Characterizing Antibody Responses to Plasmodium vivax and Plasmodium falciparum Antigens in India Using Genome-Scale Protein Microarrays

Understanding naturally acquired immune responses to Plasmodium in India is key to improving malaria surveillance and diagnostic tools. Here we describe serological profiling of immune responses at three sites in India by probing protein microarrays consisting of 515 Plasmodium vivax and 500 Plasmodium falciparum proteins with 353 plasma samples. A total of 236 malaria-positive (symptomatic and asymptomatic) plasma samples and 117 malaria-negative samples were collected at three field sites in Raurkela, Nadiad, and Chennai. Indian samples showed significant seroreactivity to 265 P. vivax and 373 P. falciparum antigens, but overall seroreactivity to P. vivax antigens was lower compared to P. falciparum antigens. We identified the most immunogenic antigens of both Plasmodium species that were recognized at all three sites in India, as well as P. falciparum antigens that were associated with asymptomatic malaria. This is the first genome-scale analysis of serological responses to the two major species of malaria parasite in India. The range of immune responses characterized in different endemic settings argues for targeted surveillance approaches tailored to the diverse epidemiology of malaria across the world.


Introduction
The burden of malaria in India has halved over the last 15 years, yet India continues to account for over 70% of malaria cases in South East Asia [1]. The 'National Framework for Malaria Elimination in India 2016-2030' has two aims: eliminating malaria throughout the country by 2030 and maintaining malaria-free status in areas where transmission has been interrupted [2]. Long-lasting insecticide-treated bed nets and artemisinin combination therapy have greatly helped to reduce malaria incidence in India. However, as transmission declines, the proportion of asymptomatic and submicroscopic infections tends to rise in a population [3]; these infections can contribute to malaria transmission [4], but they remain undetected by the standard diagnostic and surveillance tools. In order to eliminate malaria, it will be critical to develop accurate and sensitive methods for diagnosis and surveillance of asymptomatic and submicroscopic malaria infections.
The human immune response to the malaria parasite Plasmodium is multi-faceted, involving both the humoral and cell-mediated response pathways. CD8+ effector T cells can kill intra-hepatocytic stages [5], while merozoites and intraerythrocytic stages are primarily controlled by antibody-mediated responses such as interference with invasion of naïve erythrocytes, increased clearance of antibody-bound erythrocytes, and antibody-dependent cellular cytotoxicity mechanisms [6,7]. The importance of antibody-based responses against Plasmodium was first demonstrated by passive transfer of antibodies from a clinically immune adult to a symptomatic child, which conferred protection from severe disease [8,9]. Antibodies are generated rapidly to several parasite antigens immediately following infection, boosted upon subsequent infections, and are able to persist for several years after parasite clearance [10,11]. Despite being exposed to multiple infections, individuals living in endemic areas do not acquire sterile immunity to malaria; instead, they develop non-sterile transmission-and agedependent protection from clinical disease, known as 'naturally acquired immunity' (NAI). Several studies have highlighted the role of antibody-based response in NAI, encompassing protection from infection (anti-parasite immunity) and severe clinical symptoms (anti-disease immunity). The acquisition of natural immunity has been extensively demonstrated for P. falciparum, and more recently for P. vivax, in regions of both high and low transmission [12,13]. Interestingly, NAI is acquired more rapidly to P. vivax infection than to P. falciparum [12], which, hypothetically, could be attributed to the differing biology of the two parasite species, such as the ability of P. vivax to maintain a dormant state within hepatocytes [10]. Additionally, there may be a differential contribution of antibody responses to natural immunity against P. vivax and P. falciparum [14][15][16][17][18][19].
Antibodies in an individual can be indicative of recent exposure to Plasmodium parasites, current infections, or past infections, and can therefore be used to identify suitable candidates for vaccine development, and to develop tools that can estimate malaria transmission levels [20] or monitor the efficacy of treatment programs [21]. Antibody-based responses can be measured using techniques that assess responses to one or very few antigens, e.g., immunofluorescent antibody test, enzyme-linked immunosorbent assay etc., or that detect antibodies to several antigens simultaneously, such as genome-scale protein microarrays [20]. Previous attempts to characterize the antibody response to Plasmodium among Indian populations have been sparse [22], with only a few studies describing the targeted response to well-established vaccine candidates [23][24][25][26]. Protein microarrays have been used to characterize antibody responses in populations from malaria endemic regions in Africa, South East Asia, and South America [27]; however, there have been no such studies in India to date. To capture the diverse eco-epidemiology in India, we probed protein microarrays comprising~500 P. vivax and~500 P. falciparum antigens, with sera from individuals at three sites in India-Raurkela in the eastern state of Odisha, Nadiad in the north-west state of Gujarat, and Chennai in the southeastern state of Tamil Nadu. Although the two major malaria parasite species co-occur at all three sites, they differ in their prevalence. Historically, P. vivax has been predominant in Chennai and Nadiad, whereas P. falciparum has been predominant in Raurkela. The objectives of our study include: 1) to provide a descriptive analysis of seroreactivity profiles of individuals against Plasmodium antigens at three epidemiologically diverse malaria endemic sites in India; 2) to examine the relationship between age and the acquisition of antibodies against Plasmodium antigens; 3) to identify the most immunogenic Plasmodium antigens based on antibody responses at three sites in India; and 4) to identify antigens recognized with greater intensity by individuals with asymptomatic malaria.

Ethical statement
Ethical approval to conduct this study was obtained from New York University Institutional Review Board (Study #i10-00173) and the Ethics Committee of the National Institute of Malaria Research, India. All project staff completed Protection of Human Research Subjects training prior to beginning the study, and clinical samples were collected after informed consent was obtained from all participants. For all adult patients, informed written consent was obtained from literate patients, and oral consent (documented by a thumb print) was obtained from illiterate patients. For child participants, assent was obtained from the participant, in addition to written or oral consent from their parent or legal guardian.

Study sites and sample collection
Samples were collected as part of epidemiology studies at three sentinel sites in India: [1] Chennai City in the state of Tamil Nadu; [2] Nadiad town and surrounding villages in the state of Gujarat; and [3] Raurkela town and surrounding villages in Odisha, during January 2013-April 2015. These field sites are part of a National Institutes of Health-funded International Centers of Excellence for Malaria Research (ICEMR).
Chennai is the largest city in the southern state of Tamil Nadu, with a population of over 7 million in 2011. Located on the eastern coast of India, it has a tropical wet and dry climate, with a rainy season focused between mid-October to mid-December. Chennai accounts for 55.6% of all malaria cases in Tamil Nadu, and had an annual parasite index (API) of 1.79 in 2013 [28]. Entomological inoculation rate (EIR) values are not available. P. vivax is the dominant species, and although transmission is perennial, malaria cases peak between July and October. Subjects were enrolled at the Besant Nagar Malaria Clinic, or in cross-sectional surveys conducted in the Besant Nagar catchment area, which is composed of middle-and upperclass urban dwellings, a few slums, and a large coastal fishing community. Nadiad, with a population of 0.2 million in 2011 is located in the Kheda district in Gujarat state. Rainfall is primarily received between June and September. Nadiad is hypo-endemic for both P. vivax and P. falciparum species, with slightly higher prevalence of P. vivax, and an average API of 2.5 (range 0.87-4.12) and EIR of 0.05-0.21 [29]. The National Institute of Malaria Research (NIMR) Malaria Clinic enrolled subjects attending Nadiad Civil Hospital. In addition, subjects were enrolled in cross-sectional surveys conducted in rural areas in the vicinity of Nadiad town.
Raurkela, with a population of over 0.5 million, is located in Sundargarh district close to the northern border of the state of Odisha, and has a tropical climate, with high temperatures and heavy rainfall between June-September and December-January. P. falciparum is the dominant species in Raurkela, and it has the highest EIR (7.3-127) and API (average = 20, range 5.1-43.5) of the three field sites in our study [29]. Subjects were enrolled at a clinic set up in a suburb of Raurkela, and in cross-sectional surveys conducted in rural areas in the vicinity of Raurkela.
Sample collection and processing at these sites has been described in detail elsewhere [29][30][31]. To summarize, individuals between 1-69 years were enrolled after informed consent, and whole blood was collected to generate a blood smear for malaria diagnosis and to measure hemoglobin levels. The remaining blood was separated into plasma and red blood cell (RBC) fractions: DNA was extracted from the RBC fractions for species-specific detection of Plasmodium [32], and plasma samples were stored at -80˚C.

Pilot study design
Plasma samples from 353 individuals that were randomly selected from clinic and cross sectional studies at our field sites were utilized for this study ( Table 1). A total of 236 individuals were diagnosed as malaria-positive and 117 individuals were diagnosed as malaria-negative by microscopy and PCR. Individuals 15 years were categorized as children, and those > 15 years were categorized as adults. Of the 236 malaria-positive individuals diagnosed by PCR, 147 individuals with documented fever (body temperature ! 37.5˚C) or a history of fever in the past 48 hours were categorized as symptomatic; 89 individuals without documented fever or history of fever in the past 48 hours were categorized as asymptomatic. Pooled plasma samples from 20 healthy semi-immune adults from Sepik, Papua New Guinea, collected in 2004 and where transmission of P. falciparum and P. vivax was equally high, were used as positive controls, and pooled plasma samples from unexposed individuals in the United States were used as negative controls.

Protein microarrays and probing
The Pf/Pv500 protein array (Antigen Discovery Inc., Irvine, CA) comprises a total of 515 P. vivax and 500 P. falciparum antigens, expressed in cell-free in vitro transcription/translation (IVTT) reactions. Accession numbers and description of the antigens are based on the P. vivax Salvador I and P. falciparum 3D7 genome annotation from the PlasmoDB database [33]. Complete array information is publicly available through the NCBI Gene Expression Omnibus database with accession number GPL18316. Sample processing and microarray probing was performed as described elsewhere [27].

Data and statistical analysis
Microarray spot intensities were quantified using the ScanArray Express software (Perkin Elmer, Waltham, MA). The non-specific background signal for each spot was calculated as the median intensity of sample-specific no-DNA IVTT control spots. For data normalization, the raw intensity values for IVTT proteins were divided by the median intensity of corresponding IVTT controls and log 2 transformed to generate median-normalized fold-over-control (FOC) values. To generate the heat map, median-subtracted intensity values were obtained by subtracting the median intensity of IVTT controls from the raw intensity of IVTT protein spots. Significance Analysis for Microarrays [34] was carried out by comparing the median-normalized intensity of antibody binding to Indian samples with unexposed controls from the United States, to determine antigens that are specifically recognized in the Indian cohort. Individual plasma samples were considered seropositive for a particular antigen if the corresponding log 2 (FOC) ! 1. The breadth of antibody response was determined as the number of antigens an individual or group of plasma samples were seropositive to, based on the above criteria. Comparison of the breadth of responses across various groups was carried out using the Kruskal Wallis test with p-values adjusted using Dunn's correction for multiple testing. Analysis of differential antibody reactivity between groups was performed by a two-sample T test assuming unequal variance and corrected for false discovery rates using the Benjamini-Hochberg method. Statistical analyses were carried out using Microsoft Excel, R v3.0.1, and Prism v7.0. All data have been made publicly available through PlasmoDB [33].

Evaluation of seroreactivity to P. falciparum and P. vivax antigens in India
We used the Pf/Pv500 protein microarray containing 515 P. vivax and 500 P. falciparum antigens to evaluate the seroreactivity of 236 malaria-positive and 117 malaria-negative individuals (353 total) from three study sites, Raurkela (n = 168), Nadiad (n = 114) and Chennai (n = 71), in India. Table 1 provides characteristics of the individuals and study sites. Comparison of antibody responses between all plasma samples identified 638 antigens recognized with significantly higher intensity in Indian samples versus unexposed U.S. controls. All subsequent analyses were carried out using this subset of 265 P. vivax and 373 P. falciparum antigens. The global profile of antibody binding to Plasmodium antigens at the three sites is shown (Fig 1). The overall seropositivity or breadth of response against P. falciparum antigens was higher compared to P. vivax antigens.

Variation in seroreactivity at three eco-epidemiologically diverse study sites in India
To analyze observed differences in seroreactivity across the three Indian sites in more detail, we compared the breadth of response to P. vivax and P. falciparum antigens in adults. Malariapositive adults from Nadiad showed significantly higher breadth of response to P. vivax antigens compared to malaria-positive adults from Raurkela (p < 0.03) and Chennai (p < 0.0001). Malaria-negative adults from Nadiad also showed higher levels (p < 0.03) of seropositivity to P. vivax than adults from Chennai (Fig 2A). In the case of P. falciparum antigens, malaria-positive adults from Chennai had significantly lower seroreactivity compared to malaria-positive adults from Raurkela (p < 0.0001) and Nadiad (p < 0.002); a similar observation was made upon comparison between malaria-negative adults from Chennai with Raurkela (p < 0.002) and Nadiad (p < 0.03; Fig 2B).

Age-dependent changes in antibody response to Plasmodium species
We compared the breadth and intensity of antibody response between malaria-positive children (age 15 years) and adults (age > 15 years) to understand age-based differences in acquired immunity (limited to Raurkela and Nadiad due to insufficient numbers of children sampled in Chennai). Adults showed higher breadth of response against P. vivax (p < 0.0001; Fig 3A) and P. falciparum (p < 0.03; Fig 3A), as well as greater intensity of antibody binding against P. vivax (p < 0.002; Fig 3B) and P. falciparum (p < 0.03; Fig 3A) when compared to children.

Highly immunogenic Plasmodium antigens in Indian samples
We identified highly immunogenic Plasmodium proteins across all study sites. For every antigen, we calculated the average reactivity and breadth of response in malaria-positive and malaria-negative samples at each site. We identified 26 Plasmodium antigens with average log 2 (FOC) ! 1 among malaria-positive samples at all sites; these antigens were also recognized by most of the malaria-negative samples at each site. This list of 14 P. vivax and 12 P. falciparum antigens represents the most immunogenic Plasmodium antigens that are indicative of current or past malaria infections (Table 2). In the case of P. vivax, merozoite surface proteins, ETRAMP, SFT2, and a number of hypothetical proteins are recognized with the greatest intensity. Among the P. falciparum antigens recognized with the greatest intensity of antibody binding are members of the PfEMP1 family, PTP5, AMA1, and HSP70.
Plasmodium antigens recognized with greater intensity by adults with asymptomatic versus symptomatic malaria Based upon the presence or absence of fever at the time of enrollment and up to 48 hours earlier, Plasmodium-positive individuals were further categorized as symptomatic or asymptomatic. We compared the breadth and intensity of antibody response to Plasmodium antigens in adults classified into these two categories. Of the 174 malaria-positive samples, 121 were diagnosed by both PCR and microscopy, and 50 were detected only by PCR, i.e., were submicroscopic. Analysis of the 121 microscopy-and PCR-positive adults revealed that asymptomatic adults infected with P. falciparum showed significantly higher breadth (p < 0.03; Fig 4A) and intensity (p < 0.03; Fig 4B) of antibody response than symptomatic adults. In addition, we observed that the asexual parasitemia levels in asymptomatic adults were much lower (p < 0.002; Fig 4C) compared with symptomatic adults infected with P. falciparum. In the case of P. vivax, no significant differences in the breadth (Fig 4A) and intensity (Fig 4B) of antibody response were observed between adults with asymptomatic and symptomatic infection. The levels of asexual parasitemia were also comparable between asymptomatic and symptomatic adults infected with P. vivax (Fig 4C).
We were interested in identifying antigens that have significantly higher reactivity in sera from asymptomatic individuals compared to symptomatic individuals. We used log 2 (FOC) values to identify antigens that were differentially reactive between asymptomatic and symptomatic adults (p < 0.05, with Benjamini-Hochberg correction for false-discovery rate). Asymptomatic adults infected with P. falciparum displayed significantly higher reactivity to several P. falciparum antigens compared to symptomatic adults. We have identified 19 P. falciparum antigens that were recognized with 2-fold greater intensity in asymptomatic adults in India ( Table 3). On the other hand, no P. vivax antigens showed significantly higher reactivity in asymptomatic compared to symptomatic individuals infected with P. vivax.

Discussion
We present the first genome-scale analysis of seroreactivity to Plasmodium antigens in the Indian population using protein microarray technology. The three field sites in our study are distributed across eco-epidemiologically diverse regions in India, with differing prevalence of P. vivax and P. falciparum. We observed a broad response to P. vivax and P. falciparum antigens at all three sites, regardless of malaria infection status. Overall levels of antibody binding were greater in P. falciparum antigens compared to P. vivax antigens, as previously reported in regions with co-endemic occurrence of P. vivax and P. falciparum [35]. Our Indian sample cohort showed significant seroreactivity to 265 P. vivax and 373 P. falciparum antigens compared to unexposed US controls.
In the case of P. falciparum, we observed that decreasing prevalence corresponded with a decrease in the breadth of antibody response. Malaria-negative adults showed a similar pattern to malaria-positive adults at all sites, reflecting the background seroreactivity due to parasite exposure. The breadth of response to P. falciparum antigens was significantly lower in adults from Chennai compared to both Raurkela and Nadiad; although the mean breadth of response appeared to be lower in Nadiad compared to Raurkela, the decrease was not statistically significant.
We did not observe a correlation between the breadth of antibody response and species prevalence for P. vivax. Malaria-positive adults from Nadiad showed significantly larger breadth of response than adults from both Chennai and Raurkela; this was not observed in malaria-negative adults. We were puzzled that regardless of their malaria infection status, the seroreactivity of individuals from Chennai was similar to Raurkela, which has much lower prevalence of P. vivax. Since we began the study in 2012, the incidence of malaria has gone down in India, particularly in the state of Tamil Nadu where Chennai accounts for~55% of the malaria burden [28]. Thus, the reduced seroreactivity of Chennai individuals may be reflective of reduced transmission levels during the period of sample collection for our study. Interestingly, some Indian states, including Odisha and Gujarat, witnessed a spike in malaria cases in 2014 [36]. Therefore, the prevalence of P. vivax and P. falciparum in Raurkela and Nadiad may have been higher than expected during the course of our study. Hypothetically, P. vivax seroreactivity may also be affected by differential rates of P. vivax relapse; sites with a higher relapse rate may exhibit greater priming of the immune response. However, we do not have data on relapse rates at our three sites to address this conclusively. Additionally, the breadth of response could be affected by differences in the total number of individuals and the proportion of symptomatic and asymptomatic infections between the sites.
The breadth and magnitude of immune response have been known to increase with age, as a consequence of repeated exposure to the parasite. We compared the responses of children and adults from Nadiad and Raurkela to determine the influence of age on the immune response against malaria. The breadth and intensity of response against P. vivax as well as P. falciparum antigens was significantly higher in adults than in children.
Antigens recognized with the highest intensity by sera from both symptomatic and asymptomatic malaria patients may serve as indicators of exposure. Data from other ICEMR studies [27] have highlighted the need to identify country-specific indicators of exposure, as they may vary depending on the epidemiology of malaria in different parts of the world. As the  prevalence of the two major Plasmodium parasite species varies across India, we were interested in identifying highly immunogenic antigens from all three sites that could be used to develop a serological assay for countrywide routine surveillance. It is also important to elucidate the kinetics of antibody acquisition and maintenance [37] in order to distinguish between recent versus past exposure. Markers of exposure in sera from children may be better indicators of recent exposure, as adults may have had several exposure events, confounding the evaluation of responses from recent exposure. Several of the top immunogenic P. vivax proteins in the Indian population, such as ETRAMP, MSP10, MSP8, and hypothetical proteins such as PVX_117680, PVX_083560, PVX_097730, PVX_110935, PVX_118705 have also been identified as highly immunogenic in other studies [38][39][40][41][42]. Among the top immunogenic proteins in P. falciparum identified in our study, members of the PfEMP1 family, PTP5, MSP10 and HSP70 were also identified as immunogenic or recognized with greater intensity in asymptomatic malaria in other ICEMR studies [13,37,43,44]. Invasion-related proteins AMA1 and SYN6 were also highly immunogenic [45]. Asymptomatic individuals may not be protected from malaria parasite infection, but they may possess immunity against symptomatic disease. Our data indicate that adults with asymptomatic P. falciparum infection have lower average asexual parasitemia, but higher breadth as well as intensity of response, than adults with symptomatic infection. Asymptomatic P. falciparum infection was also associated with significantly higher seroreactivity to several P. falciparum antigens as compared to symptomatic infection, and these antigens may serve as novel vaccine candidates in addition to the limited repertoire of candidates currently being developed. A majority of the antigens associated with clinical immunity in P. falciparum infections are either exported to the infected red blood cell during the intraerythrocytic stages of parasite development (PfEMP1, RESA, ETRAMP, PHISTc, Hsp70-x, GEXP18) or present on the merozoite surface (MSP2, MSP4, MSP11, SERA4). These proteins are exposed to the host immune system for the longest duration of the infection, facilitating the development of a strong immune response. Some of these antigens play essential roles in vital processes such as invasion (RON2), making them promising vaccine candidates. GCN5 and LSA3, previously associated with protection from experimental challenge with sporozoites [46,47], and invasionrelated proteins, RON2 and RH2b, were recognized more strongly by asymptomatic individuals in our study MSP2, MSP4, MSP11, PHISTc, Rh2b, PfEMP1, LSA3 and SERA4 were also associated with protection from symptomatic disease in other malaria-endemic regions [13,35,43] demonstrating that the antigens associated with asymptomatic infection are common across populations, which is encouraging for the development of a vaccine.
Comparison of serological profiles of adults with symptomatic or asymptomatic malaria suggests that immunity to P. falciparum is associated with a broad and intense response to several antigens, but with low parasitemia levels. In contrast, adults with asymptomatic and symptomatic P. vivax infection had comparable breadth and intensity of antibody response, as well as similar asexual parasitemias. In addition, asymptomatic P. vivax infection was not associated with higher seroreactivity to specific P. vivax antigens. Based on these findings, it appears that unlike P. falciparum, antibody-mediated immune responses may have a much lower contribution in asymptomatic P. vivax infections. P. vivax is thought to be more pyrogenic than P. falciparum, since it stimulates a stronger fever-inducing cytokine response despite frequently presenting at a lower parasite burden than P. falciparum [48]. However, Goncalves et. al., propose that regulatory cytokines, such as IL-10 may play a more critical role in protecting P. vivax patients from severe clinical complications, while a strong inflammatory response may be critical in controlling parasite density in P. falciparum infections [14].
Our pilot study was conducted in tandem with other protein microarray projects within the ICEMR program and using the same Pf/Pv500 protein array [13,27,35,43]. Together, these studies facilitate a global comparison of seroreactivity to over 1000 Plasmodium antigens, and provide a means to identify highly immunogenic proteins and antigens recognized more strongly in asymptomatic individuals, which may subsequently be used for the development of vaccines and routine surveillance tools. We acknowledge the limitations of using this array, since the antigens have been produced from sequence information of decades-old single reference isolates (P. falciparum 3D7 and P. vivax Salvador I). Our future studies will include country-specific redesign of the arrays using sequence data from diverse parasite strains circulating at our study sites, in order to capture regional diversity in antigenic genes. In addition, appropriate study design will be extremely important to tease out the complexities of the host immune response to malaria. In particular, prospective cohort studies, with careful monitoring of parasite transmission and patient serology, are likely to be the most informative. In conclusion, these data have broadened our understanding of naturally acquired immunity against P. vivax and P. falciparum in India, and will contribute to global malaria control and eradication measures.