Macrophage Microbicidal Mechanisms In Vivo: Reactive Nitrogen versus Oxygen Intermediates in the Killing of Intracellular Visceral Leishmania donovani

Mice with a targeted disruption of the gp91^phox subunit of the NADPH-oxidase complex (phox), derived from a C57BL/6 × 129/Sv background and backcrossed six times with C57BL/6 mice (24), were provided as breeders by Dr. M. Dinauer (Indiana University Medical Center, Indianapolis, IN). C57BL/6 mice (Charles Rivers Labs.) were used as controls. iNOS^−/− KO mice (C57BL/6 × 129/Sv) were maintained as originally described (23); wild-type^+/+ littermates served as controls. Mice were 8–15 wk old when infected; both males and females were used randomly except for control C57BL/6 mice (all female).

Groups of four to five mice were injected via the tail vein with 1.5 × 10⁷ hamster spleen– derived L. donovani amastigotes (1 Sudan strain, provided by Dr. D. Sacks, National Institutes of Health, Bethesda, MD) (2, 3). Visceral infection was followed microscopically using Giemsa-stained liver imprints by counting the number of amastigotes per 500 cell nuclei × liver weight (g) (Leishman-Donovan units, or LDU) (2, 3). Granuloma formation was scored using formalin-fixed tissue sections stained with hematoxylin and eosin (1–3). In some experiments, starting 1 d after infection AG (Sigma Chemical Co.) was added at 1% (wt/vol) to acidified drinking water changed twice weekly; controls received acidified water alone (2). Differences between mean values were analyzed by a two-tailed Student's t test.

L. donovani replicated in the livers of both strains of control mice during the first 2 wk after challenge (Fig. 1); thereafter, parasite burdens declined consistent with killing and a self-healing phenotype (1). At wk 2, liver burdens in both X-CGD and iNOS KO mice were significantly higher (by 1.6- and 2.3-fold, respectively) than in control animals. Since macrophages from X-CGD mice produce RNI normally (24) and macrophages from iNOS KO mice show intact respiratory burst activity (23), the results at wk 2 suggested that (a) neither mechanism by itself was sufficient to help limit early L. donovani replication and (b) phox and iNOS may act in concert to achieve optimal initial activity.

After wk 2, C57BL/6 and wild-type control mice acquired resistance and showed near-resolution of infection by wk 8 (Fig. 1). Although initially more susceptible, X-CGD mice also controlled infection after wk 2 (Fig. 1 A), indicating that the phox-derived ROI mechanism was dispensable. In contrast, liver infection was unrestrained in iNOS KO mice and increased to high levels by wk 8 (Fig. 1 B and see Fig. 3). Thus, although an intact respiratory burst was neither required (Fig. 1 A) nor sufficient (Fig. 1 B) for acquired resistance and near-resolution of visceral infection, these responses were iNOS dependent.

To determine if the capacity of X-CGD mice to control L. donovani infection involved iNOS-derived products, AG was used as an iNOS inhibitor (2, 25). Continuous AG treatment enhanced initial susceptibility of X-CGD mice at wk 2 (Table I), pointing to an active iNOS-dependent mechanism. At wk 4, liver parasite burdens in AG-treated X-CGD mice remained elevated; nevertheless, visceral infection began to come under control (Table I). Although this latter observation raised the possibility of an antimicrobial pathway independent of both phox and iNOS and although such pathways exist (9), we did not verify the degree of iNOS inhibition induced by AG. Since iNOS KO mice failed to control L. donovani, incomplete iNOS inhibition may well explain the decrease in parasite load at wk 4 in treated X-CGD animals.

In the liver, acquired resistance to L. donovani is expressed by a granulomatous response that encircles all initially parasitized resident macrophages (Kupffer cells) (1–4). This reaction is underway by wk 2 and fully evident at wk 4 as resistance develops. The surrounding mantle of the mature leishmanicidal granuloma is formed by influxing T cells and blood monocytes (4), a process governed by Th1 cell–derived cytokines (2– 4). During resolution of visceral infection (wk 8), the tissue reaction involutes and most cellular foci are devoid of visible amastigotes and can be scored as in Table II as microscopically “empty” (5).

The kinetics of the granulomatous response and mature granuloma assembly were similar in control C57BL/6 (Fig. 2, A, C, and E) and wild-type C57BL/6 × 129/Sv (Fig. 3, A, C, and E, G) mice (Table II). In contrast, the tissue reaction in both X-CGD and iNOS KO animals was delayed and incomplete, particularly at the early stages. At wk 2, 70–81% of infected foci in X-CGD and iNOS KO livers had failed to attract any of the early mononuclear cell reaction already evident at 69–81% of parasitized Kupffer cells in control mice (Table II and Figs. 2, A and B, and 3, A and B). Thus, both phox-derived ROI and iNOS-derived RNI, presumably acting together, appear to participate in the early recruitment of T cells and monocytes into infected tissue. This observation further supports the expanding immunoregulatory spectrum of these molecules, particularly NO (10, 26–28), and in iNOS KO mice may also reflect suppressed induction of IL-12 and IFN-γ (17), key antileishmanial cytokines involved in both mononuclear cell influx and macrophage activation (2–4).

At wk 4, granuloma formation emerged at 90% of infected sites in X-CGD mice (Fig. 2 D) and at 68% of foci in iNOS KO mice (Fig. 3, D and F, and Table II). However, further assembly into mature granulomas (present at 31–37% of foci in control mice; Figs. 2 C and 3, C and E) was retarded and <10% of infected foci in either group of mutant mice could be scored as mature granulomas (Table II). In X-CGD mice, AG treatment inhibited early granuloma assembly (Table I), supporting an active role for iNOS in the initial tissue response. However, granulomas developed by wk 4 despite AG treatment, suggesting a mechanism unrelated to phox or iNOS or suboptimal pharmacologic inhibition of iNOS.

Results at wk 8 indicated that neither phox nor iNOS alone was ultimately required for cell recruitment and assembly, since granulomas developed in both types of deficient mice (Table II). In X-CGD animals, capable of killing L. donovani, the tissue reaction was widespread, comprised of both mature granulomas and inflammatory foci devoid of parasites (Fig. 2 F), and as in a prior study (29) was exaggerated at some perivascular areas (data not shown). However, despite the presence of a similarly well-developed inflammatory reaction, mature-appearing granulomas in iNOS-deficient mice contained overwhelming numbers of replicating amastigotes defining a novel structure, the “ineffective granuloma” (Figs. 3 H and 4).

Together, these results demonstrate clear-cut, probable interdigitating roles for both the respiratory burst and endogenous iNOS–derived RNI as macrophage antimicrobial mechanisms in the initial host defense response to L. donovani. However, the activity of the respiratory burst in inflammatory cell recruitment and in limiting intracellular replication is early and transient, since granulomas formed and infection resolved in X-CGD mice. Similar findings of enhanced susceptibility followed by control in X-CGD mice have been reported in a short-term model of Listeria monocytogenes infection (30).

It is possible that mechanisms related to neither phox nor iNOS, such as other sources of ROI and RNI, also emerge to complete the induction of the tissue cellular immune response to L. donovani and contribute additional antileishmanial effects. However, our results clearly illustrate an obligatory role for iNOS in intracellular killing in vivo and resolution of visceral infection. The presence in iNOS-deficient livers of numerous granulomas, well-established but heavily-laden with intracellular parasites (Fig. 4), graphically illustrates the tissue consequences of the absence of this specific macrophage antimicrobial mechanism.

We thank M. Dinauer for providing X-CGD breeders; M. Shiloh, S. Nicholson and S. Potter for their help in establishing the animal colonies; D. Sacks for providing L. donovani; and A. Delph and S. Delph-Etienne for technical assistance.

This work was supported by National Institutes of Health grants AI16963 (to H.W. Murray) and HL51967 (to C.F. Nathan).