Dimorphic Plasmodium falciparum merozoite surface protein-1 epitopes turn off memory T cells and interfere with T cell priming


  • Edwin A. M. Lee,

    1. Molecular Immunology Group, Weatherall Institute of Molecular Medicine, Nuffield Department of Medicine, University of Oxford, John Radcliffe Hospital, Oxford, UK
    Search for more papers by this author
  • Katie L. Flanagan,

    1. Molecular Immunology Group, Weatherall Institute of Molecular Medicine, Nuffield Department of Medicine, University of Oxford, John Radcliffe Hospital, Oxford, UK
    Search for more papers by this author
  • Gabriela Minigo,

    1. Vaccine and Infectious Diseases Unit, The Burnet Research Institute, incorporating The Austin Research Institute, Austin Hospital Campus, Victoria, Australia
    Search for more papers by this author
  • William H. H. Reece,

    1. Molecular Immunology Group, Weatherall Institute of Molecular Medicine, Nuffield Department of Medicine, University of Oxford, John Radcliffe Hospital, Oxford, UK
    Search for more papers by this author
  • Robin Bailey,

    1. Medical Research Council Laboratories, Fajara, Banjul, The Gambia
    Search for more papers by this author
  • Margaret Pinder,

    1. Medical Research Council Laboratories, Fajara, Banjul, The Gambia
    Search for more papers by this author
  • Adrian V. S. Hill,

    1. Molecular Immunology Group, Weatherall Institute of Molecular Medicine, Nuffield Department of Medicine, University of Oxford, John Radcliffe Hospital, Oxford, UK
    Search for more papers by this author
  • Magdalena Plebanski

    Corresponding author
    1. Molecular Immunology Group, Weatherall Institute of Molecular Medicine, Nuffield Department of Medicine, University of Oxford, John Radcliffe Hospital, Oxford, UK
    2. Vaccine and Infectious Diseases Unit, The Burnet Research Institute, incorporating The Austin Research Institute, Austin Hospital Campus, Victoria, Australia
    • Vaccine and Infectious Diseases Unit, The Austin Research Institute, Studley Road, Victoria 3084, Australia, Fax: +61-3-92870643
    Search for more papers by this author


The leading blood-stage malaria vaccine candidate antigen, Plasmodium falciparum merozoite surface protein-1 (MSP-1) occurs in two major allelic types worldwide. The molecular basis promoting this stable dimorphism is unknown. In this study, we have shown that allelic altered peptide ligand (APL) T cell epitopes of MSP-1 mutually inhibited IFN-γ secretion as well as proliferation of CD4+ T cells in 27/34 malaria exposed Gambian volunteers. Besides this inhibition of malaria-specific immunity, the same variant epitopes were also able to impair the priming of human T cells in malaria naive individuals. Epitope variants capable of interfering with T cell priming as well as inhibiting memory T cell effector functions offer a uniquely potent combination for immune evasion. Indeed, enhanced co-habitation of parasites bearing such antagonistic allelic epitope regions was observed in a study of 321 West African children, indicating a survival advantage for parasites able to engage this inhibitory immune interference mechanism.


altered peptide ligand


merozoite surface protein 1


parasite rates


tetanus toxoid


The Plasmodium falciparum merozoite surface protein 1 (MSP-1), a highly immunogenic target of host-protective responses in malaria, presents a unique pattern of dominant allelic variation manifesting itself in two main types, the MAD-20 and the WELLCOME strain 14. MSP-1 allelic polymorphism may be one of the oldest eukaryotic polymorphisms, estimated at up to 35 million years of age, and it has been argued that the MSP-1 gene must be under positive balancing selection exerted by the host immune system to maintain this polymorphism 5. Why and how malaria blood-stage antigens maintain allelic dimorphisms has been unclear. Dimorphic variants may have arisen earlier in hominoid evolution and then been maintained in certain human populations, due to a survival advantage of co-habiting allelic types. If this is true, then the consequences of allelic polymorphism must be considered for future malaria vaccine design and development 6.

In Papa New Guinea, a recombinant vaccine expressing MSP-2 in the 3D7 allelic form reduced the prevalence of parasites carrying the MSP2–3D7 allelic form, whereas a higher incidence of morbidity was found to be associated with the FC27-type allelic parasites 7. Given this lack of cross-protective immunity, it has been suggested that vaccines should incorporate both allelic or even multiple forms of polymorphic antigens to provide wider coverage of protection in the field. In support of this is the observation that parasites expressing cross-reactive epitopes have a disadvantage in their survival 8. However, some mixtures of malaria antigens compromise immunogenicity to the individual antigen mixture components 9, and studies of pre-erythrocytic malaria antigens, such as the circumsporozoite (CS) protein, show that simultaneous stimulation with certain naturally occurring T cell epitope variants turns off effector T cell immunity, resulting in enhanced parasite survival 10. Down-regulation of T cell responses by epitope variants that vary in only one or few amino acids (aa), so-called altered peptide ligands (APL), has been classically described for T cell clones expressing a specific T cell receptor structure 1012. Recent studies in autoimmune diseases, however, indicate that APL could also critically modulate polyclonal T cell responses in vivo11. Although the assays are technically similar, we call the inhibitory effect of epitope variants on polyclonal immunity ‘immune interference’, to conceptually distinguish it from APL antagonism that temporarily down-modulates the effector functions of T cell clones in vitro10, 12. Mathematical modeling further indicates that the ability to modulate polyclonal T cell responses, both at the level of naive T cells and as those induced by natural exposure to malaria parasites, would be required for potential inhibitory mechanisms to have a significant effect on a parasite population structure 10.

The frequencies of the dimorphic allelic forms of MSP-1 appear to differ in studies from East and West Africa. In Sudan and Zimbabwe, and also in Thailand and Papua New Guinea, MAD-20 frequencies are consistently reported to be 25–50% 1214. In The Gambia, MAD-20 frequencies greater than 90% are stably found 15. We have reported previously that, surprisingly, T cell reactivity by IFN-γ ELISPOT in malaria-exposed Gambians to specific MSP-1 epitopes representing allelic pairs failed to reflect the local MAD-20 parasite strain prevalence 16. In this study, we tested 22 malaria-exposed, but parasite-negative, Gambian adults for IFN-γ responses to a panel of 11 MSP-1-derived peptides from the MAD20 strain and its allelic variant from the WELLCOME strain 16. We showed that each peptide induced specific IFN-γ secretion in >18% of donors, demonstrating that these were natural epitopes. Although the MSP-1 allelic strain distribution in >200 Gambian children with clinical malaria was 88% of MAD20 strain and 12% of WELLCOME strain 16, this parasite strain distribution was not reflected in the IFN-γ responses. Based on the differential recognition of their allelic variants by T cells, we chose three of these epitope regions for further studies. We have reported 16 that reactivity to aa 310–328 (peptides M5/M6) reflected the high MAD20 strain prevalence in West Africa, with 65% and 35% of donors reacting to the MAD-20 and WELLCOME forms of the epitope, respectively. Region aa 344–357 (peptides M3/M4) demonstrated equal levels of allelic reactivity. Reactivity to region aa 1538–1550 (peptides M7/M8) showed a surprising pattern, with only 25% of individuals responding to the common (95% prevalence) MAD20 variant but 75% to the rare (5% prevalence) WELLCOME variant. The lack of correlation of allelic T cell reactivity and parasite strain prevalence led us to the hypothesis that some form of active immune suppression might be responsible for this phenomenon.

Here we show that the allelic MSP-1 epitopes that had low T cell reactivity to dominant alleles in population studies have the capacity to turn off polyclonal T cell responses in both naive and malaria exposed (West African) individuals. They do so by mutually interfering with T cell priming and T cell effector functions. Naive T cell expansion and conversion into conventional memory T cells, as well as central (proliferation) and effector (IFN-γ secretion) memory T cell effector functions were also affected. In liver-stage malaria, protective immunity can be associated with specific central or effector memory T cell subpopulations 17, 18. The existence and selection of these novel multi-level (naive, memory-central and memory-effector) antagonistic T cell epitopes within MSP-1 therefore offers an unusual and powerful immune evasion strategy for the malaria parasite. This in turn is likely to provide positive selection pressure operating to promote co-habitation and maintenance of MSP-1 allelic forms in blood-stage malaria. Indeed, the analysis of strains found in human peripheral blood in Gambian volunteers presented in this study shows a highly significant increase in co-infection by parasites bearing these antagonistic allelic epitopes.

As well as providing a mechanism that could possibly explain the distribution of dimorphic blood-stage parasites in endemic populations, these results suggest that the inclusion of both alleles of MSP-1 in a vaccine would impair rather than promote the induction of blood-stage protective immunity. The optimization of the use of largely conserved MSP-1 regions, such as the previously validated immunogenic 19-kDa fragment 19, or other largely conserved blood-stage proteins or their regions, may provide a safer alternative to provide broadly protective immunity by vaccination in malaria endemic areas.


Allelic immune interference of MSP-1-specific IFN-γ responses of West African donors

In a previous study we observed that the T cell reactivity to dimorphic MSP-1 epitope regions in malaria-exposed Gambians does not necessarily reflect the prevalence of the MAD20 strain (95% prevalence) over the WELLCOME strain (5% prevalence) 16. Three allelic peptide epitope pairs were selected for further study: allelic MAD20/WELLCOME peptide pairs M3/M4, M5/M6 and M7/M8 (Fig. 1). The peptide sequences of MSP-1 allelic variants used in this study are shown in Table 1. Out of these peptides pairs only one pair (M5/M6) reflected the MAD20 allelic prevalence with reactivity of 65% and 35% in Gambian donors, respectively 16. Reactivity to M3/M4 was equally distributed amongst donors, while higher reactivity to M8 over M7 was observed (75% and 25%, respectively) 16.

Figure 1.

Allelic immune interference of ex vivo IFN-γ responses by MSP-1 variant epitopes. PBMC from two adult Gambians were stimulated separately with (A) aa 344–357 variants, M3 (MAD-20) and M4 (WELLCOME) (donor 1), or (B) aa 1538–1550 variants, M7 (MAD-20) and M8 (WELLCOME) (donor 2) in a 16-h ex vivo IFN-γ-release assay. Results indicate the average SFU/million for two wells. Both individuals tested gave expected PPD responses (data not shown). Donor 1 responded to peptide M4 with peptide concentrations greater than 30 µg/mL giving significant IFN-γ responses (p<0.05). Donor 1 did not respond to peptide M3 at any of the peptide concentrations. Donor 2 responded to peptide M8 at peptide concentrations greater than 10 µg/mL, giving significant IFN-γ responses (p<0.05). Donor 2 did not respond to peptide M7 at any of the peptide concentrations. The results represent two similar assays. (C) In donor 1, the impact of the presence of the allelic variant M3 on the M4 IFN-γ response was assessed in an ELISPOT immune interference assay 20, 21. Index peptide was pre-pulsed first at 30 µg/mL onto PBMC and the other variant 1 h later at the peptide concentrations indicated (10–90 µg/mL). Variant peptide concentration of 0 µg/mL indicates that no variant peptide was added. Significant inhibition (>30%) 10, 21, 39 of M4 responses was observed with variant peptide M3 concentration from threefold less than index peptide concentration. (D) In donor 2, the impact of the presence of the allelic variant, M7 on the M8 IFN-γ response was assessed in an ELISPOT immune interference assay. Significant inhibition (>30%) of M8 IFN-γ responses was observed with variant peptide M7 from a concentration of more than 30 µg/mL. The results represent two similar assays.

Table 1. Peptide sequences of MSP-1 allelic variants used in this study
VariantStrainPositionAmino acid sequencea)
  1. a) Variant amino acids are in bold.


One possible explanation for the unexpected allelic T cell reactivity pattern to M3/4 and M7/8 was that some of the reactivity to these epitopes was being suppressed. In pre-erythrocytic malaria, CS protein variants can turn off some types of T cell responses to each other by APL antagonism and through immune interference with the stimulation of T cell responses in PBMC cultures 10, 20, 21. We thus investigated whether a similar mechanism is involved in the case of allelic MSP-1 variants in malaria-exposed donors. PBMC from adult Gambians were stimulated with the allelic peptide epitopes in immune interference assays. Fig. 1 shows that IFN-γ responses to one allelic variant (M4 in donor 1 and M8 in donor 2) could be inhibited by their allelic variant (M3 or M7, respectively) in two donors. In both of these specific donors, stimulation with M3 or M7 alone failed to induce IFN-γ secretion (Fig. 1A, B), indicating that the allelic variants were not cross-reactive. To rule out competition for HLA binding, we tested the peptide affinities to three HLA-DR molecules common in the Gambia (DRB1*1302, DRB1*0701 and DRB1*0101) and found that peptides M8 and M4 showed affinities similar to or higher than peptide M7 and M3, respectively (Table 2). This indicates that the inhibitory effect of the latter on reactivity to the former was unlikely to be due to competition for MHC binding promoted by stronger HLA-DR binding by M7 or M3.

Table 2. IC50 values of MSP-1 allelic epitope variants in peptide binding competition assay with purified HLA-DRB1*0101, DRB1*0701 and HLA-DRB1*1302a)
PeptideStrainPeptide binding to HLA-DR allele shown as IC50a)
  1. a) IC50 represents the concentration (in 10–3 mg/mL) of unlabeled competitor peptide required for 50% inhibition of the binding of biotinylated CLIP peptide as described in Section 4. Results are mean values of triplicates of one of two representative experiments. SD was <10% of the mean.


Table 3 summarizes results from 34 Gambian malaria-exposed donors similarly tested for immune interference by any of the three allelic variant peptide pairs M5/6, M3/4 and M7/8. The M5/6 allelic pair was not antagonistic in any of the M5 or M6 reactive donors (0/11). In contrast, immune interference was observed in the majority of the IFN-γ responders to M3 or M4 (13/17) and M7 or M8 (14/17). The mean ± SE inhibition of IFN-γ responses in immune interference assays was 47.3 ± 3.9% for M3 or M4, and 53.6 ± 4.6% for M7 or M8. A proportion of responses to M3 or M4 (23%) and M7 or M8 (17%) were neither cross-reactive nor non-antagonistic (Table 3). Thus, within the Gambian population sampled in Table 3, and as exemplified in Fig. 1, individuals could respond to either allelic variant without reacting to its allelic pair. Fig. 2 further exemplifies from the summary data in Table 3 that immune interference was mutual. Thus, in a MAD20 allele responding donor, for example, M7, the WELLCOME variant M8 turned off M7 responses (Fig. 2F), whereas in another donor M8 was the index peptide and M7 interfered with responses against it (Fig. 2H).

Table 3. Summary of antagonistic IFN-γ responses to MSP-1 allelic variant peptide pairs in malaria-exposed Gambian donors
Allelic epitopes tested for mutual immune interference
M3 and M4M5 and M6M7 and M8
  1. a) Percentages of positive reacting donors tested who gave significant IFN-γ responses to each MSP-1 epitope region (p<0.05).

  2. b) Percentages reflect the number of positive reacting donors who gave >30% inhibition of the IFN-γ response to one allelic epitope in the presence of its allelic variant in a standard antagonism assay 21.

Respondersa)17/30 (57%)11/25 (44%)17/28 (61%)
Inhibition by allelic epitopesb)13/17 (77%)0/11 (0%)14/17 (83%)
No inhibition4/17 (23%)11/11 (100%)3/17 (17%)
Figure 2.

Immune interference in the stimulation of IFN-γ release and proliferative T cell responses by costimulation with allelic variants in the same malaria-exposed individuals. Proliferation assays were set up at the same time as ex vivo IFN-γ ELISPOT immune interference assays in four malaria-exposed Gambian donors (proliferation panels A, C, E, G, I; ELISPOT panels B, D, F, H, J). All individuals tested gave strong PPD responses in both ELISPOT and proliferation assays (data not shown). The ELISPOT IFN-γ immune interference experiments were performed as described above (Fig. 1). Proliferation was monitored by [3H]thymidine incorporation over days 5–10. Proliferation results shown are peak mean cpm ± SE of triplicate samples. In all cases, APL costimulation caused inhibition of both proliferative and IFN-γ responses for allelic pairs M3/4 (A–D) and M7/8 (E–H). By contrast, (I) and (J) show representative results of three donors reactive to M5 or M6, demonstrating that costimulation with this allelic pair did not cause immune interference. Significant proliferative and IFN-γ responses over background (BG) by ELISPOT are marked as follows: *p<0.05, **p<0.005. Further statistical mathematical analyses for ELISPOT are detailed in 18. Significant inhibition of proliferative or IFN-γ responses are given by #p<0.05 and ##p<0.005).

Importantly, both WELLCOME and MAD-20 forms of the allelic epitopes could inhibit responses to the other variant, indicating that differences in binding affinity to MHC are irrelevant for the direction of immune interference (given that it is seen in both directions). In summary, MSP-1 allelic variant peptide pairs M3/4 and M7/8 initially selected on the basis of an unexpected T cell reactivity bias towards the uncommon WELLCOME variant allelic epitopes were found to be potent mutual antagonists in assays of ex vivo IFN-γ secretion.

Immune interference affects both proliferative T cell responses and IFN-γ production

Previous studies have shown that MSP-1 proliferative responses correlated with protection against clinical malaria 22. Immune interference between allelic MSP-1 epitopes was further studied in assays of proliferation, set up in parallel with ex vivo IFN-γ ELISPOT immune interference assays. Four malaria-exposed donors were studied for M3/4 and M7/8 immune interference by IFN-γ ELISPOT in parallel with proliferative responses. Fig. 2 illustrates that by both assays responses were susceptible in the same individuals. Fig. 2I and J show representative data from three donors reactive to M5/6, for whom no immune interference was observed in either assay to this allelic pair. Our data show that both IFN-γ and proliferative responses were both mutually inhibited by allelic variant epitopes M3/4 and M7/8, but not M5/6.

Importantly, similar immune interference results were obtained when peptides M3/4 or M7/8 were loaded onto the same PBMC population to serve as APC, or when presenters were pulsed with each peptide separately (in separate tubes), washed and then mixed with the responders. Results from this latter protocol are illustrated in Fig. 3, and further formally exclude the possibility that the soluble peptides compete for MHC binding on the APC.

Figure 3.

Primary proliferative responses to allelic variant MSP-1 peptide epitopes. (A) Proliferation kinetics of a primary response to allelic variant peptides, M3 and M4 in an HLA-DRB1*1302 malaria-naive donor. Proliferation to M3 or M4 was induced from PBMC by mixing peptide pre-pulsed (50 µg/mL) irradiated presenters with untreated PBMC used as responders at a ratio of 1:4. The influence of the presence of the variant on the proliferative response of the index peptide was assessed in a proliferative immune interference assay 20, 21. The index peptide and variant peptide were pre-pulsed (50 µg/mL) separately on irradiated autologous PBMC in separate Falcon tubes, washed and then mixed together at a 1:1 ratio with untreated PBMC or with each other. Thus, M3 signifies: 1 × 106 responder T cells plus 0.25 × 106 M3 pulsed and washed irradiated presenters (M3) and 0.25 × 106 untreated and washed irradiated presenters (Neg) or together M3+Neg. M3/M4 in the legend signifies: 1 × 106 responder T cells plus 0.25 × 106 M3 pulsed and washed presenters (M3) and 0.25 × 106 separately M4 pulsed and washed presenters (M4) or together M3+M4. Neg+Neg corresponds to the situation in which 0.25 × 106 untreated irradiated presenters (Neg) were mixed with 0.25 × 106 untreated irradiated presenters (Neg). Proliferative responses were monitored by [3H]thymidine incorporation over days 5–10. The results shown are the mean cpm ± SE of triplicate samples. One representative experiment of three is shown. Similar results were obtained using 50 or 100 µg/mL for direct addition or the pre-pulse, and examples of both are shown. (B) Proliferative responses of an HLA-DRB1*0101 malaria-naive donor to peptides M3 and M4 after direct addition at 100 or 50 µg/mL alone or together, or when pulsed in separate tubes at 100 µg onto irradiated PBMC for 2 h. After washing, they were used to stimulate untreated PBMC in culture alone or in combination as described above. Triplicate cultures were assessed on days 5–10 and peak proliferative responses are shown as mean cpm ± SEM representative of three similar assays. (C) Primary proliferative responses to peptide M4 in an HLA-DRB1*0101 malaria-naive donor were inhibited by its allelic variant M3 but not by peptide M5. Proliferation to M4 was induced from PBMC by separately mixing M4 pre-pulsed irradiated PBMC with untreated, M3 or M5 peptide pre-pulsed irradiated PBMC at a 1:1 ratio, and used to stimulate autologous PBMC as described above. The peptides were pre-pulsed at 50 µg/mL. Triplicate cultures were assessed on days 5–10 and peak proliferative responses are shown as mean cpm ± SEM representative of two similar assays. (E) Immune interference between allelic peptides M7 and M8 in primary proliferative responses in a HLA-DRB1*0101 malaria-naive donor. Conditions were set up as above. Triplicate cultures were assessed on days 5–10 and peak proliferative responses are shown as mean cpm ± SEM representative of two similar assays.

Population structure of MSP-1 allelic variant epitopes

Immune interference by MSP-1 allelic variant epitopes could provide an in vivo immune evasion mechanism for blood-stage parasites and promote co-habitation of APL-bearing strains. Table 4 shows the calculated expected frequencies of single and mixed infections, assuming random mixing for different values of the parasite infection rate (the proportion of the population with malaria parasites in their blood). The actual parasite rate (PR) in the locality of The Gambia from which the malaria cases were sampled (from August to November) was in the range of 10–50% 10. At all PR, the observed distribution of MSP-1 strains of antagonistic allelic variant pairs M3/M4 differed markedly from that expected under random mating without selection. In particular, co-habitation of M3/M4 was observed more than twice as frequently as would be expected from their individual prevalence in the host population (at 20% PR) (χ2 analysis, M3/4: p<10–8). This unusual parasite population structure of MSP-1 alleles is consistent with an effect of APL antagonism and immune interference 10, 21.

Table 4. Frequency of infection with MSP-1 allelic variant regions corresponding to peptides M3 (MAD-20) and M4 (WELLCOME) in 321 Gambian children with severe malaria
Expected at PRa)
MSP-1 phenotypeObserved100%50%20%10%
  1. a) The expected frequency of single (M3 only, M4 only) and mixed (M3 and M4) infections with each strain is shown for the different PR in the population. Actual PR in the locality from which the malaria cases were sampled (from August to November) is in the range of 10–50% 10. The expected values were calculated assuming a binomial distribution of observed values with the proportion of total population infected being 1–PR 10. (Observed M3 plus M4 frequency compared with expected M3 plus M4 frequency at PR between 10 and 100%: statistical significance: *χ2=>1–10, **χ2=10–50, *** χ2=50–100.) Note, given the high recombination rate of P. falciparum in natural populations, with linkage disequlibriun spanning only hundreds of base pairs in the MSP-1 gene, results from co-habitation analysis for a single allelic T cell epitope pair region cannot be extrapolated to other allelic pairings 41.

M3 only273272276277278
M4 only61233539
M3 and M442482294
p value0.14*0.0003**1 × 10–8**1 × 10–8***

Aborted T cell priming occurs by costimulation with MSP-1 allelic variant peptides

MSP-1-specific memory T cells that are susceptible to immune interference might be generated in vivo by exposure to MAD20 and WELLCOME strains. The population data described above suggests that a significant proportion of individuals may be simultaneously exposed to different immunomodulatory variants during co-infections. It is not known how memory CD4+ T cells are selected from naive precursors during blood-stage infections with APL-bearing parasite strains. To address this issue, a previously published protocol 23, 24 was used to induce primary human CD4+ T cell proliferation to the epitopes of interest from PBMC in malaria-unexposed naive donors. The question of whether costimulation with allelic epitopes would interfere with T cell priming was specifically addressed for antagonistic (M3/4 and M7/8) and non-antagonistic (M5/6) epitope pairs.

Naive donors were screened for reactivity for M3, M4, M5, M6, M7 and M8 (Table 5). Fig. 3 illustrates the induction of primary proliferative responses in malaria naive individuals. In summary, 91% (10/11) HLA-DRB1*0101, and 100% (2/2) HLA-DRB1*1302 donors responded to one of these peptides in similar primary cultures, as well as the DRB1*0701 and DRB1*1501 homozygous donors (Table 5). The only two non-responders were HLA-DRB1*0301 and HLA-DRB1*0402 homozygotes (Table 5). Twelve responders were selected for further study in proliferative immune interference assays, of whom five individuals were reactive to M5 or M6.

Table 5. Summary of peak stimulation indices (numbers shown) in malaria-unexposed naive donors in response to the MSP-1 allelic variant peptides, M3 to M8
  1. a) Positive primary proliferative stimulation indices (SI) of >2 are highlighted 42, 43. Positive responses were consistently observed in all donors upon re-testing (n=5 donors).

  2. b) Donors were molecularly typed for HLA class II antigens and are grouped into non-HLA-DR1, non-HLA-DR1302 and homozygotes / heterozygotes for HLA-DR1 and HLA-DR1302. Donors PB, JT, EM and SD are DRB1*0701, DRB1*1501, DRB1*0301 and DRB1*0402 homozygotes, respectively

SIa) in non-HLA-DR1 and DR1302 donorsb)
SI in HLA-DR1 heterozygous donorsb)
SI in HLA-DR1 homozygous donorsb)
SI in HLA-DR1302 heterozygous donorsb)

Representative experiments from three donors showing immune interference between M3/4 and M7/8 and lack of immune interference for M5/6, are shown in Fig. 3. The allelic pair of M5 and M6 was consistently non-antagonistic in all five donors. In contrast, the allelic pairs of M3/4 and M7/8 were mutually inhibitory in 6/12 responders tested (mean ± SE proliferative immune interference: 60.6 ± 3.4%).

Fig. 3A further illustrates the differences in kinetics between primary responses to both M3 and M4 in comparison to secondary responses to the control antigen tetanus toxoid (TT), and the fact that costimulation with M3 and M4, even when presented on separate APC, causes immune interference. Fig. 3B formally illustrates that presentation of M3 and M4 together to the same T cell culture, either by addition of both peptides to the well or by mixing of separately pulsed APC to stimulate the culture, results in a comparable immune interference effect. Fig. 3C shows that the immune interference effect was specific for the allelic variant pair, as reactivity to M3 peptide was turned off in this donor by co-stimulation with M4, whereas co-stimulation with M5 not only failed to cause immune interference but also promoted increased proliferative responses. Fig. 3D shows an example of immune interference for M7/8. Thus, the same non-antagonistic (M5/6) and antagonistic (M3/4 and M7/8) immune interference properties for specific allelic variant epitope pairs were observed in both malaria-naive and malaria-experienced individuals.

Immune interference is not due to a Th1 or Th2 cytokine bias of responses to allelic variants

The inhibition in proliferative responses of naive T cells by co-stimulation with allelic variant epitopes could have been due to induction of anergy or a Th1 vs. Th2 cytokine imbalance. Classically, anergy induction is characterized by a block in IL-2 synthesis 25, 26. However, the addition of exogenous IL-2 (on days 0, 1, 2 or 3 of primary culture, at 0.1, 1 or 5 U/mL of IL-2) to antagonized T cells (stimulated with M3/4 or M7/8) did not restore proliferation to levels observed with the index peptide alone (n=3 donors for M3/4, n=2 donors for M7/8) (data not shown). Similarly, the addition of exogenous IL-12 or IL-10 (at days 0, 1, 2 or 3 of primary culture, at 0.1, 1 or 5 U/mL) to antagonized T cells stimulated with M3/4 or M7/8, did not restore proliferation to levels observed with the index peptide alone (n=3 donors for M3/4, n=2 donors for M7/8) (data not shown). Use of anti-IL-10 neutralizing antibody 20 also failed to restore the proliferative responses of M3/M4 antagonized polyclonal cultures (n=2 donors) (data not shown).


The present study shows how natural allelic polymorphism in MSP-1 can lead to APL multi level (naive, memory-central and memory-effector) immune interference of polyclonal T cell responses, providing an immunological and molecular mechanism of blood-stage malaria infection that acts to promote the maintenance of the remarkable dimorphism of this protein at the blood-stage of malaria infection. Allelic variants were found to be mutually antagonistic in polyclonal responses in malaria exposed and naive individuals. Differences in MHC binding or any other potential characteristics of one single allelic variant compared to the other cannot account for our findings. APL variants with the mutual ability to cause immune interference may be seen to have evolved as a consequence of facilitating mutual parasite survival in human hosts. It could also be further argued that immune interference is beneficial to host survival in that it prevents over-activation of Th1-like cells, and thus prevents severe disease. The antagonizing effects of MSP-1 variants were observed in the majority (average 80%) of individuals from The Gambia who showed IFN-γ responses. T cell proliferation to MSP-1 recombinant proteins has been reported to correlate with protection against clinical malaria in prospective field studies 22. Therefore, the observed correlation of both IFN-γ and proliferative immune interference with allelic epitopes M3/4 and M7/8 could potentially affect protective memory T cell effector functions. Indirect protective functions include depriving the system of cognate help for antibody production 2729, and direct protective mechanisms include interfering with the IFN-γ effects on innate protective immune mechanisms mediated by monocytes and neutrophils 3032. In addition, the same antagonistic MSP-1 epitopes (M3/4 and M7/8, but not M5/6) were also capable of mutual interference during naive T cell priming.

The consequence of allelic interference with T cell priming was the induction of a state of partial activation of CD4+ T cells. The lack of priming of fully functional CD45RO memory T cells from naive precursors is likely to lead to “holes” in the effector T cell repertoire 21, 33. Preliminary studies suggested a novel partially activated T cell phenotype was induced by costimulation with allelic variants M3/M4 (data not shown). T cells with a hypo-proliferative phenotype have also been independently associated with both anergic and suppressor cell functions 34. It is possible that T cells induced by malaria parasites bearing antagonistic dimorphic MSP-1 epitopes would also have this general activity. However, to promote parasite co-habitation, it would be sufficient for cells induced by dimorphic parasites to be immunologically unresponsive 10. Naive T cells that have become fully functional effector cells can also be turned off by the same allelic epitopes through APL immune interference. The powerful combination of immune evasion strategies at two different stages of T cell activation, priming and activation of effector functions by the same mutually antagonistic APL allelic epitopes, provides a continuous population advantage to co-habiting parasites, which in turn promotes the MSP-1 dimorphism.

Recent studies have shown an increased interest in the possibility of generating vaccine-induced protective immunity to blood-stage malaria 6 through induction of a protective T cell response that can act independently of antibody 35, 36. However, allelic polymorphism in malaria antigens presents a dilemma for vaccine design. In this study, variants of MSP-1 have been identified, which can mutually turn off polyclonal naive as well as memory CD4+ T cells induced by natural malaria exposure. Thus, the inclusion of multiple allelic variants in a vaccine may be detrimental to both the initial priming, as well as the in vivo re-stimulation of pre-existing effector T cells. The use of selected conserved non-antagonistic epitopes, or largely conserved MSP-1 fragments, such as the 19-kD fragment may provide a safer development alternative for the induction of blood-stage protective immunity.

Materials and methods

Study area and volunteers blood samples

Malaria-exposed adults aged between 18 and 60 years old were recruited from the villages of Brefet, Jali and Berending in The Gambia, West Africa. Studies were approved by the Ethical Committee of the Gambia Medical Research Council. Gambian blood samples were collected both before (March to April) and after (October to November) the wet season. No statistical differences in IFN-γ reactivity patterns between samples collected during the wet and dry seasons were observed (data not shown). All malaria-exposed donors were parasite negative by microscopy of Giemsa thick and thin blood films. Donors were asymptomatic for malaria and other illnesses. Malaria-unexposed naive donors were European Caucasian volunteers aged between 21 and 50 years old, recruited at the John Radcliffe Hospital, Oxford, UK who had not previously traveled to a malaria-endemic country. These donors were asymptomatic for illness at the time of blood sampling. After informed consent, 20 mL blood was taken, and cells were separated by density gradient centrifugation on Ficoll at 800 × g for 20 min. PBMC were washed three times in RPMI 1640 medium and resuspended at 4 × 106/mL in RPMI 1640 medium supplemented with 5% heat-inactivated human AB serum, 2 mM glutamine, 100 µg/mL streptomycin and 100 U/mL penicillin (RN5) prior to use in ELISPOT assays.


Three allelic variant peptides of P. falciparum MSP-1 corresponding to the MAD-20 and WELLCOME strain 4 were commercially synthesized (Research Genetics, USA) (see njhle 1). These peptides correspond to aa 310–328 (M5/M6), aa 344–357 (M3/M4) and aa 1538–1550 (M7/M8) 2. Peptides M3, M5 and M7 correspond to the MAD-20 allelic strain, while peptides M4, M6 and M8 were derived from the WELLCOME strain. Peptides were reconstituted in sterile PBS and were non-toxic to the control antigens, purified protein derivate (PPD; Serumstatens Institut, Denmark) from Mycobacterium tuberculosis, TT (Evans Biomedical, UK) and the mitogen phytohemagglutinin (data not shown). Unless otherwise stated, peptides were used at a concentration of 30 µg/mL. At this concentration most donors samples start to reach a plateau in their responses to the malaria-derived peptides under study. The recall antigens, TT and PPD were used as positive controls antigens at 10 µg/mL.

MHC binding assay

MHC affinity of MSP-1 variants was assessed by competition binding assay with the promiscuously binding biotinylated CLIP peptide from the invariant chain (aa 96–114), as previously described 20. Briefly, test peptides were serially diluted 1:10 in pH 5 buffer [0.02 M 2-(N-morpholino)ethanesulfonic acid in 0.1 M sodium hydroxide and 0.02% sodium azide] and incubated overnight together with competitor CLIP peptide (0.1 µg) and purified HLA class II protein (0.15 µg) at 37°C. After neutralization with Tris-HCl (pH 7.5) solutions were transferred into 96-well immunoplates (Nunc, Denmark) coated with anti-HLA class II antibody (L-243; Sigma, UK) and incubated for 2 h. The binding of biotinylated CLIP peptide was detected with avidin-horseradish peroxidase (ExtrAvidin; Sigma, USA) and developed with 0.4 mg/mL o-phenylenediamine (Sigma, UK) in phosphate-citrate buffer. The reaction was stopped with 12.5% sulfuric acid and the absorption measured at 492 nm. The concentration of unlabeled test peptide required to inhibit 50% of binding of the biotinylated CLIP (IC50) was calculated. All binding assays were performed in triplicate and the mean calculated.

Ex vivo lymphokine ELISPOT assays

ELISPOT assays utilizing PBMC from malaria-exposed individuals were performed as described 37. In ELISPOT immune interference assays 20, 30 µg/mL of index peptide was added to the assay 1 h before the variant peptide (60 µg/mL) unless otherwise stated. At 30 µg/mL most donors start to plateau in their responses to the peptides under study (Fig. 1A, B). The plates were incubated at 37°C in 5% CO2/95% air for 18 h and developed as described 37. Individual cytokine-secreting cells were expressed as spot-forming-units (SFU) per million PBMC. Positive cytokine responses (p<0.05) were based on a chi-square comparison of the odds ratio of positive cells in the test well and control well 18, 38 assuming a binomial distribution of 4 × 105 cells/well. This method of determining a significant positive response is as powerful as non-parametric or parametric comparison of the means or medians of replicate wells. Significant immune interference for IFN-γ production was defined as a >30% reduction in peptide-specific responses as in previous studies 10, 21, 39. Spots were counted visually by two independent researchers. Samples were blinded prior to the readings. If there was a discrepancy, a third reader would assess the spots.

Proliferation assays

Primary T cell proliferative responses from PBMC of malaria-naive donors were performed as described 23, 24. Irradiated autologous PBMC were used as presenters and, unless otherwise stated, were pre-pulsed with 50 µg/mL peptides for 2 h. Presenters were then washed three times in sterile RPMI. Untreated PBMC were used as responders. The responder:presenter ratio was 4:1. In proliferative immune interference assays 21, the APC were pulsed separately in separate tubes with index or variant peptides (50 µg/mL) for 2 h and washed three times in RPMI. The separately pulsed presenters (index-pulsed or variant-pulsed) were then added together at a 1:1 ratio to the untreated responders. The final culture concentration was 2 × 106/mL in 48 well flat-bottom plates. Thus, 0.25 × 106 index peptide-pulsed irradiated presenters were mixed with 0.25 × 106 APL variant-pulsed irradiated presenters and mixed with 1 × 106 responders. To keep the ratio of presenters to responders consistent, untreated irradiated presenters were added (Neg) when reactivity to only one variant was tested. Thus, the single variant responses were tested by mixing 0.25 × 106 peptide pulsed irradiated presenters with 0.25 × 106 untreated irradiated presenters before adding to 1 × 106 responders. Triplicate 50-µL samples from each well were transferred between days 5 and 10 into 96-well round-bottom plates; 1 µCi [3H]thymidine was added per well and incubated for further 16 h. The incorporated radioactivity was counted using conventional scintillation methods. Proliferation data are either expressed as the mean ± SE cpm of triplicate samples or as stimulation indices (mean test cpm / mean background cpm).

Molecular detection of co-habiting parasite strains

DNA samples of Gambian children with clinical malaria were part of an earlier case-control study of malaria 40. Parts of the P. falciparum MSP-1 gene corresponding to the antagonistic allelic variant epitope pair M3/M4 was PCR amplified with primers (5′–3′): TACAATAAACAATTACAAGAAGC and TTAAATAATATTCTAAT-TCAAGTGG. As described elsewhere 10, PCR product was transferred to nitrocellulose membrane and probed with oligonucleotides specific for the two variants, MAD-20 and WELLCOME. The oligonucleotide probes (5′–3′) and wash temperatures were: M3 (MAD-20), AAAGAAATTACTTGAAGATATAG, 50°C; M4 (WELLCOME), TAAGGAATTACTTGATAAGATAA, 46°C. These were determined by a methodical optimization process and repeatedly confirmed to be uniquely allele specific by sequencing of cloned isolates (data not shown).


We thank Hilton Whittle, Keith McAdam, Momodou Sanyang, Bolong Jaiteh, Hassan Joof and all the blood donors. We thank Miles Davenport for help with the MHC binding assays. E.A.M.L. was supported by a Wellcome Trust Prize Studentship, K.L.F. by a Wellcome Trust Training Fellowship in Clinical Tropical Medicine, G.M. is a recipient of the University of Melbourne International Postgraduate Research Scholarship. A.V.S.H. by a Wellcome Trust Principal Research Fellow Award and M.P. by an NH-MRC Senior Fellow Award. The study was supported by Wellcome Trust, HHMI and NHMRC. Thanks to Dr. Patricia Mottram (The Burnet Research Institute at Austin) for reviewing and revising the manuscript.


  1. 1


  2. 2


  3. 3