Jacqueline Milet

Malgré les moyens importants de prévention et de lutte mis en place ces dernières années, le paludisme reste dévastateur avec près d’un demi-million de décès par an (405 000 en 2018, d'après le dernier rapport de l'OMS). Le rôle clé joué par les facteurs génétiques de l’hôte dans la susceptibilité et la sévérité de la maladie est admis aujourd'hui. Cependant, les bases moléculaires de la sensibilité/résistance au paludisme restent encore mal connues. Ces dix dernières années, les efforts de recherches pour l’identification de gènes impliqués dans la sensibilité au paludisme à P. falciparum se sont concentrés sur les formes graves de paludisme, avec plusieurs plusieurs études d’association sur l’ensemble du génome (Genome-Wide Association Study ou GWAS) publiées. Ce manuscrit porte sur l’extension de cette approche aux formes simples du paludisme, au travers de l’étude d’association génome entier de deux cohortes de nouveau-nés au Sud Bénin (au total 800 enfants), suivis pendant 18-24 mois par l’UMR261 (MERIT IRD/Université de Paris). Dans une première partie nous présentons les résultats de la première GWAS réalisée sur les formes simples de paludisme dans ces deux cohortes. L’association a été testée avec la récurrence des accès palustres et la récurrence de l’ensemble des infections (incluant les accès palustres et les infections asymptomatiques) en prenant en compte un risque environnemental estimé au niveau individuel. Elle met en évidence plusieurs signaux d’association forts, en lien avec des gènes dont la fonction biologique est pertinente pour le paludisme (notamment PTPRT, MYLK4, UROC1 et ACER3). La forte variabilité génétique présente au sein des populations africaines a nécessité de prendre en compte l’effet de confusion potentiel de la structure de population. Dans l’étude les formes simples de paludisme, une approche en deux étapes a été utilisée, le modèle de Cox mixte, utilisé pour l’analyse des données longitudinales, n’étant pas applicable à l’ensemble du génome du fait du temps de calcul nécessaire. Un modèle de Cox mixte a été appliqué pour construire un « effet individuel » ajusté sur les covariables, puis un modèle mixte linéaire pour tester l’association avec les polymorphismes du génome. Ceci nous a conduits à nous intéresser plus généralement aux modèles mixtes non-linéaires. Deux méthodes permettant l’estimation de l’effet des polymorphismes avec le modèle logistique mixte sont proposées, qui pourront être dans le futur généralisé à d’autres modèles, dont le modèle de Cox. Dans une dernière partie, le paludisme ayant constitué une des plus fortes pressions de sélection que l’homme ait connue dans son histoire récente, nous explorons la possibilité d’exploiter l’information de sélection naturelle pour augmenter la puissance de l’analyse, et améliorer la détection des signaux d’association. L’analyse des signaux de sélection positive récente sur l’ensemble du génome a été réalisée avec plusieurs méthodes basées sur les haplotypes longs ((iHS, nsL and XP-EHH). Celle-ci met en évidence plusieurs régions chromosomiques d’intérêt potentiel où les signaux d’association et de sélection co-localisent ; mais confirme également la difficulté à mettre en évidence les signaux de sélection liés au paludisme avec les outils disponibles actuellement.

This paper aimed to investigate the influence of polymorphisms in the FCGR2A gene encoding R131H FcgRIIA variants and in the FCGR3B gene (108G > C, 114C > T, 194 A > G, 233C > A, 244 G > A and 316G > A) encoding FcgRIIIB-NA1, -NA2 and -SH variants on malaria susceptibility and antibody responses against P. falciparum merozoite antigens in Beninese children. An active malaria follow-up was conducted in infants from birth to 24 months of age in Allada, Benin. FCGR3B exon 3 was sequenced and FCGR2A exon 4 was genotyped. Antibodies directed to GLURP and MSP3 were quantified by ELISA. Association studies were performed using mixed-effect models. Individual carriage of FCGR3B 194 AA genotype was associated with a high number of malaria infections and a low level of IgG1 against MSP3 and GLURP-R0. High parasitemia and increased malaria infections were observed in infants carrying the FCGR3B*05 108C-114T-194A-233C-244A-316A haplotype. A reduced risk of malaria infections and low parasitemia were related to the carriages of the FCGR3B 108C-114T-194G-233C-244G-316A (FCGR3B*06), FCGR3B 108C–114T–194G–233A–244A–316A (FCGR3B*03 encoding for FcgRIIIB-SH) haplotypes and FCGR3B 297 TT genotype. Our results highlight the impact of FCGR3B polymorphisms on the individual susceptibility to malaria and antibody responses against MSP3 and GLURP in Beninese children.

Background: Mixed linear models (MLM) have been widely used to account for population structure in case-control genome-wide association studies, the status being analyzed as a quantitative phenotype. Chen et al. proved in 2016 that this method is inappropriate in some situations and proposed GMMAT, a score test for the mixed logistic regression (MLR). However, this test does not produces an estimation of the variants’ effects. We propose two computationally efficient methods to estimate the variants’ effects. Their properties and those of other methods (MLM, logistic regression) are evaluated using both simulated and real genomic data from a recent GWAS in two geographically close population in West Africa.Results: We show that, when the disease prevalence differs between population strata, MLM is inappropriate to analyze binary traits. MLR performs the best in all circumstances. The variants’ effects are well evaluated by our methods, with a moderate bias when the effect sizes are large. Additionally, we propose a stratified QQ-plot, enhancing the diagnosis ofpvalues inflation or deflation when population strata are not clearly identified in the sample.Conclusion: The two proposed methods are implemented in the R packagemilorGWASavailable on the CRAN. Both methods scale up to at least 10,000 individuals. The same computational strategies could be applied to other models (e.g. mixed Cox model for survival analysis).

Aims: Schistosomiasis and malaria are endemic in sub‐Saharan Africa where Schistosoma haematobium (Sh) and Plasmodium falciparum (Pf) coinfections are thus frequent. We explored the effect of Sh infection on antibody responses directed to Pf merozoite antigens and on malaria susceptibility in Beninese children.Methods and results: A total of 268 children were followed during a malaria transmission season. Detection of Pf infection was performed by microscopy and rapid diagnostic tests. Sh infection was determined in urine by microscopy. Antimalarial antibody, cytokine and HLA‐G concentrations were quantified by ELISA. The expression of HLA‐G receptors by immune cells was assessed by flow cytometry. Children infected by Sh had higher concentrations of IgG1 directed to MSP3 and GLURP_R0, IgG2 directed to GLURP_R0 and IgG3 directed to MSP3, GLURP_R0 and GLURP_R2 and have lower Pf densities than those uninfected by Sh. No difference in cytokine and HLA‐G concentrations was observed between Sh egg carriers and non‐carriers.Conclusion: Schistosoma haematobium modulates host immune responses directed to Pf antigens. The absence of immune downregulation usually observed during helminth infections is surprising in our study. We hypothesize that the stage of Sh development could partly explain the immune pathways leading to increased antibody levels that favour better control of Pf parasitemia.

Motivation: Mixed linear models (MLM) have been widely used to account for population structure in case-control genome-wide association studies, the status being analyzed as a quantitative phenotype. Chenet al.proved that this method is inappropriate and proposed a score test for the mixed logistic regression (MLR). However this test does not allow an estimation of the variants’ effects.Results: We propose two computationally efficient methods to estimate the variants’ effects. Their properties are evaluated on two simulations sets, and compared with other methods (MLM, logistic regression). MLR performs the best in all circumstances. The variants’ effects are well evaluated by our methods, with a moderate bias when the effect sizes are large. Additionally, we propose a stratified QQ-plot, enhancing the diagnosis ofp-values inflation or deflation, when population strata are not clearly identified in the sample.Availability: All methods are implemented in the R packagemilorGWASavailable athttps://github.com/genostats/milorGWAS.Contact: herve.perdry@u-psud.frSupplementary information: Supplementary data are available atBioinformaticsonline.

Recent research efforts to identify genes involved in malaria susceptibility using genome-wide approaches have focused on severe malaria. Here, we present the first GWAS on non-severe malaria designed to identify genetic variants involved in innate immunity or innate resistance mechanisms. Our study was performed on two cohorts of infants from southern Benin (525 and 250 individuals used as discovery and replication cohorts, respectively) closely followed from birth to 18–24 months of age, with an assessment of a space- and time-dependent environmental risk of exposure. Both the recurrence of mild malaria attacks and the recurrence of malaria infections as a whole (symptomatic and asymptomatic) were considered. Post-GWAS functional analyses were performed using positional, eQTL, and chromatin interaction mapping to identify the genes underlying association signals. Our study highlights a role of PTPRT, a tyrosine phosphatase receptor involved in STAT3 pathway, in the protection against both mild malaria attacks and malaria infections (p = 9.70 × 10⁻⁸ and p = 1.78 × 10⁻⁷, respectively, in the discovery cohort). Strong statistical support was also found for a role of MYLK4 (meta-analysis, p = 5.29 × 10⁻⁸ with malaria attacks), and for several other genes, whose biological functions are relevant in malaria infection. Results shows that GWAS on non-severe malaria can successfully identify new candidate genes and inform physiological mechanisms underlying natural protection against malaria.

Lead and malaria both present significant health risks to children in Sub-Saharan Africa. Previous studies have shown that high blood lead levels in children act as a protective factor against subsequent malaria incidence. The main objective of this study was to investigate associations between blood lead level and malaria outcomes prospectively in Beninese children from 12 to 24 months of age. Two-hundred and four children were assessed for lead at 12 months and closely followed until 24 months for malaria; when symptoms and parasite density were also recorded. Univariate and multivariate negative binomial and linear regression models tested associations between blood lead level quartile and total episodes of malaria (total symptomatic and asymptomatic episodes) and parasite density, respectively. Median blood lead level among children measured at 12 months was 56.50 (4.81–578) μg/L. During the 12-month follow-up, 172 (84.31%) children had at least one malaria episode. Univariate and multivariate negative binomial and linear regressions did not reveal significant associations between blood lead level quartile and malaria outcomes. Iron deficiency was not found to be an effect modifier. Results from this prospective child-cohort study investigating associations between blood lead level and malaria did not confirm results from previous cross-sectional studies. Further research is needed to further explore this relationship and other co-morbidities due to malaria and lead.

Placental malaria has been associated with an immune tolerance phenomenon and a higher susceptibility to malaria infection during infancy. HLA-G is involved in fetal maternal immune tolerance by inhibiting maternal immunity. During infections HLA-G can be involved in immune escape of pathogens by creating a tolerogenic environment. Recent studies have shown an association between the risk of malaria and HLA-G at both genetic and protein levels. Moreover, women with placental malaria have a higher probability of giving birth to children exhibiting high sHLA-G, independently of their own level during pregnancy. Our aim was to explore the association between the level of maternal soluble HLA-G and the risk of malaria infection in their newborns. Here, 400 pregnant women and their children were actively followed-up during 24 months. The results show a significant association between the level of sHLA-G at the first antenatal visit and the time to first malaria infection during infancy adjusted to the risk of exposure to vector bites (aHR = 1.02, 95%CI [1.01–1.03], p = 0.014). The level of sHLA-G is a significant predictor of the occurrence of malaria infection during infancy consistent with the hypothesis that mother sHLA-G could be a biomarker of malaria susceptibility in children.

Substantial evidence indicates that cytophilic IgG responses to Plasmodium falciparum merozoite antigens play a role in protection from malaria. The specific targets mediating immunity remain unclear. Evaluating antibody responses in infants naturally-exposed to malaria will allow to better understand the establishment of anti-malarial immunity and to contribute to a vaccine development by identifying the most appropriate merozoite candidate antigens. The study was based on parasitological and clinical active follow-up of infants from birth to 18 months of age conducted in the Tori Bossito area of southern Benin. For 399 infants, plasma levels of cytophilic IgG antibodies with specificity for five asexual stage malaria vaccine candidate antigens were determined by ELISA in infants’ peripheral blood at 6, 9, 12 and 15 months of age. Multivariate mixed logistic model was used to investigate the association between antibody levels and anti-malarial protection in the trimester following the IgG quantification. Moreover, the concentrations of merozoite antigen-specific IgG were compared between a group of infants apparently able to control asymptomatic malaria infection (CAIG) and a group of infants with no control of malaria infection (Control group (NCIG)). Protective effect of antibodies was also assessed after 15 months of malaria exposure with a Cox regression model adjusted on environmental risk. Cytophilic IgG responses to AMA1, MSP1, MSP2-3D7, MSP2-FC27, MSP3 and GLURP R2 were associated with increasing malarial infection risk in univariate analysis. The multivariate mixed model showed that IgG1 and IgG3 to AMA1 were associated with an increased risk of malarial infection. However infants from CAIG (n = 53) had significantly higher AMA1-, MSP2-FC27-, MSP3-specific IgG1 and AMA1-, MSP1-, MSP2-FC27-, MSP3 and GLURP-R2-specific IgG3 than those from NCIG (n = 183). The latter IgG responses were not associated with protection against clinical malaria in the whole cohort when protective effect is assessed after 15 months of malaria exposition. In this cohort, merozoite antigen-specific cytophilic IgG levels represent a marker of malaria exposure in infants from 6 to 18 months of age. However, infants with resolution of asymptomatic infection (CAIG) seem to have acquired naturally immunity against P. falciparum. This observation is encouraging in the context of the development of multitarget P. falciparum vaccines.

Sickle cell trait (HbAS) confers partial protection against malaria by reducing the adhesion of Plasmodium falciparum-infected erythrocytes to host receptors, but little is known about its potential protection against placental malaria. Using flow cytometry, we assessed the recognition of HbAA and HbAS VAR2CSA-expressing infected erythrocytes, by plasma from 159 Beninese pregnant women with either HbAA (normal) or HbAS. Using multivariate linear models adjusted for gravidity, parasite infection at delivery, glucose-6-phosphate dehydrogenase deficiency, and α-thalassemia carriage, we observed significantly reduced cell surface antibody binding of HbAS-infected erythrocytes by plasma from HbAS compared with HbAA women (P < 10^–3). The difference in cell surface antibody binding was only observed when infected erythrocytes and plasma were associated according to the same hemoglobin genotype. Similar levels of VAR2CSA-specific antibody were measured by enzyme-linked immunosorbent assay in the 2 groups, suggesting that the altered interaction between VAR2CSA and HbAS women’s antibodies could reflect abnormal display of VAR2CSA on HbAS erythrocytes. Our data stress the need for assessments of erythrocyte disorders such as the sickle cell trait in a population group when studying immunological responses to P falciparum.