Phagocytosis and persistence of Helicobacter pylori

Authors


*E-mail lee-ann-allen@uiowa.edu; Tel. (+1) 319 335 4258; Fax (+1) 319 335 4194.

Summary

Helicobacter pylori is a spiral-shaped, flagellated, microaerophilic Gram-negative bacterium that colonizes the gastric epithelium of humans. All persons infected with H. pylori have gastritis, and some will develop severe disease such as peptic ulcers or gastric cancer. A characteristic feature of this infection is the pronounced accumulation of phagocytes, particularly neutrophils, in the gastric mucosa. H. pylori thrives in a phagocyte-rich environment, and we describe here how this organism uses an array of novel virulence factors to manipulate chemotaxis, phagocytosis, membrane trafficking and the respiratory burst as a means to evade elimination by the innate immune response.

Colonization of the gastric epithelium

Helicobacter pylori is the only microbe known to thrive in the hostile environment of the human stomach (Montecucco and Rappuoli, 2001). Ammonia generated by H. pylori urease buffers bacteria as they pass through the highly acidic gastric lumen (Yoshiyama and Nakazawa, 2000). Thereafter, H. pylori establishes residence in the mucus layer over the epithelium and motility in this locale is enhanced by the spiral shape of the organism and multiple polar flagella (Yoshiyama and Nakazawa, 2000). Organisms in the mucus layer secrete a ‘vacuolating cytotoxin’ called VacA (Cover and Blanke, 2005). Released VacA oligomerizes, is activated by exposure to low pH and binds to the epithelium (Cover and Blanke, 2005). Intoxicated cells accumulate large vacuoles derived from late endosomal compartments (Cover and Blanke, 2005). In addition, VacA and ammonia disrupt epithelial barrier function and a fraction of VacA reaches the host cell cytosol where it associates with mitochondria and promotes apoptosis (Cover and Blanke, 2005). Although all strains of H. pylori contain the vacA gene, not all organisms produce an active toxin (Xiang et al., 1995). The s1m1 vacA allele encodes a potent cytotoxin that causes ulceration and necrosis in mice. In contrast, s2m2 VacA is synthesized and secreted but is inactive in vivo and in vitro. A subset of H. pylori strains also contain the cag pathogenicity island which encodes a type IV secretion system (Covacci et al., 1999). Tight binding of cag-positive H. pylori to the epithelium activates the type IV secretion apparatus which delivers the effector protein CagA into the host cell cytosol. Simultaneously, a signalling pathway is activated that results in the synthesis and secretion of the potent neutrophil chemoattractant interleukin-8 (IL-8). IL-8, urease and other bacterial factors such as a cecropin-like peptide called Hp(2-20) and the H. pylori neutrophil-activating protein (HP-NAP) act in concert to recruit phagocytes to the stomach (Allen, 2000; Bylund et al., 2001; Montecucco and Rappuoli, 2001). Neutrophils traverse the epithelium and enter the mucus layer in large numbers (Allen, 2000). In contrast, macrophages encounter bacteria in regions of tissue damage and ulceration (Allen, 2000).

Both VacA and the cag pathogenicity island are central to the ability of H. pylori to cause severe disease (Blaser and Atherton, 2004). Strains lacking s1m1 vacA or the cag pathogenicity island rarely induce ulcers and the cag pathogenicity island is also essential for tumour development. Consequently, strains of H. pylori are sometimes divided into two groups. Type I (cag+/s1m1 vacA) strains cause ulceration and cancer, and type II (cag–/s2m2 vacA) organisms are common in persons with asymptomatic gastritis. Importantly, type II H. pylori induce less inflammation and tissue damage than type I bacteria and it has been suggested that these organisms occupy distinct niches in the gastric mucosa (Blaser and Atherton, 2004).

Phagocytosis

Helicobacter pylori is not efficiently controlled by phagocytes in vivo or in vitro, and substantive killing appears to occur only when neutrophils (polymorphonuclear leukocytes, PMN) are present in vast excess (Pruul et al., 1987; Andersen et al., 1993; Rautelin et al., 1994a; Allen, 2000). A large body of data indicates that H. pylori exhibits a novel mechanism of virulence that allow these organism to evade opsonization in vivo, actively retard phagocytosis and disrupt membrane trafficking and phagosome maturation (Fig. 1).

Figure 1.

Helicobacter pylori is ingested by a novel pathway and persists inside megasomes.
Left. Type I H. pylori bind to macrophages using SabA and other adhesins; tight binding also requires the type IV secretion system (T4SS). Bacteria are ingested by a slow phagocytic pathway regulated by PI3K and PKCζ. Nascent phagosomes accumulate and retain coronin and early endosome antigen 1 (EEA1). Urease-derived ammonia and VacA elevate phagosome pH, stimulate megasome formation and inhibit phagosome–lysosome fusion. Bacteria ingested by this mechanism are not eliminated.
Right. Control organisms are rapidly ingested. Forming phagosomes accumulate and retain PKCα. Bacteria traverse the endosomal pathway and are killed and degraded in mature phagolysosomes.

Adhesins

Phagocytosis is specialized form of receptor-mediated endocytosis that allows macrophages and neutrophils to engulf microorganisms. Receptors that mediate phagocytosis can bind directly to microbe surfaces or indirectly via serum opsonins or extracellular matrix proteins. In general, particle uptake begins within 30 s of receptor engagement and multiple signalling pathways co-ordinate rearrangements of the cytoskeleton, membrane expansion and phagosome closure (Aderem and Underhill, 1999; Underhill and Ozinsky, 2002). By definition, all types of phagocytosis require local actin polymerization. However, multiple mechanisms of phagocytosis exist, and it is well established that mechanism of entry can alter microbe fate (Aderem and Underhill, 1999; Underhill and Ozinsky, 2002). In this regard it is noteworthy that H. pylori uses lectin-mediated phagocytosis to enter both macrophages and PMN (Kist et al., 1993; Rautelin et al., 1994a; Chmiela et al., 1997; Allen, 2000; Teneberg et al., 2000) and is not opsonized in vivo (Darwin et al., 1996; Berstad et al., 1997).

Inefficient opsonization is not due to an absence of specific antibodies or complement factors in the gastric mucosa. Rather, low pH, mucins and other aspects of the local environment prevent antibody binding in vivo, an effect that is reversed by in vitro cultivation of the organism (Darwin et al., 1996; Berstad et al., 1997). In addition, urease retards deposition of C3 on H. pylori both in vivo and in vitro (Rokita et al., 1998), and for this reason relatively high concentrations of serum are required to opsonized H. pylori in the laboratory setting (Rautelin et al., 1993; Gonzalez-Valencia et al., 1996; Rokita et al., 1998; Berstad et al., 2001).

Receptors that mediate phagocytosis of H. pylori remain obscure. Nevertheless, it is clear that H. pylori attachment to macrophages and neutrophils is multifactorial and is mediated by a combination of sialic acid-dependent and sialic acid-independent adhesins as well as heparan sulfate-binding proteins (Chmiela et al., 1994; 1995a,b; 1997; Miller-Podraza et al., 1999). To date, three adhesins that confer binding to phagocyte glycoconjugates have been defined: HpaA which interacts with lactosylceramines; SabA which binds to sialyl-Lewis X (sLeX) dimers, and HP071 which binds to as yet uncharacterized sialic acid-containing glycoproteins (Mahdavi et al., 2002; Bennett and Roberts, 2005; Fantini et al., 2006). Defining which of these interactions is essential for phagocytosis is an area of active investigation. In this regard it is noteworthy that SabA, together with another adhesin called BabA, mediates H. pylori binding to gastric epithelial cells (Ilver et al., 1998; Mahdavi et al., 2002). BabA has no known role in bacteria–phagocyte interactions; however, SabA may be important for infection of neutrophils (Unemo et al., 2005). Moreover, phagocytosis of H. pylori does not require the macrophage mannose receptor or the β-glucan receptor; thus, H. pylori differs significantly from pathogens such as Mycobacterium tuberculosis, Leishmania donovani, Francisella tularensis and Pneumocytis carinii which use these lectins to gain entry into human and murine macrophages (Ofek et al., 1995; Steele et al., 2003; Schulert and Allen, 2006). In contrast to Neisseria gonorrhoeae (Gray-Owen et al., 1997), there is no evidence that carcinoembryonic antigens mediate H. pylori infection of neutrophils.

Phagocytosis may also be modulated by extracellular matrix proteins. H. pylori binds laminin, vitronectin, type IV collagen, heparan sulfate and hyaluronic acid (Ljungh et al., 1996). Both LPS and a SabA have been implicated in H. pylori attachment to laminin (Valkonen et al., 1994; Walz et al., 2005). How the organism interacts with other extracellular matrix components is unknown, but incubation of H. pylori with vitronectin in the presence of serum impairs attachment to phagocytes (Chmiela et al., 1997) whereas, heparan sulfate-sensitive adhesins enhance binding of some strains to macrophages (Chmiela et al., 1996).

In summary, H. pylori binding to macrophages and neutrophils is a complex, multifactorial process. How candidate adhesins function alone or in concert with other bacterial surface components remains to be determined. Ultimately, it will be important to define distinct binding patterns and to dissect how different mechanisms of attachment modulate bacterial engulfment, cell activation and resistance to intracellular killing.

Delayed phagocytosis

Helicobacter pylori uses at least two distinct mechanisms to gain entry into macrophages and PMN. Type II organisms are rapidly and efficiently ingested, and in this manner resemble model particles and other microbes (Allen et al., 2000). In marked contrast, type I H. pylori actively retards its own uptake, and there is a lag of several minutes between bacterial binding and the onset of cytoskeletal rearrangements (Allen et al., 2000). Delayed phagocytosis is observed only with live, metabolically active organisms and is prevented by opsonization with specific IgG, blockade of bacterial protein synthesis or heat denaturation of bacterial proteins (Allen et al., 2000; Ramarao et al., 2000; Ramarao and Meyer, 2001; Allen and Allgood, 2002). As noted above, a distinguishing feature of type I H. pylori is the type IV secretion system. Disruption of genes that encode components of the type IV secretion apparatus or deletion of the entire cag pathogenicity island enhances bacterial engulfment; however, the secreted effector protein CagA is not required, and other virulence factors such as VacA and urease are also dispensable (Ramarao et al., 2000; Ramarao and Meyer, 2001). These data may indicate that the type IV secretion apparatus itself acts as an adhesin. On the other hand, effectors of the type III secretion systems of Salmonella and Shigella and enteropathogenic Escherichia coli regulate the actin cytoskeleton (Mota and Cornelis, 2005), and as such it is tempting to speculate that as yet unidentified effectors of the H. pylori type IV secretion system may play a similar role during phagocytosis of this organism.

On the host cell side, slow phagocytosis of type I H. pylori is mediated by a novel signalling cascade that is defined by atypical protein kinase C-ζ (PKCζ) (Allen and Allgood, 2002). Atypical PKC isoforms lack structural motifs that confer regulation by diacyglycerol or phorbol esters and are activated by the lipid products of class IA phosphoinositide 3-kinases (PI3K). Thus, sequential activation of PI3K and PKCζH. pylori regulates local actin polymerization, and uptake by this mechanism is essential for bacterial survival (Allen and Allgood, 2002; Allen et al., 2005a). PI3Ks are also required for phagocytosis of large IgG-coated particles (Cox et al., 1999; Araki et al., 2003), but in this case signalling downstream of PI3K regulates the local exocytic events required for pseudopod extension and phagosome closure without affecting PKC or the actin cytoskeleton (Cox et al., 1999; Raeder et al., 1999; Araki et al., 2003). Rapid uptake of type II H. pylori occurs without activation of PI3K (Allen et al., 2005a), and although the pathway that confers rapid internalization of these organisms is largely undefined, it is clear that engulfed type II organisms do not survive inside phagocytes for prolonged periods of time (Allen et al., 2000).

Another distinguishing feature of H. pylori is the unusual lipid composition of the bacterial envelope which is noted for its high concentration of lysophospholipids and cholesteryl glucosides (Tannaes and Bukholm, 2005). H. pylori PldA is a highly active outer membrane phospholipase A which catalyses the conversion of phosphatidylethanolamine into lyso-phosphatidylethanolamine. Phospholipase activity is stimulated by low pH, and lyso-phosphatidylethanolamine levels can exceed 50% of total phospholipid (Tannaes et al., 2001). In addition, H. pylori acquires cholesterol from growth media or directly from the membranes of host cells and converts it into several different cholesteryl-α-glucosides (Tannaes and Bukholm, 2005; Wunder et al., 2006). Some of these cholesterol derivatives are present in all H. pylori strains whereas others accumulate only in organisms with high PldA activity. Coordinate regulation of membrane lipid composition likely controls the stability and fluidity of the envelope (Tannaes and Bukholm, 2005). Furthermore, recent data indicate that either pre-loading bacteria with cholesterol or disrupting the glycosyltransferase HP0421 markedly enhances H. pylori uptake by serum-starved J774 cells, an effect that is reversed by expression of HP0421 in trans (Wunder et al., 2006). Clearly, additional studies are needed to determine more precisely how bacterial lipids impact host cell responses and H. pylori fate and to discern the extent to which changes in lipid composition impact expression of adhesins or the activity of other virulence factors such as the type IV secretion system. This issue is also of interest as Mycobacteria, Leishmania and many other pathogens are known to infect cells via cholesterol-rich plasma membrane microdomains (Manes et al., 2003). In contrast, relatively few microorganisms accumulate sterols or their derivatives, and cholesteryl-α-glucosides may be specific to organisms of the Helicobacter genus (Haque et al., 1995).

Collectively, the data suggest that phagocytosis of H. pylori is controlled in large part by bacterial lipids and adhesins as well as host cell signalling pathways. Although much remains to be determined, it is clear that mode of entry has a dramatic effect on bacterial fate as delayed phagocytosis is linked to survival of type I organisms but rapid uptake of opsonized bacteria or type II strains is not (Allen et al., 2000; 2005b; Allen and Allgood, 2002).

Megasome formation

Another distinguishing feature of type I H. pylori infection of macrophages is the ability of these organisms to stimulate homotypic phagosome fusion (Allen, 2000; Allen et al., 2000; Zheng and Jones, 2003) and recent data suggest that this also occurs in H. pylori-infected neutrophils (L. Allen, unpubl. data). Immediately after uptake, bacteria reside in conventional phagosomes. Shortly thereafter, these organelles coalesce, and H. pylori persist inside large ‘megasomes’ for at least 24 h (Allen et al., 2000; Zheng and Jones, 2003). The mechanism of megasome formation is only partially understood. Phagosome clustering requires phagocyte microtubules, and megasomes are observed only in cells containing live, metabolically active type I H. pylori (Allen et al., 2000). The results of two studies have begun to define megasome composition, and the data demonstrate that H. pylori inhibits phagosome maturation (Zheng and Jones, 2003; Schwartz and Allen, 2006). Specifically, megasomes accumulate coronin and early endosome antigen 1, but are not strongly acidified and acquire only limited amounts of the late endosome membrane protein lamp-1.

Urease and VacA are required for inhibition of phagosome maturation, and both ΔvacA and ΔureAB organisms are eliminated in compartments with lysosomal characteristics (Zheng and Jones, 2003; Schwartz and Allen, 2006). Surface adsorption of active urease or treatment of macrophages with NH4Cl rescues ureAB mutants, but elevation of phagosome pH using chloroquine or bafilomyicin A1 does not (Schwartz and Allen, 2006). These data indicate a specific role for urease-derived ammonia (and not phagosome neutralization per se) in infection. This finding is noteworthy because ammonia enhances phagosome-early endosome fusion and is required for homotypic fusion of H. pylori phagosomes although VacA is not (Zheng and Jones, 2003; Schwartz and Allen, 2006).

Most studies of VacA have examined the effects of the pure cytotoxin on epithelial cells. In this system, VacA is endocytosed and induces the formation of moderately acidic vacuoles that accumulate rab7 and lamp-1, but not mature lysosomal hydrolases (Papini et al., 1994; Molinari et al., 1997). Importantly, cell vacuolation requires concomitant exposure of epithelial cells to NH4Cl (Ricci et al., 1997). Although the fate of VacA in macrophages has not been determined, the data support a model in which urease-derived ammonia prevents phagosome maturation early in infection, and thereafter enhances the activity of VacA secreted by intraphagosomal bacteria. Results of a recent study suggest that H. pylori uses a similar strategy to survive inside gastric epithelial cells (Terebiznik et al., 2006).

Early endosome antigen 1 is a tethering molecule required for endosome clustering and homotypic fusion, a function consistent with its retention on H. pylori compartments (Allen et al., 2005a). Conversely, the role of coronin at the H. pylori phagosome is not well defined, and the extent to which this actin-binding protein interferes with phagosome maturation is controversial (Schuller et al., 2001). Nevertheless, accumulation of coronin on H. pylori compartments is consistent with the fact that this protein is recruited to phagosomes in a PI3K-dependent manner (Didichenko et al., 2000).

Finally, manipulation of PKC signalling may also be essential for H. pylori survival. During phagocytosis of most particles and microbes PKCα is activated, and this enzyme has been linked to intracellular killing because of its roles in activation of the respiratory burst and regulation of phagosome–lysosome fusion (St-Denis et al., 1999; Larsen et al., 2000). Indeed, the ability of group B streptococci, Legionella pneumophila and L. donovani, to inhibit or downregulate PKCα is essential for their survival in mononuclear phagocytes (Jacob et al., 1994; Cornacchione et al., 1998; St-Denis et al., 1999; Holm et al., 2001), and overexpression of dominant-negative PKCα is sufficient to impair maturation of latex bead-containing phagosomes in RAW 264.7 cells (Hing et al., 2004). In this same vein, slow uptake of type I H. pylori occurs without activation of PKCα, and bacteria ingested by this mechanism persist inside megasomes (Allen et al., 2000; Allen and Allgood, 2002).

Resistance to reactive nitrogen intermediates

Exposure of murine macrophages to pro-inflammatory cytokines or LPS upregulates inducible nitric oxide synthase (iNOS), and this enzyme catalyses the conversion of l-arginine into nitric oxide (NO). Peroxynitrite (generated by the reaction of superoxide anions with NO) is a strong oxidant and nitrating agent that is highly toxic to several intracellular pathogens including M. tuberculosis and Salmonella (Bryk et al., 2000). In contrast, reactive nitrogen intermediates are relatively ineffective against H. pylori. This is not due to a failure of activated murine macrophages to synthesize iNOS (Gobert et al., 2001; Schwartz and Allen, 2006). Rather, NO fails to accumulate due to the concerted actions of H. pylori arginase, AhpC and urease (Bryk et al., 2000; Gobert et al., 2001). l-arginine is an essential nutrient for H. pylori, and bacterial arginase (RocF) catalyses the conversion of arginine into urea. Subsequent consumption of urea by H. pylori urease drives arginine uptake and catabolism (Gobert et al., 2001). Because l-arginine is also the substrate of macrophage iNOS, consumption of this amino acid by H. pylori limits NO production by substrate depletion. Thus, rocF mutants cannot prevent accumulation of reactive nitrogen intermediates and are killed by wild-type macrophages yet survive in macrophages that lack functional iNOS (Gobert et al., 2001). Whether this pathway is also important for H. pylori survival in humans has not been determined, and this question is difficult to address as human macrophages rarely generate significant amounts of NO (Weinberg et al., 1995; Thomassen and Kavuru, 2001).

All H. pylori strains tested to date contain ahpC, which encodes an alkylhydroperoxide reductase (Bryk et al., 2000). These enzymes are found in many Gram-negative bacteria and catalyse the detoxification of peroxynitirite, thereby limiting oxidative DNA damage and protein tyrosine nitrosylation (Bryk et al., 2000). H. pylori AhpC exhibits broad substrate specificity and may be essential for H. pylori survival as viability of ahpC mutants appears to require concomitant upregulation of other factors that also confer protection against oxidative stress (Olczak et al., 2002).

Neutrophil activation

A characteristic feature of H. pylori infection is a chronic, neutrophil-dominant inflammation of the gastric mucosa, and neutrophil density correlates directly with tissue damage. Bacterial extracts stimulate chemotaxis and activation of neutrophils and monocytes in vitro (Nielsen and Andersen, 1992; Allen, 2000). Moreover, ultrastructural analysis of biopsy samples indicates that large numbers of neutrophils traverse the gastric epithelium and encounter H. pylori in the mucus layer (Zu et al., 2000). Phagocytosis of H. pylori strongly activates PMNs, but ingested organisms are not eliminated, at least in part because NADPH oxidase targeting is disrupted (Allen et al., 2005b). Although much remains to be determined, it is clear that H. pylori uses its virulence factors to trigger neutrophil influx and activation and that this is central to the ability of this organism to cause disease (Fig. 2).

Figure 2.

Mechanism of neutrophil recruitment and activation. Bacteria pass through the gastric lumen and colonize the mucus layer. Urease-derived ammonia and secreted VacA damage the epithelium. Bacteria bind to the epithelium and peptidoglycan delivered into the cytosol by the T4SS triggers synthesis and secretion of IL-8 while CagA disrupts intercellular junctions. Dying bacteria release urease, HP-NAP and Hp(2-20). Urease recruits macrophages to the stomach which in turn release IL-1β and additional IL-8. Collectively, IL-1β, IL-8, HP-NAP and Hp(2-20) promote and sustain neutrophil influx. PMNs migrate into the mucus layer. Phagocytosis strongly activates neutrophils, but NAPDH oxidase targeting is disrupted and superoxide anions are released into the extracellular space. Bacteria are not eliminated and tissue damage is enhanced by toxic ROS.

Inflammation and chemotaxis

It has long been known that phagocytes follow a chemotactic gradient to reach sites of infection and that cell migration is controlled by both microbe- and host-derived factors. Moreover, a hierarchy of chemokines exists such that host-derived factors (such as IL-8) act as intermediate chemoattractants and microbe-derived factors (such as fMLF) act as end-stage chemoattractants (Heit et al., 2002). Dominance of end-stage chemoattractant signalling ensures phagocyte accumulation at sites of infection (Heit et al., 2002). During H. pylori infection a distinct cytokine milieu is generated (Zevering et al., 1999; Meyer et al., 2003), and it appears that the inflammatory response is driven primarily by bacterial virulence factors that act by Toll-like receptor-independent mechanisms (Moran et al., 1997; Lee et al., 2003; Lepper et al., 2005).

During H. pylori infection, cycloxygenase-2 and prostaglandin E2 are dramatically upregulated and drive the onset of inflammation (Meyer et al., 2003). At the same time, tight binding of H. pylori to the gastric epithelium activates the type IV secretion system, and intracellular delivery of peptidoglycan (but not CagA) triggers a signalling pathway that results in synthesis and secretion of IL-8 (Viala et al., 2004). Tissue macrophages also stimulate neutrophil influx via synthesis and release of IL-1β and IL-8 which act sequentially such that IL-1β promotes neutrophil attachment to the vascular endothelium and IL-8 favours chemotaxis (Suzuki et al., 2004). Bacterial factors that directly stimulate phagocyte chemotaxis include HP-NAP, urease, fMLF and a cecropin-like peptide called Hp(2-20) (Broom et al., 1992; Mai et al., 1992; Satin et al., 2000; Bylund et al., 2001). Urease, HP-NAP and Hp(2-20) all reside in the bacterial cytosol and are released by ‘altruistic autolysis’ of dying organisms (Allen, 2000; Bylund et al., 2001). Like fMLP, HP-NAP and Hp(2-20) activate PMNs via Pertussis toxin-sensitive G-protein-coupled receptors, and recent data indicate a specific role for formyl peptide receptor-like 1 in cell activation by Hp(2-20) (Satin et al., 2000; Bylund et al., 2001; Nishioka et al., 2003). Conversely, activation of monocytes and neutrophils by urease is formyl peptide receptor independent (Mai et al., 1992) and specific receptors for HP-NAP have not been described. Of note, intermediate and end-stage chemoattractants have distinct mechanisms of action (Heit et al., 2002), and two lines of evidence suggest that HP-NAP has features typical of an end-stage agent. First, neutrophil adhesion and directed migration are markedly impaired by inhibition of p38 MAP kinase (Nishioka et al., 2003). Second, pre-treatment with HP-NAP ablates cell migration towards the intermediate stimulus IL-8 (Nishioka et al., 2003). This issue merits further study, and signalling pathways essential for cell activation by urease and Hp(2-20) are not well defined. Nevertheless, the data suggest a model in which neutrophil-activating components of H. pylori extracts act as novel end-stage chemoattractants that sustain neutrophil migration into the gastric mucosa.

The phagocyte NADPH oxidase

The multicomponent NADPH oxidase is an essential element of innate defence that catalyses the conversion of molecular oxygen into superoxide anions. Because reactive oxygen species (ROS) are toxic to host tissue as well as microbes, enzyme is tightly controlled. In resting phagocytes the NADPH oxidase is disassembled and inactive with subunits segregated in the membrane and cytosol. Following cell activation, p47phox, p67phox, p40phox and Rac translocate to the membrane where they bind tightly to flavocytochrome b558 (gp91phox/p22phox heterodimers). Importantly, the site of oxidant generation depends on the nature of the stimulus. Soluble agonists (such as fMLF) promote oxidase assembly at the plasma membrane and superoxide is generated in the extracellular milieu. On the other hand, particulate stimuli (such as yeast and bacteria) target the NADPH oxidase to forming phagosomes, and superoxide accumulates in the phagosome lumen (Dahlgren and Karlsson, 1999; DeLeo et al., 1999; Allen et al., 2005b). Concentrating ROS in this locale enhances killing of ingested microbes and protects host cells from oxidative damage. In this regard it is noteworthy that superoxide anions generated by the NADPH oxidase are rapidly converted into more toxic ROS. Superoxide spontaneously dismutates into H2O2 and, in the presence of myeloperoxidase (MPO) released from neutrophil azurophilic granules, H2O2 is converted into highly toxic hypochlorous acid. Pathogens that resist oxygen-dependent killing must either inhibit oxidant generation or evade or withstand toxic ROS.

Phagocytosis of H. pylori disrupts NAPDH oxidase targeting

Measurement of oxygen consumption is the preferred method for studies of the neutrophil respiratory burst because it is quantitative and is not limited by spatial constraints or production of specific ROS. By this assay, H. pylori triggers a rapid and strong response in PMN that, on a per organism basis, exceeds cell activation by other bacteria (Allen et al., 2005b). At the same time, NADPH oxidase targeting is disrupted such that active enzyme complexes are present in patches at the cell surface but not on H. pylori phagosomes (Allen et al., 2005b). Consequently, superoxide accumulates in the extracellular space but not near ingested bacteria (Allen et al., 2005b). By this unusual mechanism, H. pylori evades oxidative killing and promotes tissue damage and ulceration. This finding is significant as pathogens such as F. tularensis or Anaplasma phagocytophilum can inhibit NADPH oxidase activity in PMN (Ijdo and Mueller, 2004; McCaffrey and Allen, 2006). However, H. pylori appears unique in its ability to prevent oxidase assembly at the phagosome while simultaneously promoting enzyme activation at heterologous sites.

Most H. pylori strains also activate PMN as judged by the luminol assay (Rautelin et al., 1994a,b; Teneberg et al., 2000) which detects hypochlorous acid and other ROS generated in the presence of MPO; and as such these data are in good agreement with the ability of H. pylori to stimulate PMN degranulation (Norgaard et al., 1995). Using the luminol assay, Danielsson and colleagues were the first to show that robust neutrophil activation correlated with, yet was independent of, type I strain-specific virulence factors (Rautelin et al., 1996; Danielsson et al., 2000). Indeed, neutrophil activation does not require VacA, CagA or urease and is mediated by heat-sensitive and formalin-resistant bacterial surface factors that can be bypassed by serum opsonins (Rautelin et al., 1994b; Danielsson et al., 2000; Allen et al., 2005b). Recent data suggest a role for SabA in neutrophil activation (Teneberg et al., 2000; Unemo et al., 2005). However, binding and phagocytosis of SabA-deficient organisms are also impaired (Unemo et al., 2005). For this reason it remains unclear whether SabA has a direct effect on NADPH oxidase activity.

Given the fact that H. pylori is not opsonized in the gastric mucosa, it is of considerable interest that opsonization with serum complement factors in vitro markedly reduces the magnitude of the H. pylori-triggered respiratory burst yet is sufficient to redirected NADPH oxidase complexes to the phagosome and impair survival of ingested bacteria (Allen et al., 2005b). These data suggest that robust superoxide release and bacterial persistence both require lectinophagocytosis and, more importantly, also support the notion that neutrophil activation and the associated release of ROS are important elements of virulence required for severe tissue damage and ulceration (Davies et al., 1992; Rautelin et al., 1996; Danielsson et al., 2000).

ROS and chemotaxis

Small amounts of ROS are also released by neutrophils during chemotaxis. However, several lines of evidence suggest that the respiratory burst elicited by chemoattractants differs markedly from cell activation by whole H. pylori. First, NADPH oxidase activation triggered by whole bacteria is robust and sustained, but responses to fMLF, Hp(2-20) and HP-NAP are generally of lower magnitude and very short-lived (Satin et al., 2000; Bylund et al., 2001; Allen et al., 2005b). Second, whole H. pylori stimulates both degranulation and NADPH oxidase activation but HP-NAP does not (Teneberg et al., 2000). Third, NADPH oxidase activation during phagocytosis is not affected by downregulation of formyl peptide receptors or pre-treatment of neutrophils with Pertussis toxin (Allen et al., 2005b). Fourth, the extent to which released HP-NAP and related factors associate with the surface of live organisms is controversial (Blom et al., 2001). Finally, urease is a potent chemoattractant for monocytes and to a lesser extent neutrophils. Unlike HP-NAP and Hp(2-20), urease does not activate the NADPH oxidase. However, ammonia generated by urease can react with hypochlorous acid to generated long-lived, highly toxic monochloramines (Marshall, 1991) and as such, urease-mediated tissue damage is significant but indirect. In accordance with this model, monochloramine scavengers impair monocyte influx and neutrophil activation in the gastric mucosa (Ishihara et al., 2002).

Apoptosis

In the last several years it has become apparent that bacterial pathogens can manipulate host cell death pathways to their advantage. This was first documented for Shigella flexnerii, and uptake of Shigella by macrophages induces rapid cell death via the caspase-1 inflammasome pathway (Hilbi et al., 1998). A similar pathway is triggered in macrophages containing Salmonella (Hersh et al., 1999), and release of viable bacteria from dying cells appears to be essential for bacterial persistence (Hilbi et al., 1998; Hersh et al., 1999). However, this is not always the case as apoptotic death is one mechanism by which activated macrophages control M. tuberculosis (Oddo et al., 1998).

Unlike tissue macrophages, neutrophils are short-lived cells that die by spontaneous apoptosis. Interestingly, DeLeo and colleagues described an ‘apoptosis differentiation programme’ in neutrophils that is triggered by phagocytosis of a variety of pathogenic bacteria including Staphylococcus aureus, Borrelia, Burkholderia and Listeria (DeLeo, 2004). The vast majority of ingested organisms are rapidly killed and, at the same time, phagocytosis induces a global change in neutrophil gene expression that includes upregulation of pro-apoptotic factors and suppression of receptors and other elements of innate immunity. Consequently, the rate of apoptosis of infected cells is accelerated relative to naïve neutrophils, and this may provide a mechanism to control inflammation and tissue damage during the immune response. That certain pathogens can manipulate this programme is suggested by the fact that cell death is accelerated further in PMNs infected with Streptococcus pyogenes (DeLeo, 2004). Thus, pathogen-induced apoptosis of macrophages and neutrophils can impair or enhance bacterial killing in an organism-specific manner.

In this regard, the available data suggest that H. pylori enhances neutrophil survival (Kim et al., 2001) yet promotes macrophage death (Gobert et al., 2002; Menaker et al., 2004). Macrophage viability declines ∼24 h after H. pylori engulfment (Gobert et al., 2002; Menaker et al., 2004). Dying cells exhibit morphological features of apoptosis, and death occurs, at least in part, via the intrinsic pathway (Menaker et al., 2004). Bacterial factors that trigger apoptosis are largely undefined. Disruption of vacA or cagA enhance macrophage survival by 40–55% relative to cells infected with wild-type bacteria (Menaker et al., 2004), but whether this indicates a direct effect of these virulence factors on apoptotic signalling pathways or reflects decreased intracellular survival of mutant bacteria is unclear.

The results of another study demonstrate a key role for polyamines in H. pylori-induced apoptosis (Gobert et al., 2002). Infection with H. pylori causes a rapid and specific induction of macrophage arginase II and activates ornithine decarboxylase. Together, these enzymes catalyse the conversion of l-arginine into polyamines at the expense of iNOS and NO. Death of H. pylori-infected macrophages is ablated by inhibition of arginase II or ornithine decarboxylase, and direct exposure of macrophages to polyamines (such as spermine and spermidine) restores apoptosis in a dose-dependent manner.

In marked contrast with the effects of H. pylori on macrophages, infection of neutrophils with this organism (or exposure of neutrophils to H. pylori extracts) significantly prolongs cell survival (Kim et al., 2001). Similar data have been obtained for neutrophils infected with A. phagocytophilum or Chlamydia pneumoniae (DeLeo, 2004). At the molecular level, H. pylori manipulates neutrophil viability by increasing expression of the antiapoptotic protein Bcl-XL, suppressing expression of Fas ligand and tumour necrosis factor receptor 1, and impairing activation of caspase-3 and caspase-8. A role for H. pylori virulence factors in this process has not yet been demonstrated.

Moreover, under normal circumstances macrophage phagocytosis of apoptotic neutrophils is required for resolution of infection and inflammation. However, it is becoming increasingly apparent that some pathogens, such as Leishmania, use dying neutrophils to gain entry into macrophages (van Zandbergen et al., 2004) and that phagocytosis of apoptotic neutrophils can, under certain circumstances, trigger macrophage activation (Zheng et al., 2004). The ultimate fate of H. pylori-infected neutrophils is unknown and whether phagocytosis of dying cells elicits macrophage activation and/or sustains infection remains to be determined.

Concluding remarks

Helicobacter pylori uses its virulence factors to induce a chronic neutrophil-rich inflammatory response. Phagocytes are recruited to the gastric mucosa and encounter bacteria in the mucus layer as well as regions of ulceration. Bacterial binding to phagocytes is likely multifactorial, and type I organisms retard phagocytosis. Although specific receptors have not been defined, it is clear that engulfment occurs by a unique mechanism that is modulated by bacterial lipids and adhesins as well as phagocyte signalling pathways. Thereafter, VacA and urease disrupt membrane trafficking in macrophages and thereby inhibit phagosome–lysosome fusion. Other virulence factors stimulate the neutrophil chemotaxis and the respiratory burst and detoxify reactive nitrogen intermediates generated by activated murine macrophages. Ultimately, infection prolongs PMN survival, enhances macrophage death, and local tissue damage allows H. pylori to survive in the hostile environment of the human stomach.

The data summarized here demonstrate that H. pylori can profoundly alter phagocyte function, and some strains also evade intracellular killing. At any given time, only a small fraction of organisms in the gastric mucosa are present inside neutrophils, macrophages or epithelial cells. However, the available data indicate that intracellular H. pylori reside in a protected site that not only favours inflammation but is also a key factor in antibiotic treatment failure and bacterial persistence (Andersen et al., 1993; Engstrand et al., 1997). Nevertheless, intracellular survival is not essential for H. pylori as type II strains are not eliminated by the host immune response despite the fact that these organisms are controlled by phagocytes in vitro (Blaser and Parsonnet, 1994; Tompkins and Falkow, 1995). This may be due to the fact that type II H. pylori induce much less inflammation than type I organisms, and phagocyte influx is markedly reduced (Blaser and Parsonnet, 1994; Telford et al., 1997). Thus, the fact that type II H. pylori induce gastritis, but not ulceration, may limit their contact with phagocytes and in this manner sustain infection. Altogether, the data support the hypothesis that type I and type II organisms occupy distinct niches in the gastric mucosa (Blaser and Atherton, 2004).

Acknowledgements

I apologize in advance to all investigators whose work I was unable to cite due to space limitations. Work on this topic in my laboratory is supported by funds from the Department of Veterans Affairs (Merit Review Grant).

Ancillary