The elucidation of the key mucosal defences against Giardia is guided by two central features of the infection, the luminal localization of the parasite in the small intestine, and the lack of significant mucosal inflammation after infection. The former requires that effector molecules reach the lumen in sufficient quantities to be effective. The latter suggests that inflammatory mechanisms known to be important for eradication of many enteric pathogens, e.g. neutrophils for killing of Salmonella, are not likely to be important in anti-giardial host defence. Based on these considerations, the following sections focus on the importance of specific host effector mechanisms against Giardia. Elucidation of these mechanisms is not only important for understanding mucosal immune defence against this parasite, but also provides a crucial basis in the rational development of vaccination strategies for activating the most effective host defences against Giardia.
Secretory antibodies of the IgA and IgM isotypes are attractive candidates for immune defence against Giardia, because they are secreted in large quantities into the intestinal lumen and their actions are antigen-specific. A number of studies have been performed to define the overall importance of B cells in clearing Giardia. The majority of reports suggest that B cells are indispensable for eradicating Giardia. For example, infections of humans with G. lamblia or of mice with G. muris result in the production of anti-giardial antibodies of the IgA, IgM, and IgG isotypes in mucosal secretions and serum, and specific antibody production correlates with giardial clearance (2,6,30–32). Such antibodies reach their targets in vivo, since anti-giardial IgA antibodies coat trophozoites in Giardia-infected mice (30). Mice depleted of B cells by treatment with anti-IgM antibodies and mice with X-linked immunodeficiency, which have a defect in B cell development and function, are unable to clear G. muris infection (33,34). Furthermore, B cell-deficient mice generated by gene targeting are unable to clear infections with G. muris or G. lamblia GS/M-H7 (35,36).
In contrast, other reports using animal models have suggested that B cells have only a limited role in anti-giardial immunity. For example, mice with B cell defects due to X-linked immunodeficiency can develop acquired immunity against secondary challenge with G. muris (37). A more recent study showed that B cell-deficient mice infected with either G. lamblia or G. muris controlled infection as well as normal litter-mate controls after 4 weeks, suggesting that B cells played a limited role, if any, in controlling acute G. lamblia infection in those studies (38). The reasons for the apparent discrepancy between these and other reports are not clear, but might relate to differences in the experimental design between the studies. For example, differences in the initial Giardia inocula could lead to different peak infectious loads. B cells might be more important for controlling higher infectious burdens (36), while they play a less important role in controlling lower initial and peak infectious loads (38). Other, more indirect factors could play a role. For example, the intestinal microbiota has been shown to play a role in determining susceptibility of mice to G. lamblia infection (15). The normal murine microbiota is likely to differ between different animal facilities, leading to differential innate susceptibility to Giardia and hence relative importance of B cells in giardial clearance. In any case, these data provide strong support for the existence of B cell-independent anti-giardial host defences.
Further support for the role of B cells in anti-giardial host defence comes from studies in mice genetically deficient for IgA production (36). These mice cannot eradicate infection with G. muris or G. lamblia, although they have a limited capacity to reduce the initial infectious load over extended periods (Figure 4). In contrast, mice deficient for secreted IgM show no defect in clearance of G. muris, indicating that IgM antibodies have no indispensable functions in anti-giardial host defence (36). These findings demonstrate that IgA is central for clearance of Giardia in the murine host, but also suggest the existence of IgA-independent, presently poorly defined host defences. Taken together, the preponderance of evidence in animal models suggests that B cells have a central role in anti-giardial host defence, but B cell-independent effector mechanisms also exist and contribute to host defence to a variable degree.
Figure 4. Importance of IgA in anti-giardial host defence. Mice deficient for IgA (open bars) and normal litter-mate controls (closed bars) were infected orally with G. muris cysts (top panel) or G. lamblia GS/M-H7 trophozoites (bottom panel). Trophozoite numbers in the small intestine were determined at the indicated times after infection (week 1 after G. lamblia infection represents data points between 4 and 7 days). Data are mean ± SEM of 10 or more mice. Asterisks indicate values significantly different from those in litter-mate controls (P < 0·05). Results are a combination of data reported in reference (36) and additional, more recent data.
Download figure to PowerPoint
Data from clinical studies on the incidence and severity of giardiasis in patients with selective immunodeficiencies have provided conflicting results. Patients with chronic variable immunodeficiency (CVID) have an increased susceptibility to Giardia infection (39). These patients show defects in B cell functions, but they also have T cell defects, so that the increase in Giardia prevalence cannot be attributed solely to a lack of antibodies. For patients with selective IgA deficiency, some studies found an increased incidence of infections while others did not (39–41). Selective IgA deficiency is usually defined by serum IgA levels of < 0·05 g/L (normal levels are 0·5–3·5 g/L), but many, if not most, IgA-deficient patients have low but detectable levels of serum IgA antibodies (40,42) and can produce IgA antibodies at near-normal levels in vitro (43,44). Therefore, most IgA-deficient patients are only relatively deficient and can still produce low levels of IgA, which may be sufficient for controlling G. lamblia infection. This is in contrast to IgA-deficient mice, which show a total loss of IgA production due to the deletion of key gene segments coding for IgA heavy chains (45). A simple interpretation of both clinical and experimental data is that IgA is required for clearance of Giardia, but IgA levels well below normal are sufficient. Alternative interpretations are possible, e.g. anti-giardial host defences in mice and humans differ substantially in regard to IgA dependency, or IgA-deficient patients but not mice can develop compensatory host defences against Giardia infection.
The mechanisms by which IgA exerts its anti-giardial functions are not well understood, but are likely to involve ‘immune exclusion’ (e.g. immobilization or detachment of trophozoites from the intestinal epithelium or the mucus layer) rather than direct killing. Anti-giardial IgA antibodies have been reported not to kill G. muris trophozoites in the presence or absence of complement (46), although one study suggested that anti-giardial IgA may exert cytotoxic effects on G. lamblia (47).
A number of antigens have been characterized that are recognized by anti-giardial antibodies, although only a small number of studies have focused specifically on antigens recognized by anti-giardial IgA, rather than other, functionally less relevant isotypes. Moreover, although at least 20 antigenic polypeptides ranging from 14 to 125 kDa have been identified in crude extracts of trophozoites, relatively few of these have been characterized at the molecular level (2). The best-described giardial antigens are members of the family of variant-specific surface proteins (VSPs), which constitute a major fraction of surface proteins in trophozoites (48). At least 150 genes coding for different VSPs are estimated to exist in the giardial genome, of which only one is generally produced per trophozoite (49). However, in a population of trophozoites, multiple different VSPs are expressed simultaneously and at different times during the infection (48,50). VSPs are major antigenic targets but their high variability during the course of infection suggests that antibodies against specific VSPs may have a limited role in mediating eradication of Giardia from the host, since the emergence of trophozoites expressing ‘new’ VSPs would not be prevented by such antibodies. Although much has been learned about the expression of VSPs and their role as antigenic targets, their function has remained unclear. Several other giardial antigens have been identified at the molecular level, including heat shock proteins, giardins, tubulin, and chitin (2). Some of these are intracellular proteins, which raises doubts about their importance as relevant antigenic targets and rather suggests that the development of a specific antibody response might be secondary to trophozoite lysis in the host.
In addition to secretory antibodies, other effector mechanisms are likely to operate in mucosal defence against Giardia. Such defences, like secretory antibodies, have to reach the intestinal lumen in sufficient quantities to exert their effects on the parasite. This suggests that antimicrobial products synthesized by epithelial cells in the small intestine are good candidates. One of these is nitric oxide (NO), which has antimicrobial actions on a wide range of bacterial and parasitic pathogens (51,52) and has multiple other functions, including a role in neurotransmission and regulation of mucosal barrier integrity and vascular tone in the intestine (53). NO is produced enzymatically from arginine through the action of NO synthase (NOS), which exists in three isoforms, neuronal NOS (nNOS/NOS1), inducible NOS (iNOS/NOS2), and endothelial NOS (eNOS/NOS3). In many cell types, the expression of iNOS is inducible by cytokines and microbial products, and iNOS is the major NOS isoform expressed by intestinal epithelial cells (54,55). Expression of iNOS in these cells can either be constitutive, e.g. in mouse ileum (56) and isolated normal human duodenocytes (57), or inducible in vivo during intestinal inflammation (58) or in vitro by cytokines such as interferon-γ (54) or in response to infection with invasive bacteria (55,59). In polarized intestinal epithelial cells, the stable NO end products, nitrite and nitrate, are preferentially detected at the apical side (26,55), suggesting that Giardia could be a relevant target for epithelial cell-derived NO and its metabolites. The underlying mechanisms for apical NO release are not known, but might be related to the preferential localization of iNOS at the apical side of polarized epithelial cells underneath the cell membrane (56,60).
Consistent with a role of epithelial cell-produced NO as a potential anti-giardial effector molecule, NO was found to inhibit proliferation of G. lamblia trophozoites in vitro, but not to kill them (26). Thus, NO was cytostatic rather than cytotoxic for trophozoites in these studies. Another report suggested that NO can kill trophozoites in vitro (61), although that study used minimal media (i.e. buffered salt solution) for culture, which may have exaggerated the results, as such media do not support prolonged viability and growth of trophozoites even in the absence of NO. In addition, NO also inhibited excystation and encystation of G. lamblia in vitro (26). These data show that NO exhibits complex inhibitory effects on growth and differentiation of G. lamblia in vitro (as depicted schematically in Figure 5). Inhibition of growth and excystation would be expected to reduce the number of trophozoites in the intestinal lumen, whereas inhibition of encystation might have the opposite effect, i.e. increased numbers of trophozoites in the lumen. The net outcome of these opposing effects is difficult to predict at this time. Growth inhibition may be important for the infected host, because local trophozoite growth is probably crucial for the ability of G. lamblia to establish and maintain infection of the proximal small intestine. In contrast, inhibition of encystation by NO could reduce the formation and passing of infectious cysts and, thereby, transmission to other potential hosts.
Figure 5. Model of the role of NO in the interactions between Giardia and intestinal epithelium. Intestinal epithelial cells produce NO enzymatically from arginine through the action of inducible NO synthase (iNOS). NO is released predominantly at the apical side of the cells, where it can inhibit growth of G. lamblia trophozoites, as well as encystation and excystation. On the other hand, apical arginine is required for epithelial NO production but is efficiently taken up by trophozoites and used to generate ATP. As a by-product of this reaction, ornithine is released, which competitively inhibits arginine uptake by the intestinal epithelium. Both mechanisms reduce the availability of arginine for the epithelium and thus inhibit NO production.
Download figure to PowerPoint
Evaluation of the potential importance of NO in controlling Giardia infection is complicated by the observation that trophozoites can suppress epithelial NO production in vitro. Thus, co-culture of G. lamblia trophozoites with human intestinal epithelial cells was shown to strongly inhibit inducible NO production (26). This was not due to impaired iNOS expression, but rather caused by highly effective arginine consumption by the parasite, which limited the availability of this substrate for NO production by epithelial NOS (26). Addition of sufficient exogenous arginine reversed the Giardia induced inhibition of epithelial NO production (26). Arginine is an important source of energy for the parasite (62). Consistent with this, G. lamblia possesses a highly efficient arginine transporter system, which has a comparable substrate affinity to the mammalian arginine transporter but a 10- to 20-fold higher maximal transport capacity (63–65). Therefore, G. lamblia may have an advantage over the host in taking up arginine. This is functionally relevant since epithelial NO production appears to depend largely on apical but not basolateral or intracellular arginine availability (26). These observations demonstrate that G. lamblia has strategies to counteract epithelial NO production as a potential host defence mechanism. This places NO at the centre of several, mutually inhibitory interactions between the parasite and the host intestinal epithelium (Figure 5).
Little is known about the importance of NO in controlling Giardia infection in vivo. One report suggested that NO plays no role in clearance of G. muris in the murine host, as treatment of infected mice with the NOS inhibitor, L-NMMA, did not affect trophozoite numbers in the first 12 days after infection (10). These negative data do not completely rule out an anti-giardial function of NO in vivo. It is possible that NO production was not completely inhibited by pharmacological means in that study, or that NO might exert anti-giardial activity at later times after infection. However, NO may simply not have anti-giardial actions during the course of infection in normal mice. This would suggest that the in vitro observed NO-mediated giardial growth inhibition does not occur in vivo, or that insufficient amounts of NO are produced by epithelial cells to exert those effects. In regard to the latter, it is possible that constitutive epithelial NO production may be low under non-inflamed conditions in the small intestine, similar to those observed after Giardia infection in normal mice. However, epithelial NO production can be increased during small intestinal inflammation (57). Infections with a number of other pathogens of the small intestine, such as Trichinella spiralis, are characterized by mucosal inflammation, and co-infections with multiple pathogens are common in developing countries (66). Under these conditions, increased epithelial iNOS expression caused by one pathogen may help to control infection with another pathogen, such as G. lamblia. This concept is illustrated, for example, by the observation that infection of mice with T. spiralis, which causes mucosal inflammation and increased NO production in the small intestine (67), interferes with subsequent Giardia infection (68).
Peptides and proteins with antimicrobial activity are highly conserved in evolution and appear to have an important role in innate host defence at mucosal surfaces. The defensins are one major class of antimicrobial peptides. They can be divided into two major families, α-defensins and β-defensins. α-defensins are produced by neutrophils and specialized epithelial cells, the Paneth cells, in the small intestinal crypts (69). Humans express at least two family members, human defensin (HD) 5 and HD6, in Paneth cells, whereas mice produce 20 family members, which are collectively termed cryptdins. α-defensins are stored as inactive proforms in granules in the Paneth cells. In mice, Paneth cell stimulation induces proteolytic processing through the matrix metalloproteinase (MMP)-7, and release of bioactive cryptdins into the lumen of the small intestinal crypt (70). Mice deficient in MMP-7 cannot produce bioactive cryptdins (71). In contrast to α-defensins, several β-defensins are ubiquitously expressed throughout the epithelium in the small intestine and colon, as well as other mucosal surfaces. In humans, β-defensin (HBD)-1 is expressed constitutively in epithelial cells, whereas HBD-2 is normally expressed at only low levels, but its expression can be strongly up-regulated by several proinflammatory mediators and infection with invasive bacteria (72). Another class of antimicrobial peptides are the cathelicidins, in which a cathelin domain is linked to a peptide that has antimicrobial activity. The only known cathelicidin in humans is LL-37/hCAP18, which has a restricted distribution in the intestinal tract, as its expression is limited to the differentiated surface and upper crypt cells in the colon (73). LL-37/hCAP-18 is not expressed in the small intestinal epithelium. Based on the distribution of the different antimicrobial peptides, it is conceivable that Paneth cell α-defensins and the constitutively expressed β-defensins (e.g. HBD-1) could have functions in anti-giardial host defence. In contrast, the inducible β-defensins such as HBD-2, or the cathelicidin, LL-37/hCAP18, are not likely to be relevant for controlling Giardia infection, as they are either not induced in the absence of inflammation or appear not to be expressed in the small intestine under any circumstances.
The potential importance of antimicrobial peptides in killing Giardia trophozoites has been investigated in vitro (74). This study found that cryptdins 2 and 3 killed trophozoites effectively, whereas cryptdins 1 and 6 did not, suggesting a high degree of specificity for killing despite the extensive sequence similarities between cryptdins (74). The cytotoxic activity of cryptdin 2 was lost in the presence of physiological concentrations of sodium chloride, calcium, or magnesium, which is characteristic of many defensins. This raises the question whether these molecules would be active in the intestinal lumen, given the abundant yet variable presence of these ions in the diet. However, other studies have suggested that the salt sensitivity of cryptdins may be an artifact of testing them as purified molecules, rather than in conjunction with other components normally present in Paneth cell secretions (75). In any case, elucidation of the structure/function relationship for cryptdin-dependent killing of G. lamblia might reveal new approaches for designing anti-giardial drugs, particularly since antimicrobial peptides are generally acid resistant and would be expected to be active luminally after oral administration.
No reports have been published on the physiological significance of antimicrobial peptides in controlling and clearing Giardia infection in vivo. Our laboratory found in preliminary studies that MMP-7 deficient mice, which cannot produce bioactive Paneth cell α-defensins (70), were able to control and eradicate G. muris as well as normal litter-mate controls (Figure 6). Interestingly, initial infectious load was significantly lower in these mice compared to normal controls. This could suggest that Paneth cell α-defensins affect giardial numbers indirectly by controlling the abundance and/or composition of the intestinal microbiota in the small intestine, which can inhibit giardial colonization (see below) (15). The apparent lack of inhibitory effects of Paneth cell defensins on giardial numbers may not be surprising in light of reports that G. lamblia cannot induce the release of these molecules in isolated small intestinal crypts (75). These results also suggest that Giardia might be affected by Paneth cell products under conditions when they are effectively released, such as co-infection with stimulatory enteric microbes (75). This possibility remains to be tested.
Figure 6. Giardia muris infection of MMP-7 deficient mice. Adult MMP-7 deficient mice (○), which cannot produce bioactive Paneth cell defensins, and normal litter-mate controls (•) were infected orally with 104G. muris cysts. The small intestine was removed at the indicated times after infection, chilled on ice, opened longitudinally, and agitated vigorously to release attached trophozoites. The number of viable trophozoites was determined in a counting chamber using a phase-contrast microscope. Data are means ± SEM of 6–8 mice for each point. The asterisk indicates a value significantly different from the control at the same time point (P < 0·05 by rank sum test). The grey line represents the detection limit of the assay.
Download figure to PowerPoint
Other effector mechanisms
In addition to α-defensins, Paneth cells release two other major products, lysozyme and secretory phospholipase A (76), but neither of them has been tested for their anti-giardial activity. Another antimicrobial protein is lactoferrin, which is produced mainly by mammary epithelial cells and present in the breast milk. This protein and its N-terminal peptides were shown to kill G. lamblia trophozoites in vitro, which probably contributes to the protection of nursing infants from giardiasis (77). Lactoferrin can also be produced by epithelial cells in the small intestine in humans (78,79). This suggests that it might participate in mucosal defence against Giardia after weaning, but this has not been investigated.
Although most studies have focused on specific molecules with antimicrobial functions, earlier reports some 15 years ago had suggested that host effector cells might be directly involved in Giardia killing. In vitro studies showed that G. lamblia trophozoites can be ingested by human monocytes/macrophages, and then killed through an oxidative mechanism, although with an effectiveness of less than one trophozoite per macrophage (80). For such a mechanism to be functionally relevant in vivo, macrophages must be present in the lumen where the trophozoites reside. Consistent with that, different leucocyte subsets can transmigrate across the epithelium into the intestinal lumen. For example, in mice a modest number (2 × 105/mouse) of leucocytes can be recovered from the small intestinal lumen, although this number did not differ between G. muris infected and control mice (81). More importantly, the vast majority of these cells were lymphocytes, with only a small percentage being macrophages (81,82). Given that peak infectious loads in the murine host are typically > 107 trophozoites/mouse (Figures 4 and 6), this suggests that luminal macrophages are vastly outnumbered by trophozoites during the course of infection and thus not likely to be relevant for Giardia eradication. Furthermore, although greater numbers of T cells are present in the lumen, classical cytotoxic CD8 T cells are not important for clearance of Giardia, as shown in T cell subset ablation experiments (83). Taken together, little convincing evidence exists to suggest that direct leucocyte-mediated killing of trophozoites in the intestinal lumen is a physiologically relevant mucosal defence mechanism in giardiasis.