nonlinear k1,obs vs. [H2O2] dependence is experimentally accessible for rAPX, and O–O bond cleavage is not rate-limiting under any experimental conditions. [In fact, for all APXs [17–19] and variants of rAPX [16,21,22] examined so far, a linear dependence on [H2O2] is observed and second-order rate constants are derived (k1 ≈ 107m−1·s−1).] Indeed, this intermediate has eluded detection for this very reason in other peroxidase systems and its exact structure is still not clear. For example, Baek & Van Wart [23,24] identified an intermediate, designated Compound 0 and proposed to be a hyperporphyrin (FeIII-OOH) complex, in the reaction of HRP with various peroxides under cryogenic conditions (λmax ≈ 330, 410 nm for HRP-H2O2). Transient intermediates have also been detected for HRP in polyethylene glycol [26] and for the R38L/R38G and H42L variants of HRP [25,64]. Comparison of the spectra obtained in this work with theoretical calculations [29] on the hyperporphyrin (FeIII-OOH) vs. neutral peroxide (FeIII-HOOH) structure for this intermediate, together with the similarity of the spectrum of the intermediate to that of the ferric derivative, indicates that the new intermediate for H42A is probably a neutral ferric peroxide complex. This suggests that the substrate is bound initially as a neutral peroxide, consistent with the known pKa of H2O2 (pKa = 11.6), which dictates that the peroxide molecule is predominantly in the protonated form under the conditions used in this work (deprotonation of the bound hydroperoxide species is assumed to occur before O–O bond cleavage).

The pH-dependent data for Compound I formation, Fig. 5, are particularly informative. The absence of a pH-dependence for H42A unambiguously assigns this residue as being responsible for the pH-dependent reaction between rAPX and H2O2. The pKa of His42 can also be derived directly (pKa = 4.9) from the rAPX data and indicates that this group must be deprotonated for efficient reaction with H2O2 (Scheme 1). By analogy with the rAPX data, the new pKa observed for H42E (pKa = 6.7) can be assigned as arising from titration of the glutamic acid residue at position 42: although this is slightly higher than the pKa of the free amino acid (pKa = 4.1), it is well known that protein pKa values are a sensitive function of the protein environment and that protein structure is often able to modulate thermodynamic proton-transfer events over a wide range (for example as evidenced from the range of pKa values exhibited by the distal residue in peroxidases [65]). The pKa for His42 corresponds closely to that reported previously (pKa = 5.0 [17]) for pAPX, although for the pAPX enzyme no assignment of this titratable residue was possible. For peroxidases that exhibit pH-dependent kinetics of Compound I formation, the pKa values are in a similar range to that found in this work (for example, HRP (pKa = 2.5–5.3 [66,67]), myeloperoxidase (pKa = 4.0 [68,69]) and Coprinus cinereus peroxidase (pKa = 5.0 [70]) and have been assigned as arising from titration of the distal residue. For peroxidases that exhibit pH-independent kinetics, for example, lignin [71] and manganese peroxidases [72], titration of the distal histidine presumably still influences Compound I formation, but the pKa is not experimentally accessible. For CcP, examination of the role of the distal histidine residue in Compound I formation has been complicated by specific buffer effects that alter the kinetic profile [46,62]. Where pH-dependent rate constants have been reported, the pKa values (pKa = 5.4 [62], 4.0 [62]) are in a similar range to those reported here for rAPX.

The catalytic deficiency of H42A can be partially compensated for by the use of exogenous imidazole [57]; we observed 1,2-dimethylimidazole to be more effective than imidazole in recovering activity against guaiacol. This is probably related to the relative affinities of the two imidazole derivatives for binding to the haem iron: binding of imidazole but not 1,2-dimethylimidazole is observed, leading to a linear dependence of the activity over the entire concentration range for 1,2-dimethylimidazole, but not for imidazole. The effect of exogenous imidazoles is also evident for the rate constant for Compound I formation for H42A, which was enhanced by ≈ 10-fold and ≈ 40-fold for imidazole and 1,2-dimethylimidazole, respectively. For both imidazoles, however, the rate constant for Compound I formation and the peroxidase activity are still much lower than for rAPX itself, clearly highlighting the very specific role that the distal acid–base catalyst plays in peroxidase catalysis.


This work was supported by grants from the BBSRC (Special Studentship to L. L. and project grant 91/B11469), the Royal Society (grants 18851 and 21138) and Zeneca (CASE award to L. L.). Mr J.Lamb and Professor Peter Farmer (Centre for Mechanisms of Human Toxicity, Leicester University) are gratefully acknowledged for assistance with MS analyses. Drs Ian Ashworth and Brian Cox (Syngenta formerly Zeneca) are gratefully acknowledged for supporting this work. We are also grateful to Mr Kuldip Singh for technical assistance.