Background

Malaria during pregnancy remains a major public health problem in tropical and sub-tropical regions. In 2018, an estimated 11 million pregnant women were infected by Plasmodium in sub-Saharan Africa, which resulted in about 872,000 low birthweight deliveries [1]. Pregnant women have increased susceptibility to malaria, caused by placental accumulation of Plasmodium falciparum infected erythrocytes (IEs), leading to poor outcomes for both mother and infant, including maternal anaemia, miscarriage, stillbirth, low birthweight, and preterm birth [2]. The sequestration of IEs in the placenta is mediated by a P. falciparum erythrocyte membrane protein 1 (PfEMP1) family member, VAR2CSA [3], that binds to chondroitin sulfate A (CSA) [4], a glycosaminoglycan present on the placental syncytiotrophoblast surface. VAR2CSA is a large (350 kDa) protein containing six Duffy binding-like domains (DBL1–6) and interdomain (ID) regions [5]. Although Plasmodium vivax does not sequester in the placenta [6], it has also been associated with poor pregnancy outcomes [7,8,9]. The reasons for this may include systemic increases in inflammatory cytokines, such as tumour necrosis factor α and predisposition to severe anaemia (reviewed in [9]).

In high transmission areas, primigravidae are at highest risk of P. falciparum infection [2]. By contrast, Latin America is epidemiologically heterogeneous, P. vivax generally dominates, and the region is characterized by lower malaria transmission with some foci of higher endemicity and a prevalence of malaria in pregnancy generally below 10% [10, 11]; [12, 13]. In areas of lower transmission, all pregnant women may be susceptible to P. falciparum infection [10, 13,14,15,16].

With repeated exposure, pregnant women acquire antibodies against VAR2CSA that block adhesion of IEs to CSA [17, 18], which may improve pregnancy outcomes [19, 20]. Declining malaria transmission limits the development of these protective antibodies against pregnancy malaria [21, 22].

Studies of antibodies in pregnant women from Latin America have given inconsistent findings, and few have included antibody to P. vivax antigens. One group reported a high prevalence (> 60%) of antibodies to VAR2CSA recombinant protein in both pregnant women and in never-pregnant women, men, and children [23, 24]. They proposed that these might be cross-reactive antibodies induced by a specific epitope in the P. vivax Duffy-binding protein (PvDBP) [25, 26]. A second group confirmed a high prevalence of antibodies to recombinant VAR2CSA proteins expressed in baculovirus-transfected insect cells. By contrast, antibodies targeting native VAR2CSA on whole IEs and to recombinant VAR2CSA proteins expressed in Chinese hamster ovary (CHO) cells were rare [27]. The presence of antibody epitopes in the recombinant proteins produced in insect cells rather than induction of a cross-reactive antibody to VAR2CSA could explain the previous finding [27]. The lack of parity-dependent increases in antibody responses to VAR2CSA in Colombia [24, 26, 27] could reflect the low malaria transmission intensity or may indicate the presence of non-specific antibody responses.

To address some of these knowledge gaps and contradictory findings, the current study examined antibody to VAR2CSA expressed on whole IEs, including antibody that opsonizes IEs for phagocytosis, and measured IgG to recombinant VAR2CSA proteins, non-pregnancy malaria-associated P. falciparum antigens and P. vivax proteins in a cohort of pregnant Brazilian women. Correlations between different antibody responses were sought. The effects of gravidity and number of episodes of Plasmodium infection on the seroprevalence of these antibodies were analysed, and the effect of time since last Plasmodium infection on antibody levels was investigated.

Methods

Study site and participants

A prospective cohort study was conducted in the Amazonian region of the Alto do Juruá valley, Acre, Brazil. In a previous study in this location, 8.9% of pregnant women had microscopically documented malaria infection, over 60% of which were due to P. vivax [10]. In total, 600 pregnant women uninfected or infected by Plasmodium sp. were recruited during their first antenatal care visit and followed until delivery. Malaria was diagnosed from thin and thick blood smears by 2 experienced microscopists and confirmed by PET-PCR [15]. Isolates were not genotyped. Each woman was followed up by a trained nurse, which involved at least 2 domiciliary visits during the second and third trimester, to monitor the woman’s clinical state and to provide routine antenatal care. At the time of recruitment, data were collected on socioeconomic, clinical, and obstetric variables. During the domiciliary visits, clinical and obstetric data were obtained, and a peripheral blood sample was collected. Peripheral blood plasma collected at delivery was used in this study.

A total of 192 women who presented with other infections (cytomegalovirus, rubella, toxoplasmosis, herpes simplex virus, human immunodeficiency virus, hepatitis B virus, hepatitis C virus, syphilis, dengue, chikungunya, or Zika virus) were excluded [10]. Of the remaining 408 pregnant women who were studied, 111 were infected with P. falciparum, 96 had infections with P. falciparum and P. vivax concurrently or at different time points during the current pregnancy, and 201 had no Plasmodium infection during the pregnancy. Samples from women with P. vivax mono-infection were not available.

Parasite culture

Two pregnancy malaria-associated P. falciparum lines, CS2 (which constitutively expresses VAR2CSA and binds to CSA) [28] and 3D7 (which was selected for CSA binding [29]) and a non-pregnancy associated P. falciparum line E8B-ICAM (which adheres to intercellular adhesion molecule (ICAM)–1 but not CSA) [30] were used. Parasites were cultured as previously described [31]. Cell cultures were synchronized as needed with 5% sorbitol [32] and subjected to gelatin flotation regularly to select knob-expressing IEs [33]. Adhesion to CSA and recognition by positive control sera were broadly similar for the 2 lines (not shown).

Cell culture

Pro-monocytic THP-1 cells [34] were maintained in RPMI 1640 supplemented with 10% heat-inactivated fetal bovine serum (FBS), penicillin 1 unit/mL, streptomycin 1 µg/mL, glutamine 292 µg/mL and 25 mM HEPES buffer (all from Gibco™). The cells were maintained at a density of 2 × 105 cells/ml in 75 cm2 cell culture flasks (Corning, 430641U) stored upright in a humidified 37°C incubator with 5% CO2, as previously described [35].

Measuring IgG levels against VSAs

IgG antibody levels to pregnancy-specific VSAs on CS2 and 3D7 IEs were measured by flow cytometry as previously described [36]. In brief, test plasma was incubated with CS2 and 3D7 IEs (1:10) in duplicate, followed by incubation with polyclonal rabbit anti-human IgG (Dako A0424, 1:100 dilution), and with Alexa Fluor 647 donkey anti-rabbit (Invitrogen A32795, 1:500 dilution) containing 25 µg/ml of dihydroethidium (DHE), for 30 minutes in the dark. The cells were re-suspended in 2% paraformaldehyde (PFA) and acquired by flow cytometry (CytoFLEX S, Beckman Coulter) and analysed by FlowJo® software v10. The IgG level was expressed as relative geometric mean fluorescence intensity (MFI), represented as a percentage of the MFI of the positive control (pooled sera from P. falciparum infected Malawian multigravid women, after subtraction of the negative control (median MFI of sera from 18 malaria unexposed Melbourne donors; all controls were run individually, in the same experiments as test samples).

Opsonic phagocytosis assay

The opsonic antibodies against CS2, 3D7 and E8B-ICAM were measured as previously described in [37]. In brief, purified trophozoite stage IEs stained with 25 µg/ml of DHE were resuspended at 1.67 × 107/ml and opsonized with heat inactivated test plasma (1:10). Opsonized IEs were then incubated with THP-1 cells in a 5% CO2 humidified incubator at 37°C. Phagocytosis was stopped and non-phagocytosed IEs were lysed before cells were resuspended in 2% PFA. Test samples were then analysed by flow cytometry (CytoFLEX S, Beckman Coulter) and opsonizing antibody was determined as proportion of THP-1 cells positive for DHE and expressed as a relative percentage of positive control (pooled sera from P. falciparum infected Malawian multigravid women). Samples were defined as seropositive if the relative percentage was more than two standard deviations (SD) above the mean relative percentage of the unexposed controls (sera from 15 malaria naive Melbourne donors).

Multiplex assay to measure antibodies to the recombinant VAR2CSA domains

Plasma IgG levels to VAR2CSA DBL domains (see Additional file 1) were measured using a multiplex assay as previously described in [38]. VAR2CSA DBL recombinant proteins were kind gifts from Joe Smith, Patrick Duffy, David Narum and Morten Nielsen. For antibody measurement, VAR2CSA DBL proteins coupled to Bio-Plex magnetic carboxylated microspheres (Bio-Rad) were incubated with test plasma (1:100 dilution in 1% bovine serum albumin (BSA) in phosphate-buffered saline (PBS)) in 96-well flat-bottom plates (Bio-Rad) on a shaker overnight at 4°C. Following incubation, plates were centrifuged and washed twice with PBS containing 0.05% (v/v) Tween-20 using a magnetic plate-washer (Bio-Plex Pro wash station). Then, anti-human IgG conjugated with phycoerythrin (PE) (SouthernBiotech, 9040-09, diluted to 1.3 µg/ml in 1% BSA in PBS) was added to each well and incubated for 2 hours on a plate shaker at room temperature. After washing with PBS containing 0.05% (v/v) Tween-20 and resuspending in sheath fluid (Life Technologies, MPXDF4PK), the plates were read on a Bio-Plex MAGPIX multiplex reader (Bio-Rad) and the MFI values were reported. Samples were designated antibody positive if the MFI was more than two SD above the mean MFI of the negative controls (sera from 8 malaria unexposed Melbourne donors).

Multiplex assay to measure antibodies to the recombinant P. vivax proteins

IgG levels to P. vivax recombinant proteins (see Additional file 1) were measured using a multiplex assay described elsewhere [39]. The P. vivax proteins were kind gifts from Chetan Chitnis, Eizo Takashima and Takafumi Tsuboi. Protein-coupled microspheres were incubated with test plasma (1:100 dilution in PBS containing 1% BSA and 0.05% (v/v) Tween-20) in 96-well flat-bottom plates for 30 min at room temperature on a plate shaker. Following the incubation, microspheres were washed thrice with PBS containing 1% BSA and 0.05% (v/v) Tween-20. Then, anti-human IgG detector antibody conjugated with PE (Jackson ImmunoResearch (1 mg/ml), 1:100 dilution) was added to each well and incubated for 15 minutes at room temperature on a plate shaker.

After washing and resuspending in PBS containing 1% BSA and 0.05% (v/v) Tween-20, plates were read on a Bio-Plex MAGPIX multiplex reader (Bio-Rad) and the MFI values were reported. Samples were defined as seropositive if the MFI was more than two SD above the mean MFI of the negative controls (sera from 10 malaria unexposed Melbourne donors).

IgG against non-pregnancy malaria-specific P. falciparum antigens

Antibody responses to schizont extract and P. falciparum MSP1-19 were measured by enzyme-linked immunosorbent assay (ELISA), as described elsewhere [40, 41]. PfMSP1-19 protein was cloned and expressed in Escherichia coli as previously described [40] and schizont extract was prepared from P. falciparum CS2-IEs [42]. Results were expressed relative to the Malawian positive control described above which had a value of 100 arbitrary units (AU). Samples were defined as antibody positive if the AU was more than two SD above the mean AU of the negative controls (8 malaria unexposed Melbourne donors for schizont extract and 10 malaria unexposed São Paulo donors for PfMSP1-19).

Statistical analysis

Demographic, clinical variables and antibody measurements were summarized with number and percentages for categorical variables and median and interquartile range (IQR) for continuous variables. Pearson chi-squared test was used to compare frequencies of seropositive individuals between groups, and Kruskal-Wallis rank sum tests followed by Dunn's multiple comparison test were used to compare the antibody levels amongst multiple groups. Spearman's rank test assessed the correlations between antibody to multiple antigens measured by different assays. Multiple logistic regression analysis was performed to determine the effect of Plasmodium infection and the number of episodes of P. falciparum infection on antibody seropositivity. Multivariate linear regression assessed the association between the time since the infection was diagnosed and antibody levels. P-values of less than 0.05 were considered statistically significant.

Results

Demographic and clinical characteristics of participants

Of the 408 pregnant women selected for this study, 207 (50.7%) had a Plasmodium spp. infection at some point in the current pregnancy. Of these, 111 had at least one P. falciparum infection, and 96 women had both P. falciparum and P. vivax infections. Maternal age, gravidity, and gestational age at delivery were similar across the three groups (Table 1). Women without parasitaemia episodes had higher hemoglobin levels at delivery (median (Inter quartile range (IQR)), 11.8 g/dL (11, 12.8)) compared to women with P. falciparum or with P. falciparum and P. vivax infections during pregnancy (median (IQR), 11.5 g/dL (10.7, 12.1), and 11.4 g/dl (10.5,12.2), respectively, p = 0.008). Birthweight tended to be lower in women with Plasmodium infection during pregnancy than in those without infection (p = 0.05) (Table 1).

Table 1 Demographic and clinical characteristics of participants according to malaria infection status
Full size table

Seroprevalence of antibodies to pregnancy-specific and non-pregnancy specific Plasmodium antigens

IgG antibodies to pregnancy-specific and non-pregnancy specific VSA on P. falciparum IEs were measured, and antibody seroprevalence was compared between women with P. falciparum infection, both P. falciparum and P. vivax infections, and without infection. When stratified by infection status 4–16% of women were seropositive for antibody to VSA on pregnancy-specific CSA binding CS2-IEs and 15–22% of women had opsonizing antibodies against pregnancy-specific VSA on CS2-IEs. By contrast, 43–63% of women had opsonizing antibodies to the non-pregnancy-specific VSAs of E8B-ICAM (Fig. 1A).

Fig. 1
figure 1

Seropositivity to pregnancy-specific and non-pregnancy specificP. falciparum and P. vivax antigens in pregnant women living in the Brazilian Amazon. Data presented as the percentage of women seropositive to each antigen A Percentage seropositivity of total IgG and opsonic antibodies to antigens on the surface of intact IEs B Percentage seropositivity against recombinant VAR2CSA domains. C Percentage seropositivity to non-pregnancy specific P. vivax antigens D Percentage seropositivity to non-pregnancy specific P. falciparum antigens. VSA, variant surface antigens; DBL, Duffy binding like domain; ID, interdomain region, CS2 Phago, opsonic phagocytosis of CSA-binding CS2 IEs; 3D7 Phago, opsonic phagocytosis of CSA-binding 3D7 IEs; E8B Phago, opsonic phagocytosis of ICAM-1 binding E8B IEs; RAMA, Rhoptry-associated membrane antigen; MSP, merozoite surface protein; PvTRAg, tryptophan-rich antigen; PvDBP II-sal1, Duffy binding protein region II from ‘sal1’ strain; PvDBP II-AH, Duffy binding protein region II from ‘AH’ strain; RBP2b, reticulocyte binding protein 2b; PvEBP, erythrocyte-binding protein II

Full size image

The proportion of women seropositive for antibodies to recombinant VAR2CSA DBL domains by multiplex assay varied from 60–69% (depending on exposure group) for DBL5-isolate to 21–27% (depending on exposure group) for DBL6-IT4 (Fig. 1B). Similarly, the proportions of women seropositive for antibodies to P. vivax proteins varied from 74–97% against Pv-MSP1-19 to 17–53% against PVDBPII-Sal1, (Fig. 1C). Around 90% of women with a history of previous Plasmodium infection at any time had antibodies to PfMSP1-19 and schizont extract (Fig. 1D).

Comparison of antibody levels to VAR2CSA DBL domains and P. vivax proteins

Relative levels of IgG against P. falciparum VAR2CSA DBLs and P. vivax proteins were compared between women with P. falciparum only infection, both P. falciparum and P. vivax infections, or without malaria infection. The levels of IgG to VAR2CSA DBL domains were similar across the three groups (Fig. 2). However, IgG to P. vivax proteins varied significantly, being highest in women with both P. falciparum and P. vivax infections and lowest in uninfected women (Fig. 3). Interestingly, antibody levels to all P. vivax proteins were also higher in pregnant women with P. falciparum only than uninfected women (p < 0.0003) (Fig. 3).

Fig. 2
figure 2

Levels of immunoglobulin G (IgG) antibody in pregnant women living in the Brazilian Amazon to recombinant VAR2CSA Duffy binding like (DBL) domains. The antibody levels between groups were compared by the Kruskal-Wallis test followed by Dunn's multiple comparison test. Values on the y-axis represent median IgG levels (expressed as MFI) against individual VAR2CSA domains. MFI, Median fluorescence index; ID, interdomain region. IT4, 3D7, 7G8 and FCR3 refer to the parasite strain sequence used for protein expression. *P ≤ 0.05

Full size image
Fig. 3
figure 3

Levels of immunoglobulin G (IgG) antibody in pregnant women living in the Brazilian Amazon to recombinant P. vivax antigens. The antibody levels between groups were compared by the Kruskal-Wallis test followed by Dunn's multiple comparison test. Values on the y-axis represent mean IgG levels (expressed in MFI) against individual P. vivax antigens. MFI, Median fluorescence index; Pf, Plasmodium falciparum; Pv, Plasmodium vivax; RAMA, Rhoptry-associated membrane antigen; MSP, merozoite surface protein; PvTRAg, P. vivax tryptophan-rich antigen; PvEBP, P. vivax erythrocyte-binding protein; PvDBP II-sal1, P. vivax Duffy binding protein region II from ‘sal1’ strain; PvDBP II-AH, Duffy binding protein region II from ‘AH’ strain; RBP2b, reticulocyte binding protein 2b; **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001

Full size image

Antibody responses against VAR2CSA domains correlate with each other, but not with antibodies to surface antigens on intact IEs

The correlation between the antibodies measured by different assays was investigated (Fig. 4). Antibodies to recombinant VAR2CSA DBL domains correlated significantly with each other (r ≥ 0.55, p < 0.0001) but not with antibodies to parasite antigens on surface of intact IEs (r ≤ 0.12, p > 0.06). Antibody to DBLs from different isolates (expressed in different expression systems) correlated with each other (r ≥ 0.62, p < 0.0001 for four pairs of similar constructs). Antibodies to recombinant P. vivax proteins also correlated significantly with each other (0.27 ≤ r ≤ 0.84, p < 0.0001) and showed weak to moderate correlation with antibodies to schizont extract (0.29 ≤ r ≤ 0.6, p < 0.0001) and P. falciparum MSP1-19 (0.35 ≤ r ≤ 0.47, p < 0.0001). Moreover, weak but significant correlation was observed between antibodies to P. vivax proteins and opsonic antibodies; CS2 phago (0.2 ≤ r ≤ 0.36, p < 0.0001), 3D7 phago (0.2 ≤ r ≤ 0.38, p < 0.0001), and E8B phago (0.1 ≤ r ≤ 0.31, p < 0.0001). Notably, the level of IgG to P. vivax proteins did not correlate well with the level of IgG against VAR2CSA DBLs (r ≤ 0.14, p ≥ 0.06).

Fig. 4
figure 4

Heatmap of Spearman's correlation coefficients between 28 different antibody measurements. The color bar on the right indicates Spearman’s correlation coefficient from − 1 (negative correlation, red), no correlation (white) to 1 (positive correlation, blue). The intensity of the colors and the size of the circles indicate the correlation-values, with large, deeply colored blue or red circles indicate a relatively strong correlation, while small, lightly colored blue or red circles indicate a relatively weak correlation. VSA, variant surface antigens; DBL, Duffy binding like domain; ID, interdomain region; CS2 Phago, opsonic phagocytosis of CS2 infected erythrocytes (IEs); 3D7 Phago, opsonic phagocytosis of 3D7 IEs; E8B Phago, opsonic phagocytosis of E8B IEs; MSP, merozoite surface protein; Pv, Plasmodium vivax; RAMA, Rhoptry-associated membrane antigen; PvTRAg, P. vivax tryptophan-rich antigen; PvEBP, P. vivax erythrocyte-binding protein; PvDBP II-sal1, P. vivax Duffy binding protein region II from ‘sal1’ strain; PvDBP II-AH, Duffy binding protein region II from ‘AH’ strain; RBP2b, reticulocyte binding protein 2b

Full size image

Gravidity and antibody to P. falciparum and P. vivax antigens

Next, the associations between gravidity and antibody seropositivity were investigated. Gravidity-dependent differences in seropositivity were observed for antibodies against VSA on CS2 IEs (p = 0.04). However, no gravidity dependent acquisition was observed for all other tested P. falciparum and P. vivax antigens (see Additional file 2).

Association between malaria infection and seropositivity to P. falciparum antigens

To investigate whether malaria infection was a significant determinant of seropositivity, multivariate logistic regression was performed after adjusting for gravidity, maternal age, and body mass index (BMI) at enrollment. The participants were stratified into two groups: infected (women with P. falciparum only and both P. falciparum and P. vivax infections) and uninfected.

Plasmodium-infected women were more likely to have antibodies against pregnancy-specific VSAs on CS2 IEs (AOR (95% CI) = 4.1 (1.8, 9.1), p = 0.001) than uninfected women, and they were twice as likely to be seropositive for opsonic antibodies against E8B-ICAM IEs compared to their uninfected counterparts (AOR 2.1 (1.4, 3.1), p < 0.0001). However, the proportion of women with opsonic antibodies to VSA on placental binding IEs did not vary between infection groups (Table 2). There was no significant association between seropositivity to VAR2CSA DBL domains and infection, but malaria-infected women were more likely to be seropositive for antibodies to PfMSP1-19 and P. falciparum CS2 schizont extract than uninfected women (AORs 12.9 (7.4, 22.4), p < 0.0001) and 9.7 (5.8, 16.3), p < 0.0001, respectively) (Table 2).

Table 2 Associations between Plasmodium infection status and being seropositive for antibodies to P. falciparum antigens
Full size table

Association between malaria infection and antibody seropositivity to P. vivax antigens

To investigate the association between P. vivax infection and seropositivity for antibodies against P. vivax proteins, women with both P. falciparum and P. vivax infections were compared to women without infection. Women with Plasmodium infection were more likely to be seropositive for antibodies to P. vivax proteins than uninfected women (AOR (95% CI) ranges from 5.6 (3.2, 9.7), p < 0.0001 for PVDBPII-Sal1 to 15.7 (8.3, 29.7), p < 0.0001 for PvTRAg_2; Table 3).

Table 3 Associations between P. vivax infection and being seropositive for antibodies to P. vivax antigens
Full size table

Influence of the number of P. falciparum episodes on the seropositivity

Antibody seropositivity was compared between women who had multiple episodes or a single episode of P. falciparum infection during the current pregnancy. Individuals with both P. falciparum and P. vivax infections or without Plasmodium infection were excluded from this analysis to investigate the influence of only P. falciparum. Out of 111 women infected with P. falciparum, 81 had a single infection, and 30 had multiple infections during the current pregnancy. In adjusted analyses, women who had multiple P. falciparum episodes during the current pregnancy were more likely to be seropositive for antibodies against VSAs on CS2 IEs (AOR (95% CI) = 4.0 (1.2, 13.6), p = 0.028) than women who had a single infection. The seropositivity against all other tested P. falciparum antigens did not differ significantly by the number of P. falciparum episodes (see Additional file 3).

Association between time since infection and malaria antibody levels

Multivariate linear regression was performed to investigate whether antibody levels decline from the time the infection was diagnosed until delivery. Time since last infection date was available for 97 of 111 women infected with P. falciparum only and 51 of 96 women who had both P. falciparum and P. vivax infections. The time from last infection to delivery (in weeks) ranged from (median (IQR), 0.43 (0, 4.9) for P. vivax infection to 7 (0.57, 17.6) for P. falciparum infection. Generally, the levels of antibody against P. falciparum antigens were not significantly associated with the time since the P. falciparum infection was diagnosed. The exception was for antibodies to PfMSP1-19, which declined as time since the infection increased (coefficient (95% CI) = 1.5 (− 2.3, − 0.6), p = 0.001), suggesting that anti-PfMSP1-19 antibodies could be a marker of recent malaria exposure (see Additional file 4). Similarly, there were no significant associations between the time since P. vivax infection and the antibody levels against P. falciparum antigens, except for antibody to DBL3-7G8 which increased with increasing time.

Antibody levels to the P. vivax proteins tended to be negatively associated with the time since last P. vivax infection, but these trends did not reach statistical significance (p > 0.05). Similarly, no significant association was found between the antibody levels to P. vivax proteins and time since P. falciparum infection (see Additional file 4).

Discussion

In areas of high malaria transmission, multigravid pregnant women acquire protective antibodies against pregnancy-associated malaria [43, 44]. In Latin America, transmission is often low, and P. vivax predominates, and recent studies have yielded conflicting data on the prevalence of antibody to VAR2CSA in pregnancy and non-pregnant populations [24, 26, 27]. The present study investigated the prevalence and levels of antibodies to pregnancy-associated and other malaria antigens in the Brazilian Amazon region, using a variety of assay formats and antibody targets including IEs, pregnancy-specific and non-pregnancy-specific P. falciparum recombinant proteins, and a panel of P. vivax antigens.

Antibody prevalence varied significantly depending on whether IE or recombinant antigens were used. When P. falciparum IEs were used, the prevalence of IgG antibodies and of opsonizing antibodies to pregnancy-specific VSAs was low, whereas most women had opsonizing antibodies to the VSA of a non-pregnancy-specific isolate. The low prevalence of antibody to pregnancy-specific VSAs may reflect low recent malaria transmission in the study area, while the higher recognition of E8B-ICAM is a product of higher lifetime exposure. Low levels and prevalence of antibodies to VAR2CSA antigens have been observed in settings of unstable malaria transmission [31], following a decline in malaria transmission, and in low transmission areas relative to high transmission areas of the same country [22, 45,46,47,48]. In contrast to pregnancy-associated IEs, recognition of recombinant VAR2CSA DBL proteins was common. Up to 69% of women recognized individual DBL proteins and recognition of individual DBLs was significantly correlated but seropositivity to recombinant DBL domains was not associated with Plasmodium infection. This was consistent with studies that reported a high prevalence of antibodies to VAR2CSA DBLs in Colombian pregnant women, children, and men with malaria, which did not differ between subjects with or without current P. vivax or P. falciparum infection [24, 27].

Although PfEMP1 is the predominant antigen on the surface of the IE [49], antibodies to recombinant VAR2CSA DBLs did not correlate with antibodies to antigens on the surface of VAR2CSA-expressing CS2 and 3D7CSA IEs. Previous studies have reported the existence of anti-recombinant VAR2CSA antibodies in non-pregnant populations [24, 26, 50,51,52], suggesting antibodies to recombinant VAR2CSA DBLs are not specific to pregnancy. This pattern might be related to the protein expression systems used. Post-translational glycosylation of recombinant proteins expressed in bacterial or insect cells can alter epitopes [53], and antibodies to full length VAR2CSA produced in baculovirus transfected insect cells were prevalent in Colombian pregnant women, men and children. Levels of antibodies to baculovirus-expressed VAR2CSA correlated strongly with antibodies to a similarly-expressed non-pregnancy related PfEMP1, whereas IgG towards native VAR2CSA antigen and VAR2CSA expressed in Chinese hamster ovary cells was restricted to pregnancy-associated malaria [27]. By contrast, and similar to the observations in the current study, antibodies that recognized pregnancy-specific IEs were uncommon, pregnancy-specific, and did not correlate with antibody to recombinant protein. This suggests that antibodies to recombinant VAR2CSA may recognize post-translational modifications that may be shared with non-Plasmodium proteins [27]. The VAR2CSA DBL proteins used in the present study were single domains expressed in bacterial, yeast, and insect cells (see Additional file 1). Levels of antibody did not to vary significantly with the expression system used and did not correlate with responses to IEs. They correlated moderately to strongly with one another irrespective of the expression system used. None of the proteins were expressed in Chinese hamster ovary cells [27]. Notwithstanding the high recognition of these proteins, it remains unclear whether recombinant VAR2CSA proteins are useful measures of exposure or immunity in South American populations [27].

In other studies from Colombia, antibodies to recombinant VAR2CSA domains were reported following exposure to P. vivax [26], and it has been suggested that heterologous immunity between P. vivax DBP and VAR2CSA could occur [25]. This study is only partly comparable with the earlier study, lacking non-pregnant individuals with malaria or individuals with P. vivax infections only. Antibody to different P. vivax proteins including Duffy binding protein correlated with one another, but there was no significant correlation between antibody to P. vivax antigens and VAR2CSA DBL domains. By contrast, in a cohort consisting of pregnant women from Brazil, Colombia, Guatemala, India, and Papua New Guinea correlations between antibodies to recombinant DBL proteins and recombinant P. vivax antigens were observed, although functional antibody that inhibited P. vivax Duffy binding protein interacting with its receptor did not correlate with other responses [54].

Pregnancy-specific antibodies often increase following infection and with increasing gravidity [44, 55]. In this cohort, only prevalence of IgG to CS2 IEs was higher in women with a history of P. falciparum infection compared to uninfected women and was positively associated with multiple P. falciparum infections. Antibodies to VAR2CSA DBLs did not differ in prevalence or magnitude between women with and without a history of P. falciparum infection during the index pregnancy. Similarly, only antibodies against pregnancy-specific VSAs on CS2 IEs increased with gravidity. The lack of matching associations with the 3D7CSA IEs may reflect higher VAR2CSA expression on CS2 than 3D7CSA [56, 57]. The parent isolate of CS2, It, originated in Brazil, and there could be a higher prevalence of CS2-like parasites in the population, but further studies of VAR2CSA genetic diversity [58] would be required. The present observations are consistent with the possibility that antibody to the recombinant DBLs are not pregnancy-specific, but antibody to IEs may be.

High proportions of women had antibodies to the P. vivax antigens tested, and antibody seropositivity and levels were higher in women with either P. falciparum only or both P. falciparum and P. vivax infections compared to uninfected women. The high seropositivity rate of antibodies to P. vivax antigens even among women without documented P. vivax during pregnancy may reflect the local predominance of P. vivax infection. Almost 70% of reported cases in Latin America are caused by P. vivax [59], and in the current cohort, P. vivax is the cause for 63.9% of infections [10]. In general, antibody levels to recombinant P. vivax antigens were higher in women with both P. falciparum and P. vivax infections than in those with P. falciparum only, possibly reflecting recent boosting as the P. vivax is the dominant species circulating in the area. The higher levels in women with P. falciparum than in uninfected women may reflect higher overall exposure to malaria, but are in contrast to findings from a multi country pregnancy cohort in which there was no evidence of higher antibody levels to P. vivax antigens in women with P. falciparum mono-infections compared to women without malaria infection at the time of sampling [54].

The relationship between the time since infection and the antibody level at delivery was also investigated. There were no significant associations between the time since either P. falciparum or P. vivax infection and antibody levels against pregnancy-specific P. falciparum antigens. On the other hand, antibodies to PfMSP1-19 decreased as the time since the last P. falciparum infection increased, suggesting that these antibody responses are relatively short-lived in pregnant women. This is broadly similar to a study from Thailand, which found that antibodies to merozoite antigens declined faster than antibody to VAR2CSA DBL5 in pregnant women [60], although the confidence intervals were large in the previous study. Furthermore, antibody levels to most P. vivax antigens also trended downwards as time since P. vivax infection increased, suggesting that repeated exposure may be required to maintain these antibody responses. The small sample size may have limited these findings, but they are in line with identification of specific proteins such as RBP2b (reticulocyte binding protein 2b) as markers of P. vivax infections in the prior 9 months [39], although these markers have not yet been validated in pregnant women.

Strengths of the study included the large number of women with documented P. falciparum and P. vivax infection, the longitudinal study design and the examination of antibody response using multiple experimental approaches. Weaknesses include the lack of sequential samples, which could have allowed more definitive conclusions regarding antibody decline following infections, and the relatively small number of women with multiple episodes of infection during pregnancy. Timing of infection during pregnancy relative to sampling varied quite widely. Infections were not genotyped, so it was not possible to confirm whether they were recrudescent or new infections. Inclusion of women with P. vivax alone would have added another dimension to the study.

Conclusions

In summary, pregnant women in this region of Brazil had limited acquisition of antibodies to pregnancy-specific antigens on surface of whole IEs, despite having relatively high levels of antibodies to non-pregnancy associated P. falciparum antigens, recombinant DBL proteins, and P. vivax antigens. The lack of correlation between antibody to VAR2CSA DBL domains and antibody to surface antigen on IEs, or between antibody to VAR2CSA DBL domains and gravidity or malaria infection as markers of exposure, suggests they have limited utility in measuring immunity in this setting. Further studies are needed to define antibody responses that are robust markers of recent P. falciparum infection and/or may be correlates of protection from adverse pregnancy outcomes in low and unstable malaria transmission settings such as the Brazilian Amazon.