Restriction of Toxoplasma gondii Growth in Human Brain Microvascular Endothelial Cells by Activation of Indoleamine 2,3-Dioxygenase

Abstract

One of the first steps in the development of cerebral toxoplasmosis is the penetration of the blood-brain barrier, which is comprised of microvascular endothelial cells. We examined the capacity of human brain microvascular endothelial cells (HBMEC) to interact with Toxoplasma gondii. We found that stimulation of HBMEC with gamma interferon (IFN-γ) resulted in the induction of toxoplasmostasis. The capacity of HBMEC to restrict Toxoplasma growth after IFN-γ stimulation was enhanced in the presence of tumor necrosis factor alpha (TNF-α). In addition, we found that IFN-γ induced a strong induction of indoleamine 2,3-dioxygenase (IDO) activity in HBMEC, and this enzyme activity was enhanced by costimulation with TNF-α. The addition of excess amounts of tryptophan to the HBMEC cultures resulted in a complete abrogation of the IFN-γ–TNF-α-mediated toxoplasmostasis. We therefore conclude that IDO induction contributed to the antiparasitic effector mechanism inducible in HBMEC by IFN-γ and TNF-α.

Prenatal transmission of the obligate intracellular parasite Toxoplasma gondii may result in congenital toxoplasmosis, of which the most serious manifestation is toxoplasma encephalitis. In order to enter the brain, the parasites must cross the blood-brain barrier by one of two possible routes. First, T. gondii may penetrate the brain tissue through infected cells, such as monocytes and macrophages, that are capable of penetrating the blood-brain barrier. Second, the parasites may infect and destroy endothelial cells. Since it is known that most human monocytes are able to kill T. gondii without prior activation, the first possibility seems unlikely (27), but more recent data show that at least some subpopulations of monocytes are able to transport the parasites throughout the body (9). In this report, we analyze the capacity of human brain microvascular endothelial cells (HBMEC) to support the replication of the parasite.

HBMEC have been previously used to study the pathogenesis of central nervous system infections by meningitis-causing bacteria such as Escherichia coli, Citrobacter spp., Streptococcus pneumoniae, and group B streptococci. Several E. coli-HBMEC interactions have been shown to contribute to crossing of HBMEC (13, 20, 23). However, T. gondii replication and the role of cytokines in controlling intracellular multiplication of the parasite in HBMEC have not yet been investigated.

The HBMEC used in this study were isolated from a brain biopsy of an adult female with epilepsy by methods previously described (23). These cells were positive for factor VIII-Rag, carbonic anhydrase IV, and Ulex europaeus agglutinin I, took up fluorescent-labeled acetylated low-density lipoprotein, and expressed gamma glutamyl transpeptidase, thus demonstrating brain endothelial cell characteristics (24). HBMEC were subsequently immortalized by transfection with simian virus 40 large T antigen and maintained their morphological and functional characteristics (25). Directly after thawing, HBMEC were cultured in medium supplemented with 10% heat-inactivated fetal calf serum and 10% NuSerum (Becton Dickinson, Bedford, Mass.). Thereafter, cells were cultured in Iscove's or RPMI 1640 medium supplemented with 5% fetal calf serum in culture flasks (Costar, Cambridge, Mass.) and split weekly in 1:10 ratios by using trypsin-EDTA (Gibco, Grand Island, N.Y).

T. gondii strain BK was obtained from Seitz and Saathoff (Institut fur Medizinische und Parasitologie, Bonn, Germany) and was propagated in the mouse fibroblast cell line L929 (American Type Culture Collection, Rockville, Md.). The parasites were usually harvested after 3 to 5 days of incubation, resuspended in RPMI 1640 with or without l-tryptophan (Gibco), and used for infection experiments.

The most important cytokine involved in the control of intracellular pathogens is gamma interferon (IFN-γ). It is known that IFN-γ, especially in combination with tumor necrosis factor alpha (TNF-α), is able to inhibit the growth of T. gondii in cells like macrophages, fibroblasts, and astrocytes (5, 19). We therefore stimulated HBMEC (3 × 10⁴ cells/well) with different amounts of IFN-γ (0 to 500 U/ml) in the absence or presence of TNF-α (400 U/ml) for 3 days. Thereafter, HBMEC cultures were infected with T. gondii (1 × 10⁴ cells/well). Three days after infection, parasite growth was determined by [³H]uracil uptake. Data from one of five independent experiments are shown in Fig. 1 and indicate that the parasites are able to replicate in HBMEC. Figure 1 also shows that IFN-γ induces a dose-dependent antiparasitic effect which is enhanced in the presence of TNF-α, while treatment of HBMEC with TNF-α alone did not influence parasite growth.

FIG. 1.

Synergistic effect of TNF-α and IFN-γ on the induction of toxoplasmostasis in HBMEC. Aliquots of 3 × 10⁴ HBMEC were stimulated with different amounts of IFN-γ (0 to 500 U/ml) in the absence or in the presence of TNF-α (400 U/ml) for 3 days. The total volume of medium was 200 μl in each well. Thereafter, 1 × 10⁴ Toxoplasma were added to each well, and parasite growth was monitored 3 days later by pulsing the cultures with [³H]uracil for 18 h. Data are given as mean cpm ± the standard error of triplicate cultures.

In vivo, the major source of IFN-γ is T cells, and we have previously shown that Toxoplasma antigen-specific T cells are capable of producing IFN-γ in amounts sufficient to induce toxoplasmostasis (4). Within the first day of infection, the main source for IFN-γ in vivo is the NK cell (22). Furthermore, T cells and NK cells are capable of producing TNF-α, and several other cells like macrophages and astrocytes produce TNF-α after infection with T. gondii (10, 14). In addition, immunohistological studies have shown the presence of both cytokines in the brain and cerebrospinal fluid of Toxoplasma-infected mice (21), and we therefore suggest that the conditions used in our in vitro study are similar to those occurring in vivo.

The capacity of HBMEC to control Toxoplasma growth after IFN-γ stimulation is not unique, as this capacity has been demonstrated for fibroblasts, astrocytes, macrophages, microglia, and epithelial cells (6). However, the effector mechanisms used by different cells after IFN-γ stimulation are different. In human fibroblasts and human glioblastoma cells, the induction of the tryptophan-degrading enzyme indoleamine 2,3-dioxygenase (IDO) is the only effector mechanism active against T. gondii, and a supplementation of the culture medium with excess amounts of l-tryptophan results in a complete abrogation of IFN-γ-induced toxoplasmostasis (5, 19). In contrast, the effector mechanism induced in murine macrophages is the induction of the inducible form of nitric oxide synthase (iNOS), and the addition of N(G)-monomethyl-l-arginine [N(G)MMA] completely blocks the IFN-γ-induced antiparasitic effect (1). Human macrophages and murine astrocytes are also capable of blocking Toxoplasma growth after activation with IFN-γ; however, in these cells the addition of tryptophan and NGMMA did not block the IFN-γ-induced toxoplasmostasis, indicating that neither iNOS induction nor IDO-mediated tryptophan depletion is the active effector mechanism (11, 16). We therefore analyzed iNOS and IDO induction in HBMEC after stimulation with IFN-γ. Aliquots of 3 × 10⁴ HBMEC/well were stimulated with IFN-γ in the absence or presence of TNF-α. After 3 days of stimulation the culture supernatants were analyzed for the presence of nitrites by the use of Griess reagent (8) or for the tryptophan degradation product kynurenine by the use of Ehrlich's reagent, as described previously (3). We found that the nitrite concentration in the supernatant of stimulated cells was below the detection limit of the assay used (<1 μM). In contrast, we found an IDO activity in HBMEC cultures after stimulation with IFN-γ which was enhanced in the presence of TNF-α (Fig. 2A). The costimulatory effect of TNF-α on IFN-γ-induced IDO activity is dose dependent and reaches a maximal effect at 100 to 200 U/ml (data not shown). In additional experiments, we analyzed IFN-γ-mediated IDO mRNA induction in HBMEC using a PCR technique. The detection of iNOS and IDO mRNA by reverse transcription-PCR (RT-PCR) was performed as follows: total RNA from unstimulated and stimulated (IFN-γ at 200 U/ml) cells was extracted with guanidinium thiocyanate followed by ultracentrifugation on a CsCl cushion. One microgram of total RNA was used for first-strand synthesis with the Advantage RT-for-PCR kit (Clontech, Heidelberg, Germany) according to the instructions of the manufacturer. PCR was carried out with the following specific iNOS, IDO, and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) primers: INOS, sense primer 5′ TG GGG CAG CGG GAT GAC TTT 3′ and antisense primer 5′ GT GAT GGC CGA CCT GAT GTT GC 3′; IDO, sense primer 5′ GCA AAT GCA AGA ACG GGA CAC 3′ and antisense primer 5′ TCA GGG AGA CCA GAG CTT TCA CAC 3′; GAPDH, sense primer 5′ ATG GGG AAG GTG AAG GTC GGA GTC 3′ and antisense primer 5′ CAG CGT CAA AGG TGG AGG AGT GG 3′.

FIG. 2.

(A) Synergistic effect of TNF-α and IFN-γ on the induction of IDO activity in HBMEC. Aliquots of 3 × 10⁴ HBMEC were stimulated with IFN-γ (0 to 1,000 U/ml) in the absence or in the presence of TNF-α (200 U/ml) for 3 days. Thereafter, culture supernatants were harvested and the kynurenine content was determined by the use of Ehrlich's reagent. Data are given as mean optical density at 492 nm (OD₄₉₂) ± the standard error of triplicate cultures. (B) Detection of IDO mRNA in IFN-γ-activated HBMEC. HBMEC were stimulated with or without IFN-γ (200 U/ml), and RT-PCR was performed as described in the text, 18 h after cytokine stimulation. As a control, GAPDH mRNA was analyzed in parallel. The data shown indicate that IFN-γ is a strong inducer of IDO mRNA in HBMEC. (C) Synergistic effect of TNF-α on IFN-γ-induced IDO protein production. HBMEC were stimulated in the absence or presence of IFN-γ (800 U/ml) or TNF-α (200 U/ml). After 3 days cells were harvested and subjected to Western blot analysis. As a control, GAPDH was analyzed in parallel. The data shown indicate that TNF-α is unable to induce IDO activity in HBMEC, but it enhances IDO expression in cells costimulated with IFN-γ.

Denaturation time in all RT-PCRs was 3 min at 94°C; cycling time was 30 s at 94°C; annealing times were 1 min at 64°C for iNOS, 45 s at 62°C for IDO, and 45 s at 60°C for GAPDH. Synthesis in all RT-PCRs was 1 min at 72°C for 30 cycles and a further 4 min at 72°C. As shown in Fig. 2B, we found a strong IDO signal in IFN-γ-activated cells, while in unstimulated cells no IDO transcript was found. In contrast, RT-PCR for iNOS was negative in unstimulated and stimulated cells (data not shown). Furthermore, IDO protein was detected by Western blot analyses. The detection of IDO protein was performed with HBMEC cultures grown to confluence in a culture flask and stimulated with 800 U of IFN-γ/ml or IFN-γ with TNF-α (200 U/ml) for 72 h. The cells were lysed and then were subjected to sodium dodecyl sulfate–9.5% polyacrylamide gel electrophoresis. The proteins were transferred to nitrocellulose using a semidry electroblotting apparatus (Phase, Mölln, Germany). The membrane was incubated in 5% skim milk in phosphate-buffered saline for 1 h at room temperature and then incubated for 1 h with IDO-specific mouse monoclonal antibodies (O. Takikawa, Department of Chemistry and Australian Cataract Research Foundation, University of Wollongong, Wollongong, Australia). Afterwards, the membrane was washed with phosphate-buffered saline and incubated for 1 h with goat anti-mouse immunoglobulin G-horseradish peroxidase-conjugated antibodies (Dianova, Hamburg, Germany). Detection was performed with an ECL kit from Amersham.

Western blot analysis done with HBMEC lysates from IFN-γ and IFN-γ–TNF-α-treated cells showed an intensive band of the IDO protein. As a control for the protein load, GAPDH was analyzed in parallel in each sample by using a mouse anti-GAPDH monoclonal antibody (Biotrend, Cologne, Germany). The Western blot data shown in Fig. 2C indicate that TNF-α induces an enhanced IDO protein expression in IFN-γ-activated HBMEC. In untreated cells, IDO was not detectable.

In summary, the data shown in Fig. 2 indicate that IFN-γ induces the expression of IDO activity in HBMEC, which is enhanced in the presence of TNF-α. We have previously shown that TNF-α enhances IDO-mediated mRNA expression in astrocytoma cells (4), and here we have shown that TNF-α enhances IDO protein expression in IFN-γ-activated HBMEC.

Human macrophages are able to restrict Toxoplasma growth and show a strong IDO activation after stimulation with IFN-γ. We found that IDO activity inducible in human macrophages is active as an antibacterial effector mechanism (15). However, IDO does not seem to be responsible for the toxoplasmostasis in human macrophages, since addition of l-tryptophan to the cultures did not influence IFN-γ-induced toxoplasmostasis (16). Similar data were also published by Woodmann et al. (28). Those investigators analyzed IFN-γ-induced toxoplasmostasis in human umbilical vein endothelial cells (HUVEC) and found that the IFN-γ-induced toxoplasmostasis was not abrogated in the presence of l-tryptophan. However, these authors did not analyze possible IDO activation in HUVEC with biochemical or molecular biological methods.

We therefore examined whether the IFN-γ-induced IDO activity is responsible for the observed toxoplasmostasis in HBMEC. We stimulated HBMEC as described for Fig. 1 with IFN-γ or IFN-γ with TNF-α. After 3 days of stimulation the cultures were infected with Toxoplasma (1 × 10⁴ cells/well) in the absence or presence of l-tryptophan (100 μg/ml). As shown in Fig. 3, the supplementation with tryptophan resulted in a complete abrogation of the IFN-γ-mediated toxoplasmostasis. In addition, we found the same blocking effect of supplemental l-tryptophan in the immortalized HBMEC line stimulated with a combination of IFN-γ and TNF-α. Comparable results were obtained in three additional experiments. We therefore suggest that IFN-γ-mediated IDO induction is also the main effector mechanism controlling the growth of T. gondii in native HBMEC.

FIG. 3.

Antagonistic effect of supplemental l-tryptophan on toxoplasmostasis induced in HBMEC by IFN-γ and TNF-α. Aliquots of 3 × 10⁴ HBMEC were stimulated with different amounts of IFN-γ in the presence or absence of TNF-α (100 U/ml). After 3 days of incubation, 1 × 10⁴ Toxoplasma cells were added to each well in the absence or presence of supplemental l-tryptophan (100 μg/ml). Parasite growth was monitored by [³H]uracil incorporation; data are given as mean cpm ± the standard error of triplicate cultures.

After the penetration of brain endothelial cells, T. gondii comes into contact with astrocytes. We have previously shown that the same effector mechanism described here for HBMEC is also relevant in the neighboring astrocytes (6). We have previously shown that IDO inhibited Toxoplasma growth within astrocytoma cells but did not reduce the number of cells infected by the parasites (2). IDO activity in HBMEC results in an inhibition of Toxoplasma growth and therefore might reduce the number of parasites reaching the astrocyte layer.

In a recent review, Denkers (7) stated that it remains unclear how important IFN-γ-mediated tryptophan starvation is as a mechanism against T. gondii. He suggested that IDO may not be an important effector mechanism against this parasite in general, since IDO activity is found mainly in fibroblasts and in a restricted number of other cell types. Here we have shown that IDO is also active in HBMEC, and previously it has been reported that IDO is also effective against T. gondii in astrocytoma cells (2, 5). We suggest that microvascular endothelial cells and astrocytes cooperate in the inhibition of T. gondii growth. The amino acid tryptophan has to cross the blood-brain barrier to reach the astrocytes. The IDO-positive HBMEC are able to cleave tryptophan to kynurenine, and thereby they reduce the transport of tryptophan to the astrocytes. Since IDO is the main effector mechanism in astrocytes against T. gondii, a reduced tryptophan influx enhances the antimicrobial effect of IDO-positive astrocytes. We therefore suggest that IDO activation in HBMEC might play an important role in the antiparasitic defense, especially in humans.

We agree with Woodman et al. (28), who reported that iNOS activation is not responsible for toxoplasmostasis in HUVEC, since we did not find iNOS activation in HBMEC after stimulation with IFN-γ and TNF-α. However, there seems to be a difference in the antiparasitic defense between HBMEC and HUVEC. Woodman et al. (28) reported that tryptophan was not able to abrogate toxoplasmostasis in HUVEC after stimulation with IFN-γ, while tryptophan supplementation blocked the toxoplasmostasis induced in HBMEC. There might be several explanations for these different findings. Although both cell populations are endothelial cells, it is possible that HUVEC and HBMEC have different defense mechanisms against Toxoplasma. It might be possible that in HUVEC cells, as in macrophages, kynurenine is not the final product in the tryptophan degradation pathway (12, 26). Kynurenine degradation in HUVEC might result in the generation of tryptophan metabolites that are toxic for Toxoplasma. Such toxic effects would not be antagonized by tryptophan supplementation. In addition, HBMEC and HUVEC are relevant for different settings. HBMEC constitute the blood-brain barrier, whereas HUVEC are only relevant during pregnancy. It is known from murine models that IDO activity is of particular importance during pregnancy since IDO inhibition leads to spontaneous abortion (17, 18). Therefore, it is possible that IDO induction and activity is regulated in different ways in HBMEC and HUVEC.