|Home | About | Journals | Submit | Contact Us | Français|
Plasmodium falciparum imposes an enormous burden upon the developing world, with 300 to 500 million cases and 1 to 2 million deaths per year (94). Despite extensive research efforts, development of parasite drug resistance is a growing problem, and an effective vaccine is still lacking. Individuals living in areas of high P. falciparum transmission acquire protective immunity to severe malaria during early childhood after only a few symptomatic infections yet remain susceptible to uncomplicated disease and asymptomatic infection into adulthood (65). Thus, sterile immunity that prevents infection may never develop, but significant antidisease immunity is acquired relatively rapidly. While the protective targets of antidisease immunity are largely unknown, the parasite variant antigens exposed at the erythrocyte surface are considered strong candidates.
Key virulence factors and prime candidates for antidisease vaccines have been identified in a family of clonally variant surface antigens collectively termed P. falciparum erythrocyte membrane protein 1 (PfEMP1), encoded by about 60 var genes per haploid genome (9, 40, 92, 97). P. falciparum-infected erythrocytes (IEs) bind host endothelium and other host cells, in turn sequestering infected cells away from the spleen, which would otherwise destroy them. Switching of var gene expression allows the parasite to modify the antigenic and functional properties of IEs, thereby evading immunity and affecting infection outcome.
How antidisease immunity could be achieved rapidly against variant surface antigens is a deep mystery. Unraveling the basis for this protection represents a promising direction for antidisease malaria vaccines. This review considers how var gene organization may shape the functional and antigenic properties of PfEMP1 variants and regulate their expression during infection.
The var gene repertoires have been compared in three geographically diverse P. falciparum clones: 3D7, HB3, and IT4/25/5 (IT4), representing Africa, Central America, and Southeast Asia, respectively (52). Of the three, var repertoire coverage is most complete for the genome reference isolate, 3D7 (40). HB3 has been sequenced to 8× genome coverage (105), and IT4 var genes were identified by targeted gene cloning. Overall, 3D7 has 61 var genes, HB3 has 54 full and partially sequenced var genes (including six pseudogenes), and 48 full and partially sequenced var genes have been cloned from IT4.
Despite having distinct geographic origins, the general organization of var genes appears to be similar between isolates. In 3D7, approximately two-thirds of the 61 var genes are located at subtelomeric regions of the 14 chromosomes, with the remainder in chromosome central regions (40). Each chromosome end typically contains one, two, or three var genes, followed by a group of rif, stevor, and other multigene families. Many subtelomeric regions have two var genes arranged in tail-to-tail orientation relative to each other with one or more rif genes in between (Fig. (Fig.1A).1A). Chromosome central var genes can appear singly or in groups that are nearly always tandem arrays (head to tail), containing from three to seven var genes in genomes sequenced to date. Superimposed on this general organization, the chromosomal location and transcription orientation of a var gene can be predicted from its 5′ noncoding region sequence (106). Based upon sequence similarity, the 5′ promoter regions can be defined into four major upstream (Ups) sequence groups, UpsA, UpsB, UpsC, and UpsE. The former UpsD is now grouped with UpsA (52). The functional significance of these different promoter types is unclear. However, in all three parasite genomes where the gene location has been mapped, UpsC var genes are chromosome central, UpsB var genes are either subtelomeric and transcribed away from the telomere or chromosome central in tandem arrays with other UpsB or UpsC var genes, and UpsA and UpsE-type var genes are subtelomeric but transcribed towards the telomere in the opposite direction to UpsB var genes (Fig. (Fig.1A).1A). There is only one known exception, a UpsA var gene that is predicted to be in a central chromosome cluster in HB3 (52), but this may represent a sequence assembly artifact. This overall conservation of var gene organization based on 5′ noncoding sequences, in an organism where one-to-one pairings between true alleles might only occur in meiotic selfing, may reflect either recombinational constraints and/or host selection pressures. Functional requirements for controlled var gene expression in each chromosomal location might also select for the conservation and particular arrangement of var promoter types.
While recent studies have made great progress in investigating genetic diversity across the P. falciparum genome (45, 48, 71, 105), these studies have not been able to address the highly diverse and largely nonallelic var gene family. Thus, the mechanisms driving var gene diversity remain only partially understood. Plasmodium parasites are haploid during the vertebrate stage of infection and replicate asexually by mitotic division. Sexual (meiotic) recombination only occurs during the mosquito stage of the infection. Ectopic gene recombination among var genes occurs during meiosis (34, 98) and potentially during mitotic division, although the latter has not been definitively shown. Sequence comparisons and restriction fragment length polymorphism analysis of parasite crosses and population studies suggest that both small (~100 to 200 nucleotides) and larger recombination events contribute to var gene evolution (7, 13, 34, 52, 98, 100, 108). Also, recent studies have shown that higher rates of recombination events occur in the chromosomal regions near or surrounding var genes (70, 71), further confirming the role of recombination in generating var gene diversity.
P. falciparum telomeres form four to seven clusters of chromosome ends, called “bouquets” (34). It has been speculated that telomere end clustering may allow var genes with similar 5′-flanking sequence and gene orientation to preferentially line up and facilitate recombination within the different var groups (53, 59, 88). This may involve some members of the rif multigene family, because many rif genes are linked head to head with UpsA var genes, and the entire intergenic region containing both genes’ promoters is highly conserved (59). Repertoire-wide comparisons of 3D7, HB3, and IT4 show that coding sequences of UpsA var genes have diverged from UpsB and UpsC var genes (52). While the extent to which subtelomeric group B and central group C var genes might be recombining is unknown, central var genes have been shown to be located within the nuclear periphery (78) and may be colocalizing with telomere bouquets (107). However, it is uncommon to find larger segments of gene similarities, greater than 500 bp, shared between subtelomeric and central var genes (52), suggesting that exchange between these groups may be infrequent. These sequence relationships support the hypothesis that recombination preferentially occurs within genes that have a common genome location and gene orientation and is also likely influenced by 5′-flanking region and gene coding similarity (Fig. (Fig.2).2). This recombination hierarchy may be shaping the var gene repertoire and influencing the evolution of the family.
The var gene repertoire also includes three unusual semiconserved var genes (var1CSA, var2CSA, and type 3 var) that are found in most parasite isolates. Interestingly, these distinctive genes share little sequence identity with other members of the var gene family and may therefore primarily undergo self-self recombination (52). Remarkably, homologs of var1CSA and var2CSA are present in the chimpanzee malaria parasite Plasmodium reichenowi (100), although it is believed to have diverged from P. falciparum ~5 to 7 million years ago (29). Therefore, these isolate-transcendent genes have ancient origins and may be under special selection to be maintained in the parasite population.
It has been hypothesized that an original ancestral var gene was duplicated and diverged into three main var groups (A, B, and C) and subsequently into transitional groups (59). This interpretation is supported by analyses showing similar categories of var genes in all three parasite isolates (Fig. (Fig.3A).3A). Interestingly, sequence comparisons of var2CSA alleles from a global collection of parasite isolates including P. reichenowi revealed high levels of gene mosaicism and extensive overlap of polymorphic segments (100). Thus, ancient polymorphic segments have been carried forward into existing parasite populations, possibly from a burst of polymorphism that occurred in the common progenitor to these isolates. Small stretches of shared polymorphism have also been observed between less-conserved var genes and potentially contribute to cross-reactive antibody responses (15, 108). These findings suggest that recombination of slowly mutating polymorphic segments has an important role in var gene diversification and that recombining genes may be more likely to share antigenic relationship. Furthermore, ancestral polymorphism may maintain antigenic similarities between otherwise diverse PfEMP1 variants even across global isolates.
Figure Figure44 shows a possible scenario of var gene evolution in which a gene duplication event occurred very early in the evolution of the gene family, followed by gene diversification and inversion into conserved orthologs, such as var1CSA and var2CSA and the var groups, before P. falciparum and P. reichenowi split. Future studies testing if similar patterns of var groupings exist in P. reichenowi as in P. falciparum will be useful in shedding light on the origins and evolution of the var gene family.
PfEMP1 proteins are encoded in two exons, with exon 1 encoding the exposed extracellular portion (Fig. (Fig.1B)1B) (97). This region is extremely variable in both sequence and length, although it consists of a few fundamental building blocks put together with some minimal rules (93). A short region of the N terminus (NTS) contains sequence features sufficient for transport beyond the parasitophorous vacuole that surrounds the intraerythrocytic parasite (42, 66). Other regions of the PfEMP1 protein may assist in final transport to the erythrocyte membrane (50). The remainder of the extracellular region consists of two main adhesion domains: Duffy binding like (DBL) and cysteine-rich interdomain regions (CIDR), which are classified according to sequence similarity.
The diverse exon 1 structures of PfEMP1 variants can be categorized by their domain combinations, typically ranging from two to seven DBL domains and one or two CIDR domains. Thirty-one different architectural types were described in the three sequenced parasite var repertoires (52), with certain tandem domain combinations consistently preserved (DBLα-CIDR1, DBLβ-c2, and DBLδ-CIDR) (Fig. (Fig.1).1). Most PfEMP1 variants have a semiconserved protein head structure consisting of NTS-DBLα-CIDR1 domains (40).
While seven protein architectural types are shared among the three isolates, most PfEMP1 proteins have overall amino acid identities of less than 50% in individual domains, even among proteins of the same architectural type (52). In addition, there is minimal overlap of DBLα tags in population surveys of parasite isolates (7, 13, 31, 100). The only exceptions are the three isolate-transcendent vars, which have identities greater than 75% over multiple domains (53). The vast antigenic diversity of PfEMP1 proteins in the parasite population may help explain why individuals are repeatedly susceptible to P. falciparum infections and never develop sterilizing immunity. Nevertheless, although the diversity of variant antigens is indeed large, hyperimmune human sera from distinct geographic regions are able to recognize IEs from East or West Africa (2), suggesting that some epitopes are globally related, possibly due to ancestral polymorphism and gene recombination.
Despite having distinct var repertoires, the three parasite genomes have approximately the same number of genes in each var group (Fig. (Fig.3A).3A). However, the ratio of small to large PfEMP1 proteins differs between isolates, and there is an overrepresentation of distinct architectural types in the different genotypes (Fig. (Fig.3B).3B). Consistent with the idea that var genes may be functionally and structurally diversifying under a gene recombination hierarchy, UpsA PfEMP1 variants tend to have larger, more complex domain architectures and encode a distinct protein head structure. In addition, the UpsA group lacks the type 1 PfEMP1 protein architecture, which is otherwise the most common PfEMP1 type in all three parasite genomes (Fig. (Fig.3B)3B) (52).
Relatively few PfEMP1 adhesion traits have been mapped to specific domains or proteins (summarized in Fig. Fig.1B).1B). The best-characterized binding interaction is between the CIDR1 domain and the host receptor CD36 (8). Considered the primary receptor for IE binding to blood microvessels, the CIDR1 domains from UpsB and UpsC PfEMP1 proteins tend to bind CD36 (80). In contrast, the UpsA PfEMP1 proteins tend not to bind CD36, due either to primary sequence differences in CIDR1 in the UpsA-type-associated PfEMP1 (80) or complete lack of CIDR domains in type 3 var and var2CSA (40). Significantly, var2CSA is specifically upregulated in infected erythrocytes that have switched away from CD36 binding and sequester in the placenta (36, 86). There is considerable interest in understanding whether specific groups of PfEMP1 proteins may have different roles in binding and disease and how this may relate to var groups and recombination potential.
Of the different malaria disease syndromes, the role of PfEMP1 proteins is best understood for pregnancy-associated malaria (PAM) (reviewed in references 38 and 81). During pregnancy, women who have previously developed malaria immunity become susceptible to IEs that bind low-sulfated chondroitin sulfate A (CSA) in the placenta (3, 36). After one or two pregnancies, women develop protection to the placental form of the disease. This protection is correlated with the development of antibodies that recognize placental parasites from different geographical regions (26, 37, 96), suggesting that the surface molecules expressed by placental infected erythrocytes may have unique and conserved features that may be utilized in the development of a pregnancy malaria vaccine.
One such “conserved” antigen and possible vaccine candidate has been identified, UpsE-associated var2CSA. VAR2CSA is conserved at approximately 70 to 80% amino acid identity across global isolates (53, 87, 100) and is transcriptionally upregulated in placental isolates and parasites selected to bind CSA (23, 24, 39, 87, 101). Disruption of var2CSA causes infected erythrocytes to lose their ability to bind CSA (25, 104). Furthermore, Salanti et al. showed that high levels of anti-VAR2CSA antibodies correlated with a lower risk of delivering low birth weight children (86). Detailed serological comparisons have shown that the human immunoglobulin G response against PAM appears to be highly focused on polymorphic regions in VAR2CSA (6, 10, 20). Significantly, sequence analysis indicates there is extensive overlap of VAR2CSA polymorphism between globally diverse parasite isolates and that polymorphic loops assume a relatively limited diversity of basic types (11, 100). Whether PAM immunity is conferred by a repertoire of antibodies that collectively recognize polymorphic epitopes shared between different VAR2CSA alleles, or if low level antibody responses develop to highly conserved but “cryptic” epitopes, is the subject of investigation.
Similar to PAM disease, it is being investigated whether variant surface antigen types associated with other malaria disease syndromes may be antigenically restricted. Severe childhood malaria encompasses several disease syndromes (severe anemia, cerebral malaria, respiratory distress, and hypoglycemia) and has been linked to sequestration of infected erythrocytes to many tissues (63). Also, severe malaria has been associated with parasite phenotypes such as rosetting (binding of infected erythrocytes to uninfected erythrocytes), adhesion to brain microvasculature, and autoagglutination (clumping of infected erythrocytes bridged by platelets) (14, 73, 76, 79, 83, 102). To determine if specific PfEMP1 proteins are important to one or more of the severe malaria syndromes, researchers are analyzing var gene expression during infections, characterizing the antibody response to the infected erythrocyte surface, and investigating the binding properties of PfEMP1 proteins.
To date, only five studies representing a total of 88 patients with differing forms of severe malaria have been done to investigate the types of var gene(s) that are expressed during disease (13, 47, 49, 58, 82). Differences in epidemiology, severe disease characterization, and var classification make comparisons across studies difficult. In three studies, expressed DBLα sequence tags were classified by the number of cysteines encoded (49) and other features (13, 58). Genes with two cysteines in this region (2cys/DBLα1 type) are likely to represent UpsA var genes, or a small subgroup of UpsB var genes (58), whereas those with four cysteines are either UpsB or UpsC. So far, expression of 2cys/DBLα1/UpsA sequence variants correlates with rosetting phenotype (13, 58), with severe cerebral malaria in children (58) and noncerebral severe malaria in adults (49). Real-time PCR analysis determining the expression of var genes with the different promoter types showed a correlation between expression of both UpsA and UpsB var and severe malaria cases in Tanzanian children (82), but in Papua New Guinea only UpsB var expression correlated with severe disease (47). However, the latter study was in an area where ~79% of the population is deficient in complement receptor 1 (19), a major receptor for infected erythrocyte rosetting, and the rosetting phenotype does not associate with severe disease (5). Therefore, it is possible that human genetic polymorphisms in cytoadhesion receptors may influence PfEMP1 disease associations. Although strict correlations between any group of var genes and disease status have not been found, expression of both UpsA var and UpsB var has been associated with severe disease at different geographic sites.
Although many different parasite genotypes are potentially virulent (22, 65, 69), severe malaria syndromes are a relatively infrequent complication of malaria infections (estimated to be about 1% of infections) (65), suggesting that isolate-transcendent disease immunity can develop rapidly. Given the limited overlap of variant antigen repertoires, with only three known isolate-transcendent variants, it can be questioned whether PfEMP1 immunity is an important factor in the rapid development of disease immunity. Although the variant antigen repertoire is vast, serological evidence suggests that the variant surface antigens associated with disease may be antigenically restricted (12, 13, 46, 74). By analogy to PAM immunity, we hypothesize that a “patchwork” epitope relationship between disease-promoting PfEMP1 variants could contribute to antigenic cross-reactivity, especially var genes in the same gene recombination group, because these are more likely to share regions of overlapping polymorphism. Therefore the concept of a var gene recombination hierarchy has implications for investigating antidisease immunity and vaccine development. To test this hypothesis it will be necessary to develop a better understanding of the sequence and antigenic relationship of PfEMP1 disease variants (13, 46, 62, 64).
Information about var gene regulation may provide further insight into the factors influencing gene expression during infection or lead to drug interventions that could prevent PfEMP1 expression. var genes are expressed in a mutually exclusive manner, with only one PfEMP1 protein expressed by any individual parasite (60, 89). The relatively limited var repertoire in any haploid, clonal parasite population requires that antigenic variation take place at a rate sufficient to maintain infection in face of the acquired host immune response, but not so rapidly that the repertoire is exhausted before the parasite can be transmitted via mosquito bite to another host. Whether the parasite is able to prolong blood-stage infections by var mutation/recombination is unknown, as is the relative contribution of mitotic (vertebrate host) versus meiotic (mosquito host) processes to var diversification. Antigenic variation for PfEMP1 comprises both memory for expression of the same variant in most progeny parasites and switching to expression of new types at variable rates (43). The paradigm for study of antigenic variation in protozoan parasites is the variant surface glycoprotein (vsg) of Trypanosoma brucei (99). With few parallels between the two parasites’ antigenic variation systems, and the big picture describing var gene regulation still uncertain, at least a few details are now clear.
While P. falciparum parasites have a 48-h blood-stage asexual life cycle, full-length var transcripts are only detected up to about 20 h postinvasion (56). The var protein, PfEMP1, is synthesized early and relies on a slow trafficking pathway to arrive at the IE surface (54). One variant, var1CSA, is unusual in that it is transcribed constitutively in all parasites, even as a truncated pseudogene (PFE1640w in 3D7), and thus falls outside the controls of mutually exclusive gene expression (57). Whether the protein encoded by var1CSA is exposed at the red cell surface is unknown, and its role in antigenic variation is currently unclear.
Unlike T. brucei vsg, P. falciparum var genes do not undergo rearrangement or gene conversion into an active expression site in order to be switched on, although there are examples of var gene deletion accompanied by adjacent var gene activation (21, 44). Specific var gene expression is activated in situ (89, 92) and is controlled at the level of transcription initiation soon after the parasite invades a red blood cell (55, 90).
Critical features of var gene exclusive expression have been identified through transfection of parasites with plasmids containing both (i) a var 5′ promoter driving a drug resistance marker, and (ii) a second promoter in the form of either the var intron or a heterologous promoter driving a second drug resistance marker (28, 107). By selecting for expression from the var 5′ promoter/drug resistance marker, all endogenous var gene expression is shut down. The var 5′ promoter-driven drug resistance marker appears to “occupy” the single var expression site, filling the place normally occupied by an endogenous 5′ var promoter. Depending on the interpretation, the presence of the var 5′ promoter is either sufficient for the drug resistance marker to be counted as a var gene by the exclusive expression mechanism (107) or the interaction between the 5′ var promoter and a second promoter (var intron or heterologous promoter) is required (28; reviewed in reference 32). Either way, the strict counting mechanism relies not on negative feedback from the var protein product itself but on noncoding information. Although the stage-specific expression of var intron transcripts suggested that they may have a potential role in var silencing, it has not been possible to establish a relationship between these transcripts and the active or silent state of a var locus (55, 56, 77, 97).
All var genes, regardless of chromosomal location or transcription status, appear to physically reside at the nuclear periphery (27, 67, 78, 107), usually in telomeric clusters (34). By using parasite populations known to be highly homogeneous for expression of a single var gene, it has been shown that the active var gene moves away from the “silent” cluster to another position at the nuclear periphery (78). However, from studies of parasites transfected with various episomally maintained telomere-homing constructs or drug resistance genes under control of a var-specific promoter (27, 67, 107), it appears that active var promoters can associate in the bouquet structures.
Whatever the explanation for these differences, the idea that an actively transcribed var locus undergoes nuclear repositioning to an active transcription location is reminiscent of the T. brucei expression site body (72), the extranucleolar site of vsg transcription. Although it is well-established that the polycistronic vsg-containing mRNA is transcribed by RNA polymerase I, it has recently been shown that var genes are transcribed by RNA polymerase II (55, 90), like all other protein-coding genes that have been investigated in P. falciparum (68). Moreover, any active var gene would usually be linked to at least one silent var gene within several kilobase pairs (15 single var genes versus 46 var genes in groups in the 3D7 genome). Therefore, mechanisms must exist to prevent transcriptional upregulation of nearby var loci following nuclear repositioning. The physical movement of a var gene to a transcription-permissive region of the nuclear periphery cannot universally explain mutually exclusive regulation: further layers of control must operate.
Histone modifications can influence gene expression by altering DNA accessibility or recruiting other nonhistone proteins (51). Chromatin structure may hold some clues to var gene regulation, as histone hypoacetylation correlates with var gene silencing (35). The P. falciparum homologue of the histone deacetylase SIR2 (PfSIR2) protein associates with silent var 5′ promoter types UpsE and UpsB, but not UpsC (35), consistent with a telomere-silencing association for this protein. Knock-out of the PfSir2 gene results in activation of only certain subtelomeric var genes, some UpsA and the UpsE-type var genes (27), which suggests that the UpsB-type var genes are subject to a further layer of silencing. Recent studies in yeast and T. brucei have shown links between the factors involved in telomere gene silencing (including SIR2) and DNA repair (4, 61, 75, 103). These connections raise the possibility that factors involved in regulating the transcription of var genes may also facilitate recombination between family members.
The modified histone H3K9me3 (H3 trimethylated at lysine 9) appears to be an epigenetic marker for both subtelomeric and central var gene silencing (18). The inheritance of this unmethylated pattern in progeny parasites could provide a marker to trace switching events and provides yet another clue leading to the gene-specific control mechanisms.
Observation of parasite var gene switching in vitro has shown that different var genes switch on and off at different rates, resulting in an apparent overall hierarchy that remains to be defined. Switch rates are intrinsic to each var gene sequence and presumably to its associated noncoding sequence. Although Horrocks and colleagues found no correlation between switch rates and 5′ var promoter type (43), the study was done in the IT4 isolate, for which there was limited sequence information at that time. Transgenic parasites that have switched all var genes off due to occupation of the single var expression site by a drug resistance marker are an exciting new tool for studying switch rate phenomena (28, 107). With this approach it has been shown that var genes have different intrinsic switch rates that correlate with chromosomal location and that var genes located in central chromosomal regions had extremely stable expression patterns but that subtelomerically located var genes readily switched to alternative var loci (33).
While the function of PfEMP1 proteins is best understood during the asexual cycle in erythrocytes, var gene expression is not limited to this stage of parasite development. PfEMP1 proteins are also expressed by early stage gametocyte-infected erythrocytes (41, 91), the transmissible form of the parasite that is infective to mosquitoes. PfEMP1 tryptic fragments were also identified in mosquito salivary gland sporozoites (30). Therefore, PfEMP1 proteins may have multistage functions, including a role in cytoadhesion and transmission of gametocyte-infected erythrocytes and potentially other stages.
In summary, mutually exclusive expression of var genes appears to involve multiple layers of control, potentially including nuclear repositioning, histone modifications, and interaction between the var 5′ promoter and var intron promoter (Fig. (Fig.5).5). These and other as-yet-unidentified processes may be involved in packaging of chromatin into functional domains with defined boundaries, the regulation of which may be the key to control of var gene expression. Additional mechanisms are required to explain switching of var gene expression in progeny cells and stage specificity. The relative influence of various control mechanisms may be different for specific var groups. Thus, the conserved organization of var genes within the parasite genome may be, to some extent, maintained by requirements for tightly regulated var gene expression.
PfEMP1 proteins have a central role in P. falciparum immune evasion and pathogenesis. While considered strong candidates for disease intervention, we are only beginning to glimpse the complex relationship between PfEMP1 expression, disease, and immune acquisition. The best-understood disease syndrome is pregnancy malaria, which has been associated with a uniquely conserved PfEMP1 variant, termed VAR2CSA (81, 86). Whether other severe malaria syndromes are similarly restricted in PfEMP1 variants or antigenic diversity of protective epitopes is the subject of intense investigation. Recent progress in understanding var gene organization suggests that gene recombination hierarchies may have an important role in shaping the antigenic and functional properties of PfEMP1 variants. Further information on both the acquisition of immunity to PfEMP1 variants and PfEMP1 virulence phenotypes may lead to potential disease interventions.
Published ahead of print on 20 July 2007.