Factors Associated with Immunoglobulin G Subclass Polarization in Naturally Acquired Antibodies to Plasmodium falciparum Merozoite Surface Proteins: a Cross-Sectional Survey in Brazilian Amazonia

ABSTRACT We investigated immunoglobulin G (IgG) subclass antibody responses to Plasmodium falciparum merozoite surface protein 1 (MSP-1) and MSP-2 in 112 malaria-exposed subjects in Brazil. IgG3 polarization was primarily epitope driven, being little affected by cumulative or current exposure to malaria and not affected by a subject's age and Fcγ receptor IIA genotype.

The polarization of antibody responses towards immunoglobulin G1 (IgG1) and IgG3 subclasses, which bind to Fc␥ receptors (Fc␥R) on the surface of monocytes, macrophages, and neutrophils, is believed to play a key role in immunity to blood-stage Plasmodium falciparum infection (2). Cytophilic antibodies mediate parasite-killing responses such as opsonization and phagocytosis of extracellular parasites or parasitized red blood cells (8,14) and antibody-dependent cellular inhibition of intracellular parasites (2). Antibody-mediated protection is further affected by polymorphism in Fc␥RIIA, one of the three receptors for human IgG (3). The replacement of arginine (R) with histidine (H) at position 131 defines a Fc␥RIIA allotype with increased avidity for IgG2 and IgG3; since only Fc␥RIIA-H131 interacts efficiently with IgG2, this polymorphism may determine whether parasite-specific IgG2 cooperates with effector cells (1). In addition, human monocytes bearing the Fc␥RIIA-H131 allotype are more efficient in the phagocytosis of IgG3-opsonized parasitized red blood cells (29).
We studied 112 adults (78.6% males), aged 18 to 52 (mean, 33.4) years, living in an opencast gold-mining area (Garimpo Satélite) in Mato Grosso, northwestern Brazil. Subjects were mostly migrants living in areas where malaria is endemic for 18.6 years on average (range, 2 to 50 years). This area is characterized by year-round transmission of both P. falciparum and Plasmodium vivax, with an increase in incidence at the end of the rainy season (March to April); the main malaria vector is Anopheles darlingi. The inhabitants of Apiacás experienced, throughout the 1990s, an average of one symptomatic malaria attack each year, and half of the diagnosed malaria infections were due to P. falciparum (11,26). At the time of the field survey, 17% of the population (n ϭ 527) had malaria parasites, as detected by thick-smear microscopy (11). P. falciparum was detected by microscopy or species-specific, PCR-based amplification of the 18S rRNA gene (20) in 57 plasma donors (50.9%) studied here; the remaining 55 subjects were free of malaria parasites. Subjects presenting with an acute febrile disease and P. falciparum infection (median load, 3,500 parasites per l of blood; range, Ͻ10 to 71,000 parasites/l) and individuals with P. falciparum infection (median load, 1,400 parasites per l of blood; range, Ͻ10 to 13,275 parasites/l) but without malaria symptoms at the time of blood collection were analyzed. These subjects were clinically reexamined Ն72 h after the initial parasite detection, and all remained infected (as detected by thick-smear microscopy and PCR) but free of any symptoms. We found no association of frequencies or levels of antibodies with clinical status (data not shown).
Recombinant antigens for enzyme-linked immunosorbent assays (ELISA) included three versions (MAD20, 3D7, and RO33) of polymorphic block 2 of MSP-1 (4), one version (Wellcome) of the conserved C-terminal end of MSP-1 (MSP-1 19 ) (19), and six versions (25, AM89, FUP/CP, 3D7, S20, and FC27) of polymorphic blocks 2 and 3 of MSP-2 (31, 32) (Fig.  1A). Except for MSP-1 19 (expressed in Saccharomyces cerevisiae), antigens were expressed in Escherichia coli fused to Schistosoma japonicum glutathione S-transferase (GST). Antibodies were measured essentially as described previously (25,31); results for MSP-1-derived antigens were partially reported MSP-1 block 2 antigens represent the three major variants (MAD20, K1, and RO33) of this domain found in local parasites (proportions, 38, 42, and 23%, respectively [25]). MSP-2 antigens represent the central region (blocks 2 and 3) of two major allelic families: four IC1-type variants and two FC27-type variants. Three MSP-2 antigens (25, AM89, and S20) represent "local" variants commonly found in Amazonian isolates of P. falciparum (24,32), while three "foreign" variants (FUP/CP, 3D7, and FC27) are commonly found on other continents but less frequently in Brazil. (B) Proportions (%) of malaria-exposed adult subjects living in northwestern Brazil with naturally acquired antibodies of each IgG subclass, detected by ELISA, to recombinant antigens derived from P. falciparum MSP-1 and MSP-2. IgG1, IgG2, IgG3, and IgG4 subclass antibodies are represented by white, hatched, gray, and black boxes, respectively. Error bars indicate the upper limit of exact 95% confidence intervals of proportions. For all antigens but MSP-1 19 , the proportion of subjects with detectable IgG3 antibodies significantly exceeds that of subjects with IgG1 antibodies (by the MacNemar test, P was 0.24 for MSP-1 19 and P was Ͻ0.01 for all other antigens). (C) Mean levels of naturally acquired antibodies of each IgG subclass, measured as ELISA reactivity indices, to recombinant antigens derived from P. falciparum MSP-1 and MSP-2 among malaria-exposed adult subjects living in northwestern Brazil. Error bars indicate the upper limit of 95% confidence intervals of means. Note that levels of antibodies to RO33 (a highly conserved MSP-1 block 2 variant) are substantially higher than those to any other antigen tested. The IgG1/IgG3 ratio for each antigen is given in parentheses. For all antigens except MSP-1 19 , levels of IgG3 antibodies significantly exceed those of IgG1 antibodies (by the Wilcoxon test, P was 0.30 for MSP-1 19 and P was Ͻ0.005 for all other antigens).  (16); frequencies of H131 and R131 alleles were 0.69 and 0.31, respectively. As expected, antibodies to all antigens except MSP-1 19 were IgG3 biased, but in contrast to African populations (30), our subjects had antibodies of other IgG subclasses to most antigens detected (Fig. 1B and C). To test whether IgG3 polarization was correlated with cumulative exposure to malaria (measured as the time in years that each individual lived in an area where malaria is endemic), we calculated IgG1/IgG3 RI ratios for each antigen and compared IgG3 versus IgG1 RIs across terciles of cumulative exposure (2 to 10 years, 11 to 20 years, and 22 to 50 years). Levels of IgG3 antibodies to eight antigens significantly exceeded those of IgG1 in all terciles (P Ͻ 0.05, Wilcoxon's test), with similar IgG1/IgG3 ratios (ranges, 1:1.2 to 1:2.8 for MSP-2 and 1:3.5 to 1:8.9 for MSP-1 block 2) across exposure strata (data not shown). The exceptions were MSP-1 19 (no significant excess of IgG3 in any exposure stratum [IgG1/IgG3 ratios between 1:1.1 and 1:1.3]) and 3D7-type MSP-2 (IgG3 excess in only the second and third strata [IgG1/ IgG3 ratios, 1:1.9 and 1:4.0]). Accordingly, the 3D7-type MSP-2 variant, relatively rare in local parasites (24,32), was poorly recognized in this ( Fig. 1B and C) and other Amazonian populations (18,32).
This first comparison of IgG subclass responses in the same population to two P. falciparum vaccine candidate antigens known to elicit IgG3 polarization showed that antibodies to the block 2 domain of MSP-1 are much more biased towards IgG3 than those to MSP-2 ( Fig. 1B and C). Since antibodies to the nonrepetitive RO33 variant were also IgG3 biased, the short repeat sequences in MSP-1 are not required for inducing IgG3 polarization. The IgG3/IgG1 ratio for RO33 is the lowest among block 2 antigens (Fig. 1C) but still higher than the corresponding ratio found for MSP-2-derived antigens. The fact that RO33 was recognized in more subjects than other block 2 antigens (MAD20 and K1) is not surprising, since RO33-type variants are highly conserved (no polymorphism found in Brazil) but MAD20-type and K1-type variants display extensive within-family diversity (25). These results implicate MSP-1 block 2, a major target for vaccine development (6,22), as a useful model to investigate mechanisms underlying IgG subclass polarization in malaria-exposed populations.
We next used multiple linear regression models to investigate the independent contribution of a subject's age, Fc␥RIIA alleles, and cumulative exposure to malaria to IgG3 polarization. Because IgG3 antibodies are short-lived (9), the presence of current P. falciparum infection was also included as a covariate. We found little or no independent association of these parameters with levels of antibodies of any IgG subclass to most antigens tested (Table 1). Age correlated positively with levels of IgG1 (but not IgG3) antibodies to two MSP-2 variants, leading to a decreased IgG3 bias. However, since only adults were studied, age-related changes in IgG subclass antibody distribution could not be fully evaluated. Cumulative or current exposure to malaria parasites affected levels of IgG1 or IgG3 for only 3 of 10 antigens tested (Table 1). No significant association was found between Fc␥RIIA polymorphism and levels of any IgG subclass antibody for any antigen, in contrast to data recently reported for African children for IgG2 (21).
We conclude that the IgG subclass distribution of naturally acquired antibodies to P. falciparum merozoite surface proteins in adults exposed to low to moderate levels of malaria transmission is primarily epitope driven. IgG3 polarization was more evident for polymorphic domains of MSP-1 than those of MSP-2, being little affected by cumulative or current exposure to malaria and not affected by the subject's age and Fc␥RIIA a The independent variables used in multiple linear regression analysis of reactivity indices of each IgG subclass antibody to each antigen were as follows: age (in years; continuous variable), length of exposure to malaria (in years of residence in an area where malaria is endemic), presence of current P. falciparum infection (1 ϭ yes), and presence of the Fc␥RIIA allele (1 ϭ yes). Fc␥RIIA polymorphism was entered in separate models as genotype data (H131/H131, H131/R131, and R131/R131) with quite similar results. Only variables associated with P values of Ͻ0.05 are shown. Analyses were made with version 13.0 of the SPSS software (SPSS Inc., Chicago, IL).
genotype. These findings have clear implications for the rational design and evaluation of antimalarial vaccines that induce antibody-mediated protection (12).