Serologic responses to the PfEMP1 DBL-CIDR head structure may be a better indicator of malaria exposure than those to the DBL-α tag

Plasmodium falciparum erythrocyte membrane protein-1 (PfEMP1) antigens are critical to parasite sequestration, and their genetic diversity likely facilitates host immune system evasion. The var gene family encodes PfEMP1 antigens, whose extracellular binding region is made up of Duffy-binding-like (DBL) domains interspersed with cysteine-rich interdomain regions (CIDRs). A 300–400 variable nucleotide region surrounded by semi-conserved motifs in the DBL-α domain, the first N-terminal var domain, acts as a unique “fingerprint” specific to individual vars. DBL-α “tag” sequencing facilitates analyses of both PfEMP1 genetic diversity [1, 2] and seroreactivity [3, 4].

Antibody responses to locally derived DBL-α tags in Papua New Guinea increased in magnitude and prevalence with age in children, suggesting that DBL-α tags correlate with malaria exposure and potentially the development of natural immunity to malaria [4]. In contrast, a subsequent study of seroreactivity to 3D7 strain PfEMP1 antigens found that sera from children with cerebral malaria or severe malarial anaemia did not differ in recognition of DBL-α tags compared to sera from matched uncomplicated malaria controls in Bandiagara, Mali, but did with respect to larger extracellular PfEMP1 fragments [5].

Protein microarray construction followed a four-step process that included: (1) PCR amplification of each complete or partial P. falciparum open reading frame, (2) in vivo recombination cloning, (3) in vitro transcription/translation (IVTT), and (4) microarray chip printing [6]. This IVTT protein microarray platform has produced antibody responses strongly correlated with those from ELISA assays featuring purified proteins of several malaria vaccine candidate antigens [7]. The microarray included nine DBL-α tags that encode Group A, B, B/A, and C PfEMP1 antigens from the 3D7 reference genome; nine matched parent DBL-CIDR head structures; and 18 matched downstream PfEMP1 fragments. Each DBL-α tag was expressed as a single protein fragment, and the corresponding DBL-CIDR head structure was expressed as a single protein fragment.

The protein microarrays were probed with sera from 18 adults aged 18–55 years old in the control arm of a trial of an apical membrane antigen 1 (AMA1) vaccine adjuvanted to GlaxoSmithKline’s AS02A (FMP2.1/AS02A) [8] and from 35 children aged 1 to 6 years in the control arm of a Phase II AMA1 vaccine (FMP2.1/AS02A) trial [9] using published methods [6]. These serum samples were collected with respect to two malaria transmission seasons in rural Mali [adults: June 2005 (pre-malaria season) and December 2005 (post malaria season); children: May 2007 (pre-malaria season) and September 2007 (peak-malaria season)]. Sera from 11 US blood donors were used as negative controls. The data were background subtracted using the mean of the no-DNA controls, and negative fluorescence intensities were zeroed.

All participants or guardians of participants provided written informed consent, and the trial was conducted under the Declaration of Helsinki. The institutional review boards of the Faculty of Medicine, Pharmacy and Dentistry, Bamako, Mali, and the University of Maryland approved the study protocol.

Fluorescence intensities to DBL-α tags and their parent DBL-CIDR head structures were measured using sera collected from Malian children and adults and North American malaria-naïve adult controls (Fig. 1). Overall, increased serologic responses to the DBL-CIDR domains compared to their corresponding DBL-α tags were observed for both children and adults.

DBL-α tags have been used as a unique “fingerprint” reflecting PfEMP1 diversity in gene expression, serologic, and population genomics studies [4, 13‐15]. These DBL-α tag sequences serve as a useful means of PfEMP1 classification, because each PfEMP1 contains only one DBL-α domain and most PfEMP1 antigens include a DBL-α domain. The DBL-α tag consists of a 300 to 400 base pair variable region flanked by conserved motifs that provide targets for degenerate primers [16, 17]. The findings here suggest that such a unique sequence tag may not perform well as a sensitive marker for malaria exposure, potentially due to the diverse nature of the protein fragment. Seroreactivity to the parent DBL-CIDR head structure containing more conserved sequences better identified age and seasonal malaria exposure.

These findings contrast with results from a study done in Papua New Guinea that found that serologic responses to DBL-α tags revealed differences in malaria exposure. One hundred twenty-three DBL-α tag sequences from Papua New Guinea populated a protein microarray that was used to seroprofile Papua New Guinea adults and children, revealing that children’s antibody responses peaked between four and 15 years of age [4]. In this Papua New Guinea population, DBL-α tags served as a marker for age-related exposure to malaria. Of note, DBL-α tags from Papua New Guinea are less diverse than DBL-α tags from global populations. In a population genomic analysis of DBL-α tags from both Papua New Guinea and other malaria-endemic regions, the number of unique DBL-α tags plateaued with 30 Papua New Guinea isolates examined, but no such plateau was evident with more than 1000 DBL-α tag sequences from 59 global isolates [2]. Serologic responses to Papua New Guinea DBL-α tags may therefore be more informative than responses to DBL-α tags in locations with more genetically diverse DBL-α tags. In the current study, age-related Malian serologic responses to DBL-α tags were not observed. However, we evaluated fragments from nine PfEMP1 antigens compared to the 123 evaluated from Papua New Guinea. It is possible that with a large enough pool of DBL-α tags on a microarray and a larger sample size, differences in age-related exposure may be discerned.

Antibody responses to particular PfEMP1 domains have been associated with protection from clinical malaria in several studies [7, 18‐20], and may play a role in protection from severe malaria disease. A recent study of antibody responses in Malian children with severe malarial anaemia and/or cerebral malaria found gaps in immunity to particular PfEMP1 antigens compared to uncomplicated malaria controls [5]. Sera from children with severe malaria recognized fewer PfEMP1 antigens and reacted less intensely to particular PfEMP1 fragments compared to controls with uncomplicated malaria. Immunologic gaps included several DBL-CIDR domains, but only two associated DBL-α tags. In fact, immunologic gaps to particular DBL-CIDR head structures did not predict a gap to a particular DBL-α tag, and vice versa. This is consistent with the findings here that serologic responses to DBL-α tags were not as predictive as markers of malaria exposure compared to responses to larger PfEMP1 fragments such as head structures.

Earlier microarray studies have found that more conserved PfEMP1 fragments are markers of malaria exposure. In a study of Malian children and adults followed during the malaria season, the most differentially seroreactive fragments in adults compared to children contained the intracellular ATS domain, which is more conserved than extracellular domains [11, 21]. Similarly, in this study, paediatric sera recognized all three of the ATS domains. In Malian P. falciparum genomes, DBL-α and DBL-β domains have more sequence conservation than other DBL domains [22]. Here, children recognized head structures containing DBL-α and DBL-β domains while other extracellular PfEMP1 domains with lower sequence conservation remained unrecognized. Taken together, these findings suggest that when using PfEMP1 antigens as microarray markers of malaria exposure, the most conserved portions may be the most informative for malaria exposure. Such ordered serorecognition of PfEMP1 domains based on sequence conservation may also underlie the order of PfEMP1 domain recognition identified in Tanzanian children during the first 2 years of life [23, 24].

This study had some limitations. The examined DBL-α domains were from the reference strain 3D7, not from circulating Malian parasites in the population. In spite of this, individuals had seroreactivity to head structures and to the downstream PfEMP1 fragments, particularly Malian adults, who as a group recognized all DBL-CIDR head structures and downstream fragments. This suggests that the PfEMP1 antigens from 3D7 may represent the much larger circulating PfEMP1 population in Mali to some degree. Another limitation is the limited DBL-α tag population studied: DBL-α tags from nine 3D7 PfEMP1 antigens were examined, which represents only a small subset of the 60 3D7 PfEMP1 antigens, although at least one member of each PfEMP1 subgroup was included. This limited representation of PfEMP1 antigens could be another possible explanation for the limited serorecognition by children. It is possible that DBL-α tags from other 3D7 PfEMP1 antigens could have elicited more serorecognition in Malian children, but were not represented on the microarray. Whereas proteins expressed on this microarray platform have been shown to induce antibody responses similar to those from purified proteins, these proteins are produced with an Escherichia coli system and lack post-translational modifications such as glycosylation [7]. Correct folding of the PfEMP1 proteins produced by the Escherichia coli expression system has not been confirmed.

To further examine the role of conserved PfEMP1 domains in inducing serologic responses, peptide microarrays are a potentially useful tool to identify reactive epitopes. Such an approach may provide insight distinct from protein microarrays, where entire domains are included and may elicit differential seroreactivity. A protein microarray does not permit identification of the specific roles of either more conserved or more diverse regions within a protein fragment. Peptide arrays have already been used as a tool for studying epitopes in variant surface antigen proteins in Plasmodium falciparum. In particular, they have recently been used to pinpoint epitopes targeted by antibodies in a DBL-α domain that mediates rosetting [25] and epitopes within particular domains of STEVOR and RIFIN antigens reflecting malaria exposure [26]. This approach could be used to identify epitopes targeted by protective PfEMP1 antibodies and thereby inform vaccine or therapeutics design.

Malian serologic responses to a PfEMP1’s DBL-α tag, a semi-conserved short fragment of the DBL-α domain, did not correlate with seasonal malaria exposure or responses to the parent DBL-CIDR head structure in either adults or children. Larger, more conserved PfEMP1 domains may be better indicators of malaria exposure than short, variable PfEMP1 fragments such as DBL-α tags. PfEMP1 head structures that include conserved sequences are thus well suited for study as serologic predictors of malaria exposure on protein microarrays. Further support for this finding would include testing a broader panel of PfEMP1 antigens, including those from non-3D7 reference strains and from endemic parasite genomes.