Abstract

BackgroundVariant surface antigens (VSA) on Plasmodium falciparum–infected erythrocytes are potentially important targets of immunity to malaria. We previously identified a VSA phenotype—VSA with a high frequency of antibody recognition (VSAFoRH)—that is associated with young host age and severe malaria. We hypothesized that VSAFoRH are positively selected by host molecules such as intercellular adhesion molecule 1 (ICAM1) and CD36 and dominate in the absence of an effective immune response. Here, we assessed, in 115 Kenyan children, the potential role played by in vivo selection pressures in either favoring or selecting against VSAFoRH among parasites that cause malaria

MethodsWe tested for associations between VSAFoRH and (1) the repertoire of VSA antibodies carried by children at the time of acute malaria and (2) polymorphisms in ICAM1 (K29M) and CD36 (T188G) that could potentially reduce the positive selection of VSAFoRH

ResultsAn expected negative association between VSA antibody repertoire and VSAFoRH was observed in children with nonsevere malaria. However, this association did not extend to children with severe malaria, many of whom apparently had well-developed VSA antibody responses despite being infected by parasites expressing VSAFoRH. There was no evidence for involvement of CD36 or ICAM1 in positive selection of VSAFoRH. On the contrary, a weak positive association between carriage of the CD36 (T188G) allele and VSAFoRH was observed in children with severe malaria

ConclusionThe association between the VSAFoRH parasite phenotype and severe malaria cannot be explained simply in terms of the total repertoire of VSA antibodies carried at the time of acute disease

In sub-Saharan Africa, Plasmodium falciparum infection causes considerable mortality. Because naturally acquired immunity is critical to defense against the pathogen, the burden of disease falls mainly on young children. Despite receiving considerable attention over the last 10 years, naturally acquired immunity is still poorly understood. Even less is understood about why, despite similar levels of exposure to infectious mosquito bites, some children die of severe malaria and others do not succumb to the life-threatening disease [1]. There is a need to identify immune responses associated with protection from malaria

The variant surface antigens (VSA) expressed on the surfaces of P. falciparum–infected erythrocytes are candidate targets for naturally acquired immunity [2, 3]. Several recent reports have supported the idea that naturally acquired immunity develops through the piecemeal acquisition of a repertoire of specific VSA antibody responses and that clinical episodes of malaria correspond to gaps in this developing repertoire [4, 5]

The only well-characterized VSA, the P. falciparum erythrocyte membrane protein 1 (PfEMP1) family, is encoded by ∼60 var genes per parasite genome [6]. This family of proteins appears to play a role in the host-parasite interaction. First, the PfEMP1 variants are made up of combinations of different domains, each of which mediate a specific range of interactions with various host cell molecules, such as CD36, intercellular adhesion molecule 1 (ICAM1), and complement receptor 1 [7–11 ]. Second, the PfEMP1 variants are antigenically diverse and, through a poorly understood mechanism of gene regulation, they undergo clonal antigenic variation [12]

Because of the clear potential for a relationship between cytoadherence and the pathogenesis of malaria [13, 14], relatively little attention has been paid to variation in the antigenic properties of infected erythrocytes. However, recent studies have shown that different parasite isolates are recognized by antibodies carried by P. falciparum–exposed children at different frequencies and that this frequency of recognition (FoRVSA) is dependent on the immune status of the infected host, being associated both negatively with host age and positively with disease severity [15–18 ]. It has been proposed that this is due to the existence of a restricted subset of VSA that are maintained in the parasite population because of their optimal cytoadherence characteristics [15, 19]. Here, we will call this putative subset “VSAFoRH” (VSA with a high frequency of recognition), and VSA with a low frequency of recognition we will call “VSAFoRL.” FoRVSA has elsewhere been termed “agglutination frequency” [16], and VSAFoRH have elsewhere been termed “common” [15] and “VSA associated with severe malaria (VSASM)” [20]

Because VSAFoRH are associated with low host immunity, they could prove to be useful as markers by which protective components of the host immune response can be identified. The first step is the identification of host responses that are negatively associated with VSAFoRH expression. Because existing data suggest that antibodies impose negative selection pressure on the parasite VSA that they recognize [4], the main aim of the present study was to test the hypothesis that the repertoire of VSA antibodies present in the host at the time of acute malaria is the main determinant of the FoRVSA of a parasite. This would provide a simple underlying explanation for the variation in FoRVSA observed among isolates from both children of different ages and children with malaria of different levels of severity

Two potential in vivo selection pressures that could maintain VSAFoRH in the parasite population were also considered. A genetic approach was used to assess the potential role of 2 important PfEMP1-binding host molecules, CD36 and ICAM1. Polymorphisms in the genes for both molecules are common in African populations, and both potentially alter the concentration of VSAFoRH-binding sites in vivo. In the case of CD36 the T188G polymorphism leads to the introduction of a nonsense codon and results in CD36 deficiency in the homozygous state [21, 22]. In the case of ICAM1 the K29M polymorphism alters the cytoadherence properties of the molecule [23, 24]. We therefore tested whether these polymorphisms are associated with reduced FoRVSA

Subjects, Materials, and Methods

Study siteThe present study was conducted at Kilifi District Hospital on the Kenyan coast. The hospital has an outpatient department, a pediatric ward, and a high-dependency ward that is equipped to treat patients with severe illness

Study population and sample collectionThe present study extends a previous analysis of a group of 115 children with malaria who received treatment at the hospital between January 1997 and March 1999 [16]. Children who received a primary diagnosis of malaria were divided into 3 categories. Those who visited the outpatient department but were not admitted to the hospital were defined as children with nonsevere mild malaria (n=24; median age, 36 months [interquartile range {IQR}, 14–50 months]); those who were admitted to the pediatric ward were defined as children with nonsevere moderate malaria (n = 49; median age, 14 months [IQR, 9–34 months]); and those who were prostrated or had signs of respiratory distress were admitted to the high-dependency ward and were defined as children with severe malaria (n=42; median age, 28 months [IQR, 15–49 months]) [25]

Children were recruited into the study if they received a primary diagnosis of malaria and had ⩾1 trophozoite/100 erythrocytes (1%) [16]. Isolates were processed and selected for inclusion as described elsewhere [15]. For each isolate, a sample of acute plasma was stored at −20°C. Informed consent was obtained from the parents or guardians of the study children, and the Kenya Medical Research Institute guidelines for the conduct of clinical research were followed

Characterization of parasite isolatesParasite isolates were tested for FoRVSA by an agglutination assay against a panel of 15 heterologous blood group A plasma samples from Kenyan children, as described elsewhere [16]. The plasma samples are referred to by the letters ao

Characterization of plasma samplesPlasma samples from the 115 children were tested for their VSA antibody repertoires by the agglutination assay against a panel of 6 well-recognized heterologous parasite isolates (isolates 4513, 4518, 1759, 4542, 4508, and 4528) from children with malaria who had received treatment at the hospital either between January and August 2000 or, in the case of isolate 1759, in December 1995. Methods for the agglutination assay have been described elsewhere [16]

Allele carriageBecause of a shortage of DNA for 3 of the samples, only 112 children were tested for the ICAM1 (K29M) allele, and only 114 children were tested for the CD36 (T188G) allele. ICAM1 was amplified by polymerase chain reaction with the primers described by Bellamy et al. [26]. The 127-bp-long product was digested with NlaIII (New England Biolabs) under the conditions recommended by the manufacturer. The K29 products were 50 bp and 80 bp, and the 29M products were 50 bp, 25 bp, and 55 bp. The CD36 (T188G) allele was detected as described elsewhere [21]

Data analysisThe FoRVSA of a parasite isolate was quantified as the number of samples from the above-described panel of 15 heterologous plasma samples that scored positive against that isolate by the agglutination assay. The host VSA antibody repertoire was quantified as the number of parasites from the above-described panel of 6 well-recognized heterologous parasites that were agglutinated by a plasma sample. Parasite data were stored and analyzed by use of Stata (version 5.0; Stata Corporation). Univariate analysis was conducted with nonparametric tests—either the Mann-Whitney U test (for the comparison of 2 populations) or Spearman’s&amp;rank correlation coefficient (rs). To correct for age as an independent variable and to assess the role played by statistical interaction, logistic regression was used. For the logistic regression analysis, the parasite isolates that expressed VSAFoRH were defined as those that were recognized by ⩾2 of the 15 heterologous plasma samples, and the parasite isolates that expressed VSAFoRL were defined as those that were recognized by <2 of the 15 heterologous plasma samples. Children were considered to be antibody positive if their plasma sample recognized at least 1 of the 6 heterologous parasite isolates. Statistical interaction was tested by comparing logistic regression models with and without the interaction variable by a likelihood ratio test. Age was recorded for only 112 of the 115 children

Results

Relationship between VSA antibodies and VSAFoRHAs previously reported, among the 115 children included in the present study, there was a negative association between FoRVSA and host age and a positive association between FoRVSA and disease severity [16]. We first tested the simple hypothesis that parasite FoRVSA is determined primarily by the repertoire of host VSA antibodies carried at the time of acute disease. As described above, we used the number of parasite isolates recognized by a plasma sample obtained from each child at the time of acute malaria as a semiquantitative measure of his or her VSA antibody repertoire. Because, by definition, VSAFoRH are the most likely VSA to be recognized by a plasma sample from any child, the number of parasite isolates recognized should also be a good measure of his or her VSAFoRH-specific antibody repertoire. We therefore tested a plasma sample from each of the 115 children for its ability to agglutinate a panel of 6 heterologous parasite isolates

We tested for a negative association between parasite FoRVSA and host VSA antibody repertoire at the time of acute malaria. Inconsistent with our simple hypothesis, there was no evidence for such an association (rs=-0.08; P=.38). To test whether this was universally the case, we divided the children into those with severe malaria and those with nonsevere malaria. In the children with severe malaria, there again was no evidence for an association between parasite FoRVSA and VSA antibody repertoire (rs=0.16; P=.31) (figure 1B and 1D and figure 2B). However, a negative association was observed in the children with nonsevere malaria (rs=-0.29; P=.014) (figure 1A and 1C and figure 2A). This association was independent of host age and whether the children had nonsevere mild or nonsevere moderate malaria (odds ratios for being infected with parasites expressing VSAFoRH if antibody positive, 0.32 [95% confidence interval {CI}, 0.11–0.91] [P=.032] and 0.27 [95% CI, 0.08–0.89] [P=.032], with and without correction, respectively; logistic regression analysis)

Figure 1

Agglutination profiles of 115 parasite isolates in relation to the reactivity profiles of the host plasma. One hundred and fifteen children with malaria are listed vertically, with their identification numbers indicated. Parasites from these children were tested against a panel of 15 heterologous blood group A plasma samples (denoted ao), which are shown in the first 15 columns and are ordered from left to right with respect to increasing overall reactivity [16]. Positive agglutination of the parasites by the plasma samples from the study children is indicated by hatched boxes. Plasma from each child was subsequently tested against 6 well-recognized heterologous parasite isolates (isolates 4513, 4518, 1759, 4542, 4508, and 4528). The presence of heterologous antibodies to members of this panel is indicated by black boxes. The children are divided into those with nonsevere malaria (panels A and C) and those with severe malaria (panels B and D) as well as into those who were antibody positive (panels A and B) and those who were antibody negative (panels C and D). Within each subgroup, the children are ordered from top to bottom with respect to decreasing overall frequency of recognition of the variant surface antigens (FoRVSA) expressed by the parasites that infected them

Figure 1

Agglutination profiles of 115 parasite isolates in relation to the reactivity profiles of the host plasma. One hundred and fifteen children with malaria are listed vertically, with their identification numbers indicated. Parasites from these children were tested against a panel of 15 heterologous blood group A plasma samples (denoted ao), which are shown in the first 15 columns and are ordered from left to right with respect to increasing overall reactivity [16]. Positive agglutination of the parasites by the plasma samples from the study children is indicated by hatched boxes. Plasma from each child was subsequently tested against 6 well-recognized heterologous parasite isolates (isolates 4513, 4518, 1759, 4542, 4508, and 4528). The presence of heterologous antibodies to members of this panel is indicated by black boxes. The children are divided into those with nonsevere malaria (panels A and C) and those with severe malaria (panels B and D) as well as into those who were antibody positive (panels A and B) and those who were antibody negative (panels C and D). Within each subgroup, the children are ordered from top to bottom with respect to decreasing overall frequency of recognition of the variant surface antigens (FoRVSA) expressed by the parasites that infected them

Figure 2

Host selection for and against variant surface antigens (VSA). A and B Relationship between host VSA antibody repertoire at the time of acute malaria and the frequency of recognition of the VSA (FoRVSA) expressed by parasites. Children with nonsevere malaria and those with severe malaria are shown in panels A and B, respectively. The host VSA antibody repertoire was quantified as the number of parasites from a panel of 6 well-recognized heterologous parasites that were agglutinated by a plasma sample, and the FoRVSA of a parasite isolate was quantified as the number of samples from a panel of 15 heterologous plasma samples that scored positive against that isolate by an agglutination assay. Spearman’s&amp;rank correlation coefficients (rs) are indicated, as are linear regression lines. For comparison, panels C and D show the relationship between parasite FoRVSA and host age. For these analyses, the Spearman’s&amp;rank correlation coefficients and the linear regression lines were calculated only with data from children >5 months old, to negate the effect of maternal antibodies. In panels A and C, children with mild nonsevere malaria are indicated by black circles, and children with moderate nonsevere malaria are indicated by white circles. E–H Relationships between host age, VSA antibody repertoire, and parasite FoRVSA, in children with nonsevere malaria and those with severe malaria. Panels F and H show only children infected with parasites expressing VSA with a high frequency of recognition (VSAFoRH; those recognized by ⩾2 of the 15 heterologous plasma samples), and panels E and G show only children infected with parasites expressing VSA with a low frequency of recognition (VSAFoRL; those recognized by <2 of the 15 heterologous plasma samples). I Parasite FoRVSA in relation to carriage of the CD36 (T188G) allele by children with nonsevere malaria and those with severe malaria. Children homozygous for the wild-type (wt) CD36 (T188) allele are indicated by white circles; heterozygous children are indicated by black circles. J and K Hypothetical models of within-host selection pressures on parasite VSA. On each side of the figure, the entire circle represents all of the VSA expressed by the infecting parasite population; the shaded inner portion represents a putative subset of VSA (VSAFoRH), and the unshaded outer portion represents VSAFoRL. The arrows pointing to the right and left represent opposing within-host selection pressures that favor VSAFoRH and VSAFoRL, respectively. In the basic model tested, shown in panel J, VSAFoRH are maintained in the parasite population by the transmission advantage they confer to the parasite, the nature of which is unknown. Because of their restricted diversity, VSAFoRH are preferentially selected against by the developing VSA antibody response. Panel K shows a modified working model. CD36 may play a role in selection against VSAFoRH. During severe malaria, VSAFoRH are under strong selection pressure that overrides the effect of VSA antibodies. Alternatively, antibodies (dashed line) are not of the correct specificity or subclass

Figure 2

Host selection for and against variant surface antigens (VSA). A and B Relationship between host VSA antibody repertoire at the time of acute malaria and the frequency of recognition of the VSA (FoRVSA) expressed by parasites. Children with nonsevere malaria and those with severe malaria are shown in panels A and B, respectively. The host VSA antibody repertoire was quantified as the number of parasites from a panel of 6 well-recognized heterologous parasites that were agglutinated by a plasma sample, and the FoRVSA of a parasite isolate was quantified as the number of samples from a panel of 15 heterologous plasma samples that scored positive against that isolate by an agglutination assay. Spearman’s&amp;rank correlation coefficients (rs) are indicated, as are linear regression lines. For comparison, panels C and D show the relationship between parasite FoRVSA and host age. For these analyses, the Spearman’s&amp;rank correlation coefficients and the linear regression lines were calculated only with data from children >5 months old, to negate the effect of maternal antibodies. In panels A and C, children with mild nonsevere malaria are indicated by black circles, and children with moderate nonsevere malaria are indicated by white circles. E–H Relationships between host age, VSA antibody repertoire, and parasite FoRVSA, in children with nonsevere malaria and those with severe malaria. Panels F and H show only children infected with parasites expressing VSA with a high frequency of recognition (VSAFoRH; those recognized by ⩾2 of the 15 heterologous plasma samples), and panels E and G show only children infected with parasites expressing VSA with a low frequency of recognition (VSAFoRL; those recognized by <2 of the 15 heterologous plasma samples). I Parasite FoRVSA in relation to carriage of the CD36 (T188G) allele by children with nonsevere malaria and those with severe malaria. Children homozygous for the wild-type (wt) CD36 (T188) allele are indicated by white circles; heterozygous children are indicated by black circles. J and K Hypothetical models of within-host selection pressures on parasite VSA. On each side of the figure, the entire circle represents all of the VSA expressed by the infecting parasite population; the shaded inner portion represents a putative subset of VSA (VSAFoRH), and the unshaded outer portion represents VSAFoRL. The arrows pointing to the right and left represent opposing within-host selection pressures that favor VSAFoRH and VSAFoRL, respectively. In the basic model tested, shown in panel J, VSAFoRH are maintained in the parasite population by the transmission advantage they confer to the parasite, the nature of which is unknown. Because of their restricted diversity, VSAFoRH are preferentially selected against by the developing VSA antibody response. Panel K shows a modified working model. CD36 may play a role in selection against VSAFoRH. During severe malaria, VSAFoRH are under strong selection pressure that overrides the effect of VSA antibodies. Alternatively, antibodies (dashed line) are not of the correct specificity or subclass

One interpretation of this finding is that children with severe malaria and those with nonsevere malaria differ in the degree to which VSA antibodies carried at the time of acute malaria can control parasites expressing VSAFoRH. Consistent with this interpretation, disease severity statistically modified the association between VSA antibody repertoire and FoRVSA (age-corrected P=.020, logistic regression analysis). Also, there was no evidence for a trend toward reduced VSA antibody repertoire in children as disease severity increased (for the comparison of children with nonsevere mild, nonsevere moderate, and severe malaria, rs=0.06 and P=.56), despite a trend toward higher FoRVSA (rs=0.28; P=.003)

Figure 2 shows this in more detail. Figure 2A and 2B contrast the relationship between FoRVSA and VSA antibody repertoire in the children with nonsevere malaria and the children with severe malaria, respectively. Although VSA antibody repertoire appears to be a poor predictor of FoRVSA in the children with severe malaria, there was, nonetheless, a negative relationship between FoRVSA and host age in both the severe and nonsevere malaria groups, suggesting that an age-dependent selection pressure exists in both (figure 2C and 2D). Figure 2E–2H explores the relationships between host age, VSA antibody repertoire, and FoRVSA. The children are divided into those infected with parasites expressing VSAFoRH (figure 2F and 2H) and those infected with parasites expressing VSAFoRL (figure 2E and 2G), and the distribution of these 2 parasites phenotypes is shown with respect to both host age (with a cutoff of 2 years) and the repertoire of VSA antibodies at the time of acute malaria. Among the children with nonsevere malaria, parasites expressing VSAFoRL were well distributed between the 2 age groups and between those with a low and those with a high VSA antibody repertoire at the time of acute malaria, suggesting a lack of restriction with respect to whom these parasites can infect (figure 2E). In contrast, with only 1 exception, the parasites expressing VSAFoRH infected only children <2 years old or those who were antibody negative, suggesting that these parasites can cause clinical infection only in immunologically naive children (figure 2F). Among the children with severe malaria, the distribution of parasites expressing VSAFoRL was not strikingly different from that observed among the children with nonsevere malaria (figure 2G). However, in contrast to what is shown in figure 2F the parasites expressing VSAFoRH did not exhibit any indication of being restricted to immunologically naive children (i.e., either the children <2 years old or those who were antibody negative) (figure 2H)

Relationship between parasite VSAFoRH and polymorphisms in the genes for host receptor moleculesTo test whether host ICAM1 or CD36 play a role in the selection of parasites expressing VSAFoRH, we tested for associations between parasite FoRVSA and (1) host carriage of 0, 1, or 2 copies of the ICAM1 (K29M) allele and (2) host carriage of a copy of the CD36 (T188G) allele. There was no evidence for an association between parasite FoRVSA and host carriage of the ICAM1 (K29M) allele, whether we considered the children with severe malaria (rs=-0.08; P=.63), the children with nonsevere malaria (rs = 0.21; P=.07), or all of the children together (rs = 0.12; P=.20). Similarly, there was no evidence for an association between parasite FoRVSA and carriage of the CD36 (T188G) allele, both in the children with nonsevere malaria (P=.56) and when all of the children were considered together (P=.16, Mann-Whitney U test); however, a weak positive association was evident in the children with severe malaria (P = .039, Mann-Whitney U test) (figure 2I), although this association did not remain significant after correction for multiple comparisons (P>.05). This raises the possibility that CD36 plays a role in selection against VSAFoRH

Discussion

To survive the dynamic environment encountered when P. falciparum infects the human host, the parasite expresses VSA on the surfaces of infected erythrocytes that can switch among a large repertoire of different antigenic and adhesive phenotypes [12, 27]. Dissecting the various causal pathways leading to the expression of different VSA phenotypes in different children may provide important information about why, despite similar levels of exposure to infectious mosquito bites, some children die of severe malaria and others do not succumb to the life-threatening disease. A first step is the identification of VSA phenotypes that are associated with measurable features of the within-host environment. VSAFoRH is one such phenotype

The basic model we set out to test in the present study is summarized in figure 2J. We expected that children carrying a well-developed repertoire of VSA antibodies at the time of acute malaria would tend to not be infected by parasites expressing VSAFoRH. When we tested for a negative association between the FoRVSA of the 115 parasite isolates and VSA antibody repertoire at the time of acute malaria, the expected association was observed in the children with nonsevere malaria but not in the children with severe malaria. In addition, there was no overall difference in VSA antibody repertoire between the children with severe and those with nonsevere malaria. These observations are consistent with the results of previous studies. A recent study of FoRVSA in 5 children with cerebral malaria and 4 children with noncerebral malaria also demonstrated the simultaneous presence of a well-developed VSA antibody repertoire in a child with cerebral malaria who was infected with well-recognized parasites [18]. In the Gambia, recognition of heterologous parasite VSA at the time of acute malaria was found to not differ between children with severe cerebral malaria and those with mild malaria [28]. A recent longitudinal study in Kenya found that recognition of a panel of heterologous VSA was not associated with protection from subsequent severe malaria [29]

Two plausible explanations could explain why detectable VSA antibodies would be unable to control parasites expressing VSAFoRH in children with severe malaria (figure 2K): (1) VSAFoRH are under particularly strong positive selection and (2) these children, despite carrying VSA antibodies, do not recognize all VSAFoRH variants and remain vulnerable to infection with parasites that express them, whether because of a chance lack of exposure or a specific inability to develop responses or because the antibody response is not of sufficient quality or is dominated by the wrong subclass. In support of the latter explanation, both a small study in Kenya and a study in Gabon have suggested that children who are susceptible to severe malaria may have altered antibody responses to infection [29, 30]

The former explanation is difficult to test in the absence of any information on the function of VSAFoRH and the types of selection pressures that maintain well-recognized VSA in the parasite population. One possibility is that they are maintained because of their optimal cytoadherence characteristics with respect to major PfEMP1-binding molecules, such as ICAM1 and CD36. To test this possibility, we used the fact that polymorphisms in the genes for these 2 molecules are prevalent in African populations [21, 31]. We reasoned that, if these host molecules are important in selection for VSAFoRH, then (1) any reduction in the normal binding-site concentration in vivo would reduce the overall FoRVSA of the infecting parasite population and (2) this effect would be amplified in children with severe malaria if increased positive selection by these molecules is responsible for the breakdown in VSA antibody control of VSAFoRH. There was no evidence for an association between the ICAM1 (K29M) allele and parasite FoRVSA. However, in the children with severe malaria, a weak positive association was observed between the CD36 (T188G) allele and parasite FoRVSA. This finding is inconsistent with the idea that CD36 is involved in the positive selection of VSAFoRH, but it raises the possibility that CD36 is normally involved in the negative selection of VSAFoRH. Because of low numbers, the association was not significant after correction for multiple comparisons. This finding, therefore, needs to be confirmed in larger studies

Parasite binding to CD36 is mediated by the cysteine-rich interdomain region of PfEMP1. Recent analysis of the genome sequence of the P. falciparum laboratory isolate 3D7 shows that the var genes encoding PfEMP1 fall into distinct subfamilies [32–34 ]. The genes can be crudely divided into a restricted class of genes encoding long PfEMP1 molecules that do not bind to CD36 and a much broader class of genes encoding short PfEMP1 molecules that do bind to CD36 [35]. The following hypothetical working model would, therefore, be consistent with the above observations: VSAFoRH-specific antibodies and CD36 selection pressure both favor parasites that express VSAFoRL, which correspond with members of the class of CD36-binding PfEMP1 molecules; in contrast, a lack of effective VSAFoRH antibody responses and reduced CD36 selection pressure both favor parasites that express VSAFoRH, which correspond with members of the class of non–CD36-binding PfEMP1 molecules. Clearly, the host selection pressures that favor VSAFoRH in the absence of these selection pressures remain to be determined. Several lines of evidence support this model. First, parasites from children with severe malaria in China tend to express PfEMP1 molecules that are longer when resolved by SDS-PAGE [36]. Second, the Duffy binding–like–α domains of PfEMP1 molecules from children with severe malaria in Columbia have sequence characteristics similar to those of the non–CD36-binding PfEMP1 molecules of 3D7 [37]. Finally, parasites that were isolated from semi-immune children and that were selected in vitro for binding to IgG tended to express VSAFoRH and to both exhibit reduced CD36 binding and a bias toward the restricted class of non–CD36-binding PfEMP1 molecules [20, 38]. In future studies, it will be of great interest to determine whether differential expression of these 2 classes of var genes could account for the observed variation in FoRVSA among parasites isolated from hosts with different levels of immunity. However, the results of the present study emphasize the need, in the first instance, for a more detailed characterization of both the VSA antibody response and the interactions that take place between VSA and host cells during severe malaria

Acknowledgments

We thank Norbert Peshu, director of the Kenya Medical Research Institute (KEMRI) Centre for Geographic Medicine Research, Coast, Kilifi, together with the unit staff, parents, and children who were involved in the present study; Britta Urban, Sue Kyes, and Paul Horrocks, for critical comments on the manuscript; and Greg Fegan, for statistical advice

References

1.
Marsh
K
Malaria—a neglected disease?
Parasitology
 , 
1992
, vol. 
104
 
Suppl
(pg. 
S53
-
69
)
2.
Marsh
K
Otoo
L
Hayes
RJ
Carson
DC
Greenwood
BM
Antibodies to blood stage antigens of Plasmodium falciparum in rural gambians and their relation to protection against infection
Trans R Soc Trop Med Hyg
 , 
1989
, vol. 
83
 (pg. 
293
-
303
)
3.
Dodoo
D
Staalsoe
T
Giha
H
, et al.  . 
Antibodies to variant antigens on the surfaces of infected erythrocytes are associated with protection from malaria in Ghanaian children
Infect Immun
 , 
2001
, vol. 
69
 (pg. 
3713
-
8
)
4.
Bull
PC
Lowe
BS
Kortok
M
Molyneux
CS
Newbold
CI
Marsh
K
Parasite antigens on the infected red cell are targets for naturally acquired immunity to malaria
Nature Medicine
 , 
1998
, vol. 
4
 (pg. 
358
-
60
)
5.
Giha
HA
Staalsoe
T
Dodoo
D
, et al.  . 
Antibodies to variable Plasmodium falciparum‐infected erythrocyte surface antigens are associated with protection from novel malaria infections
Immunol Lett
 , 
2000
, vol. 
71
 (pg. 
117
-
26
)
6.
Gardner
MJ
Hall
N
Fung
E
, et al.  . 
Genome sequence of the human malaria parasite Plasmodium falciparum
Nature
 , 
2002
, vol. 
419
 (pg. 
498
-
511
)
7.
Udeinya
IJ
Schmidt
JA
Aikawa
M
Miller
LH
Green
I
Falciparum malaria‐infected erythrocytes specifically bind to cultured human endothelial cells
Science
 , 
1981
, vol. 
213
 (pg. 
555
-
7
)
8.
Pain
A
Ferguson
DJ
Kai
O
, et al.  . 
Platelet‐mediated clumping of Plasmodium falciparum‐infected erythrocytes is a common adhesive phenotype and is associated with severe malaria
Proc Natl Acad Sci USA
 , 
2001
, vol. 
98
 (pg. 
1805
-
10
)
9.
Udomsangpetch
R
Wahlin
B
Carlson
J
, et al.  . 
Plasmodium falciparum‐infected erythrocytes form spontaneous erythrocyte rosettes
J Exp Med
 , 
1989
, vol. 
169
 (pg. 
1835
-
40
)
10.
Rowe
JA
Moulds
JM
Newbold
CI
Miller
LH
Plasmodium falciparum rosetting is mediated by a parasite‐variant erythrocyte membrane protein and complement‐receptor 1
Nature
 , 
1997
, vol. 
388
 (pg. 
292
-
5
)
11.
Urban
BC
Ferguson
DJ
Pain
A
, et al.  . 
Plasmodium falciparum‐infected erythrocytes modulate the maturation of dendritic cells
Nature
 , 
1999
, vol. 
400
 (pg. 
73
-
7
)
12.
Roberts
DJ
Craig
AG
Berendt
AR
, et al.  . 
Rapid switching to multiple antigenic and adhesive phenotypes in malaria
Nature
 , 
1992
, vol. 
357
 (pg. 
689
-
92
)
13.
MacPherson
GG
Warrell
MJ
White
NJ
Looareesuwan
W
Warrell
DA
Human cerebral malaria: a quantitative ultrastructural analysis of parasitized erythrocyte sequestration
Am J Pathol
 , 
1985
, vol. 
119
 (pg. 
385
-
401
)
14.
Pongponratn
E
Turner
GD
Day
NP
, et al.  . 
An ultrastructural study of the brain in fatal Plasmodium falciparum malaria
Am J Trop Med Hyg
 , 
2003
, vol. 
69
 (pg. 
345
-
59
)
15.
Bull
PC
Lowe
BS
Kortok
M
Marsh
K
Antibody recognition of Plasmodium falciparum erythrocyte surface antigens in Kenya: evidence for rare and prevalent variants
Infect Immun
 , 
1999
, vol. 
67
 (pg. 
733
-
9
)
16.
Bull
PC
Kortok
M
Kai
O
, et al.  . 
Plasmodium falciparum–infected erythrocytes: agglutination by diverse Kenyan plasma is associated with severe disease and young host age
J Infect Dis
 , 
2000
, vol. 
182
 (pg. 
252
-
9
)
17.
Nielsen
MA
Staalsoe
T
Kurtzhals
JA
, et al.  . 
Plasmodium falciparum variant surface antigen expression varies between isolates causing severe and nonsevere malaria and is modified by acquired immunity
J Immunol
 , 
2002
, vol. 
168
 (pg. 
3444
-
50
)
18.
Lindenthal
C
Kremsner
PG
Klinkert
MQ
Commonly recognised Plasmodium falciparum parasites cause cerebral malaria
Parasitol Res
 , 
2003
, vol. 
91
 (pg. 
363
-
8
)
19.
Hviid
L
Staalsoe
T
Malaria immunity in infants: a special case of a general phenomenon?
Trends Parasitol
 , 
2004
, vol. 
20
 (pg. 
66
-
72
)
20.
Staalsoe
T
Nielsen
MA
Vestergaard
LS
Jensen
AT
Theander
TG
Hviid
L
In vitro selection of Plasmodium falciparum 3D7 for expression of variant surface antigens associated with severe malaria in African children
Parasite Immunol
 , 
2003
, vol. 
25
 (pg. 
421
-
7
)
21.
Pain
A
Urban
BC
Kai
O
, et al.  . 
A non‐sense mutation in CD36 gene is associated with protection from severe malaria
Lancet
 , 
2001
, vol. 
357
 (pg. 
1502
-
3
)
22.
Aitman
TJ
Cooper
LD
Norsworthy
PJ
, et al.  . 
Malaria susceptibility and CD36 mutation
Nature
 , 
2000
, vol. 
405
 (pg. 
1015
-
6
)
23.
Adams
S
Turner
GD
Nash
GB
Micklem
K
Newbold
CI
Craig
AG
Differential binding of clonal variants of Plasmodium falciparum to allelic forms of intracellular adhesion molecule 1 determined by flow adhesion assay
Infect Immun
 , 
2000
, vol. 
68
 (pg. 
264
-
9
)
24.
Craig
A
Fernandez‐Reyes
D
Mesri
M
, et al.  . 
A functional analysis of a natural variant of intercellular adhesion molecule‐1 (ICAM‐1Kilifi)
Hum Mol Genet
 , 
2000
, vol. 
9
 (pg. 
525
-
30
)
25.
Marsh
K
Forster
D
Waruiru
C
, et al.  . 
Indicators of life‐threatening malaria in African children
N Engl J Med
 , 
1995
, vol. 
332
 (pg. 
1399
-
404
)
26.
Bellamy
R
Kwiatkowski
D
Hill
AV
Absence of an association between intercellular adhesion molecule 1, complement receptor 1 and interleukin 1 receptor antagonist gene polymorphisms and severe malaria in a West African population
Trans R Soc Trop Med Hyg
 , 
1998
, vol. 
92
 (pg. 
312
-
6
)
27.
Smith
JD
Chitnis
CE
Craig
AG
, et al.  . 
Switches in expression of Plasmodium falciparum var genes correlate with changes in antigenic and cytoadherent phenotypes of infected eryhtrocytes
Cell
 , 
1995
, vol. 
82
 (pg. 
101
-
10
)
28.
Erunkulu
OA
Hill
AVS
Kwiatowski
DP
, et al.  . 
Severe malaria in Gambian children is not due to lack of previous exposure to malaria
Clin Exp Immunol
 , 
1992
, vol. 
89
 (pg. 
296
-
300
)
29.
Bull
PC
Lowe
BS
Kaleli
N
, et al.  . 
Plasmodium falciparum infections are associated with agglutinating antibodies to parasite‐infected erythrocyte surface antigens among healthy Kenyan children
J Infect Dis
 , 
2002
, vol. 
185
 (pg. 
1688
-
91
)
30.
Tebo
AE
Kremsner
PG
Piper
KP
Luty
AJ
Low antibody responses to variant surface antigens of Plasmodium falciparum are associated with severe malaria and increased susceptibility to malaria attacks in Gabonese children
Am J Trop Med Hyg
 , 
2002
, vol. 
67
 (pg. 
597
-
603
)
31.
Fernandez‐Reyes
DC
Craig
AG
Kyes
SA
, et al.  . 
A high frequency African coding polymorphism in the N‐terminal domain of ICAM‐1 predisposing to cerebral malaria in Kenya
Hum Mol Genet
 , 
1997
, vol. 
6
 (pg. 
1357
-
60
)
32.
Voss
TS
Kaestli
M
Vogel
D
Bopp
S
Beck
HP
Identification of nuclear proteins that interact differentially with Plasmodium falciparum var gene promoters
Mol Microbiol
 , 
2003
, vol. 
48
 (pg. 
1593
-
607
)
33.
Lavstsen
T
Salanti
A
Jensen
AT
Arnot
DE
Theander
TG
Sub‐grouping of Plasmodium falciparum 3D7 var genes based on sequence analysis of coding and non‐coding regions
Malar J
 , 
2003
, vol. 
2
 pg. 
27
 
34.
Kraemer
SM
Smith
JD
Evidence for the importance of genetic structuring to the structural and functional specialization of the P. falciparum var gene family
Mol Microbiol
 , 
2003
, vol. 
50
 (pg. 
1527
-
38
)
35.
Robinson
BA
Welch
TL
Smith
JD
Widespread functional specialization of Plasmodium falciparum erythrocyte membrane protein 1 family members to bind CD36 analysed across a parasite genome
Mol Microbiol
 , 
2003
, vol. 
47
 (pg. 
1265
-
78
)
36.
Bian
Z
Wang
G
Antigenic variation and cytoadherence of PfEMP1 of Plasmodium falciparum‐infected erythrocyte from malaria patients
Chin Med J (Engl)
 , 
2000
, vol. 
113
 (pg. 
981
-
4
)
37.
Kirchgatter
K
Portillo Hdel
A
Association of severe noncerebral Plasmodium falciparum malaria in Brazil with expressed PfEMP1 DBL1 alpha sequences lacking cysteine residues
Mol Med
 , 
2002
, vol. 
8
 (pg. 
16
-
23
)
38.
Jensen
AT
Magistrado
P
Sharp
S
, et al.  . 
Plasmodium falciparum associated with severe childhood malaria preferentially expresses PfEMP1 encoded by group A var genes
J Exp Med
 , 
2004
, vol. 
199
 (pg. 
1179
-
90
)
Potential conflicts of interest: none reported
Financial support: Wellcome Trust Advanced Training Fellowship in Tropical Medicine (grant 060678 to P.C.B.); Wellcome Trust Senior Fellowship (grant 631342 to K.M.); United Kingdom National Blood Service (D.J.R.)
This article is published with the permission of the director of the Kenya Medical Research Institute
Present affiliations: Wellcome Trust Sanger Institute, Cambridge, United Kingdom (A.P.); National Institute for Medical Research, London, United Kingdom (F.M.N. and S.M.K.)