TNF-α mediates the development of anaemia in a murine Trypanosoma brucei rhodesiense infection, but not the anaemia associated with a murine Trypanosoma congolense infection

Abstract

Development of anaemia in inflammatory diseases is cytokine-mediated. Specifically, the levels of tumour necrosis factor-α (TNF-α), produced by activated macrophages, are correlated with severity of disease and anaemia in infections and chronic disease. In African trypanosomiasis, anaemia develops very early in infection around the time when parasites become detectable in the blood. Since the anaemia persists after the first waves of parasitaemia when low numbers of trypanosomes are circulating in the blood, it is generally assumed that anaemia is not directly induced by a parasite factor, but might be cytokine-mediated, as in other cases of anaemia accompanying inflammation. To clarify the role of TNF-α in the development of anaemia, blood parameters of wild type (TNF-α+/+), TNF-α-null (TNF-α–/–) and TNF-α-hemizygous (TNF-α–/+) trypanotolerant mice were compared during infections with the cattle parasite Trypanosoma congolense. No differences in PCV, erythrocyte numbers or haemoglobin were observed between TNF-α-deficient and wild type mice, suggesting that the decrease in erythrocytes was not mediated by TNF-α. Erythropoetin (EPO) levels increased during infection and no significant differences in EPO levels were observed between the three mouse strains. In contrast, during an infection with the human pathogen Trypanosoma brucei rhodesiense, the number of red blood cells in TNF-α-deficient mice remained significantly higher than in the wild type mice. These data suggest that more than one mechanism promotes the development of anaemia associated with trypanosomiasis.

Introduction

A significant pathological feature of African trypanosomiasis is severe anaemia. In one study of infected cattle it was pronounced the primary cause of death [1] and a negative correlation between anaemia and productivity has been demonstrated in infected cattle [2]. The capacity to control anaemia is regarded as probably the most important trait of the more resistant ‘trypanotolerant’ cattle [2]. It is dependent on the presence of haemopoietic cells from trypanotolerant cattle [3] and may therefore be either due to a haemopoietic system that can respond more efficiently to an anaemic state or to a phagocytic system that is less prone to develop anaemia.

The aetiology of trypanosomiasis-associated anaemia in cattle is multifactorial, with extravascular haemolysis and noncompensatory erythropoiesis playing a major part [1,4,5]. Anaemia is also a significant feature in mouse models of the disease [6]. A rapid decrease in erythrocyte numbers occurs during the first wave of parasitaemia and is followed by a chronic phase, characterized by stabilization of the anaemia with persistent, low parasitaemia. It is therefore assumed that the anaemia is not directly dependent on the numbers of parasites, but is controlled by host factors, such as cytokines. Several hypotheses have been suggested [1,4]. Increased red cell damage by haemolysins, proteases [7], phospholipases [8,9], neuraminidase [10,11] or their products have been implicated. Also immunological mechanisms have been advanced to explain increased removal of erythrocytes. Autologous IgM and IgG antibodies and C3 have been demonstrated on the surface of red cells from infected cattle [12–14]. The kinetics suggested that autoantibodies appeared after the first parasitaemia peak and correlated with the decline in PCV. The red cell surface may bind auto- or polyreactive antibodies, or may be sensitized by absorption of immune complexes. Alternatively red cells may passively absorb trypanosome molecules, followed by binding of antitrypanosome antibodies and removal from the system [15]. Yet, immunological competence is not essential for the development of anaemia. Irradiated rats still became anaemic after T. brucei infection [4] and when cattle were depleted for T cells, specific and nonspecific antibody production was seriously reduced and delayed, yet anaemia was a consistent feature [16].

The possibility that cytokines might mediate the loss of erythrocytes is reminiscent of anaemia associated with inflammation, which accompanies infectious diseases, such as malaria and HIV. There is strong evidence that anaemia of inflammation is mediated by TNF-α and other inflammatory cytokines [17]. However, the term ‘anaemia of inflammation’ covers a range of disease states and probably covers a range of different pathways as well. Recent studies suggested that anaemia of infection involving hypoferremia is caused by IL-6 and hepcidin [18]. But this is unlikely to be the cause of the anaemia associated with trypanosomiasis: serum iron levels are not significantly lower in infected cattle, at least not in the earlier phases of the disease [19].

Few data are available that report a link between TNF-α and trypanosome-induced anaemia. There is some weak evidence for a role of TNF-α in severity of anaemia in trypanosome-infected cattle [20]. Furthermore, TNF-α-deficient mice have been reported to be highly susceptible to infections with T. congolense, a major parasite of cattle and ruminants, and showed reduced survival rates and higher parasitaemias [21,22]. In a separate study, it was shown that these TNF-α-deficient mice also developed much higher parasitaemias than the wild-type mice when infected with a different trypanosome, T. b. brucei [23], although their survival times were not shorter. These observations suggest that TNF-α plays an important role in parasitaemia control in murine trypanosomiasis, but its contribution to pathogenesis, including anaemia, may be different between T. congolense and T. brucei-infections. To clarify the roles of TNF-α in anaemia development during experimental trypanosomiasis, we examined the kinetics of anaemia in mice deficient for one and two TNF-α-gene copies. We compared the development of anaemia during an infection with T. congolense, a major cattle parasite, and with T. b. rhodesiense, a parasite infective for humans.

Materials and methods

Mice and parasite challenge

Infections with T. congolense clone IL-1180 were carried out as described previously [21]. Trypanosomes from a frozen stabilate were grown in sublethally irradiated A/J mice and blood was taken by cardiac puncture when parasitaemia reached about 10⁷ parasites/ml blood. Experimental mice were infected intraperitoneally with 1 × 10⁴ parasites in 200 µl blood diluted in phosphate buffered saline (PBS, pH 8·0) containing 1% glucose.

The T. b. rhodesiense stabilate was obtained from a human isolate made on 16·8.1961 from Uhunya, Yimbo, in Nyanza province, Kenya. The human parasites were given 1 passage in rats and then frozen as stabilate IL-2025 after 29 days. Infection was done in the same conditions as the T. congolense, but with only 1 × 10³ parasites.

The generation of TNF-α-knock-out mice on a C57BL/6 background by gene-targeting and subsequent inbreeding was described previously [24]. The mice for the present experiments were generated in our laboratory by intermating hemizygote mice, and the different genotypes of TNF-α were selected by DNA genotyping, using a PCR amplification procedure [21].

Determination of blood parameters

Hematocrit or packed cell volumes (PCV) were compared between infected mice from three groups of mice: wild type, hemizygous and homozygous TNF-α-deficient mice. Six mice from each group, three male and three female, were infected intraperitoneously with T. congolense clone IL 1180. An additional four mice from each group underwent the same recording as the infected mice, but were not infected. Blood for PCV measurements was taken twice weekly by tail snip and gently squeezing the tail. The blood was collected by capillary action in 100 µl micro haematocrit tubes coated with heparin-Na. PCV was determined using a standard microhematocrit method. Plasma samples were stored at −75°C until used in radioimmunoassay for EPO. In a second experiment 50 wild type and 55 TNF-α-knock-out mice were infected and groups of 6 mice killed at regular intervals up to 4 weeks after infection. Blood samples were tested for erythrocyte counts and relative haemoglobin concentration. Erythrocyte numbers were enumerated by haemocytometer under phase-contrast microscope. The relative haemoglobin concentrations were measured spectrophotometrically at 540 nm. Samples of 2 µl of blood were collected from the tail and diluted in 150 µl of distilled water in a plate with 96 round bottom wells (Costar 3799, Corning Incorporated, Corning NY, USA). After 30 min at room temperature, the plate was centrifuged at 600 g for 10 min, after which 100 µl of supernatant was transferred to a new plate and the optical density measured at 540 nm in an ELISA plate reader (Multiscan MCC/340, Titertek Instruments, Huntsville, AL, USA). Measurements were carried out in triplicate.

In a second series of experiments, red blood cell counts were compared in wild type and TNF-α-knock-out mice during infections of T. b. rhodesiense (IL-2025) and T. congolense (IL-1180), using 10 mice of each genotype with the first parasite and eight mice with the latter parasite.

Radioimmunoassay of plasma erythropoietin

EPO levels were also compared between infected mice from three genotypes of mice in the first experiment. Two pooled samples, one from the three males and one from the three females, were analysed. The same was done with the uninfected control samples.

EPO was assayed by radioimmunoassay using a Recombigen EPO Kit (Nippon DPC Corporation, Tokyo, Japan). Recombinant mouse EPO (Boehringer Mannheim) was used as a standard and the detection limit was 3·0 mU/ml. One unit is defined as the amount of EPO that is required to produce an equivalent ³H-thymidine incorporation into spleen cells from phenylhydrazine treated mice to that expressed by one unit of the WHO-EPO reference standard (2nd International Reference Preparation).

Statistical analyses

Analyses of variance (anova) was performed on anaemia parameters (haematocrit, RBC counts and relative haemoglobin titres) and EPO concentrations using Genstat 5 software, with fixed effects for mouse genotype, day, and genotype × day, where day is days after infection.

Results

T. congolense-infected mice

Severe anaemia was observed after T. congolense infection in all infected mice, with PCVs decreasing by up to 50% of the preinfection level (Fig. 1a). A slight decrease in PCV was also observed in uninfected mice, up to 25% in the homozygous TNF-α-knock-out mice by day 10 postinfection and about 20% in the other mice, due to the removal of blood at regular intervals. By day 10 parasitaemia reached its first peak in the blood: 1·7 × 10⁷ in wild type, 3·1 × 10⁷ in hemyzygous mice and 6·1 × 10⁷ in homozygous TNF-α-deficient mice. By that time the PCV had already dropped 34% in wild type infected mice and 31% in hemizygous infected mice. In this experiment, the decrease started a bit later in TNF-α-deficient mice, attaining 20% by day 10, but reaching similar levels as wild type and hemizygous mice a week later. There was no statistical difference when the PCV values of the three groups were analysed over the entire period. In all three groups of infected mice, the EPO concentrations increased significantly by day 10 (P < 0·01) and fluctuated between 100 and 250 mU/ml from day 14 onwards (Fig. 1b). No significant differences in PCV and EPO titres existed between the three genotypes of mice.

Fig. 1.

Comparison of PCV from wild type (•), homozygous TNF-α-deficient (○) and hemizygous (+) C57BL/6 mice (a) infected with T. congolense or (b) uninfected. (c) Titres of erythropoietin of infected mice.

In the second experiment, red blood cell numbers and relative haemoglobin levels were compared between T. congolense-infected wild type and TNF-α-knock-out C57BL/6 mice (Fig. 2). Again, no significant differences were observed between the two genotypes, suggesting that anaemia development is similar in the presence and absence of TNF-α. The first parasitaemia wave peaked at day 11 (average 3·3 × 10⁷ in TNF-knock-out mice compared to 1·4 × 10⁷ in wild type mice)

Fig. 2.

Comparison of relative haemoglobin titres (a) and erythrocyte counts (b) from wild type (•) and TNF-α-knock-out (○) C57BL/6 mice infected with T. congolense (——) or uninfected (- - - -). There were no significant differences (P > 0·05) between parameters of wild type and TNF-α-knock-out mice at any time point.

T. brucei rhodesiense-infected mice

During the T. brucei rhodesiense infections, the decrease in red blood cell numbers differed between wild type and TNF-α-deficient mice (Fig. 3). The development of anaemia was less severe in the TNF-α-deficient mice than in the wild type mice (P < 0·001). Maximum parasitaemia during the first wave was observed at day nine, a mean of 24·1 × 10⁷ in TNF-α-knock-out mice and 18·1 × 10⁷ in wild type mice.

Fig. 3.

Mean red blood cell (RBC) counts and standard deviations from wild type (•) and TNF-α-knock-out (○) C57BL/6 mice, either trypanosome-infected (——) or not (- - - -). (a) T. congolense (infected n = 8, uninfected n = 1). (b) T. brucei rhodesiense (infected n = 10, uninfected n = 4). Significant differences between infected wild type and TNF-α-knock-out mice are indicated by *P < 0·05 or **P < 0·01.

Discussion

Anaemia developed similarly in wild type and TNF-α-knock-out mice during T. congolense infections, suggesting that TNF-α is not an essential mediator in its induction. This contrasts with the anaemia associated with malaria [25] and chronic disease [26], which is mediated by TNF-α. In a previous publication [23], it was mentioned that no anaemia was obvious in TNF-α-deficient mice infected with T. brucei brucei. Other signs of pathology were also lacking in the T. brucei-infected TNF-α-knock-out mice: they were as active as uninfected controls and showed no deterioration in their coat condition [23]. In contrast, the coat condition of the T. congolense-infected TNF-α-deficient mice in our experiments was as poor as that of the infected wild type. This suggests that in our host-parasite combination, TNF-α does not mediate anaemia and major pathology. One of the ways that TNF-α influences erythrocyte levels is by its capacity to regulate EPO production [17]. But as EPO levels increased to similar levels in the wild type, the hemyzygous and the homozygous TNF-α-knock-out mice, it is unlikely that TNF-α has a major effect on the production of EPO during trypanosomiasis.

However, during T. b. rhodesiense infections, the decrease in red blood cells was significantly less in TNF-α-knock-out mice compared to the wild type animals, suggesting a role for this cytokine in anaemia development in this host-parasite combination, and implying that the pathology induced by T. brucei parasites in mice is different from that of T. congolense. It will be interesting to see how EPO levels change in the case of a T. brucei infection.

Other pathways induced by activated macrophages have been described as causing anaemia during trypanosomiasis. One is the production of nitric oxide (NO). Bone marrow cells from T. brucei-infected C3H/He mice exhibited increased NO synthase activity. Blocking NO production resulted in a reduction of anaemia [6]. Mice with a disrupted IFN-γ receptor did not respond with increased NO synthesis after infection and were better at controlling anaemia [27]. Other mechanisms are erythrophagocytosis, which is known to occur in bone marrow of trypanosome-infected cattle [28], and the production of oxidative species.

Anaemia associated with trypanosome infections is multifactorial and the relative contribution of each mechanism will differ according to the host-parasite model, the phase of anaemia development and the severity of infection. When pathology and parasitaemia rise, it is likely that a plethora of mechanisms, as referred to in the introduction, will be triggered and contribute to anaemia, such as those that cause erythrocyte lysis (haemolysins and immune-mediated mechanisms) or those that prevent erythropoiesis (reduction of iron levels or inhibition of haemopoiesis) or by retention and accumulation of erythrocytes in the spleen. However, the initial phase of anaemia development is fast, is probably caused by massive extravascular erythrophagocytosis by an expanded mononuclear phagocytic system [4], and, as shown in this paper, can be cytokine-mediated. Our data suggest that also the initial phase of anaemia is the result of a mixture of responses and that the role of TNF-α in the induction of anaemia and pathology will depend on the particular host-parasite combination. It seems likely that a TNF-dependent mechanism is triggered in the T. b. rhodesiense-infected mice, but not, or to a lesser degree, in T. congolense-infected mice. T. b. rhodesiense produces a more virulent infection than T. congolense, as higher parasitaemias were observed and this despite the fact that fewer parasites were used to establish infection. It is possible that this larger number of parasites or a specific T. brucei molecule triggers more TNF-α production, which may result in a larger contribution of TNF-α towards development of anaemia. As anaemia is a critical factor in the innate resistance to trypanosomiasis, it is important to identify the different pathways that can lead to anaemia [16]. Comparison of these different mouse infection models may allow us to dissect the distinct mechanisms that induce anaemia and pathology.