Plasmodium falciparum infection of the placenta affects newborn immune responses

Abstract

The effects of exposure to placental malaria infection on newborn immunological responses, in particular Th1/Th2 cytokines and antigen-presenting cell (APC) function, were compared between cord blood mononuclear cells (CBMC) from parasitized and non-parasitized placentas of Gambian women. Cells were analysed in vitro for their ability to respond to mitogens [phorbol myristate acetate (PMA)/ionomycin, phytohaemagglutinin (PHA)], a malaria-unrelated test antigen [purified protein derivative of Mycobacterium tuberculin [purified protein derivative (PPD)] and Plasmodium falciparum schizont extracts. Mitogens induced strong proliferation and secretion of high concentrations of both IL-13 and sCD30 in CBMC from both groups. Conversely, significantly lower amounts of IFN-γ were induced in the parasitized group in response to low doses of PHA. Protein antigens induced very low amounts of all tested cytokines, in particular IFN-γ. However, a significantly higher release of sCD30 was observed in response to schizont extracts in the parasitized group. Addition of LPS to activate APC to low doses of PHA or schizont extracts increased the IFN-γ production in both groups but levels remained lower in CBMC from the parasitized group. This result correlates with the lower production of IL-12 found following lipopolysaccharide (LPS) stimulation in this group. Taken together, these data show that placental infection with P. falciparum affects Th1 differentiation and sCD30 priming of neonatal lymphocytes and that the probable mode of action is via APC.

INTRODUCTION

Malaria is a public health problem in more than 90 countries, and 40% of the global population is exposed to this potentially fatal disease [1,2]. Pregnant women are highly susceptible to malaria, largely because the transient depression of their cell-mediated immunity that allows retention of the fetal allograft also interferes with resistance to infections [3,4]. Placental malaria stimulates the production of inflammatory mediators, shifting immune responses away from the Th2 cytokines associated with healthy pregnancy towards Th1 type cytokines [5]. The prevalence of placental malaria varies from 30 to 60% in many malaria-endemic areas, including The Gambia, and in this region is associated with increased risk of abortion, stillbirths, low birth weight and infant mortality [6]. Several epidemiological studies have demonstrated that these effects are most marked in primigravidae [7,8], probably because specific antibodies to parasites able to cytoadhere to human syncytiotrophoblasts develop only after several pregnancies [9,10].

Maternal Plasmodium falciparum (Pf) infection can also impact on the immunological responses of the newborn. The Pf exoantigen, GLURP, has been detected in cord blood indicating transplacental passage of malarial antigens [11], and priming for Pf antigens has been reported in human cord blood lymphocytes from neonates born in areas holoendemic for malaria [12,15]. In a mouse-model system, neonatal exposure to major T cell sites of Pf circumsporozoite protein also induces specific T cell tolerance [16], and vaccination using formalin-fixed malaria parasites fails in mice born to immune mothers, due apparently to suppression by maternally derived IgG [17]. Maternal malaria may thus lead to antigen-specific tolerance or priming, but more generally placental infection may also affect the Th2 bias found usually in newborn infants. A well-described facet of this Th2 bias is the weak Th1 differentiation of the newborn, compared to the adult, when challenged by protein antigens [18–20]. This has been ascribed to the immaturity of the neonatal immune system and has also been attributed to a functional defect of antigen-presenting cells (APC) and impaired IL-12 production by dendritic cells (DC) and monocytes [21], rather than being ascribed to T cells directly [18,19]. However, the impact of Pf infection on the Th1/2 balance of newborn immune responses has been little studied. In humans allergens may cross the placenta, inducing the differentiation of fetal T lymphocytes into Th2 type cells [22], and responses to BCG vaccination are reduced in infants whose mothers had helminthic infections during pregnancy [23]. Such Th2 priming was also reported for HIV^– infants born to HIV⁺ mothers [24] and, in this instance, appeared linked to inhibition of IL-12 production by APC, even in the absence of vertical transmission [25]. Thus maternal HIV infection may increase the Th2 bias usually present in newborn infants. In addition, the observation that Pf inhibits DC maturation [26] led us to hypothesize that placental Pf infection may further reduce IL-12 production in newborns, leading to decreased IFN-γ levels and affect the Th1/Th2 balance. To substantiate these hypotheses, we investigated the effect of active Pf infection in the placenta on immune responses of cord blood mononuclear cells (CBMC) collected from pregnant women enrolled at delivery in a peri-urban village in The Gambia. We measured specifically IL-12, IFN-γ and IL-13 responses and also included sCD30, as this has been shown to be related to the T helper/cytotoxic 2-type immune responses [27].

MATERIAL AND METHODS

Population

In collaboration with the Gambian Ministry of Health, a study was undertaken in the peri-urban village of Sukuta, 11 km from the MRC laboratories in Fajara, during September 2001–January 2002. The catchment area of the Sukuta health centre has about 20 000 inhabitants and there are approximately 800 deliveries each year. A field assistant was posted to the health centre to recruit primiparous and multiparous pregnant women into the study, administer a simple questionnaire, prepare and examine malaria slides, measure packed cell volume (PCV) and collect cord blood. After obtaining written informed consent, up to 50 ml of cord blood was collected upon delivery. Active Pf infection was assessed by microscopic examination of Giemsa-stained thick smears of cord blood, mother's venous blood and imprints of the placenta. All slides were read twice by experienced microscopists and discrepancies resolved by a third reader. One hundred fields at ×1000 magnification were examined before declaring a slide negative (limit of detection, two parasites/µl) [28]. The study was approved by the joint Gambian Government/MRC Ethics Committee.

Parasite cultivation and schizont collection

Pf culture strains Pf164 and K1 [29] were maintained in vitro by standard techniques [30,31]. Uninfected red blood cells (RBC) of group ‘O’ donors were collected into acid citrate dextrase ( ACD) [0·6% disodium hydrogen citrate, 0·4% dextrose (d-glucose), Sigma, St Louis, MO, USA]. Washed ‘O’ cells were used for subculturing and fresh RBC were added to cultures at least every 7 days. Appropriate amounts of infected cells were added to uninfected ‘O’ cells to give a desired parasitaemia and the culture flask(s) were incubated in a candle jar at 37°C. The medium was changed daily and a smear made to determine the parasitaemia and parasite stage(s). When there were sufficient parasites, cultures were washed and schizonts were concentrated on 72% Percoll (Sigma), resuspended at 20% PCV, frozen and thawed at least four times and protein concentration determined by optical density. Extracts were titrated with peripheral blood mononuclear cells (PBMC) from semi-immune adults. An optimal induction of activation assessed by proliferation and IFN-γ production by PBMC was obtained with 10 µg/ml. Extract from uninfected RBC were used at the same concentration as a negative control. Parasite cultures were tested at regular intervals and found free of contamination with Mycoplasma spp. using a polymerase chain reaction (PCR)-based mycoplasma detection kit (American Type Culture Collection; ATCC, Manassas, USA).

Measurement of cytokine production in vitro

Cord blood mononuclear cells (CBMC) were isolated by density centrifugation of heparinized blood on Lymphoprep (Nycomed, Oslo, Norway), washed with RPMI-1640 (Sigma) and resuspended in RPMI supplemented with 20 µg/ml gentamycin (Sigma), 2 mm l-glutamine (Sigma) and 10% AB human serum (Sigma).

CBMC (triplicate of 2·10⁵ cells/200 µl) were cultured for 3 days in the presence of phorbol myristate acetate (PMA)/ionomycin, 50 ng/ml; ionomycin 0·5 µg/ml; Sigma Chemicals, UK), phytohaemagglutinin (PHA) (PHA-L, 2·5 µg/ml, Sigma), PHA and lipopolysaccharide (LPS) from Escherichia coli (1 µg/ml, serotype: 0128:B12, Sigma), or cultured overnight with LPS alone (1 µg/ml).

CBMC were also cultured for 6 days in the presence of the purified protein derivative of Mycobacterium tuberculosis [purified protein derivative (PPD) RT49, 10 µg/ml, Statens Serum Institut, Denmark], schizont extracts of Pf strains (Pf164 and K1) or an extract of uninfected RBC from the same preparation as the negative control, all at 10 µg/ml. In parallel, the CBMC were preincubated with schizont extracts for 2 h before adding LPS (1 µg/ml) and culturing for 6 days.

Activation and cytokine profiles raised in response to these stimuli were assessed using proliferative assays and/or ELISA. IFN-γ, IL-12, IL-13 cytokine and soluble sCD30 release were measured to assess the effect of placental Pf infection on the cytokine profile.

IFN-γ, ΙL-12, sCD30 and IL-13 levels in culture supernatants were determined using ELISA kits from Biosource (IFN-γ, IL-12; BioSource Europe, Fleurus, Belgium), MedSystems (sCD30; MedSystems Diagnostics GmbH, Vienna, Austria) and Diaclone (IL-13: Diaclone, Besançon, France), following the manufacturer's instructions.

Proliferation assays

Methyl-[³H]thymidine (1 µCi/well, Amersham Life Science, UK) was added on day 3 or day 6 for the final 16–18 h of culture to assess cell proliferation. Thymidine incorporation was measured by liquid scintillation using a Betaplate reader (LKB1205, Turku, Finland). Proliferative responses were expressed using a stimulation index calculated from triplicate values as the geometric mean counts per minute (cpm) stimulated wells/mean cpm negative control wells.

Flow cytometric analysis

CBMC were analysed either ex vivo or after 18 h stimulation with LPS (1 µg/ml). For immunophenotyping, CBMC were washed in PBS supplemented with 0·5% BSA and 10 mm NaN₃ and incubated for 30 min at 4°C with murine MoAbs. Cells were stained for cell surface markers CD3, CD4, CD8, CD45-RA, CD45-RO, CD16, γδ-TCR, CD14 and CD19. HLADR and CD40 were also analysed on unstimulated and LPS-activated CBMC. CD45RA, CD4 and CD14 were purchased from Biosource (BioSource International, Camarillo, CA, USA), CD40 from PharMingen (PharMingen, San Diego, CA, USA), CD3 and HLA-DR from Sigma and CD14, CD19, CD8, CD45RO and CD16 from Beckton-Dickinson (Beckton-Dickinson, San Diego, CA, USA). As controls, cells were stained with corresponding isotype-matched control MoAbs. Analysis was performed using a FACSCalibur flow cytometer (Becton Dickinson).

Statistical analysis

Paired and non-paired t-tests were performed on log-transformed data to compare the significance of results from groups with and without detectable parasite infection in the placenta.

RESULTS

Study population

Seventy-three pregnant women were enrolled into the study and 25 had active placental Pf infection at delivery (34·3%). As has been found previously, the prevalence of placental Pf infection was higher in first pregnancies (7/17, 41·2%versus 18/56, 32·2%) (Table 1). Mean maternal age, parity, placental weight, birth weight and reported chloroquine use were similar in the two groups (Table 1). As would be expected in asymptomatic individuals sampled once, maternal parasitaemia was detected in the majority, but not all, cases of active placental Pf infection; 72% of mothers with active placental Pf infection had peripheral parasitaemia. Almost half (12/25) of the cord blood samples from parasitized placenta were positive for Pf, but parasitaemia was low (= or>2500 parasites/µl) and all the infants were aparasitaemic at 1 week of age.

Table 1.

Selected descriptors of study population

	All	Active placental malaria	No active placental malaria	P-value
Total (n,%)	73	25 (34·3%)	48 (65·8%)	–
Age (mean, s.d.)	29·5 (19·6)	30·5 (22·6)	28·9 (17·9)	0·8
Parity (mean, s.d.)	3·4 (2·2)	3·2 (2·1)	3·6 (2·2)	0·5
Primipara (n,%)	17 (23·3%)	7 (41·2%)	10 (58·8%)	–
Multipara (n,%)	56 (76·7%)	18 (32·2%)	38 (67·8%)	–
Birth weight (g, s.d.)	3136 (436)	3108 (325)	3152 (493)	0·7
Placenta weight (g, s.d.)	477 (152)	512 (176)	457 (135)	0·2
Reported CQ use in last 3 months of pregnancy	45·7%	36·4%	50·0%	0·5

Effect of placental P. falciparum infection on proliferative responses of cord blood T cells

To compare cord blood T cell activation between placenta parasitized and non-parasitized groups, CBMC were tested for their ability to proliferate in response to mitogens and protein antigens (Fig. 1). Proliferative responses to PMA/ionomycin, PPD or Pf schizont extracts did not show any significant differences associated with Pf infection. However, for PHA, and in agreement with reported data [12], a significantly higher proliferative response was observed (P = 0·01) in the infected group. Taken together, these results indicate normal activation of T cells by all stimuli tested.

Fig. 1.

Proliferation of CBMC in response to mitogens and protein antigens. CMBC from newborns of mothers with Pf infected (n = 21) or uninfected placenta (n = 34) were cultured for 3 days with PHA or PMA/ionomycin or 6 days with PPD or schizont extracts. Medium alone or RBC extracts were used as negative controls for mitogens/PPD or schizont extracts, respectively. 1 µCi/well of methyl-[³H]thymidine was added for an extra 16 h, and thymidine uptake was measured (cpm). Proliferative responses were expressed using a stimulation index (SI) calculated from triplicate values as the geometric mean cpm stimulated wells (mitogens, proteins)/mean cpm negative control wells. Results are shown as means of stimulation index ± s.e.m. No significant differences in SI were observed between the groups except for PHA (*P = 0·01).

Effect of placental P. falciparum infection on cytokine production in response to mitogens

Cytokine production was also measured after 3 days of culture with mitogens ( Fig. 2). CBMC from both parasitized and non-parasitized placenta responded to PHA and PMA/ionomycin by producing high amounts of all tested cytokines. PHA and PMA/ionomycin induced high and equal amounts of IL-13 in both groups. In addition, concentrations of the Th2 marker sCD30 induced by these stimulants were also similar between the two groups. However, in the parasitized placenta group, significantly less IFN-γ was produced in response to PHA (P = 0·01). This difference in IFN-γ production was found using a low concentration of PHA (2·5 µg/ml) and decreased when higher concentrations, 5–10 µg/ml, were used (results not shown). CD3 depletion of CMBC almost abrogated IFN-γ induction in response to mitogen and demonstrated that CD3⁺ T cells and not NK cells are the major IFN-γ producers (results not shown).

Fig. 2.

Cytokine production by CBMC in response to mitogens. CBMC from parasitized (n = 21) and non-parasitized (n = 30) placenta were cultured as described in Fig. 1 and assessed for IFN-γ and IL-13 cytokines and sCD30 release in response to mitogens. Results are shown as mean ± s.e.m. Significantly less IFN-γ was produced by CBMC from the placenta parasitized group in response to PHA (2·5 µg/ml) (*P = 0·01), while the levels of the other cytokines were similar.

The observation that IL-13 and sCD30 production was similar in the two groups excludes at least major T cell malfunction or apoptosis in those from parasitized placenta. This was confirmed by flow cytometry analysis using the cell death dye propidium iodide, as no difference was found between both groups either ex vivo ( Table 2) or after culture with mitogens (results not shown). Furthermore, analysis of ex vivo CBMC by flow cytometry using markers for T cell subsets (CD3, CD4, CD8) and other cell populations [γδ T cells, NK (CD16), APC (CD19, CD14, MHC-II)] did not show any significant difference between the groups (Table 2).

Table 2.

Analysis of cord blood cell subsets by flow cytometry

	Mean ± s.d.
Placental infection
	PI	CD3	CD4	CD8	CD16	γδ-TCR	CD45-RA	CD45-RO	MHC	CD14	CD19
–	0·9 ± 0.01	30 ± 18	25 ± 13	9 ± 5	7 ± 3	2 ± 2	50 ± 22	15 ± 4	17 ± 3	11 ± 6	15 ± 6
+	0·5 ± 0·08	42 ± 29	40 ± 31	20 ± 17	10 ± 9	1·5 ± 1	55 ± 22	14 ± 8	18 ± 9	9 ± 4	19 ± 8

CBMC were analysed ex-vivo by flow cytometry and the percentage of dead cells (PI) or different cell subsets determined in both groups (n = 8 in each group). Results represent mean ± s.d.

Increased levels of sCD30 in response to P. falciparum antigens

CBMC from parasitized and non-parasitized placenta were also tested for their response to PPD and Pf164 or K1 schizont extracts. CBMC cultured in medium alone or in presence of extracts from uninfected RBC were used as negative controls, respectively ( Fig. 3). In comparison to mitogen-induced responses, both PPD and Pf antigens induced lower concentrations of tested cytokines. This is in agreement with the reported weak response of CBMC to protein antigens [18–21].

Fig. 3.

Cytokine production by CBMC in response to P. falciparum antigens. CBMC from parasitized (n = 14) and non-parasitized (n = 21) placenta were assessed for IFN-γ and IL-13 cytokines and sCD30 released in response to PPD and schizont extracts from two Pf strains, Pf164 and K1. Results are shown as mean ± s.e.m. In both groups IFN-γ production was either low or not induced in response to PPD or Pf antigens, respectively. Both PPD and Pf antigens induced IL-13 production. A specific and significant higher production of sC30 was found in response to Pf antigens from both Pf strains (P = 0·001), while sCD30 was below the detection limits of our assay in response to PPD.

Interestingly, although there was no increase of IFN-γ production in either group in comparison to the control uninfected RBC extracts, there was a significant increase of sCD30 in response to Pf extracts in the parasitized group (P = 0·001).

This cytokine profile, i.e. induction of IL-13 and sCD30 and absence of IFN-γ, was associated with placental Pf infection independently of maternal parasitaemia or parity state (results not shown).

Effect of placental P. falciparum infection on neonatal antigen-presenting cells

LPS activates monocytes and DC [32,33], inducing IL-12 production and up-regulation of co-stimulatory molecules, which can increase T cell activation and IFN-γ production. When LPS was combined with low-dose PHA (Fig. 4a) to stimulate CBMC, the concentration of IFN-γ detected increased significantly compared to PHA alone and the difference between the Pf infected and non-infected group was no longer statistically significant (P = 0·7).

Fig. 4.

(a) IFN-γ production in response to PHA and LPS. CBMC from parasitized (n = 10) and non-parasitized (n = 10) placenta were either stimulated with PHA alone or PHA and LPS. Results are shown as mean ± s.e.m. The difference found between both groups for IFN-γ production in response to PHA (P = 0·01) was still observed, but decreased to a non-significant level when LPS was combined with PHA (P = 0·26). (b) IFN-γ induction subsequent to APC activation with LPS. CBMC from parasitized (n = 12) and non-parasitized (n = 25) placenta were loaded with schizont extracts from both Pf strains Pf164 and K1, then stimulated or not with LPS, cultured for 6 days and assessed for IFN-γ release. Results are corrected for background release induced by RBC extracts. Results are shown as mean ± s.e.m. LPS enhanced IFN-γ production in both groups. (c) IL-12 production in response to LPS. CBMC from parasitized (n = 8) and non-parasitized (n = 24) placenta were stimulated with LPS and supernatants tested for IL-12 (p-40) production. Results are shown as mean ± s.e.m. LPS induced IL-12 production in both groups but levels were lower in the parasitized group.

Furthermore, when CBMC were preincubated with schizont extracts for 2 h before adding LPS and the cytokine responses measured on day 6, LPS increased T cell responses to Pf antigens and enhanced IFN-γ production in both groups (Fig. 4b). In both instances IFN-γ production remained lower in the parasitized group (Fig. 4a,b).

Accordingly, we assessed the contribution of APC in this system by measuring the effect of LPS on IL-12 production and up-regulation of co-stimulatory molecules. MHC-II and CD40 expression increased following LPS stimulation but no difference could be seen between the groups ( Table 3). Despite variability between donors, we found a significantly lower production of IL-12 (p40) (P = 0·04) in the CBMC from the Pf-infected group (Fig. 4c). Using quantitative reverse transcriptase (RT)-PCR, we also found a lower concentration of IL-12(p40) mRNA in the parasitized group (results not shown).

Table 3.

MHC-II and CD40 expression on CBMC following LPS stimulation

Placental infection		HLA+ CD40+ Mean ± s.d.
–	Medium	227 ± 691,3
	LPS	431 ± 1951,4
+	Medium	326 ± 1332,3
	LPS	485 ± 2362,4

CBMC were cultured for 18 h either in medium alone or with LPS (1 µg/ml) and MHC-II⁺ CD40⁺ cells were analysed by flow cytometry. LPS induced a significant up-regulation of HLA-II and CD40 cell markers in both non-infected (

¹P = 0·02) and infected group(

²P = 0·02). No significant difference was observed between both groups either in medium alone(

³P = 0·09) or after LPS treatment(

⁴P = 0·3). Results represent means and standard deviations of mean of fluorescence intensities (MFI) from seven Pf infected and eight non-infected placenta.

The observation that the LPS-induced increase of IFN-γ production in the presence of either PHA or Pf antigens correlates with the increase of IL-12 p40 (results not shown)

DISCUSSION

Pf infection during pregnancy increases infant morbidity and mortality further and this may reflect an exacerbation of the newborn's poor Th1 differentiation. The current study thus measured cytokine responses indicative of the Th1/Th2 bias of CBMC from infants born to mothers with or without active placental infection. Our data suggest strongly that placental infection with Pf impacts on Th differentiation. Neonatal T cells stimulated by PMA and ionomycin or PHA appeared fully competent, as they were able to proliferate and produce high concentrations of Th1 (IFN-γ) and Th2 (IL-13) cytokines or markers (sCD30). In contrast, PHA induced lower IFN-γ levels in the parasitized groups while IL-13 and sCD30 remained comparable in each group. PHA at low doses may mimic physiological conditions, as it requires APC co-stimulation to activate T cells, while PMA/ionomycin was used at higher doses as an internal control for an APC-independent T cell activation. This may account for the lower concentrations of IFN-γ in the Pf-parasitized group and indicate APC involvement. The reported inhibitory effect of Pf on DC maturation [26] is in concordance with this possibility. Pf schizont extracts and PPD, which require processing and presentation by APC, induced very low quantities of all tested cytokines in both groups. In particular, we found low IFN-γ production in response to PPD and none in response to Pf antigens. These results agree with those published previously showing poor APC function and Th1 differentiation in human neonates [18–21].

We studied the possible influence of APC further by examining the effect of adding LPS to activate APC to PHA and the Pf antigens. First, when LPS was combined with PHA, IFN-γ levels increased and the difference in IFN-γ production between the two groups decreased to a non-significant level. Secondly, Pf extracts induced IFN-γ production when combined with LPS. In both instances, even though the differences were not statistically significant, less IFN-γ was induced in the parasitized group.

The contribution of APC to this cytokine profile was assessed directly by comparing the effect of LPS on APC function in both groups. In this case, while co-stimulatory molecules increased on the surface of stimulated cells and IL-12 was induced in both groups, IL-12 concentration remained lower in the parasitized group. The levels of IL-12 correlated with an enhanced IFN-γ production by T cells in response to PHA and Pf extracts in both groups and also lower IFN-γ production in the parasitized group. The increased production of IFN-γ in response to these stimuli indicated indirectly the bioactivity of IL-12. These results show that Pf exacerbate the poor APC function, reducing further their IL-12 production which in turn impacts negatively on IFN-γ production. This may also favour sCD30 release, and thus the Th2 priming we found in response to Pf antigens in the infected group.

There was considerable variation in responses to the protein antigens within the groups. There are no doubt many reasons for this, including the highly variable genetic background of this population [34]. This may also reflect different in utero exposure of the fetus to parasite antigens, e.g. passage of mainly Pf schizont antigens through the placenta or passage of intact parasites and subsequent exposure of the fetus to all Pf erythrocyte stages. In utero priming may also be modulated by passively transferred maternal antibodies. Chloroquine is an immunomodulatory molecule [35,36] and may have increased within group variation, but as its use was divided equally between our two groups it cannot explain the differences found between the groups. In this study population it is noteworthy that these effects were associated with placental malaria and not with maternal parasitaemia, but this may vary with different levels of malaria endemicity.

Taken together, these results indicate that active infection of the placenta by Pf, on one hand, primes newborn immune responses for a Th2 (sCD30) response to Pf antigens and on the other hand, exacerbates the impaired function of a newborn's APC by reducing their IL-12 production and thus reducing IFN-γ production by neonatal T lymphocytes. These effects may impact on Th1 differentiation, which is critical for protection against infectious diseases in human newborns and infants. A prospective study in infants is needed to investigate the duration of this effect and determine whether it is due to the particular environment present at birth or is of longer duration, as a Th2 bias for the first few months of life might jeopardize immune responses to intracellular pathogens and vaccines of the Expanded Programme of Immunization (EPI).