Plasmodium falciparum Line-Dependent Association of In Vitro Growth-Inhibitory Activity and Risk of Malaria

Abstract

Plasmodium falciparum's ability to invade erythrocytes is essential for its survival within the human host. Immune mechanisms that impair this ability are therefore expected to contribute to immunity against the parasite. Plasma of humans who are naturally exposed to malaria has been shown to have growth-inhibitory activity (GIA) in vitro. However, the importance of GIA in relation to protection from malaria has been unclear. In a case-control study nested within a longitudinally followed population in Tanzania, plasma samples collected at baseline from 171 individuals (55 cases and 116 age-matched controls) were assayed for GIA using three P. falciparum lines (3D7, K1, and W2mef) chosen based on their erythrocyte invasion phenotypes. Distribution of GIA differed between the lines, with most samples inhibiting the growth of 3D7 and K1 and enhancing the growth of W2mef. GIA to 3D7 was associated with a reduced risk of malaria within 40 weeks of follow-up (odds ratio, 0.45; 95% confidence interval [CI], 0.21 to 0.96; P = 0.04), whereas GIA to K1 and W2mef was not. These results show that GIA, as well as its association with protection from malaria, is dependent on the P. falciparum line and can be explained by differences in erythrocyte invasion phenotypes between parasite lines. Our study contributes knowledge on the biological importance of growth inhibition and the potential influence of P. falciparum erythrocyte invasion phenotypic differences on its relationship to protective immunity against malaria.

INTRODUCTION

Plasmodium falciparum is responsible for the most-severe forms of malaria, a disease that was responsible for approximately 225 million clinical cases and around 781,000 deaths in 2009 (54). The development of an efficacious vaccine is widely viewed as a useful advancement toward controlling this disease. However, mimicking naturally acquired immunity that is complex (18) has this far been challenging (51). Preliminary data from the ongoing phase III clinical trials of the leading vaccine candidate, RTS,S, show that the vaccine confers only partial protection against malaria (1). The development of malaria vaccines is largely hampered by the incomplete understanding of the immune responses that are responsible for naturally acquired immunity. For instance, whereas the importance of antibody responses has been demonstrated by passive antibody transfer studies (10), the specific targets and effector mechanisms of most antibodies, such as those to merozoite antigens, are largely unknown (21). Vaccine development is also hampered by the lack of in vitro functional assays that correlate clearly with protective immunity in vivo.

As an obligate intracellular parasite, P. falciparum's ability to invade erythrocytes is essential for its survival within the human host. Immune factors, such as antibodies, that impair this ability are therefore expected to contribute toward immunity against the parasite. This is supported by results of studies in animals in which antibodies against several merozoite antigens generated by vaccination have been shown to inhibit growth of the parasite in vitro (7, 31, 45) and to be associated with protection against homologous blood stage challenge (16, 45).

P. falciparum merozoites can use different pathways for erythrocyte invasion, mediated by variation in the expression and/or use of erythrocyte-binding antigens (EBAs) and P. falciparum reticulocyte-binding homolog (PfRh) proteins (12, 39). Invasion phenotypes of P. falciparum lines can be broadly classified based on their sensitivity to cleavage of erythrocyte surface receptors by enzymes such as neuraminidase and trypsin (as reviewed in reference 37). Sialic acid-dependent invasion (neuraminidase sensitive) involves the EBAs and PfRh1, whereas PfRh2 and PfRh4 are important in sialic acid-independent invasion (19, 46).

In humans, several studies have described the growth-inhibitory activity (GIA) of the total immunoglobulin G (IgG) fraction (4, 27, 28, 36) and of malaria antigen-specific fractions (20, 24, 28, 29, 34) of serum from people living in areas where malaria is endemic. Antibodies that inhibit P. falciparum growth in vitro have also been shown to be present in individuals who are clinically immune to the parasite (6, 9). However, relatively few studies have addressed the association between the GIA of plasma from people living in areas of endemicity and protective immunity. Some of these studies have reported significant associations between GIA and reduced risk of malaria (13, 17, 25), while other studies have been inconclusive (11, 26–28, 36). The variations in the results may be due to differences in study design, malaria transmission intensities, GIA methods, and the choice of parasite line. This is because erythrocyte invasion by merozoites involves several interactions (as reviewed in reference 37). Apart from the interaction between PfRh5 and basigin (14), none of the other known ligand-receptor interactions are required by all parasite lines, implying different invasion pathways. Inhibition of the in vitro growth of P. falciparum by sera from individuals who are naturally exposed to the parasite has been reported to show isolate specificity (35, 53). This implies that the choice of parasite line may influence the association between GIA and malaria risk.

The aim of this study was to assess the relationship between GIA and the risk of malaria by using P. falciparum lines that differ in their erythrocyte invasion phenotypes. We hypothesized that the breadth of GIA (number of parasite lines to which individuals have high GIA) would be predictive of protection from malaria. We found that high GIA to one parasite line, but not to the other lines, was associated with a reduced risk of malaria. We did not find an association between the breadth of GIA and the risk of malaria. These results show that GIA, as well as its association with protection from malaria, is dependent on the P. falciparum line.

MATERIALS AND METHODS

Location and study population.

The population of Nyamisati village, Rufiji District, Tanzania, has been longitudinally followed as part of an epidemiological study of malaria. A cross-sectional survey was conducted in March and April 1999, just before the rainy season. The survey included 890 individuals aged between 1 and 84 years. The overall parasite prevalence was 27% by microscopy and 46% by PCR, with the highest prevalence (74% by PCR) observed in children aged 3 to 5 years (2). Venous blood samples, from which packed erythrocytes and plasma were prepared and frozen, were collected in EDTA. The presence of P. falciparum infections at the time of the survey was investigated by microscopy and by genotyping of the P. falciparum merozoite surface protein 2 gene (msp2) by PCR as described before (2). The genotyping strategy allowed for the grouping of the P. falciparum infections into those that belong to the IC (corresponding to 3D7) and the FC27 allele families. All the participants were monitored for the subsequent 40 weeks, and episodes of malaria were recorded through a passive case detection system by the research team that also operated the only health facility in the village. The case definition of malaria for this study was fever (axillary temperature of >37.5°C or a history of hot body within 24 h) together with a parasite density of ≥5,000 parasites/μl in peripheral blood.

In total, 700 of the 890 individuals were asymptomatic, including having no record of malaria 4 weeks before or 1 week after the survey, at the time of the cross-sectional survey. Among these, 60 individuals had a recorded malaria episode during the 40 weeks of follow-up, of whom 55 were included as cases in the current study and 5 were excluded from the study (2 individuals were seropositive for HIV and 3 individuals had inadequate plasma volumes). Age-matched controls (n = 116) were randomly selected among individuals who did not have malaria during follow-up. The project was approved by the National Institute for Medical Research in Tanzania and the Stockholm Ethical Review Board (Dnr 00-084).

Parasite lines.

The 3D7 parasite line is a cloned line derived from NF54. K1 is of Southeast Asian origin and was obtained from David Walliker's laboratory (Department of Animal and Population Genetics, University of Edinburgh). W2mef parasites were obtained from the Malaria Research and Reference Reagent Resource Center (MR4; Manassas, VA). The molecular pathways employed by these parasite lines for erythrocyte invasion have been previously described (49). 3D7 uses a sialic acid-independent invasion pathway and expresses more PfRh2 and PfRh4 and less PfRh1 than K1 and W2mef. Erythrocyte invasion by W2mef is sialic acid dependent and relies on the use of PfRh1 and the EBAs.

Treatment of plasma samples.

Plasma samples (150 μl) were dialyzed against phosphate-buffered saline (PBS) in dialysis tubes with a 50-kDa molecular mass cutoff MMCO (G-Biosciences) and subsequently reconcentrated to the original starting volume using centrifugal concentration tubes with a 100-kDa MMCO (Pall Corporation).

Growth inhibition assays.

GIA was measured using a previously described method (38). P. falciparum-infected erythrocytes were cultured at pH 7.4 in plastic culture flasks using human group O erythrocytes, at 5% hematocrit, in RPMI HEPES medium supplemented with 50 μg/ml hypoxanthine, 25 mM NaHCO₃, 20 μg/ml gentamicin, and 5% (vol/vol) heat-inactivated pooled human sera from healthy Swedish donors. Cultures were maintained at 37°C in 5% O₂, 5% CO₂, and 90% N₂ and were tightly synchronized by resuspension of culture pellets in 5% d-sorbitol before the start of the growth inhibition assay. Fifty microliters of parasite suspensions at 0.3% parasitemia and 1% hematocrit was then transferred into 96-well U-bottom plates (Techno Plastic Products). Thereafter, 5 μl of test serum was mixed into each well and incubated at 37°C. After 48 h, 10 μl of culture medium was added to each well and incubation was continued. After 2 growth cycles, parasitemia was determined using flow cytometry. Parallel cultures of the respective parasite lines were maintained and monitored microscopically for parasite development to determine the end of the 2-cycle growth inhibition assay (when parasites were in the late ring or early trophozoite stage of the second cycle). For flow cytometry, 100 μl of 10 μg/ml hydroethidine (Invitrogen Corporation) in phosphate-buffered saline (PBS; pH 7.4) was mixed with 25 μl of parasite culture and incubated for 1 h in darkness at room temperature. After centrifugation, the supernatant was discarded, cells were resuspended in 200 μl/well of PBS, and the samples were processed using a FACSCalibur cytometer (BD Bioscience). The percentage of infected erythrocytes was evaluated using CellQuest software (BD Bioscience) by first gating for intact erythrocytes by side scatter and forward scatter parameters and subsequently determining the proportion of ethidium bromide-positive cells. Wells with uninfected erythrocytes were included to control for cross-contamination between wells. The GIA for each plasma sample was expressed as a percentage relative to the parasite growth in control wells in which PBS was added instead of a plasma sample.

ELISAs.

Enzyme-linked immunosorbent assays (ELISAs) against P. falciparum schizont extract prepared from the cultures of the three P. falciparum lines were performed according to a previously described protocol (40). A significant modification to this protocol was that serial dilutions of purified malaria immunoglobulin (prepared by cold ethanol fractionation as previously described [47]) were run on each ELISA plate to allow for standardization of plate-to-plate variations and to allow for the conversion of optical densities to antibody concentrations.

Statistical analysis.

Data analysis was performed using STATA version 11.0 and GraphPad version 5.02. Spearman's rank correlation coefficients were used to assess the relationships between continuous variables. Continuous variables between groups were compared by the Kruskal-Wallis test or the Mann-Whitney test where appropriate. Equality of the distribution of GIA and anti-schizont extract antibody concentrations across the three parasite lines was first tested using the Friedman test. Where this test showed a significant difference in the distribution, the Wilcoxon signed-rank test was then used to test the equality of the distributions between two parasite lines at a time. Each individual was classified as a low or high responder to each parasite line depending on whether its GIA data were above or below the median. The breadth of GIA was assessed by assigning each individual a GIA breadth score (between 0 and 3) depending on the number of parasite lines to which they had GIA that was above the median. The chi-square test for trend was used to test the relationship between the breadth of GIA score and whether an individual experienced an episode of malaria or not. Considering the case-control design of the study, conditional logistic regression was used to assess the associations between GIA and the risk of malaria, which were reported as odds ratios (OR).

RESULTS

Characteristics of the study population.

The median age, sex, and hemoglobin levels were similar between the 55 cases and the 116 age-matched controls (Table 1). The proportions of cases that were P. falciparum parasite positive at the time of the baseline cross-sectional survey were 7.3% and 34.6% by microscopy and PCR, respectively, compared to 24.1% and 50%, respectively, for controls (Table 1). Considering the superior sensitivity of PCR over microscopy, we based further analysis on the PCR data.

Table 1.

Population characteristics

Characteristic	Value for:
	Cases	Controls
n	55	116
Median age (yr) (interquartile range)a	11 (6–26)	11 (7–26)
No. of females (%)	26 (47)	58 (50)
No. (%) positive for P. falciparumb
By microscopy	4 (7.3)	28 (24.1)
By PCR	19 (34.6)	58 (50)
Median hemoglobin level (mg/ml) (interquartile range)	113 (101–123)	110 (99.5–119)

^aControls were matched on age.

^bNumber and proportion of individuals who were P. falciparum positive at the time of the baseline cross-sectional survey.

Distribution of growth inhibitory activity.

The distributions of growth-inhibitory activity (GIA) were significantly different between the three parasite lines (Fig. 1). The plasma samples exhibited higher GIA against 3D7 than against the K1 and W2mef parasite lines (Wilcoxon signed-rank test; P < 0.001). The median GIA was 18% using the 3D7 parasite line (range, −10% to 89%), 7% for the K1 line (range, −48% to 52%), and −12% for the W2mef line (range, −51% to 42%). A negative GIA value is obtained when the parasite growth in a plasma sample well is greater than the growth in PBS control wells. A majority of the plasma samples inhibited parasite growth using the 3D7 and K1 parasite lines (91.2% and 63.8%, respectively) and thus gave positive growth inhibition values, whereas using the W2mef line enhanced parasite growth in a majority of the samples (80.7%). The GIA breadth scores, i.e., the number of parasite lines to which an individual had GIA that was above the median, were similar in the cases and controls (chi-square test for trend P = 0.92) (Fig. 2). There was a weak correlation between the GIAs determined using the respective parasite lines for the controls (Fig. 3D to F) that was not seen for the cases, perhaps due to a lower number of samples from the cases (Fig. 3A to C). There was no significant correlation between GIA to any of the parasite lines and age among the cases (Fig. 4A to C). Among the controls, there was a weak correlation between age and GIA to 3D7 and W2mef but not to K1 (Fig. 4D to F). Among the controls, the distributions of GIA to all three parasite lines were similar in individuals with and without asymptomatic parasitemia at baseline, and among the individuals with asymptomatic parasitemia, this distribution was not affected by the msp2 allelic family of these parasites (see Fig. S1D, E, and F in the supplemental material). The same was true among the cases, with the exception of the GIA to the K1 parasite line, in which the distribution differed (see Fig. S1B in the supplemental material). Among the cases, GIA to the K1 parasite line was lower in individuals infected with parasites belonging to the FC27 msp2 allele family than in aparasitemic individuals (Mann-Whitney test; P = 0.036) and individuals infected with parasites belonging to both msp2 allele families (P = 0.046). Individuals infected with parasites belonging to the IC-1 msp2 allele family had lower GIA to the K1 parasite line than individuals infected with parasites belonging to both msp2 allele families (P = 0.039) (see Fig. S1B in the supplemental material).

Fig 1.

Comparison of growth-inhibitory activities of plasma samples on 3D7, K1, and W2mef P. falciparum lines. Box plot distribution of GIA data among the cases (malaria during follow-up) and controls (no malaria during follow-up) as determined using the 3D7, K1, and W2mef P. falciparum lines. Horizontal bars in the middle of each box indicate the median percentages of growth inhibition. The top and bottom of each box represent the upper and lower quartiles, respectively. The whiskers show the 5th and 95th percentiles. Data points outside the 5th and 95th percentiles are shown as dots above and below each plot. P values were determined by the Wilcoxon signed-rank test.

Fig 2.

Breadth of growth-inhibition activity for the cases and controls. Individuals were assigned a GIA breadth score between 0 and 3 depending on the number of parasite lines to which their plasma had growth-inhibitory activity above the median. Chi-square test for trend (P = 0.92).

Fig 3.

Correlations between growth-inhibitory activities of the plasma samples on 3D7, K1, and W2mef P. falciparum lines. Scatter plots that show the relationships between the growth-inhibitory activities of the plasma samples of the cases (A, B, and C) and those of the controls (D, E, and F) on the three P. falciparum lines. The correlation coefficients were determined by Spearman's rank test.

Fig 4.

Correlations between growth-inhibitory activity and age. Scatter plots that show the relationships between age and the growth-inhibitory activities of the plasma samples of the cases (A, B, and C) and the controls (D, E, and F) on the three P. falciparum lines. The correlation coefficients were determined by Spearman's rank test.

Antibody titers to parasite schizont extract and growth-inhibitory activity.

The levels of antibodies to parasite schizont extract were similar for the three parasite lines (Fig. 5) and increased similarly with age among both the cases and the controls (Spearman's rank test [r_s]; P < 0.001) (Fig. 6A to F). There was no association between the risk of malaria and the antibody responses to schizont extract of the lines after adjusting for age and whether an individual was parasite positive by PCR at the time of the survey: 3D7 had an OR of 1.19 (95% confidence interval [CI], 0.56 to 2.58), K1 had an OR of 0.83 (95% CI, 0.38 to 1.77), and W2mef had an OR of 0.97 (95% CI, 0.46 to 2.07). There was no significant correlation between the GIA and the antibody titers to schizont extracts of the lines: 3D7 had an r_s of 0.02 (P = 0.75), K1 had an r_s of 0.06 (P = 0.43), and W2mef had an r_s of 0.11 (P = 0.15).

Fig 5.

Comparison of the distributions of antibody responses to the parasite schizont extracts of the 3D7, K1, and W2mef P. falciparum lines. Box plot distribution of the levels of IgG to parasite schizont extracts of the 3D7, K1, and W2mef parasite lines among the cases and controls. Horizontal bars in the middle of each box indicate the median antibody concentration. The top and bottom of each box represent the upper and lower quartiles, respectively. The whiskers show the 5th and 95th percentiles. Data points outside the 5th and 95th percentiles are shown as dots above and below each plot.

Fig 6.

Correlation between age and the levels of IgG to parasite schizont extracts of the 3D7, K1, and W2mef P. falciparum lines. Scatter plots show the relationships between age and the levels of IgG to parasite schizont extract of the 3D7, K1, and W2mef parasite lines in the plasma samples from the cases (A, B, and C) and the controls (D, E, and F). The correlation coefficients were determined by Spearman's rank test.

Association between growth-inhibitory activity and risk of malaria.

Analysis by conditional logistic regression showed that GIA determined using the 3D7 parasite line was associated with a reduced risk of malaria; the univariate analysis OR was 0.52 (95% CI, 0.26 to 1.03) and the multivariate analysis OR was 0.45 (95%, CI 0.21 to 0.96) after being adjusted for age, antibody response to parasite schizont extract, and whether an individual was parasite positive by PCR at the time of the survey (Table 2). The multivariate analysis was adjusted depending on whether an individual was parasite positive by PCR at the time of the survey because being parasite positive at the time of the survey was associated with a reduced, but marginally significant, risk of malaria during 40 weeks of follow-up (OR, 0.56; 95% CI, 0.30 to 1.02; P = 0.059). Using the K1 and W2mef parasite lines, there was no association between GIA and the odds of experiencing malaria during follow-up in either univariate or multivariate analysis (Table 2).

Table 2.

Growth inhibition and risk of clinical malaria^a

P. falciparum line	Unadjusted OR (95% CI)	P value	Adjusted ORb (95% CI)	P value	Adjusted ORc (95% CI)	P value	Adjusted ORd (95% CI)	P value
3D7	0.52 (0.26–1.03)	0.062	0.46 (0.22–0.95)	0.037	0.51 (0.25–1.02)	0.059	0.45 (0.21–0.96)	0.040
K1	1.25 (0.65–2.41)	0.503	1.28 (0.65–2.51)	0.471	1.30 (0.67–2.54)	0.440	1.37 (0.68–2.81)	0.379
W2mef	0.95 (0.49–1.88)	0.900	1.06 (0.53–2.14)	0.860	0.95 (0.49–1.88)	0.900	0.95 (0.46–1.96)	0.894

^aGIAs for each parasite line were categorized into “low” and “high” using the median as the cutoff. The odds ratios (ORs) represent the odds of experiencing an episode of malaria for individuals with GIA in the “high” category relative to those with GIA in the “low” category.

^bOdds ratios adjusted for age.

^cOdds ratios adjusted for antibody response to parasite schizont extract.

^dOdds ratios adjusted for age, antibody response to parasite schizont extract, and whether an individual was parasite positive or negative by PCR at the time of the survey.

DISCUSSION

In this study, growth-inhibitory activities of plasma from individuals living in an area of endemicity in Tanzania were significantly different when tested on different parasite lines. The majority of plasma samples inhibited the growth of the 3D7 and K1 lines but enhanced the growth of W2mef. Moreover, GIA at baseline was associated with a reduced risk of malaria during 40 weeks of follow-up using the 3D7 parasite line but not using the K1 and W2mef lines.

So far, significant associations between GIA and reduced risk of malaria have been reported only by studies that used the 3D7 parasite line, as in our study and that by Crompton et al. (13). However, another study did not find such an association (27). Using other parasite lines, Marsh et al. (26), using a Gambian P. falciparum isolate (GAM 83/1), and Perraut et al. (36), using the parasite line FCR3, did not find associations between GIA and risk of malaria. Similarly, we did not find associations between GIA and the risk of malaria using the K1 and W2mef parasite lines. We also did not find a significant association between the number of parasite lines to which an individual had high GIA and the risk of malaria, likely due to the divergent associations with risk for the different lines.

Considering that the inhibitory effect of plasma on in vitro growth of P. falciparum has been shown to mainly be due to erythrocyte invasion rather than inhibition of the intraerythrocytic growth (38) and that erythrocyte invasion by merozoites is a rapid process, inhibitory antibodies have to be present in high concentrations and/or possess high affinity to exert their inhibitory effects (43). These observations imply that the concentration/amount of antibodies used in growth inhibition assays may affect the outcome of these assays. Therefore, it can be argued that the differences in the methods used to perform the growth inhibition assays may also contribute to the inconsistencies in the association between GIA and the risk of malaria reported so far. Crompton et al. purified total immunoglobulin G (IgG) fractions from the respective plasma samples and used the IgG fractions at a standardized final concentration of 6.3 mg/ml. We used the method described by Persson et al. (38) in which plasma is dialyzed against PBS, reconstituted to its original volume, and used in the growth inhibition assay at a 1-in-10 dilution in the final test well. Therefore, using this method, the amount of antibodies added to each well varies depending on the concentration of antibodies in each of the original plasma samples. However, this argument is negated by the fact that our study and that by Crompton et al. found a significant association between GIA and a reduced risk of malaria using the 3D7 parasite line even though the two studies used different methods to perform the growth inhibition assays.

There was no correlation between age and GIA to any of the three parasite lines among the individuals who had malaria during follow-up (cases). The correlation among controls, i.e., without malaria, was significant, but weak, for 3D7 and W2mef but not for K1. This was despite the age-associated increase in the levels of antibodies to schizont extracts of all three parasite lines in both the cases and the controls. Previous studies have reported conflicting associations between GIA and age. Two studies have reported a negative correlation (17, 27), whereas one study showed a positive correlation but only among individuals who were aparasitemic at the time when plasma samples were collected (13). It has been suggested that growth-inhibitory antibodies are acquired early in life and do not appear to be boosted by ongoing exposure (27). This may explain the positive correlation between age and GIA observed by Crompton et al. (13) in looking at individuals aged between 2 and 10 years and the negative association observed in studies that included relatively older study populations (17, 27). We found only weak correlations between the GIAs of the plasma samples on the three parasite lines (Fig. 3), suggesting that much of the antibody responses that contribute to growth inhibition may not be directed at epitopes that are conserved across the three parasite lines.

The distributions of GIA to each of the parasite lines were not significantly different between parasitemic and aparasitemic individuals. This is in agreement with one previous study (27) but differs with another study that showed higher GIA in the parasitemic individuals than in aparasitemic individuals (13). Whether being parasite positive reflects higher exposure and thus immunity and whether concurrent parasitemia contributes to growth inhibition remain to be established. Considering that naturally acquired immunity to malaria is, to a considerable extent, strain specific, we sought to find out if GIA shows any strain specificity in relation to the major allelic family dimorphism of msp2. Among the parasitemic individuals, we found no evidence that the msp2 allele family of the parasites affected the distribution of GIA to the 3D7 and W2mef parasite lines. In the case of the K1 line, the potential effect of the msp2 allele family of the parasites present at baseline on the distribution of GIA was evident only among the cases and not among the controls. This finding is likely to be coincidental, owing to the small number of cases infected with parasites belonging to the FC27 allele family or concurrently infected with parasites belonging to both allele families. We therefore did not find substantial evidence that the msp2 allele family of the parasites affects the distribution of GIA.

Interestingly, we found significant differences in the distributions of GIA of plasma to the three parasite lines, with a majority of the plasma samples that inhibited the growth of 3D7 enhancing the growth of the W2mef parasite line (Fig. 1). This contrasts with the findings of earlier studies, which show that plasma from individuals living in areas where malaria is endemic is generally inhibitory to laboratory-adapted Plasmodium falciparum lines (17, 35).

This observation might be due to several factors, some of which may not be antibody related. The lack of significant correlations between GIAs determined using any of the parasite lines and antibody levels to schizont extract made from cultures of the same parasite line suggests that the GIA might be partly due to nonantibody host or exogenous factors in plasma. For instance, traces of pyrimethamine in the plasma samples may affect in vitro growth of the three parasite lines differently due to various susceptibilities to this drug. At the time of sample collection, sulfadoxine-pyrimethamine (SP) was the drug of choice for treatment of uncomplicated malaria and it was not possible to dialyze pyrimethamine out of the plasma samples. The K1 and W2mef parasite lines have triple mutations in the dihydrofolate reductase gene (42), and these mutations confer resistance to pyrimethamine, whereas 3D7 does not have any of these resistance-conferring mutations. However, the possibility of pyrimethamine explaining our results is low, since the study clinic was the only health facility in the village and individuals who either were symptomatic at the time of the survey or had malaria in the preceding 4 weeks were excluded from our study.

Our observation can be partly explained by the findings of studies done using monoclonal antibodies (MAbs). Some of these studies have shown that although some MAbs inhibit invasion by preventing the requisite processing of some of the proteins necessary for parasite invasion, other MAbs either are neutral or block the activity of the inhibitory ones (3, 7, 50). Additionally, human studies have shown that naturally acquired antibodies to merozoite surface protein 1 are comprised of inhibitory, neutral, and blocking fractions (23, 33). These studies, together with another that reports that a monoclonal antibody to a P. falciparum asparagine-rich protein enhances the in vitro growth of P. falciparum (22), suggest that there might be antibodies in the plasma of people living in areas of malaria endemicity that enhance parasite growth. This proposition is further supported by reports that antibodies purified from strongly inhibitory sera may enhance parasite growth in vitro (5, 44, 48). However, the findings of these studies do not fully explain the fact that the plasma samples we tested inhibited the growth of one parasite line and enhanced the growth of the other two.

It has been proposed that rosetting can facilitate merozoite invasion (52), which would then enhance parasite growth in vitro and thus impact the growth inhibition assay. It can therefore be argued that differences in rosetting phenotypes between parasite lines can explain differences in the distributions of GIA of plasma to different parasite lines. However, previous studies have shown that rosetting does not affect erythrocyte invasion rates under static in vitro parasite culture conditions either in the presence (15) or in the absence (8) of invasion-inhibitory antibodies. These two studies (8, 15), together with the fact that we did not observe any rosetting or agglutination by microscopy with any of the three parasite lines, argue against the influence of rosetting on the differences between the distributions of GIA of plasma to the three parasite lines that we observed.

A plausible explanation of our observation is that the three parasite lines differ in the molecular pathways that they employ for erythrocyte invasion. Except for the interaction between PfRh5 and basigin (14), none of the other known ligand-receptor interactions involved in erythrocyte invasion by merozoites are required by all parasite lines. This implies that different parasite lines employ invasion pathways that are, to a large extent, not common to each other. 3D7 is less sialic acid dependent than K1, which, in turn, is less sialic acid dependent than W2mef (49). They also differ in their relative expression of proteins, such as PfRh1, that are necessary for erythrocyte invasion, with 3D7 expressing less PfRh1 than K1, which expresses less PfRh1 than W2mef (49). We observed a sequential decrease in the inhibitory activity of the plasma samples we tested to the three parasite lines, with 3D7 being more inhibited than K1, which was, in turn, more inhibited than W2mef (Fig. 1). The differences in the distributions of GIA to the three parasite lines may be explained by the differences in the relative expression of PfRh1. All P. falciparum lines express this protein but with different levels of expression (32, 49). The difference in expression has been shown to be due to amplification of the PfRh1 gene copy number (49). P. falciparum lines with elevated PfRh1 gene copy numbers grow faster under in vitro culture conditions than lines without this elevation (41), suggesting that the gene amplification is of biological relevance. The W2mef parasite line has three copies of the PfRh1 gene, while 3D7 and K1 have single copies (49). Nair et al. have studied 202 single-clone natural P. falciparum isolates and found that none of them had elevated PfRh1 gene copy numbers and that PfRh1 gene amplification arose in P. falciparum laboratory lines following establishment in culture (30). These findings suggest that P. falciparum employs variation in PfRh1 gene copy number to enhance its survival in culture and that P. falciparum lines with elevated PfRh1 gene copy numbers, such as W2mef, grow better and may compensate for the inhibitory factors in plasma better than those lines without an elevated PfRh1 gene copy number, such as 3D7 and K1.

In summary, our study contributes to the debate on growth inhibition assays as potential in vitro assays that correlate with in vivo protective immunity against malaria. In particular, it outlines the importance of considering the choice of parasite line(s) in the design of studies investigating associations between GIA and the risk of malaria.