Pathogenesis of Intestinal Amebiasis: From Molecules to Disease

Superficial epithelial erosion.Once the mucus barrier has been broken down, E. histolytica reaches the luminal surface of enterocytes and initially produces a contact-dependent focal and superficial epithelial erosion (Fig. 3). In vitro, time-lapse microcinematography has shown that the aggressive mechanism of E. histolytica is a complex multifactorial phenomenon that includes adhesion, a contact-dependent “hit-and-run” damage to the plasma membrane of effector cells, phagocytosis, and intracellular degradation of ingested cells (86). The adhesion and cytolytic events have been related to three types of molecules: lectins, amebapores, and proteases.

(i) Gal-GalNAc lectin.Adhesion of the parasite occurs mainly through a surface Gal-GalNAc lectin which binds to exposed terminal Gal-GalNAc residues of target cell glycoproteins (116). Other molecules include a 220-kDa lectin (131), a 112-kDa adhesin (10), and a surface lipophosphoglycan (151). It has been suggested that the Gal-GalNAc lectin could be translocated to the basolateral surface of enterocytes before the parasite exerts its cytolytic effect (77), but the benefits of this action for the parasite remain unclear.

The Gal-GalNAc adhesin is a multifunctional protein composed of an heterodimer of heavy (170-kDa) and light (35/31-kDa) subunits (116). Evidence for the participation of this molecule in the adhesion event of the parasite has been demonstrated by reduced amebic adherence to human erythrocytes, neutrophils, colonic mucins, and epithelia and to certain bacteria when the lectin is inhibited by galactose (17, 29, 33, 124). Complete inhibition of E. histolytica adherence to target cells or colonic mucins is not observed, even when the lectin is blocked with high (100 to 500 mM) concentrations of galactose or GalNAc monomers (42, 77, 115). It has been suggested that the spacing of multiple GalNAc residues on the surface of target cells is important for optimal lectin binding (1). In addition, other molecules (mentioned above) could participate in the adhesion event. In addition to its role in amebic adherence, the adhesin may participate in the cytolytic event, since contact-dependent target cell lysis is reduced in the presence of galactose and a monoclonal antibody against the heavy subunit is capable of partially inhibiting cytolysis without blocking adherence (133). Considering that the purified lectin has no cytotoxic effect even at high concentrations, it has been proposed that the adhesin is involved in the signaling of cytolysis, probably through the stimulation of actin polymerization (84, 133). Furthermore, the adhesin binds to purified C8 and C9 components of complement and blocks the assembly of the complement membrane attack complex on the amebic plasma membrane, suggesting a role in mediating amebic resistance to complement lysis through components C5b through C9 (18).

The heavy subunit of the Gal-GalNAc adhesin is encoded by five genes inE. histolytica strain HM1:IMSS (83, 121, 155). The derived amino acid sequences suggest that the heavy subunit is an integral membrane protein with a small cytoplasmic tail and a large extracellular N-terminal domain containing a cysteine-rich region. The cytoplasmic tail has sequence identity with β₂-integrin cytoplasmic tails and seems to be involved in the inside-out signaling that controls the extracellular adhesive activity of the amebic lectin (169). On the other hand, the cysteine-rich region seems to be essential for several activities described for the lectin. With the aid of monoclonal antibodies, several antigenic determinants that affect adherence and complement resistance have been mapped to this region (reviewed in reference 92). Moreover, the carbohydrate recognition domain is contained within the cysteine-rich region of the lectin (43). The light subunit shows two 35- and 31-kDa isoforms, with the latter having a glycosylphosphatidylinositol anchor. This subunit is also encoded by a gene family that have 79 to 85% nucleotide sequence identity (123). Its function is not yet clear but seems to involve the modulation of parasite cytopathic activity (7, 109).

Comparison between E. histolytica and E. disparreveals very similar Gal-GalNAc adhesins that efficiently bind to target cells and colonic mucins (29, 42, 119). Derived amino acid sequences encoded by homologous genes show that E. dispar heavy- and light-subunit clones are 86 and 79% identical in their primary structures, respectively, to their E. histolytica counterparts (42, 119). In vitro, E. dispar exhibits adherence and cytotoxicity to target cells that is mediated by the Gal-GalNAc lectin (42), although to a lesser extent than E. histolytica (42, 49). These observations are in accordance with the ability of E. disparto produce superficial erosions in the colonic mucosa of experimental animals (28) but leaves unanswered the question why this ameba does not invade any further.

(ii) Amebapores.Once E. histolyticaestablishes contact with mammalian cells in vitro, a rapid cytolytic event takes place that results in swelling, surface blebbing, and lysis of the inadvertent target cell, including lymphocytes, polymorphonuclear leukocytes, and macrophages, leaving the parasite unharmed. The similarity of this event to the perforin-mediated lysis of target cells by cytotoxic T lymphocytes (162) initially suggested the participation of a channel-forming protein called the amebapore, whose activity had been identified in E. histolytica lysates (82, 132, 178).

The amebapore of E. histolytica is a channel-forming peptide of 77 amino acid residues, which has now been purified; the protein has been sequenced, and the respective genes have been cloned (68, 69). Three isoforms, amebapores A, B, and C, are present at a ratio of 35:10:1, respectively, with the genes showing 35 to 57% deduced amino acid sequence identity. The molecules share six cysteine residues at identical positions and a histidine residue near the C terminus. Structural modeling suggests a compact tertiary structure composed of four α-helical structures stabilized by three disulfide bonds (72). Thus, amebapores are different from the much larger (65- to 70-kDa) perforins that contain three amphipathic segments, two α-helices and one β-sheet (162). However, similarities at the structural and functional levels have been found between amebapores and NK-lysin, a polypeptide present in natural killer (NK) cells and cytotoxic T lymphocytes of pigs (reviewed in references 73 and 75).

Like other pore-forming peptides, amebapores are readily soluble but are capable of rapidly changing into a membrane-inserted stage (75). Early observations suggested that the molecule forms multistate channels with similar properties to those found in the barrel-stave aggregates of toxins such as alamethacin (65). With the elucidation of the primary and secondary structures, it is now believed that amebapores aggregate through the arrangement of their amphipathic α-helices. In a model proposed by Andrä and Leippe (5), amebapores bind to negatively charged phospholipids via protonated lysine residues; this is followed by the insertion of the peptide into the lipid bilayer driven by the negative membrane potential of the target membrane. Oligomerization of the peptide occurs with the participation of a key histidine residue (His⁷⁵), that may either interact with another monomer through the formation of hydrogen bonds or stabilize the predicted fourth α-helix. The oligomer forms a channel through the plasma membrane, allowing the passage of water, ions, and other small molecules and thus lysing the target cell.

In vitro, amebapores exert cytolytic activity against several human cultured cell lines. Amebapore C seems to be the most effective, while amebapore A is not efficient in lysing erythrocytes. In addition, the peptides show potent antibacterial activity against gram-positive bacteria by damaging their surface membranes. Damage to the outer membrane-shielded gram-negative bacteria requires high concentrations of amebapore or removal of the wall with lysozyme (6, 71, 72). With the use of synthetic peptides, of the four α-helices present in the molecule, helix 3 was found to be responsible for the membrane penetration, displaying the highest antibacterial activity. Interestingly, helix 3 is also the most highly conserved domain in the three isoforms (71). The peptide derived from isoform C was the most active of several synthetic peptides studied, reaching a magnitude of activity in the range of the whole molecule (6).

Amebapores are localized in cytoplasmic vesicles, as evidenced by positive immunofluorescence staining and by the presence of typical signal peptides of intracellular transport in its primary translation products. The peptides show maximum activity at acidic pH, which is consistent with previous observations that lysis of target cells byE. histolytica required a pH of 5.0 within amebic vesicles (69).

A peptide homologous to E. histolytica amebapore A has been identified in E. dispar (70). The molecules have common structural and functional properties, such as insertion into negatively charged liposomes, highest activity at low pH, localization in cytoplasmic vesicles, 95% identity of primary structures, and a high degree of similarity of secondary-structure predictions. In spite of these similarities, the specific activity of the E. dispar amebapore is 60% lower than that of the one in E. histolytica. This may be related to a shortened amphipathic helix in the former (70).

Surprisingly, in spite of all the advances in the biochemistry and molecular biology of amebapores, their participation in the cytolytic event produced by E. histolytica has not yet been demonstrated. Amebapores are not spontaneously secreted from viable trophozoites (74). Whether the molecule is able to insert into target cell membranes upon adherence in vivo remains to be established.

The presence of pore-forming activity in the noninvasive E. dispar suggests that the primary function of amebapores is to destroy phagocytosed bacteria, the main intestinal source of nutrients of amebas; thus, they have a similar function to defensins found in mammalian phagocytes that kill bacteria and fungi to prevent intracellular microbial growth within digestive vacuoles (reviewed in references 75 and 158). The anaerobic environment found in the colon may favor oxygen-independent mechanisms as a means of destroying bacteria, rather than using oxygen metabolites and nitric oxide. Amebapores and other proteins, like the recently characterized ameba lysozyme that colocalizes to the same cytoplasmic granules of amebapores (61, 105), could synergistically enhance the antibacterial activity.

Interglandular foci of microinvasion.Focal superficial epithelial erosions produced by E. histolyticain humans are followed by small interglandular foci of microinvasion. Electron microscopy studies of experimentally infected rodents have confirmed the invasion of trophozoites through the interglandular epithelium (Fig. 3), where pronounced shedding of desquamating epithelial cells takes place. The superficial desquamation seems to render this area particularly susceptible to invasion, which occurs through an active displacement of amebas with large pseudopodia extending toward the basal epithelial layers (88, 149). During their passage to deeper layers of the intestine, trophozoites must lyse surrounding cells and degrade the extracellular matrix (ECM) components of the colonic mucosa. Thus, this stage of lesion is characterized by continuing lysis of cells, penetration through locomotion, and proteolytic degradation of ECM (Fig. 2). The way in which the parasite exerts its cytolytic action is described above. The last two events, locomotion and ECM degradation, seem to be closely related and possibly occur through cycles of anchorage to ECM components, forward movement, and degradation of ECM.

(i) Invasion through locomotion.In mammals, cell locomotion involves at least three processes: (i) extension of a leading edge, (ii) attachment to the substrate surface through adhesion plaques, and (iii) pulling forward of the remainder of the cell. The motile force underlying cytoplasmic streaming is a Ca²⁺-activated interaction between actin and myosin, similar to the one described for skeletal muscle contraction. InE. histolytica, actin polymerization occurs during pseudopod extension (47). No reports are available on fluctuations in intracellular Ca²⁺ concentration ([Ca²⁺]_i) during locomotion, but the presence of myosin II in leading lamellae (47) suggests that the mechanism for leading-edge extension in amebas is similar to that in mammalian motile cells.

Interaction of E. histolytica trophozoites with fibronectin (FN) and other extracellular matrix substrates induces the formation of adhesion plaques. Ligand recognition and binding seem to involve a 37-kDa FN-binding protein (152) and a 140-kDa integrin-like receptor (153, 154). Isolated amebic adhesion plaques are composed mainly of actin filaments and at least four actin binding proteins: vinculin, α-actinin, tropomyosin, and myosin I (166). Vinculin may serve as a link between actin and an integral component during anchoring of the cytoskeleton to the plasma membrane in higher eukaryotes. Its identification in amebic adhesion plaques suggests a similar function in this parasite. The presence of α-actinin in E. histolytica had already been reported (13). However, the identification of this gel-forming protein in adhesion plaques suggests a role in the stabilization of actin filaments. In mammalian cells, tropomyosin is not present in adhesion plaques but is found in stress fibers, which are absent in amebas. In muscle cells, this cofilamentous protein sterically blocks the interaction of actin and myosin but shifts its position when the [Ca²⁺]_i is raised, thus allowing contraction to take place. As discussed below, adhesion ofE. histolytica trophozoites to FN induces a sustained rise of [Ca²⁺]_i. The unexpected finding in amebic adhesion plaques of myosin I, a molecule known to participate in the intracellular transport of vesicles in higher eukaryotes, awaits further explanation.

The mechanism of the third process involved in cell motility, pulling forward of the remainder of the cell, is still obscure in E. histolytica. During locomotion, the cell body of amebas move toward the pseudopod, leaving a trailing uroid. Actin and myosin II have been identified in the uroid (9), but no regulatory proteins have been reported.

The most mysterious aspect of locomotion, however, is related to its control. Only recently have signaling pathways in E. histolytica started to be explored. As stated above, interaction of E. histolytica with FN induces a marked reorganization of the actin cytoskeleton, with the formation of adhesion plaques. FN binding transduces information into the trophozoites, activating a protein kinase C (PKC) pathway with the production of inositol triphosphate and the phosphorylation of several proteins, probably including PKC itself (137). Activation of PKC through binding of FN or direct stimulation with phorbol myristate acetate shows a direct relationship to actin polymerization and assembly of various actin binding proteins in the adhesion plaques (reviewed in reference 94). Although the targets of PKC have not been identified, its activation sheds some light on one signaling pathway in the parasite.

A second signaling pathway seems to occur via a focal adhesion kinase, pp125^FAK, identified in immunoblots of isolated amebic adhesion plaques induced by FN (166). Moreover, adhesion ofE. histolytica trophozoites to collagen induces the phosphorylation of several proteins, one of which has been identified as pp125^FAK, further suggesting its participation in signal transduction (111). It has been proposed that activation of pp125^FAK might initiate a signal transduction pathway that activates the mitogen-activated protein kinase cascade, allowing the flow of information from the ECM to the cell interior. This is supported by the identification of p42^MAPK in tyrosine-phosphorylated polypeptides induced by collagen-stimulated trophozoites (111).

Incubation of trophozoites with FN produces a sustained rise in [Ca²⁺]_i, which promotes the stabilization of adhesion plaques and focal contacts by polymerizing soluble actin. External Ca²⁺ influx is responsible for the increased [Ca²⁺]_i, since cytoplasmic Ca²⁺stores are rapidly depleted (30). In the absence of Ca²⁺ following the addition of chelating agents, poor adhesion of trophozoites is observed. Since tropomyosin has been identified in isolated adhesion plaques, it is possible that this protein regulates actin-myosin interactions, as occurs in muscle cells, where increased Ca²⁺ levels induce a conformational change in troponin that shifts the position of tropomyosin, thus allowing the interaction between actin and myosin.

Recent studies indicate that actin organization might have different levels of regulation, namely, changes in actin gene expression and balance between the monomeric G-actin and polymerized F-actin configurations. Treatment of E. histolytica trophozoites with drugs known to increase cytoplasmic cyclic AMP levels (forskolin and dibutyryl cyclic AMP) or to activate PKC (phorbol myristate acetate) produces a shift in the actin equilibrium to the F-actin form, which in turn increases actin mRNA levels (85). This suggests an actin feedback-regulatory mechanism, previously described in other cells, where the gene product and the configuration of the microfilaments could directly regulate actin gene expression (85). Evidence for another regulatory mechanism in actin organization is provided by the recent isolation and cDNA cloning of the E. histolytica profilin basic isoform (16). Profilin acts as a buffer in actin organization by either inhibiting or promoting actin filament formation.

(ii) Degradation of ECM components by cysteine proteases.Once regarded as a passive support element for cells and tissues, the ECM is now considered a highly active tissue that participates in cellular migration, proliferation, differentiation, and immune system signaling. In the human colon, the basement lamina underlying the epithelium consists mainly of type IV collagens, laminins, and proteoglycans and contains a number of basement lamina-associated molecules such as fibronectin, tenascin C, and entactin (114). In addition, fibronectin, laminin, and collagens type I, III, IV, and VI have been identified in the lamina propria and muscularis mucosae (3).

The amebic adhesion plaques described above seem to be important not only in the adhesion and locomotion of the parasite but also in the degradation of ECM components. This is supported by the detection of several protease activities in isolated adhesion plaques (166).

Cysteine proteases are the most abundant proteases in the parasite. Potent cysteine protease activities ranging from 16 to 116 kDa are found in amebic extracts electrophoresed on substrate gels (11, 38, 108, 112, 138, 147). Cell fractionation studies indicate that although present in the cytosol, they are enriched in the plasma and internal membranes. In addition, secreted protease activities of 16, 26, and 56 kDa have been reported (64, 80, 81). A total of six distinct genes (EhCP1 through EhCP6) encoding prepro forms of cysteine proteinases have been identified. The enzymes encoded by three of these genes have been purified and characterized, namely, EhCP1 (ACP3; previously known as amebapain) (138, 156), EhCP2 (ACP2; earlier reported as histolysin) (80, 157), and EhCP5, a membrane-bound protease (62). These three enzymes, all with a molecular mass of ∼30 kDa, account for ∼90% of cysteine protease transcripts and for virtually all cysteine protease activity found in E. histolytica lysates (24). It remains to be established whether the different molecular masses reported for other protease activities represent different enzymes or the expression of different states of one of the six genes identified (97).

Amebic cysteine proteases are active against a variety of substrates and increased activity has been reported in clones of high virulence (103). Of the ECM components that E. histolyticaencounters during colonic invasion, laminin, collagen types I and IV, and FN are good targets for amebapain, histolysin, EhCP5 membrane-bound protease, and the neutral 56-kDa protease (Table1).

E. dispar seems to possess four homologous cysteine protease genes, although none of the enzymes (EdCP2, EdCP3, EdCP4, and EdCP6) have been purified. In addition, the presence of the neutral 56-kDa protease in E. dispar has not been reported yet. If E. dispar indeed lacks several of the most potent E. histolytica cysteine proteases (EdCP1, EdCP5, and the neutral protease), it is possible that this difference could partially explain its noninvasive nature.

In addition, E. histolytica possesses a membrane-bound metallocollagenase that degrades collagen types I and III and is more active against the former (100). The degree of collagenolytic activity has been linked to the virulence of different isolates (53, 101, 145, 164). Activation of the enzyme apparently results in translocation of the enzyme from internal membranes to the plasma membrane (87).

In vitro, incubation of trophozoites on collagen-coated wells induces the formation and release of electron-dense granules (EDGs) through a cytoskeleton-driven, Ca²⁺-calmodulin-dependent mechanism (87, 102, 146). In addition to a collagenolytic activity, EDGs seem to contain at least 25 polypeptides with acidic pIs (6.06 to 6.59), 9 gelatinase activities, actin, small molecules including inorganic phosphate (P_i) and pyrophosphate (PP_i), and several ions. Six of these polypeptides (108, 106, 104, 97, 68, and 59 kDa) and two protease activities of 40 and 85 kDa have been claimed to be detected in EDGs but not in total-trophozoite extracts (76). Whether some of the above-mentioned components represent contamination with cytoplasmic molecules not found in EDGs remains to be established.

In summary, E. histolytica possesses the necessary machinery to degrade the ECM components it encounters during invasion. However, the participation of the different proteases during in vivo infections has yet to be demonstrated.

Neutrophil infiltration of the lamina propria.The last observation of the early invasive lesion with superficial ulceration is a mild to moderate infiltration of the lamina propria. At this stage, cell infiltration around invading amebas leads to rapid lysis of inflammatory cells and tissue necrosis. These observations have been confirmed in rodent models of intestinal amebiasis (88, 149).

Neutrophils, representing over 90% of the circulating granulocytes, respond to a variety of cytokines and soluble factors during the inflammatory response. C5a, the cleavage product of complement component C5, is one of the most potent chemotactic and activation molecules for phagocytes. In amebic intestinal lesions, however, the participation of C5a in the neutrophil infiltration of the lamina propria is uncertain, since E. histolytica trophozoites release a neutral cysteine protease that degrades this molecule (128).

In recent years it has become evident that gut inflammation is not only the result of the immune system response to an intestinal insult but also the result of a complex interplay of immune and nonimmune cell interactions that involve epithelial, mesenchymal, endothelial, and nerve cells, as well as components of the ECM (51). Studies on mucosal immunity have provided growing evidence that intestinal epithelial cells, for instance, constitutively express or can be induced to express a number of immunologically active cytokines and soluble factors, including interleukin-8 (IL-8), monocyte chemoattractant protein 1, granulocyte-macrophage colony-stimulating factor, and tumor necrosis factor alpha (TNF-α) (93). Thus, in addition to their fundamental absorptive and secretory functions, intestinal epithelial cells are integral and essential components of the host's innate and acquired immune system.

Neutrophils are rapidly recruited and activated in response to the proinflammatory cytokine IL-8. In human colon epithelial cell lines,E. histolytica trophozoites increase the secretion of IL-8 and TNF-α (45, 179). At least three possible mechanisms have been described for the upregulation of IL-8. Some colon epithelial cell lines increase the secretion of IL-8 in response to the preformed IL-1α released from the cells lysed by E. histolytica. In other cell lines lacking preformed IL-1α, increased IL-8 production seems to be mediated by the adherence of trophozoites through the Gal-GalNAc lectin by a mechanism that involves, at least partially, an increase in [Ca²⁺]_i (45). In a third mechanism, E. histolytica secreted products could induce the secretion of IL-8 in the absence of trophozoite contact (179). Whether any of these mechanisms take place during natural intestinal infection remains to be established, but further support for the importance of intestinal epithelial cells as producers of inflammatory cytokines comes from studies using a severe combined immunodeficient (SCID) mouse-human intestinal xenograft model (148). In this model, increased IL-1α and IL-8 production and extensive neutrophilic infiltration followed infection of the xenograft with E. histolytica. Moreover, both cytokines were of epithelial origin from the human intestinal xenograft (148).

Finally, a chemoattractant activity for neutrophils has been found inE. histolytica (32). Although the molecule responsible for this activity has not been purified, enzymatic treatment suggests that it corresponds to a membrane-bound peptide.

Role of neutrophils.Cell infiltration around invading amebas leads to rapid lysis of inflammatory cells followed by tissue necrosis. The reasons for the inability of neutrophils to destroyE. histolytica are unknown. In other parasitic infections, neutrophils kill invading microorganisms by both O₂-dependent and -independent mechanisms; they produce an intense oxidative burst, and their secretory granules contain highly cytotoxic molecules. E. histolytica could avoid the oxidative burst through an iron-containing superoxide dismutase that can be induced by superoxide anions to produce H₂O₂ (22). In addition, a bifunctional NADPH:flavin oxidoreductase containing NADPH-dependent disulfide reductase and H₂O₂-forming NADPH oxidase activities could aid in the detoxification of hydroperoxides produced during an oxidative stress (23, 27). Although the ameba does not contain catalase, a cysteine-rich 29-kDa protein in amebas has been shown to eliminate H₂O₂(21, 26, 52). There is no evidence, however, on how E. histolytica might escape the action of the cytotoxic molecules released by the neutrophils. The parasite must possess still unexplored mechanisms of protection against its own lytic peptidases, but whether it uses them to avoid attack by external cytotoxic molecules is unknown.

Alternatively, it could be speculated that neutrophils present in humanE. histolytica lesions are not properly activated. As stated above, humans are the only natural host for E. histolytica. Unfortunately, no experimental model has been developed to date that reproduces the invasive intestinal amebic lesions seen in human intestinal amebiasis. Although the results are highly reproducible with the best model available, the gerbil, they can only mimic the initial stages of infection but resolve in 96 h (149). This has been interpreted as an innate resistance to E. histolytica. Recent studies using the model of induction of amebic liver abscess have started to explore the cellular basis of this innate resistance in mice (167). Following intrahepatic inoculation of trophozoites, neutropenic mice developed larger liver lesions than did normal mice, suggesting a role for neutrophils, albeit partial, in the mechanism of resistance in these rodents (167). Also, in the highly susceptible hamster liver model, in contrast to what occurs withE. histolytica, neutrophils effectively eliminate E. dispar within 24 h (48). Comparison of the responses induced by the two species of amebas in experimental liver and intestinal infection might shed some light on their strikingly different pathogenic behaviors.

Neutrophils not only fail to resist E. histolytica but may in fact contribute to host tissue damage through their destruction and release of cytotoxic granules (88, 163). In vitro, human neutrophils are killed even by low-virulence strains of E. histolytica in a 400:1 ratio of neutrophils to amebas, although many of these parasites succumb in the process. Highly virulent strains, however, emerge victorious after incubations with 3,000 neutrophils per ameba (55).

In addition to the contribution of neutrophils, amebas play a direct role in tissue necrosis. Evidence for this is available only from the model of induction of amebic liver abscess, where the parasite is able to produce liver damage in the absence of an inflammatory reaction in neutropenic mice (167). In the same report (167), images suggesting apoptosis of hepatocytes close to the inflammatory infiltrate or in areas of the liver parenchyma distant from amebas or inflammatory cells are shown.

Role of macrophages.The role of macrophages in acute intestinal amebiasis remains obscure. In vitro, human polymorphonuclear leukocytes, peripheral blood mononuclear cells, and monocytes are rapidly killed by E. histolytica trophozoites (134). An amebicidal effect has been reported only after activation of human monocytes and murine spleen, peritoneal, and bone marrow-derived macrophages with high doses of gamma interferon (IFN-γ) (10² to 10³ U/ml), TNF-α (10⁴ to 10⁵ U/ml), or colony-stimulating factor 1 (10⁴ U/ml), with a macrophage-to-ameba ratio of 200:1 (40, 78, 134). However, a high proportion of activated macrophages are still killed during this interaction (40, 134). The amebicidal effect exerted by activated macrophages is contact dependent and involves both oxygen-dependent (NO, H₂O₂, O₂⁻) and -independent (cytolytic protease) pathways (40, 78, 79, 134, 143).

In contrast to these in vitro observations with activated macrophages, a deficiency in parasite-specific cell-mediated immunity occurs in human amebiasis, which gradually recovers following anti-amebic therapy (107, 135). Experimental models confirm that E. histolytica infections are associated with suppression of cell-mediated immunity and particularly with impaired macrophage effector functions. Thus, in the rodent model of induction of amebic liver abscess, macrophages isolated from amebic liver granulomas are refractory to IFN-γ and lipopolysaccharide activation, defective for the production of H₂O₂, IL-1, and TNF-α, and ineffective in amebic killing (39, 173, 174). During the course of cecal amebiasis in gerbils, peritoneal exudate cells are significantly more cytotoxic to trophozoites than are macrophages isolated from mesenteric lymph nodes (31). This higher activity has also been reported for spleen and peritoneal macrophages isolated from intrahepatically infected animals (39), suggesting a local suppression of macrophage functions while distant macrophage populations remain unaffected. In vitro, exposure of naive macrophages to amebic proteins inhibits the production of TNF-α, IL-1, H₂O₂, and NO₂⁻as well as the IFN-γ-induced macrophage surface Ia antigen synthesis and IAβ-mRNA expression (39, 173-175). The possible mechanisms for this amebic-specific transient anergy state have only started to be explored.

Macrophage activation is a complex phenomenon whose effector functions depend on environmental and maturational effects and on the lymphokines and inflammatory stimuli to which they are exposed. To become activated, macrophages respond to external signals through signal transduction pathways like the PKC and PKA cascades. PKC has been linked to macrophage activation, while PKA is associated with decreased TNF-α mRNA accumulation in macrophages. The c-fos gene is among the early genes expressed during macrophage activation, with expression being rapid and transient in PKC stimulation and stable for hours in PKA stimulation. In vitro, naive bone marrow-derived macrophages from BALB/c mice exposed to E. histolyticaproteins increase the expression of c-fos and TNF-α mRNA via the PKC pathway. However, both mRNAs are rapidly degraded and no TNF protein is secreted, suggesting a mechanism whereby the parasite can suppress macrophage function (142).

Surface expression of class II major histocompatibility complex (MHC) molecules is necessary for the effective participation of macrophages as antigen-presenting cells to CD4⁺ helper T lymphocytes. Transcription of class II MHC genes in macrophages is induced by IFN-γ. This molecule also stimulates the production of TNF-α, which, together with IFN-γ, induces l-arginine-dependent cytotoxic effector mechanisms through the production of nitric oxide (NO) (110). During the development of amebic liver abscess in gerbils, macrophages isolated from the lesions are refractory to IFN-γ and LPS stimulation and show decreasing levels of TNF-α production as the infection progresses (173, 174). Thus, these macrophages could be impaired in their antigen-presenting function (see also below) and in the production of cytotoxic molecules like NO.

Prostaglandin E₂ (PGE₂) is a modulator of the immune response, having proinflammatory effects that enhance edema and leukocyte infiltration while inhibiting some macrophage functions like IL-1 and TNF-α production and class II MHC gene complex expression (118). In the gerbil model, amebic liver abscess-derived macrophages display high basal levels of PGE₂ production, which can be further enhanced upon stimulation with live trophozoites or amebic proteins (172). In addition, naive peritoneal and bone marrow-derived macrophages are induced to secrete PGE₂following incubation with E. histolytica proteins (172, 175), which results in decreased TNF-α production in response to lipopolysaccharide stimulation (173). Production of enhanced levels of PGE₂ by macrophages in response to amebic proteins is in part responsible for the suppression of IFN-γ-induced macrophage surface Ia antigen synthesis and IAb-mRNA expression from the class II MHC gene complex, since cotreatment with the cyclooxygenase inhibitor indomethacin partially restores surface Ia antigen expression (175). Further support for the participation of PGE₂ in the downregulation of macrophages has been obtained in vivo. Although the effect of PGE₂ has not been tested in animal intestine models, results obtained from rodents inoculated intrahepatically with amebas suggest a role for this molecule in the induction of amebic liver abscess. Thus, hamsters treated with indomethacin show decreased levels of PGE₂ in plasma and reduction in size of abscess lesions by 30% (136).

How can the in vivo macrophage anergy state be reconciled with the in vitro enhanced ability of activated macrophages to kill E. histolytica? As detailed above, amebic proteins inhibit naive macrophage activation (39, 173-175). However, if macrophages are activated prior to exposure to live trophozoites, there is a 10-fold increase in TNF-α production (173), explaining the enhanced amebicidal activity observed in vitro.

Another topic of interest in the late invasive lesion is the presence of trophozoites in the internal muscular layer in the absence of tissue damage, which seems to occur quite frequently in biopsy samples from patients with invasive amebiasis (91). It has been speculated that tissues might have different susceptibilities to the aggressive mechanisms of the parasite. However, amebas can invade and destroy virtually every tissue in the organism. It has also been suggested that there may be a requirement for a minimal number of trophozoites to produce tissue damage. This could be a possibility, since histopathological sections of intestinal tissue from humans with amebiasis show that in the muscle layer amebas tend to be isolated rather than forming aggregates. However, this does not explain the absence of the inflammatory infiltrate, because in experimental models strikingly large foci of acute inflammation are elicited by very few parasites (163).