The present follow-up is part of a study concerning 1,183 pregnant women participating in Malaria in Pregnancy Preventive Alternative Drugs, a randomized trial of intermittent preventive treatment (IPTp) in Benin (18). The first 154 infants for which flow cytometry data were available were enrolled from January 2010 to June 2011 and followed throughout the first 2 years of life in the TOLIMMUNPAL project (19). At delivery, active placental malaria was determined by impression smear from the placental blood stained with Giemsa. At birth, newborn’s gender, weight and axillary temperature were recorded. At 6, 9, 12, 18 and 24 months, a medical questionnaire was filled-out. A thick blood smears (TBS) was performed monthly to detect asymptomatic malaria carriage. A rapid diagnosis test (RDT) for malaria was performed during the follow-up if an infant was febrile. Fever was defined as an axillary temperature greater or equal to 37.5°C. Symptomatic malaria was defined as a combination of a malaria infection determined by a positive RDT and/or a positive TBS and the presence of fever (or a 24 hours’ history of fever) since infants are particularly vulnerable to symptomatic malaria. However, we cannot exclude that non-malaria infections can also participate to the observed fever in case of coinfection. Malaria attacks were treated with artemether-lumefantrine combination, as recommended by the Beninese National Malaria Control Program. An asymptomatic infection was defined as a positive monthly systematic TBS with no fever or no history of fever.

The design of the study follow-up children for 2 years based on two arguments. First, epidemiological investigations from our group demonstrated that the risk of first Plasmodium falciparum infection associated with placental malaria is observed during the first 2 years of life (5, 20). Second, Malhotra et al, demonstrated the association between placental malaria and the modulation of T cell responses in children up to 3 years of age (10). Important modulation of T cell responses related to exposed not sensitized (putatively tolerant) children were observed at 18 and 24 months of age in this study. These findings led us to select these two time points for assessment of LILIRB1 and LILRB2 expression in immune cell populations. Therefore, peripheral blood at 18 and 24 months was collected to perform the cellular immune cell phenotyping including LILRB1 and LILRB2 expression, to determine antibodies, cytokines and HLA-G concentrations.

We previously showed the importance to take into account the local malaria transmission to demonstrate the immune tolerance hypothesis (20). In the present study, Center for Disease Control light traps were used on two consecutive nights per month between April 2011 and February 2013. The traps were installed at dark and collected at dusk in 180 of the 400 infant’s bedrooms. The selection of 180 bedrooms was made on the basis of geographical equidistribution criteria. All mosquitoes were brought to the laboratory for identification (genus and species) and all Anopheles mosquitoes were kept individually for further analysis (21). Climatic factors as well as household characteristics were recorded throughout the study in all household. The following information was systematically collected: soil type, vegetation index, proximity of a watercourse, type of roof, number of windows and doors, number of inhabitants, ownership of a bed-net or insect repellent, presence of abandoned objects susceptible to be used as oviposition sites for female mosquitoes and presence of areas where building constructions are ongoing with tools or holes representing potential breeding habitats for anopheles. From these data, a predictive regression model was developed to predict the spatiotemporal malaria exposure risk for each infant (22).

The Enzyme Linked ImmunoSorbent Assay (ELISA) standard operating procedures developed by the African Malaria Network Trust was used to assess antibody concentrations to MSP1, MSP2, MSP3, AMA1 and GLURP antigens. All details about the recombinant proteins and the procedure are described by Dechavanne et al. (23).

Concentrations of IL-4, IL-10, TNF-α and IFN-γ were determined using BD OptEIA ELISA Set (BD Biosciences, San Diego, CA, USA) following the manufacturer’s protocol and as previously described by our team (24). The levels of soluble HLA-G isoforms HLA-G1 and HLA-G5 were determined by ELISA assay as we described previously (17).

Phenotype analysis was performed on infant peripheral blood (1 mL) the day of collection. The expression of LILRB1 and/or LILRB2 inhibitory receptors on the surface of the following cell populations was analyzed: T helper cells (CD3⁺CD4⁺), T cytotoxic cells (CD3⁺CD8⁺), T regulatory cells (CD3⁺CD4⁺CD25^highCD127^-), T effector cells (CD3⁺CD4⁺CD25⁺CD127⁺), γδ T cells (CD3⁺Vd2⁺) and B cells (CD19⁺). NK cell subsets were characterized by expression of CD56 and CD16 among the mononuclear lymphoid cells. Blood monocyte subpopulations were characterized by expression of CD14 and CD16 markers in the population of mononuclear cells. Neutrophils and eosinophils are respectively CD16⁺ and CD16^- in the polymorphonuclear cell population. A classical gating strategy was used to investigate monocyte dysregulation during Plasmodium falciparum infection in children. A second analysis by contour plot was also applied to validate the gates of our flow cytometry analysis for all samples. Due to the inter-individual variability in the proportions of monocyte populations or the variations induced by parasite infection, the selection of monocyte subsets was adapted for each sample. To define positive and negative events, isotype-matched control antibodies were used. Flow cytometer data were analyzed using FlowJo software version 10. The gating strategy is presented in Figure 1 .

The study was conducted in the context of the MiPPAD “Malaria in Pregnancy Preventive Alternative Drugs,” (http://clinicaltrials.gov/ct2/show/NCT00811421) and TOLIMMUNPAL programs. Both projects were approved by the Ethics Committee of the Faculté des Sciences de la Santé de Cotonou. The TOLIMMUNPAL study protocol and informed consent were also approved by the Comité Consultatif de Déontologie et d’Éthique (CCDE) of the Institut de Recherche pour le Développement (IRD, France).

Confounders were environmental exposure (22), infant sex, infant age, birth weight, health centre, ethnic group, maternal anemia and maternal IPTp. A Chi-square test was performed to assess the differences between maternal or neonatal characteristics. The risk of malaria infection between 18 and 24 months of age was compared between infants born to mother with and without PM using the total number of Pf infections during the 6-month period and a negative binomial regression adjusted on confounders.

Univariate analysis was performed at 18 and 24 months of age using a linear regression to assess (i) the effect of PM on the frequency of immune cell populations and on LILRB1 and LILRB2 expression and (ii) the effect of active malaria infection on LILRB1 and LILRB2 expression, on cytokines and HLA-G levels.

For multivariate analysis, a mixed linear model was used, taking into account the measurement at 18 and 24 months of age together in one analysis. The effect of PM or of an active malaria infection was assessed on the frequency of monocytes, LILRB1 and LILRB2 expression, cytokines, HLA-G and anti-malarial antibody concentrations. All models were adjusted on potential confounders and on the frequency of each subset of immune cell in the LILRB1 and LILRB2 expression analyses. For antibody and cytokines levels analysis, an interaction term was included to determine if the effect of an active malaria infection at the moment of blood draw depends on whether the infant is born to mother with or without PM. Data were analysed with Stata^® Software, Version 13 (StatCorp LP, College Station, TX, USA) and the graphs were done using Graph Pad Prism (Version 8.1.2).

Funders had no role in study design; collection, analysis and interpretation of data; writing of manuscript; and in the decision to submit the paper for publication.

Raw experimental data associated with the figures presented in the manuscript are available from the corresponding author upon reasonable request.

The objective of this study was to investigate the immune response at 18 and 24 months of age in infants born to mothers with or without PM. None of the main characteristics of the studied population were found to differ between infants born to mothers with a Pf-infected or an uninfected placenta ( Table 1 ). However, infants born to mothers with PM had a significantly higher risk of developing symptomatic malaria (adjusted negative binomial regression IRR=1.533; [1.020;2.305], p=0.04) between 18 and 24 months of age. There were no differences in the numbers of asymptomatic infections between the two groups of infants.

Lymphoid and myeloid immune cells were analyzed in order to assess the effect of PM on infant immunity at 18 and 24 months of age. To this end, a flow cytometry gating strategy was set-up to analyze the three monocyte subsets defined as classical (CD14⁺⁺CD16^-), intermediate (CD14⁺⁺CD16⁺) and non-classical (CD14⁺CD16⁺⁺) ( Figure 1 ) as well as lymphoid populations, including CD4 T cells, CD8 T cells, NK cells, γδ T cells or B cells, neutrophils and eosinophils ( Figures S1 , S2 ). As shown in Figure 2 , PM did not alter the proportions of CD4 T cells, CD8 T cells, NK cells, γδ T cells, neutrophils, eosinophils or B cells in infants. Interestingly, PM was associated with a higher frequency of classical monocytes and a lower frequency of intermediate and non-classical monocytes. In the multivariate analyses, the proportion of non-classical monocytes remained significantly lower in infants born to mothers with PM, independently of active malaria infection at the time of blood sample collection ( Table 2 ). Of note, malaria infection at the time of sample collection was associated with a reduced frequency of classical monocytes and an increased frequency of intermediate and non-classical monocytes ( Table 2 ). LILRB2 expression was higher on non-classical monocytes, and enhanced to a lesser extent on intermediate monocytes in infants born to mothers with PM as compared to infants from mothers without PM ( Table 3 ). PM did not seem to influence LILRB1 expression by either classical, intermediate or non-classical monocytes ( Table 3 ). Active malaria infection at the time of blood collection can influence LILRB1 and/or LILRB2 expression ( Figures 3 , 4 ). Therefore, infants with active malaria infection at the time of blood collection were excluded from the multivariate analyses ( Table 3 ).

LILRB1 and LILRB2 receptors were expressed at higher levels on intermediate and non-classical monocytes in infants with active Pf infection ( Figures 3 , 4 ). In a multivariate analysis, we confirmed that during active Pf infection, both receptors were highly expressed on intermediate and non-classical monocytes ( Table 4 ) and LILRB1 expression decreased with age. Moreover, levels of LILRB1 and LILRB2 expression were higher during symptomatic attacks than during asymptomatic infections ( Figures 3 , 4 ). Finally, out of all the other cell populations evaluated in this project, the levels of LILRB1 were only higher on γδ T cells in infants infected with Pf than in infants without malaria infection at blood draw ( Figure 5 ). The multivariate analysis including the level of LILRB1 expression at 18 and 24 months confirmed that during active Pf infection, LILRB1 receptor was highly expressed on γδ T cell surface in infant with malaria infections (p=0.003; coef. 0.72; adjusted mixed model). A decrease of LILRB1 expression was also observed with age on γδ T cell surface in the mixed model (p=0.05; coef.=-0.051).

At 18 and 24 months of age, cytokines (IL-4, IFN-γ, TNF-α and IL-10) and soluble HLA-G concentrations were quantified. No differences were observed as a function of PM or of active Pf infection except for IL-10 levels. IL-10 levels were higher in those with active malaria infections [febrile and/or asymptomatic ( Figure 6 )]. These observations were confirmed by multivariate mixed models for malaria infections (Coef=2.08; [1.587;2.579]; p<0.001), symptomatic infections (Coef=2.02; [1.455;2.950]; p<0.001) and asymptomatic infections (Coef=1.68; [0.967;2.387]; p<0.001), and mostly in infants with PM (interaction active infection*PM; Coef=2.58; [1.155;4.012]; p<0.001).

Higher anti-malarial antibody concentrations between 18 and 24 months of age were observed during active malaria infections ( Table 5 ) mostly IgG1 and IgG3 responses against all vaccine candidate antigens (except for IgG3 to MSP1) whereas almost no IgG2 or IgM responses were affected. To understand how anti-malarial antibody levels vary in infants born from women with PM during an active Pf infection, we introduced an interaction term (interaction active infection*PM, Table 5 ) in the model. Infants born to mothers with PM seemed less able to mount an antibody responses during active infections ( Table 5 ). Indeed, IgG1 to AMA1, MSP2 and MSP3, IgG2 to MSP2 and IgG3 to AMA1 and MSP2 were significantly lower in Pf-infected infants born to women with PM ( Table 5 and Figure 7 ).