Skip to main content
UKPMC Funders Author Manuscripts logoLink to UKPMC Funders Author Manuscripts
. Author manuscript; available in PMC: 2008 Jul 23.
Published in final edited form as: J Infect Dis. 2006 Jan 30;193(5):721–730. doi: 10.1086/500145

Antigenic Differences and Conservation among Placental Plasmodium falciparum-Infected Erythrocytes and Acquisition of Variant-Specific and Cross-Reactive Antibodies

James G Beeson 1,2, Emily J Mann 2, Timothy J Byrne 2, Aphrodite Caragounis 2, Salenna R Elliott 2, Graham V Brown 2, Stephen J Rogerson 2,3
PMCID: PMC2483301  EMSID: UKMS2099  PMID: 16453269

Abstract

Background

Pregnant women are infected by Plasmodium falciparum with novel antigenic phenotypes that adhere to chondroitin sulfate A (CSA) and other receptors in the placenta. The diverse and variant parasite protein P. falciparum erythrocyte membrane protein 1 (PfEMP1), which is encoded by var genes, is a ligand for CSA and a major target of antibodies associated with protective immunity.

Methods

Serum samples from pregnant women exposed to malaria were tested for immunoglobulin G, adhesion-inhibitory antibodies, and agglutinating antibodies to different CSA-binding isolates expressing conserved var2csa-type genes and to parasite isolates from infected placentas. Parasite isolates also were examined to assess PfEMP1 expression, the effect of trypsin treatment of infected erythrocytes on parasite adhesion and cleavage of PfEMP1, and inhibition of adhesion by rabbit antiserum raised against a CSA-binding isolate.

Results

Findings demonstrated that (1) there are significant antigenic differences between CSA-binding isolates that correspond with polymorphisms in var2csa; (2) there are differences in the properties of PfEMP1 and antibody reactivity between CSA-binding and placental isolates, which express multiple PfEMP1 forms; (3) acquired antibodies target diverse and cross-reactive epitopes expressed by CSA-binding infected erythrocytes, and cross-reactive antibodies are not necessarily cross-inhibitory; and (4) the breadth of antibody reactivity is greater among multigravidae than among primigravidae.

Conclusions

Immunity may be mediated by a repertoire of antibodies to diverse and common epitopes. Strategies based on vaccination with a single domain or isolate might be hindered by antigenic diversity.


Although immunity to malaria may exist before pregnancy, malaria is more prevalent and severe during pregnancy, especially among primigravid women. Plasmodium falciparum infection is characterized by the accumulation in the placenta of mature-stage parasite-infected erythrocytes (IEs) [1, 2], mediated through adhesion to chondroitin sulfate A (CSA) [3, 4], and other molecules, such as hyaluronic acid (HA) and nonimmune immunoglobulins [5-8]. Early in a first pregnancy, women generally lack antibodies to placental-binding IEs, which suggests that these parasites represent novel variant antigens to which women have not been exposed previously [4, 9-11]. Reduced susceptibility is observed in women who have had several pregnancies exposed to malaria, because of the acquisition of specific immunity [12]. Antibodies to surface antigens expressed by placental isolates and isolates that adhere to CSA or both CSA and HA are acquired after exposure to placental malaria [4, 9-11, 13, 14]. These antibodies generally are more prevalent in multigravidae than in primigravidae [4, 9-11], corresponding with a reduced risk of malaria during pregnancy, and, in cross-sectional studies, there is some association between these antibodies and improved pregnancy outcomes among women at delivery [15, 16].

The major target of antibodies to antigens expressed on the surface of IEs is the highly diverse Plasmodium falciparum erythrocyte membrane protein 1 (PfEMP1), which is encoded by the var multigene family [17-19]. Antigenic variation of PfEMP1, through switching expression of different var genes, facilitates evasion of host immune responses, and specific variants of PfEMP1 mediate adhesion to CSA [20, 21]. Recently, var2csa-type genes [22], which are relatively conserved between isolates, have been identified as encoding CSA-binding PfEMP1. This gene type appears to be expressed among a range of CSA-binding and placental parasite isolates [22-24], and pregnant women exposed to malaria acquire antibodies to recombinant var2csa proteins [23]. Together, these findings suggest that var2csa PfEMP1 is an important target of acquired and possibly protective antibodies.

The degree of antigenic diversity or conservation of antigens expressed by placental IEs is currently unclear. Serum samples obtained from women in different geographic regions inhibited adhesion to CSA of the same placental isolates, suggesting the expression of conserved antigens [9]. However, it is unresolved whether this finding reflected the presence of cross-reactive antibodies to conserved epitopes or, alternatively, the presence of antibodies with multiple specificities in individual serum samples. In contrast, in a separate study, agglutinating antibodies to antigens expressed on the surface of placental IEs were found to be variant specific to a significant extent [4]. In studies of children and nonpregnant adults, the presence and significance of cross-reactive antibodies to IE surface antigens have, in general, been controversial [25]. Here, we have investigated these issues by examining antigenic diversity and differences among defined CSA-binding and CSA-HA-binding isolates known to be expressing var2csa as the dominant var gene, by comparing defined CSA-binding isolates and IEs harvested from infected placentas and by examining whether acquired antibodies comprise variant-specific and/or cross-reactive antibodies to placental IEs.

SUBJECTS, MATERIALS, AND METHODS

P. falciparum culture and placental isolates

P. falciparum was cultured in medium supplemented with either 10% vol/vol pooled human serum (obtained from donors who were residents of Australia) or 5% serum and 0.25% Albumax II (Gibco), as described elsewhere [4]. P. falciparum isolates CS2, HCS3, and 3D7-CSA are genetically distinct and were generated by selection for adhesion to CSA, as described elsewhere [8]. The specificity of adhesion to CSA and HA has been established elsewhere [5, 8, 26]. Isolate E8B adheres to CD36 and intercellular adhesion molecule 1 and is isogenic to CS2 [8]. Clonality and genetic identity were determined by polymerase chain reaction analysis of msp-1 and msp-2 alleles, by use of established methods [27]. Cultures were free of Mycoplasma species, as determined by polymerase chain reaction. Placental P. falciparum isolates were recovered from pregnant women attending the labor ward of the Queen Elizabeth Central Hospital in Blantyre, Malawi [4]. Written, informed consent was obtained from all women who participated in the study. P. falciparum IEs (predominantly mature trophozoites) were harvested from freshly delivered infected placentas, as described elsewhere [4].

Parasite adhesion and inhibition assays

Adhesion assays and assays to measure the inhibition of adhesion by serum and/or plasma were performed, as described elsewhere [4, 13], using pigmented-trophozoite IEs at 3%-8% parasitemia. For inhibition, parasites were incubated with human serum at 1:5 dilution or with rabbit serum at 1:10-1:50 dilution for 45 min at 37°C, before testing for adhesion to immobilized CSA was done. Rabbit antiserum against CS2 IEs and control rabbit serum were adsorbed against uninfected red blood cells before use. Samples were coded and tested in a blinded manner.

Antibodies to the surface of P. falciparum IEs

Measurement of IgG binding to the surface of IEs was performed using flow cytometry (FACSCalibur; Becton-Dickinson) with established methods [13]. Pigmented-trophozoite IEs were tested at 3%-4% parasitemia and a hematocrit of 0.1%. Cells were sequentially incubated with test serum or plasma at 1:10-1:20 dilution, with goat or rabbit anti-human IgG (Fc-specific; Caltag or Dako) at 1:50 dilution, and with fluorescein isothiocyanate-conjugated anti-goat or anti-rabbit immunoglobulins (Caltag or Dako) at 1:50 dilution with 10 μg/mL ethidium bromide in darkness.

Agglutination assays were performed, as described and validated elsewhere [13, 28], using mature-stage IEs enriched to 60%-90% parasitemia with serum at 1:10 dilution in PBS. Samples were considered to be positive if >3 agglutinates of ≥5 P. falciparum IEs were observed in the absence of agglutinates of red blood cells. Agglutination assays comparing CS2 IEs and placental isolates were performed, as described elsewhere [4], at 1%-10% parasitemia and a hematocrit of 5%, and IEs were labeled with 10 μg/mL ethidium bromide. A sample was considered to be positive if it contained >3 agglutinates of ≥3 IEs. Samples were coded and tested in a blinded manner.

In mixed agglutination assays [29], CS2 and HCS3 IEs at 5%-8% parasitemia were separately labeled with ethidium bromide or 4′,6-diamidino-2-phenylindole, mixed together in equal numbers, and incubated with serum or plasma as described above. Agglutinates were examined using combined light and fluorescence microscopy, and single-color agglutinates (denoting isolate-specific antibodies) and mixed-color agglutinates (denoting cross-reactive antibodies) were identified. A sample was considered to be positive if it contained ≥5 agglutinates.

Trypsin treatment of IE surface proteins

IEs from culture were washed 3 times in PBS (pH 7.2) and were incubated with trypsin (TPCK-treated; Worthington Biomedical) in PBS for 15 min at room temperature [5, 17]. Control samples were incubated with PBS alone. Digestion was stopped by washing with 10% human serum in RPMI 1640-HEPES, and IEs were washed 3 times in RPMI-HEPES containing 10% pooled human serum.

Western blot analyses of PfEMP1

Western blot analyses were performed using Triton X-100-insoluble, SDS-soluble protein extracts of IEs, as described elsewhere [20]. Triton X-100-insoluble proteins were resuspended in 2% SDS in PBS (to a final concentration equivalent to ∼108 IEs/mL) and were fractionated by SDS-PAGE performed on 6% polyacrylamide gels (Tris-glycine gels) or precast 3%-8% polyacrylamide gradient gels (Tris-acetate buffer; Invitrogen). Nitrocellulose membranes were probed with affinity-purified rabbit antiserum against a conserved sequence in the acidic terminal segment (ATS) of PfEMP1 or an anti-ATS monoclonal antibody (6H1) [20, 30]. Parasite isolates from infected placentas were enriched for pigmented-trophozoite IEs by use of Percoll density gradients [4], and they then were stored in liquid nitrogen until use.

Human serum and plasma samples and rabbit antiserum

Peripheral blood serum and plasma samples were collected from pregnant women receiving routine care at the antenatal clinic and labor ward of the Queen Elizabeth Central Hospital in Blantyre, Malawi, from January 1998 through November 2000 [4, 13]. For samples collected at delivery, placental infection was determined using a combination of histological tests and analyses of blood smears [13]. Peripheral blood samples were also collected from donors who had not been exposed to malaria and who were residents of Australia. Written, informed consent was obtained from all donors, and ethical approval was obtained from the College of Medicine Research Committee, University of Malawi, and the Clinical Research Ethics Committee, Royal Melbourne Hospital Research Foundation. Antiserum (R1945) was raised to the surface of CS2 IEs [20, 31] by vaccination of rabbits with whole CS2 IEs (pigmented trophozoite-stage IEs). Approval for animal use was granted by the Animal Ethics Committee of the Walter and Eliza Hall Institute of Medical Research.

RESULTS

Adhesion phenotypes and expression of PfEMP1 by CSA-binding and placental isolates

The properties and origin of the 3 genetically distinct CSA-binding isolates used in these studies are shown in table 1. All 3 isolates demonstrated high levels of adherence to CSA (i.e., a high density of IEs bound to the immobilized receptor) [8] and expressed var2csa-type genes as dominant transcripts [24, 31]; CS2 and HCS3, but not 3D7-CSA, also showed high levels of binding to HA [8]. There was little or no binding of these isolates to intercellular adhesion molecule 1 and CD36, and they did not form erythrocyte rosettes [8].

Table 1. Summary of the properties of defined Plasmodium falciparum isolates used.

Isolate Origin Adhesion phenotypea Dominant var gene(s) expressedb
CS2 It line CSA, HA var2csa homologuec
3D7-CSA 3D7 line CSA var2csa
HCS3 Southeast Asiad CSA, HA var2csa homologue
E8B It line CD36, ICAM-1 varEHA, FCR3S1.2var11, others

NOTE. CSA, chondroitin sulfate A; HA, hyaluronic acid; ICAM-1, intercellular adhesion molecule 1.

a

Major adhesive properties only; minor phenotypes are not indicated (see [8] for details)

b

For details, see [24] and [31].

c

Amino acid homology with Duffy binding-like domains of 3D7 var2csa is 66%-84% [32].

d

Recent clinical isolate adapted to in vitro culture [33].

On Western blots of IE membrane extracts, antibodies to the conserved ATS region of PfEMP1 labeled a single dominant band in CS2, HCS3, and 3D7-CSA (figure 1A). Bands were not observed in extracts of uninfected red blood cells. The dominant PfEMP1 species expressed by the 3 isolates were of very similar relative molecular mass (Mr) (∼300 kDa). By contrast, anti-ATS antibodies labeled bands of lower Mr in isolate E8B (figure 1B), which adheres to CD36 and intercellular adhesion molecule 1 rather than to CSA and HA [5]. We confirmed that the PfEMP1 bands labeled in Western blots are surface exposed (see below). PfEMP1 proteins were labeled with anti-ATS antibodies in Western blots from 4 placental isolates that had not been cultured in vitro before use (figure 1C). There was diversity in the Mr of PfEMP1 bands, and multiple bands were present in extracts of different isolates. Some bands had an Mr similar to that of the single dominant band observed in extracts of CS2 IEs, whereas other bands of higher or lower Mr were observed.

Figure 1.

Figure 1

Demonstration, on Western blots, that 3 isolates selected for adhesion to chondroitin sulfate A (CSA) express Plasmodium falciparum erythrocyte membrane protein 1 (PfEMP1) of the same relative molecular mass, whereas placental isolates appear to express multiple PfEMP1 forms. Membrane-bound proteins were extracted from P. falciparum-infected erythrocytes and were probed, in Western blots, with a PfEMP1-specific antibody raised against a conserved sequence of the acidic terminal segment. A, Isolates selected for adhesion to CSA (CS2, HCS3, and 3D7-CSA). Each isolate appears to express a single dominant form of PfEMP1 with a similar molecular weight (∼300 kDa; SDS-PAGE performed with Tris-acetate buffer on 3%-8% gradient gel). B, Labeling of PfEMP1 bands of different relative molecular masse with extracts from the CS2 isolate, compared with isolate E8B, which adheres to CD36 and intercellular adhesion molecule 1 (SDS-PAGE was performed with Tris-glycine buffer on 6% acrylamide gel). C, Labeling of multiple PfEMP1 bands in extracts from different placental isolates. Isolate CS2 was included for comparison (SDS-PAGE was performed with Tris-acetate buffer on 3%-8% gradient gel). Molecular weight markers (expressed in kilodaltons) appear to the left of each blot. No labeling of proteins was observed with extracts from uninfected erythrocytes (i.e., red blood cells [RBCs]).

Differences in the properties of IE surface antigens

PfEMP1 bands of all 3 CSA-binding isolates were sensitive to trypsin treatment of the surface of intact IEs, as was indicated by the appearance of smaller-sized bands labeled with anti-ATS antibodies in the trypsin-treated samples but not in the control samples (figure 2). The sensitivity of PfEMP1s to trypsin treatment of IEs confirmed that the labeled PfEMP1s are expressed on the surface of IEs. Not all of PfEMP1 is exposed at the IE surface and is accessible to trypsin [34]; therefore, the ∼300-kDa band of intact PfEMP1 is still observed. Further confirming that surface antigens are cleaved by trypsin, agglutination of CS2 IEs by serum samples obtained from pregnant women was substantially reduced by trypsin treatment (mean reduction in agglutination score, 51% [compared with control]). For 2 of 11 serum samples tested, agglutination was unchanged by trypsin treatment (difference in the agglutination score vs. control, <5%), and, for 1 sample, agglutination was increased.

Figure 2.

Figure 2

Demonstration, on Western blots, of cleavage of Plasmodium falciparum erythrocyte membrane protein 1 (PfEMP1) as a result of trypsin treatment (10 μg/mL for 15 min) of the surface of intact infected erythrocytes (IEs) that adhere to chondroitin sulfate A (CSA). Membrane-bound proteins were extracted from trypsin-treated and control IEs and were probed with a PfEMP1-specific antibody raised against a conserved sequence of the acidic terminal segment. Bands of lower relative molecular mass (Mr) that were labeled in the trypsin-treated samples (Tryp) were not observed in the control samples (Con). A, Yielding of 2 bands of low Mr by trypsin in both CS2 and HCS3. The Mr of the cleavage products also differed between the 2 isolates. An additional lane for HCS3 (far right; Con) is shown at reduced exposure to reveal the position of the major high-molecular-weight band. B, Detection of only 1 additional band in 3D7-CSA in trypsin-treated samples. This band was of lower Mr than were bands detected in CS2. An additional lane for 3D7-CSA (far right; Con) is shown at reduced exposure, to reveal the position of the major high-molecular-weight band. No labeling of proteins was observed with extracts from uninfected erythrocytes (i.e., red blood cells [RBCs]). Molecular weight markers (expressed in kilodaltons) appear to the left of the blots. SDS-PAGE was performed with Tris-acetate buffer (3%-8% gradient gel).

The patterns and/or sizes of ATS antibody-positive trypsin cleavage bands were different for each isolate, suggesting differences in the PfEMP1 forms expressed. Trypsin generated 2 bands in CS2 and HCS3 (figure 2A) but only 1 band in 3D7-CSA (figure 2B). Bands in HCS3 and 3D7-CSA had a lower Mr than did those in CS2, and all trypsin cleavage bands differed in the Mr between the 3 isolates. On the basis of the predicted ATS and transmembrane domain of var2csa, which are not surface exposed, the minimum-size band generated by enzyme digestion of the IE surface would not be less than ∼45 kDa. The effect of trypsin treatment of intact IEs on adhesion to CSA also varied among the 3 isolates (figure 3A), with the difference in sensitivity being most marked between CS2 and HCS3. Adhesion of HCS3 was highly sensitive to trypsin (at a concentration of 1 μg/mL, the mean ± SEM inhibition of adhesion was 86% ± 7%; at a concentration of 10 μg/mL, the mean inhibition of adhesion was 99%), whereas CS2 was largely resistant to trypsin (figure 3A). The 2 isolates had equivalent levels of binding to CSA (mean ± SEM, 407 ± 27 and 535 ± 5 IEs/mm2 for HCS3 and CS2, respectively [tested in parallel]). Adhesion of both isolates to HA is sensitive to a trypsin concentration of 1-10 μg/mL [5, 8]. 3D7-CSA was resistant to trypsin at a concentration of 1 μg/mL, and the mean inhibition of adhesion was 57% at a concentration of 10 μg/mL and was >90% at a concentration of 100 μg/mL.

Figure 3.

Figure 3

Differences between isolates in adhesion to chondroitin sulfate A (CSA) after trypsin treatment of the surface of intact infected erythrocytes (IEs), as well as differences in the inhibition of adhesion of isolates by antiserum. A, Effect of trypsin treatment of CS2, HCS3, and 3D7-CSA IEs on adhesion to CSA. Adhesion of CS2 IEs was largely resistant to trypsin, whereas adhesion of HCS3 IEs was highly sensitive to trypsin at all concentrations. 3D7-CSA was partially inhibited at a trypsin concentration of 10 μg/mL and was inhibited >90% at a concentration of 100 μg/mL. Values expressed relative to adhesion of control-treated IEs are the mean ± SEM of 2 experiments performed in triplicate. B, Effect of rabbit antiserum raised against CS2 IEs on adhesion of CS2 and HCS3 IEs to CSA. The CS2 antiserum completely inhibited adhesion of CS2 IEs at 1:10 and 1:20 dilution, whereas there was little inhibition of adhesion of HCS3 IEs. Values expressed relative to adhesion using control rabbit serum are the mean ± SEM of 2 experiments performed in triplicate. Differences between inhibition of CS2 and HCS3 IEs were statistically significant at all serum dilutions (P < .01, Wilcoxon rank-sum test).

The adhesion inhibitory effect of rabbit antiserum R1945, which reacts with the surface of CS2 but not with the parent non-CSA-binding isolate E8B [20, 31], also demonstrated differences between the CSA-binding isolates. Antiserum R1945 strongly inhibited adhesion of CS2 IEs in a concentration-dependent manner, relative to nonimmune rabbit serum, whereas there was little effect on adhesion of HCS3 IEs (figure 3B).

Evidence of isolate-specific and cross-reactive antibodies in serum samples obtained from pregnant women

Serum samples obtained from Malawian primigravidae with active or past placental infection were used for comparisons of IgG binding to the surface of CS2, HCS3, and 3D7-CSA IEs (figure 4). For the same serum samples, IgG binding was significantly correlated between CS2 and HCS3 (r = 0.837; n = 47; P < .001), CS2 and 3D7-CSA (r = 0.839; n = 46; P < .001), and HCS3 and 3D7-CSA (r = 0.792; n = 46; P < .001). For most samples, if IgG to one of the isolates was detected, then IgG to the other isolates was also detected, particularly when high levels of IgG binding were observed. However, some individual serum samples demonstrated either substantial differences in IgG binding to the different isolates or significant IgG binding to only 1 or 2 of the different isolates. Differences in the reactivity of serum samples against isolates were confirmed by repeat testing. All 3 isolates were recognized, in a parity-dependent manner, by serum antibodies in pregnant women; this finding suggests that the isolates express antigens associated with placental malaria ([13]; data not shown).

Figure 4.

Figure 4

Distinct and overlapping reactivity to different chondroitin sulfate A (CSA)-binding isolates by serum samples obtained from pregnant women. Binding of IgG to the surface of CS2-, HCS3-, and 3D7-CSA-infected erythrocytes (IEs) was determined using immunofluorescence with flow cytometry. All isolates were tested against the same panel of serum samples obtained from primigravid Malawian women. IgG binding is expressed as a ratio of the fluorescence of IEs to the fluorescence of uninfected erythrocytes. A relative score of 1, denoted by broken lines in the charts, indicates that no significant IgG binding to IEs was detected, compared with uninfected erythrocytes. Correlations are as follows: r = 0.839 and P < .001 for CS2 and 3D7-CSA (top; n = 46); r = 0.837 and P < .001 for CS2 and HCS3 (middle; n = 47); and r = 0.792 and P < .001 for HCS3 and 3D7-CSA (bottom; n = 46).

We extended comparisons between CS2 and HCS3, because these isolates had clear differences defined using trypsin and rabbit antiserum samples. Many of the serum samples obtained from uninfected women in midpregnancy and from primigravidae at the time of delivery demonstrated similar effects on adhesion of the isolates, but some serum samples differentially inhibited adhesion of the 2 isolates (figure 5). Inhibition of the 2 isolates showed an overall correlation in both sample sets (r = 0.420 and P = .058 [figure 5A]; r = 0.56 and P = .037). A selection of serum samples underwent a second round of testing to confirm the cross-inhibitory effect of some serum samples and the differential inhibitory effect of others (figure 5B). Of the serum samples obtained, at delivery, from pregnant women with or without placental infection, 54 of 96 samples tested in agglutination assays were found to agglutinate CS2 and/or HCS3. Of the positive samples, 32 (59.3%) agglutinated both CS2 and HCS3, and 22 samples agglutinated only 1 of the isolates (9 agglutinated HCS3 only, and 13 agglutinated CS2 only; figure 5C). Of the positive samples, a greater proportion of those obtained from multigravidae and secundigravidae had agglutinating antibodies to both isolates (P < .05, for primigravidae vs. secundigravidae and multigravidae), which suggests that multigravidae and secundigravidae had a greater repertoire of agglutinating antibodies than did primigravidae. For both isolates, agglutinating antibodies were more prevalent among multigravidae and were rarely detected among men; these findings suggest that these antigenic determinants are predominantly restricted in their expression to malaria during pregnancy.

Figure 5.

Figure 5

Isolate specificity and overlap in adhesion inhibitory activity and agglutinating antibodies for CS2, HCS3, and placental isolates among serum samples obtained from pregnant women. A, Representative example of the correlation between the inhibitory effect of samples on adhesion to chondroitin sulfate A (CSA) of CS2- vs. HCS3-infected erythrocytes (IEs) (samples from primigravid women were obtained at delivery; r = 0.42; P = .058). B, Inhibition of adhesion of CS2 and HCS3 IEs to CSA by serum samples obtained from multigravidae (MG) and primigravidae (PG) during the second trimester; values are the mean of 2 experiments performed in duplicate, and they are expressed relative to values for control samples. The gravidity of the donors is indicated below the X-axis. C, Antibodies to CSA-binding isolates CS2 and HCS3 in the serum samples obtained from pregnant women, as measured by agglutination. Data show the proportion of serum samples positive to CS2 only, to HCS3 only, or to both. A greater proportion of secundigravidae (SG) and MG had antibodies to both isolates, compared with PG (P < .05, χ2 test). The proportion of serum samples that were negative for both isolates was 50%, 33%, and 44% for PG, SG, and MG, respectively. D, Effect of serum samples obtained from pregnant women on the adhesion to CSA of IEs from isolate CS2 and a placental isolate. Gravidity of the donors is indicated below the X-axis. neg, negative; pos, positive.

To directly test for the presence of antibodies that are cross-reactive to the surface of CS2 and HCS3, we performed mixed agglutination assays [29] with selected serum samples obtained from primigravidae (table 2). Most serum samples that reacted with CS2 and HCS3 formed only single-color isolate-specific agglutinates. Some serum samples for which a reaction was noted did result in mixed agglutinates (formed by CS2 and HCS3 IEs), and, among these serum samples, single-color agglutinates were also prevalent, suggesting that both isolate-specific and cross-reactive antibodies were present in the sample. Mixed agglutinates were typically formed by a similar proportion of CS2 and HCS3 IEs. Of the IEs in agglutinates, only a minority of CS2 and HCS3 IEs were observed in mixed agglutinates.

Table 2. Results of mixed agglutination assays for the detection of cross-reactive antibodies against CS2 and HCS3 Plasmodium falciparum-infected erythrocytes (IEs), for serum samples obtained from primigravidae.

Agglutinates
Proportion of agglutinates that were mixed, %
Serum sample CS2 only HCS3 only Mixed
P104 + +
P98 + + + 20a
P92 +
CS290 +
P160 + +
P62 +
P169 + +
CS263 + + + 43b
CS79 + NA
CS294 + NA
P156
P73

NOTE. All serum samples were obtained from primigravidae at delivery. Parasite isolates were tested at 5%-8% parasitemia. NA, not assessed; +, positive for agglutination; −, negative for agglutination.

a

A total of 51.5% of agglutinates were formed by CS2 only, and 28.8% were formed by HCS3 only. Of all CS2 IEs in agglutinates, 21.6% were in mixed agglutinates; of all HCS3 IEs in agglutinates, 37.4% were in mixed agglutinates; and, among mixed agglutinates, 51% of IEs were CS2.

b

A total of 29.3% of agglutinates were formed by CS2 IEs only, and 27.6% were formed by HCS3 only. Of all CS2 IEs in agglutinates, 21.6% were in mixed agglutinates; of all HCS3 IEs in agglutinates, 37.4% were in mixed agglutinates; and, among mixed agglutinates, 54% of IEs were CS2.

Serum samples differed in their agglutination of CS2 versus placental isolates. Some serum samples that agglutinated CS2 did not agglutinate any placental isolates tested, and some that did not agglutinate CS2 did agglutinate some of the placental isolates. As reported elsewhere [4], panagglutinating antibodies against different placental isolates were not observed with any serum samples from pregnant women, and agglutinating antibodies against placental parasites were, to a large extent, isolate specific (table 3). A comparison of the CSA adhesion inhibitory activity of serum samples against CS2 IEs and fresh placental IEs (figure 5D) demonstrated that some serum samples inhibited both CS2 and placental isolates, whereas others differentially inhibited the two.

Table 3. Agglutination of infected erythrocytes of different placental Plasmodium falciparum isolates and the chondroitin sulfate A-binding isolate CS2, for serum samples obtained from pregnant women.

Parasite isolate
Serum sample A B C D E F G CS2
075 + ... ... ...
116 + ... ... ...
124 + ... ... ...
154 + + + ... ... ... +
164 ... ... ... ... + +
169 ... ... ... ... + +
171 + ... ... ... +
175 ... ... ... ... +
199 ... ... ... ...
205 ... ... ... ...

NOTE. Placental isolates A-G were tested without intervening culture. Placental isolates and serum samples were obtained from Malawian women. Serum samples 116, 075, 199, and 205 were obtained from primigravid donors, and other serum samples were obtained from multigravid donors (gravidity, 3 or 4). +, positive for agglutination; −, negative for agglutination.

DISCUSSION

Convincing evidence of antigenic differences between CSA-binding and placental P. falciparum IEs was obtained from serologic assays and analysis of the PfEMP1 variants expressed. We measured antibodies by agglutination assay, adhesion inhibition assay, and indirect immunofluorescence with flow cytometry, because these different assays measure antibodies targeting separate and overlapping epitopes [13]; therefore, examining reactivity by use of all assay types is more informative than examining reactivity by use of a single assay type. Our prior studies have shown that, on their own, agglutinating antibodies do not account for inhibition of adhesion [13]. Antigenic differences and isolate-specific antibodies were further confirmed using mixed agglutination assays, which suggested that isolate-specific agglutinating antibodies either were more common or were present at higher levels than were cross-reactive agglutinating antibodies. Differences between CS2 and HCS3 IEs were further confirmed using rabbit antiserum against CS2 IEs. The effect of trypsin on CSA adhesion and PfEMP1 cleavage differed among the 3 CSA-binding isolates. Examination of the trypsin cleavage patterns of PfEMP1 on Western blots may be a useful tool for comparing isolates expressing homologous var genes. We also observed differences between placental and CSA-binding isolates in antibody binding, inhibition of adhesion by serum, and PfEMP1 expression. Furthermore, placental isolates appeared to express multiple PfEMP1 types, whereas CSA-binding isolates appeared to express only 1 PfEMP1 form. Antigenic differences between placental isolates have also been suggested in other studies [4, 35]. Placental isolates in the population studied here typically consist of multiple genotypes [27] and mixed phenotypes [4], which may explain the multiple PfEMP1 forms observed. Our data suggest that a single infection would give rise to a repertoire of antibodies with different specificities.

Antibodies that are cross-reactive to different CSA-binding isolates can develop after exposure to malaria, and IgG binding to the 3 CSA-binding isolates was closely correlated. However, results from our mixed agglutination assays suggest that only a component would result from cross-reactive antibodies binding to conserved epitopes. Use of highly selected isolates excluded the possibility that mixed agglutinates would result from agglutination of minor phenotypes present in isolates, as was suggested to have occurred in other studies using clinical isolates [25]. Adhesion domains might be more conserved or antigenically restricted than other domains; therefore, antibodies targeting receptor-binding sites might have broader activity. These possibilities are considered in detail elsewhere [36]. Prior studies involving nonpregnant individuals have found that agglutinating antibodies are largely variant specific, and few studies have provided evidence of substantial cross-reactivity [29, 37-39]. Cross-reactive antibodies to the surface of CSA-binding IEs have been induced in vaccinated animals [31, 40], but they have not been previously reported in human serum samples. Our data suggest that antibodies that cross-react with different isolates do not necessarily cross-inhibit, which may influence their role in immunity. This finding was clearly shown with rabbit antiserum, which labels the surface of HCS3 [31] but does not inhibit adhesion (figure 3B), as well as with serum from pregnant women.

Antigenic differences between isolates may arise from the expression of different placental-binding or pregnancy-specific PfEMP1 types and/or polymorphisms among relatively conserved var genes, such as var2csa-type genes [36]. Antibodies that are cross-reactive to CS2 and HCS3 IEs probably reflect the presence of conserved epitopes within the different var2csa-type genes expressed. PfEMP1 appears to be the major antigen on the IE surface, but expression of other IE surface proteins, such as rifins [41, 42], may contribute to antigenic properties. It is likely that var2csa PfEMP1 is the major target of antibodies to the IE surface of the defined CSA-binding isolates, because var2csa-type genes were the dominant full-length var transcripts in CS2, 3D7-CSA, and HCS3 [24, 31]. Our data demonstrate that polymorphisms in var2csa correlate with variation in antigenic phenotype, influencing the binding of acquired antibodies and differences in the properties of PfEMP1 and parasite adhesion. Identity between the Duffy binding-like domains of 3D7-CSA and CS2 var2csa is 66%-84% [32]; HCS3 var2csa is not yet fully sequenced. 3D7-CSA var2csa has a 12-aa deletion in the Duffy binding-like 3x domain, compared with that expressed by CS2 [43]. 3D7-CSA IEs do not adhere to HA, whereas CS2 IEs do [8]. These data have important implications for potential vaccine development and for identification of the targets of protective antibodies. Several observations suggest that differences in the binding of antibodies to the isolates are not simply explained by possible differences in the level of PfEMP1 expression or the number of epitope copies between isolates. These observations include differential binding or inhibition of adhesion of isolates by serum antibodies (figures 4 and 5) and rabbit antiserum (figure 3B), variation in the patterns of trypsin cleavage of PfEMP1 (figure 2), and inhibition of adhesion by trypsin treatment (figure 3A). Relative adhesion to CSA was generally highest for HCS3, and it was lowest for 3D7-CSA [8].

Among women exposed to malaria who had antibodies to CSA-binding isolates, a greater proportion of secundigravidae and multigravidae had antibodies to both CS2 and HCS3 IEs, compared with primigravidae. This suggests that successive exposures occurring during pregnancy may enhance the acquisition of cross-reactive antibodies and/or expand the repertoire of variant-specific antibodies, which may contribute to the development of protective immunity. Among women with antibodies to the surface of CS2 IEs, we previously found that the inhibitory activity of serum antibodies among primigravidae and multigravidae did not differ substantially [13].

Elucidating the extent of the diversity or conservation of placental IE antigens, CSA-binding sites, and antibody epitopes is essential to understanding the basis of acquired immunity and to developing preventive vaccines. In conclusion, our findings clearly indicate that major antigenic differences exist between defined CSA-binding isolates expressing var2csa as the dominant var gene and between CSA-binding and placental isolates. The basis for antigenic differences appears to be explained, at least in part, by differences in the PfEMP1 forms expressed. After exposure, pregnant women appear to acquire a repertoire of variant-specific antibodies, some of which cross-react with different placental isolates, and the breadth of reactivity appears to be greater among women with greater exposure. Immunity may be mediated by a repertoire of antibodies to both diverse and common epitopes, and strategies based on vaccination with a single domain or isolate might be hampered by the antigenic diversity present in the parasite population. The identification and detailed characterization of conserved epitopes would contribute enormously to understanding immunity and developing effective interventions against malaria during pregnancy.

Acknowledgments

We thank Ebby Chalaluka, Labes Njiragoma, Maxwell Kanjala, and Patrick Mkundika for assistance with sample collection and processing in Blantyre, Malawi. We also thank Malcolm Molyneux and Terrie Taylor, the staff of the antenatal clinic and delivery suite of the Queen Elizabeth Central Hospital (Blantyre, Malawi), and all individuals who participated in the study. Human erythrocytes and serum samples used for in vitro culture were kindly provided by the Australian Red Cross Blood Service (Melbourne, Victoria, Australia).

Financial support: National Health and Medical Research Council of Australia (grants 145677 and 215201; Neil Hamilton Fairley Fellowship to J.G.B.); The Wellcome Trust, UK (Senior Research Fellowship to S.J.R.); Walter and Eliza Hall Institute of Medical Research (Miller Fellowship to J.G.B.).

Footnotes

Potential conflicts of interest: none reported.

References

  • 1.Walter PR, Garin Y, Blot P. Placental pathologic changes in malaria: a histologic and ultrastructural study. Am J Pathol. 1982;109:330–42. [PMC free article] [PubMed] [Google Scholar]
  • 2.Beeson JG, Amin N, Kanjala M, Rogerson SJ. Selective accumulation of mature asexual stages of Plasmodium falciparum-infected erythrocytes in the placenta. Infect Immun. 2002;70:5412–5. doi: 10.1128/IAI.70.10.5412-5415.2002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Fried M, Duffy PE. Adherence of Plasmodium falciparum to chondroitin sulfate A in the human placenta. Science. 1996;272:1502–4. doi: 10.1126/science.272.5267.1502. [DOI] [PubMed] [Google Scholar]
  • 4.Beeson JG, Brown GV, Molyneux ME, Mhango C, Dzinjalamala F, Rogerson SJ. Plasmodium falciparum isolates from infected pregnant women and children are associated with distinct adhesive and antigenic properties. J Infect Dis. 1999;180:464–72. doi: 10.1086/314899. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Beeson JG, Rogerson SJ, Cooke BM, et al. Adhesion of Plasmodium falciparum-infected erythrocytes to hyaluronic acid in placental malaria. Nat Med. 2000;6:86–90. doi: 10.1038/71582. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Flick K, Scholander C, Chen Q, et al. Role of nonimmune IgG bound to PfEMP1 in placental malaria. Science. 2001;293:2098–100. doi: 10.1126/science.1062891. [DOI] [PubMed] [Google Scholar]
  • 7.Pouvelle B, Buffet PA, Lepolard C, Scherf A, Gysin J. Cytoadhesion of Plasmodium falciparum ring-stage-infected erythrocytes. Nat Med. 2000;6:1264–8. doi: 10.1038/81374. [DOI] [PubMed] [Google Scholar]
  • 8.Beeson JG, Brown GV. Plasmodium falciparum-infected erythrocytes demonstrate dual specificity for adhesion to hyaluronic acid and chondroitin sulfate A and have distinct adhesive properties. J Infect Dis. 2004;189:169–79. doi: 10.1086/380975. [DOI] [PubMed] [Google Scholar]
  • 9.Fried M, Nosten F, Brockman A, Brabin BJ, Duffy PE. Maternal antibodies block malaria. Nature. 1998;395:851–2. doi: 10.1038/27570. [DOI] [PubMed] [Google Scholar]
  • 10.Maubert B, Fievet N, Tami G, Cot M, Boudin C, Deloron P. Development of antibodies against chondroitin sulfate A-adherent Plasmodium falciparum in pregnant women. Infect Immun. 1999;67:5367–71. doi: 10.1128/iai.67.10.5367-5371.1999. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Ricke CH, Staalsoe T, Koram K, et al. Plasma antibodies from malaria-exposed pregnant women recognize variant surface antigens on Plasmodium falciparum-infected erythrocytes in a parity-dependent manner and block parasite adhesion to chondroitin sulfate A. J Immunol. 2000;165:3309–16. doi: 10.4049/jimmunol.165.6.3309. [DOI] [PubMed] [Google Scholar]
  • 12.Fried M, Duffy PE. Maternal malaria and parasite adhesion. J Mol Med. 1998;76:162–71. doi: 10.1007/s001090050205. [DOI] [PubMed] [Google Scholar]
  • 13.Beeson JG, Mann EJ, Elliott SR, et al. Antibodies to variant surface antigens of Plasmodium falciparum-infected erythrocytes and adhesion inhibitory antibodies are associated with placental malaria and have overlapping and distinct targets. J Infect Dis. 2004;189:540–51. doi: 10.1086/381186. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.O’Neil-Dunne I, Achur RN, Agbor-Enoh ST, et al. Gravidity-dependent production of antibodies that inhibit binding of Plasmodium falciparum-infected erythrocytes to placental chondroitin sulfate proteoglycan during pregnancy. Infect Immun. 2001;69:7487–92. doi: 10.1128/IAI.69.12.7487-7492.2001. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Staalsoe T, Shulman CE, Bulmer JN, Kawuondo K, Marsh K, Hviid L. Variant surface antigen-specific IgG and protection against clinical consequences of pregnancy-associated Plasmodium falciparum malaria. Lancet. 2004;263:283–9. doi: 10.1016/S0140-6736(03)15386-X. [DOI] [PubMed] [Google Scholar]
  • 16.Duffy PE, Fried M. Antibodies that inhibit Plasmodium falciparum adhesion to chondroitin sulfate A are associated with increased birth weight and the gestational age of newborns. Infect Immun. 2003;71:6620–3. doi: 10.1128/IAI.71.11.6620-6623.2003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Leech JH, Barnwell JW, Miller LH, Howard RJ. Identification of a strain-specific malarial antigen exposed on the surface of Plasmodium falciparum-infected erythrocytes. J Exp Med. 1984;159:1567–75. doi: 10.1084/jem.159.6.1567. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Biggs BA, Goozé L, Wycherley K, et al. Antigenic variation in Plasmodium falciparum. Proc Natl Acad Sci USA. 1991;88:9171–4. doi: 10.1073/pnas.88.20.9171. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.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 erythrocytes. Cell. 1995;82:101–10. doi: 10.1016/0092-8674(95)90056-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Reeder JC, Cowman AF, Davern KM, et al. The adhesion of Plasmodium falciparum-infected erythrocytes to chondroitin sulfate A is mediated by P. falciparum erythrocyte membrane protein 1. Proc Natl Acad Sci USA. 1999;96:5198–202. doi: 10.1073/pnas.96.9.5198. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Gamain B, Smith JD, Avril M, et al. Identification of a 67-amino-acid region of the Plasmodium falciparum variant surface antigen that binds chondroitin sulphate A and elicits antibodies reactive with the surface of placental isolates. Mol Microbiol. 2004;53:445–55. doi: 10.1111/j.1365-2958.2004.04145.x. [DOI] [PubMed] [Google Scholar]
  • 22.Salanti A, Staalsoe T, Lavstsen T, et al. Selective upregulation of a single distinctly structured var gene in chondroitin sulphate A-adhering Plasmodium falciparum involved in pregnancy-associated malaria. Mol Microbiol. 2003;49:179–91. doi: 10.1046/j.1365-2958.2003.03570.x. [DOI] [PubMed] [Google Scholar]
  • 23.Salanti A, Dahlbäck M, Turner L, et al. Evidence for the involvement of VAR2CSA in pregnancy-associated malaria. J Exp Med. 2004;200:1197–203. doi: 10.1084/jem.20041579. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Duffy MF, Byrne TJ, Elliott SR, et al. Broad analysis reveals a consistent pattern of var gene transcription in Plasmodium falciparum repeatedly selected for a defined adhesion phenotype. Mol Microbiol. 2005;56:774–8. doi: 10.1111/j.1365-2958.2005.04577.x. [DOI] [PubMed] [Google Scholar]
  • 25.Roberts DJ. Understanding naturally acquired immunity to Plasmodium falciparum malaria. Infect Immun. 2003;71:589–90. doi: 10.1128/IAI.71.2.589-590.2003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Chai W, Beeson JG, Kogelberg H, Brown GV, Lawson AM. Inhibition of adhesion of Plasmodium falciparum-infected erythrocytes by structurally defined hyaluronic acid dodecasaccharides. Infect Immun. 2001;69:420–5. doi: 10.1128/IAI.69.1.420-425.2001. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Kamwendo DD, Dzinjalamala FK, Snounou G, et al. Plasmodium falciparum: PCR detection and genotyping of isolates from peripheral, placental, and cord blood of pregnant Malawian women and their infants. Trans R Soc Trop Hyg Med. 2002;96:145–9. doi: 10.1016/s0035-9203(02)90284-1. [DOI] [PubMed] [Google Scholar]
  • 28.Mann EJ, Rogerson SJ, Beeson JG. An alternative agglutination assay to measure antibodies to variant surface antigens of Plasmodium falciparum-infected erythrocytes. Trans R Soc Trop Hyg Med. 2003;97:717–9. doi: 10.1016/s0035-9203(03)80111-6. [DOI] [PubMed] [Google Scholar]
  • 29.Newbold CI, Pinches R, Roberts DJ, Marsh K. Plasmodium falciparum: the human agglutinating antibody response to the infected red cell surface is predominantly variant specific. Exp Parasitol. 1992;75:281–92. doi: 10.1016/0014-4894(92)90213-t. [DOI] [PubMed] [Google Scholar]
  • 30.Crabb BS, Cooke BM, Reeder JC, et al. Targeted gene disruption shows that knobs enable malaria-infected red cells to cytoadhere under physiological shear stress. Cell. 1997;89:287–96. doi: 10.1016/s0092-8674(00)80207-x. [DOI] [PubMed] [Google Scholar]
  • 31.Elliott S, Duffy MF, Byrne TJ, et al. Cross-reactive surface epitopes on chondroitin sulfate A-adherent Plasmodium falciparum-infected erythrocytes are associated with transcription of var2csa. Infect Immun. 2005;73:2848–56. doi: 10.1128/IAI.73.5.2848-2856.2005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Kraemer SM, Smith JD. Evidence for the importance of genetic structuring to the structural and functional specialization of the Plasmodium falciparum var gene family. Mol Microbiol. 2003;50:1527–38. doi: 10.1046/j.1365-2958.2003.03814.x. [DOI] [PubMed] [Google Scholar]
  • 33.Beeson JG, Chai W, Rogerson SJ, Lawson AM, Brown GV. Inhibition of binding of malaria-infected erythrocytes by a tetradecasaccharide fraction from chondroitin sulfate A. Infect Immun. 1998;66:3397–402. doi: 10.1128/iai.66.7.3397-3402.1998. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Waterkeyn JG, Wickham ME, Davern KM, et al. Targeted mutagenesis of Plasmodium falciparum erythrocyte membrane protein 3 (PfEMP3) disrupts cytoadherence of malaria-infected red blood cells. Embo J. 2000;19:2813–23. doi: 10.1093/emboj/19.12.2813. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Tuikue Ndam NG, Fievet N, Bertin G, Cottrell G, Gaye A, Deloron P. Variable adhesion abilities and overlapping antigenic properties in placental Plasmodium falciparum isolates. J Infect Dis. 2004;190:2001–9. doi: 10.1086/425521. [DOI] [PubMed] [Google Scholar]
  • 36.Beeson JG, Rogerson SJ, Elliott SR, Duffy MF. Targets of protective antibodies to malaria during pregnancy. J Infect Dis. 2005;192:1647–50. doi: 10.1086/496895. [DOI] [PubMed] [Google Scholar]
  • 37.Marsh K, Howard RJ. Antigens induced on erythrocytes by P. falciparum: expression of diverse and conserved determinants. Science. 1986;231:150–3. doi: 10.1126/science.2417315. [DOI] [PubMed] [Google Scholar]
  • 38.Chattopadhyay R, Sharma A, Srivastava VK, et al. Plasmodium falciparum infection elicits both variant-specific and cross-reactive antibodies against variant surface antigens. Infect Immun. 2003;71:597–604. doi: 10.1128/IAI.71.2.597-604.2003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Giha HA, Staalsoe T, Dodoo D, et al. Overlapping antigenic repertoires of variant antigens expressed on the surface of erythrocytes infected by Plasmodium falciparum. Parasitology. 1999;119:7–17. doi: 10.1017/s0031182099004485. [DOI] [PubMed] [Google Scholar]
  • 40.Lekana Douki J-B, Traore B, Costa FTM, et al. Sequestration of Plasmodium falciparum-infected erythrocytes to chondroitin sulfate A, a receptor for maternal malaria: monoclonal antibodies against the native parasite ligand reveal pan-reactive epitopes in placental isolates. Blood. 2002;100:1478–83. doi: 10.1182/blood-2002-01-0315. [DOI] [PubMed] [Google Scholar]
  • 41.Kyes SA, Rowe JA, Kriek N, Newbold CI. Rifins: a second family of clonally variant proteins expressed on the surface of red cells infected with Plasmodium falciparum. Proc Natl Acad Sci USA. 1999;96:9333–8. doi: 10.1073/pnas.96.16.9333. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42.Fernandez V, Hommel M, Chen Q, Hagblom P, Wahlgren M. Small, clonally variant antigens expressed on the surface of Plasmodium falciparum-infected erythrocytes are encoded by the rif gene family and are targets of human immune responses. J Exp Med. 1999;190:1393–403. doi: 10.1084/jem.190.10.1393. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Gamain B, Trimnell AR, Scheidig C, Scherf A, Miller LH, Smith JD. Identification of multiple chondroitin sulfate A (CSA)-binding domains in the var2CSA gene transcribed in CSA-binding parasites. J Infect Dis. 2005;191:1010–3. doi: 10.1086/428137. [DOI] [PubMed] [Google Scholar]

RESOURCES