We have previously reported that the addition of exogenous GM-CSF to nonactivated MPM substantially reduces T. cruzi
infection in vitro (28
). This reduction is correlated with higher levels of TNF-α production and occurs in the absence of detectable NO production. Thus, NO is not solely involved in the control of T. cruzi
infections, at least in vitro. Furthermore, neutralization of endogenous GM-CSF with a neutralizing MAb aggravates whereas injection of rmGM-CSF decreases both parasitemia and cumulative mortality of T. cruzi
-infected mice (27
). These observations raised the possibility of a combined, direct effect of GM-CSF and TNF-α on T. cruzi
trypomastigotes. Indeed, besides exerting pleiotropic effects on mammalian cells (18
), these two cytokines have been shown to interact directly with parasites; for instance, GM-CSF acts as a growth factor for promastigotes of Leishmania mexicana amazonensis
), and TNF-α was reported to stimulate the growth of Schistosoma mansoni
The direct effects of rmGM-CSF and rmTNF-α were thus tested separately on trypomastigotes. Our results indicate that the two cytokines affect T. cruzi trypomastigotes in different ways. (i) rmGM-CSF rapidly changes the morphotype of the trypomastigotes and strongly reduces their ability to infect MPM, although most of them remain alive. rmGM-CSF also lyses trypomastigotes but only after longer incubation periods (16 h). (ii) rmTNF-α has also a cytolytic effect on T. cruzi trypomastigotes, and lysis occurs after short incubation periods (7 h). Furthermore, TNF-α reduces the ability of trypomastigotes to infect MPM without affecting their morphology. The antiparasite action of both cytokines is specific for the infective form of the parasite because the vector form (epimastigotes) was completely resistant to the parasitocidal activities of rmGM-CSF and rmTNF-α.
Incubating the parasites in a medium containing rmGM-CSF cause rapidly the morphological changes of slender forms into amastigote-like forms which are noninfectious. Such changes also occur in medium devoid of cytokines. However, this transformation requires long incubation periods (24 to 48 h), and these amastigote-like forms are infectious (2
). Incubation of T. cruzi
trypomastigotes with rmIL-2 had no effect on the parasite, showing that rmGM-CSF had a specific activity because the two cytokines were prepared by the same procedure.
The cytolysis of T. cruzi
by TNF-α is similar to the trypanolytic activity of this cytokine on African trypanosomes such as T. brucei
. This trypanolytic activity is mediated by the lectin-like domain of TNF-α and not by other TNF-α domains that bind to physiological TNF-α receptors on mammalian cells (22
). The trypanolytic activity of TNF-α against T. cruzi
also involves the lectin-like domain of TNF-α because this activity was sharply reduced after incubation with N
) and an anti-TNF-α TIP MAb (22
) but not by the neutralizing anti-TNF-α MAb 1F3F3 (21
). Although TNF-α had only a weak trypanolytic activity against T. cruzi
trypomastigotes (30% ± 5% lysis after 7 h of incubation), its effect on parasite infectivity was significant (77% ± 5% reduction). This activity was further lowered upon preincubation of the cytokine with N
′-diacetylchitobiose and the anti-TNF-α TIP MAb (23
) but not with the anti-TNF-α MAb 1F3F3 (21
). The data suggest that glycosylated molecules may act as receptors or ligands for TNF-α. Lectins such as concanavalin A have been reported to have cytolytic activity against T. cruzi
); furthermore, T. cruzi
trypomastigotes derived from the mammalian host have more and distinct concanavalin A receptors of various types than noninfective epimastigotes (10
). Thus, the lectin-like domain of TNF-α may be specific for glycosylated moieties that are selectively produced on T. cruzi
It is not clear which GM-CSF domain is implicated in the observed activity. At the optimum concentration (i.e., 62.5 ng/ml), one of the two neutralizing anti-GM-CSF MAbs (MP1-22E9) tested reduced marginally the biological activities of GM-CSF against T. cruzi trypomastigotes. Thus, GM-CSF may also interact with T. cruzi via a lectin-like domain that is not recognized by the currently available anti-GM-CSF MAb.
The different activities of the two cytokines against T. cruzi trypomastigotes may reflect differences in the acceptor molecules for TNF-α and GM-CSF on the parasite and/or different mechanisms underlying cytotoxicity, morphotype transitions, and reduced infectivity. Indeed, the trypomastigotes used in this study were harvested from infected rats. Thus, this population probably contains parasites at various stages of development, and there may be some heterogeneity in the putative cytokine receptors on the parasite membrane. Treatment of parasites with a combination of both rmGM-CSF and rmTNF-α resulted in a higher percentage of lysed parasites than with rmTNF-α alone (Table ). The rmGM-CSF-induced morphotype may be more susceptible to lysis by TNF-α.
The individual and combined activities of GM-CSF and TNF-α were recorded in vitro and may not reflect their activities in vivo. However, parasitemia and cumulative mortality are lowered by injecting rmGM-CSF (500 pg per mouse every 2 days) into T. cruzi
-infected mice (27
). If the cumulative effect of these repeated injections and the synthesis of endogenous GM-CSF are taken into account, concentrations in vivo are probably similar to those used in our in vitro experiments. Similarly, the concentrations of rmTNF-α used in our experiments (39 to 2,500 U/ml) may be also physiologically relevant. Indeed, it has been reported that T. cruzi
infection causes a large increase in TNF-α production in mice and that TNF-α concentrations reach 3,200 to 6,400 U/ml in the serum of T. cruzi
-infected mice (35
). In addition, transgenic mice producing high levels of soluble TNF-α receptors, which neutralize the effects of TNF-α in vivo, are highly susceptible to T. cruzi
Carbohydrate residues exposed at the surface of bacteria and parasites may bind many soluble factors with various effects. The binding of soluble cytokines to the parasite surface may be a major pathway for leukocyte recognition and activation. Cytokines bound to the surface carbohydrates of parasites may produce opsonin-like signals mediating attachment and phagocytosis by effector cells and may also trigger leukocyte cytotoxicity (12
). This activity may be similar to the direct lysis, mediated by anti-T. cruzi
antibodies, which is inhibited by carbohydrates such as melibiose (9
). High levels of T. cruzi
parasites in hosts with neutralized GM-CSF or TNF-α may be due to an impaired cytokine-dependent immunoprotective response such as NO release. Alternatively, a direct parasitocidal activity of TNF-α and GM-CSF on trypomastigotes could lead to lower infectivity in vivo even in the absence of NO release.