ABSTRACT
Placental malaria, caused by sequestration of Plasmodium falciparum-infected erythrocytes in the placenta, is associated with increased risk of maternal morbidity and poor birth outcomes. The parasite antigen VAR2CSA (variant surface antigen 2-chondroitin sulfate A) is expressed on infected erythrocytes and mediates binding to chondroitin sulfate A, initiating inflammation and disrupting homeostasis at the maternal-fetal interface. Although antibodies can prevent sequestration, it is unclear whether parasite clearance is due to antibodies to a single Duffy binding-like (DBL) domain or to an extensive repertoire of antibodies to multiple DBL domains and allelic variants. Accordingly, plasma samples collected longitudinally from pregnant women were screened for naturally acquired antibodies against an extensive panel of VAR2CSA proteins, including 2 to 3 allelic variants for each of 5 different DBL domains. Analyses were performed on plasma samples collected from 3 to 9 months of pregnancy from women living in areas in Cameroon with high and low malaria transmission. The results demonstrate that high antibody levels to multiple VAR2CSA domains, rather than a single domain, were associated with the absence of placental malaria when antibodies were present from early in the second trimester until term. Absence of placental malaria was associated with increasing antibody breadth to different DBL domains and allelic variants in multigravid women. Furthermore, the antibody responses of women in the lower-transmission site had both lower magnitude and lesser breadth than those in the high-transmission site. These data suggest that immunity to placental malaria results from high antibody levels to multiple VAR2CSA domains and allelic variants and that antibody breadth is influenced by malaria transmission intensity.
INTRODUCTION
Plasmodium falciparum evades host immune clearance in the spleen by sequestering to microvascular endothelial cells (32, 42). In pregnant women, P. falciparum-infected erythrocytes (IE) bind to chondroitin sulfate A (CSA) (24, 35), a glycosaminoglycan abundantly expressed in the intervillous space and on placental syncytiotrophoblast cells (1, 2, 40, 48). Binding is mediated by VAR2CSA (variant surface antigen 2-chondroitin sulfate A), a variant antigen of the PfEMP1 family expressed on the surface of IE by the parasite (45, 46).
P. falciparum infections during pregnancy have been associated with maternal anemia and poor pregnancy outcomes, specifically, low birth weight babies (4, 6, 19, 27, 29). Low birth weight is a major cause of neonatal and infant mortality in both developed and developing countries (25, 36, 55) and may be associated with long-term health consequences (28, 30, 36, 43). Furthermore, babies born to mothers who have placental malaria are more likely to have clinical malaria earlier in life than babies born to placental-malaria-negative (PM−) mothers (11, 27, 49). Despite relentless efforts against malaria, children younger than 5 years of age in sub-Saharan Africa still account for the majority of malaria-related deaths worldwide (16, 52, 56). Thus, preventing malaria in pregnant women is important, not only to reduce maternal morbidity but also to minimize the negative impact of malaria on the health of young children.
Antibodies (Ab) in immune sera have been shown to inhibit IE from binding to CSA in vitro and to placental tissues ex vivo (23, 26, 31). Production of Ab to placental IE is gravidity dependent (45, 54). In a low-transmission area with 13 infectious bites per year, antiadhesion Ab were detected in primigravid women starting around 20 weeks of pregnancy, compared to 12 weeks for multigravidae (44). The delay and/or lack of Ab in primigravid women appears to account, in part, for the increased prevalence of placental malaria in these women (18, 31, 37, 38) and supports a protective role of Ab in controlling placental infections.
Antibody recognition of CSA-binding parasites and anti-CSA adhesion activity are strain transcending, as Ab from women of different geographic origins, e.g., Africa, Asia, and South America, inhibit placental parasite isolates of diverse origins from binding to CSA in vitro and ex vivo (13, 15, 23, 26, 31). These data support the idea that a universal vaccine can be developed. However, it is unknown whether strain-transcending responses are due to Ab that target conserved epitopes or a repertoire of Ab against polymorphic epitopes, since polymorphisms exist in VAR2CSA (5, 17, 53). Thus, polymorphism in VAR2CSA may need to be taken into consideration when assessing protective Ab responses, as well as in vaccine development.
Though the protective role of Ab in placental malaria is well recognized, the specificity of Ab that prevents parasite sequestration in the intervillous space is unknown. Most studies have not found a difference in the amount of Ab to CSA-binding parasites at delivery between women with and without placental malaria (44); instead, higher Ab levels often correlated with active malaria infections (15, 21, 31, 51). VAR2CSA is a large, ∼350-kDa transmembrane protein with six Duffy binding-like (DBL) domains and a cysteine-rich interdomain region referred to as CIDRPAM (5, 17). A previous study found that women with high levels of Ab to DBL5 delivered higher birth weight babies, but only two of the six VAR2CSA domains were examined (45). Therefore, it remains unclear whether Ab to one or multiple DBL domains are associated with parasite clearance. In this study, we examined the kinetics of Ab acquisition to VAR2CSA domains in settings with low and high malaria transmission. We evaluated whether the malaria transmission intensity affected the magnitude and breadth of Ab production to VAR2CSA domains and strain variants and whether responses to specific VAR2CSA domains correlated with the absence of placental malaria at delivery.
MATERIALS AND METHODS
Study sites.A prospective cohort study was conducted between 2001 and 2005 in the small rural village of Ngali II and in the capital city Yaoundé in Cameroon. Malaria transmission at both sites is perennial, with 2 wet and dry seasons. Entomological inoculation rates were estimated to be 256 infectious bites per person per year in Ngali II (33) and 13 infectious bites per person per year in Yaoundé (34).
Study populations.Pregnant women were recruited at government-operated hospitals and health clinics during their first trimester of pregnancy. After providing informed consent, women received a comprehensive health evaluation, and clinical histories were obtained along with a peripheral blood sample. Participants were followed at the antenatal clinics, where each month, pregnancy- and malaria-related information and peripheral blood samples were collected. Information on the baby, including birth weight, was obtained at delivery. Women who became blood smear-positive for P. falciparum were provided with antimalarial drugs and iron supplements, based on the government policy for treatment of pregnant women. Initially, chloroquine was the drug of choice, but due to increasing drug resistance, artemisinin in combination with amodiaquine was adopted as the first-line malaria drug in 2004, the last year of sample collection (33). None of the women received antimalarial treatment close enough to delivery to affect their placental malaria status. The study was completed before the implementation of intermittent preventive treatment (IPTp) and bed nets in Cameroon. Thus, the clearance of placental infection before delivery was largely due to naturally acquired immunity. Participants gave written or oral informed consent. The study was approved by the National Ethics Committee of Cameroon and the Institutional Review Board of Georgetown University. Use of the coded archived samples in the current study was found to be exempt from human subject research by the Committee on Human Studies, University of Hawai'i, Mānoa.
Sample selection.In this study, sera from 39 women in Ngali II and 50 women in Yaoundé were selected for evaluation. The inclusion criteria included available demographic information on the woman's age, gravidity, and malarial status during each visit, confirmed malaria infection during ≤6 months of pregnancy (MoP) based on slide or PCR data, and ≥4 samples from the participant collected throughout the course of pregnancy (6 ± 1.2 samples [mean ± standard deviation] per woman). Thus, all women were documented to have had at least one blood-positive P. falciparum infection by the end of the second trimester and the infections were early enough in pregnancy to allow them to mount an Ab response and clear IE from the placenta prior to delivery. Information and plasma samples collected at delivery were available for only 27 of the 39 women from Ngali II and 48 of the 50 women from Yaoundé. Plasma samples from 20 males in Ngali II and 20 in Yaoundé were used as negative controls in the serological assays.
Diagnosis of placental malaria.Thick and thin blood smears of placental intervillous space blood and impression smears of placental tissue were prepared at delivery. All slide preparations were read by two laboratory members, and a woman was determined to have placental malaria if IE were observed in either the blood or impression smear. Past infection, i.e., the presence of hemozoin pigment in macrophages but no IE, was considered to be placental malaria negative.
Recombinant proteins.A panel of recombinant VAR2CSA proteins representing 5 individual DBL domains and up to 3 allelic variants was used in the serological assays. Each of the domains has been characterized in previous studies (Table 1). Cloning and production of recombinant proteins for DBL1 through DBL6 (except for DBL2) from parasite strains 7G8 (South America), 3D7 (origin ambiguous), and FCR3 (origin ambiguous, genotype most close to South East Asia) (39) were performed in Pichia pastoris (Pp in protein designations) as previously described (7, 10). All proteins contained a histidine tag at the C terminus. The identity of recombinant proteins was confirmed by mass spectrometry analysis. Separate sets of recombinant proteins for DBL1, DBL1+2, DBL3, DBL4, DBL5, and DBL6 from parasite strain FCR3 were produced in insect cells with baculovirus (BV) vector pAcGP67-A (BD Biosciences) containing a histidine tag at the C-terminal end (41). Protein sequencing and mass spectrometric peptide mapping were performed by Alphalyse (Denmark). An additional DBL3 protein from FCR3 was produced in Escherichia coli (Ec in protein designations). Detailed information on the construct boundaries of the 16 VAR2CSA recombinant proteins, representing 13 different DBL variants, is provided in Table 1. Two additional antigens were used to evaluate general humoral responses to P. falciparum: recombinant protein from the C-terminal regions of the merozoite surface protein-1 (MSP142) of the FVO strain expressed in E. coli and recombinant apical merozoite antigen-1 (AMA-1) of the 3D7 strain expressed in yeast. These were provided by the Malaria Vaccine Development Branch, National Institute of Allergy and Infectious Disease, National Institutes of Health, Rockville, MD.
Recombinant proteins used in the study
Measurement of Ab levels.IgG levels to each recombinant protein were measured using the bead-based multianalyte profiling assay. In brief, saturating amounts (1 μg) of each protein were covalently coupled to 1 million SeroMAP beads (Luminex Corp., Austin, TX). Coupling was performed at pH 7 overnight at 4°C as previously described (22), followed by blocking with 1% bovine serum albumin (BSA) in phosphate-buffered saline (PBS, pH 7.0) for 24 h. Since each of the recombinant proteins has approximately the same molecular weight and maximal amounts of antigen were coupled, the assay sensitivity was similar for each of the proteins. In optimizing the assay, DBL domains and variants were tested individually and in different combinations (multiplex). Based on the results, pools of 8 to 12 antigen-coupled beads with different spectral addresses were created, so that interference among DBL domains in the multiplex assay did not occur. For Ab measurements, pools of antigen-coupled beads were incubated with 50 μl of plasma samples (diluted to 1:500 with 1% BSA–PBS) for 1 h on a titer plate shaker (Lab-Line Instruments, Inc., Melrose Park, IL) at speed 5 at room temperature (RT). After 3 washings with 1% BSA–PBS, 100 μl of R-phycoerythrin-conjugated, affinity-purified F(ab′)2, heavy chain-specific goat anti-human IgG (Jackson ImmunoResearch) at 2 μg/ml was added to the beads and incubated for 1 h with shaking at RT. After washing, median fluorescence intensities (MFI) were obtained using a LiquiChip 100 analyzer (Luminex Corp., Austin, TX). Median MFI values are based on values from at least 100 beads per antigen. Plasma pooled from 20 adult Cameroonian males living at each site were used as negative controls for VAR2CSA antigens and plasma from U.S. adults who had never been exposed to malaria served as the negative controls for MSP142 and AMA1. The cutoff for seropositivity for the VAR2CSA domains was the mean plus 2 standard deviations (SD) of the results for the male controls after the exclusion of outliers. Outliers (≤10% of total data points) were excluded in order to minimize false-negative results in pregnant women, as a previous study has shown that an occasional male can have Ab against CSA-binding IEs (14) (see Fig. S1 and S2 in the supplemental material for details). Exact cutoff values are shown in the legends of Fig. 1 and 2. The cutoff for MSP142 and AMA1 was the mean plus 2 SD based on the results for the U.S. adult controls. For ease of comparison, Ab results were categorized into 4 groups: Ab negative (MFI below the cutoff) and low (above the cutoff but <5,000 MFI), intermediate (5,000 to 10,000 MFI), and high (>10,000 MFI) levels. Women with low, intermediate, and high Ab levels were defined as seropositive, and women with intermediate and high Ab levels were considered to have strong Ab responses.
Prevalence of antibodies to VAR2CSA domains in women living in the high-transmission site of Ngali II. Sera collected longitudinally from 39 women (6 ± 1.2 samples per woman) between 3 and 9 months of pregnancy were screened for IgG Ab against 13 variants of VAR2CSA. Antibody results were grouped into 4 categories: Ab negative (less than the mean + 2 SD of the results for 20 adult males living in Ngali II); low (>cutoff but <5,000 MFI), intermediate (5,000 to 10,000 MFI), and high (>10,000 MFI). The total number of data points per month of pregnancy ranged from 19 to 37 serum samples. MSP-142 and AMA1 were included as non-VAR2CSA controls. Note: women with intermediate and high Ab levels were considered to have strong antibody responses. Specific cutoffs were as follows: DBL1 7G8-Pp (1,247 MFI), DBL1 3D7-Pp (1,122 MFI), DBL1 FCR3-Pp (438 MFI), DBL3 7G8-Pp (884 MFI), DBL3 FCR3-BV (1,207 MFI), DBL3 FCR3-Ec (768 MFI), DBL4 7G8-Pp (534 MFI), DBL4 FCR3-BV (3,490 MFI), DBL4 FCR3-Pp (502 MFI), DBL5 7G8-Pp (893 MFI), DBL5 FCR3-BV (542 MFI), DBL5 3D7-Pp (772 MFI), DBL6 7G8-Pp (3,202 MFI), DBL6 FCR3-BV (1,725 MFI), DBL6 FCR3-Pp (1,370 MFI), and DBL1+2 FCR3-BV (1,584 MFI). MFI, median fluorescence intensity.
Prevalence of antibodies to different VAR2CSA domains in women living in the low-transmission site of Yaoundé. Sera collected longitudinally from 50 women (6 ± 1.2 samples per woman) were screened for IgG antibodies against the 13 variant domains of VAR2CSA. Ab values were grouped into 4 categories as described in the Fig. 1 legend, using a cutoff of the mean + 2 SD of the results for 20 adult males living in Yaoundé. The number of data points per month of pregnancy ranged from 39 to 46 serum samples. Specific cutoffs were as follows: DBL1 7G8-Pp (1,046 MFI), DBL1 3D7-Pp (854 MFI), DBL1 FCR3-Pp (425 MFI), DBL3 7G8-Pp (419 MFI), DBL3 FCR3-BV (598 MFI), DBL3 FCR3-Ec (288 MFI), DBL4 7G8-Pp (184 MFI), DBL4 FCR3-BV (1,620 MFI), DBL4 FCR3-Pp (427 MFI), DBL5 7G8-Pp (408 MFI), DBL5 FCR3-BV (187), DBL5 3D7-Pp (369 MFI), DBL6 7G8-Pp (1,636 MFI), DBL6 FCR3-BV (751 MFI), DBL6 FCR3-Pp (1,089 MFI), and DBL1+2 FCR3-BV (626 MFI). MFI, mean fluorescent intensity.
Breadth of the Ab response.The 13 recombinant proteins represent 5 DBL domains (DBL1, -3, -4, -5, and -6). In determining the breadth of the response to the 5 DBL domains, a woman was considered to be seropositive for the domain if she had Ab to one or more of the strain variants. DBL1+2 was excluded from this analysis because it was impossible to distinguish between DBL1 and DBL2. Accordingly, a woman could have Ab to 0 to 5 DBL domains. The 13 recombinant proteins included variants from different geographical regions, and each woman was scored as having Ab to 0 to 13 different variants.
Statistical analysis.Comparisons between the two sites for maternal age and length of pregnancy were made using the Student t test (unpaired). Fisher's exact test was used to compare the percentages of primigravid women and slide positivity between the two study sites, as well as for comparison between domain/variant Ab reactivity. Longitudinal Ab levels between women with and without placental malaria were analyzed using multilevel polynomial regression analysis. In this analysis, longitudinal Ab levels, breadth of Ab response to DBL domains, and allelic variants were dependent variables, while placental malaria status was the independent variable. Also, changes in Ab levels and breadth were analyzed between and within groups. Antibody levels and breadth to domains/variants were dependent variables, while length of pregnancy (linear and quadratic) and the interaction between malaria status at delivery, gravidity, and time were independent variables. The Student t test was used for analysis of Ab breadth between women with and without placental malaria at 3 to 4 and 8 to 9 MoP. Results with P values of ≤0.05 were considered to be significant. Bonferroni correction for the P values was applied when appropriate.
RESULTS
Description of women.The characteristics of the 89 women whose plasma were used in the study are shown in Table 2. No significant difference in age, length of pregnancy, or proportion of primigravidae was found between women enrolled at the two sites. Although the levels of prevalence of malaria during pregnancy and placental malaria at delivery were higher in women living in Ngali II than Yaoundé, the differences did not reach significance.
Description of women in the study
Domain-specific VAR2CSA IgG acquisition throughout pregnancy.Samples collected during the course of pregnancy from 39 women in Ngali II (Fig. 1) and 50 women in Yaoundé (Fig. 2) were screened for Ab against 16 recombinant proteins representing 13 different DBL variants, as well as MSP142 and AMA1. Considerable variation in the kinetics of the Ab responses of the women was observed (see Fig. S1 and S2 in the supplemental material), but a general pattern of responsiveness was present.
In Ngali II, DBL5 was the best recognized domain based on early Ab production, magnitude of the Ab response, and seroprevalence (Fig. 1). At 3 MoP, >70% of Ngali-II women had Ab to DBL5, and by 4 months, >90% had Ab to all three DBL5 allelic variants (Fig. 1). The proportion of women with strong Ab responses to DBL5 (7G8-Pp and FCR3-BV) increased rapidly from 25 to 30% at 3 MoP to 75 to 80% at 9 MoP (Fig. 1). At 6 MoP, more women had strong responses to DBL5 than to DBL1, DBL4, and DBL6 (P < 0.0001 for all 3 variants) and DBL3 (7G8-Pp and 7G8-Ec; P = 0.04). Thus, most women had Ab to DBL5 early in pregnancy and their levels increased rapidly and remained high throughout pregnancy. The second most commonly recognized domain was DBL3 (7G8-Pp and 7G8-Ec; P = 0.0004). At 3 MoP, 50% of Ngali-II women had Ab to DBL3, and by 6 MoP, 90% were seropositive for both DBL3 variants with >40% having strong responses (Fig. 1). In contrast, Ab responses to DBL1, DBL4, and DBL6 were delayed and not as strong (Fig. 1). Only 20 to 35% of the women had Ab to the DBL1 variants early in pregnancy, and although 60 to 80% of them ultimately developed DBL1 Ab, <25% produced strong DBL1 responses prior to delivery (Fig. 1). A similar response to the DBL1+2 domain was observed, with 40% of women having Ab early in pregnancy, 80% ultimately developing Ab to DBL1+2, and up to 30% having strong responses by term. Unlike the other domains, the Ab response to DBL6 differed among the 2 allelic forms (7G8 and FCR3), with 5% to 42% of women having Ab to DBL6 at 3 MoP, 5% to 82% producing Ab by term, and 0% to 32% developing strong responses depending on the recombinant protein (Fig. 1). These data suggest that DBL5 and DBL3 are the most immunogenic domains in VAR2CSA and that Ab levels to all other VAR2CSA domains increase slowly over pregnancy at the high-transmission site (Fig. 1).
In the low-transmission site of Yaoundé, a similar trend was observed, with the highest prevalence of Ab being to DBL5 (50 to 60%) and DBL3 (37 to 57%), followed by the other domains (Fig. 2); however, no statistical difference in either the prevalence or levels of Ab was found among the 16 recombinant proteins. All women were infected with P. falciparum at least once during pregnancy, yet 40% of them failed to produce Ab to any of the domains (Fig. 2).
A comparison of Ab responses between village women (Ngali II) and city women (Yaoundé) suggests that transmission intensity directly influenced the Ab response to VAR2CSA domains. For example, by 6 MoP, over 20% of women in Ngali II had strong responses to DBL5 (all 3 variants), DBL3 (7G8-Pp and FCR3-Ec), DBL4 (FCR3-BV), DBL6 (FCR3-BV), and DBL1+2 (FCR3-BV) (Fig. 1), whereas in Yaoundé, only 10% of women had a strong response to only DBL5 but not to any of the other domains (Fig. 2). As expected, essentially all women in Ngali II and Yaoundé were seropositive for MSP142 and AMA-1, showing they had Ab to these antigens prior to pregnancy and maintained Ab levels to these antigens during pregnancy.
Correlation between antibody levels during pregnancy and the absence of placental malaria.To determine if Ab levels to one or more DBL domains were associated with clearing parasites from the placenta, Ab levels throughout the course of pregnancy for all women with known delivery outcomes were compared between women who were placental malaria positive (PM+) and those who were placental malaria negative (PM−) at delivery (Fig. 3). Women in the two groups were similar with respect to age and length of pregnancy (Table 3). Compared to PM+ women, PM− women in Ngali II had higher Ab levels throughout pregnancy to DBL3 (7G8-Pp, P = 0.014; FCR3-BV, P = 0.023; and FCR3-Ec, P = 0.007), DBL4 (7G8-Pp, P = 0.037; and FCR3-BV, P = 0.025), DBL5 (7G8-Pp, P = 0.037; and FCR3-BV, P = 0.048), and DBL6 (FCR3-BV, P = 0.030) (Fig. 3) but not to DBL1 (data not shown). In contrast, in Yaoundé, no differences in Ab levels were found for any of the recombinant proteins except that PM+ women had higher levels of Ab to DBL1 (3D7-Pp, P = 0.017; and FCR3-Pp, P = 0.037) (data not shown). Thus, higher levels of Ab to 4 of the 6 DBL domains but not to a single domain were associated with absence of placental malaria in pregnant women in the high- but not the low-transmission setting. These results clearly show that malaria transmission intensity affects the development of protective anti-VAR2CSA Ab.
Correlation between antibody levels and absence of placental malaria in women in Ngali II. Mean antibody levels throughout the course of pregnancy were compared between women who were placental malaria positive (PM+) at delivery (n = 14) and those who were placental malaria negative (PM−) (n = 13) using multilevel polynomial regression analysis. Significant associations and their corresponding P values are shown. Individual data points are the mean ± standard error of the mean (SEM) for a minimum of 7 values.
Description of women with and without placental malaria at deliverya
Correlation between the breadth of the antibody response and absence of placental malaria at delivery.Since VAR2CSA allelic polymorphism exists between parasite isolates, effective immunity may require an extensive Ab repertoire. To test the association of Ab breadth with parasite clearance, we evaluated whether having Ab to more DBL domains or allelic variants, i.e., the breadth of the response, was associated with the absence of placental malaria at delivery (Fig. 4). In the high-transmission setting, longitudinal analysis showed that PM− women had Ab that recognized more DBL domains (P = 0.003) and variants (P = 0.033) throughout the course of pregnancy than PM+ women. That is, at 3 to 4 MoP, women in the PM+ group recognized fewer domains (PM+, 2.9 ± 0.5, versus PM−, 4.2 ± 0.2; P = 0.022) and variants (PM+, 6.2 ± 1.2, versus PM−, 7.9 ± 0.5), but the breath of the response increased significantly during pregnancy (P < 0.0001) (Fig. 4a and b). By the end of pregnancy, the breadth of the response was similar between PM+ and PM− women. Different results were observed in Yaoundé. As predicted, the breadth of the Ab response was less than in Ngali II (Fig. 4). Furthermore, the breadth of the response throughout pregnancy was higher in PM+ women than PM− women (DBL domains, P = 0.025), and the breadth of the response did not expand during pregnancy. These results suggest that the number of P. falciparum infections has a direct effect on the development of Ab breadth to VAR2CSA.
Breadth of IgG response to the 5 DBL domains and 13 variants stratified by placental malaria status. At each month of pregnancy, the maximal number of DBL domains and strain variants each woman had Ab to was determined. The breadth of the response was compared between women who were placental malaria negative (PM−) and placental malaria positive (PM+). Results are presented as mean ± SEM. Maximum number of domains is 5; maximum number of variants is 13. P values are based on multilevel polynomial regression analysis.
Influence of gravidity on antibody levels and breadth of the response.Ab levels to the four VAR2CSA domains associated with the absence of placental malaria (DBL3, -4, -5, and -6) were compared between primigravid and multigravid women (Fig. 5). In Ngali II, the analysis could only be made for 3 groups, since there were no PM− primigravidae (Table 3). As expected, mean Ab levels were lower at the end of the first trimester in PM+ primigravidae than PM+ multigravidae (Fig. 5); however, between the second and third trimesters, there were no significant differences in Ab levels over time for each of the four domains between PM+ primigravid and PM+ multigravid women (all P > 0.05), and by the end of pregnancy, both groups of women had similar Ab levels (Fig. 5). The small sample size prevents direct comparison between multigravidae. Overall, these data show that Ab levels were lower in PM+ than in PM− women (Fig. 3) and that Ab levels are similar in women of different gravidities who had PM. As expected, no relevant differences were found for women in Yaoundé (data not shown).
Comparison of Ab levels in primigravidae and multigravidae living in Ngali II. Antibodies to DBL3, DBL4, DBL5, and DBL6 were compared between primigravidae (PG) and multigravidae (MG) who were placental malaria positive (PM+) and placental malaria negative (PM−). All PM− women in Ngali II were MG (n = 13). Similar Ab levels were found between PM+ PG (n = 7) and MG (n = 7). The same trend was observed with the other strain variants of DBL3, DBL4, and DBL5. The arrow on the y axis and associated horizontal dotted line indicate the cutoff value for seropositivity (mean ± 2 SD of the results for male controls).
The influence of gravidity on the breadth of the Ab response was also evaluated (Fig. 6). In Ngali II, both PM+ primigravid and PM+ multigravid women recognized similar numbers of DBL domains (Fig. 6a) and variants (Fig. 6b) throughout pregnancy. While there is no significant difference between the Ab breadth of PM+ primigravid and PM+ multigravid women, the response of PM+ women as a group was significantly lower than the response of PM− women, as shown in Fig. 4a and b. Thus, the restricted response of PM+ women seen in the results in Fig. 4 was observed in both primi- and multigravidae. On the other hand, in Yaoundé, both PM+ primigravidae and multigravidae recognized more domains than PM− primigravidae and multigravidae (Fig. 6).
Influence of gravidity on the breadth of IgG responses. The breadth of Ab responses to VAR2CSA DBL domains and variants during pregnancy were compared between primigravidae (PG) and multigravidae (MG) who were placental malaria negative (PM−) and placental malaria positive (PM+). Results are represented as mean ± SEM.
Again, the results show that the breadth of the response was associated with the presence/absence of placental malaria and not gravidity per se, and greater Ab breadth and magnitude were associated with lack of PM at delivery. Clearly, the level of malaria transmission influenced the breadth of the response in that, by the end of pregnancy, women in Yaoundé still had lower Ab breadth (domains and variants, P < 0.0001) than women in Ngali II, regardless of gravidity and placental malaria status (Fig. 6). As a proxy for gravidity, we further analyzed how mean Ab levels to VAR2CSA domains as well as the Ab breadth of each woman might be influenced by age. We found no correlation between mean Ab levels and Ab breadth and age (data not shown).
Although it was not possible to directly compare responses between PM+ and PM− primigravidae, it is clear that in the high-transmission setting, PM+ women had lower levels of Ab to VAR2CSA domains and a more restricted Ab repertoire throughout pregnancy than PM− women, which supports the importance of anti-VAR2CSA Ab in clearing parasites from the placenta.
DISCUSSION
Protection from placental malaria is acquired over successive pregnancies, with the prevalence of placental malaria being lower in multigravid than primigravid women. Although Ab to placental-type parasites have been associated with improved pregnancy outcomes (20, 45), it remains unclear to which part(s) of the VAR2CSA molecule protective Ab bind and whether protection from placental malaria requires a large repertoire of Ab to multiple domains and/or different allelic variants of VAR2CSA. Data from this study were obtained using samples collected prior to implementation of intermittent preventive treatment (IPTp). Today, IPTp is the standard of care in many African countries; therefore, studies similar to the one reported herein would be difficult to repeat. IPTp reduces the number of infections during pregnancy and the Ab response to placental isolates (3). Although the number of women followed was relatively small (n = 89 women), the sample set included parasitological data and serum samples from 6 ± 1.2 time points per women, thereby allowing consistent and reliable analyses of the natural evolution of Ab to VAR2CSA. Accordingly, this study provides much needed information about the natural acquisition of immunity to placental malaria that would be difficult to obtain today.
Exposure to P. falciparum differed between the two sites. Women living in Ngali II are bitten, on average, every other night by a P. falciparum-infected mosquito and thus were exposed to placental isolates very early in pregnancy. Epidemiological studies show that multigravid women in Ngali II are slide positive fewer times during pregnancy and have fewer and lower-density placental infections than primigravidae (33), indicating that multigravidae have acquired significant protective immunity. The situation is different in the lower-transmission area in the city of Yaoundé, where women are exposed to ∼1 infectious mosquito bite per month and may not become infected until the middle to end of the second trimester. Thus, it was expected that the kinetics of Ab response to VAR2CSA would differ in these two populations due to the frequency and timing of placental infections.
In both transmission settings, DBL5 was the most recognized domain, followed by DBL3, including allelic domains expressed in different vector systems (Table 1 and Fig. 1 and 2). This observation is consistent with other studies showing that most monoclonal Ab derived from pregnant women target the DBL5 or DBL3 domains (12). Since women in both sites had significant gains in their VAR2CSA Ab repertoires during pregnancy (Fig. 4) and more women produced higher levels of Ab to multiple VAR2CSA variants over time (Fig. 1 and 2; also see Fig. S1 and S2 in the supplemental material), the natural acquisition of Ab to VAR2CSA is broad spectrum and follows a domain hierarchy.
This is the first study to report that the magnitude and breadth of anti-VAR2CSA Ab levels are affected by transmission intensity and that absence of placental malaria correlates with Ab breadth. In the high-transmission setting, primigravidae in Ngali II produced a relatively broad VAR2CSA Ab response during the course of a single pregnancy. Many of them acquired Ab to multiple VAR2CSA domains, and by the end of pregnancy, there was little difference in VAR2CSA Ab levels between PM+ primigravidae and PM+ multigravidae (Fig. 5 and 6). Although the response was rapid, it may not have reached the necessary threshold soon enough to allow primigravidae to completely clear parasites from the placenta prior to delivery. Some multigravidae in Ngali II also had placental malaria. Their Ab response appears to be less mature than that of PM− multigravid women (Fig. 5 and 6), suggesting that some multigravidae are still acquiring pregnancy-associated immunity. A strong immune response may be required to eliminate placental malaria prior to delivery when women are bitten every other night by infected mosquitoes. In comparison, anti-VAR2CSA immunity developed slowly in the lower-transmission site of Yaoundé. The finding that 40% of women in Yaoundé who became infected did not produce Ab to any of the DBL domains is of interest. Placental-malaria-positive women in the city had higher Ab levels and broader repertoires than PM− women. As a result of infection, PM+ women were exposed to additional new variants and produced Ab against them, resulting in a larger Ab repertoire. Interestingly, women in the city who were PM− at delivery had the lowest Ab levels and breadth. It is possible that these women had strong antimalarial immunity prior to pregnancy, quickly cleared their new infections, and were not exposed to VAR2CSA, i.e., they cleared their infections prior to parasites switching to var2csa expression. Further studies are needed to evaluate the role of non-VAR2CSA Ab in protection from PM in very-low-transmission areas. Overall, the results from the study show that having Ab early in pregnancy is necessary to provide sufficient time for parasite clearance prior to delivery.
Consistent with these observations, we recently found that parasites containing multiple var2csa gene copies were more prevalent in multigravid woman in Ngali II (47) who had high levels of anti-VAR2CSA Ab. Taken together, these data suggest that there is an interaction between malaria endemicity and acquisition of Ab to VAR2CSA and that parasites carrying more than one var2csa gene copy may possess a selective advantage as VAR2CSA Ab breadth expands.
Given that the introduction of IPTp and bed nets have reduced malaria intensity in many African settings from high to moderate or from moderate to low, the results from women living in the low-transmission site are of high importance. They show that Ab responses to DBL domains were delayed and less strong compared to those in village women (Ngali II). Although these results were less informative about the hierarchy of the response, they provided unique data on the natural development of anti-VAR2CSA immunity in women living in urban settings. These data should be applicable to many similar socioeconomic settings throughout Africa.
In conclusion, this study shows that women make Ab to all DBL domains of VAR2CSA, beginning as early as the second trimester of pregnancy for some domains. Antibody levels and Ab breadth were influenced by malaria transmission intensity and expanded over multiple pregnancies. Finally, high levels of Ab to multiple VAR2CSA domains and allelic variants early in pregnancy were associated with the absence of placental infections at delivery.
ACKNOWLEDGMENTS
We acknowledge the support of the Malaria Research Team at the Biotechnology Center, University of Yaoundé I, Cameroon, for their outstanding work. A special thank you goes to the women and their families who participated in the studies. We also thank Ian Pagano, Cancer Research Center, University of Hawaii, for statistical support and Kazutoyo Miura for helpful suggestions in manuscript preparation.
The work was supported by grants UO1AI43888 and RO1AI071160 (D.W.T. and R.G.F.L.) and U19AI089688 (J.D.S.) from NIAID, NIH, by the intramural research program of the Division of Intramural Research, NIAID, NIH (K.S. and C.A.L.), and by FP7/2007-2013 under grant agreement no. 200889 (STOPPAM) (A.S.).
FOOTNOTES
- Received 25 January 2012.
- Accepted 30 January 2012.
- Accepted manuscript posted online 13 February 2012.
Supplemental material for this article may be found at http://dx.doi.org/10.1128/IAI.00071-12.
- Copyright © 2012, American Society for Microbiology. All Rights Reserved.
