Alterations in early cytokine-mediated immune responses to Plasmodium falciparum infection in Tanzanian children with mineral element deficiencies: a cross-sectional survey

This study was conducted in a lowland area around Segera village (S 05° 19.447', E 38° 33.249'), Handeni District, north-eastern Tanzania, in May-July 2006. Malaria is highly endemic in this area, with virtually all infections being due to P. falciparum. The residents in the study population mostly comprise poor farmer families growing maize and cassava for subsistence use. Such populations are prone to deficiencies of zinc and iron because they have cereal-based diets that are rich in natural dietary constituents that inhibit the absorption of these trace metals [17]. At the time of our study, only one health centre in Segera was available to serve all of the surrounding area. The study was approved by Ethics Review Committees in The Netherlands and Tanzania (reference numbers for KCMC and the National Health Research Ethics Review sub-Committee: 094 and NIMR/HQ/R.8a/VolIX/540, respectively). Informed consent was obtained from community leaders and local government officials, and from parents or guardians.

A census list was made with all resident children aged 6-72 months in the study area. Using this list, 16 children were randomly selected from 19 communities, resulting in a total of 304 subjects. Further details are provided elsewhere [18].

All children were examined by a clinical officer, who also measured axillary temperature by electronic thermometer. Subjects were eligible when they had no fever, and showed no signs of other severe disease or severe malnutrition (weight-for-height z-score below -3 SD). Plasmodium infection was detected both by microscopy and rapid immunochromatographic assay (Vista Diagnostics Int. Kirkland, WA, USA, based on antibodies developed for OptiMAL test by Flow Inc., Portland, OR, USA). Although this dipstick test cannot be used to determine the duration of infection, it detects lactate dehydrogenase (pLDH) produced by live parasites only, either P falciparum or any human Plasmodium species [19, 20]. For blood samples with > 50 P. falciparum parasites/mL (0.001% parasitaemia), it has been found that the OptiMAL assay has a sensitivity of approximately 96% [20]. Venous blood (6 mL) was collected in containers suitable for mineral element analysis with sodium heparin as anticoagulant (Becton-Dickinson, Franklin Lakes, NJ). Immediately upon collection, the cap was sprayed with ethanol and allowed to dry; approximately 1.3 mL blood was then drawn by sterile syringe. This aliquot was centrifuged and plasma samples were stored and transported to The Netherlands at -80°C for subsequent measurement of mineral element concentrations. The remainder of the blood sample was kept at 20-25°C during transport the same day to the laboratory in Moshi, at approximately 300 km distance, for collection of additional plasma and PBMCs. Children were treated for common childhood infections and anaemia according to guidelines of Tanzanian Ministry of Health.

Plasma samples were diluted 20 times in milliQ [21], and concentrations of zinc, magnesium and copper were measured by inductively-coupled plasma atomic emission spectrometry (ICP-AES) (Vista Axial, Varian, Australia). To determine variability in outcomes, measurements were replicated five times. With mean values set at 100%, measurements varied between 97% to 102% for zinc, 99% to 102% for magnesium, and 97% and 102% for copper. Because we found no evidence for copper deficiency as assessed by plasma copper concentrations < 7.1 μmol/L (unpublished data), only the results for zinc and magnesium are reported in this paper.

PBMCs were isolated by Ficoll density gradient centrifugation, cells were transferred to 10% v/v DMSO in fetal calf serum, cooled at -1°C/minute in an isopropyl-loaded device (Nalgene, Rochester, NY, USA) and preserved in liquid nitrogen [22]. For a 16-h period during transport to Wageningen University, The Netherlands, the PBMCs were kept on dry ice, and immediately thereafter stored again in liquid nitrogen until stimulation experiments.

Routinely prepared asexual stage parasitized erythrocytes were obtained from the Department of Medical Microbiology, Radboud University, Nijmegen [23]. Briefly, human O and rhesus-negative erythrocytes from healthy blood donors (Sanquin, Nijmegen, The Netherlands) were cultured in medium to which live P. falciparum parasites (NF54 strain) produced in a continuous culture were added [23, 24]. After two to four days, when ~8-10% of erythrocytes were parasitized by asexual Plasmodium stages, the culture was concentrated by centrifugation at 625 × g for 5 min; parasitized erythrocytes were separated on a 67% Percoll gradient as reported elsewhere [25] and washed twice in phosphate-buffered saline (PBS). Purified parasitized erythrocytes were preserved at a concentration of approximately 15 × 10⁷/mL in 13% glycerol/PBS in a freezing container at -80°C. Glycerol (50% w/v) was added to the parasitized erythrocytes to avoid mechanical damage of the cells through ice formation. Unparasitized erythrocytes were processed similarly but without adding parasites to serve as a control. Both in parasitized and unparasitized erythrocyte cultures, we confirmed the absence of mycoplasma contamination by polymerase chain reaction. Both parasitized and unparasitized erythrocytes were counted by flow cytometry, and compared regarding their size and internal complexity to the counting beads and PBMCs. Aliquots were made and stored at -8°C until needed for PBMCs stimulation.

Malaria antigens differ in their capabilities to stimulate PBMCs: intact parasitized red blood cells (pRBC) are capable of inducing more rapid and intense pro-inflammatory responses from PBMCs than freeze-thaw lysates of P. falciparum [26]. To simulate in vivo malaria-specific responses as closely as possible, P. falciparum pRBC were used, with an adapted protocol for stimulation of PBMCs by Jeurink et al [22]. In brief, PBMCs were cultured at 10⁶ cells/well in sterile polystyrene 48-well plates with flat-bottom wells (Corning Inc, Corning, NY, USA). Based on initial optimization experiments, aliquots of pRBC were thawed and cultured with PBMCs in IMDM with glutamax containing Yssel's supplements [27] with 2% human AB serum, 1% penicillin/streptomycin and 1% fungizone (Gibco-BRL), at a PBMCs:pRBC ratio of 1:2. PBMCs were also cultured under similar conditions with unparasitized erythrocytes (uRBC) (2 × 10⁶ cells/well) as a negative control, and with soluble antibodies to CD3 and soluble antibodies to CD28 (Cat. No.555336 and 555725, Becton-Dickinson, Alphen aan den Rijn, The Netherlands) as a positive control. Monoclonal anti-CD3 and anti-CD28 antibodies provide co-stimulatory signals and polyclonal stimulation required for maximal proliferation of T lymphocytes [22, 28]. Cell culture plates were incubated at 37°C in a humidified atmosphere containing 5% CO₂. After PBMCs culturing for 1 day, we aspirated 75 μL of the supernatant per well to measure cytokine concentrations.

Concentrations of IL-1β, IL-10, IL-12p70 and TNF were determined on a FACSCanto II flow cytometer by cytometric bead array system and analysed with FCAP software (all from Becton-Dickinson).

Data were entered and analysed using SPSS for Windows (version 15.0. SPSS Inc., Chicago, IL, USA). Zinc deficiency and low zinc status were defined as plasma zinc concentrations < 9.9 μmol/L and < 10.7 μmol/L, respectively. Low magnesium status was defined by magnesium concentration < 750 μmol/L [29, 30]. Iron deficiency anaemia was defined by co-existing iron deficiency (plasma ferritin concentration < 12 μg/L) and anaemia (haemoglobin concentration < 110 g/L). Fisher's Exact Test was performed to determine the association between inflammation (CRP levels) and sex, age, malaria status as well as nutritional status. Cytokine concentrations were ln-transformed to obtain normally-distributed values. Group differences in these values were analysed assuming t-distributions. Interactions between malaria and micronutrient indicators were assessed using multiple linear regression models on log-transformed cytokine data; the resulting effect sizes were exponentiated and expressed as percentage values. Linear regression analyses were also carried out to explore the associations between IL-1β, TNF and IL-10, and to what extent these associations were influenced by nutrient status and malaria infection status. The association between concentrations of TNF and IL-10 were considered as a measure of balance between the pro-inflammatory responses and the regulatory response. The analyses of the cytokine responses to pRBC are reported. As expected, the average response to uRBC (negative control) was less than to pRBC, whereas the average response to CD3/CD28 (positive control) was higher. Correction for these responses does not change the estimates of the associations between nutrient status and cytokine responses, or between malarial infection status and cytokine responses.

Peripheral blood was collected from 135 boys and 169 girls; these had similar age distributions. The following prevalence values (n) were found: low zinc status: 63.1% (188); zinc deficiency: 48.3% (144); low magnesium status: 65.1% (194); iron deficiency anaemia: 9.4% (26); malaria: 46.1% (140). Malaria and age were associated with inflammation (determines as CRP levels); however, there was no evidence that inflammation was associated with zinc deficiency, magnesium deficiency or iron deficiency anaemia (Fisher's Exact Test). Detailed characteristics of the study population by malarial infection status are summarised in Table 1. In addition, the associations between nutrient status and supernatant cytokine concentrations, and between malaria infection status of the child at the time of blood collection and supernatant cytokine concentrations, following 24 h of PBMCs stimulation with P. falciparum-infected erythrocytes are also summarized (Figure 1). Adjustment for age class, sex and/or magnesium deficiency did not lead to marked changes in the associations between zinc deficiency and supernatant cytokine concentrations shown in Figure 1; conversely, adjustment for age class, sex and/or zinc deficiency did not lead to marked changes in the associations between magnesium deficiency and those supernatant cytokine concentrations.

When analysing IL-10 concentrations, all individuals with IL-10 concentrations below the detection limit were excluded. In some instances, differences in cytokine concentrations between nutrient replete and deficient children (Figure 2) seemed to depend on malarial infection status at the time of blood collection (Table 2). The profile of supernatant cytokine concentration appeared different between subjects with deficiencies in zinc, magnesium and with iron deficiency anaemia. In the absence of malaria infection at the time of blood collection, zinc deficiency was associated with marginal reductions in concentrations of TNF, IL-1β and IL-10. Amongst donors with malaria infection at the time of blood collection, zinc status was not associated with altered concentrations of IL-1β or IL-10, but low plasma zinc concentrations were associated with an increase in TNF concentration by 74% (11% to 240%, 95% CI). Malaria infection at the time of blood collection seemed to determine the magnitude of the association between low plasma zinc concentration and TNF concentration (9% reduction in children without malaria, as compared to 74% increase in their peers with malaria; although the statistical evidence for this difference was weak (P = 0.15).

Magnesium deficiency, on the other hand, was associated with increased concentrations of IL-10; this increase was 46% in children without malaria, as compared to only 6% in their peers with malaria (Figure 2). Low magnesium concentrations seemed associated with reduced concentrations of TNF and IL-1β by -25% (95% CI: -64% to 55%; P = 0.79) and -44% (-70% to 6%; P = 0.13), respectively, in children with malaria infection at the time of blood collection, although these differences may have been due to chance. These results are a reverse of the situation in zinc deficiency. The numbers of individuals with both iron deficiency anaemia and malaria (Table 1) were too low to compare groups meaningfully.

No evidence was found that the associations between concentrations of TNF and IL-10 depended on zinc, magnesium or malaria status at time of blood collection, as indicated by differences in slopes of 17% (-44% to 147%; 95% CI, P = 0.67) for zinc status, 10% (-47% to 127%; 95% CI, P = 0.80) for magnesium status, or 3% (-24% to 39%; 95% CI, P = 84) for malaria (Figure 3). There was a tendency, albeit not significant, that iron deficiency anaemia (IDA) influenced the relationship between concentrations of TNF and IL-10, as indicated by the difference between slopes of 119% (35% to 637%; 95% CI, P = 0.20).

Additional linear regression analyses (Figure 4) showed evidence that zinc status influenced the association between concentrations of IL-1β and IL-10, as indicated by differences in slopes of 118% (4% to 359%; 95% CI, P = 0.04). There was no evidence that malaria infection influenced the association between concentrations of TNF and IL-1β (Figure 4, malaria panel), as indicated by a difference in the slopes of regression lines of 9% (-13% to 35%; P = 0.47). In summary, these results show no evidence of influence on associations among innate cytokines, by the various conditions of micronutrient and malaria status at time of blood collection except for zinc status, for which there was some evidence that it influenced the association between IL-1β and IL-10. There was no evidence of an influence of micronutrient status and malaria on associations in other relationships. There were insufficient cases in all groups to explore and meaningfully compare the associations between IL-12 and other cytokines.

The biochemical data showed that most children in this study had nutrient deficiencies, particularly in zinc and magnesium and to a lesser extent iron deficiency anaemia. Zinc deficiency was associated with increased TNF responses in children with malaria infection at the time of blood collection but not in those without infection. TNF is a pro-inflammatory cytokine resulting in pathology if not properly regulated. In children with malarial infection, zinc deficiency was associated with increased production of IL-1β and IL-10, even if this increase did not bring the levels to those reached by individuals in the non-infected group. This is important because IL-10 is required to limit the production of pro-inflammatory cytokines, so that they do not lead to pathological consequences [31]. The low production of IL-10, however, could be due to the fact that the cytokine is said to be produced late (in vivo) following infection relatively to the innate cytokines. The initial production of TNF could also be the triggering factor by feedback mechanisms for production of IL-10 although Ramharter et al [32] reported increased responsiveness of in vivo primed cells as compared to malaria-naïve cells, with a tendency towards increased production of TNF. This can possibly explain the difference between subjects who were exposed or non-exposed at the time of blood collection, in response to in vitro stimulation in our study. These results show possible alterations in innate cytokine production particularly TNF and IL1-β due to the reported impaired macrophage functions and NK-cells activity in zinc deficiency [1, 2, 13, 33]. Interaction between these cells leads to the production of innate cytokines in the early stages of infections.

The relatively higher cytokines levels in individuals with malarial infection as compared to their uninfected peers (Figure 2), however, can be explained by the priming of the immune system by malaria. Exposure of T cells to a plethora of Plasmodium antigens leads to priming, so that these cells during subsequent exposures even with subsets of these antigens can produce greatly increased amounts of IFN-γ. This cytokine is necessary for up-regulation of production of TNF and other pro-inflammatory cytokines by monocytes, but also Th subsets and even natural killer cells, in malaria infection [26, 34]. The cellular source of these abundantly produced cytokines, like IL-1β and IL-10 remain to be established in future studies, as besides monocytes/macrophages and B-cells, also various T-cell subsets will be involved. The increase in innate cytokine production in zinc-deficient individuals with malarial infection can be the result of a shift in activated monocytes towards a pro-inflammatory immune response, associated increased levels of TNF and IL-1β, due to zinc deficiency in combination with prior priming of these cells due to previous exposure to malaria, as has been suggested before [35]. The initial contact with the pathogen directs towards production of pro-inflammatory cytokines to limit infection. Loharungsikul et al [36] proposed Toll-like receptors (TLRs) to play a role in innate immune recognition in which the differential expression of TLRs on antigen presenting cells (APCs) could be regulated by the P. falciparum parasite. This could account for the increase in levels of TNF in malaria-positive individuals regardless of micronutrient status (Figure 1). Glycosylphosphatidyl inositols (GPIs) that anchor P. falciparum merozoite surface protein 1 (MSP1) and merozoite surface protein 2 (MSP2) were described to be the pathogen associated molecular patterns (PAMPs) preferentially recognised by TLR-2 and TLR-4 [37]. The recognition and the interaction between these molecular patterns signal the induction of pro-inflammatory cytokine production. In addition, it is possible that parasite DNA attached to malarial pigment (haemozoin) produced in the course of infection further activates the innate immune response through TLR-9 engagement [38]. The expression of TLRs has been found to differ between malaria-infected and uninfected individuals, with higher expression being observed in infected patients [24, 39]. These recent studies have further indicated TLR-2 to be highly expressed in mononuclear cells, particularly monocytes of P. falciparum-infected children and that TLR-2 are well responsive following stimulation with pRBCs resulting into stronger signals with consequential change in cytokine production profiles.

Unfortunately, the design of this study did not allow to ascertain the duration of infection at the time of blood collection. The children investigated may have been infected for some time, and may not have relied on the innate cytokines measured to control infection at the time of blood collection. It is also important to note that this concerned a cross-sectional study, that samples that were collected in a single time point, and that cytokines measured had accumulated in supernatant following 24-h stimulation of cultured cells. The design of our study does not allow time-dependent production of cytokines, or to establish specific cell subsets are sources of the cytokines measured.

The interesting result in this study is that the impact of magnesium deficiency on early cytokine responses followed a different profile from that observed with zinc status. Magnesium deficiency seemed to be associated overall with low TNF concentrations, low concentrations of IL-1β and higher concentrations of IL-10 in uninfected but not infected donors (Figure 2). Low levels of pro-inflammatory cytokines in malaria are critical because they reduce the ability of the initial innate immune response to limit infection. These results imply that magnesium deficiency directs early cytokine responses towards anti-inflammatory rather than pro-inflammatory cytokine responses, although further studies are still needed to confirm this hypothesis. The significantly increased IL-10 and variable alteration in levels of TNF and IL-1β in both malaria-negative and malaria-positive subjects with magnesium deficiency may explain the imbalance in cytokine production as a result of magnesium deficiency modulated by malaria status.

Methodological differences may explain contradictions between our findings and those from previous studies [7, 9, 40]. Parasitized erythrocytes were used to simulate the in vivo infection, whereas others used mitogens, lipopolysaccharides (LPS), phytohemagglutinin (PHA) and polyclonal stimulation. In addition, we used Ficoll-isolated PBMCs that had been stored for several months under frozen conditions, whereas whole blood stimulated within 15 minutes of collection was also used in some of the previous studies. McCall et al [24] stimulated freshly prepared PBMCs from adult naïve volunteers ex vivo with P. falciparum antigens. These findings suggest that zinc and other micronutrients can protect against malaria infection by a different means such as targeting specific pathogenic processes of infection in vivo [41]. Nevertheless, the idea that zinc can also reduce production of pro-inflammatory cytokines by inhibiting signal transduction in monocytes in healthy human subjects [42], particularly IL-1β and TNF [2, 43, 44], should be further explored. The latter idea is also supported by in vitro studies [45, 46] in other conditions than malaria.

The results from this study and those conducted by others [47, 48], IL-12 concentrations were below the detection limits. The most probable reason is the time required for maximal priming of pathogen recognition receptors (e.g. TLRs) on PBMCs by P. falciparum-parasitized erythrocytes. McCall et al [24] have shown that pro-inflammatory priming effects of P. falciparum require up to 48 hours to develop maximally, whereas we measured cytokines after 24 hours of stimulation. This priming is lacking in our culture system despite the reported poor in vitro induction of IL-12 by P. falciparum [49]. IL-12 levels obtained in vitro from stimulated monocytes and macrophages are generally low and zinc deficiency is reported to be associated with further decreased IL-12 production [50]. Early IL-12 activity is also liable to suppression by transforming growth factor (TGF)-β [51, 52] that has been reported to variably influence and result in weak IL-12 activation and production, at least in vivo. Most of our donors responded towards production of IL-1β rather than TNF and IL-10. This is interesting since although different arguments reveal the pathological effect of IL-1β on cerebral malaria and severity of the disease in children [53], IL-1β together with other pro-inflammatory cytokines like IFN-γ and IL-6 is said to be protective against malaria by inducing parasite killing by monocytes, macrophages and neutrophils [54]. Production of IL-1β is induced by direct interaction between zinc and monocytes through activation of interleukine-1 receptor associated kinase (IRAK) which is dose-dependent [44]. Lower in vivo zinc levels, partially inhibit IRAK leading to diminished but not completely inhibited normal T-cells IL-β response. Results from this study may also reflect that stimulation of cryopreserved PBMCs by pRBCs results in a gradual production of innate cytokines preceded by IL-1β from the monocytes.

The association between IL-1β and IL-10 was found to be influenced by zinc status (Figure 4). The two innate cytokines TNF and IL-1β are a prerequisite in early responses to malaria infection and IL-10 is an important regulatory cytokine affected by nutrient deficiencies and malaria infection status. This is critical under tropical situations where both micronutrients deficiencies and malaria prevail, posing a challenge to the early immune response to infections.