Assessing the Binding of Four Plasmodium falciparum T Helper Cell Epitopes to HLA-DQ and Induction of T-Cell Responses in HLA-DQ Transgenic Mice

Abstract

A subunit vaccine for Plasmodium falciparum malaria will need to contain well-defined T helper cell epitopes that induce protective immune responses to the parasite. One major barrier to the use of subunit vaccines is the requirement for T helper cell epitopes to be presented by the HLA class II molecules that are present in the population being vaccinated. Since the majority of malaria studies have focused on HLA-DR, little information on the role of HLA-DQ in the binding and immune response to malarial epitopes is available. This study used an in vitro peptide-binding assay to predict the extent of HLA-DQ binding of four conserved T helper cell epitopes identified from asexual-stage malaria vaccine candidate antigens. Epstein-Barr virus (EBV)-transformed human B-cell lines expressing 14 different DQ molecules (DQ2.1, -2.2, -4.1, -4.2, -5.1 to -5.3, -6.1, -6.2, -6.4, -7.1, -7.3, -8, and -9) representing all broad serological specificities, including common DQ molecules present in populations in areas where malaria is endemic, were used in the binding assay. Moreover, an HLA-DQ transgenic mouse model was employed to evaluate the correlation between the in vitro DQ binding of the peptides and the generation of in vivo immune responses following peptide immunization. This study identified two broad DQ-binding peptides, ABRA#14 and SERA#9. ABRA#14 also induced T-cell proliferation and Th1-associated cytokine production in DQ8⁺ transgenic mice. The combination of peptide binding to EBV-transformed cell lines and DQ transgenic mice provides a method for identifying additional T-cell epitopes for inclusion in a vaccine.

Despite efforts by the World Health Organization to control and eradicate Plasmodium falciparum malaria since 1950, between 300 and 500 million people are infected by malaria worldwide (39). Accordingly, a substantial effort is being made to develop an effective vaccine; however, the complexity of the malarial life cycle together with antigenic variation makes this a difficult task. One approach is the development of a subunit vaccine in which well-defined conserved epitopes that induce protective helper T-cell responses will have to be included. Since T-cell epitopes have to be presented by HLA class II molecules in order to be recognized by T cells, it is essential to include a combination of peptides or a promiscuous peptide that can be bound by at least one HLA molecule expressed in an individual in order for the peptide to generate an effective immune response. The inclusion of such a T helper epitope(s) linked to an appropriate B-cell determinant can potentially induce antibody responses in a genetically heterogeneous human population (17).

Studies of HLA class II restriction of T-cell epitopes from malarial parasites have mainly focused on HLA-DR (3, 6, 9, 14, 29, 33). Less is known about the role that HLA-DQ plays in the immune response to malaria (3, 28). DQ alleles have been associated with susceptibility and resistance to a number of diseases (21, 23), including one report of a possible association between DQB1*0501 and protective immunity to severe malarial infection (12). It is likely that DQ plays a role different from that played by DR in the immune response to the parasite. Both alpha and beta chains of DQ are polymorphic, forming a binding site which may be structurally different from the DR antigen binding groove (18, 27). It has been postulated that the observed strong linkage disequilibrium between DR and DQ may be driven by selection for complementarity between DR and DQ antigen binding profiles (10).

The objective of this study was to demonstrate a strategy for evaluating epitopes for inclusion in a subunit vaccine based on their interaction with HLA-DQ. Four conserved T helper cell epitopes (MSP-1#2, MSP-1#3, SERA#9, and ABRA#14) were used in this study. These epitopes were identified using computer algorithms to predict potential T-cell epitopes from conserved regions of blood stage proteins with vaccine potential, the major merozoite surface protein 1 (MSP-1), a serine-rich antigen (SERA), and an acidic-basic repeat antigen (ABRA) (25, 26). These peptides induced recall proliferative responses in a significant percentage (∼30 to 45%) of western Africans living in the Ivory Coast (I. A. Quakyi, personal communication) and in ∼10 to 45% of Cameroonians (N. Pimtanothai, C. K. Hurley, C. Y. Ginsberg, M. E. O'Brien, R. Leke, W. Klitz, and A. H. Johnson, unpublished data). Furthermore, mice immunized with two of these epitopes (SERA#9 and ABRA#14) have been shown to be primed for help for antibody production upon subsequent exposure to an extract of P. falciparum (M. Parra, unpublished data). Since these peptides can induce recall responses in a relatively high percentage of western Africans, they might bind to multiple HLA class II molecules or to an HLA class II molecule found at a high frequency in the African population. An in vitro peptide-binding assay was used to test the binding of the four peptides to 14 different HLA-DQ molecules, including those present in regions endemic for malaria. In addition, in order to correlate in vitro binding with the ability to induce immune responses in vivo, HLA-DQ transgenic mice were immunized with the peptides and the resulting immune responses were evaluated.

MATERIALS AND METHODS

Peptides.

Malarial peptides ABRA#14, SERA#9, MSP-1#2, and MSP-1#3 (Table 1) were synthesized by AnaSpec Labs (San Jose, Calif.). Biotinylated peptides used in the binding assay were amino-terminally linked to a long-chain biotinyl group. The peptides were purified by reverse-phase high-pressure liquid chromatography (>90% purity) and analyzed by mass spectrometry.

TABLE 1.

Synthetic peptides used in the study

Peptide	Sequence	Locationa
ABRA#14	DSNIMNSINNVMDEIDFFEK	ABRA, aa 487–506
SERA#9	DDYTEYKLTESIDNILVKMFKTN	SERA, aa 391–411
MSP-1#2	FGYRKPLDNIKDNVGKMEDYIKK	MSP-1, aa 250–271
MSP-1#3	SKLNSLNNPHNVLQNFSVFFNKK	MSP-1, aa 1101–1121

^aABRA, acidic-basic repeat antigen; SERA, serine-rich antigen; MSP-1, major merozoite surface protein 1; aa, amino acid. 

MAbs and cell lines.

The monomorphic monoclonal antibody (MAb) SPVL3, specific for all DQ molecules, was kindly provided by R. Dewall (DNAX Research Institute, Palo Alto, Calif.) and F. Koning (University Hospital, Leiden, The Netherlands). Homozygous Epstein-Barr virus (EBV)-transformed human B-lymphoblastoid cell lines (B-LCLs) were obtained from the 12th International Histocompatibility Workshop (Table 2). Class II-deficient human B-cell line BLS-1 (13) was a gift from S. Rosen Bronson (Georgetown University, Washington, D.C.). MAbs specific for HLA-DQ (IVD-12), CD4 (GK1.5), CD8a (53-6.7), and I-A^b (AF6-120.1) and a control irrelevant MAb (R35-95; Pharmingen, San Diego, Calif.) were used in the antibody-blocking assay. Labeled MAbs specific for mouse CD4 (fluorescein isothiocyanate [FITC]; GK1.5), mouse CD8a (R-phycoerythrin [PE]; Ly-2), mouse I-A^b (FITC; AF6-120.1) (Pharmingen), HLA-DQ (FITC; Leu-10) (Becton Dickinson, San Jose, Calif.), and mouse B cells (PE; B220) (Caltag, San Francisco, Calif.) were used in flow cytometric analyses.

TABLE 2.

Homozygous EBV-transformed human B-LCLs^a

B-LCL	DQA1*/DQB1*b	DQ molecule
MOU	0201/0202	DQ2.1
COX	0501/0201	DQ2.2
LKT3	0301/0401	DQ4.1
RSH	0401/0402	DQ4.2
LWAGS	0101/0501	DQ5.1
KAS011	0102/0502	DQ5.2
TEM	0104/05031	DQ5.3
E4181324	0103/0601	DQ6.1
MGAR	0102/0602	DQ6.2
WT47	0102/0604	DQ6.4
JHAF	0301/0301	DQ7.1
RML	0501/0301	DQ7.3
BOLETH	0301/0302	DQ8
T7526	0302/03032	DQ9
BLS-1

^aFrom the 12th International Histocompatibility Workshop cell panel (19). 

^bSee Kimura et al. (15). 

Binding assay with EBV-transformed B-LCLs.

The cell binding assay was described previously by Kwok et al. (16). In brief, B-LCL cells were washed twice with Hanks' balanced salt solutions (HBSS) (Biofluids, Rockville, Md.), incubated with 0.5% paraformaldehyde in phosphate-buffered saline (PBS) for 10 min at room temperature, and then washed twice with HBSS. The fixed cells (1.5 × 10⁶), in duplicate, were incubated with various concentrations of biotinylated peptides at 37°C for 24 h in a binding buffer (150 mM citrate-phosphate buffer [pH 4.4, 5.6, or 7.0], 5 mM EDTA, 1 mM iodoacetamide, 1 mM benzamidine, 1 mM Pefabloc [Boehringer Mannheim, Indianapolis, Ind.]). After incubation with the peptides, the cells were washed five times with PBS–0.05% Tween 20 and lysed using 100 μl of lysing buffer (0.5% Nonidet P-40, 0.15 M NaCl, 50 mM Tris [pH 8.0] containing protease inhibitors [1 mM Pefabloc, 1 mg of pepstatin/ml, and 1 mg of leupeptin/ml]). Lysates were cleared by centrifugation. Each sample was neutralized with 100 μl of 0.5 M Tris, pH 8.0, containing 0.02% n-dodecyl-β-d-maltoside (Sigma Chemical Co., St. Louis, Mo.), and then samples were transferred to 96-well plates previously coated with HLA-DQ-specific MAb SPVL3 (2 μg/well) and incubated for 18 h at 4°C. After the plates were washed five times with PBS–0.05% Tween 20, europium-labeled strepavidin (200 μl/well; 1:1,000) (Wallac, Gaithersburg, Md.) was added and the plates were incubated at room temperature for 4 h. The plates were then washed and incubated with enhancement buffer (200 μl/well) (Wallac) for 1 h. Binding was assessed by fluorescence measured in a Delfia 1234 fluorometer (Wallac). The level of binding of peptide to B-LCLs (in mean fluorescence units) was compared with the binding of peptide to a cell without HLA-DQ (BLS-1). The BLS-1 cell line exhibited a background binding of ∼1,000 to 2,000 fluorescence units. In this assay, peptide-DQ binding with a mean fluorescence >10,000 fluorescence units was defined as high, binding with a mean fluorescence between 5,000 and 10,000 fluorescence units was defined as intermediate, and binding with a mean fluorescence <5,000 fluorescence units was considered negative.

Immunization and in vitro proliferation.

HLA-DQ8 transgenic mice (DQ8⁺ mice) and control HLA-DQ8-negative littermates (DQ8⁻ mice) with a major histocompatibility complex class II (MHC-II)-deficient background (H-2 A^b0) were provided by C. David (Mayo Clinic, Rochester, Minn.). Expression of the DQ8 molecule was confirmed by PCR-based DNA typing using DQ8-specific oligonucleotide probes (Pimtanothai et al., unpublished data) and splenocytes staining with a DQ-specific MAb (Leu-10). DQ8⁺ and DQ8⁻ mice (6 to 12 weeks old) were immunized subcutaneously on both sides of the base of the tail with an 83.5 μM peptide emulsified in complete Freund's adjuvant. Mice immunized with adjuvant alone were included as a negative control. The experiments were conducted using three mice per group and were repeated at least twice to confirm the results. Ten days after immunization, the draining lymph nodes and spleens were removed and prepared for cell culture. For proliferation assays, lymph node cells (LNC) and splenocytes (SPC) were suspended at a concentration of 2 × 10⁶/ml in RPMI 1640 medium (Biofluids) supplemented with 10% heat-inactivated fetal calf serum, 25 mM HEPES buffer, 2 mM glutamine, 100 U of penicillin/ml, 100 mg of streptomycin/ml, and 50 mM β-mercaptoethanol. The cell suspension (200 μl), containing 4 × 10⁵ cells, was added to each flat-bottom microtiter well (Costar, Cambridge, Mass.) in the presence of medium alone or 1 or 10 μM peptide. The cells were incubated for 72 h (37°C, 5% CO₂) and then pulsed with 0.5 μCi of [³H]thymidine (37 MBq/ml; Amersham, Arlington Heights, Ill.) during the final 16 to 18 h of culture. The cells were harvested, and the extent of [³H]thymidine uptake was determined using a liquid scintillation counter (Wallac). The geometric mean (Gm) of the counts per minute for six replicate wells was determined following outlier analysis. The stimulation index (SI) was calculated as the Gm for treatment divided by the Gm for medium alone. Results were considered positive when (i) peptide incubation gave a SI ≥2.0 and (ii) the t values comparing the treatment and the medium control were ≥95%. For in vitro antibody-blocking studies, purified MAbs specific for HLA-DQ8 (IVD-12), mouse CD4 (GK1.5), mouse CD8a (53-6.7), and I-A^b (AF6-120.1) and an irrelevant control antibody (R35-95) (Pharmingen) were added at various concentrations (0.01 to 10 μg/ml) to the cell cultures.

Cell selection using magnetic beads.

Depletion of either CD4⁺ or CD8⁺ T cells was performed using Dynabeads coated with antibodies specific for mouse CD4 (L3T4) or CD8 (Lyt2) (Dynal, Inc., Lake Success, N.Y.) according to the manufacturer's instructions. Single cell suspensions from draining lymph nodes and spleen (prepared as described above) were mixed with prewashed Dynabeads (cell/bead ratio, ∼1:10) and rotated for 20 min at 4°C. Following removal of the CD4 or CD8 cells, the cell-depleted suspensions were analyzed by flow cytometry and were used in a cell proliferation assay as described above.

Flow cytometry.

A FACStar Plus (Becton Dickinson) was used to assess the purity of cells following magnetic bead separation by determining expression of CD4 and CD8. In addition, a phenotypic study of cells from mice immunized with peptide was conducted. Freshly isolated draining LNC and SPC from DQ8⁺ and DQ8⁻ mice were measured for the expression of CD4, CD8, the B-cell marker (B220), I-A^b, and HLA-DQ. The same cell populations were also measured in LNC and SPC from both types of mice following in vitro cultures with ABRA#14 for 4 days.

Cytokine determination.

Lymph node and spleen cell suspensions were prepared and cultured as described for the in vitro proliferation assay. Culture supernatants were collected from each well after 24, 48, and 72 h of incubation at 37°C. The cultures were centrifuged to remove cells, and supernatants were immediately stored at −20°C and assayed within a week. In vitro cytokine release was assessed in duplicate by quantitative enzyme-linked immunosorbent assay using gamma interferon (IFN-γ), interleukin-2 (IL-2), IL-4, and IL-12 minikits from Duoset (Genzyme Diagnostics, Cambridge, Mass.) according to the manufacturer's instructions in 96-well flat-bottom microtiter plates (MaxiSorp; Nunc, Roskilde, Denmark). Assay results were read in a microtiter autoreader (LEX800; Bio-Tek, Winooski, Vt.) at 450 nm. Supernatant cytokine levels were quantified by comparison to standards run in parallel on the same plate.

RESULTS

In vitro peptide binding to DQ defines candidate broad binding peptides.

To determine the binding capacities of the four malarial peptides to HLA-DQ, in vitro peptide-binding assays were performed using paraformaldehyde-fixed B-LCLs expressing all broad serological specificities, including common DQ molecules present in populations in areas where malaria is endemic (7, 11, 20, 24). A simplified nomenclature for the DQA1/DQB1 composition of the DQ molecules used in this study (DQ2.1, -2.2, -4.1, -4.2, -5.1 to -5.3, -6.1, -6.2, -6.4, -7.1, -7.3, -8, and -9) is described in Table 2. Since previous studies had found that peptide binding is influenced by pH (2, 32), peptide binding was measured at pH 4.4, 5.6, and 7.0. Each experiment was repeated at least three times. Biotinylated ABRA#14 peptide at 10 μM bound at a high level to DQ2.1, DQ2.2, DQ5.1, DQ5.2, DQ6.4, and DQ8 molecules at all three pHs, while it bound at a high level to DQ4.1, DQ5.3, and DQ6.1 only at certain pHs (Fig. 1A). Intermediate binding of ABRA#14 to DQ6.2 and DQ9 was observed at at least one pH tested. The binding level of ABRA#14 was higher at acidic pH for most of the DQs tested except for DQ2.1, DQ5.1, and DQ6.1. The binding patterns were found to be the same when ABRA#14 was tested at 50 μM (data not shown). In summary, ABRA#14 bound at a high level to 9 of the 14 DQ molecules tested.

FIG. 1.

Binding specificities of malarial blood stage T-cell epitopes ABRA#14 (10 μM), SERA#9 (10 μM), MSP-1#2 (50 μM), and MSP-1#3 (50 μM) to DQ allelic products expressed on B-LCLs including class II-negative control line BLS-1. Binding was performed at pH 4.4, 5.6, and 7.0. The solid horizontal line indicates a high mean fluorescence, and the dashed line indicates an intermediate mean fluorescence (5,000 fluorescence units). The DQA1/DQB1 alleles encoding each DQ molecule are shown in Table 2. Data represent the means ± standard deviations of three independent experiments.

SERA#9 bound at a high level to only DQ5.3 at all three pHs and to DQ2.1, DQ4.2, DQ5.1, and DQ6.4 at 10 μM, at pH 4.4 (Fig. 1B). SERA#9 also bound to DQ6.4 at a high level at pH 5.6 but not at pH 7.0. Intermediate binding with DQ2.1 at pH 5.6 and 7.0 was observed. Similar binding patterns were observed when SERA#9 peptides were tested at 50 μM (data not shown). SERA#9 bound at a high level to 5 of the 14 DQ molecules tested mostly at low pH.

Although MSP-1#2, at 10 μM, did not bind well to most of the DQ molecules tested in this study (data not shown), MSP-1#2 peptide was able to bind, at 50 μM, at a high level to DQ5.3 and DQ6.4 at all three pHs (Fig. 1C). DQ7.1 also bound well to MSP-1#2 at pH 4.4, but only intermediate binding was observed at higher pHs. Intermediate-level bindings to DQ5.1 at pH 7.0 and to DQ5.2 at pH 4.4 and 7.0 were observed. MSP-1#3 did not bind to any DQ tested at either 10 or 50 μM at all pHs (Fig. 1D). In general, only 3 of the 14 DQ molecules tested showed a high level of binding to MSP-1#2 and none bound to MSP-1#3.

Immune response of DQ transgenic mice to peptide correlates with peptide-binding profiles.

The immunogenicities of peptides in vivo were compared to those from in vitro binding studies. DQ8 mice lacking endogenous murine class II antigens were used as a model in this study since the DQ8 (DQA1*0301/DQB1*0302) molecule was shown in the above studies to bind ABRA#14, but not SERA#9, MSP-1#2, and MSP-1#3 peptides (Fig. 1). Mice expressing other DQ types were not available for testing. Groups of DQ8 mice were immunized with either ABRA#14, SERA#9, MSP-1#2, or MSP-1#3, and then T-cell recall responses and cytokine levels (IFN-γ, IL-2, and IL-4) were measured in vitro with homologous peptides. The responses to each peptide were assessed using cells from both the lymph nodes and spleen.

As predicted by the binding results, only ABRA#14 induced recall proliferative responses in the DQ8 mice (Fig. 2); immunization with SERA#9, MSP-1#2, and MSP-1#3 did not result in T-cell proliferation (Fig. 2). The response was dose dependent although not in a linear fashion, i.e., there was a higher response when cells were stimulated with 10 μM peptide (LNC: mean response, 16,105 cpm; SPC: mean response, 10,598 cpm) than when cells were stimulated with 1 μM peptide (LNC: mean response, 10,384 cpm; SPC: mean response, 8,390 cpm). The response from SPC was lower than the response from LNC, possibly due to the route of immunization and the effect of complete Freund's adjuvant (Fig. 2). Control studies verified that naive and adjuvant control mice did not respond to any of the peptides at any of the concentrations used (data not shown).

FIG. 2.

Responses to malaria peptides in HLA-DQ8 transgenic mice correlate with the DQ8 binding result. In vitro proliferative responses of LNC and SPC from HLA-DQ8⁺/H-2 A^b0 mice immunized with ABRA#14, SERA#9, MSP-1#2, and MSP-1#3 as described in Materials and Methods and cultured with the same peptide at 1 or 10 μM compared to those of cells incubated with medium alone. Data are means ± standard deviations from six replicate well cultures from one out of two representative experiments.

Immunization with SERA#9, MSP-1#2, and MSP-1#3 also did not induce any proliferative responses in the DQ8-negative littermates (DQ8⁻) (data not shown); however, an unexpected proliferative response was observed in DQ8⁻ mice immunized with ABRA#14 (Fig. 4B). Proliferative responses <50% of the level of responses in DQ8⁺ mice were observed from both LNC and SPC (Fig. 4B). Since no responses to ABRA#14 were observed in DQ8⁺ or DQ8⁻ mice immunized with adjuvant alone, SERA#9, MSP-1#2, or MSP-1#3 (data not shown), the proliferative responses in DQ8⁺ mice appear to be specific for ABRA#14.

FIG. 4.

The response to ABRA#14 peptide in HLA-DQ8 transgenic mice is mediated by CD4⁺ cells and the HLA-DQ molecule, with a background response by CD8⁺ cells in HLA-DQ8-negative littermates. LNC and SPC from HLA-DQ8⁺/H-2 Ab⁰ mice (A) and HLA-DQ8⁻/H-2 Ab⁰ mice (B) were challenged with ABRA#14 (10 μM) in the presence of the indicated MAbs. Results of responses in the absence of MAb (none) and unstimulated cultures are also indicated. Only results for MAbs at 1 μg/ml are shown here. The inhibition level was not altered by using a higher concentration of each MAb. The data are means ± standard deviations of six replicate well cultures from one out of two representative experiments.

Culture supernatants from LNC and SPC of HLA-DQ8⁺ and DQ8⁻ transgenic mice immunized with each peptide were assayed for the presence of IFN-γ, IL-2, and IL-4 cytokines following in vitro culture with the homologous peptide. Similar to proliferation results, cytokines were only detected in cell cultures from DQ8⁺ transgenic mice immunized with ABRA#14 (Fig. 3). Specifically, upon recall stimulation with ABRA#14, cultures of LNC from DQ8⁺ mice contained IFN-γ at high levels (92 to 1,013 pg/ml), with a peak at 72 h of culture (Fig. 3A). In cultures of SPC from DQ8⁺ mice, a marginal level of IFN-γ was also detected (180 pg/ml) at 72 h (Fig. 3B). IL-2 was detected in LNC cultures from DQ8⁺ mice (32 pg/ml) as early as 24 h after stimulation with the peptide, while the peak level was detected in the SPC culture (77 pg/ml) at 72 h (Fig. 3C and D). No IL-4 was detected in DQ8⁺ mice after in vitro recall stimulation with either ABRA#14 (Fig. 3E and F) or the other peptides (data not shown). DQ8⁻ mice, as expected, did not produce any cytokine in response to SERA#9, MSP-1#2, and MSP-1#3. As noted above, not only did ABRA#14 induce an unexpected proliferative response in DQ8⁻ mice but also low levels of IFN-γ (15 to 59 pg/ml) and IL-2 (12 to 25 pg/ml) were produced in response to this peptide (Fig. 3A to D).

FIG. 3.

Cytokine production by LNC and SPC of HLA-DQ8 transgenic mice and HLA-DQ8-negative littermates in response to immunization with the ABRA#14 peptide. Cells were cultured in vitro with medium alone or with ABRA#14 (10 μM) and harvested at 24, 48, and 72 h of culture. Cytokine profiles were measured by enzyme-linked immunosorbent assay. Data represent the means of duplicate wells from one out of two representative experiments.

Since in vitro cell cultures from DQ8⁺ and DQ8⁻ mice immunized with ABRA#14 were positive for IL-2 and IFN-γ, IL-12 (Th1 inducer) production in these mice was further assayed. Similar amounts of IL-12 secretion were detected in LNC cultures from both DQ8⁺ and DQ8⁻ mice, although a lower level of IL-12 was observed in cultures with peptide (300 to 440 pg/ml) than in cultures with media alone (380 to 760 pg/ml) (Fig. 3G). The levels of IL-12 secretion in SPC cultures from both DQ8⁺ and DQ8⁻ mice were also the same and were in the range of 70 to 280 pg/ml (Fig. 3H).

In vitro activation of cells from ABRA#14-immunized DQ8 transgenic mice is mediated by CD4⁺ T cells and is dependent on HLA-DQ8.

To determine the molecule and T-cell subset(s) involved in the recall responses described above, both antibody blocking of T-cell proliferation and enrichment of each T-cell subset were performed. For the antibody-blocking experiment, LNC and SPC from HLA-DQ8⁺ and DQ8⁻ littermates previously immunized with ABRA#14 were incubated in vitro with 10 μM ABRA#14 10 days postimmunization. At the time of initiation of the cultures, MAbs at various concentrations were added to the wells and T-cell activities were measured after 72 h of culture. Figure 4 shows the results using one MAb concentration. An HLA-DQ-specific MAb (IVD12) resulted in >50% inhibition of proliferation in cells from the DQ8⁺ mice but not in cells from the DQ8⁻ mice, and the addition of the CD4-specific MAb (GK1.5) inhibited the responses by approximately 50% in cells from DQ8⁺ mice (Fig. 4A) but had no effect on cells from the DQ8⁻ mice (Fig. 4B). The addition of the CD8-specific MAb (53-6.7) also inhibited the responses by approximately 50% in cells from DQ8⁺ mice (Fig. 4A) and completely inhibited the unexpected proliferative response observed in cells from DQ8⁻ mice (Fig. 4B). This background CD8 T-cell response may explain why only a 50% inhibitory effect was seen with the CD4-specific MAb treatment in cells from DQ8⁺ transgenic mice. A MAb specific for I-A^b (AF6-120.1) and an irrelevant control (R35-95) only caused a marginal inhibitory effect in cells from both groups of mice. Results showed that the response in cells from DQ8⁺ mice was mediated by both CD4 and CD8 T cells and the DQ8 molecule while the response in cells from DQ8⁻ mice was solely mediated by CD8 T cells.

In order to further confirm the cell type responsible for the recall response in DQ8⁺ transgenic mice, magnetic beads coated with antibodies specific for CD4 or CD8 were used for a cell depletion study. The depleted populations of either CD4 or CD8 T cells (containing <5% CD4 or CD8 cells as confirmed by two-color flow cytometry analysis) from LNC and SPC preparations were used in a recall proliferative assay. The CD4⁺ cell population gave a positive response (SI = 3 to 4) to ABRA#14 (Fig. 5). The response of the CD8⁺ cell population, while positive (SI = 1.5 to 2.5), was lower than the CD4 response (Fig. 5). The response was more obvious in LNC. No responses were detected in cell cultures of CD4⁺ and CD8⁺ cell populations from unimmunized mice (data not shown), showing that the addition of antibody-coated magnetic beads did not have a nonspecific effect on cell activation. The proportions of the response attributed to CD4 appear to differ in the two assays in that CD8 T cells appear to play a greater role in the antibody-blocking assay. The disparity between the two assay results might be due to the continued presence of the antibody-bound T cells in the blocking assay and the absence of T cells in the depletion assay. For example, if CD8 T-cell proliferative responses required cytokines produced by CD4 T cells, then the two assays might show different levels of CD8 responses.

FIG. 5.

Cell cultures depleted of either CD4 or CD8 T cells from lymph nodes (A) and spleens (B) of immunized HLA-DQ8 transgenic mice were used in a cell proliferation assay. Proliferation levels of unseparated cells cultured without magnetic bead treatment (CD4⁺/CD8⁺), CD4-depleted cells (CD8⁺), and CD8-depleted cells (CD4⁺) are shown. Bars represent incubation with the ABRA#14 peptide at 1 and 10 μM and incubation with medium alone. Data are means ± standard deviations of six replicate well cultures from one out of two representative experiments.

In addition, LNC and SPC from immunized mice before and after peptide activation for 4 days in culture were stained with fluorescent antibodies directed to CD4, CD8, a B-cell marker, HLA-DQ, and I-A^b (Table 3). As indicated, the CD4- and CD8-positive cells were also increased in peptide-stimulated cells from DQ8⁺ mice. The percentage of CD8 cells, but not that of CD4 cells, in cells from the immunized DQ8⁻ control mice was increased. There was no increase of B cells after peptide stimulation in cultures from both types of mice. HLA-DQ and I-A^b phenotype analysis confirmed DQ expression only in DQ8⁺ mice and the lack of mouse MHC molecules in both DQ8⁺ and DQ8⁻ mice. In summary, analyses of the responding cells revealed that proliferative T cells in DQ8⁺ mice were stimulated by DQ8 molecules, ABRA#14, and CD4 T cells, with a portion of the response also being mediated by CD8 T cells.

TABLE 3.

Surface phenotype^a of cells from DQ8⁺ transgenic mice and DQ8⁻ littermates

Cellsb	% of cells positive for indicated surface phenotypec
	CD4	CD8	B cell	HLA-DQ	I-Ab
DQ8+ mice
LNC	18.1	44.2	31.1	28.4	2.1
Activated LNC	33.3	54.4	22.0	13.8	5.0
SPC	12.1	17.2	37.2	38.0	4.2
Activated SPC	18.2	36.6	27.1	27.2	3.5
DQ8− mice
LNC	5.3	75.2	17.3	0.1	0.6
Activated LNC	4.7	90.6	18.4	0.9	2.4
SPC	4.9	15.5	33.4	0.02	1.4
Activated SPC	5.0	33.8	30.5	0.05	1.8

^aData were averaged from two independent experiments. 

^bActivated LNC and SPC were cultured with ABRA#14 for 4 days. 

^cValues in boldface represent an increase compared to the corresponding values for nonactivated cells. 

DISCUSSION

This study reports the HLA-DQ binding properties of four malarial T helper cell epitopes and the use of DQ8 transgenic mice as a model to test immune responses in vivo. The results have several important implications. The present study characterized the binding of malarial peptides to 14 DQ molecules, including DQ5.1, which was reported to be associated with resistance to malaria (12), and DQ6.2, which is found at very high frequency (∼30%) in African populations (11; Pintanothai et al., submitted). Interestingly, three out of four peptides, ABRA#14, SERA#9, and MSP-1#2, bound to DQ5.1, although at various levels. Unfortunately, the immunogenicity of these peptides in the context of DQ5.1 could not be tested, due to the unavailability of DQ5.1 transgenic mice. None of the peptides bound to DQ6.2. Two of the malarial peptides, ABRA#14 and SERA#9, bound well to multiple DQ molecules, whereas MSP-1#2 and MSP-1#3 peptides were more limited in their binding. The binding results are consistent with the T-cell responses in a malaria-exposed population in Cameroon in which most of the population responded to broader DQ-binding peptides, ABRA#14 (45%) and SERA#9 (25%), but not to peptides with more limited DQ binding, MSP-1#2 and MSP-1#3 (10 to 12.5%) (A. H. Johnson, unpublished data). In contrast, a study in Ivory Coast (I. A. Quakyi, unpublished data) demonstrated that 40% of the population responded to the MSP-1#2 peptide. The DQ binding analyzed revealed that MSP-1#2 can bind to the DQ6.4 molecule (DQα1*0102-DQβ1*0604). While rarely seen in Cameroon, DQ6.4 is quite common in the Ivory Coast (A. H. Johnson, unpublished data).

These peptides had been previously tested in several strains of congenic mice for the ability to induce specific T-cell responses (M. Parra, unpublished data). MSP-1#2 shows a highly H-2-restricted response in mice, whereas SERA#9 and ABRA#14 are more permissive. These results correlate well with the DQ-binding profile reported here. Interestingly MSP-1#3 appeared to be restricted by the murine HLA-DQ homologue I-A but did not bind to any HLA-DQ molecule tested in this study. Although it is possible that MSP-1#3 might bind to other DQ molecules not tested in this study, it seems likely that the I-A restriction noted in mice does not correlate with the DQ restriction in humans.

Second, these data reflect the influence of pH on the interaction between peptides and MHC molecules. Although protein molecules are thought to be naturally processed and loaded on MHC-II molecules in an acidic compartment (1, 37), a recent study suggests that exogenous peptides may bind mainly to class II molecules on the cell surface (22). However, previous studies have reported that the intrinsic properties of both peptides and MHC molecules determine the proper pH for their interaction (2, 32). The differences in pH-dependent binding to various DQ molecules suggest that peptides are loaded into the MHC binding groove over a range of pHs, i.e., those from the acidic compartments including lysosomes (pH 4.4 to 5.0), late endosomes (pH 5.0 to 5.6), and early endosomes (pH >6.0) and the neutral pH on the cell surface (4). Therefore, we have evaluated binding at three different pHs to assess all the possibilities of positive binding that can occur in vivo. Interestingly, our data indicate that the four tested peptides bind at higher levels to most DQ molecules in acidic pH. This observation suggests that these predicted epitopes reflect the peptides naturally selected in acidic compartments.

Another implication comes from the fact that HLA binding is not the only factor that determines the effective activation of T cells. It is believed that individuals with particular HLA types also require the right T-cell repertoire to recognize the peptide-HLA complex. In this study, DQα1*0301/DQβ1*0302 (DQ8) transgenic mice, on a murine class II knockout background, were used as a model to evaluate in vivo immune responses. Since the T-cell repertoire in these mice was selected under the influence of a human HLA-DQ8 molecule, the mature murine T-cell repertoire should be similar to the DQ8-selected T-cell repertoire of humans. In this study, the DQ8-mediated responses by CD4⁺ T cells and cytokine production in vivo correlate with in vitro DQ8-peptide binding as predicted. That is, DQ8⁺ mice responded to ABRA#14 but not to the other three peptides. Although immunogenicity of these peptides remains to be tested in the context of other DQ molecules, the finding for DQ8 transgenic mice supports the use of the in vitro binding assay in combination with the HLA transgenic mouse model as the effective tool to screen peptides for their MHC-restricted properties.

Since the generation of protective immunity can be greatly affected by cytokines (34, 36, 38), it is useful to identify types of cytokines generated upon immunization with any specific antigen or peptide before conducting a human trial. In this study, cell cultures from the immunized mice stimulated with ABRA#14 in vitro produced a Th1-like response with a high level of IFN-γ and a low level of IL-2 but not IL-4. The background responses from CD8 T cells observed in this study, however, may have affected the cytokine pattern. IL-12, a cytokine involved in Th1 differentiation (31, 35), showed an unexpectedly low level in cultures with ABRA#14 compared to cultures with medium alone. This observation might be the result of IFN-γ acting as a negative regulator and suppressing the production of IL-12 from its major source, the myeloid dendritic cell (30).

Finally, DQ8-negative littermates, which do not express any endogenous MHC-II molecules, responded to the ABRA#14 peptide. Antibody blocking, cell isolation experiments, and phenotypic analysis verified that this response was mediated by CD8⁺ cells. A study of Leishmania infection using the same class II-deficient mice also demonstrated proliferative responses mediated by CD8⁺ cells (5). These results suggest that either classical or nonclassical murine MHC-I molecules might be able to present ABRA#14 to CD8 T cells. Although MHC-I molecules primarily present cytosolic peptides, there are several observations that extracellular peptides can induce MHC-I-dependent responses (8). Further studies are required to clarify this observation and to determine whether ABRA#14 contains a cytotoxic T-lymphocyte epitope.

In summary, with a focus on HLA-DQ, and considering the HLA distribution within a population in an area where malaria is endemic, an in vitro peptide-binding assay was shown to be an effective tool to assess the HLA binding profiles of candidate peptides. With the DQ8⁺ mice as a model, a direct correlation between the in vitro binding properties of the peptides and the in vivo response was demonstrated. However, although the HLA transgenic mouse model is a helpful tool to assess the immunogenicities of the peptides prior to human trial, the lack of availability of transgenic mice expressing different HLA molecules is a significant limitation. One practical approach is to develop transgenic mice expressing HLA molecules common in the populations where vaccines are most needed. The positive binding and ability to induce immune responses to malaria epitopes in association with DQ molecules described in this study will help elucidate the role of DQ in malarial immunity and its contribution to malarial subunit vaccine development.