Cytokine production and apoptosis among T cells from patients under treatment for Plasmodium falciparum malaria

Abstract

Available evidence suggests that Plasmodium falciparum malaria causes activation and reallocation of T cells, and that these in vivo primed cells re-emerge into the periphery following drug therapy. Here we have examined the cytokine production capacity and susceptibility to programmed cell death of peripheral T cells during and after the period of antimalarial treatment. A high proportion of peripheral CD3⁺ cells had an activated phenotype at and shortly after time of admission (day 0) and initiation of therapy. This activation peaked around day 2, and at this time-point peripheral T cells from the patients could be induced to produce cytokines at conditions of limited cytokine response in cells from healthy control donors. Activated CD8^hi and TCR-γδ⁺ cells were the primary IFN-γ producers, whereas CD4⁺ cells constituted an important source of TNF-α. The proportion of apoptotic T cells was elevated at admission and peaked 2 days later, while susceptibility to activation-induced cell death in vitro remained increased for at least 1 week after admission. Taken together, the data are consistent with the concept of malaria-induced reallocation of activated T cells to sites of inflammation, followed by their release back into the peripheral blood where they undergo apoptotic death to re-establish immunological homeostasis as inflammation subsides. However, the high proportion of pre-apoptotic cells from the time of admission suggests that apoptosis also contributes to the low frequency and number of T cells in the peripheral circulation during active disease.

INTRODUCTION

Acute Plasmodium falciparum malaria is associated with increased plasma levels of a wide range of cytokines [1–6] and markers of T cell activation and endothelial inflammation [7][8–13], but also T cell lymphopenia [14–16] and impaired ability of peripheral blood T cells to produce cytokines in vitro [17]. Initiation of drug treatment of acutely ill malaria patients causes a rapid increase in the frequencies and absolute numbers of CD3⁺, CD4⁺ and CD8⁺ cells in the peripheral blood [16,18]. This process changes the disease-induced lymphopenia to a transitory state of lymphocytosis, before homeostasis is re-established within a period of about 2 weeks.

We have hypothesized that these perturbations reflect the re-emergence in the peripheral circulation of previously sequestered in vivo activated T cells [16]. Under this hypothesis, T cells obtained shortly after initiation of antimalarial chemotherapy can be used to examine functional aspects of cells that have been engaged in the in vivo immune response to the infection. To this end we have examined the activation status, cytokine production capacity and susceptibility to apoptosis of T cells obtained at various time-points from Ghanaian P. falciparum malaria patients.

MATERIALS AND METHODS

Study population

All the children were admitted as in-patients to the Department of Child Health, Korle-Bu Teaching Hospital, Accra, Ghana with a diagnosis of P. falciparum malaria. Only patients that could be categorized as having cerebral malaria (CM) or uncomplicated malaria (UM), according to a set of strict inclusion and exclusion criteria described elsewhere, were admitted [5]. Patients categorized as having severe malarial anaemia were specifically excluded from the present study, as the necessary blood transfusions in these children affect frequencies and absolute numbers of T cells in the peripheral blood [16]. All patients were treated with standard regimens of chloroquine or artesunate [5] and all recovered fully from their infection. In addition to the malaria patients, we also included a group of age-matched, healthy children from a nearby community (Dodowa) [5].

The study was approved by the Ethical and Protocol Review Committee, University of Ghana Medical School and by the Ghanaian Ministry of Health.

Sample collection and processing

Peripheral blood samples were obtained on days 0 (admission), 1, 2, 4 and 21 (activation data), on days 0, 2 and 7 (apoptosis data) or on day 0 (serology) or day 2 only (cytokine production capacity). From the healthy control donors, only a single blood sample was obtained. Samples for analysis of T cell activation and cytokine production by flow cytometry were processed directly for assays on fresh cells within 4 h of sample collection. Plasma for serology was separated from the heparinized blood and stored at –20°C until analysis. Analysis of apoptosis was done on peripheral blood mononuclear cells (PBMC) isolated by gradient centrifugation and cryopreserved at –196°C until time of assay, as described previously [19].

Flow cytometry analysis

Expression of T cell activation markers. 100-μl aliquots of full blood were labelled with antibodies to CD3 (UCHT1; Dako, Glostrup, Denmark) and either CD69 (L78; BD PharMingen, San Diego, CA, USA) or HLA-DR (L243; BD PharMingen) for 20 min at room temperature. Erythrocytes were subsequently lysed (FACS Lysing Solution, BD PharMingen), the samples washed twice in PBS, and 10 000 events acquired on a FACScan flow cytometer (BD PharMingen). Data analysis was performed using WinList software (Verity, Topsham, ME, USA).

Intracellular cytokine production. 1-ml aliquots of full blood were incubated with monensin (1·5 μm; Sigma, St Louis, MO, USA), ionomycin (1 μm) and PMA (50 μg/ml) for various time periods, as indicated. Following this incubation, the cells were surface labelled with antibodies to CD3, CD4 (MT310; Dako), CD8 (DK25; Dako) or TCR-γδ(11F2; BD PharMingen) and the erythrocytes lysed as above. The cells were then washed twice in a freshly made saponin buffer (PBS/BSA/NaN₃ containing 0·1% (w/v) saponin (Sigma)) and intracellular cytokines labelled with anticytokine (IFN-γ, TNF-α or IL-10; BD PharMingen) antibodies for 30 min in the dark (4°C). Finally, the cells were washed twice in saponin buffer, twice in staining buffer, resuspended in the same buffer and analysed by flow cytometry as described above.

Apoptosis. PBMC were labelled sequentially with Annexin V (BD PharMingen) and 7-aminoactinomycin D (7AAD; Sigma) and analysed by flow cytometry as above. In some experiments, PBMC were preincubated with immobilized CD3 antibody for 18 h.

Plasma levels of soluble Fas and Fas ligand

Plasma levels of soluble Fas (sFas) and soluble Fas ligand (sFasL) were measured by ELISA (Naka-ku, Nagoya, Japan).

Statistical analyses and data presentation

Groups were compared by one-way or two-way anova followed by Tukey’s post-hoc test or by Student’s t-test for paired data. Values of P < 0·05 were considered significant.

Preliminary data analysis revealed that there were no significant differences between CM and UM patients in the parameters investigated, and consequently data from both groups are presented together. Results are presented as means and 95% confidence intervals.

RESULTS

T cells from malaria patients under drug cure have an activated phenotype

We and others have demonstrated previously that after initiation of chemotherapy the initial lymphopenia of acutely ill P. falciparum malaria patients is rapidly replaced by a transitory lymphocytosis that lasts for about a week before return to predisease homeostasis [16,18]. The first set of experiments was designed to determine in detail the time-course of the proportion of in vivo activated T cells in the peripheral circulation following drug treatment in our study population. Figure 1 shows the proportion and expression intensity of T cells expressing the early (CD69) and late (HLA-DR) activation markers from time of admission until 3 weeks later. Compared to expression on day 21, the proportion of CD3⁺ cells expressing CD69 was significantly higher (1-way anova, P < 0·001 followed by Tukey’s test, P < 0·05) on days 0–2, while the proportion of HLA-DR⁺ cells was significantly higher on days 1 and 2 (Fig. 1, main panels). By the same method, the expression level of CD69⁺ and HLA-DR⁺ cells was significantly increased on days 0–2 and on day 1, respectively (Fig. 1, inserts). Taken together, these results pointed to day 2 as the time of peak in vivo activation among peripheral T cells in our setting, and the cytokine production capacity among cells obtained at this time-point was next studied in more detail.

Fig. 1.

Ex vivo expression of CD69 (a) and HLA-DR (b) by CD3⁺ peripheral blood T cells obtained from 20 children at various points after admission to hospital with a diagnosis of P. falciparum malaria. The percentages of cells expressing marker (main panels) and the expression level of cells positive for marker (inserts) are shown as means and 95% confidence intervals.

Peripheral T cells from malaria patients under drug cure can be induced to produce cytokines by suboptimal in vitro stimulation

We could not detect cytokine-producing cells ex vivo among cells obtained at day 2, although a substantial proportion of these cells had an in vivo activated phenotype (Fig. 1). In these experiments, we assayed full blood samples following incubation at 37°C, 5% CO₂ for 30 min, in the presence of monensin but without in vitro stimulus.

As a consequence, we next compared the proportion of cytokine-producing CD3⁺ cells from patients and healthy control donors following stimulation of the cells by PMA/ionomycin for various periods of time. This was performed to identify the optimal time of stimulation allowing discrimination between responses of in vivo activated cells from the patients and resting cells from the control donors. Peak production of both TNF-α, IFN-γ and IL-10 occurred between 90 and 120 min of incubation. In the case of IFN-γ and TNF-α a substantially higher proportion of T cells obtained from patients expressed cytokine intracellularly compared to cells from healthy control donors (Fig. 2, main panels), and the difference between the two donor groups also peaked at this time (Fig. 2, inserts). For these two cytokines, both donor category and incubation time varied significantly (P < 0·001, two-way anova), with significant differences between donor categories at incubation times 60, 90 and 120 min (P < 0·01 in all cases, Tukey’s test). Only few (<5%) IL-10 containing T cells could be identified in either patients or controls at any time-point, and the differences between donor groups were not significant (data not shown).

Fig. 2.

Cytokine-containing CD3⁺ peripheral blood T cells from seven children under treatment (day 2) for P. falciparum malaria (•) and seven healthy, age-matched children from a nearby community (○) following in vitro stimulation by PMA/ionomycin for various periods of time. The percentages of cells (main panels) containing IFN-γ (a) and TNF-α (b) and the difference between patients and controls (inserts) are shown as means and 95% confidence intervals.

Different subsets of T cells have distinct cytokine production profiles

Based on the above findings (Fig. 2), we completed a third series of experiments to investigate in more detail the cell subsets producing cytokines after 90 min of in vitro stimulation. Using an FSC/SSC lymphocyte gate we could show that the large majority of cytokine-producing cells were T lymphocytes (Table 1). In fact, T cells were the only important lymphocyte source of IFN-γ and TNF-α. About half of all IFN-γ containing cells expressed CD8 at high levels, corresponding to TCR-αβ⁺ CD8⁺ cells. This was more than three times (paired t-test, mean 3·3×, 95% CI: 2·8×–3·9×, P < 0·001) the overall proportion of this subset among the peripheral lymphocytes (Table 1), and identified CD8^hi cells as an important source of IFN-γ. In line with this, a third of CD8^hi cells contained IFN-γ, compared to only about 10% of CD4⁺ cells (Fig. 3). In contrast, CD4⁺ cells dominated among TNF-α⁺ T cells, and a significantly higher fraction of TNF-containing cells were CD4⁺ compared to the overall percentage of these cells (paired t-test, mean 2·2 ×, 95% CI: 1·8×–2·6×, P < 0·001). Furthermore, twice as many CD4⁺ as CD8^hi cells contained TNF-α (Fig. 3).

Table 1.

Subset composition of peripheral blood lymphocytes ^* from 21 children under treatment (day 2) for P. falciparum malaria. Cells were stimulated by PMA and ionomycin in vitro for 90 min prior to analysis. Means and 95% confidence intervals are shown

	CD4	CD8**	TCR-γδ
All cells	31·0 ± 3·9	17·1 ± 2·7	22·4 ± 4·1
IFN-γ+ cells	24·3 ± 4·8‡	51·7 ± 5·2‡	35·5 ± 6·2
TNF-α+ cells	62·6 ± 6·0‡	17·9 ± 5·4‡	9·1 ± 2·1

^*Lymphocytes were identified by FSC/SSC scatter signal.

^**Only CD8 ^hi cells were included, to exclude TCR-γδ cells expressing CD8 at low density [48].

^†Significantly different from percentage of all cells in subset (paired t-test, P < 0·005).

Fig. 3.

Proportion of cytokine-containing cells within different subsets of T cells obtained from 21 children under treatment (day 2) for P. falciparum malaria. Cells were stimulated by PMA and ionomycin in vitro for 90 min prior to analysis. Means and 95% confidence intervals (a) and examples of cytokine profiles (c–d) are shown). □, IFN-γ; ▪, TNF-α; , IL-10.

TCR-γδ⁺ cells were over-represented among IFN-γ containing cells (paired t-test, mean 1·8 ×, 95% CI: 1·4 ×–2·3 ×, P < 0·001), and about half of the TCR-γδ cells contained IFN-γ, a higher proportion than either CD4⁺ or CD8^hi cells (Fig. 3 and two-way anova, P < 0·001 followed by Tukey’s test, P < 0·05).

Only few IL-10 containing T cells could be identified. All of these were CD4⁺ cells (data not shown) and less than 10% of CD4⁺ cells contained IL-10 after 90 min of stimulation (Fig. 3).

Lymphocytes from malaria patients express markers of apoptosis and are susceptible to activation-induced cell death in vitro

As shown in Fig. 4a, a higher proportion of cells obtained from the patients early during antimalarial chemotherapy (days 0 and 2) were pre-apoptotic (Annexin V⁺, 7AAD-negative) compared to cells obtained on day 7 (one-way RM-anova on ranks, P < 0·006). Similarly, the levels of necrotic, 7AAD⁺ cells were significantly higher on day 0 compared to day 7 (P < 0·007). Pre-incubation of cells with immobilized CD3 antibody to activate the T cells in vitro did not change the proportion of apoptotic cells (Fig. 4b), but the proportion of necrotic cells was markedly higher at later than earlier time-points (P < 0·007), indicating an increasing susceptibility to activation-induced cell death with time.

Fig. 4.

Spontaneous (a) and anti-CD3 induced (b) apoptosis (Annexin V⁺, 7AAD-negative; filled squares) and cell death (7AAD⁺; open squares) of lymphocytes obtained from 25 P. falciparum malaria patients at various time points after admission and initiation of antimalarial chemotherapy (day 0). Means and 95% confidence intervals are shown ▪, Annexin V⁺; □, 7AAD⁺.

Plasma levels of soluble Fas ligand but not soluble Fas are elevated in malaria patients

Plasma levels of soluble Fas ligand (sFasL) were higher in the P. falciparum patients at admission (day 0) compared to plasma from a group of healthy, age-matched children from a nearby community (Student’s t-test, P < 0·001) (Fig. 5). In contrast, plasma levels of soluble Fas (sFas) in the patients were not significantly different from control levels (Student’s t-test, P =0·31) (Fig. 5).

Fig. 5.

Levels of soluble Fas (a) and soluble FasL (b) in plasma obtained at admission (day 0) from 30 P. falciparum malaria patients (▪) and 10 healthy, age-matched control donors (□). Means and 95% confidence intervals are shown.

DISCUSSION

Our study shows that a substantial proportion of peripheral T cells obtained early after initiation of chemotherapy in children with P. falciparum malaria have an activated phenotype (Fig. 1). In combination with previous data [9,11,16,20], this finding is consistent with the hypothesis that these cells represent previously sequestered cells that have been engaged in the immune response to the infection, and which are returning to the peripheral circulation as the result of treatment. T cells obtained 2 days after initiation of antimalarial chemotherapy thus appear useful in examining functional aspects of cells that have been engaged in the in vivo immune response to the infection at sites not accessible for investigation.

Under conditions where only a limited response from cells obtained from healthy, age-matched controls was observed (Fig. 2), we found that a large proportion of CD8^hi cells, presumably activated in vivo as a result of the P. falciparum malaria episode, produced IFN-γin vitro (Fig. 3 and Table 1). This finding is in line with previous studies pointing to substantial in vivo activation of this cell subset [21,22]. Furthermore, data from several studies implicate CD8⁺ cells in the naturally acquired immune response to P. falciparum infection [23–25], which is mediated at least partially through IFN-γ[26–28]. Our data extend these observations by showing that peripheral CD8^hi cells from P. falciparum malaria patients are poised for IFN-γ production. In addition, γδ T cells appear to be an important source of IFN-γ (Fig. 3 and Table 1). This subset of T cells is increased in healthy, and particularly in malarious, individuals from endemic areas compared to healthy adults from non-endemic areas [29–31]. The role of γδ T cells in the immune response to malaria infections remains unclear. However, the evident malaria-induced γδ T cell perturbation, which in endemic populations is dominated by Vδ1⁺γδ T cells [31,32], does not appear to be driven by particular antigens [31] and resembles the γδ T cell response in HIV infection [33]. Taken together, the available evidence suggests that Vδ1 γδ T cell have an autoregulatory function rather than responding directly to invading microorganisms [34].

Malaria is associated with an increase in TNF production [2,3,35–37]. Although myeloid cells are generally thought to be the main producer of this cytokine, our data suggest that CD4⁺ cells are another important source of TNF-α in P. falciparum malaria (Fig. 3 and Table 1), in line with recent in vitro findings [38,39].

The role of apoptosis in the malaria-induced perturbation of T cell subsets and the subsequent return to homeostasis has been debated [40,41]. We have not been able to detect apoptotic cells using the methodology employed in the study from Senegal (unpublished data), but found evidence of transiently increased frequencies of apoptotic cells in our malaria patients using an alternative approach (Fig. 4). This is in contrast to the persistent changes seen in the earlier study [40]. In vitro activation by CD3 caused a proportion of the T cells to undergo activation-induced cell death (AICD), and this proportion was higher using cells obtained at later compared to earlier time points (Fig. 4). Malaria-induced apoptosis of lymphocytes has been suggested to be the result of direct parasite contact [42,43] and in addition activated cells express both Fas (CD95) and FasL, which may induce apoptosis [44,45]. We found increased plasma levels of sFasL at admission (Fig. 5), which rapidly normalized following treatment. Soluble FasL is produced by activated lymphocytes and can protect cells from apoptosis by inhibiting Fas/FasL ligation [46]. Thus, an alternative explanation for the increased susceptibility to AICD during convalescence is unmasking of Fas as levels of sFasL subside. Soluble Fas may act in a similar way [47], but we could not detect increased plasma Fas levels in the patients (Fig. 5).

In conclusion, we have shown that many circulating T cells obtained during antimalarial chemotherapy have an activated phenotype, are primed for cytokine production and are susceptible to AICD. The data are consistent with the hypothesis that these cells include previously sequestered cells that have been engaged in the immune response to the infection, and are re-emerging into the peripheral circulation as the result of therapy. Finally, our observations suggest that apoptosis and AICD are involved in the return to immunological homeostasis during convalescence from acute malaria, but the presence of substantial levels of pre-apoptotic cells at the time of admission suggest that apoptosis also contributes to the lymphopenia characteristic of this disease.