Role of IFN-α/β and IL-12 in the activation of natural killer cells and interferon-γ production during experimental infection with Trypanosoma cruzi

Abstract

Control of Trypanosoma cruzi infection depends largely upon the production of interferon (IFN)-γ. During experimental infection this cytokine is produced early, mainly by natural killer (NK) cells and later by T cells. As NK cells have been reported to participate in defence against T. cruzi, it is of importance to study the regulation of NK cell functions during infection with the parasite. Several innate cytokines regulate NK cell activity, among them being interferon (IFN)-α and IFN-β (type 1 IFNs) and interleukin (IL)-12, which have all been reported to be involved in protection against T. cruzi. The role of these cytokines in regulation of NK cell functions and disease outcome were studied by infection of mutant mice lacking the IFN-α/β receptor (IFNα/βR^–/–) or IL-12 (IL-12^–/–) with T. cruzi. IFNα/βR^–/– mice were unable to activate the cytotoxic response but produced IFN-γ, and were not more susceptible than controls. IL-12^–/– mice were extremely susceptible and failed to produce T cell-derived IFN-γ and nitric oxide (NO), although NK cytotoxicity was induced. The results indicate that IL-12 protects against T. cruzi by initiating T cell-mediated production of IFN-γ, but that endogenous IFN-α/β and NK cell cytotoxicity are not of major importance in defence.

INTRODUCTION

The innate immune response to Trypanosoma cruzi, an obligate intracellular protozoan parasite, involves production of cytokines that have the capacity to regulate natural killer (NK) cell activity. Some of these, such as interferon (IFN)-α, IFN-β (type 1 interferons) and IL-12 have also been implied in protection against the parasite [1–3]. Natural killer cells participate in protection against T. cruzi by mechanisms that are not completely understood [1,4]. Part of the protective role of innate cytokines may therefore be through regulation of NK cell functions. NK cells are cytotoxic cells but they can also produce cytokines [e.g. IFN-γ and tumour necrosis factor (TNF)-α] upon activation [5]. The protective property of NK cells may thus involve cytotoxicity against the parasite or infected cells. It may also depend on NK cell-derived IFN-γ that activates macrophages to produce nitric oxide (NO) and other microbicidal agents. Furthermore, it is possible that production of innate IFN-γ regulates the later adaptive response that is necessary to eventually control T. cruzi infection. These mechanisms are not mutually exclusive.

Cytotoxicity and production of cytokines, both features of activated NK cells, do not necessarily coincide. Differential regulation of these functions has been demonstrated in several systems [6–10]. Both NK-cytotoxicity and IFN-γ production by NK cells are induced during the first days after exposure to T. cruzi in vivo [4,11,12]. IFN-α/β are known to stimulate NK cell cytotoxicity [13], but have also been implied in positive [14–16] and negative [6] regulation of IFN-γ production by NK cells. IL-12 is a strong stimulator of IFN-γ production during many infections and is also known to stimulate NK cytotoxicity [17,18]. IL-12 is considered a key cytokine in T helper 1 (Th1) cell development, even though IL-12 independent production of IFN-γ has been described [19,20].

We have previously shown that the cytotoxic function of NK cells is activated during at least 1 week after infection of mice with a high dose of T. cruzi trypomastigotes [10]. IFN-γ production by NK cells in the same animals peaks at 24 h after infection. A second wave of IFN-γ production appears a few days later, but 8 days after infection NK cells do not contribute substantially to the production of IFN-γ, which is then derived mainly from CD4⁺ cells. In order to study the role of type 1 IFNs and IL-12 in regulation of NK cells and IFN-γ production during T. cruzi infection mutant mice lacking cytokines or their receptors were infected with the parasite. Production of early IFN-γ and NK cytotoxicity were analysed. Parasitaemia and mortality was monitored. The production of later, CD4⁺ cell-derived IFN-γ was measured as well as production of NO.

MATERIALS AND METHODS

Mice and parasites

IL-12 p40^–/– and IFN-α/βR^–/– mice were generated by embryonic stem cell technology [21,22]. Control mice were C57BL/6 for IL-12 p40^–/– and 129/sv/ev for IFN-α/βR^–/–. All mice were bred at the animal facility at the Microbiology and Tumor Biology Center (MTC). The animals were 4–10 weeks old when inoculated intraperitoneally with trypomastigotes of the Tulahuén strain. T. cruzi parasites were obtained from peripheral blood of infected C57BL/6 mice. Control mice were injected with PBS. Parasitaemia was measured periodically and mortality was recorded daily.

Cytotoxic assay

Spleens were removed surgically, suspended in RPMI medium containing 5% fetal calf serum and distributed in appropriate numbers to 96-well V-bottomed microtitre plates. ⁵¹Cr-labelled target cells (5 × 10³) were added to the plates containing effector cells, and the plates were incubated for 4 h at 37°C. After incubation, plates were centrifuged and 100 µl of the supernatant was harvested and transferred to precipitin tubes. The amount of released ⁵¹Cr was measured and the percentage of specific lysis was calculated from the standard formula [23].

Competitive polymerase chain reaction (PCR)

Spleen cells (10⁷) were dissolved in 1 ml Ultraspec RNA solution (Biotecx Laboratories, USA) and total RNA was purified according to the manufacturer's instructions. Two µg of RNA was denatured at 94°C for 5 min, reverse transcribed at 40°C for 45 min and treated at 94°C for 5 min. The reverse transcription was carried out in a total volume of 40 µl with 7·5 mm DTT (Gibco BRL, Paisley, UK), 0·5 mm nucleotides (dNTP, Pharmacia, Sweden), 1 U/µl RNAsin (Promega, USA), 5 µm random hexanucleotides (pd(N)6, Pharmacia) and 10 U/µl M-MLV reverse transcriptase (Gibco BRL).

Each cDNA sample was amplified in a competitive PCR assay. The PCR reaction was performed in 20 µl containing 0·2 mm dNTP, 25 mU/ml Taq polymerase (Perkin-Elmer Roche, USA), 0·5 µm sense and antisense primers, cDNA and 2 µl of competitor fragments of different lengths, but with the same primer binding sequences as the target DNA. The competitors for each cytokine were diluted in a series of at least five threefold dilutions for every sample. A negative control containing no template, a competitor control and cDNA controls were included in every assay. Primer sequences, annealing temperatures and competitors for β-actin and IFN-γ, were the same as reported previously [24]. PCR products were loaded in a 2% agarose gel, electrophoresed with ethidium bromide and photographed. The original concentration of the competitor was known in all cases. The concentrations of cDNA were calculated from cDNA and competitor bands with equal intensity. The cytokine mRNA concentrations were expressed as percentage of the β-actin mRNA concentration for each sample.

Immunoassay for IFN-γ in serum

IFN-γ was measured using a capture ELISA. Rat antimouse IFN-γ MoAb R4-GA2 (Immunokontact, Oxon, UK) was used as the capture antibody (5 µg/ml), and biotinylated antimouse IFN-γ MoAb XMG1·2 (4 µg/ml) (Immunokontact) was used for detection. Quantification of IFN-γ in serum was calculated by comparison with a reference linear regression with a known concentration of recombinant IFN-γ (PharMingen, San Diego, USA).

NO measurement

Endogenously synthesized NO is oxidized to stable nitrite and nitrate. Nitrite entering the vascular system reacts rapidly with oxyhaemoglobin with formation of methaemoglobin and nitrate. We therefore assayed NO in vivo by serum nitrite measurements by the Griess reaction [25] after reducing nitrate to nitrite with Aspergillus sp. nitrate reductase [26]. The nitrate concentration was determined by a standard curve using NaNO₃.

RESULTS

Course of disease in T. cruzi infected mice

In normal mice, infection with the Tulahuén strain of T. cruzi results in a peak of parasitaemia 3–5 weeks after inoculation of 50 trypomastigotes [1]. During this period a varying percentage of the mice die, depending partly on the genotype. When mice deficient in IL-12 were infected with T. cruzi, they rapidly developed high parasitaemia and all animals died within 4 weeks (Fig. 1a,b). C57BL/6 control mice also displayed a high mortality in these experiments but some animals survived and their parasitaemias were much lower. On the contrary, almost all the mice deficient in the receptor for IFN-α/β survived the same challenge (Fig. 2b). 129/sv/ev (which is the control for IFN-α/βR^–/–) had higher mortality. This was also reflected in the parasite growth, as 129/sv/ev mice developed a somewhat higher parasitaemia than mice lacking the receptor for IFN-α/β (Fig. 2a).

Fig. 1.

Parasitaemia (a) and mortality (b) of IL-12^–/– and C57BL/6 control mice after i.p. infection with 50 trypomastigotes of the Tulahuén strain. Error bars represent s.e.m.

Fig. 2.

Parasitaemia (a) and mortality (b) of IFN-α/βR^–/– and 129/sv/ev control mice after i.p. infection with 50 trypomastigotes. Error bars represent s.e.m.

Activation of NK cell cytotoxicity

Both C57BL/6 and 129/sv/ev mice infected with a high dose (10⁵ trypomastigotes) of the Tulahuén strain displayed increased NK cytotoxic activity in the spleen 24 h after inoculation (Fig. 3a,c). In order to examine the importance of IL-12 and type 1 IFNs for this activation, mutant mice were infected and the cytotoxicity of splenocytes to the NK sensitive thymoma cell line Yac1 was assayed. IL-12 was not necessary for activation of cytotoxicity (Fig. 3b). However, IFN-α/βR^–/– mice were unable to mount increased cytotoxic NK activity in response to T. cruzi (Fig. 3d), demonstrating that IFN-α or IFN-β is required for activation of early NK cytotoxic functions.

Fig. 3.

Lysis of Yac1 by spleen cells 24 h after infection with 10⁵ trypomastigotes.

Wild-type mice retain activated cytotoxicity throughout the first week of infection. This is not necessarily a result of the events leading to the initial activation. Other activating factors that appear later may influence the lytic capacity. To study this, IL-12^–/– and IFN-α/βR^–/– mice were infected with 10⁵ trypomastigotes and cytotoxicity against Yac1 was measured, now 8 days after inoculation. At this stage the elevated NK cytotoxic capacity was still independent of IL-12 (Fig. 4b). Furthermore, it could now be observed in IFN-α/βR-/– mice (Fig. 4d). Thus other factors become involved in activation of NK cytotoxicity after the first day of infection.

Fig. 4.

Lysis of Yac1 by spleen cells 8 days after infection with 10⁵ trypomastigotes.

IFN-γ production

The stimuli that cause IFN-γ production by NK cells early during T. cruzi infection were examined by infection of IL-12^–/– and IFN-α/βR^–/– mice. IFN-γ in serum was assayed by ELISA 24 h after inoculation of the parasite. Surprisingly, both mutant mice had appreciable amounts of IFN-γ in serum at this time, indicating that neither IL-12 nor IFN-α/β are absolutely necessary for induction of IFN-γ production by NK cells in response to T. cruzi(Fig. 5a,b). Furthermore, mRNA for IFN-γ was readily detectable in splenocytes from both infected IL-12^–/– and IFN-α/βR^–/– mice (Fig. 6).

Fig. 5.

Concentrations of IFN-γ in serum 24 h (a,b) and 8 days (c,d) after infection with 10⁵ trypomastigotes as measured by ELISA. Error bars represent s.e.m.; n = 3–5.

Fig. 6.

Accumulation of mRNA for IFN-γ in spleen 24 h after infection with 10⁵ trypomastigotes as measured by competitive PCR. Data are expressed as moles of IFN-γ mRNA per mole of β-actin mRNA in percent ± s.e.m.; n = 3–5.

The situation was different 8 days post-infection, when IFN-γ is mainly T cell-derived [10]. Serum from IFN-α/βR^–/– mice contained considerable amounts of IFN-γ when assayed after 8 days of infection (Fig. 5d). In contrast, the concentrations of IFN-γ in serum from IL-12^–/– mice were below the detection level (Fig. 5c). Thus, IL-12 but not type 1 IFNs are necessary for activating T cell production of IFN-γ in this system.

NO production

IFN-γ activates macrophages and other cells to produce microbicidal NO through the enzymatic activity of inducible nitric oxide synthase (iNOS). NO in serum becomes oxidized and appears in the form of nitrate. There is always some nitrate in serum but the concentration increases when iNOS is activated. Serum from IL-12^–/– and IFN-α/βR^–/– mice was assayed for nitrate in order to examine how NO production is regulated by the respective cytokines during T. cruzi infection. After 24 h of infection, nitrate concentrations remained at control levels (data not included) but 7 days later, infected control mice displayed increased amounts (Table 1). Serum from infected IFN-α/βR^–/– mice contained equal or higher levels than infected controls. On the contrary, IL-12^–/– mice were not able to produce NO.

Table 1.

Nitrate concentrations in serum

	Nitrate in serum (μM)
Mice	Control	Infected
C57BL/6	50 ± 9	100 ± 25
IL-12–/–	51 ± 9	51 ± 8
129/sv/ev	31 ± 8	100 ± 6
IFN-α/βR–/–	31 ± 5	159 ± 11

Groups of 3–5 mice were infected i.p. with 10⁵ trypomastigotes and blood was collected 8 days later. Data are shown as mean concentration ± s.e.m.

DISCUSSION

Resolution of T. cruzi infection requires induction of a cell-mediated immune response involving production of IFN-γ that activates inducible nitric oxide synthase (iNOS) to produce NO [27–29]. Availability of IFN-γ during the first week after infection is important, because treatment with monoclonal antibodies to IFN-γ at the same time as the parasites are inoculated or the following days results in dramatically increased parasitaemia, whereas treatment during the second week has no effect [4]. NK cells produce IFN-γ only hours or days after T. cruzi infection [4,10,12]. Depletion of NK cells prior to infection with T. cruzi leads to increased parasitaemia and mortality [1,4]. Some of the most important cytokines in regulation of NK cells are IFN-α/β and IL-12 [30]. In this study we investigated their role in activation of NK cells during infection with T. cruzi. IFN-γ was present in the serum of IFN-α/βR^–/– mice both at 24 h and 8 days after infection. We have demonstrated previously that the 24 h peak is mainly NK cell-dependent and that CD4⁺ cells produce most of the IFN-γ after 8 days [10]. IFN-α/β thus do not seem to be required for induction of IFN-γ production in either of the cell types. The activation of the early (24 h) cytotoxic response was completely inhibited in IFN-α/βR^–/– mice but 8 days after infection infected knockout mice displayed higher NK cell cytotoxicity than uninfected controls, showing that (1) IFN-α/β are necessary for the initial lytic response and (2) other factors are operating to stimulate NK cell cytotoxicity later during infection.

When IL-12^–/– mice were infected with T. cruzi, a different picture emerged. The cytotoxic response did not differ from that of controls, but IFN-γ was absent in serum after 8 days of infection. This suggests that IL-12 is necessary for development of a Th1 response in this model. Surprisingly, IL-12^–/– mice expressed substantial amounts of IFN-γ in serum after 24 h of infection. This means that factors other than IL-12 stimulate IFN-γ production by NK cells. Both IFN-α and IL-18 have been shown to promote production of innate IFN-γ [16,31].

When T. cruzi infected IFN-α/βR^–/– mice were assessed for parasitaemia and mortality, they were not more susceptible than wild-type mice, suggesting no protective function of IFN-α/β in this model. This apparently contradicts some earlier work [1,3]. In those studies, however, type 1 interferons were either induced by chemicals or added exogenously. This may have resulted in higher concentrations or different kinetics of IFN-α/β induction than that induced by the parasites themselves. It should be emphasized that T. cruzi induces considerably lower concentrations of type 1 IFNs than some other pathogens, viruses in particular [10,30].

IL-12^–/– mice were considerably more susceptible to T. cruzi infection than controls. The parasitaemia was notably higher and all animals succumbed within 4 weeks of infection. It seems that the absence of an appropriate CD4⁺ T cell response, reflected by the lack of IFN-γ and caused by the lack of IL-12, determines susceptibility. This was supported further by measurements of nitrate concentrations in the blood of mice after 8 days of infection. The infected IFN-α/βR^–/– mice, like infected wild-type mice, had increased concentrations of NO₃ in serum, demonstrating the presence of NO as an effector mechanism. At this time, IL-12^–/– mice only displayed background levels of NO₃ in serum. These data are in accordance with the general notion that control of T. cruzi requires an adequate Th1 response [32].

The importance of the NK-derived IFN-γ remains uncertain. In this model, it is not crucial for development of a Th1 response because NK depletion only affects early IFN-γ production [10]. Neither does early NK cell cytotoxicity seem to be important, as IFN-α/βR^–/– mice that fail to induce such activity are not more susceptible than controls. These results are in accordance with studies of perforin deficient mice, and indicate that cytotoxicity is not a major defence mechanism against T. cruzi [33]. Others have demonstrated a limited role for cytotoxicity [34]. However, none of these studies are conclusive concerning the involvement of NK cells as CD8⁺ T cells, which also depend partly on perforin for killing, are protective against T. cruzi [35,36]. These data disagree with the results in another study in which NK1·1⁺ cells were protective only in the presence of a functional thymus [37]. In this study the NK1·1⁺ cells appeared to promote the generation of activated and memory T cells as well as T dependent antibody responses. The discrepancy between the studies may be explained by the activity of NK1·1 positive T cells (NK T cells). The reports on the role of NK T cells in T. cruzi infection are conflicting. Duthie et al. found that CD1d^–/– mice, lacking NK T cells, are more susceptible to the CL strain of T. cruzi than controls, and suggest that this may be due to early IFN-γ produced by the NK T cells [38]. NK T cells have also been shown to activate NK cells rapidly and their protective functions could therefore be double [39]. Other studies of CD1d^–/– mice do not support a protective role for NK T cells. Miyara et al. found CD1d^–/– to be more resistant than control mice to T. cruzi [40]. In our laboratory no significant difference was observed between the susceptibility of CD1d^–/– mice and controls (Andersson and Une, unpublished observation). In the two latter studies the Tulahuén strain of T. cruzi was used. This may explain the differences. Increased resistance of CD1d^–/– mice could alternatively be explained by the absence of NK T-derived IL-4. The protective function of NK cells in T. cruzi infection may be to stimulate effector functions, such as generation of NO, in the early stage of infection when parasites are multiplying exponentially in the form of amastigotes and spreading in the blood.

In conclusion, we show that T. cruzi infection of mutant mice lacking expression of IL-12 display extremely high susceptibility which is related to the failure of T cells, but not NK cells, to produce IFN-γ. A concomitant inability to generate NO was also observed. Mice that cannot respond to type 1 interferons were unable to respond to infection by a rapid increase in NK cell cytotoxicity, but showed normal IFN-γ production and survival, arguing against an important function for IFN-α/β in defence against T. cruzi.