Background
Malaria along with HIV/AIDS and Tuberculosis is one of the major causes of mortality and morbidity in tropical areas and consequently the primary obstacle to social and economic improvement for developing countries in these regions [
1]. Despite recent efforts to control infections based on the use of insecticide-treated bed nets alone or combined with indoor residual spraying [
2] and artemisinin-combined treatment [
3], almost 300 million cases and 800,000 deaths are annually registered [
1,
3]. Most of these deaths occur in Sub-Saharan Africa in children under 5 years and pregnant women as a consequence of
P. falciparum infection [
3]. Given the looming resistance of the parasite against the derivatives of artemisinin, concentrated and continuous efforts are necessary to contain the disease. These include the facilitated access to effective treatment, the introduction of novel drugs and also the development of efficient vaccines.
Important success has been achieved in the development of vaccines based on pre-erythrocytic targets using the circumsporozoite protein [
4] or liver stage targets employing knockout parasite lines [
5]. The finding that the passive transfer of immunoglobulins led to the suppression of parasite multiplication and temporary cure [
6] supports the view that relevant targets are also found on blood stage parasites. Proteins on the infected red blood cell which are mostly variant and with relatively low immunogenicity can be targets of antibodies which recruit the infected cells for phagocytosis. Proteins on the merozoite actively participate in the successive coordinated events that culminate in the erythrocyte invasion [
7]. Several merozoite surface proteins and others secreted by the apical organelles interact with erythrocyte ligands. This interaction favors adhesion, apical reorientation and creation of a moving junction that allows the invagination of this parasite form into the erythrocyte, leading to the formation of a parasitophorous vesicle membrane (PVM) in which the parasite resides after invasion (reviewed in [
8]). Many of the merozoite antigens act as targets of the natural antibody immune response [
9,
10] and several of them have been implicated in the development of clinical protection [
11‐
17] and therefore are present in the anti-blood stage vaccine formulations which are currently being tested [
18‐
21].
In the Brazilian Amazon, the transmission and incidence of
P. falciparum malaria is quite different from Africa [
22]. Many localities which experienced periods of high malaria transmission in the past [
23], presently show low and seasonal
P. falciparum transmission [
24]. There is also a significant genetic structuring in parasites from the Western Amazon suggesting a paucity of different circulating strains [
25,
26]. Coincidently, the repertoire of variant genes of
P. falciparum is restricted and redundant [
27]. This special situation of exposure to restricted numbers of antigens in addition to the observation of uncomplicated infections probably explains previous data regarding a high incidence of asymptomatic carriers and/or persons to which the majority of new infections are without symptoms [
28]. Regardless of the apparently functional immune protection of these asymptomatic carriers, they do represent a persistent source of
P. falciparum infections [
29] in Amazonian settings.
On the basis of this epidemiologic background characterized as a high incidence of asymptomatic infections with the occurrence of sporadic symptomatic cases in the same population in the Amazon, we set out to analyze which parameter of the humoral immune response against merozoite antigens is decisive for the observed outcome during a P. falciparum malaria infection. To do this, we focused on the recognition of a number of antigens involved in the erythrocyte invasion process, namely proteins of the MSP and EBL families and AMA1. In order to test the response against relevant versions of target proteins, we analyzed the circulating alleles of merozoite genes from field isolates present in the blood of sympatric symptomatic and asymptomatic carriers and produced parts of them as recombinant antigens. Then, the humoral immune response against these antigens was measured by ELISA and correlated to disease outcome and epidemiological parameters.
Methods
Study site
The study was conducted in a riverside area of Porto Velho, the capital of Rondônia state, in the Western Brazilian Amazon. Four localities on the riverbanks of the Madeira River were chosen, Vila Candelaria, Bate-Estaca, Santo Antonio and Engenho Velho. The first three are located one after the other on the right bank of the Madeira River and the last, on the left bank. As in other endemic settings in Brazil, the majority of malaria cases in these areas are caused by
Plasmodium vivax[
22] and the overall endemicity for Malaria is relatively low
. The annual distribution of malaria cases in these areas paralleled the seasonal rain distribution with a peak of incidence in the beginning (October-November) and other after the peak of rainy season (February-March) [
24]. The principal vector in Brazilian malaria endemic areas is
Anopheles darlingi[
22]
.
Study design
During the low malaria transmission season (August-September), a cross sectional study was conducted in the four localities by the IPEPATRO field team and FUNASA technicians. Children older than one year to seniors, both male and female, were invited to participate and more than 50% of residents from each locality agreed to participate. After each participant (or the legal guardian for children) signed a specific consent form, a 5 ml blood sample was collected by peripheral venipuncture, independent of the presence of malaria symptoms. Individuals with any malaria symptoms (fever >37.8°C, prostration, headache, chills, tremors, myalgia, nausea and vomiting) and P. falciparum positive thick blood films were immediately treated after blood was taken and were included in the symptomatic group. The P. vivax symptomatic patients were immediately treated and were not included in the study. The participants without symptoms but positive by PCR for P. falciparum were followed up for 1 month by a medical team to verify the development of symptoms. When symptoms appeared, a thick blood film was performed and individuals which were positive for falciparum malaria were treated immediately and included in the symptomatic group. Persons who did not develop symptoms even with P. falciparum positive PCR were classified asymptomatic and treated after the follow-up. Persons with positive PCR for P. vivax without symptoms were treated immediately after the parasite diagnosis and not included in the study. Only one asymptomatic patient presented parasites in the thick blood smear (routinely, 50 fields per blood smear were monitored) while the remaining asymptomatic infections were only diagnosed by PCR. Finally, 28 samples were obtained from symptomatic patients (28 erythrocyte concentrates and 27 plasma samples) and 25 from asymptomatic patients (25 erythrocyte concentrates and 24 plasma samples). In order to calculate reactivity indices, 13 plasma samples from persons living in non-endemic regions of Brazil and never exposed to malaria were included as the negative control group.
Blood sampling
The blood samples were collected in sterile tubes with EDTA, centrifuged (400 g, 5 minutes, room temperature) and fractioned into plasma and packed erythrocytes. One volume of glycerol was added to each plasma sample and these samples were maintained at −80°C. 250 μl of the packed erythrocytes were submitted to
Plasmodium genomic DNA (gDNA) purification by the Proteinase K/phenol-chloroform method [
30].
Identification of Plasmodiuminfection by PCR, locus amplification
Each gDNA extracted was submitted to nested PCR in order to verify the presence of the
Plasmodium genus by amplification of ssrRNA fragments as described previously [
30]. According to the PlasmoDB database (v5.5), primer pairs covering polymorphic regions and/or immune epitopes of MSP4, MSP5, MSP6, MSP7, MSP9, MSP10, AMA1, EBA140, EBA175 and EBA181 were designed on the basis of the 3D7 genome sequences. Primers described previously to amplify MSP1 block 2 [
31], the polymorphic regions of MSP2 [
32] and MSP3 [
11] and the M2 domain of MAEBL [
33] were used. A
BamHI or BglII restriction site was added at the 5′ extremity of each forward oligonucleotide. The sequences of the primer pairs used as well as the amplified regions are shown in Additional file
1: Table S2 and in Additional file
2: Figure S2, respectively. Field isolate gDNAs extracted from blood from symptomatic and asymptomatic infections were submitted to 35 amplification cycles of denaturation at 94°C (1 minute), annealing at temperatures specific for each pair of primers (1 minute), polymerization at 72°C (1 minute) and a final polymerization at 72°C for 10 minutes. Taq Polymerase enzyme (Fermentas) was used according to the manufacturer’s instructions. In all reactions, 3D7 gDNA was used as a positive control and an amplification solution without gDNA as negative control.
Cloning and sequence analysis
All amplified fragments from each gDNA were A/T cloned in pGEM-T easy vector (Promega) according to the manufacturer’s instructions. Ligations were transformed in
E. coli DH10B cells. Five recombinant clones were analyzed by automated DNA sequencing using the BigDye 3.1 Terminator Cycle Sequencing kit (Applied Biosystems) according to the manufacturer’s instructions and sequencing was conducted in an ABI 3100 automatic sequencer (Applied Biosystems). The identity of each sequence was confirmed by BlastN analysis. All obtained
P. falciparum sequences of each merozoite gene were loaded in
Clustal X (1.83) and aligned. An identity matrix was generated for each sequence group and sequences were clustered into groups with identities of 95% or more which were then considered identical. This number was chosen based on the assumption that differences above this cutoff would most probably generate proteins which are recognized similarly by antibodies present in human sera. One representative sequence from each group of >95% identical sequences was subcloned via BamH1/EcoR1 digestion and ligated in pGEX2T (Amersham Pharmacia). Data regarding the sequence diversity of MSP1 block 2 [
34] and MSP2 [
35] were used to define allelic groups and variants.
Protein expression
Recombinant pGEX2T plasmids were transformed into E. coli BL21 DE3 pLys Codon Plus RIL cells and grown on a shaker at 37°C in 5 ml aliquots overnight in LB supplemented with ampicillin (100 μg/μl) and chloramphenicol (34 μg/μl). 1 ml of the preinoculum was then transferred into 100 ml LB supplemented with ampicillin (100 μg/μl) and grown to optical densities (600 nm) of ≥ 0,6 and induced by the addition of IPTG to a final concentration of 0,1 mM for 3 hours at 37°C on a shaker at 200 rpm. Bacterial cultures were then centrifuged for 20 minutes at 4°C and 2300 g in a Sorvall R7 benchtop centrifuge. The bacterial pellet was dissolved in 5 ml PBS/1% Triton X-100 (pH 7.4) and Lysozyme (0,1 mg/ml, Sigma) was added. The solution was then incubated rocking for 5 minutes at room temperature and frozen for 2 hours or overnight at −20°C. Afterwards, the material was thawed and submitted to five sonication cycles (Branson sonifier, 30 seconds, force 40, on ice) intercalated by 30 seconds without sonication. 5 ml of PBS 1X/1% Triton X-100 was added and centrifuged as described above. Supernatants were then incubated with 100–200 μl of Glutathione-sepharose resin (Amersham Pharmacia) pre-washed with PBS (pH 7.4), under continuous agitation at room temperature. After one hour, the resin was washed once with PBS/1% Triton X-100 and twice with PBS. Then, 500–750 μl of elution buffer (0,1 M Tris–HCl pH 8, 0,12 M NaCl and 10 mM reduced gluthatione) were added to the resin in 1,5 ml micro tubes, and incubated at RT under continuous agitation for 1 hour, after which the supernatant was harvested by centrifugation for 1 minute, 12000 rpm. Recombinant proteins were submitted to electrophoresis in SDS-PAGE under reducing conditions. All expressed peptides were submitted to Western Blotting analyses with anti-GST mouse serum to test the correct size of the fusion protein (data not shown). The concentration of recombinant proteins was calculated by the Bradford method.
ELISA assays
Flat wells plates (medium binding) were coated overnight at 4°C with 50 μl polypeptide solution (125 ng/well) in sodium carbonate buffer (50 mM, pH 9.6). After one wash with PBS/0.05% Tween 20 (washing solution), 200 μl/well of blocking solution (PBS, 0.05% Tween 20 and 4% skimmed milk) was added for 2 hours at 37°C. Plates were washed three times with washing solution and incubated for one hour at 37°C using 1/200 diluted plasmas from asymptomatic and symptomatic patients. This incubation occurred in blocking solution with 1% skimmed milk (incubation solution). All plasmas were tested in duplicates in wells coated with fusion peptide and background binding to the GST fusion was estimated by wells which contained only rGST. After five washes, plates were incubated with 1/1500 diluted goat anti-human IgG HRP antibodies (KPL) in incubation solution for one hour at 37°C. After five washes, 50 μl/well of TMB substrate (KPL) was added and 10 minutes later the reaction was stopped with 50 μl/well of 1 M HCl. The OD450nm was read in a spectrophotometer using the OD595nm value as reference. Thirteen plasmas from persons without previous contact with Plasmodium sp. were used as negative controls. IgG subclass-specific ELISA assays for each peptide were performed as follows: Monoclonal mice antibodies against each anti-human IgG subclass (anti-IgG1 (2C11, ab1927), anti-IgG2 (3C7, ab1935), anti-IgG3 (5G12, ab1928) e anti-IgG4 (5C7, ab1930) (Abcam, Inc., Cambridge, Massachusetts, USA) were added after the primary incubation, diluted 1/3000 in incubation solution. Afterwards, goat anti-mouse IgG HRP antibodies (KPL) were added, 1/2500 in incubation solution.
Statistical analyses
Microsoft Excel 2007 spreadsheets were used to collect the raw data and perform OD cutoff calculations while statistical analyses were run in SPSS (15.0). The quantitative variables for the epidemiologic data were: age, time living in an endemic malaria area (TLEA), time living at the same actual address (TLSA), number of previous symptomatic falciparum malaria infections (NPMF), time since the last symptomatic malaria falciparum infection (TLMF), number of recombinant antigens recognized in the ELISA assays (NRAR). These variables were compared between symptomatic and asymptomatic groups by Mann-Whitney’s U test for comparisons between two independent samples. p values lower than 0.05 were considered significant.
Instead of optical densities, we used reactivity indices (RIs) to estimate the specific antibody reaction, thereby eliminating signals from residual E. coli derived components in the recombinant protein preparations. For this, the optical density (OD) values from the negative plasma control group were used to estimate the reaction cut-off for each antigen tested. The average ODfinal (ODfinal = ODantigen-ODGST) plus 2 SD (standard deviations) was taken as the cutoff value. ODantigen is the optical density that was measured using all plasmas on the GST-fused antigen while ODGST is the optical density that was measured the same plasmas on GST-coated wells. The RI for each reactive plasma and each antigen (OD > cutoff) was calculated by dividing the OD value by the cut-off reaction value. The RIs from symptomatic and asymptomatic plasmas for each tested antigen were compared by Mann-Whitney’s U test for comparisons between two independent samples. For comparisons between different alleles or variants, the RIs from each clinical group were compared by Wilcoxon signed rank test for comparisons between paired samples. p values lower than 0.05 were considered significant for the two tests. The OD values obtained in IgG subclass ELISA assays were compared between symptomatic and asymptomatic individuals by Mann-Whitney’s U test.
Cross tab analysis was used to evaluate the frequency of reactive plasmas for each antigen in groups of symptomatic and asymptomatic patients. This analysis was also applied to analyze the frequencies of different alleles of polymorphic genes present in parasites collected from these patient groups. The Chi-square (Χ2) method or, whenever necessary, Fisher’s exact test identified the differential frequencies (significant when p < 0.05).
Association between qualitative dichotomous variables “symptoms” (Yes-symptomatic or No-asymptomatic) and “antigen recognition” (Yes-recognized or No-not recognized) was also verified by cross tab analysis. The strength of association was calculated using the V Cramér coefficient that varies from 0 to 1 (0-no association to 1-strong association). Odds ratio (OR) values were obtained from tabs showing true association among variables, verified by the X2 method (p < 0.05). OR values represent the ratio of odds of non antigen-recognizing individuals to present an asymptomatic profile in relation to antigen-recognizing individuals. Values lower than 1 indicate that antigen-recognizing individuals are more likely to remain free of symptoms compared to individuals who do not react against tested antigens. Association between the asymptomatic profile and intensity of antigen recognition was verified by logistic regression. The response variable (symptoms) was categorized in 0 (yes = symptomatic) or 1 (no = asymptomatic) and the explanatory variables (RI values) were categorized in terciles. Firstly, each explanatory variable was analyzed by a likelihood ratio test in the univariate model to verify association or not with the response variable. The association force was calculated using the Wald test. The OR points to the ratio of odds (the chances in relation to the non-chances) that each explanatory variable offers for the response variable assuming the “1” value (asymptomatic profile). In the multivariate model, we then verified which explanatory variables were maintained independent in offering chances to present an asymptomatic profile, and this was done by the partial likelihood ratio test.
Discussion
Immunity to blood stage malaria is seen as a process that takes several years of continued exposure to infection and reasons for this delay to a protective response may lie in the suppression of an effective T cell response [
36,
37] and the great diversity of antigens exposed to the host immune system. There is striking evidence that the immune response mediated by antibodies against either the infected red blood cell surface or the merozoite is of prime importance in the observed partial protection of individuals [
38]. It appears that at least two types of a humoral response against blood stage malaria may lead to clinical protection. On one hand, the vaccine-induced production of function-inhibiting antibodies against otherwise rarely recognized antigens, such as the PfRh5 antigen [
39], may lead to protection. On the other hand, a significant degree of protection mediated by antibodies is induced through years of exposure to natural infection with multiple parasite lines and antigens. In our study we tried to elucidate which antigen or combination of antigens was recognized in order to provide partial protection (asymptomatic status). In contrast to the high throughput assay performed by Crompton and colleagues [
40], we analyzed antigens which are expected to contribute to immunity since they are exposed on the merozoite’s surface.
In the studied population, the individuals who suffered from symptomatic infections showed almost identical epidemiologic features when compared to the asymptomatic individuals. The variable TSLE (time since the last falciparum symptomatic episode, Tables
1,
2 and
3) pointed to the magnitude of the clinical immunity developed by asymptomatic patients. A significant number of individuals were not only asymptomatic at the onset of the study but indeed had been symptom-free for up to 30 years regardless of continued exposure to the parasite. The symptomatic group consisted of individuals with relatively mild malaria symptoms and low parasitemias and no single case of truly severe malaria was observed in the area. Severe malaria itself is a rare event in the Brazilian Amazon and the reasons for this may be rapid access to adequate individual antimalarial treatment, restricted virulence antigen repertoires in circulating strains [
27], low transmission rates with
Anopheles darlingi[
24,
41] being the principal vector, and partly due to successful efforts to control malaria in the Amazon region during the last 20 years [
22]. Given the similar history of exposure to
P. falciparum in the herein tested symptomatic and asymptomatic patients groups, it can be assumed that the former group is in the process of acquiring clinical immunity to malaria and will ultimately become part of the asymptomatics group. The difference between the clinical status of each group lies in the step at which their members are at in the process to clinical immunity. Importantly, the sample number of 51 in our experiments is low compared to studies in other endemic settings such as found in Africa or South East Asia. Therefore, it would be important to confirm our results in future studies using larger sample sizes. However, in the current situation of low transmission and decreasing incidence of
Plasmodium falciparum, it is logistically difficult to achieve this in most areas of the Amazon.
When amplifying merozoite-protein related genes, many negative results in the amplification step were obtained and these occurred mostly in samples from asymptomatic infections. There are two main possibilities which may account for this observation: Firstly, the low parasitemia in asymptomatic infections and consequently the small quantity of recovered plasmodial genomic DNA may have been insufficient to permit the amplification of determined genes. Secondly, it is possible that distinct polymorphic sequences - different from the 3D7 genome – were present at the primer binding sites in isolates especially from asymptomatic infections. However, even when analyzing the alleles of the highly polymorphic block 2 of MSP1, the quantity of different sequences found were much smaller than previously described in high transmission malaria areas in Africa [
34], but comparable to previous studies in the Amazon [
42]. The rather small number of different sequence types in the circulating field isolates supports previously published data regarding the diversity of the repertoire of
var variant gene DBL-α sequences in circulating isolates from the same area [
27].
In
P. falciparum donor samples from the same area, a differentiation of parasite haplotypes associated to asymptomatic or symptomatic donors was observed [
43]. When we clustered the obtained sequences for each analyzed gene according to the criterion “symptomatic” or “asymptomatic”, we detected a significant accumulation of isolates with MSP1, block 2 sequence type K1 in parasites from asymptomatic patients while symptomatic patients showed a higher proportion of parasites with the RO33 type sequence. The reason for this is unclear and, to our knowledge, has never been described in other reports. All other sequences from antigens included in the study showed high similarity to sequences from the 3D7 genome, reinforcing a limited diversity of relevant protein sequences in the Amazon.
We then asked if there were antigens (or groups of antigens) which were recognized differentially by asymptomatic or symptomatic carriers. One has to take into account that the response in symptomatic patients may reveal higher antibody titers against certain antigens. This may lead to the probably erroneous interpretation that the presence of antibodies is detrimental for the health condition of the respective carrier. Also, the quality of antigens used may not permit exact conclusions about the functionality of the detected antibodies. Many of the antigens contain conformational epitopes in their native full-length form and these were probably not reproduced in our recombinant antigens, likely restricting the assay to the detection of antibodies against linear epitopes. At least rMSP1
19 and perhaps rEBA175, however, contained conformational epitopes which could be recognized by antibodies. This was seen by the heat-mediated denaturation of these proteins leading to a decrease in their recognition (Additional file
2: Figure S3).
Testing of the leading erythrocyte stage vaccine candidates MSP1
19, MSP2, MSP3 and AMA1 [
20] has revealed interesting features in the development of clinical immunity to falciparum malaria in these low endemic areas. The first falciparum episodes normally are sufficient to generate IgG antibodies to MSP1
19. Intriguingly, some asymptomatic patients exhibited no antibodies to this antigen suggesting that antibodies against solely this domain are not critical for an asymptomatic outcome, regardless of its importance in the invasion process and parasite survival [
44]. Nevertheless, this antigen was largely recognized in almost all patients in both groups and the antibody response was maintained over time. The IC1 allele of MSP2 was largely recognized in both groups and the same was true for the rarely occurring FC27 allele, although to a lesser extent but still at a higher rate than may be expected from the genotyping of circulating strains. The presence of antiMSP2_FC27 either corroborates the presence of shared epitopes between polymorphic sequences or a previous, more abundant presence of the FC27 allele. A similar result was observed in the study by Jordan and colleagues regarding the occurrence of antibodies against determined allelic forms of MSP3 [
45] and this seems to be a feature for areas with low transmission conditions such as those found in the Western Amazon. It is known that MSP1 and MSP6 genotypes fluctuate in the Amazon area over time [
42,
46] and this is probably true for MSP2 genotypes. No differences regarding the intensity of the immune response of both clinical groups were observed, despite the higher RI values to MSP2 antigens found in the asymptomatic group. To our knowledge, no study aiming to associate the anti-MSP2 response and protection has previously been done in the Amazon. In a study analyzing immune responses and protection from clinical malaria in the Gambia, a combined higher immune response against MSP2 alleles and AMA1 was associated to protection [
47]. Polley and colleagues also found a strong correlation between antibodies to MSP2 and protection [
48]. IgG3 was the predominant subclass for IC1 and also for rMSP1
19 among asymptomatic plasmas and may be associated to the asymptomatic profile, interacting with phagocytes and favoring ADCC mechanisms, such as the respiratory burst by neutrophils [
49]. This is in concordance with the data from other groups in settings of higher endemicity (e.g. [
50] and papers cited therein).
The polypeptide MSP3_3D7-like, the major MSP3 allele found among the Rondonian isolates, was broadly recognized by plasmas from asymptomatic and symptomatic patients, however, with no differences that were statistically significant. In a recent study monitoring indigenous populations from the Venezuelan Amazon, smaller numbers of individuals with antibodies against MSP3 (11.2%) compared to our study (70%) were found, which is possibly due the fact that the study groups in that work contained only a few parasite carriers [
51]. Another study in the Peruvian Amazon also detected a predominance of one distinct MSP3 genotype and a similar recognition frequency of the MSP3 HB3 allelic form (very similar to the 3D7 sequence) as found herein [
45]. In that study, the responses to the N-terminal and the conserved C-terminal region were measured separately and the authors detected a higher recognition of the polymorphic N-terminal portion of MSP3 compared to the C-terminus. Our study used a peptide including the variant N- and the conserved C-terminal region of the MSP3_3D7 antigen, making us unable to dissect the response either against the polymorphic N or the conserved C-terminus as done in the study of Jordan and colleagues. Notably, the same group also observed the occurrence of MSP3 genotypes similar to those found in our study [
52].
The interaction of antiMSP3 specific immunoglobulins with phagocytes and their role in ADCI mechanisms has been described previously [
53] and the higher portion of IgG3 in asymptomatic individuals compared to symptomatic persons may underscore their importance in disease outcome. However, the antibody response to this important vaccine candidate seemed short-lived as shown by a fast decrease of the response in plasmas collected 30 or 60 days after the treatment. Perhaps for this reason, the IgG3 subclass in some asymptomatic plasma became predominant. These residual anti-MSP3 IgG3 antibodies may contribute to the maintenance of the asymptomatic profile, pointing to its importance in the acquisition of clinical immunity. In a study conducted in Senegal, the presence of IgG3 was significantly associated with protection [
11]. Importantly, a MSP3 based vaccine proved efficient in a phase 1b vaccine trial [
21].
The longer AMA1 peptide (rAMA1_B = M1 and M2 domains) showed a higher frequency of recognition than the shorter (rAMA1_A) corresponding to the M1 domain and the region between M1 and M2 domains. Analysis of AMA1 recognition by plasmas from Papua New Guinea showed that immune epitopes present in the M2 and M3 domains depend on the conformational structure of the protein for antibody interaction. The cysteine-rich M1 and M2 domains and their loop structures [
54] contain conformational epitopes and are generally not well reconstituted in a prokaryotic expression system. The relatively weak recognition of the two peptides expressed (IR < 5 for AMA1_A and IR < 10 for AMA1_B) could be a consequence of the probable linear structure of the recombinant peptides used in our trials. When monitoring antibody titers in subsequent samples after treatment, no difference in recognition by plasmas from asymptomatic and symptomatic individuals was observed. Interestingly, the number of reactive plasmas among those collected 30 and 60 days after treatment increased over time. It is known that there is an extensive polymorphism in the AMA1 peptide which is critical for immune escape [
55]. Of note, the expressed peptides also contain this region but in our isolates there was only one AMA1-allele in circulation, in contrast to what was found in other endemic regions [
56]. Another tested antigen that probably suffers from incorrect folding during prokaryotic expression is MAEBL. Although we found only two different sequences, we noticed that these showed several amino acid changes present in their M1 and M2 domains.
The immune response developed to MSP1 block 2 has been associated with protection by some groups [
57,
58], but not by others [
34,
59,
60].
Aotus and
Saimiri monkeys immunized with native or recombinant MSP1 were protected when challenged with homologous or heterologous
P. falciparum strains [
57,
61]. These results indicated that the C-terminal as well as the N-terminal regions of MSP1 can be associated to protection. This also raises the possibility of immune crossreaction among the allele families. Here, among the 24 recombinant peptides tested in ELISA assays, the RO33 type-sequence of MSP1 block 2 generated the highest RI values (RI > 100), but these were not restricted to one clinical group. Regardless of the decrease in RI values observed in plasmas from asymptomatic patients collected 30 and 60 days after treatment, the subclasses identified in plasmas collected before treatment showed a predominance of cytophilic antibodies IgG1 and IgG3 in both clinical groups, but asymptomatics exhibited higher IgG3 quantities than symptomatics. Again, it is possible that better protection is due to the interaction of IgG3 and Fc receptors on the phagocytes’ surface, triggering mechanisms of ADCI described previously for this antigen [
62]. Interestingly, an IgG4 response was observed against the RO33 type-sequence of MSP1 block 2, and this was stronger in asymptomatic than in symptomatic patients and extended to MSP6 and MSP7 which form a complex together with MSP1. When measured, the antibody response to MSP7 decreased over time, but in the plasmas collected before treatment, IgG1 and IgG4 were the predominant antigens with no differences among the clinical groups. Few studies have addressed the humoral response against MSP6 and 7 in populations, and none have been done in Brazil. The recognition frequency of MSP6 and 7 seen in our study group was significantly lower than in malaria patients tested in India [
63], perhaps due to different exposure of patients or conformational issues of the proteins used. In a recent study from the Peruvian Amazon, a comparable recognition of MSP6 (C-terminal fragment as used herein) was found [
64]. MSP6 is found in two allelic forms in
P. falciparum, and we found a predominance of the 3D7 sequence, in accordance with results by Neal and colleagues who analyzed isolates from the Iquitos region [
46].
The only antigens for which the intensity of the antibody response revealed differences in the recognition between symptomatic and asymptomatic plasmas were rMSP4, rMSP5, rMSP9 and rEBA175. The first two have a conserved sequence among different strains of
P. falciparum but unknown functions [
65]. The acid-basic repeat antigen (ABRA or MSP9) has been linked to the proteolytic processing of band 3 protein [
66], the putative erythrocyte receptor for MSP1
19[
67]. EBA175 is the most important antigen implicated in the invasion of erythrocyte by the sialic acid dependent invasion pathway [
68]. rMSP4 and rEBA175 were widely recognized by plasmas from both clinical groups but revealed a stronger response in plasmas from asymptomatic patients. Although rMSP5 was readily amplified in parasites from asymptomatic and symptomatic infections, the antibody response was restricted to the asymptomatic group. Independently from the low RI observed in the ELISA assays using rMSP5, the presence of antibodies to this antigen and its intensity of recognition predicted significantly higher chances for the absence of symptoms and the development of the asymptomatic profile, respectively. Notably, MSP5 is not essential for parasite survival
in vitro[
69]. The antibody response against MSP4 and MSP5 has never previously been tested in Amazonian settings, and the overall recognition of MSP5 was comparable to a recent study done in Papua New Guinea (PNG), although we found much less recognition in symptomatic individuals compared to the study in PNG [
70]. Perhaps for this reason, that study also did not reveal a differential intensity of recognition between asymptomatic and symptomatic individuals. A result similar to what was found for rMSP5 was observed for rEBA175 while the magnitude of antibody response to rMSP9 only predicted higher chances for the absence of symptoms. High levels of antibodies against EBA175 were also shown to be protective against symptomatic malaria in children in PNG [
71], while no comparable study has previously been done in the Amazon. Also, and to our best knowledge, no study has been conducted which measured PfMSP9-specific antibodies in natural infections. The correlation between the antibody response against it and the asymptomatic status surely makes this molecule an interesting target.
Our analysis revealed that the main potential vaccine antigens MSP119, MSP2, MSP3 and AMA-1 were recognized equally in number and intensity by both groups of patients. Despite all concerns regarding the quality of the expressed antigens and/or the restricted antigen repertoire in the Brazilian Amazon, it appears that the recognition of some of these antigens is dispensable for the development of an asymptomatic profile or is another indication that we are observing very similar groups in relation to the clinical immunity to falciparum malaria. On the other hand, what is the real importance of developing antibodies against MSP4, MSP5, MSP9 and EBA-175, which were the only antigens tested in our study recognized more by asymptomatic patients? Is the recognition of only one single antigen as effective as the recognition of all four antigens?
To answer these questions, we performed the cross tab and the logistic regression analysis. The multivariate model was employed in the logistic regression to measure the independent contribution of each antigen to the asymptomatic profile and this was achieved by the estimation of the odds for this outcome. The only antigens independently associated with the asymptomatic profile were MSP5, MSP9 and EBA-175. This means that the presence of antibodies against each of them is necessary to generate or maintain the asymptomatic profile. Due to this, a potential vaccine based on single merozoite antigen would not likely induce the desirable protective status, while a vaccine based on multiple merozoite antigens may be better. Accordingly, the multiple target approach proved more effective in
in vitro approaches to measure the quality of vaccines (e.g. [
72]). Considering that the several existent parasite invasion routes are redundant, the development of the asymptomatic condition will require the recognition of multiple antigens involved in each of these routes. Two antigens, implicated in two different invasion routes (MSP9 and EBA-175) were independently classified as important for the development of the asymptomatic status, while the third protein, MSP5, has still no assigned function and is also non-essential [
69]. It is important to note that the univariate model of logistic regression defined odds ratios associated MSP3_3D7 and MSP4 to the asymptomatic status (data not shown). However, the multivariate model which permits the visualization of the independent variables for a specific outcome did not confirm this result. Further tests, perhaps with higher sample numbers, are necessary to elucidate this issue especially when considering that MSP3 is a vaccine candidate already used in clinical field trials [
21].
Authors’ contributions
MMM and GW conceived the study, MMM, WLF and GW performed the experiments, MMM, RCDM, THK and LHPdS collected samples, recorded and analyzed patient information and provided infrastructure for the processing of samples, MMM and GW wrote the manuscript. All authors read and approved the final manuscript.