Potent Reversible Inhibition of Myeloperoxidase by Aromatic Hydroxamates*

Background: Myeloperoxidase causes oxidative damage in many inflammatory diseases. Results: New substituted aromatic hydroxamates are identified as potent, selective, and reversible inhibitors of MPO. Conclusion: Binding affinities of hydroxamates to the heme pocket determine the potency of inhibition. Significance: Compounds that bind tightly to the active site of myeloperoxidase have potential as therapeutically useful inhibitors of oxidative stress. The neutrophil enzyme myeloperoxidase (MPO) promotes oxidative stress in numerous inflammatory pathologies by producing hypohalous acids. Its inadvertent activity is a prime target for pharmacological control. Previously, salicylhydroxamic acid was reported to be a weak reversible inhibitor of MPO. We aimed to identify related hydroxamates that are good inhibitors of the enzyme. We report on three hydroxamates as the first potent reversible inhibitors of MPO. The chlorination activity of purified MPO was inhibited by 50% by a 5 nm concentration of a trifluoromethyl-substituted aromatic hydroxamate, HX1. The hydroxamates were specific for MPO in neutrophils and more potent toward MPO compared with a broad range of redox enzymes and alternative targets. Surface plasmon resonance measurements showed that the strength of binding of hydroxamates to MPO correlated with the degree of enzyme inhibition. The crystal structure of MPO-HX1 revealed that the inhibitor was bound within the active site cavity above the heme and blocked the substrate channel. HX1 was a mixed-type inhibitor of the halogenation activity of MPO with respect to both hydrogen peroxide and halide. Spectral analyses demonstrated that hydroxamates can act variably as substrates for MPO and convert the enzyme to a nitrosyl ferrous intermediate. This property was unrelated to their ability to inhibit MPO. We propose that aromatic hydroxamates bind tightly to the active site of MPO and prevent it from producing hypohalous acids. This mode of reversible inhibition has potential for blocking the activity of MPO and limiting oxidative stress during inflammation.

The neutrophil enzyme myeloperoxidase (MPO) promotes oxidative stress in numerous inflammatory pathologies by producing hypohalous acids. Its inadvertent activity is a prime target for pharmacological control. Previously, salicylhydroxamic acid was reported to be a weak reversible inhibitor of MPO. We aimed to identify related hydroxamates that are good inhibitors of the enzyme. We report on three hydroxamates as the first potent reversible inhibitors of MPO. The chlorination activity of purified MPO was inhibited by 50% by a 5 nM concentration of a trifluoromethyl-substituted aromatic hydroxamate, HX1. The hydroxamates were specific for MPO in neutrophils and more potent toward MPO compared with a broad range of redox enzymes and alternative targets. Surface plasmon resonance measurements showed that the strength of binding of hydroxamates to MPO correlated with the degree of enzyme inhibition. The crystal structure of MPO-HX1 revealed that the inhibitor was bound within the active site cavity above the heme and blocked the substrate channel. HX1 was a mixed-type inhibitor of the halogenation activity of MPO with respect to both hydrogen peroxide and halide. Spectral analyses demonstrated that hydroxamates can act variably as substrates for MPO and convert the enzyme to a nitrosyl ferrous intermediate. This property was unrelated to their ability to inhibit MPO. We propose that aromatic hydroxamates bind tightly to the active site of MPO and prevent it from producing hypohalous acids. This mode of reversible inhibition has potential for blocking the activity of MPO and limiting oxidative stress during inflammation.
Myeloperoxidase (MPO) 6 is a vital component of host defense. This heme enzyme produces hypochlorous acid (HOCl) as part of the microbicidal attack on invading organisms by the neutrophil. It is apparent, however, that MPO activity exacerbates many inflammatory diseases including atherosclerosis, glomerulonephritis, multiple sclerosis, rheumatoid arthritis, asthma, and cystic fibrosis (1). Evidence is also mounting for its role in promoting oxidative stress in Alzheimer disease, Parkinson disease, diabetes mellitus, and some cancers (2)(3)(4). Therefore, MPO inhibitors may be useful for the treatment of a broad range of human diseases. Despite the growing understanding of its complex enzymology and pharmacology, few therapeutically suitable inhibitors have been discovered that specifically target MPO.
A number of different inhibitors of MPO have been reported over the last four decades. These can be classified into three main categories: those that promote accumulation of Compound II, suicide substrates, and those that bind reversibly to the native enzyme. The first two types of inhibitors serve as alternative substrates that divert MPO from its normal catalytic cycle (Fig. 1). Inhibitors that cause accumulation of Compound II are poor peroxidase substrates that react well with Compound I but slowly with Compound II. These include dapsone (5), tryptamines (6), tryptophan analogues (7), and nitroxides (8,9). Such inhibition is unlikely to be effective in a normal physiological environment because of an abundance of better peroxidase substrates such as ascorbate (10) and urate (11) that will efficiently convert any accumulated Compound II back to the active native MPO state. The plasma protein ceruloplasmin is an endogenous inhibitor of MPO that also acts by promoting accumulation of Compound II (12). However, it also prevents reduction of Compound II so MPO becomes trapped in this redox state.
Suicide substrates, or mechanism-based irreversible inhibitors, of MPO include 4-aminobenzoic acid hydrazide (13) and 2-thioxanthines (14). Oxidation of these inhibitors by MPO promotes inactivation either by destruction or covalent modification of the heme prosthetic groups of the enzyme. Other redox-based inhibitors include paracetamol (15) and isoniazid (16). They are reversible inhibitors that divert MPO from its halogenation cycle. In the process they produce radical intermediates. With all of the substrate-based inhibitors, whether irreversible or reversible, there is possible generation of undesirable, reactive by-products of the oxidized inhibitor. As MPO is a heme peroxidase with extremely powerful oxidizing abilities (17,18), it is indeed not surprising that the majority of known inhibitors are oxidized by the enzyme. Reactive radicals formed during inhibition may promote local toxic chain reactions or lead to hapten formation in vivo (16,19,20). This feature places major restrictions on the feasibility of inhibitors as therapeutic agents. However, the problem is minimized for the most potent 2-thioxanthine compounds because they inactivate MPO within a single turnover of the enzyme (14).
Reversible inhibitors that bind to the native enzyme differ from the substrate-based inhibitors in that they compete with MPO substrates by occupying the heme binding pocket. As an alternative mechanism, this is an attractive means of inhibition because the oxidizing capability of the enzyme is simply blocked without permanent changes to the enzyme or production of unwanted by-products. Salicylhydroxamic acid (SHA) was identified as a reversible inhibitor of MPO (21) after earlier observations of broad peroxidase inhibition by substituted aromatic hydroxamates (22). However, SHA performed poorly in MPO inhibition assays in comparison with benzoic acid hydrazides (23).
Proof of the competitive nature of SHA-enzyme binding (24) and the subsequent crystal structure of the MPO-SHA complex (25) spawned the hypothesis that modified hydroxamates could be identified as new, more potent reversible inhibitors of MPO. For this type of inhibitor, the critical feature is the docking of the molecule in the heme binding pocket of MPO. In this study, we aimed to explore different substituted aromatic hydroxamates to identify compounds with stronger binding affinities and improved specific inhibition of the halogenation activity of MPO. Our results show that the strength of hydroxamate-MPO binding correlated with the inhibition of MPO activity. We have solved the crystal structure of the MPO-hydroxamate complex and determined the mechanism of inhibition by heme spectral analysis and substrate competition kinetics. We present new compounds, in particular hydroxamate HX1, as highly potent and reversible inhibitors of MPO.

EXPERIMENTAL PROCEDURES
Materials-Human MPO (EC 1.11.2.2) purified from human blood (purity index (A 430 /A 280 ) Ͼ0.84) was purchased from Planta (Wien, Austria). Human recombinant thyroid peroxidase (TPO; purity Ͼ95% by SDS-PAGE) was purchased from RSR Ltd. (Cardiff, UK). Bovine lactoperoxidase (LPO; purity index (A 412 /A 280 ) Ͼ0.88) was purchased from Sigma. For structural characterization of complexes between MPO and inhibitors, MPO was purified from HL-60 cells, which were obtained from American Type Culture Collection (Manassas, VA). Cells were grown in DMEM/F-12 (Invitrogen) plus 5% fetal calf serum and 5 mM glutamine in a 50-liter reactor to a cell density of 1.7 ϫ 10 6 /ml. The purification was a modification of the protocol described previously (26). In the modified protocol, the ammonium sulfate precipitation steps were excluded, and the final purification was achieved using Superdex 200 (GE Healthcare) size exclusion chromatography. Purity and identity of MPO were determined by 10% SDS-PAGE and N-terminal sequencing. Hydroxamates were prepared by the Department of Medicinal Chemistry, AstraZeneca Research and Development Charnwood. Pronase was from Roche Diagnostics, and human serum albumin was the clinical product Albumex 4 from CSL Ltd. (Australia). All other reagents were of the highest purity commercially available and were from Sigma unless otherwise stated.
Myeloperoxidase Assays-Enzyme activity was determined as the production of HOCl via accumulation of taurine chloramine, which was detected using iodide-catalyzed oxidation of 3,3Ј,5,5Ј-tetramethylbenzidine (27). Assays were performed at 22°C with 2 nM MPO and 10 M hydrogen peroxide (H 2 O 2 ) in 20 mM NaH 2 PO 4 buffer, pH 6.5 containing 140 mM NaCl, 10 mM taurine, and 1 mM L-tyrosine. Inhibitor compounds were preincubated with MPO for 15 min prior to the addition of H 2 O 2 , and the accumulation of taurine chloramine was determined after 1 min. Inhibitory effects of compounds are expressed as percentage of control activity in the absence of compound, and curves were fitted to data using Origin 7.5 (Origin Labs). The concentration of inhibitor giving 50% of the full enzyme activity measured in the absence of inhibitor is the IC 50 value.
The consumption of H 2 O 2 by MPO was measured with the ferrous oxidation of xylene orange (FOX) assay (28). The assay was modified to imitate the protein-rich environment of plasma with the inclusion of 200 M urate, 50 M tyrosine, 50 M tryptophan, and 1 mg/ml albumin in 50 mM phosphate buffer, pH 7.4 containing 140 mM NaCl and 1 mM methionine to scavenge any HOCl. Reactions were performed at room tem- perature in Eppendorf tubes and started by adding 20 M H 2 O 2 to 5 nM MPO in the presence or absence of inhibitor. Each 200-l reaction was stopped after 15 min by addition (on a Vortex mixer) of one-third volume (67 l) of FOX developer (400 M xylene orange, 1 mM ferrous ammonium sulfate, 400 mM sorbitol in 200 mM H 2 SO 4 ). Aliquots (200 l) were transferred to a microtiter plate, and the absorbance was measured after 45 min at 560 nm. Time course experiments showed that ϳ10 M H 2 O 2 was consumed in 15 min (not shown). Each reaction was blanked against a control without MPO, and inhibition was expressed as a ratio of the change in absorbance in the presence of inhibitor to that in the absence of inhibitor.
Reversibility of inhibition was determined in a system using immobilized MPO. MPO (10 g/ml in 100 mM sodium carbonate buffer, pH 10) was immobilized onto the well surface of protein immobilizer plates (Exiqon, Vedbaek, Denmark) according to the manufacturer's instructions, and HOCl production was assessed by taurine chloramine assay prior to and after extensive washing in enzyme assay buffer.
The halogenation of NADH by MPO was monitored to determine the kinetics for competing substrates. This assay, detecting the initial rate of bromohydrin production at 275 nm using ⑀ 275 of 11,800 M Ϫ1 cm Ϫ1 , is a direct measure of the formation of hypohalous acid by MPO (29). Bromide was chosen because of faster reaction rates compared with other halides (30). Briefly, 20 nM MPO was incubated at room temperature in 20 mM phosphate buffer, pH 7.4 containing 100 M NADH and varying concentrations of inhibitor and NaBr. The absorbance changes were monitored upon addition of H 2 O 2 to start the reaction. Initial rates were measured over the 1st min of reaction, and K m and V max were determined using non-linear regression (Sigma Plot, Jandel Scientific).
Cell Assays-Human neutrophils were purified from peripheral venous blood (36) and then resuspended in 10 mM NaH 2 PO 4 buffer, pH 7.4 containing 140 mM NaCl, 0.5 mM MgCl 2 , 1 mM CaCl 2 , and 1 mg/ml D-glucose (Hanks' buffer). Production of HOCl was measured by the taurine chloramine assay (27) using cells at 1.4 ϫ 10 6 /ml with 5 mM taurine included in the buffer and stimulated with 30 ng/ml phorbol 12-myristate 13-acetate (PMA) for 40 min at 37°C. Superoxide production was measured as the rate of cytochrome c reduction (37) using PMA-stimulated cells as above with 2.5 mg/ml cytochrome c added to the buffer. Absorbance readings were taken at 550 nm at 1-min intervals for 15 min at 37°C.
Neutrophils (2 ϫ 10 6 /ml in Hanks' buffer) were stimulated with PMA (100 ng/ml) in the presence of human serum albumin (0.5 mg/ml), and the chlorination of tyrosine residues was measured by mass spectrometry. After 40 min at 37°C, cells were pelleted, and the supernatant was removed and spiked with internal standards including 1 nmol of [ 13 C 6 ]tyrosine and 500 fmol of 3-chloro[ 13 C 9 ]tyrosine. The samples were then lyophilized prior to Pronase digestion in 100 mM Tris, pH 7.5 containing 10 mM CaCl 2 for 18 h with a 5:1 excess of protein to protease. Samples (ϳ100 g of protein) were lyophilized again and reconstituted in 10 mM phosphate buffer at pH 7.4 for detection of 3-chlorotyrosine and tyrosine by liquid chromatography with mass spectrometry (LCMS).

3-Chlorotyrosine Measurement by LCMS/MS-
The method of analysis was similar to that published previously (38) with additional monitoring of 3-chlorotyrosine by the 3:1 ratio of its 35 Cl and 37 Cl isomers. High performance liquid chromatography (HPLC) was performed on a Dionex Ultimate 3000 pump with a 3-m Hypercarb column (250 ϫ 2.1 mm) with an identical guard column and an SDS guard cartridge (all Thermo Scientific). Detection was on an Applied Biosystems (Ontario, Canada) 4000 QTRAP electrospray mass spectrometer via stable isotope multiple reaction monitoring for tyrosine and its chlorinated derivatives. Use of the internal standards [ 13 C 6 ]tyrosine and chloro[ 13 C 9 ]tyrosine enabled complete quantification as well as monitoring any artifactual chlorination of tyrosine. For tyrosine, the fragment transitions that were monitored had m/z values of 182 to 136, 188 to 142, and 191 to 144 for [ 12 C]tyrosine, [ 13 C 6 ]tyrosine, and [ 13 C 9 ]tyrosine, respectively. Correspondingly, for 3-chlorotyrosine the transitions had m/z values of 216 to 170, 222 to 176, and 225 to 178 for the 35 Cl isotope of each species and 218 to 172, 224 to 178, and 227 to 180 for the 37 Cl isotopes. Standard curves were generated using known standards, and results were calculated as mol of 3-chlorotyrosine/1000 mol of tyrosine.
Measurement of Compound Binding Kinetics-Binding kinetics were determined by surface plasmon resonance (SPR) using a Biacore S51 (Biacore, Uppsala, Sweden). MPO (50 g/ml dissolved in 10 mM sodium acetate, pH 5.0) was immobilized onto the surface of CM5 sensor chips (Biacore) using surface amine coupling. One of the spots on the sensor surface was left without MPO to control for nonspecific binding. The signal observed in response to analyte binding was as expected linearly related to the amount of immobilized ligand, and 10,000 response units was routinely used to characterize compound binding (data not shown). Compounds were dissolved in binding buffer (10 mM HEPES, pH 7.4, 150 mM NaCl, 3 mM Na 2 EDTA, 0.005% (w/v) surfactant P20, 1% (v/v) DMSO final), and association was assessed during 60 -210-s injections. After this time, analyte injection was terminated, and the chip surface was perfused with binding buffer at 30 l/min for 4 -12 min to monitor compound dissociation.
Compound binding responses were determined as a change in signal following solvent correction and subtraction of baseline responses using the Biacore S51 Evaluation program. Specific saturation binding data were fitted to a logistic equation assuming an interaction of the compound with a single class of binding site to obtain estimates of the dissociation constant pK D (Ϫlog 10 K D where K D ϭ k d /k a ). Kinetic data were fitted to single exponential association and dissociation equations using Biacore S51 Evaluation. An interaction with a single population of binding sites was assumed, and estimates of association and dissociation rates (k a and k d ) and the half-life for dissociation (t1 ⁄ 2 ϭ 0.69/k d ) were obtained. Mean values were generated by taking the arithmetic mean of the individual estimates.
Crystal Structure Determination-For crystallization, the protein buffer was exchanged with 20 mM sodium acetate buffer, pH 5.5 containing 50 mM ammonium sulfate and 2 mM CaCl 2 , and the MPO sample was concentrated to about 10 mg/ml. The compound was added to the protein sample to a final concentration of ϳ1 mM in 2% DMSO. After a 6-h incubation, excess ligand precipitate was removed by centrifugation. Crystals were obtained by the hanging drop vapor diffusion technique. The protein sample (1 l) was mixed with 1 l of a well solution containing 18% PEG 3350 and 0.1 M NaCl. The drop was allowed to equilibrate over a reservoir containing well solution. Prior to data collection, the drops containing crystals were supplied with glycerol as a cryoprotectant. The crystals were then quickly removed from the drop and flash cooled in liquid nitrogen. Data were collected at beam line ID14 EH4 at a wavelength of 0.939 Å. The data were processed using MOSFLM (39) and scaled and further reduced using the CCP4 suite of programs (40). For statistics, see Table 1.
Initial phasing was done by molecular replacement using a high resolution ligand-free structure of MPO (Protein Data Bank code 1CXP (41)) as a starting model. The F o Ϫ F c difference map showed positive residual density in the distal heme cavities in each half of the molecule corresponding to the bound HX1 (see Fig. 6B). Although the difference map allowed unambiguous modeling of the HX1 molecule, it was clear that the site was not fully occupied, and it was possible to outline the solvent structure of the ligand-free enzyme superimposed on the ligand structure. When the ligands were refined assuming full occupancy of the ligand, the B factors were refined to values almost twice the average B factor for protein atoms, and the occupancy was therefore set to 0.5. Model rebuilding was performed within O (42), and refinement was performed using REFMAC5 (40). For statistics for the final models, see Spectrophotometric Analyses-UV-visible absorbance spectra were recorded on an Agilent 7500 diode array spectrophotometer operated at room temperature. Spectra between 190 and 1100 nm at 1-nm intervals were recorded every 30 s (and 250 -700 nm is reported) for analysis of spectral changes due to the MPO-HX interaction. On this spectrophotometer, each spectrum is the average of 10 readings taken over 1 s.

Inhibition of MPO by Aromatic
Hydroxamates-We set out to design more potent and selective ferric state MPO inhibitors based initially on SHA. Three substituted aromatic hydroxamates have been named HX1, HX2, and HX3 (Fig. 2). Inhibitor potency against production of HOCl was routinely estimated in the presence of 10 M H 2 O 2 and 1 mM tyrosine. The inclusion of tyrosine ensures optimal turnover of the enzyme and prevention of Compound II accumulation as it is a good peroxidase substrate (43). SHA was found to be a weak inhibitor of HOCl production, and the other inhibitors evaluated were significantly more potent (Fig. 3A). HX1 was by far the most potent inhibitor (IC 50 ϭ 5 nM) with the following rank order of potency obtained across the group: HX1 Ͼ HX2 Ͼ HX3 Ͼ SHA.
The selectivity of the inhibitors for MPO was tested against a panel of enzymes (Table 2) including the closely related heme peroxidases LPO and TPO. Generally, the new inhibitors all showed high selectivity for the heme peroxidases over the other redox enzymes and the more disparate targets tested. HX1, HX2, and HX3 were tested in a panel of more than 30 standard in vitro activity assays 7 covering a diverse range of receptors, ion channels, transporters, and enzymes at a single concentra-    Table 2). HX1 was the most selective inhibitor identified with Ͼ300-fold higher potency against MPO compared with any of the other targets investigated.
Next we sought to test the efficiency of HX1 as an inhibitor of MPO in an assay that more closely resembled the physiological environment. This assay would demonstrate whether effective inhibition could be achieved in the presence of protein and other potential substrates of the enzyme. We measured the consumption of hydrogen peroxide in a modified FOX assay (28) that included typical plasma concentrations of urate, tyrosine, tryptophan, and albumin (Fig. 3B). Hydrogen peroxide consumption was linear and dependent on the presence of MPO. HX1 showed strong inhibition of MPO in this system with an IC 50 of 50 nM. Inhibition was immediate and constant over time (not shown). Under the same conditions, TX1, the 2-thioxanthine that irreversibly inhibits MPO (14), was required at a 10-fold higher concentration to achieve the same inhibition.
Reversibility of Inhibition-The inhibition of MPO displayed by all the compounds under evaluation was shown to be fully reversible. Representative data for HX1 and SHA are given in Fig. 3C. In this assay, MPO was immobilized onto an ELISA plate, and its chlorination activity was determined in the presence of inhibitor. The plate was then washed to remove the inhibitor, and the chlorination activity was remeasured. The chlorination activity of the immobilized MPO was lowered upon incubation with inhibitor in line with previously determined IC 50 values. However, upon extensive washing of the plate, enzyme activity was restored, indicating that the interaction and inhibition were reversible.
Inhibition of Neutrophil MPO by Hydroxamates-Inhibitor potencies for blocking the production of HOCl by neutrophils are presented in Table 3. Peripheral blood neutrophils were incubated with increasing concentrations of inhibitor, and the effect on PMA-stimulated oxidant production was measured.  The inhibitors blocked HOCl generation with the following rank order of potency as determined by IC 50 values: HX1 Ͼ HX2 Ͼ HX3 Ͼ SHA. To test for specificity of inhibition in the cell environment, compounds HX1-HX3 were also assessed for their effect on superoxide production. There was no significant effect on superoxide production by neutrophils with concentrations well in excess (5-200-fold) of the IC 50 values for HOCl production.
To test the effect of HX1 on chlorination of tyrosine residues in proteins, neutrophils were stimulated in the presence of human serum albumin under conditions that produce substantial formation of 3-chlorotyrosine in proteins (44). When proteins in the supernatant of stimulated neutrophils were digested with Pronase, it was apparent that 3-chlorotyrosine was formed because a product of the requisite mass co-eluted with the stable isotopes of authentic 3-chloro[ 13 C]tyrosine (Fig.  4A). The product also co-eluted with a second product that was 2 mass units higher and was present at a third of its abundance, which is characteristic of chlorine isotopes. The formation of 3-chlorotyrosine was quantified by a sensitive LCMS/MS method, which demonstrated that, under the reaction conditions, stimulated neutrophils chlorinated approximately three tyrosine residues per 1000 and that this was almost completely inhibited by 1 M HX1 (Fig. 4B, inset). HX1 inhibited chlorination of tyrosine residues in a dose-dependent manner with an estimated IC 50 of 150 nM (Fig. 4B). No artifactual chlorination during sample handling was detected. Also, HX1 did not inhibit chlorination by reagent HOCl (data not shown), indicating that it inhibited MPO rather than scavenging HOCl.
Inhibitor-Enzyme Binding Characteristics-The ability of the hydroxamate analogues to bind to ferric MPO was investigated by SPR. HX1, HX2, and HX3 all caused concentration-dependent, saturable increases in SPR responses (Fig. 5), and the rank order of affinity was the same as the rank order of potency as enzyme activity inhibitors ( Table 4). The observed increase in binding affinity from HX3 to HX1 was reflected by significantly slower off-rates and on-rates. For example, the dissociation phase for HX3 was much faster than that of HX1. In contrast, the calculated on-rate constants were similar (within 4-fold) for HX2 and HX3. This suggests that inhibitor optimization (decreasing K D ) from HX3 to HX1 is largely driven by increasing the stability of the bound complex.
Crystal Structure of the Ferric MPO-Inhibitor Complex-To investigate the nature of the interaction between ferric MPO and HX1, we determined the x-ray crystal structure of the complex to 2.0-Å resolution. The overall structure of the protein chain in the inhibitor complex was identical to that found in the ligand-free enzyme (41). HX1 was unambiguously modeled into the electron density (Fig. 6). The pyridine ring of HX1 is positioned almost parallel to the plane of pyrrole ring D of the MPO heme, and the trifluoromethyl aromatic ring is bent, occupying the hydrophobic pocket at the entrance of the active site. The hydroxamic acid group lies in the center of the distal cavity, and both the carbonyl and hydroxyl oxygens form hydrogen bonds with amino acid side chains in the distal cavity. One of the pyrimidine nitrogens provides the main interaction   DECEMBER 20, 2013 • VOLUME 288 • NUMBER 51

Reversible Inhibition of Myeloperoxidase
with the heme propionate group, whereas the oxygen of the hydroxyl group is well positioned for a hydrogen bond with Arg-239. The trifluoromethyl aromatic ring "tail" of the mole-cule extends toward the surface of the enzyme. There are no hydrogen bonds to the protein outside the active site, but the trifluoromethyl groups are likely contributing weak electrostatic interactions with Thr-238 (C-O⅐⅐⅐F-C distance around 3.4 Å). The carbonyl oxygen binding site overlaps with the previously identified halide binding site (41).

Kinetic Studies of the Inhibition of MPO Halogenation
Activity-To examine the kinetics of the inhibition of halide oxidation by MPO, the effect on the initial rate of the bromination of NADH was assessed at varying concentrations of bromide and hydrogen peroxide. The NADH bromohydrin has a distinctive absorption spectrum that gives this assay high sensitivity with respect to the halogenation activity of MPO. HX1 inhibited the formation of NADH bromohydrin with an IC 50 of 70 nM using optimal concentrations of 50 M H 2 O 2 and 10 mM NaBr. At 100 nM, HX1 had a marked effect on the rate of formation of the bromohydrin with increasing concentrations of both halide (Fig. 7A) and hydrogen peroxide (Fig. 7B). Kinetic constants were obtained from these curves ( Table 5) and showed that the inhibitor decreased the catalytic production of bromohydrin, k cat , by 82 and 78% for halide and hydrogen peroxide, respectively. The ratio indicating the catalytic efficiency, k cat /K m (45), of MPO in this system also decreased significantly with respect to the two substrates by 43 and 59% for bromide and hydrogen peroxide, respectively. This indicates that the hydroxamate HX1 acts as a mixed-type inhibitor with respect to both halide and hydrogen peroxide.
Spectral Changes upon MPO-Hydroxamate Interaction-To further understand the mechanism of inhibition of MPO by these compounds, their interaction was studied spectrophotometrically. MPO was incubated with hydroxamate HX1 or HX2 (Fig. 8). The inhibitors had no effect on the absorption spectrum of ferric MPO. However, upon adding hydrogen peroxide, there were changes in the heme spectrum of MPO. In the case of HX1, there was a decline in absorbance at 430 nm and a slight increase in absorbance between 450 and 500 nm followed by slow decay back to the ferric spectrum (Fig. 8A). The difference spectrum of the transient form revealed peaks at 466 and 632 nm (Fig. 8B). The spectral changes were much more pronounced for the interaction with HX2, showing a stable shift to a spectral form with peaks at 468 and 637 nm (Fig. 8C). The spectrum also showed changes in the far-UV region consistent with oxidation of the inhibitor HX2 (Fig. 8D). The effect on the heme spectrum, although different in magnitude for HX1 and HX2, gave a common result. The interaction between MPO and these hydroxamates gave rise to the characteristic spectrum of a nitrosyl complex with ferrous MPO with Soret maxima at 467 and 635 nm (46). Together with the UV spectral changes, the formation of nitrosyl ferrous MPO indicates that these hydroxamates are to some extent metabolized by MPO. Compound HX2 showed greater formation of this complex than HX1, which is the inverse of their inhibitory potency. Hence, the loss of activity does not correlate with the degree of the formation of the nitrosyl complex of MPO.

DISCUSSION
We have identified three aromatic hydroxamates that have unprecedented high potency as reversible inhibitors of the hal- MPO was immobilized to a CM5 sensor chip, and compound binding was evaluated for HX1, HX2, and HX3. The panels show specific binding traces representative of three to six separate experiments. Each panel shows the overlay of multiple serial sensorgram traces for nine different compound concentrations in a series of 3-fold dilutions ranging from 0.3 M to 30 pM for HX1 and HX2 and from 30 M to 3 nM for HX3. From t ϭ 0, compound was continuously perfused over the sensor chip leading to a clear net association of compound to the immobilized MPO. Subsequently, compound perfusate was replaced with buffer at t ϭ 60 -210 s, leading to a loss of response due to net compound dissociation. RU, response units.

TABLE 4 Affinity and kinetic estimates for hydroxamate binding to ferric MPO
Dissociation constants (K D ) were derived from parameter logistic curve fitting to [compound] versus binding measurements. Association and dissociation rate constants (k a and k d ) were measured by SPR from which the half-life for dissociation (t1 ⁄ 2 ) was calculated. Data are presented as mean Ϯ S.E. from three to six independent experiments. ND, not determined. ogenation activity of MPO. Of particular interest is the trifluoromethyl-substituted compound HX1, which is by far the best inhibitor of MPO currently known. Its physical occupancy of the active site is the defining feature by which it inhibits MPO.
The new hydroxamates showed specific inhibition of MPO when screened against other human enzymes and offer the highest seen reversible inhibition with an IC 50 of 50 nM under physiological conditions. Our findings demonstrate that this type of inhibitor has potential as a therapeutic agent against the detrimental activity of MPO and should also provide useful information about the active site of heme peroxidases. The hydroxamates inhibit MPO by binding to the active site of ferric MPO and blocking the access of substrates. This mode of action is supported by our findings that hydroxamates bind tightly to the enzyme, occupy the active site and perturb binding of substrates, cause reversible inhibition, and act as mixedtype inhibitors with respect to hydrogen peroxide and halides. Also, their abilities to inhibit and bind to MPO were directly related. Inhibition of this type is likely to occur in vivo because the hydroxamates not only exhibited potent inhibition of the purified enzyme but also consistently inhibited MPO in more physiological systems including a multisubstrate MPO assay. With PMA-stimulated neutrophils, they were effective inhibitors of HOCl production as well as chlorination of tyrosine residues in proteins. The IC 50 values in the latter systems were significantly higher than in the simple assay with purified enzyme, but the trend in potency remained unchanged with HX1 Ͼ HX2 Ͼ HX3 Ͼ SHA. The raised IC 50 values in the more    Fig. 9 that prevents turnover of the enzyme in both its halogenation and peroxidation cycles. The crystal structure of the complex formed by ferric MPO and HX1 was solved to high resolution. Although HX1 binds tightly to MPO (K D ϭ 15 nM), the occupancy was only ϳ0.5. It is likely that the crystallization conditions have influenced ligand binding. In particular, the low pH of crystallization (pH 5.5) may partially protonate His-95, which is involved in hydrogen bonding to HX1, and this would have an effect on ligand occupancy. Nevertheless, the crystal structure provides clear evidence of the nature of the interactions between HX1 and MPO. The hydroxamate is located in the substrate binding pocket of MPO without any significant conformational changes to the native structure of the enzyme. The position of the hydroxamic acid group in the distal cavity is similar to that described previously for the MPO-SHA complex (25). However, the planar tilt angle differs by ϳ20°due to additional interactions between the pyridine ring of HX1 and the heme propionate group. Also, the second ring of HX1 with its trifluoromethyl groups creates a hydrophobic tail that contributes to improved affinity over SHA and provides additional steric hindrance for substrate access to the active site. The second ring system also enhances selectivity for MPO over other heme peroxidases. That is, for SHA, the potency of inhibition is more than 10-fold lower for MPO than for TPO and LPO, whereas HX1 is greater than 2 orders of magnitude more potent toward MPO than both TPO and LPO. The structure of TPO is not known, but the improved selectivity for MPO over LPO can be rationalized by comparing the shape of the cavity adjacent to  the active site. The loop corresponding to residues 407-415 in MPO adopts a different conformation in the structure of both caprine and bovine LPO (47) that would prevent binding of HX1.
The reversibility of the inhibition displayed by HX1 and SHA (Fig. 3C) confirms that hydroxamate inhibitors simply dock at the active site of MPO unlike the 2-thioxanthine inhibitors that become irreversibly covalently bound to the heme (14). Notably, the 2-thioxanthine series of potent inhibitors also features multiring structures with a bent tail that extends into the hydrophobic pocket.
Another aspect of the interaction between hydroxamates and MPO that is of interest to their pharmacology is that they are also potential peroxidase substrates. We found spectral evidence of hydroxamate oxidation by MPO in the case of HX2 with discernible losses in the fa-UV region attributable to MPO metabolizing this substrate. It is long established that hydroxamates such as SHA can serve as redox substrates of peroxidases such as horseradish peroxidase (22,48) as well as MPO (21,24). Hydroxamates can undergo oxidation to a transient nitroxide radical (RC(O)NHO ⅐ ) (49) and therefore are potential reductants in the classical peroxidase cycle. The reactions of hydroxamates (RC(O)NHOH) with MPO intermediates are summarized in Fig. 9 and can be regarded as secondary to the inhibitory complex formation.
Our spectral analyses with the new substituted aromatic compounds revealed the formation of a nitrosyl adduct form of ferrous MPO ( Fig. 9; NO-Fe(II)). This was previously discovered by the direct reaction of gaseous NO with MPO, yielding Soret maxima at 467 and 635 nm (46). We observed this NO-Fe(II) spectrum after reaction of MPO with hydrogen peroxide in the presence of HX1 or HX2 (Fig. 8). Ferric nitrosyl MPO with maxima at 433 and 630 nm (46) was not observed. The formation of a ferrous nitrosyl heme intermediate occurs in normal catalytic activity of nitric-oxide synthase (50) but has not been reported with other substrates of MPO. We have proposed a mechanism for its production via oxidation of hydroxamates with concomitant formation of nitric oxide and ferrous enzyme (Fig. 9). The extent to which the NO-Fe(II) complex is formed depends on the ease of oxidation of a particular hydroxamate to the nitroxide radical (RC(O)NHO ⅐ ). This radical promotes the reduction to ferrous MPO and upon hydrolysis of its oxidized form will also lead to HNO and NO (49,51). NO binds reversibly to ferrous MPO (46) and in our aerobic conditions was stable for a few minutes only (Fig. 8). Alternatively, the ferrous NO complex could form by reaction of nitroxyl with the ferric enzyme (52). There was a significant difference between the spectral changes seen for HX1 and HX2. It was evident that HX2 formed the most NO-Fe(II), but it was not as potent a binder or inhibitor of MPO as HX1. Therefore, the formation of NO-Fe(II) is counterinhibitory and indicates inhibitor instability. These results have implications for development of inhibitors as robust pharmaceutical agents regarding mechanisms of drug breakdown. The propensity of hydroxamates to act as NO donors is a recognized problem in the generation of all hydroxamate-based drugs (49).
The most potent inhibitors of MPO reported previously are the 2-thioxanthine family of suicide substrates (14). These compounds are notable in that they are mechanism-based inhibitors that do not release reactive free radicals from the active site of MPO. The production of unwanted side products is also avoided by reversible inhibitors provided there is no concurrent oxidation of the bound inhibitor. This issue of inhibitor metabolism by the enzyme is a potential shortcoming of the new aromatic hydroxamates. Another limitation of these compounds is that they undergo slow hydrolysis, which decreases their pharmacological efficacy. However, their mode of binding to MPO and their extreme potency signify that reversible inhibition is potentially the best strategy for limiting the activity of MPO in vivo.
We conclude that modified hydroxamates have proven to be highly potent and specific reversible inhibitors of MPO. The differently substituted double-ring hydroxamates have achieved higher potency because of increased polar interactions with the MPO heme and a modified bent shape suited to filling the active site cavity. This leads to stronger binding at the heme and better interference of the access of substrates to the active site. These new potent reversible inhibitors demonstrate a valuable alternative mechanism for MPO inhibition to that of irreversible mechanism-based inhibitors exemplified by the 2-thioxanthines (14). Without permanently crippling the enzyme or generating multiple radical chain reactions and by-products, this reversible type of inhibitor should be ideal for therapeutic inhibition of unwarranted MPO activity and oxidative damage. In the case of inflammatory disorders characterized by episodes of heightened neutrophil attack such as cystic fibrosis, this benign but efficient type of reversible inhibitor could be administered to control transient fluxes of released MPO. Development of pharmacologically stable inhibitors that bind reversibly to the active site of MPO but are not substrates would be of great value for the treatment of inflammatory diseases.