Parity-Dependent Recognition of DBL1X-3X Suggests an Important Role of the VAR2CSA High-Affinity CSA-Binding Region in the Development of the Humoral Response against Placental Malaria

Abstract

Plasmodium falciparum multidomain protein VAR2CSA stands today as the leading vaccine candidate against pregnancy-associated malaria (PAM). Most of the studies aiming to decrypt how naturally acquired immunity develops have assessed the immune recognition of individual VAR2CSA Duffy-binding-like (DBL) domains, thus overlooking the presence of conformational epitopes resulting from the overall folding of the full-length protein. In order to characterize the development of humoral immunity toward VAR2CSA, we made use of a large cohort of 293 Senegalese pregnant women to assess the level of recognition by plasma IgG of the full-length VAR2CSA protein of the 3D7 parasite strain (3D7-VAR2CSA), the CSA-binding multidomains 3D7-DBL1X to -DBL3X (3D7-DBL1X-3X), and the CSA nonbinding multidomains 3D7-DBL4ε to -DBL6ε (3D7-DBL4ε-6ε), as well as individual 3D7-DBL domains. Our results revealed a parity-dependent recognition of the full-length 3D7-VAR2CSA and of the CSA-binding region, 3D7-DBL1X-3X. Indeed, multigravid women possess significantly higher levels of antibodies directed against these constructs than primigravidae. Our results suggest an important role of antibodies targeting the CSA-binding region in the development of immunity against PAM, therefore providing new insights on how natural protection might be acquired and further information for the design of VAR2CSA-based vaccines.

INTRODUCTION

Each year, pregnancy-associated malaria (PAM) is responsible for the deaths of as many as 363,000 neonates and for at least 10,000 maternal deaths worldwide (1). The massive accumulation of Plasmodium falciparum-infected erythrocytes (IEs) to the syncytiotrophoblast layer composing the maternal face of the placenta limits maternal-fetal exchanges, leading to clinical complications for both mother and child (2). This phenomenon is mediated by the VAR2CSA protein, a member of the P. falciparum erythrocyte membrane protein 1 (PfEMP1) adhesins, that is preferentially expressed by placental parasites (3–5) and binds to the placental chondroitin-4-sulfate A (CSA) (6). After one or two pregnancies, women acquire antibodies targeting VAR2CSA that inhibit IE sequestration and decrease the negative outcomes of PAM (7, 8). In vitro assays have shown that antibodies from multigravid women are able to block the interactions between CSA and IEs originating from different parts of the world (9), showing the existence of a cross-reactive antibody response and suggesting that a relative conservation of important inhibitory epitopes within VAR2CSA proteins may exit.

VAR2CSA is a 350-kDa transmembrane protein composed of six Duffy-binding-like (DBL) domains. Previous works have described the CSA-binding properties of single DBL domains (3, 10, 11). Insight derived from structural studies of DBL domains revealed the importance of the DBL3X domain in VAR2CSA CSA binding but also suggested a higher-order structure organization of VAR2CSA where the positioning of different surfaces involved in chondroitin sulfate proteoglycan (CSPG) binding forms a specific binding pocket (12–14). Nevertheless, the overall folding of the DBL domains and their interactions with each other remain unclear. Recent studies have reported that the full-length extracellular region of VAR2CSA binds to CSA with higher specificity and affinity than individual domains due to the presence of a single high-affinity CSA-binding site involving multiple domains (15, 16). The demonstration that the CSA-binding properties of the DBL1X to DBL3X (DBL1X-3X) region of the VAR2CSA protein are similar to those of the full-length protein DBL1X to DBL6ε (DBL1X-6ε) led us to consider DBL1X-3X a PAM vaccine candidate equally as potent as DBL1X-6ε in generating adhesion-blocking antibodies (17). Additional work confirmed that the core of the CSA-binding site is indeed located within the N terminus of VAR2CSA (18) and, more precisely, in the DBL2X domain and parts of the flanking interdomain (ID) regions ID1-ID2a (19). Taken together, these data indicate that CSA binding most likely relies on the formation of a specific conformational pocket involving multiple domains and not, as thought previously, on independent low-affinity binding sites located on different domains and forming multivalent interactions that together cause placental sequestration by avidity effects. These results are of importance as they redefine our experimental approaches to characterize the development of naturally acquired humoral immunity to VAR2CSA. Indeed, most of the previous studies on acquisition of anti-VAR2CSA antibodies from natural infections have made use of individual DBL domains, thus disregarding the presence of immunoglobulins that are able to recognize conformational epitopes resulting from the overall folding of the full-length VAR2CSA protein or the CSA-binding multidomain DBL1X-3X (20).

A study done on a limited number of plasma samples from Cameroonian pregnant women (n = 89) showed that high levels of antibodies directed toward the full-length VAR2CSA of P. falciparum FCR3 (FCR3-VAR2CSA) play a preponderant role in protection against PAM (21). Furthermore, another study also performed in Cameroon but on a larger number of pregnant women revealed that the response to ID1-ID2a (which span the DBL2 domain) was not pregnancy specific but predominantly directed against unrelated cross-reactive epitopes which may have been induced by other PfEMP1 variants during previous infections (22).

Here, we examine a large cohort of 293 Senegalese pregnant women to characterize the development of humoral immunity toward the 3D7-VAR2CSA variant. We then compared the recognition patterns of the full-length 3D7-VAR2CSA, the CSA-binding 3D7-DBL1X-3X multidomains, and the CSA-nonbinding multidomains of 3D7-DBL4ε to -DBL6ε (3D7-DBL4ε-6ε) as well as of individual 3D7-DBL domains, using plasma samples obtained from primigravid and multigravid women during the course of pregnancy as well as at the time of delivery.

MATERIALS AND METHODS

Ethics statement.

For the complete study, the human experimentation guidelines of both French and Senegalese health authorities were followed. The study design, the sampling protocol, and the method of collecting informed consent were approved by the ethical committee of the Senegalese Ministry of Health.

Study populations.

Pregnant women were enrolled in a prospective cohort study between July and October 2001 in Thiadiaye, Senegal (23). Malaria in Thiadiaye is seasonally transmitted during the rainy season, with an estimate of 10 infected bites/individual/year (24). Women pregnant for less than 6 months were enrolled only if they were not infected with malaria parasites at the time of enrollment. Women were followed by active and passive malaria detection through monthly antenatal care visits and through weekly home visits until delivery. Malaria diagnosis was based on thick smears of peripheral blood to identify plasmodial parasites. At the time of delivery, the placental malaria diagnosis was determined by the presence of P. falciparum-infected red blood cells in the placental blood. Double-read routine microscopic examinations were performed by experienced technicians. According to the Ministry of Health policy at the time of the study, women were advised to take chloroquine prophylaxis while they were pregnant, but the drug was not given to them. All women from the cohort showing fever and a positive thick blood smear were given curative treatment with chloroquine. A total of 293 pregnant women could finally be considered for the study.

Plasma samples from 24 Beninese men ([BM] mean age, 32.6 years) living in southwest Benin and 24 plasma samples from French adult (FA) donors never exposed to P. falciparum were used as control groups. Malaria in southwest Benin is seasonally transmitted during the rainy season, with an estimate of 35 to 60 infected bites/individual/year (25).

Recombinant proteins.

The 3D7-VAR2CSA multiple-domain constructs DBL1X-3X, DBL4ε-6ε, and DBL1X-6ε (Fig. 1) were cloned into the pTT3 vector from a synthetic 3D7-VAR2CSA gene optimized for human codon usage. Proteins were expressed in HEK 293-F cells and purified as previously described (16, 17, 26).

FIG 1.

Expression of VAR2CSA recombinant proteins. (A) Schematic representation of the 3D7-VAR2CSA domain organization and sequence boundaries of the recombinant proteins used in this study. (B) SDS-PAGE under reducing and nonreducing conditions of purified proteins visualized by Coomassie blue staining.

For the single domains DBL5ε and DBL6ε, the encoding gene sequences from the 3D7 parasite strain were cloned into the pET15b vector, and the gene sequence encoding 3D7-DBL2X was cloned into a modified pET21b vector. All proteins were expressed in the Escherichia coli SHuffle strain (NEB) and purified as previously described (17). The AMA1 recombinant protein of clone FVO (FVO-AMA1) was a kind gift of Louis Miller (NIH, Bethesda, MD) (27).

Enzyme-linked immunosorbent assay (ELISA).

Antigens were coated overnight at 1 μg/ml in phosphate-buffered saline (PBS). Then, wells were blocked with PBS–4% bovine serum albumin (BSA) for 1 h at 37°C. Plasma samples, diluted 1/100 in PBS–2% BSA, were added to the wells and incubated for 1 h at 37°C. After three washes with PBS–0.5% Tween 20 (PBST), anti-human IgG (Fcγ specific), horseradish peroxidase (HRP) conjugated (Jackson ImmunoResearch) and diluted at 1/4,000 in PBS–2% BSA, was added to each well and incubated for 1 h at 37°C. Following three washes with PBST, the HRP substrate tetramethylbenzidine (Bio-Rad) was added to the wells, and optical densities (OD) were measured at 655 nm.

A pool of plasma from multigravid women living in an area of malaria endemicity previously identified to be highly reactive toward the VAR2CSA protein (23) (positive pool) as well as a pool of plasma from adult French donors never exposed to P. falciparum and not reacting or poorly reacting with the VAR2CSA protein (negative pool) was used in each ELISA plate for arbitrary unit conversion of OD values.

Statistical analysis.

For each ELISA, the OD was converted into arbitrary units (AU) using the following formula (28): AU = 100 × (lnOD_test − lnOD_{negative pool})/(lnOD_{positive pool} − lnOD_{negative pool}). Group comparisons were performed using a Mann-Whitney rank sum test. Three group comparisons were performed using a Kruskal-Wallis one-way analysis and Dunn's method. Given the multiplicity of tests achieved, we considered significant P values of ≤0.01 instead of the usual threshold value of 0.05.

RESULTS

Multigravidae are more protected than primigravidae against PAM.

In order to assess the development of the humoral immunity toward VAR2CSA, 293 women, pregnant for less than 6 months, were enrolled in the study and followed by active and passive malaria detection through monthly antenatal care visits and through weekly home visits until delivery. Fifty-nine primigravidae and 234 multigravidae (mean age, 21.4 and 27.6 years, respectively) were enrolled in the study (Table 1). The P. falciparum infection rates during the course of pregnancy were similar between primigravid and multigravid women. However, placental sequestration at delivery was more frequently observed in primigravidae than in multigravidae (26.5% and 10.6%, respectively). Consistent with these data, 71.4% of multigravidae versus 42.8% primigravidae infected at least once during the follow-up had no parasites in the placenta at delivery. Furthermore, the percentage of women for which the presence of parasites in the placenta was detected at delivery was inversely correlated to the number of pregnancies (Table 2). Taken together, these results indicate an increasing degree of protection against PAM following multiple pregnancies and legitimate the use of this cohort to assess how humoral immunity toward VAR2CSA develops.

TABLE 1.

Characterization of cohort and status of infection in women

Parameter	Value for the group
	Primigravid (n = 59)	Multigravid (n = 234)
Mean age (yr [range])	21.4 (16–35)	27.6 (16–46)
Mean gravidity (range)	1	4.2 (2–10)
No. (%) of women infected during the follow-up	21 (35.6)	84 (35.9)
No. of infections during the follow-up	1–4	1–6
Mean no. of infections per infected woman during follow-up	1.9	1.6
No. (%) of women not infected during pregnancy and presenting no sign of P. falciparum placental infection at delivery (group a)	20 (58.8)	103 (64)
No. (%) of women infected at least once during pregnancy and presenting no sign of P. falciparum placental infection at delivery (group b)	5 (14.7)	41 (25.4)
No. (%) of women infected at least once during pregnancy and presenting parasite sequestration in the placenta at delivery (group c)	9 (26.5)	17 (10.6)

TABLE 2.

Acquisition of PAM protective immunity

Parameter	Value
Graviditya	1	2–4	5–7	8–10
No. of women	59	145	61	26
% women presenting parasites in the placenta at delivery	26.5	13.1	9.8	7.7

^aThe precise gravidity status of 2 of the 234 multigravid women could not be established.

The levels of antibodies recognizing VAR2CSA single and multiple domains increased during the course of pregnancy in women living in areas of malaria endemicity.

In this study, we performed an in-depth characterization of the development of humoral immunity in pregnant women living in areas of malaria endemicity toward the PAM-specific VAR2CSA protein using different VAR2CSA-derived highly purified recombinant proteins as target antigens (Fig. 1A and B). Before analyzing both the impact of gravidity and P. falciparum infection on antibody recognition against the target antigens, we first needed to ensure that their recognition by plasma IgG corroborated the intimate link existing between VAR2CSA and pregnancy. We first assessed the levels of antibodies toward VAR2CSA-derived recombinant proteins in plasma samples from primigravid women at enrollment (PE group), Beninese men (BM), and French adult (FA) donors never exposed to P. falciparum. FVO-AMA1, a P. falciparum-specific antigen involved in parasite invasion, was used as a non-PAM-related control throughout this work. The levels of antibodies directed toward the VAR2CSA single and multiple domains were not statistically different between Senegalese primigravid women at inclusion and men from Benin. However, we observed a statistically significant higher level of antibodies to FVO-AMA1 in Beninese men than in Senegalese primigravidae. This difference is most likely due to a higher level of exposure to P. falciparum in Benin (35 infected bites per year compared to 10 in Senegal). In contrast to these results, statistically significant differences were observed between both malaria-exposed populations and French adults never exposed to P. falciparum. Indeed, none of the malaria antigens were substantially recognized by plasma of the French adults (Fig. 2).

FIG 2.

No significant difference in the levels of antibodies recognizing VAR2CSA proteins is observed between primigravid women at inclusion and men. Reactivity of plasma IgG from primigravid women at the time of enrollment (PE) (n = 59) was compared to that of men (BM) (n = 24) and French naive donors (FA) (n = 24) on 3D7-VAR2CSA single and multidomains. FVO-AMA1 was used as a non-PAM-specific plasmodial antigen. ELISA results are expressed in arbitrary units (AU). A group comparison was performed using Kruskal-Wallis one-way analysis and Dunn's method. P values of ≤0.01 were considered statistically significant (*). The table summarizes the P values for the pairwise comparisons.

We next assessed the levels of antibodies in plasma of 293 pregnant Senegalese women at the time of enrollment and at delivery regardless of their gravidity status and their P. falciparum infection history during pregnancy and at delivery.

We observed that, in addition to FVO-AMA1, all VAR2CSA protein constructs were markedly recognized by plasma IgG from pregnant women at the time of enrollment and at delivery (Fig. 3). Furthermore, the women tended to present higher levels of antibodies toward the VAR2CSA constructs at delivery than at enrollment, with statistical significance being reached for DBL6ε, DBL4ε-6ε, and DBL1X-6ε. The antibody levels to FVO-AMA1 remained unchanged. Taken together, these results confirmed the specific association between VAR2CSA and PAM, validating all our protein constructs as suitable antigens to analyze the acquisition of the humoral immunity toward VAR2CSA in pregnant women living in areas of malaria endemicity.

FIG 3.

Levels of antibodies recognizing VAR2CSA single and multiple domains are increased during the course of pregnancy in women living in areas of malaria endemicity. Reactivity of plasma IgG from Senegalese women at the time of enrollment (E) (n = 274) was compared to that at delivery (D) (n = 203) on 3D7-VAR2CSA single and multiple domains. FVO-AMA1 was used as a non-PAM-specific plasmodial antigen. ELISA results are expressed in arbitrary units (AU). A group comparison was performed using a Mann-Whitney rank sum test. P values of ≤0.01 were considered statistically significant (*).

DBL1X-3X and DBL1X-6ε immune recognition is parity dependent at the time of delivery.

Antibodies toward VAR2CSA are acquired in a parity-dependent manner in pregnant women exposed to P. falciparum infection (29). We therefore analyzed the impact of parity on the immune recognition of different 3D7-VAR2CSA constructs using plasma samples originating from either primigravid or multigravid women, regardless of P. falciparum infection (Fig. 4A and B). At enrollment, multigravidae possessed higher levels of IgG able to recognize the full-length VAR2CSA-DBL1X-6ε (P = 0.005) than primigravidae (Fig. 4A). However, at the time of delivery, multigravidae presented higher levels of antibodies recognizing DBL1X-3X as well as DBL1X-6ε than primigravidae (P = 0.009; P = 0.01, respectively) (Fig. 4B), suggesting a B cell memory-boosting effect toward these proteins. Although multigravidae tended to have in general more IgG able to react with all the other VAR2CSA constructs than primigravidae, these differences did not reach statistical significance.

FIG 4.

DBL1X-6ε and DBL1X-3X immune recognition is parity dependent at the time of delivery. Reactivity of plasma IgG from primigravidae (P) was compared to that of multigravid (M) Senegalese women at the date of enrollment (A) and at delivery (B) on 3D7-VAR2CSA single and multiple domains. FVO-AMA1 was used as a non-PAM-specific plasmodial antigen. ELISA results are expressed in arbitrary units (AU). A group comparison was performed using a Mann-Whitney rank sum test. P values of ≤0.01 were considered statistically significant (*). Primigravidae enrollment, n = 54; multigravidae enrollment, n = 218; primigravidae delivery, n = 36; multigravidae delivery, n = 166.

Antibodies raised in response to P. falciparum infection during pregnancy target the CSA-binding region DBL1X-3X.

Next, we assessed the impact of P. falciparum infection during the course of pregnancy on the immune recognition of VAR2CSA-derived antigens at delivery. Parasitological data gathered at each prenatal visit allowed us to further group primigravidae and multigravidae in relation to P. falciparum infection history during the course of pregnancy. Three subgroups were defined as follows: group a, women that were not infected during pregnancy and did not present with a P. falciparum placental infection at delivery; group b, women that were infected at least once during pregnancy and did not present with a P. falciparum placental infection at delivery; group c, women that were infected at least once during pregnancy and presented with a P. falciparum placental infection at delivery.

Considering the primigravidae (Fig. 5A), we observed that women who were not infected during pregnancy had lower levels of antibodies toward all the VAR2CSA constructs than women infected at least once. Comparison between women from groups a and c revealed significant differences in the levels of antibodies recognizing DBL1X-6ε (P = 0.005), DBL1X-3X (P = 0.001), and DBL2X (P = 0.001), with the level of recognition higher in women infected at least once during pregnancy and presenting a P. falciparum placental infection at delivery. No significant difference was observed between primigravid women who were not infected during pregnancy or who were infected at least once but did not present a P. falciparum placental infection at delivery (groups a and b). However, a similar analysis performed on multigravidae (Fig. 5B) showed significant differences in the levels of antibodies recognizing DBL1X-6ε (P = 0.002), DBL1X-3X (P = 0.002), DBL2X (P = 0.001), and DBL6ε (P = 0.01) between women from groups a and b but not between women from groups a and c. Interestingly, women from group c that were infected at least once during pregnancy and presented a P. falciparum placental infection tended to have lower levels of antibodies able to recognize all the VAR2CSA-derived antigens than women from group b that had no detectable placental infection at delivery. Nevertheless, these differences did not reach statistical significance. The levels of anti-FVO-AMA1 IgG between groups a, b, and c remained unchanged.

FIG 5.

Antibodies raised in response to P. falciparum infection during pregnancy are directed toward DBL1X-6ε, DBL1X-3X, DBL2X, and DBL6ε. Reactivity of plasma IgG at delivery from primigravid (A) and multigravid (B) Senegalese women on 3D7-VAR2CSA single and multiple domains considering their P. falciparum infection history was determined for the following groups: group a, women noninfected during pregnancy and presenting a P. falciparum-negative placental biopsy specimen at delivery; group b, women infected at least once during pregnancy and presenting a P. falciparum-negative placental biopsy specimen at delivery; group c, women infected at least once during pregnancy and presenting a P. falciparum-positive placental biopsy specimen at delivery. FVO-AMA1 was used as a non-PAM-specific plasmodial antigen. A multiple-group comparison was performed using Kruskal-Wallis one-way analysis and Dunn's method. P values of ≤0.01 were considered statistically significant (*). Primigravidae: n = 20 (a), n = 5 (b), and n = 9 (c); multigravidae, n = 103 (a), n = 41 (b), and n = 17 (c). The tables summarize the P values for the pairwise comparisons.

DISCUSSION

To date, many data support VAR2CSA as the leading vaccine candidate against PAM (30–33). Current vaccine strategies aim to elicit a potent and long-lasting humoral response against placental parasites that could inhibit the binding of infected erythrocytes to the placental syncytiotrophoblastic ligand CSA and/or actively participate in parasite clearance by mediating opsonic phagocytosis. It is therefore of crucial importance to get a better understanding on how natural immunity against VAR2CSA develops.

Many studies have assessed the development of humoral immunity against placental malaria using VAR2CSA single domains. For instance, the 3D7-DBL5ε domain was identified as a VAR2CSA module displaying a parity-dependent recognition and therefore was considered an interesting vaccine candidate against PAM (34).

Nevertheless, recent work performed using the full-length extracellular part of VAR2CSA revealed the functional importance of the overall folding of the protein (15, 16). These findings indicate that antibody response to single domains may not fully reflect the natural immune response to the native VAR2CSA proteins. Indeed, a recent study showed that high levels of antibodies directed toward the full-length FCR3-VAR2CSA play a preponderant role in protection against PAM (21). We have recently demonstrated that the N-terminal region (DBL1X-3X) of VAR2CSA presents binding properties to CSA similar to those of the full-length protein and that the DBL4ε-6ε region is not directly implicated in CSA binding (17).

Based on this evidence, we assessed the humoral immunity development toward the full-length VAR2CSA, the multidomains DBL1X-3X and DBL4ε-6ε, and single DBL domains from the 3D7 variant using plasma samples from 293 Senegalese pregnant women as well as Beninese men.

Several studies reported the presence of antibodies against VAR2CSA in plasma from men and children living in areas of malaria endemicity (35, 36), and the plasma samples from the 293 pregnant Senegalese women of our study also presented nonnegligible antibody levels directed toward all VAR2CSA antigens tested at the time of enrollment, regardless the gravidity status of the women (Fig. 3). CSA proteoglycans have been reported to be present in the placenta by the 12th week of pregnancy, and recent work has shown that VAR2CSA-expressing parasites can be preponderantly detected in the mother's peripheral blood as early as in the first trimester of pregnancy (37). It has to be noted that most women were enrolled in the study during the second trimester of pregnancy and therefore might already have been exposed to P. falciparum infection early on. We also failed to observe a significant difference in the levels of antibodies directed toward the VAR2CSA recombinant proteins between Senegalese primigravid women at enrollment and Beninese men, indicative of either a comparable level of exposure in both populations to VAR2CSA or to cross-reactive epitopes present on other PfEMP1 variants. Indeed, although VAR2CSA protein architecture singularly differs from other PfEMP1s, the presence of DBL domains remains a common feature between all the different variants. Therefore, previous P. falciparum infections and exposure to non-VAR2CSA PfEMP1s could partially explain the observed reactivity of the plasma from primigravid women and men toward the antigens considered in this study.

As the levels of antibodies recognizing VAR2CSA single and multiple domains increased during the course of pregnancy in women living in areas of malaria endemicity, our results confirmed the specific association of VAR2CSA with PAM and validated our recombinant proteins as suitable antigens to further analyze the acquisition of the humoral immunity toward VAR2CSA.

Using these antigens, we showed that the recognition of the full-length VAR2CSA was significantly higher at enrollment in multigravid than in primigravid women (Fig. 4A). However, at delivery, DBL1X-3X and DBL1X-6ε proteins were recognized in a parity-dependent manner (Fig. 4B). Taken together, these results suggest that a B cell memory-boosting effect is more effective toward these proteins and that the parity-dependent recognition of VAR2CSA relies mostly on antibodies targeting epitopes carried by the CSA-binding region DBL1X-3X. These results are in line with previous studies showing that multigravidae have higher levels of antibodies able to recognize VAR2CSA-expressing parasites than primigravidae (5, 38).

We next assessed the immune recognition of VAR2CSA-derived antigens considering the P. falciparum infection history during the course of pregnancy (Fig. 5). Primigravid women not infected since their enrollment had lower levels of antibodies directed against the VAR2CSA constructs at delivery than women infected at least once (Fig. 5A). Interestingly, the immune recognition of DBL1X-6ε, DBL1X-3X, and DBL2X was higher in primigravidae presenting parasites in the placenta at delivery than in women without signs of infection during the course of pregnancy. Furthermore, multigravid women not infected since their enrollment had lower levels of antibodies toward VAR2CSA than women infected at least once and presenting a P. falciparum-negative placenta at delivery (Fig. 5B). Once again, significance was observed for the immune recognition of DBL1X-3X, DBL1X-6ε, and DBL2X and also for DBL6ε. Since we did not have access to histological data at delivery, such as the presence of hemozoin in placental macrophages or of fibrinogen in the intervillous space, the frequency of placental malaria developed early in pregnancy might be slightly underestimated in both groups (primigravidae and multigravidae).

As previously discussed, immune recognition data generated using single domains must be interpreted with care. Therefore, the statistical significance observed for DBL6ε does not necessarily reflect a real association with protection as the observed differences in antibody levels could originate from the exposure/recognition of nonrelevant epitopes. Indeed, these recombinant constructs might not harbor the same folding as in the context of the full-length protein, thus eventually exposing epitopes not relevant for protection. Corroborating this hypothesis, no statistical difference in antibody levels was observed between women from the groups A and B for the multidomain DBL4ε-6ε.

Our results indicate that infection during pregnancy boosted the immune response toward VAR2CSA and notably against the CSA-binding region. These results confirm that the acquisition of antibodies against placental parasites and then against VAR2CSA start developing as early as in first pregnancy (20, 38, 39) and highlight the importance of the N-terminal region of VAR2CSA (DBL1X-3X) in the development of the humoral response against PAM. This is borne out by other studies revealing the detailed mechanisms of other malaria-derived protein/ligand interactions (40–43).

The observed cross-reactivity falls in line with the hypothesis that the N-terminal region of VAR2CSA is under conformational constraint to maintain the high-affinity CSA-binding activity (44), leading to the exposure of conserved conformational epitopes that are shared between the 3D7-VAR2CSA recombinant proteins used in this study and the VAR2CSA variants expressed by the field parasites encountered by the pregnant women.

A recent study showed that the response to ID1-ID2a, which spans the DBL2 domain (central core of the DBL1X-3X region), is not recognized specifically during pregnancy (22). Indeed, antibody levels did not change during the course of pregnancy and did not increase with gravidity. Furthermore, no significant differences were found in antibody levels between infected and uninfected women at delivery. These results are in line with our results showing no parity-dependent recognition of the DBL2X domain. An explanation could be that these recombinant proteins do not expose the proper conformational epitopes associated to PAM or are hidden in the context of the full-length VAR2CSA.

Taken together, our data suggest that high levels of antibodies against VAR2CSA recombinant proteins encompassing the high-affinity CSA-binding region are associated with the control of placental infection. These results highlight the key role of the CSA-binding N-terminal DBL1X-3X region of VAR2CSA in the acquisition of a cross-reactive immune response and therefore provide novel insights on how protection against PAM is acquired. In the context of a PAM vaccine development, this study provides new information for the rational design of a VAR2CSA vaccine based on the N-terminal DBL1X-3X region of the protein.