Modeling the Reactions of Superoxide and Myeloperoxidase in the Neutrophil Phagosome

Neutrophils kill bacteria by ingesting them into phagosomes where superoxide and cytoplasmic granule constituents, including myeloperoxidase, are released. Myeloperoxidase converts chloride and hydrogen peroxide to hypochlorous acid (HOCl), which is strongly microbicidal. However, the role of oxidants in killing and the species responsible are poorly understood and the subject of current debate. To assess what oxidative mechanisms are likely to operate in the narrow confines of the phagosome, we have used a kinetic model to examine the fate of superoxide and its interactions with myeloperoxidase. Known rate constants for reactions of myeloperoxidase have been used and substrate concentrations estimated from neutrophil morphology. In the model, superoxide is generated at several mm/s. Most react with myeloperoxidase, which is present at millimolar concentrations, and rapidly convert the enzyme to compound III. Compound III turnover by superoxide is essential to maintain enzyme activity. Superoxide stabilizes at ∼25 μm and hydrogen peroxide in the low micromolar range. HOCl production is efficient if there is adequate chloride supply, but further knowledge on chloride concentrations and transport mechanisms is needed to assess whether this is the case. Low myeloperoxidase concentrations also limit HOCl production by allowing more hydrogen peroxide to escape from the phagosome. In the absence of myeloperoxidase, superoxide increases to >100 μm but hydrogen peroxide to only ∼30 μm. Most of the HOCl reacts with released granule proteins before reaching the bacterium, and chloramine products may be effectors of its antimicrobial activity. Hydroxyl radicals should form only after all susceptible protein targets are consumed.

Neutrophils kill micro-organisms by ingesting them into phagocytic vacuoles (phagosomes). Phagocytosis is accompanied by the activation of the NADPH oxidase, an enzyme complex that assembles in the phagosomal membrane and converts oxygen into the superoxide radical anion (O 2 . ). O 2 . is generated at the external surface (i.e. inside the phagosome) with reducing equivalents supplied by intracellular NADPH. There are two apparently distinct mechanisms of killing, one requiring activation of the oxidase and the other involving antimicrobial peptides (1). Superoxide production is important for an effective antimicrobial defense, as illustrated by the recurrent infections seen in chronic granulomatous disease (2). The specific mechanisms and species responsible for oxidative killing are the subject of continuing debate (3). The O 2 .
produced by the oxidase undergoes dismutation to produce hydrogen peroxide (H 2 O 2 ). At the same time, myeloperoxidase (MPO) 2 is released into the phagosome. MPO catalyzes the formation of hypochlorous acid (HOCl) from chloride and H 2 O 2 . It also oxidizes a wide range of reducing substrates to their corresponding radicals (4). HOCl is strongly microbicidal (1), has been detected in the neutrophil phagosome (5,6), and is widely considered to be the major oxidative weapon of the neutrophil (1, 7). However, it is possible that sufficient HOCl may not be generated to kill ingested bacteria on its own (6,8), and in view of the benign clinical consequences of hereditary deficiency (9), the importance of MPO has been questioned (10 -12). Alternative explanations for antimicrobial activity, especially in MPO deficiency, include the build up of high concentrations of H 2 O 2 that might kill directly or via formation of hydroxyl radicals (13,14). Production of hydroxyl radicals (13) and singlet oxygen (15,16) in secondary reactions of HOCl is possible. Ozone formation by neutrophils has also been suggested (17), although this proposal is controversial (18,19). An alternative view is that O 2 . generation has an electrogenic function of bringing potassium ions into the phagosome to increase ionic strength and aid solubilization of granule proteins. According to this mechanism, oxidants are incidental to the killing process (11,12,20). However, this mechanism has been challenged in recent publications (21,22). To understand oxidative killing, knowledge of the fate of O 2 .
and reactions of MPO in the phagosomal environment is required. MPO cycles through redox intermediates that undergo a complex array of reactions (Fig. 1). Oxidation of most substrates (including phenolic compounds, nitrite and ascorbate) occurs via 1-electron step involving compounds I and II. Chloride undergoes a 2-electron oxidation by compound I. This reaction is inhibited by formation of compound II * This work was supported by the Health Research Council of New Zealand.
This work was presented in part at a conference on The Peroxidase Multigene Family of Enzymes (80). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. □ S The on-line version of this article (available at http://www.jbc.org) contains supplemental Fig. S1. 1 To whom correspondence should be addressed. Tel.: 64-3-3640564; Fax: 64-3-364-1083; E-mail: christine.winterbourn@chmeds.ac.nz.
or the O 2 . adduct, compound III (4,23). Superoxide is able to react with all the redox intermediates of MPO in ways that can enhance or inhibit chlorination (4,24). 3 It converts native MPO to compound III (25,26); it reacts with compound II (23,24) and, analogous to horseradish peroxidase (27), also with compound I. There is also evidence for a reaction with compound III (28). H 2 O 2 also reacts with compounds I and II (29,30). How MPO acts in the phagosome will depend on the interplay of these reactions. The phagocytosing neutrophil spreads over the surface of the targeted bacterium, creating a very narrow space where oxidants are released and killing occurs (31)(32)(33). Very little extracellular medium is taken up (34), and the major constituents are released granule contents. This creates an environment with very high concentrations of granule proteins, including MPO, into which O 2 . is released at a high rate. These conditions are impossible to mimic in pure enzyme systems and difficult to investigate directly as this requires probes that distinguish the phagosome from the cytoplasm or extracellular surroundings. We have used a kinetic model to explore the generation and fate of oxidants in the phagosome. In the model, O 2 . is generated at a rate determined from measured rates of oxygen uptake by neutrophils, and subsequent reactions of O 2 . and its dismutation product H 2 O 2 are followed. These include losses by diffusion as well as interactions with the redox intermediates of MPO. The volume of the phagosome and the total concentration of MPO are deduced from morphological data plus direct analyses of neutrophil MPO content. Where data are not available, we have assumed initial values and considered the impact of variations. The model has been used to assess the following:

EXPERIMENTAL PROCEDURES
Construction of a Kinetic Model of the Phagosome-The kinetic model of the phagosome considers rates of O 2 . generation and disappearance and includes the MPO redox transformations shown in Fig. 1. These and all the other reactions considered are shown in Table 1. Modeling was carried out using the Simulink feature of Matlab (The Math Works Inc, Natick, MA). Phagosome Volume-Setting reactant concentrations requires knowledge of the phagosome volume. Killing of most micro-organisms takes place within minutes of phagocytosis (5,35,36) when degranulation but little phagosomal swelling occurs. For our standard conditions (Table 2), we used a volume of 1.2 ϫ 10 Ϫ15 liters (1.2 m 3 ). This is equivalent to a 0.25-m space between a 1-m diameter bacterium and the phagosomal membrane and includes the volume of the granules that are released. The neutrophil contains ϳ1500 azurophil granules of 0.3 m diameter and twice as many specific granules of half that diameter (31,37,38). Assuming maximum phagocytosis of ϳ20 particles (5) is accompanied by 80% degranulation, granule contents will comprise almost half the phagosome volume. These dimensions will vary somewhat depending on the size and number of micro-organisms ingested but are consistent with electron micrographs of ingested bacteria (31,32,38).
Chloride-The phagosomal chloride concentration has recently been estimated as 73 mM for neutrophils in medium containing 122 mM chloride (40). For initial modeling we have used 100 mM (the same as in extracellular fluid) and then assessed the impact of variations.
Oxygen Consumption Rate and Site of Superoxide Generation-When neutrophils phagocytose particles, oxygen is reduced on the phagosomal membrane and O 2 . is released into the lumen, with very little detected in the medium (41). This continues over a period of minutes during which killing takes place. We have assumed that oxygen consumption occurs at a constant rate, based on the maximum uptake by 10 6 cells being 3-4 nmol/min (5,35). This was observed with 15-20 particles ingested per cell and corresponds to about 3 ϫ 10 Ϫ18 mol of oxygen per phagosome/s or with a phagosome volume of 1.2 ϫ 10 Ϫ15 liters at 2.5 mM/s. This is a net rate, as oxygen is regenerated from the O 2 . and H 2 O 2 is formed. As it is a measured value, it takes into account any limitation in oxygen availability because of diffusion. We have assumed that oxidase activity is not limited by NADPH availability. A more complex analysis would be required to test the effects of rapid oscillations in NADPH concentration as observed by Petty and co-workers (42,43). We have also assumed that charge compensation for the electrogenic effect of O 2 . production is met primarily by proton transport associated with activation of the oxidase, and this provides sufficient protons for O 2 . dismutation and maintenance of pH (22,44). Myeloperoxidase Reactions-When killing occurs, the phagosomal pH is 7.4 -7.8 (45), so we have used rate constants measured at neutral pH. The majority of these are known or can be estimated from literature data ( Table 1). The rate constants for compounds I and II reacting with peroxidase substrates vary depending on the substrate. For most substrates, k 5 is between 10 6 and 10 7 M Ϫ1 s Ϫ1 ; values of k 6 are less and vary more widely (46). Good substrates (high k 6 , e.g. tyrosine) and poor substrates that are potential inhibitors (low k 6 , e.g. tryptophan and nitrite) have been modeled.  Association of MPO with the surface of ingested bacteria has been observed microscopically (47). This could be a way of directing HOCl for effective killing (1). MPO has been shown to bind to the surface of a number of species, and for Actinobacillus actinomycetemcomitans, 4500 high affinity binding sites per cell have been measured (48). A comparable value has been obtained for eosinophil peroxidase binding to Staphylococcus aureus (49). However, this is sufficient to accommodate less than 1% of the estimated 7 ϫ 10 5 molecules of MPO released into a phagosome. Therefore, a small fraction of the MPO products could be localized to the bacterial surface, but even with binding, most of the H 2 O 2 would react with unbound MPO.
Efflux of Oxidants from the Phagosome-The model allows for losses of O 2 . and H 2 O 2 by diffusion into the neutrophil cytoplasm and the ingested bacterium. For H 2 O 2 , the permeability constant (P) was assumed to be the same for the phagosomal and plasma membranes and was determined by measuring the rate at which neutrophils consume exogenous H 2 O 2 . Neutrophils were prepared from the blood of healthy human donors (50) and suspended in 0.14 M NaCl containing 10 mM phosphate buffer, pH 7.4, 1 mM CaCl 2 and MgCl 2 , and 5 mM glucose. H 2 O 2 (5-400 M) was added to 10 5 to 10 6 cells per ml at 37°C, and the concentration remaining was determined at intervals over 10 min using the xylenol orange assay (51

Reaction
Rate constant a Ref.

RESULTS AND DISCUSSION
Modeling Oxidative Reactions in the Phagosome-Each simulation begins with oxygen being reduced to O 2 . at a given rate and with MPO present in its native form. Under the standard conditions defined in Table 2 and with the reactions described in Table 1, O 2 . and H 2 O 2 build up within seconds to steady state concentrations where the rates of production and consumption are the same ( Fig. 2A and Table 3 Fig. S1), rate constants used in the phagosome model gave reasonable agreement with these data.   It is usually assumed that in MPO deficiency, H 2 O 2 accumulates to high concentrations and kills either directly or via secondary products such as hydroxyl radicals produced within the bacterium (14). This is not supported by the model. Greater than millimolar H 2 O 2 concentrations need to be added to kill most bacteria in cell-free systems (59), and bacteria in the phagosome would be exposed to much lower concentrations than this.  If chloride is consumed by MPO in the model and is not replenished, HOCl production rapidly ceases within a minute. Although chloride is regenerated when HOCl oxidizes substrates such as thiols and methionines, it would be depleted if long lived chloramines and other chlorinated products such as chlorotyrosine are formed. It would also decline if any HOCl leaves the phagosome. As discussed below, chloride-depleting reactions are likely to occur and ongoing production of HOCl in the phagosome should therefore require a continuous supply of chloride.
Measurements of chlorination of phagocytosed probes (5, 6) imply that chloride is present in the neutrophil phagosome. Some chloride would enter with the extracellular fluid taken up during phagocytosis and be diluted, probably to about a half, with released granule contents. Granules are packed with proteins anchored to a negatively charged sulfated proteoglycan matrix (62,63), so their chloride content could be low. Recent studies with a chloride-sensitive probe indicated an intraphagosomal chloride concentration of about 60% that of the medium (40), which is consistent with such a dilution. These studies were carried out in the presence of azide and so did not assess the impact of MPO activity. Neutrophil cytoplasm contains 80 mM chloride (64), and agonists such as tumor necrosis factor or phorbol 12-myristate 13-acetate cause rapid efflux through specific chloride channels (65,66). If these channels were activated in the phagosomal membrane in response to particle ingestion, they could provide an ongoing supply of chloride for HOCl production. More information on chloride distribution in the neutrophil is needed to assess whether it limits HOCl production. although giving 99% compound III, still allows more than halfmaximal HOCl production (Table 4). This rate is too low to fit our experimental data (Table 2) or the findings of Wever and co-workers (28), and it is therefore unlikely that turnover of compound III is a major constraint for HOCl production. The model is insensitive to a 3-fold increase or a 10-fold decrease in the rate of reaction of O 2 . with compound II (r 14 ), which is well beyond the measured limits (not shown). The reaction of O 2 . with compound I (r 13 ) occurs in competition with chloride. Decreasing k 13 to the lower limit of measured values has little impact, but at the upper limit, competition starts to decrease the efficiency of HOCl production (Table 4). This becomes more marked at lower chloride concentrations, particularly if the MPO concentration is also low and the O 2 .

Reactions of Hydrogen
concentration consequently higher.
Alternative Myeloperoxidase Substrates-If other MPO substrates are present in the phagosome, they are oxidized and decrease the efficiency of HOCl production to an extent that depends on their reaction rate with compound I (tyrosine ϳ tryptophan Ͻ nitrite Ͻ serotonin; see Table 1) and on their concentration relative to chloride (Fig. 3A). These reactions only impact HOCl production if there is continuous replenish-  (Fig. 3B). Tyrosine and serotonin, which react well with compound II, have little impact. However, nitrite or tryptophan, which react slowly with compound II, give an initial phase of efficient HOCl production, until ϳ98% of the MPO is converted to compound II. Turnover of this form is slow, and HOCl production is strongly curtailed. The initial burst is shorter for nitrite because of its higher k 13 , but tryptophan reacts more slowly with compound II and eventually causes greater inhibition. As shown for nitrite (Fig. 3B) (Table 5). These reactions have been proposed as the source of the small amounts of hydroxyl radicals detected in neutrophils (72) and in a proposed mechanism for ozone generation from singlet oxygen (17). If no proteins or other biological targets for HOCl are included in the phagosome model, hydroxyl radicals are major products. In contrast, extremely little singlet oxygen is generated because of the slower reaction rate with H 2 O 2 and its low concentration. However, the phagosome is packed with granule proteins. Based on the estimated concentrations and rate constants in Table 5, these will be major targets. Modeling shows that cysteines and methionines are almost exclusively lost first, followed by amine groups and disulfides (Fig. 4). Only then are hydroxyl radicals formed. Direct reactions of HOCl with tyrosine or tryptophan residues are less favored. If there is ongoing protein release from either the neutrophil or the ingested bacterium, 1% of the estimated reactive protein groups in Table 5 is sufficient to maintain hydroxyl radical production at Ͻ1%.  Based on the concentrations and reactivities in Table 5, the diffusion distance of HOCl can be calculated 4 as 0.03 m when it can react with protein thiols and methionines and 0.1 m for amines. Both are less than the width of the phagosome. Therefore, HOCl should react close to its site of generation. As already noted, association of MPO with the bacterial surface could direct some HOCl to the bacterium, but most of the HOCl should be generated by unbound MPO throughout the phagosomal space. A majority of this will react with granule proteins before reaching the bacterium. Initial reactions with methionine and cysteine could be considered as scavenging, but further oxidation generates protein chloramines. These have some cytotoxic action, and break down to give aldehydes, ammonia, and probably ammonia chloramine, which are more bactericidal (73,74). MPO-dependent aldehyde formation has been detected in neutrophil phagosomes (75), and as suggested in these early papers, aldehyde and chloramine products may be major effectors of the antimicrobial action of HOCl. The findings of Segal and co-workers (67) that granule proteins at phagosomal concentrations protect bacteria against HOCl cannot be taken as evidence that neutrophils do not use HOCl to kill, as they used only enough HOCl to oxidize a small fraction of the methionine and cysteine residues.
A striking inference from these simulations is that HOCl generated in the phagosome would oxidize susceptible groups on degranulated proteins within a short time. These proteins are required to degrade ingested material, and it is unlikely that oxidation could be sustained without inactivation. Granule enzymes, including MPO, are inactivated by HOCl (76), and the NADPH oxidase contains thiols that could be oxidant-sensitive (77). Oxidant production, therefore, may need to be transient. A possible scenario is that a burst of HOCl production initiates the process of microbial killing and digestion. Production could terminate as a result of intrinsic mechanisms that deactivate the NADPH oxidase complex (78)  accumulation. Some H 2 O 2 will enter the bacterium but directly bactericidal concentrations would not be reached. It is possible for the majority of the oxygen consumed in the phagosome to be converted to HOCl, but the efficiency of this process is sensitive to variations in parameters about which we have limited knowledge. Low chloride, particularly Approximations are based on an estimated total concentration of proteins released into the phagosome by degranulation of 200 g/liter, based on granules consisting of 50% protein and being diluted 2.5-fold. Taking an average mass of 130 kDa, this corresponds to 1.5 M amino acid residues. An average amino acid composition of 8.3% Lys ϩ His, 3% Met and reduced Cys. 9% Tyr ϩ Arg (91) was assumed.

Reaction
k obs M ؊1 s ؊1 Ref.   Tables 1 and  2 with the concentrations of targets and reactions in Table 5.
in combination with a low MPO concentration, decreases HOCl production, and ongoing production requires replenishment of chloride consumed through HOCl escaping or undergoing chlorination reactions. Phagosomal chloride concentrations and transport mechanisms stand out as priorities for further investigation. More direct information on the efficiency of MPO release and the extent to which it becomes inactive is also needed. The model has provided insight into likely reactions of HOCl in the phagosome environment. Some should be directed toward the ingested bacterium by MPO bound to the surface, but calculations based on numbers of binding sites lead to the conclusion that the majority will be generated in the lumen of the phagosome. Most of this HOCl should react with proteins released during degranulation and not reach the bacterium. Although it is possible that these reactions could inhibit microbicidal activity, it is more likely that they contribute via the generation of protein chloramines and secondary products such as aldehydes and ammonia chloramine. Hydroxyl radicals should become significant only when reactive groups (thiols, methionines, amines, and disulfides) are consumed, and singlet oxygen formation is not favorable. Ongoing production of HOCl should cause extensive oxidation and probable inactivation of released granule proteins. This is counterintuitive to the requirement for active degradative enzymes and raises the possibility that O 2 . or HOCl production in each phagosome is short lived and self-limiting. Experiments are now needed to test these predictions directly.