Antibody Biomarkers Associated with Sterile Protection Induced by Controlled Human Malaria Infection under Chloroquine Prophylaxis

Abstract

Immunization with sporozoites under chloroquine chemoprophylaxis (CPS) induces distinctly preerythrocytic and long-lasting sterile protection against homologous controlled human malaria infection (CHMI). To identify possible humoral immune correlates of protection, plasma samples were collected from 38 CPS-immunized Dutch volunteers for analysis using a whole Plasmodium falciparum proteome microarray with 7,455 full-length or segmented protein features displaying about 91% of the total P. falciparum proteome. We identified 548 reactive antigens representing 483 unique proteins. Using the breadth of antibody responses for each subject in a mixture-model algorithm, we observed a trimodal pattern, with distinct groups of 16 low responders, 19 medium responders, and 3 high responders. Fifteen out of 16 low responders, 12 of the 19 medium responders, and 3 out of 3 high responders were fully protected from a challenge infection. In the medium-responder group, we identified six novel antigens associated with protection (area under the curve [AUC] value of ≥0.75; P < 0.05) and six other antigens that were specifically increased in nonprotected volunteers (AUC value of ≤0.25; P < 0.05). When used in combination, the multiantigen classifier predicts CPS-induced protective efficacy with 83% sensitivity and 88% specificity. The antibody response patterns characterized in this study represent surrogate markers that may provide rational guidance for clinical vaccine development.IMPORTANCE Infection by Plasmodium parasites has been a major cause of mortality and morbidity in humans for thousands of years. Despite the considerable reduction of deaths, according to the WHO, over 5 billion people are still at risk, with about 216 million worldwide cases occurring in 2016. More compelling, 15 countries in sub-Saharan Africa bore 80% of the worldwide malaria burden. Complete eradication has been challenging, and the development of an affordable and effective vaccine will go a long way in achieving elimination. However, identifying vaccine candidate targets has been difficult. In the present study, we use a highly effective immunization protocol that confers long-lasting sterile immunity in combination with a whole P. falciparum proteome microarray to identify antibody responses associated with protection. This study characterizes a novel antibody profile associated with sterile protective immunity and trimodal humoral responses that sheds light on the possible mechanism of CPS-induced immunity against P. falciparum parasites.

Keywords: CHMI; antibody; malaria; preerythrocytic immunity; protein microarrays; sterile protection; vaccines.

FIG 1

CPS immunization induces broad and variable humoral responses against P. falciparum antigens. (A) Plasma samples from 38 CPS-immunized volunteers from 3 clinical trials (n = 9 from study 1, n = 5 from study 2, and n = 24 from study 3) were probed on whole P. falciparum proteome microarrays. Thirty CPS-immunized volunteers were protected and eight were not protected from malaria challenge. RBC, red blood cells. (B) Plasma samples collected preimmunization (I1-7) and postimmunization/prechallenge (C-1) were probed, and signals were quantified. The heat map represents the signal intensity at time point C-1, corrected for background reactivity by subtracting the signal intensity at I1-7, against 548 immunogenic P. falciparum features (or 483 unique proteins) printed on proteome microarrays. For an antigen to be considered immunogenic, (i) its levels had to have increased 100% at C-1 compared to I1-7 and (ii) its seroprevalence had to be >15%. The specimens are ordered in clinical groups, nonprotected (n = 8) and protected (n = 30). Samples are arranged with increasing reactivity from left to right, while the antigens are arranged with decreasing reactivity from top to bottom. A color gradient is used to display the signal intensity detected for each antigen, as shown by the key. (C) Comparison of aggregated C-1 reactivity of the 548 immunogenic antigens among the 3 different immunization dose groups. During the 3 rounds of immunization, 10, 9, and 19 volunteers received 5, 10, and 15 mosquito bites, respectively. Dots show individual data points, center lines show the medians, box limits indicate the 25th and 75th percentiles, and whiskers extend 1.5 times the interquartile range (IQR). Aggregate antibody levels were compared using the Kruskal-Wallis rank sum test. To detect only malaria-specific responses, the I1-7 responses were subtracted from the C-1 responses.

FIG 2

CPS immunization induces trimodal IgG responses in previously malaria-naive adults. (A) Three-component mixture model for breadth counts observed in CPS-immunized individuals (n = 38). The solid curved lines show the fitted breadth count distributions in the mixture model. The dashed curve line shows the mixture-model fit for components 1 (orange), 2 (blue), and 3 (green). The two dashed straight lines show the estimated cutoff points at 180 and 681 antigens, which distinguish the different responder groups (low, medium, and high responders). The histogram represents the frequency of the breadth counts in the different bins. Breadth counts are ranked from the lowest to the highest (range, 9 to 881). (B) Box plots of low-, medium-, and high-responder groups showing protected and nonprotected volunteers. Comparison of breadth counts within the medium-responder group was performed by a Wilcoxon rank sum test. Gray dotted lines show the cutoff points between responder groups. Dots show individual data points, center lines show the medians, box limits indicate the 25th and 75th percentiles, and whiskers extend to show the minimum and maximum breadth counts. (C) Humoral reactivity was compared prechallenge (C-1) and post-CHMI challenge (C+35). Mean antibody signal intensities measured against the top 50 most reactive antigens after CHMI challenge are shown, comparing low, medium, and high responders. The medium-responder group is further stratified into protected (“Med-P”) and nonprotected (“Med-NP”) volunteers. The mean reactivities of the top 5 most seroprevalent antigens, CSP, LSA1, LISP2, EXP1, and MSP2, are shown. Black dotted lines indicate the other boosted antigens after challenge (see Table S2 in the supplemental material).

FIG 3

Stage of expression of proteins printed on the microarray. Publicly available tandem mass spectrometry (MS/MS) data on protein expression were obtained from PlasmoDB. (A) Venn diagram showing stage-specific profiles of the proportions of proteins expressed by sporozoite and blood stages with MS/MS evidence of expression and printed on the microarray. (B and C) Bar plots showing numbers of P. falciparum proteins uniquely or commonly expressed at sporozoite and/or blood stages and recognized by at least 25% of nonprotected or protected volunteers (B) or low, medium, and high responders (C). The names and gene annotations of these proteins are indicated in Table S1 in the supplemental material.

FIG 4

Novel antigens associated with CPS immunization-induced sterile protection. (A) Differential signal intensities of 229 antigens at C-1 in plasma samples of protected (n = 7) and nonprotected (n = 12) medium responders. The red dotted line indicates the threshold P value of <0.05; dots in the green area and dots in the orange area represent antigens significantly more reactive in protected and nonprotected subjects, respectively. (B) Gene identifications of the antigens associated with protection and susceptibility in medium responders. Wilcoxon rank sum test P values (Wilcox Pval), Wilcoxon rank sum test P values with Benjamini-Hochberg correction for multiple comparisons (Wilcox Pval BH), area under the receiver operating characteristic curve (AUC) values, and leave-one-out cross-validation (LOOCV) AUC values for protected and nonprotected volunteers are shown. (C and D) Specific antigens associated with protection (C) or susceptibility (D). Boxes represent individual antigens in nonprotected (orange) (n = 7) and protected (green) (n = 12) subjects. Dots show individual data points, center lines show the medians, box limits indicate the 25th and 75th percentiles, and whiskers extend 1.5 times the IQR. Comparisons were tested by a Wilcoxon rank sum test.

FIG 5

Breadth of antigen response as a correlate of protection. (A and B) Breadth counts of protective (A) and nonprotective (B) antigens in protected (n = 30) and nonprotected volunteers (n = 8). Dots show individual data points, center lines show the medians, box limits indicate the 25th and 75th percentiles, and whiskers extend 1.5 times the IQR. Antibody intensities were compared using the Wilcoxon rank sum test. (C) The difference between protective and susceptible seropositivity counts was used to construct an ROC curve. LOOCV analysis was also performed. An AUC value of 0.5 indicates that there is no difference between the groups.