Natural immune response to Plasmodium vivax alpha-helical coiled coil protein motifs and its association with the risk of P. vivax malaria

Protein α-helical coiled coil structures are known to induce antibodies able to block critical functions in different pathogens. In a previous study, a total of 50 proteins of Plasmodium vivax erythrocytic asexual stages containing α-helical coiled coil structural motifs were identified in silico, and the corresponding peptides were chemically synthesized. A total of 43 peptides were recognized by naturally acquired antibodies in plasma samples from both Papua New Guinea (PNG) and Colombian adult donors. In this study, the association between IgG antibodies to these peptides and clinical immunity was further explored by measuring total IgG antibody levels to 24 peptides in baseline samples from a longitudinal study of children aged 1–3 years (n = 164) followed for 16 months. Samples were reactive to all peptides tested. Eight peptides were recognized by >50% of individuals, whereas only one peptide had < 20% reactivity. Children infected at baseline were seropositive to 23/24 peptides. No significant association was observed between antibody titers and age or molecular force of infection, suggesting that antibody levels had already reached an equilibrium. There was a strong association between antibody levels to all peptides and protection against P. vivax clinical episodes during the 16 months follow-up. These results suggest that the selected coiled coil antigens might be good markers of both exposure and acquired immunity to P. vivax malaria, and further preclinical investigation should be performed to determine their potential as P. vivax vaccine antigens.


Introduction
Plasmodium vivax is the second most important Plasmodium species in terms of epidemiological significance, with an estimate of 13.8 million malaria cases globally in 2015, and about half of the total number of malaria cases occurring outside Africa [1]. Although this parasite has been classically considered benign, several features make it difficult to control and eliminate. First, severe and lethal P. vivax malaria cases have been reported [1][2][3][4]; second, chloroquineresistant strains have recently emerged with at least one case confirmed in 10 countries [1]; third, this parasite species produces hypnozoite-forms that upon periodic reactivation induce clinical relapses [5,6], even in individuals who have left endemic regions; and fourth, gametocytes emerge early during the erythrocytic cycle possibly increasing its transmissibility [7,8]. Because of the difficulty in controlling P. vivax, in areas where the two species coexist its incidence appears to decrease more slowly than that of Plasmodium falciparum [1].
Due to limitations associated with classical malaria control measures, vaccination against malaria is currently considered a potentially valuable cost-effective complement for malaria control activities that would significantly contribute to its elimination [9]. During the last 2-3 decades, significant efforts have been invested on developing P. falciparum [10], and more recently P. vivax vaccines [11]. However, discovery of new potential vaccine candidates is required. The use of bioinformatics tools has allowed to explore the malaria genome/proteome databases, and to identify parasite proteins containing specific domains with functional importance for the parasite that could be immunologically targeted and therefore represent novel candidate antigens for vaccine development.
Protein α-helical coiled coils are stable structures capable of eliciting antibodies able to block functional domains in different microorganisms [12][13][14]. These motifs have been investigated in influenza virus [13,15], HIV-1 [12], coronaviruses [14] and malaria parasites [16]. In the case of P. falciparum, 170 α-helical coiled coil motifs have been identified in silico, from proteins predicted to be in different cellular locations such as the cytoplasm, the nucleus, the mitochondria, and the peroxysomes and in addition, some of them have trans-membrane segments. Therefore, synthetic peptides containing these motifs were synthesized, and tested for their reactivity in serum obtained from adult donors from Burkina Faso, Tanzania and Colombia [17]. The most recognized antigens were selected, and specific human IgG antibodies were affinity purified and tested in vitro using antibody-dependent inhibition (ADCI) assays, showing that several of them were active in inhibiting in vitro parasite growth [17,18]. Association of antibody responses with protection against infection was also observed [18,19]. In the case of P. vivax, 43 coiled coils segments with extensive homology to the P. falciparum counterparts were identified in silico, and the synthesized segments were tested for their reactivity with serum of individuals from malaria endemic areas [20]. However, due to the lack of P. vivax in vitro cultures, the functional activity of antibodies elicited to these antigens has not yet been studied in this parasite species. Similarly, there are no reports on the relationship between antibody responses to P. vivax coiled coil antigens and risk of infection or clinical malaria.
In order to explore the association between total IgG antibodies to P. vivax coiled coil segments (selected based on their antigenicity) and risk of disease, baseline plasma samples from a cohort of children from Papua New Guinea (PNG) [21] were tested using a multiplexed bead array assay. In this cohort, 264 children aged 1-3 years were enrolled and actively followed for up to 16 months to identify factors associated with either risk of or protection from infection and disease, and the molecular force of blood-stage infection ( mol FOB) at the individual level was also assessed [22,23]. Exposure to P. vivax infections varied greatly (between 0 and 38 clones acquired over the entire study period), and children were found to acquire an average of 15 new P. vivax blood-stage clones/child/ per year-at-risk [22]  resolution genotyping parasites. Despite the high endemicity, the incidence of clinical P. vivax malaria decreased from approximately 3.5 episodes to 1.5 episodes per year, suggesting acquisition of clinical immunity at young age [21]. In the present study, we identified for the first time a number of P. vivax coiled coil segments strongly associated with protection against clinical disease. Future studies will help to assess the value of the coiled coil antigens identified here as markers of acquired immunity or vaccine candidates.

Study description
Blood samples were collected in 2006, in a rural area near Maprik, East Sepik Province of PNG [21], including children aged between 1-3 years at enrollment. Children were actively followed-up every 2 weeks, with clinical examination for malaria signs and symptoms, and blood-samples collected every 8 weeks for a period of 16 months. In addition, samples were collected through passive case detection at health centers. Malaria cases were confirmed by rapid diagnostic test (RDT) and all samples screened for P. vivax, P. falciparum, P. malariae and P. ovale by semi-quantitative post-PCR ligase detection reaction-fluorescent microsphere assay (LDR-FMA). All P. vivax positive samples were genotyped using the markers msp1F3 and MS16 [22] and all P. falciparum positive samples using the marker msp2 [24]. A total of 164 plasma samples collected at baseline, selected from children who completed follow-up were used to test antibody reactivity against 24 P. vivax coiled coil peptides. P. vivax clinical malaria episodes were defined as the presence of fever (i.e. axillary temperature >37.5˚C) plus parasitemia (500 parasites/μL) [21].

Ethics statement
Written informed consent was obtained from all parents or guardians prior to recruitment of each child.

Coupling of polypeptides to fluorescent beads
Bio-Plex carboxylated beads (Bio-Rad) were covalently coated with the different coiled coil polypeptides following the manufacturer's instructions (Bio-Plex Amine Coupling Kit). Briefly, microspheres (1.25 × 10 6 beads/mL) were activated with a mixture of N-Hydroxysulfosuccinimide (5mg/mL) and N-(3-Dimethylaminopropyl)-N 0 -ethylcarbodiimide hydrochloride (5mg/mL) in H 2 O for 20 min at room temperature (RT) in the dark. Microspheres were then washed twice in 250 μL PBS, pH7.4, re-suspended in 500 μL of PBS, pH 7.4, and 1 μg of each polypeptide was added. Microspheres were incubated at 4˚C overnight and re-suspended in 500 μL of blocking buffer (PBS + 1% BSA). After further incubation under rotation at RT for 30 minutes, beads were washed with PBS and stored at 4˚C protected from light until use.

Antibody assays
Coupled beads were used to analyze plasma reactivity as described previously [26]. Briefly, experimental plasma samples and controls were vortexed, centrifuged at maximum speed for 5-10 minutes in a microcentrifuge (Eppendorf), and diluted 1:100 in PBS plus tween 0.1% (PBT), 50 μL were then added to 50μL of previously mixed beads, and incubated at room temperature and protected from light for 30 min. After washing 3 times with 100 μL of PBT, aliquots of 100 μL of R-phycoerythrin conjugated anti-human IgG produced in Donkey (Jackson Immuno Research) diluted 1:100 were added and incubated for 15 min. Beads were washed and re-suspended in 125 μL of PBT and read on a Bio-Plex 200. Each plate also contained a blank (wells containing only beads and PBT). Negative control consisted on a pool of plasma samples from individuals never exposed to malaria diluted 1:100, and positive control consisted on serial dilutions (1:50-1:25,600) of pooled serum (n = 20) from PNG immune adults.

Statistical analysis
Luminex median fluorescence intensity (MFI) values were converted into arbitrary antibody units based on the parameters estimated from a standard curve made with dilutions of the highly-immune PNG positive control pool, as previously described [27]. Antibody units ranged from 1.95x10-5 (i.e., equivalent to 1:51,200 dilution of the immune pool) to 0.02 (1:50). A cut-off for positivity was calculated as three standard deviations above the dilution value of negative controls for each antigen. Individuals were considered sero-positive if the dilutions of the experimental plasma samples were higher than the cut-off of negative controls cut-off. All samples were tested in duplicates in two independent experiments. Differences in the prevalence of IgG antibodies with age, infection status and exposure were assessed using Fisher's exact test. The Spearman rank correlation coefficient was used to investigate the relationship among antibody responses to the coiled coil antigens. A heat map was created using the corrplot function in R. The association between antibody levels to P. vivax coiled coil antigens and protection against clinical malaria (defined as fever and parasite density >500/μL) [21], as well as the association between number of recognized antigens and the risk of P. vivax clinical episodes was assessed using negative binomial GEE models with exchangeable correlation structure and semi-robust variance estimator [27,28]. IgG levels were classified into terciles and analyses done by comparing children with low versus medium and high antibody levels. All GEE models were adjusted for seasonal trends, village of residency, age and individual differences in exposure ( mol FOB). Multivariate analysis with all antigens was also performed using a backward stepwise regression for the selection of the final model. Analyses were performed using STATA version 12 (StataCorp) or R version 3.2.1 [29].

Prevalence of IgG antibodies and their association with age, infection status and village of residency
As expected based on the selection criteria for these 24 peptides, plasma samples collected at enrollment were reactive to all 24 peptides, with variable frequencies (  (Fig 1). All plasma samples (n = 19) that reacted with all antigens were infected at sampling time, although there were a few samples (8/36) infected at sampling that did not react with any of the antigens. There were some differences between Ilahita and Sunuhu villages regarding malaria prevalence and age at enrollment [21]. The percentage of infected children at enrollment was higher in Sunuhu than in Ilahita (33/62; 53% and 39/102; 38% respectively, P = 0.033). For this reason, reactivity of total IgG from children in these two villages were compared. Children from Sunuhu were more likely to have IgG antibodies reactive to 22 peptides, although significant differences were observed only for seven of them (Pv12, Pv63, Pv82.03, Pv95, Pv106, Pv121 and Pv145) (S1 Table). Higher positivity was observed in two hamlets (Ilahita 5 and Sunuhu 1) for 23 of the 24 antigens (S2 Table).

Association between antibody titer and prospective risk of P. vivax malaria
During the 16 months of follow-up, children experienced an incidence rate of 1.28 (95% CI 1.08-1.52) clinical malaria episodes. An adjusted GEE model was applied to test whether responses to the coiled coil antigens were associated with risk of clinical disease. Significantly lower risk of clinical P. vivax malaria was associated with high levels of IgG for all coiled coil antigens (Table 4) (adjusted incidence rate ratio (aIRR) ranging from 0.351 to 0.598). When adjusting for confounders such as age, village of residence, seasonality and individual differences in exposure, IgG levels to each of coiled coil antigens remained associated with protection with aIRR ranging from 0.314 to 0.548. Medium and high levels of IgG to 16 coiled coil fragments were associated with protection. For Pv5, Pv27, Pv42, Pv43, Pv82.02, Pv92 and Pv96.01, only high (but nor medium) IgG levels were significantly associated with protection (Fig 3; S5 Table). In general, there was a strong association between increasing antibody levels and protection for all coiled coil fragments except Pv45 and Pv90. In addition, the association between the number of recognized antigens and the risk of P. vivax clinical episodes was assessed, identifying that for each additional recognized antigen there was a reduction in P. Responders defined as individuals whose plasma dilution was above the cut-off for positivity for a given antigen.
Bold peptides correspond to those reacting with >50% of the tested sera IgG responses to any P. vivax coiled coil antigens were not associated with protection against P. falciparum clinical episodes, suggesting that these associations observed are speciesspecific (S6 Table).

Discussion
Repeated exposure to malaria infection results in development of clinical immunity, which is a multifactorial process where antibodies play a major role [30]. Both polyclonal and monoclonal antibodies have shown the potential to prevent malaria infections or to completely clear parasitemia [31,32]. Plasmodium coiled coil domains have been previously been proposed as potential targets for human malaria vaccine development, since such domains fold into stable structures that are capable of eliciting antibodies reactive against parasite functional epitopes, and are in general non-or little polymorphic [17,33].
Down selection of P. vivax antigens could represent a challenge due to the technical difficulties and limitations presented by the lack of an in vitro culture system for this parasite species. Previously, naturally acquired antibodies to the P. vivax merozoite surface protein 1 (Pv-MSP1) [34] and binding-inhibitory antibodies response to P. vivax Duffy binding protein (PvDBP) [35] have been associated with reduced risk of P. vivax clinical manifestations in individuals from the Brazilian Amazon region. In addition, the contribution of antibody titers to the acquisition of protection against clinical disease has been assessed in a cohort of children from PNG. In this population, immunity against clinical P. vivax malaria was acquired at a young age with a 3-fold decrease in episodes between the ages of 1 and 4 years [21,36,37]. In previous studies using this cohort, it have been demonstrated a strong association between specific IgG to merozoite surface proteins PvMSP3a, PvMSP9, [28], three novel merozoite proteins (PVX_081550, P12 and P41, [27]) as well as the reticulocyte specific binder RBP2b and RBP1a and clinical protection [38]. Here, by using the same PNG cohort we demonstrated that P. vivax coiled coil antigens are also targeted by natural acquired immunity and reactive with plasma from all Plasmodium species including both P. vivax and P. falciparum exposed and infected children. This antibody reactivity observed at recruitment time was associated with acquisition of protective immunity against only P. vivax clinical disease. It was found that children with high antibody levels to all the coiled coil antigens had 2-3 fold lower risk of clinical P. vivax malaria during follow-up. In addition, a positive association between the number of recognized antigens and protection against P. vivax clinical episodes was also found. This corroborates the role of acquired immunity as explanation for the decrease in incidence of P. vivax clinical disease with an increase in age, despite constant exposure [21,36,37]. Multivariate analysis revealed antibody levels to Pv95 as the only antigen significantly associated with a reduced risk of clinical malaria, which may indicate that this fragment is particularly targeted by natural immunity or a good marker of immunity. Given the high correlation between antibody responses to the coiled coil fragments, it is difficult to differentiate co-acquisition of antibodies from cross reactivity in multivariate analyses. This correlation is observed even though most of the antigens are derived from completely different proteins with limited sequence homology among the peptides,  [20]. Although cross-reactivity is likely, 24 peptides were tested using 164 individuals samples and for several individual donors a unique set of peptides was recognized (with the only exceptions for those that recognized none or most of the peptides) (S1 Fig). In addition, by using affinity-purified antibodies no cross-reactivity was observed between the P. falciparum protein sequences orthologous to P. vivax [17]. Therefore, further research in different populations exposed to P. vivax is however warranted to confirm this association.
Serum samples from PNG were reactive to all P. vivax coiled coil peptides tested, and in most cases, prevalence was comparable to that previously observed using adult's sera from a different PNG region [20]. Moreover, about 25% of children were infected at baseline, and their plasma showed reactivity to most of the antigens. These data suggest that some of these antigens could be markers of recent exposure.
Increased reactivity to P. vivax coiled coil peptides was observed in children living in villages where exposure to P. falciparum is higher (Ilahita 5 and Sunuhu 1) [21]. This observation could be explained by the high homology observed between the P. falciparum and P. vivax coiled coil antigens [20]. Indeed, cross-reactivity to P. vivax peptides was observed in sera from African donors (Supplementary results, S7 Table). However, in the PNG cohort there was no association between antibody response to P. vivax antigens and protection from clinical P. falciparum malaria. Even though the numbers of clinical episodes with P. falciparum and P. vivax were similar, there was no evidence of acquisition of clinical immunity to P. falciparum in the age group covered by this cohort [21]. These results correlate with previous studies using this study cohort, where no significant associations was observed between antibody responses to any of the P. vivax proteins and risk of P. falciparum clinical episodes [27,28]. This could be explained by differences in the age of immunity acquisition, the decrease in risk of infection to P. falciparum is only seen in adolescence and early adulthood [39].
Data presented here, along with those previously published [20], point to the 24 selected coiled coil P. vivax antigens as good markers of acquired immunity to P. vivax and to an important potential source of malaria vaccine candidates. Further insight may be provided by studies in animal models i.e. primates to determine experimentally the protective efficacy of these antibodies against P. vivax blood infection as well as to further assess the reactivity between peptides and select the most promising candidates for vaccine development.