Intracellular Neutrophil Oxidants: From Laboratory Curiosity to Clinical Reality

Reactive oxygen species (ROS) are chemically reactive metabolites of oxygen, and small amounts of such chemical species are produced in all living cells as by-products of aerobic metabolism. Phagocytic cells are equipped with a specialized enzyme system, the phagocyte NADPH oxidase (sometimes referred to as NOX2), that provides these cells with the capacity to produce much higher levels of ROS than those generated by mitochondria. The NADPH oxidase is an electron transport chain that reduces molecular oxygen to superoxide, which in turn serves as a precursor for formation of hydrogen peroxide and other secondary ROS. The induced ROS production because of activation of the NADPH oxidase is accompanied by a marked peak in cellular oxygen consumption, and the process is, therefore, often referred to as a “respiratory burst.” Among the human phagocytes, neutrophils are especially competent to produce ROS.

The basic structure of the phagocyte NADPH oxidase has been uncovered primarily through studies of cells from patients with chronic granulomatous disease (CGD; described in more detail below), which is a rare, inherited condition characterized by the inability of phagocytes to produce ROS because of a dysfunctional NADPH oxidase (1). The NADPH oxidase is a multicomponent enzyme consisting of a membrane-bound, heterodimeric b-type cytochrome (cytb₅₅₈) comprising the subunits gp91^phox and p22^phox, together with several cytoplasmic subunits (p47^phox, p40^phox, and p67^phox) (2). Upon activation, the cytosolic components translocate to the membrane and associate with the cytb₅₅₈, together with a rac GTPase (Fig. 1).

When the NADPH oxidase is active, cytoplasmic NADPH (generated by the hexose monophosphate shunt) provides electrons to the cytb₅₅₈, which transports the electrons across the membrane to molecular oxygen on the opposite (noncytoplasmic) side. The process is dependent on the FAD unit of gp91^phox. The electrons reduce oxygen to superoxide anion that spontaneously dismutates to hydrogen peroxide. These two primary ROS can then be further transformed into a wide variety of secondary ROS, (i.e., through reactions catalyzed by the neutrophil enzyme myeloperoxidase [MPO]) (3).

Neutrophils are microbe killers, and their antimicrobial components are internally organized in different subcellular organelles, or granules. Part of the neutrophil cytb₅₅₈ is localized in the plasma membrane, but the main portion is present in granule membranes, most notably those of the specific granules (4). Depending on which pool of the cytb₅₅₈ that is engaged in formation of the active NADPH oxidase, the generated superoxide can either be secreted extracellularly from activation in the plasma membrane or produced inside membrane-enclosed intracellular compartments, tentatively by activation of specific granule-associated NADPH oxidase (5). The predicted site for intracellular ROS production would typically be the phagolysosome, formed after particle engulfment and fusion of specific granules (containing cytb₅₅₈) with the phagosomal membrane (Fig. 1). True enough, neutrophils do produce massive amounts of ROS during phagocytosis of different microorganisms (6) at sites that are out of reach to large, membrane-impermeable ROS scavengers (e.g., superoxide dismutase [SOD]). In other words, the ROS are formed intracellularly (7, 8).

Intracellular ROS generation was long thought to be linked, by necessity, to phagocytosis. It is clear, however, that intracellular ROS can also be formed in the absence of phagosome formation [i.e., after stimulation of cells with the phorbol ester PMA (5, 9, 10) or with lectins, such as galectins or wheat germ agglutinin (11–13)]. The exact identity of the intracellular compartment where such ROS are produced is not yet known, but an organelle formed by heterotypic fusion of specific granules (providing the cytb₅₅₈) and azurophil granules (providing MPO) has been suggested (14, 15) (Fig. 1). It is also clear that NADPH oxidase activation in intracellular membranes relies on slightly different signaling pathways, as compared with activation in the plasma membrane, suggesting that the NADPH oxidase in different membranes may be regulated by different cues (5, 9). Phagocyte ROS production is indeed crucial for our well-being, and, in all likelihood, the localization of ROS production will determine their functionality. Regarding the intracellular, nonphagosomal ROS, it seems reasonable that considerable ROS production in a cellular compartment (fused granules) devoid of any microbe to kill could substantially affect and direct cell signaling and function.

Various types of stimuli can induce activation of the neutrophil NADPH oxidase, often resulting in production of ROS at distinct sites. For instance, high doses of neutrophil chemoattractants (such as formylated peptides) typically activate the plasma membrane-localized NADPH oxidase, which results in extracellular ROS release (16). There are a number of soluble stimuli that give rise to intracellular ROS production in combination with various levels of extracellular ROS release (5). PMA is, for example, a widely used soluble stimulus that directly activates protein kinase C, resulting in potent induction of ROS generation extracellularly as well as intracellularly, in nonphagosomal compartments (5, 9, 10). One group of soluble neutrophil activators of special interest in this regard is the galectins, endogenous lectins with affinity for lactosamine-containing glycoproteins. Galectins are present in a variety of inflammatory exudates and have many different effects on immune regulation (17, 18). At least three human galectins, galectin-1, -3, and -8, trigger distinct activation of the intracellular NADPH oxidase in addition to inducing extracellular ROS release. Interestingly, galectins only activate the NADPH oxidase if the cells have been primed prior to stimulation by, for example, in vivo transmigration (11, 13, 19).

With regards to the intracellular ROS triggered during phagocytosis, it is likely that the main part stems from the phagolysosomal compartment (see above), but experimental data show that other ROS-producing compartments are also present in cells exposed to particulate stimuli (20, 21). As will be discussed below, methodological shortcomings have prevented detailed analysis of the localization of ROS production during phagocytosis, suggesting that there is more to learn.

CGD is a rare immunodeficiency associated with bacterial and fungal infections, originally called “fatal granulomatous disease of childhood” when described some 60 y ago (22). Following the identification and study of more patients, the disease was shown to be associated with phagocytes that are unable to mount a respiratory burst and that are defective in microbial killing, thus explaining the predisposition to infection. As antibiotic regimes became more effective and CGD patients survived into adulthood, it was obvious that they were not only susceptible to infections but also suffered from a variety of inflammatory complications indicative of a malfunction in the mechanisms that control inflammation. It should be stressed that the clinical manifestations of CGD are quite varied (even when the causative mutations are identical), as both genetic and environmental factors likely influence the progression and severity of clinical symptoms (23, 24).

In classic CGD, neutrophils lack NADPH oxidase activity because of mutations in the genes encoding gp91^phox (the majority of cases), p47^phox, p67^phox, or p22^phox. Mutations in the gene for the cytosolic subunit p40^phox have only recently been characterized and represent a distinct (nonclassic) subtype of CGD (see below). In vitro, neutrophils from patients with classic CGD display reduced microbial killing, resulting in recurrent and severe infections with certain bacterial and fungal species (1). Although phagocytes from many patients with classic CGD are entirely unable to form ROS, residual ROS production may occur for certain mutations, and as little as 1% of normal ROS production seems to confer milder disease and increased survival (25). To our knowledge, all classic CGD mutations affect the plasma membrane- and granule-localized NADPH oxidase equally, meaning that the ROS deficiency is manifested both extracellularly and intracellularly.

As mentioned above, increased susceptibility to infection is not the only clinical manifestation of CGD. Patients with classic CGD are also frequently plagued by various inflammatory disorders [e.g., granuloma formation, inflammatory bowel disease, and lupus-like symptoms (26–28)], comprising clear indications of dysregulated inflammatory reactions. This suggests that the ability to properly regulate inflammatory signaling is somehow dependent on NADPH oxidase-derived ROS, which will be discussed below.

Among the structural NADPH oxidase subunits, p40^phox was for a long time never found among the mutated genes causing CGD. In 2009, a first patient was described with a genetic deficiency in this cytoplasmic subunit (29). Interestingly, this patient had no history of severe infections, but, instead, presented with granulomatous (Crohn) colitis, a condition that is indicative of an inability to limit inflammatory reactions. Quite recently, a description of 24 other p40^phox-deficient patients from 12 families was presented by van de Geer and coworkers (30). Also, in these patients there was a strong association with hyperinflammation and significantly less severe (invasive) infections than in patients with classic CGD (30).

A most remarkable finding from patients with p40^phox deficiency is that their neutrophils displayed a selective inability to generate intracellular ROS, whereas the ability to produce extracellular ROS was intact (29, 30). This cellular phenotype is markedly distinct from classic CGD but correlates well with experimental data from mice and cell lines, showing that p40^phox is of minor importance for extracellular ROS production but critical for phagocytosis-induced (intracellular) ROS production (31, 32). Collectively, these findings imply that p40^phox is of special importance for the NADPH oxidase when activated in internal membranes (Fig. 1). Clearly, p40^phox deficiency stands out from deficiencies in the other oxidase subunits. Compared with classic CGD, p40^phox deficiency is associated with a distinctly different neutrophil behavior in vitro and a pathologic condition seemingly characterized by less severe infections but just as pronounced inflammatory symptoms.

Apart from mutations in genes for the structural components of the NADPH oxidase, deficiencies in certain auxiliary components may also render phagocytes incapable of mounting an oxidative burst. One very recent and exciting report from Iceland describes CGD-like pathologic conditions with particularly pronounced inflammatory problems in the gastrointestinal tract in eight patients, caused by a homozygous loss-of-function mutation in a gene named CYBC1 (for cytochrome b chaperone 1) (33). Almost simultaneously, another report described one Saudi Arabian patient with a different mutation in the same gene (34). Concurring with the Icelandic patients, this patient also had a clinical history that is compatible with CGD.

Not much is known about the protein CYBC1, also called EROS [essential for ROS (34)], except that it is a transmembrane protein which appears to function as a chaperone that facilitates assembly and correct folding of gp91^phox and p22^phox into a functional, heterodimeric cytb₅₅₈ (33). Because gp91^phox and p22^phox in their monomeric forms are known to be degraded quite rapidly by the cytosolic proteasome (35), lack of CYBC1 should lead to decreased expression of cytb₅₅₈ and, thus, impaired ROS production. This has been demonstrated to be the case in murine phagocytes (36) and was also corroborated in the two human studies (33, 34).

Interestingly, the relative importance of CYBC1 for cytb₅₅₈ expression appears to be larger in monocytes/macrophages than in neutrophils (33, 36), which could imply that cells of the neutrophil lineage express additional chaperone proteins capable of partially backing up CYBC1. The infectious profile of the described patients (33, 34) seems slightly unusual as compared with classic CGD, with a possible inclination for intracellular pathogens (Mycobacterium tuberculosis/BCG and Listeria). CYBC1/EROS–deficient mice were also shown to display increased sensitivity to intracellular infections (36), and it is tempting to speculate that this reflects the central position of macrophages for controlling intracellular infections in particular. Whether CYBC1-deficiency could be viewed as a cell type-specific CGD (with complete penetrance in monocytes/macrophages and partial penetrance in neutrophils) is presently not clear, and, to our knowledge, nothing is known regarding where in neutrophils the remaining cytb₅₅₈ is actually expressed and whether ROS production at different sites are equally affected.

As concluded by the pathological features of classic CGD, NADPH oxidase-derived ROS production is clearly central for our ability to fight infections, but it is also essential to control inflammatory reactions. By comparing the newly characterized p40^phox CGD, in which neutrophils specifically lack the ability to form intracellular ROS, with classic CGD, in which ROS production is lacking extracellularly as well as intracellularly, a picture is emerging that implies that specific roles are played by neutrophil ROS generated in intracellular compartments.

Phagocytic uptake of microbes constitutes an important step in immune eradication of invading pathogens, and a number of microbes have evolved strategies to resist uptake or, alternatively, allowing phagocytic uptake but resisting the bactericidal arsenal of the phagosome, a characteristic of, for example, the intracellular pathogens Fracisella tularensis (37) and M. tuberculosis (38, 39). The logical link between phagocyte ROS production and microbial killing was established when it was shown that human phagocytes generate superoxide (40) and that such ROS production was lacking in phagocytes isolated from CGD patients (41, 42). Also, it was soon established that a significant portion of the ROS generated during phagocytosis was out of reach for extracellular scavengers, hence intracellular in location (7, 8), suggesting that they are produced inside the phagosome and contribute to killing at this site. The primary ROS produced by the NADPH oxidase, superoxide, is not very bactericidal per se, but its killing effect can be markedly increased by its transformation into hydrogen peroxide, especially in the presence of MPO and halides (43). These findings gave rise to the residing view that what is mostly responsible for the killing of microbes inside the phagosome is hypochlorous acid (44). This acid is formed by an MPO-catalyzed reaction between hydrogen peroxide and chloride, and it is very reactive and quite capable of killing microbes (45).

However, a number of observations challenge the view that intracellular hypochlorous acid is critical for microbial killing in the phagosome. For example, MPO-deficient individuals are less sensitive to infections than would be expected (46), at least much less than patients with classic CGD. Additionally, the MPO/hydrogen peroxide/halide system has been shown to affect host cell components rather than the ingested bacteria per se (47, 48). This confusion aside, if intracellular ROS production (with or without accompanying MPO processing) is indeed critical for phagosomal killing, the expectation would be that phagocytes lacking such ROS would display diminished killing activity. However, the p40^phox deficiency described above, in which neutrophils specifically lack the ability to form ROS in intracellular compartments (including the phagosome), seems noticeably less associated with severe infection than classic CGD (29, 30). In addition, in vitro killing seems markedly more effective in p40^phox-deficient neutrophils as compared with those from classic CGD patients (30).

One possible way to reconcile conflicting reports on the importance of intracellular ROS as microbe killers is to consider the idea that different microbes may be differently sensitive to specific ROS at specific sites. It could be that extracellular ROS are more important for bacterial killing than previously thought, which has, for example, been demonstrated for Escherichia coli strains resistant to phagocyte engulfment (49). It has also been suggested that the actual ROS are not at all responsible for bacterial killing in the phagosome, but that the primary role of the NADPH oxidase is, instead, to provide optimal ionic conditions and pH inside the phagosome for the neutrophil serine proteases (e.g., cathepsin G and elastase) to kill the microbes (50). Accordingly, the electron transport and the subsequent membrane polarization that control a charge compensatory flux of protons and other ions would be the most important function of the NADPH oxidase, and the actual killing would primarily be mediated by the serine proteases (51). Opposing this model, patients with Papillon–Lefèvre syndrome, the neutrophils of whom completely lack serine protease activity, are not particularly immunocompromised; this disease is not nearly as associated with infections as CGD (52). However, even if the electrogenic action of the NADPH oxidase cannot fully explain microbial killing in the phagosome, it may be that ion fluxes and pH regulation are of importance for other processes that are abnormal in CGD.

Clearly there is still much to learn regarding phagosomal killing and intracellular NADPH oxidase activity, and a detailed, mechanistical explanation for how ROS kill microbes inside phagosomes is still a rather open question (53–55).

In a seminal paper from 2004, neutrophils were shown to undergo a spectacular form of cell death to capture and eliminate microbes by casting out neutrophil extracellular traps (NETs) (56). This violent cell death, subsequently called NETosis, is distinct from necrosis and has in recent years attracted great interest (57–60). The NETs are web-like structures that consist of a backbone of nuclear material (DNA and histones) decorated with proteins originating from the neutrophil granules, such as MPO and elastase (56, 61). Although NETs were initially described as an antibacterial defense strategy, their regulatory potential in inflammation has become evident in a number of studies; diseases such as systemic lupus erythematosus, psoriasis, small vessel vasculitis, rheumatoid arthritis, gout, venous thrombosis, and cardiovascular disease all show the presence of NETs in vivo (62–64).

The best characterized example of NETosis describes a coordinated series of cellular events in which ROS production is a crucial step (65). Such ROS-dependent NETosis, requiring a functional NADPH oxidase, occurs as a result of interactions with a variety of microbes (61, 66–68). In vitro, the standard procedure to trigger NETosis by this ROS-dependent pathway is to expose cells to PMA. We recently showed that the ROS species crucial for PMA-triggered NETosis are those generated intracellularly in nonphagosomal compartments and that these ROS need to have been processed internally by MPO to drive NET formation (69). Because MPO and cytb₅₅₈ are localized in different neutrophil granules (azurophil and specific granules, respectively) a likely scenario is that ROS and MPO become colocalized through heterotypic granule fusion (Fig. 1).

Our experimental findings recently gained support through the characterization of the p40^phox-deficient CGD (see above) (30). Similar to classic CGD neutrophils, neutrophils lacking p40^phox are unable to form NETs in response to PMA (30) despite the fact that these cells produce rather normal levels of extracellular ROS (29, 30). This finding fits perfectly with the idea that intracellular ROS production following PMA stimulation is truly central for the NETosis process (69). It is difficult to discern the exact identity of the intracellular ROS that affect NETs release by available techniques (described below and in the Tables). However, there is strong evidence that the NETs-triggering ROS species have to be processed by MPO intracellularly because MPO-deficient individuals are unable to form NETs (70) and only membrane-permeable MPO inhibitors and scavengers are capable of blocking NETosis (69).

In general, ROS have the potential to react with a wide variety of biomolecules and take part in cell signaling. Initially, NADPH oxidase-derived ROS were argued to be essential for, for example, NF-κB activation leading up to the production of multiple proinflammatory cytokines. These conclusions were, however, largely based on experiments conducted with (often notoriously unspecific) NADPH oxidase inhibitors. In contrast, in vitro experiments with primary classic CGD phagocytes established that not only were these cells fully capable of activating NF-κB (71) or the NLRP3 inflammasome (72, 73), but they also produced significantly elevated levels of cytokines spontaneously as well as in response to stimulation (71, 72, 74–78). Exaggerated cytokine production has been confirmed in a cell line lacking gp91^phox (74, 77), which implies that the inflammatory complications in CGD are not because of accompanying (subclinical) infections but, rather, represent an inherent hyper- (or even auto-) inflammatory phenotype. This fits perfectly well with the fact that classic CGD patients typically suffer from a wide variety of hyperinflammatory symptoms.

Pronounced hyperinflammatory pathologic condition is also seen in the p40^phox-deficient CGD patients (29, 30), which suggests that the ROS that regulate inflammatory signaling are primarily the ones generated at intracellular sites. To our knowledge, whether p40^phox-deficient phagocytes produce elevated levels of proinflammatory cytokines in vitro, similar to primary phagocytes from patients with classic CGD, has not been directly tested. Given the clinical features of patients with p40^phox deficiency (29, 30), it is likely that p40^phox-deficient phagocytes are equally hyperinflammatory as classic CGD phagocytes.

Because redox regulation of signaling pathways with a definite role of ROS as drivers of cytokine production is well established, the source of cytokine-inducing ROS is hardly the NADPH oxidase. If they were, nonphagocytic cells (or CGD phagocytes) would not be able to produce any proinflammatory cytokines at all. Instead, other cellular ROS sources, such as mitochondrial ROS, are more likely to promote NF-κB and inflammasome activation. We recently showed that classic CGD phagocytes (as well as a gp91^phox-deficient cell line) displayed increased mitochondrial ROS (77), which could potentially reconcile the observations that ROS are critical for inflammatory signaling, yet CGD is a hyperinflammatory disease. To bring clarity to how (and where) NADPH oxidase-derived ROS modulate cytokine production and, thus, hyperinflammation, more emphasis needs to be given to identification of the source and localization of phagocyte ROS.

Over the years, a number of techniques have been developed to monitor phagocyte superoxide (Table I), hydrogen peroxide (Table II), or just undefined ROS production (Table III), all of which have both advantages and limitations (the Tables and (79) for a review). In the context of intracellular ROS production, some aspects of commonly used techniques are important to mention. The probe DHR-123 (80), used, for example, in the studies of p40^phox-deficient neutrophils (29, 30) and the CYBC1 deficiency (33), is a membrane-permeable dye that gains in fluorescence (as measured by flow cytometry) when oxidized by ROS. Thus, in theory, DHR-123 mainly detects intracellular ROS. However, extracellular hydrogen peroxide produced by the plasma membrane-localized NADPH oxidase may diffuse across membranes to react with DHR-123 intracellularly (81). Also, ROS from other cellular sources, such as mitochondria (77), may contribute to DHR-123 fluorescence. There are, accordingly, risks for overestimation of NADPH oxidase-derived intracellular ROS using this probe. Moreover, oxidation of DHR-123 is, in part, dependent on the granule enzyme MPO (80), but the extent of the MPO dependency of this dye is, to our knowledge, not entirely clarified. Thus, it is clear that in order for the signals generated by dyes like DHR-123 to reflect intracellular NADPH oxidase-derived ROS, membrane-impermeable ROS scavengers should be added and MPO inhibitors (such as azide) should be avoided in the experimental set-up. Another rhodamine-based probe (R19-S) was recently introduced to directly measure the production of MPO-derived hypochlorous acid, and the dye has been used to determine phagosomal ROS activity (82). It should be noticed, however, that most of the fluorescence signal detected during phagocytosis of opsonized zymosan was recovered in cell supernatants, indicating that it is the result of extracellular reactions (82), possibly because of limited membrane permeability of the probe.

To determine the subcellular localization of neutrophil ROS production, luminol/isoluminol–based techniques are, sometimes, better alternatives than DHR-123. These techniques are based on the excitation of luminol or isoluminol by superoxide anion, generating detectable chemiluminescence (CL). Luminol, the most commonly used CL probe, is cell permeable and can be used to measure intracellular ROS. However, to achieve selectivity, it is crucial to add membrane-impermeable ROS scavengers (such as SOD and catalase) to neutralize extracellular ROS (79, 83). The related CL probe isoluminol differs from luminol only in the position of the amino group on the phthalate ring of the molecule, making it membrane impermeable. Isoluminol can thus be used to selectively determine extracellularly released ROS (79, 83). Both luminol and isoluminol are completely dependent on the processing of ROS by a peroxidase; the activity of endogenous MPO is absolutely critical for the detection of intracellular ROS (69), whereas the isoluminol-amplified system should contain an exogenous peroxidase (preferentially HRP) in order for the CL signal not to be critically dependent on released cellular MPO.

For detailed analyses of neutrophil ROS production, it is recommendable to use complementary methods, taking into account their limitations and paying close attention to methodological details. The future development of novel techniques and reagents along the lines outlined in Erard et al. (84) would, however, be a great contribution to the field.

The research community is increasingly willing to accept the idea that part of the ROS generated during the neutrophil respiratory burst is in fact produced and retained intracellularly and often generated independently of phagocytosis and phagolysosome formation. In light of the clinical phenotype of the patients with p40^phox CGD and the experimental findings using their neutrophils, we foresee an increased future interest in these particular ROS. Judging by the clinical picture of the patients with p40^phox CGD (highly inflamed but largely devoid of severe invasive infections) as compared with that of patients with classic CGD, it seems that intracellular ROS are specifically important for controlling inflammatory reactions but, surprisingly, not particularly important for microbe killing. The contribution to phagolysosomal killing of a p40^phox-independent, plasma membrane-localized NADPH oxidase is, from this perspective, an interesting notion well worth examining.

It is clear that granule-localized ROS are vital for the expulsion of NETs in vitro, but how and exactly where these ROS drive NETosis is still not established. If and how ROS-dependent NET formation truly contributes to health and disease is a matter of some controversy (85). However, the fact that neutrophils treated with cell-permeable MPO inhibitors or scavengers of intracellular, MPO-processed ROS failed to generate NETs (69), as did neutrophils from patients with p40^phox CGD (30), inspires further exploration of this question in animal models.

We look forward to the development of methodologies to measure intracellular NADPH oxidase-derived ROS independently of MPO. This would allow for more detailed characterization of the subcellular compartment(s) in which intracellular ROS are produced and, hopefully, facilitate the unraveling of mechanism(s) whereby they mediate their cellular effects and regulate inflammation. Because of the methodological difficulties in measuring intragranular ROS today (especially in a clinical setting), it is possible that specific defects in the generation of intragranular ROS may be under-detected among hyper- or autoinflammatory disorders. It will be truly exciting to follow this field of investigation in the years ahead.

We thank all past and present members of the Phagocyte Research Group at the Sahlgrenska Academy for valuable discussions about neutrophil physiology in general and ROS production in particular.

The authors have no financial conflicts of interest.