Naturally Exposed Populations Differ in Their T1 and T2 Responses to the Circumsporozoite Protein of Plasmodium falciparum

Abstract

T-cell responses directed against the circumsporozoite protein (CS) of Plasmodium falciparum can mediate protection against malaria. We determined the frequency of T cells reactive to different regions of the CS in the blood of donors naturally exposed to P. falciparum by examining T1 (gamma interferon [IFN-γ] ELISPOT assay), T2 (interleukin 4 [IL-4] ELISPOT assay), and proliferative T-cell responses. The proliferative responses were weak, which confirmed previous observations. The responses to the CS in the IL-4 and IFN-γ ELISPOT assays were also weak (<40 responding cells per 10⁶ cells), much weaker than the response to the purified protein derivative of Mycobacterium tuberculosis in the same donors. Moreover, a response in one assay could not be used to predict a response in either of the other assays, suggesting that although these assays may measure different responding cells, all of the responses are weakly induced by natural exposure. Interestingly, the two different study populations used had significantly different T1 and T2 biases in their responses in the C terminus of the protein, suggesting that the extent of P. falciparum exposure can affect regulation of the immune system.

CD8⁺ (27) and CD4⁺ (19) T cells specific to the circumsporozoite protein (CS) of Plasmodium are important mediators of protection in animal models of malaria (16, 21, 30). Attempts to understand pre-erythrocytic immunity to infection in humans have led many researchers to focus on T cells specific for CS epitopes from P. falciparum CS.

T cells can be divided into two subsets called T1 and T2 depending on the cytokine secretion patterns (20; for a review see reference 23), and production of a T1 type of response, as measured by the cytokine gamma interferon (IFN-γ), can determine whether animals are protected in models involving both natural resistance to malaria and vaccination with irradiated sporozoites (30). However, the role of the T2 type of response, which includes production of interleukin-4 (IL-4), has not been addressed directly with regard to pre-erythrocytic immunity in humans. Since a T2 type of response can, on the one hand, inhibit many of the type T1 functions (for a review see reference 23) and, on the other hand, induce antibodies (for a review see reference 5), which can be protective (25), it may play a significant role in determining immunity against malaria. Indeed, during blood stage infection, in which the protective importance of antibodies is clear, production of IL-4 in response to the malaria antigen pf155/RESA correlates with protection against malaria better than IFN-γ production correlates with such protection (1). The levels of another T2 type of cytokine, IL-10, positively correlated with protection in a study of pre-erythrocytic immunity (14).

In two previous studies of CS the workers used proliferation and IFN-γ secretion to determine the pattern of T-cell reactivity to CS in humans (8, 12). In these studies, Good et al. identified two immunodominant regions in donors from The Gambia (12), while Doolan et al. found very little reactivity to any region of the CS in donors from Papua New Guinea (8). However, it is possible that a low level of reactivity as determined by proliferation and/or IFN-γ production may simply reflect a different cytokine secretion pattern in response to this protein and not necessarily an overall lack of reactivity. We therefore assessed the pattern of T1 and T2 reactivity to CS using the ELISPOT technique with naturally exposed donors in The Gambia. Since the ELISPOT technique measures rapid production of IFN-γ (18) in response to CS, which correlates with protection in animal models (22, 24, 29), this response is likely to be important in human immunity. We used a panel of overlapping peptides spanning the CS to identify novel T1- and T2-cell responses with no bias.

We present data here on the T1 and T2 reactivities to CS of two study populations from The Gambia that differed in terms of exposure to P. falciparum. The populations differed not only in the level of response, as has been observed previously, but also in the relative numbers of IL-4- and IFN-γ-secreting cells to the C terminus of the protein. The results of neither assay used correlated with the proliferative response, suggesting that the assays measured different sets of CS-reactive cells and hence that immunity may be dependent on the balance among these responses.

MATERIALS AND METHODS

Panel of peptides.

A panel of peptides covering all of the CS was synthesized by Research Genetics Inc. Five of the peptides tested came from P. falciparum strain 7G8 (peptides 4, 5, 6, 7, and 18) (National Center for Biotechnology Information accession no. C60657 [15]), while the rest of the peptides came from strain NF54 (National Center for Biotechnology Information accession no. S05428 [6]). The positions of the peptides on CS are shown in Fig. 1.

FIG. 1.

Sequence of the CS from NF54, showing the positions of the peptides used in this study. The shaded region is the region covered by the peptides to which Good et al. found the most frequent responses (Th2R and Th3R), and the letters and numbers are the designations of the peptides. The arrows indicate regions where peptides from strain 3D7 were used, while the dashed lines ending in solid circles indicate regions where peptides from strain 7G8 were used. The letters in parentheses indicate the sequences of 7G8 which differ from the 3D7 sequence.

Donors.

A total of 66 donors from two villages (Brefet and Bakadagy) in The Gambia were enrolled in the study. Both sets of donors were enrolled towards the end of the dry season in March 1998. The most recent measurements of the entomological inoculation rates (EIR) in the areas that contained these villages were made in 1994 by Thomson et al. (31), who estimated that the EIR around Brefet was 0.92 to 3.24 and that the EIR in Bakadagy was 7.75 to 11.15. The cohort from Brefet in the Western Division contained 13 males and 17 females who were 18 to 65 years old, while the cohort from Bakadagy in the Upper River Division contained 36 males who were 18 to 45 years old. ELISPOT assays were performed for peptides from the C terminus with the samples from Bakadagy, while all three assays were performed in parallel with the Brefet samples.

Preparation of peripheral blood mononuclear cells.

Venous blood was collected into heparinized tubes, and peripheral blood mononuclear cells were separated by Ficoll gradient centrifugation (Lymphoprep; Nycomed) and used in the ELISPOT and proliferation assays.

ELISPOT assays.

ELISPOT assays were performed with an IFN-γ or IL-4 ELISPOT kit (Mabtech) according to the manufacturers' recommendations. Briefly, nitrocellulose-backed 96-well plates (MAIP S45; Millipore) were coated with capture antibody (15 μg ml⁻¹ in phosphate-buffered saline [PBS]) overnight at 4°C and then blocked with RPMI medium containing 10% fetal calf serum for at least 1 h. Then 4 × 10⁵ cells in RPMI medium containing 5% human AB serum, 100 U of penicillin/streptomycin per ml, and 2 mM l-glutamine were added to each well, and stimulants were added at final concentrations of 25 μg ml⁻¹ (for peptides) and 10 μg ml⁻¹ (for the purified protein derivative of Mycobacterium tuberculosis [PPD] [Statens Serum Institut] or phytohemagglutinin [Sigma]). The plates were cultured for 18 to 20 h for the IFN-γ assay and for 24 h for the IL-4 assay and then washed and incubated overnight at 4°C with the biotinylated detector antibody at a concentration of 1 μg ml⁻¹ in 50 μl of PBS. The plates were washed again and incubated for 1 h with 1 μg of streptavidin alkaline phosphatase per ml and then developed with the precipitating alkaline phosphatase substrate (Bio-Rad), and spots were counted by using a microscope. A person was considered to react to a stimulant if the probability of a spot appearing in the test well was significantly different (P ≤ 0.05) from the probability of a spot appearing in the control well, assuming a Poisson distribution.

Proliferation assays.

A total of 1 × 10⁵ cells were added to each well of a U-bottom 96-well tissue culture plate in 200 μl of proliferation medium (minimal essential medium [αMEM] containing 10% human AB serum, 2 mM l-glutamine, and 100 U of penicillin/streptomycin per ml) and incubated for 6 days at 37°C in 5% CO₂-95% air. Then 100 μl of supernatant was removed and replaced with proliferation medium containing 10 μCi of [³H]thymidine per ml. The cells were harvested after an additional 16 h of incubation. A stimulation index (SI) was calculated by dividing the geometric mean for the test cultures by the geometric mean for the background cultures (4), and a response was considered positive if the SI was greater than 2.

Statistics.

Results are expressed as means ± 1 standard error, and P values were calculated by using the nonparametric Mann-Whitney U test to compare groups or Spearman's test to calculate correlation coefficients.

RESULTS

Differential epitope mapping by proliferation and IFN-γ ELISPOT assay.

To determine whether T-cell epitopes capable of stimulating proliferation could also induce rapid IFN-γ and IL-4 production, we tested overlapping peptides covering the length of CS for different effector functions in adult Gambians. Using the standard definition (4) that a positive proliferative response is an SI of >2, we confirmed that most of the proliferative responses were responses to the variable Th2R region of CS (12). However, the responses which we observed were much weaker than those observed previously by Good et al., since only four of these responses would have been considered positive in their analysis (Fig. 2). The pattern of responses determined by the rapid IFN-γ ELISPOT assay, on the other hand, was very different from that determined by proliferation analysis (Fig. 3). Whereas proliferation analysis detected an immunodominant region, the limited number of IFN-γ ELISPOT responses were distributed throughout the CS, and only one of six comparable IFN-γ responses coincided with a proliferative response. Even when a response was present, the average number of CS-reactive IFN-γ-secreting cells (23 ± 11 cells per 10⁶ cells) was much less than the number of cells reactive to PPD (>104 ± 80 cells per 10⁶ cells; P < 10⁻⁹) and the number of cells previously reported for responses to peptides from blood stage antigen Pf155/RESA in The Gambia (210 ± 115 cells per 10⁶ cells; P < 10⁻⁷) (9). These data suggest that natural exposure to P. falciparum does not readily induce CS-reactive IFN-γ-producing cells and that these cells are different from the cells that proliferate.

FIG. 2.

Proliferative responses to peptide pairs. A total of 100,000 cells were incubated with pairs of peptides at concentrations of 25 μg ml⁻¹ or with PPD at a concentration of 10 μg ml⁻¹. On day 6, 1 μCi of [³H]thymidine was added to each well for 16 h. Experiments were performed in triplicate. SIs of greater than 2 are shown, and the geometric means of the background counts are indicated. Shading indicates that the response was positive when the method of calculation described by Good et al. (12) was used. Briefly, the log of the counts was determined and a mean and a standard deviation were calculated for the PBS counts; a result was positive if it differed by more than 3 standard deviations from the PBS mean. F, female; M, male.

FIG. 3.

Ex vivo IFN-γ ELISPOT responses to peptide pairs. A total of 400,000 cells were cultured for 18 to 20 h with peptides at concentrations of 25 μg ml⁻¹ or with PPD at a concentration of 10 μg ml⁻¹. A response was considered positive if the probability of a spot appearing in a well was greater than the probability of a spot appearing in the control well (P < 0.05), assuming a Poisson distribution of spots. The values are the absolute numbers of IFN-γ secreting cells in the wells, and results are shown only for the background responses and the responses which were considered positive (shaded). ND, not determined; F, female; M, male.

T2 responses to CS.

In addition to IFN-γ production and proliferation, T cells may also respond to CS by secretion of T2 cytokines. To determine which regions of CS may induce this type of response, we tested donors from Brefet and Bakadagy for rapid secretion of IL-4 using the ELISPOT assay. A total of 9 of 54 donors had a response, and 29 of the 768 responses tested were positive to part of CS as determined by this assay (Fig. 4). Like the IFN-γ ELISPOT responses, the IL-4 responses were found to be responses to many regions of the CS. Interestingly, none of the positive IL-4 responses coincided with a positive IFN-γ response, while only 1 of 22 responses coincided with a proliferative response, suggesting that the IL-4 ELISPOT assay measured yet another set of reactive cells.

FIG. 4.

Ex vivo IL-4 ELISPOT responses to peptide pairs. A total of 400,000 cells were cultured for 24 h in medium containing peptides at concentrations of 25 μg ml⁻¹ or with phytohemagglutinin (PHA) at a concentration of 10 μg ml⁻¹. A response was considered positive if the probability of a spot appearing in a well was greater than the probability of a spot appearing in the control well (P < 0.05), assuming a Poisson distribution of spots. The values are the absolute numbers of IL-4-secreting cells in the wells, and results are shown only for background responses and the responses which were considered positive (shaded). ND, not determined; F, female; M, male.

T1/T2 ratios differ in two Gambian populations.

When responses to the C terminus of CS were examined, significant differences were observed between donors from the two villages involved in the study. In the Brefet donors, the response was biased towards IL-4 production; the frequency of IL-4 responses (28 of 527 tests [5.3%]) was higher than the frequency of IFN-γ ELISPOT responses (10 of 540 tests [1.85%]) (P < 0.005) (Fig. 3 and 4). By contrast, in the Bakadagy donors, the responses were biased towards IFN-γ production; 1 of 241 (0.4%) responses were positive for IL-4, and 13 of 361 (3.2%) responses were positive for IFN-γ (P < 0.01). When the responses to each of the cytokines individually were considered, the Brefet donors had a higher frequency of IL-4 responses than the Bakadagy donors (P < 0.005), and although the difference between the proportion of positive IFN-γ responses in the Bakadagy donors and the proportion of positive IFN-γ responses in the Brefet donors was not significant, the proportion of positive responses was higher in the Bakadagy group. To confirm these comparisons taking into account the size of the responses, as well as the frequency of positive responses, we tested for differences between villages, sexes, and peptides by using a generalized estimating equations procedure. This procedure took into account the nonnormal distribution of the ELISPOT responses since we used the robust estimates of the model's coefficients and standard errors. The procedure is adequate when it is used with repeated ELISPOT measurements for an individual (32), and by including the background values for the IL-4 measurements and IFN-γ measurements as offsets, we were able to account for the effect of differences between backgrounds for each individual. The parameters of the model are shown in Table 1. No estimates are shown for the peptide pair A+B since the effects of the other peptides were quantified against this pool. This choice of reference parameter was arbitrary, and the same conclusions would have been reached if any other peptide pool had been used as the reference parameter. The intercept of the model was negative, indicating that there was a significant difference between IL-4 and IFN-γ production. The Z scores for each pair of peptides and for sex were all less than 3, indicating that there was not a significant effect for any pair of peptides or for sex on the ratio of IL-4 production to IFN-γ production. There was, however, a significant effect for village, indicating that in the two areas studied there are significant differences in the levels of IL-4 and IFN-γ production.

TABLE 1.

Parameter estimates for differences between IL-4 and IFN-γ ELISPOT responses

Parameter	Coefficient estimate	SE	Z score	P value
Intercept	−35.45	7.135	−4.97	10−6
C+D	−0.267	0.548	−0.487	0.626
E+F	−0.467	0.632	−0.738	0.461
G+H	−0.067	0.705	−0.095	0.924
I+J	−0.567	0.520	−1.090	0.276
K+L	0.233	0.541	0.431	0.666
1	0.638	0.619	1.031	0.303
2+3	0.04	0.691	0.059	0.953
4+5	0.739	0.733	1.01	0.312
6+7	1.199	0.669	1.792	0.073
8+9	−0.294	0.664	−0.443	0.658
10+11	0.293	0.639	0.460	0.646
12+13	0.120	0.669	0.179	0.858
14+15	0.402	0.620	0.648	0.517
16+17	−0.564	0.703	−0.802	0.423
18+19	−0.171	0.838	−0.203	0.839
20+21	1.30	1.045	1.244	0.213
22+23	1.653	0.956	1.729	0.084
Sex	1.933	1.350	1.431	0.152
Village	15.66	3.497	4.478	10−5

DISCUSSION

Different T-cell reactivity patterns were found for the CS in a naturally exposed population depending on the different effector functions studied. This may reflect the fact that proliferation provides a cumulative measure of both antigen-specific division and nonspecific division in culture, while the ELISPOT assays are more direct measures of specific T-cell activity.

While the proliferation assay confirmed the immunodominant region found by Good et al. in 1988, the pattern of reactivity determined by both IFN-γ and IL-4 ELISPOT assays was very different, with no clear focus of reactivity. Responses to both variant and invariant regions of the protein were found. It is possible that there was a weakly immunodominant region for the cells detected by the ELISPOT assay and that we could not detect it due to the low frequency of responses in this population. However, if such a region exists, it is not overridingly immunodominant since it does not prevent observation of responses to other regions of the protein (Fig. 3 and 4).

The responses in all the assays were weak compared to the PPD responses and to previously described results for peptides from other malaria antigens (9), suggesting that natural exposure to P. falciparum does not induce strong T-cell reactivity to the CS in these populations. Four other studies have examined T-cell responses either to peptides covering all of the CS (8, 12) or to the immunodominant regions (10, 26). The proliferative responses observed in our study were comparable to those observed by Esposito et al. (10), Doolan et al. (8), and Riley et al. (26), although they were weaker than those observed by Good et al. (12). It is unlikely that this low level of T-cell reactivity is due to sequestration of the T cells to sites of infection since our samples were taken 4 months after the end of the malaria transmission season and the rate of parasitemia in both of the populations studied was low (0% in Brefet and 17% in Bakadagy). It may be necessary to improve the sensitivity of the T-cell assays to quantify the effects of these low T-cell responses on protection against malaria.

Neither IFN-γ production nor IL-4 production correlated with proliferation, so different T1/T2 ratios are unlikely to explain the difference between the proliferative responses and IFN-γ production observed in this study and reported previously (8, 10). Interestingly, considering the number of peptide-specific cells in all the tests, there was a weak (correlation coefficient, 0.182) but significant (P < 10⁻⁵) positive Spearman's rank correlation between IL-4 production and IFN-γ production, indicating that although the assays measured different sets of cells, the cells were probably not mutually inhibitory.

In the C terminus of CS, the T1/T2 ratios were different for the two populations described here; while IFN-γ production predominated in Bakadagy, IL-4 production predominated in Brefet. Thus, the two populations differed not only in the frequency of the responses but also in the types of responses. Several factors have been associated with altered T1/T2 ratios; these factors include age (for a review see reference 11), sex (3, 28), ethnicity (8), level of parasite exposure (10), and frequency of mosquito feeding by Culex pipiens and Aedes aegypti (33). Our study populations included people of similar ages (18 to 65 years for Brefet and 18 to 45 years for Bakadagy) and of the same ethnicity (both populations are Mandinka). Sex differences also cannot explain the observed T1/T2 ratio differences since strong T2 responses were seen when just the males in Brefet were tested for T1/T2 ratio bias (16 of 234 [4.7%] IL-4 tests were positive, compared to 1 of 234 [0.4%] IFN-γ tests that were positive [P < 0.001]). As the frequency of mosquito feeding increased in mouse studies, the response became biased towards the T2 response (33). However, in our study, the T2 responses were stronger in the population with the lower biting frequency (31) so that feeding frequency is unlikely to explain the differences which we observed. The level of parasite antigen exposure may also influence the T1/T2 ratio; increasing the antigen load at low doses of antigen can lead to more IFN-γ production (13), while at high doses it leads to higher IL-4 production (7, 13). Bakadagy is in a region with a higher malarial load than Brefet (31) and is more biased towards a T1 type of response, which fits the hypothesis that the antigen load due to natural exposure to P. falciparum is low. Parasite dose has been shown to regulate the T1/T2 ratio in animal models of other infections, such as Leishmania infections (7, 17) and Trichuris muris infections (2). However, in both of these models, the T1/T2 ratio decreased with increasing infectious dose, which is different from the change that we observed, probably because the overall antigen burdens in these model infections were much higher than that resulting from natural exposure to P. falciparum sporozoites.

Thus, P. falciparum can induce different types of peripheral CS-specific T-cell immune responses, albeit at low precursor frequencies. The low level of reactivity is most likely due to the low level of antigen exposure. It is possible that variations in exposure may also explain variations in the T1/T2 ratio bias of naturally induced immune responses. Since there is no correlation between either of these assays and proliferation, the differences between IFN-γ production and proliferation cannot be explained simply by the existence of T2 cells. Rather, there are many different types of reactive cells in each donor, and the finding that the results of the assays used are not correlated improves the chances of finding a correlate of protection in prospective studies, which should significantly aid the process of vaccine design. Indeed, cells reactive in the ELISPOT and proliferation assays can be differentially regulated (Flanagan et al., submitted for publication), and it may be possible to target the generation of one type of immune response without the other. The complex interaction between subsets is more likely to explain immunity than measurements obtained in individual assays, and in the future investigation of the subsets may provide invaluable information for the design of vaccines against malaria.