Functional antibodies against Plasmodium falciparum sporozoites are associated with a longer time to qPCR-detected infection among schoolchildren in Burkina Faso

Background: Individuals living in malaria-endemic regions develop immunity against severe malaria, but it is unclear whether immunity against pre-erythrocytic stages that blocks initiation of blood-stage infection after parasite inoculation develops following continuous natural exposure. Methods: We cleared schoolchildren living in an area (health district of Saponé, Burkina Faso) with highly endemic seasonal malaria of possible sub-patent infections and examined them weekly for incident infections by nested PCR. Plasma samples collected at enrolment were used to quantify antibodies to the pre-eryhrocytic-stage antigens circumsporozoite protein (CSP) and Liver stage antigen 1 (LSA-1). In vitro sporozoite gliding inhibition and hepatocyte invasion inhibition by naturally acquired antibodies were assessed using Plasmodium falciparum NF54 sporozoites. Associations between antibody responses, functional pre-erythrocytic immunity phenotypes and time to infection detected by 18S quantitative PCR were studied. Results: A total of 51 children were monitored. Anti-CSP antibody titres showed a positive association with sporozoite gliding motility inhibition (P<0.0001, Spearman’s ρ=0.76). In vitro hepatocyte invasion was inhibited by naturally acquired antibodies (median inhibition, 19.4% [IQR 15.2-40.9%]), and there were positive correlations between invasion inhibition and gliding inhibition (P=0.005, Spearman’s ρ=0.67) and between invasion inhibition and CSP-specific antibodies (P=0.002, Spearman’s ρ=0.76). Survival analysis indicated longer time to infection in individuals displaying higher-than-median sporozoite gliding inhibition activity (P=0.01), although this association became non-significant after adjustment for blood-stage immunity (P = 0.06). Conclusions: In summary, functional antibodies against the pre-erythrocytic stages of malaria infection are acquired in children who are repeatedly exposed to Plasmodium parasites. This immune response does not prevent them from becoming infected during a malaria transmission season, but might delay the appearance of blood stage parasitaemia. Our approach could not fully separate the effects of pre-erythrocytic-specific and blood-stage-specific antibody-mediated immune responses in vivo; epidemiological studies powered and designed to address this important question should become a research priority.

ρ=0.76). Survival analysis indicated longer time to infection in individuals displaying higher-than-median sporozoite gliding inhibition activity (P=0.01), although this association became non-significant after adjustment for blood-stage immunity (P = 0.06).
In summary, functional antibodies against the Conclusions: pre-erythrocytic stages of malaria infection are acquired in children who are repeatedly exposed to parasites. This immune response does Plasmodium not prevent them from becoming infected during a malaria transmission season, but might delay the appearance of blood stage parasitaemia. Our approach could not fully separate the effects of pre-erythrocytic-specific and blood-stage-specific antibody-mediated immune responses ; in vivo epidemiological studies powered and designed to address this important question should become a research priority.

Introduction
The most advanced malaria vaccine, RTS,S (trade name, Mosquirix), induces immune responses that target P. falciparum circumsporozoite protein (CSP), and thereby the pre-erythrocytic stages of malaria, and has been shown to be partially effective in delaying the time to clinical malaria episodes 1 . Alongside the RTS,S subunit vaccine, several other pre-erythrocytic stage vaccines are under development, based on subunit or whole-parasite vaccination 2-5 . Vaccination with the attenuated sporozoite vaccine PfSPZ resulted in a protective efficacy of ~48%, as quantified by differences in time to first positive blood smears in malariaexperienced adults in Mali 6 . The results of this and other vaccine trials and the efficient immunisation of malaria-naive individuals with multiple infected mosquito bites while receiving chloroquine 7 contrast with the limited epidemiological evidence of naturally acquired functional immunity to Plasmodium pre-erythrocytic stages, which could be possibly linked to the lower number of sporozoites in natural parasite inoculations or to the frequency of host-vector contacts. Individuals living in malaria-endemic regions can develop naturally acquired immunity against severe malaria disease and death 8,9 , but it is unclear whether immunity that reduces, entirely or partially, the probability of blood-stage infection after parasite inoculation develops following natural exposure 9 . The high incidence of blood-stage re-infection after effective anti-malarial treatment in adults living in malaria-endemic regions suggests that sterilizing immunity does not develop even after years of repeated infection 10,11 . Similarly, cohort studies that have analysed the relationship between age and risk of P. falciparum infection showed no evidence for complete protection against infection and conflicting evidence on whether naturally acquired immunity can result in a different time to patency 12,13 . One of the most detailed studies on this topic reported clear negative associations between age and the risk of clinical malaria or microscopy-detected malaria infection, but similar times to PCR-detected infection for all age groups. The study concluded no or very limited evidence for an agedependent acquisition of immunity that protects from infection 13 .

Number of individuals
Screened and parasite-free by microscopy 58 Presence of parasites post-treatment 6 Parasitological and immunological data from the remaining 51 children followed intensively were analysed (Table 1). Every week these study participants were screened for incident infections. All but one study participant had P. falciparum parasites detected by 18S qPCR during follow-up. Malaria infection caused clinical disease in 43/50 children. One child developed symptomatology suggestive of malaria, but no parasites were detected in samples collected before and during the clinical episode; data from this child were censored after the onset of

Amendments from Version 1
The manuscript has been modified to address the comments made by the three reviewers. The main modifications in this version are related to the discussion and interpretation of the results. We now mention that our approach might not have fully addressed the correlation between immune responses against asexual stage antigens and pre-erythrocytic immunity, and that the multivariate analysis performed that included both types of immunity revealed a non-significant association between infection risk and functional pre-erythrocytic-stage immune responses -these modifications can be seen in the Abstract, Results and Discussion section. In this version, we have also removed the term 'sporozoite-specific', that was used in the first version to describe the assay that assessed antibody responses against whole sporozoites, since we did not show the specificity of these responses. Additionally, we have included descriptive analysis on the relationship between sporozoite gliding and hepatocyte invasion assays. We have also included a few sentences that discuss the potential role of mosquito saliva as a factor influencing our results. symptoms. The median time from confirmation of the absence of parasites (i.e. 3 weeks after anti-malarial treatment) to infection detection by nested PCR or onset of symptoms was 28 days; one child who did not have parasites detected by nested PCR was not included in this calculation. Similarly, the median time to parasite detection by 18S qPCR was 30 days. In Figure 1, both the times of first 18S qPCR positive result and, if applicable, of development of clinical disease are presented for all study participants.
Naturally acquired IgG and IgM antibodies targeting sporozoites Malaria antigen-specific antibodies to pre-erythrocytic antigens CSP, liver stage antigen (LSA-1) and to asexual lysate were determined in naturally exposed children and malaria-naive European donors by ELISAs. Antibody titres to the CSP preerythrocytic antigen were on average low in naturally exposed children and not significantly different from malaria-naive donors (Figure 2A, P=0.11; non-parametric tests were used for all comparisons), while LSA-1 antibody levels were significantly higher in malaria-exposed children compared to malaria-naive donors ( Figure 2B, P=0.006). As expected, asexual blood stagespecific antibody titers in naturally exposed children were higher compared to malaria-naive donors ( Figure 2C, P<0.0001). In addition to antigen-specific assays, IgG and IgM antibodies recognizing Plasmodium sporozoites were quantified using fluorescently labelled anti-IgG and anti-IgM antibodies by flow cytometric analysis. Levels of antibodies targeting whole sporozoites were significantly higher in the study participants compared to malaria-naive donors ( Figure 2D, E, P=0.01 and 0.04 for IgG and IgM, respectively). Strong correlations between IgG and IgM antibodies targeting whole sporozoites ( Figure S1A in Extended data 14 , P=0.0005, Spearman's ρ=0.75), and between CSP-IgG antibody levels and whole sporozoite IgG antibodies ( Figure S1B in Extended data 14 , Spearman's ρ=0.83, P<0.0001) were also observed.
Naturally acquired antibodies in children neutralize in vitro sporozoite gliding motility The neutralizing activity of naturally acquired antibodies against sporozoite motility was determined in an in vitro assay. Sporozoites pre-treated with PBS showed an average gliding trail surface of 7,392 (95% confidence interval [CI] 4,151-9,388) pixels.
To characterize the effect of human plasma on sporozoite gliding motility that is independent of naturally acquired immunity, sporozoites were incubated in the presence of plasma from malaria-naive individuals (n=5) and showed an average gliding trail surface of 1,427(95% CI 646.6-2,207) pixels, significantly greater than the gliding trail surface of sporozoites incubated with 30 μg/ml of a monoclonal anti-CSP antibody, our positive control, 141 (95% CI 71.9-225.4) pixels.
Plasma from the majority of cohort participants reduced in vitro sporozoite gliding motility with a median gliding inhibition of 59.6% (IQR 27.8-77.0%); incubation of sporozoites with plasma from three participants resulted in lower gliding trail surfaces than the positive control. We defined two groups: one with poor gliding inhibition (individuals whose plasma inhibited less than 20% of sporozoite gliding motility, N=8) and the other with strong gliding inhibition (more than 80% gliding inhibition, N=8). The first group had similar gliding trail surface (median, 1,353; IQR, 1,311.7-1,639.2) compared to malaria-naive donors (median 1605.7, Figure 2. Naturally acquired pre-erythrocytic antibody levels and their functionality against sporozoite infectivity in vitro. Malaria antigen-specific antibody levels in children from Burkina Faso (n=51) and European malaria-naive adults (n=9) to the pre-erythrocytic antigens (A) circumsporozoite protein (CSP), (B) liver-stage antigen-1 (LSA-1), and (C) asexual lysate as an internal control were determined by ELISAs and expressed as arbitrary units (AU). The amount of (D) IgG and (E) IgM antibodies recognizing sporozoites was determined by flow cytometry and shown as the geometric mean fluorescence intensity (MFI). To this end, P. falciparum NF54 sporozoites were pre-treated with 10% heat-inactivated plasma from children from Burkina Faso (N=16) and malaria-naive adults (N=8) and stained with fluorescently labelled antibodies against IgG and IgM antibodies. (F) The gliding motility of P. falciparum NF54 sporozoites, pre-treated with plasma from children from Burkina Faso (N=51) and malaria-naive adults (N=5), was determined by in vitro gliding motility assays. Gliding trail length of sporozoites incubated with plasma from malaria-naive donors (N=5) or naturally exposed children who were poor (N=8) or strong (N=8) inhibitors of gliding motility are shown. (G) The percent hepatocytes invaded by P. falciparum NF54 sporozoites pre-treated with plasma from children from Burkina Faso (N=16) who were shown to be either poor (n=8) or strong (n=8) gliding inhibitors and malaria-naive adults (N=9) was determined by in vitro invasion assays in human hepatoma cells. Comparisons between multiple groups were tested with Kruskal-Wallis test with Dunn's multiple comparison post hoc test.
IQR 824.3 -1,652) ( Figure 2F), whilst strong gliding inhibitors had a median trail surface of 186.2 (IQR 9.4 -208.6), which was significantly lower compared to malaria-naive donors ( Figure 2F, P<0.01). In vitro gliding inhibition did not correlate with LSA-1 IgG antibody levels (P=0.11, Spearman's ρ=0.23), but correlated with CSP-specific IgG antibodies ( Figure S2A in Extended data 14 , P<0.0001, Spearman's ρ=0.76), whole sporozoite IgG ( Figure S2B in Extended data 14 , P<0.0001, Spearman's ρ=0.81) and IgM antibodies ( Figure S2C in Extended data 14 , P=0.01, Spearman's ρ=0.61). While levels of IgG and IgM antibodies targeting whole sporozoite did not significantly differ between poor gliding inhibitors and malaria-naive adults, whole sporozoite antibody levels of strong gliding inhibitors were significantly higher compared to poor gliding inhibitors ( Figure S2D in Extended data 14 , P<0.0001 and P=0.005 for IgG and IgM antibodies, respectively) and malaria-naive adults ( Figure S2D, E in Extended data 14 , P<0.0001 and P=0.0006 for IgG and IgM antibodies, respectively). This suggests that quantitative differences in measured immune responses might explain variation in these functional phenotypes.
Antibodies in malaria-exposed children neutralize in vitro sporozoite infectivity of hepatocytes The inhibitory effect of naturally acquired antibodies on in vitro sporozoite invasion of hepatocytes was also assessed in a selected number of samples shown to inhibit gliding motility strongly (n=8) or poorly (n=8). In vitro invasion was inhibited by naturally acquired antibodies (median invasion inhibition, 19.4% [IQR, 15.2-40.9%]), and plasma from children categorized as strong gliding inhibitors (see previous section) also prevented hepatocyte invasion more effectively compared to malaria-naive donors ( Figure 2G, P<0.001). 7/8 plasma samples from strong gliding To assess the effect of immune responses on infection risk, children were classified based on whether the results of their assays were higher or lower than the study population median (see Results): in (A), vertically aligned cells represent the same participant, and orange cells indicate that assay results are higher than the median (see also Figure S4, that shows a scatter plot of circumsporozoite protein (CSP) antibody levels and immunity against asexual lysate). In (B), Kaplan-Meier curves for children with high and low circumsporozoite protein CSP responses and gliding inhibition (GI) phenotypes are presented (N = 51); in (C), curves for participants categorized based on their responses to asexual stage antigens (AL) are shown (N = 51). In both panels (B) and (C), the y-axis corresponds to the percentage of the population uninfected at different time points.
Evidence of natural risk-modifying pre-erythrocytic immunity For each immunological assay, ELISA or sporozoite gliding motility assays, children were categorized in two groups: participants with high antibody responses (or high sporozoite gliding inhibition activity) were those with assay values higher than the study population median ( Figure 3A); children considered to have low antibody responses or low sporozoite gliding inhibition capacity had values lower than the median. Based on this categorization, study subjects with high CSP responses and those whose plasma inhibited sporozoite gliding movement acquired blood-stage P. falciparum infection (qPCR-based parasitaemia ≥ 0.1 parasites/μl) later compared to children with lower CSP responses and less efficient gliding inhibitory activity ( Figure 3B; P=0.05 and P=0.01 for CSP responses and sporozoite gliding inhibition, respectively). High anti-LSA-1 antibody levels, on the other hand, did not influence time to infection (P=0.31). Blood-stage immunity, i.e. high response in the asexual stage lysate assay, was also associated with longer time to PCR-detected infection ( Figure 3C, P=0.005). We repeated these analyses excluding children with haemoglobin S and haemoglobin C mutations, as these conditions might influence both immunity 15 and parasite carriage 16,17 . Despite the limited number of individuals included in this analysis (n=36), similar results were obtained (P=0.01 for both CSP and asexual stage lysate assays and P=0.05 for the sporozoite gliding inhibition Since children with high CSP responses were more likely to have high antibody responses to asexual blood stage parasite lysate compared to low CSP responders (60% versus 38.5%, respectively; P=0.12), multivariate Cox proportional hazards models were fit to assess the mutually adjusted effects of these immune responses on time to infection detection (Table 2). In a model that also included the results of the asexual stage lysate assay, the relationship between anti-CSP responses and time to malaria infection was not statistically significant. Despite a clear trend for a protective effect, the association between inhibition of in vitro sporozoite gliding motility and time to falciparum infection also did not reach statistical significance after adjustment for blood-stage immunity (hazard ratio, 0.55; 95% CI, 0.29-1.01).

Discussion
In this longitudinal study in Burkina Faso, we analysed the associations between functional immune responses against pre-erythrocytic stages of malaria and the incidence of qPCRdetected infections in children aged 5-10 years. We observed that (i) schoolchildren develop antibody responses that can interfere with sporozoite motility and infectivity in vitro, and (ii) those children with antibodies that more efficiently reduced in vitro sporozoite gliding motility remained uninfected for longer periods of time. Our epidemiological data together with the in vitro data provide evidence that there is partially effective pre-erythrocytic immunity that influences individual level infection incidence, although we cannot exclude the contribution of immunity against asexual blood-stage antigens to this observation.
Sterile protection against malaria can be readily demonstrated in human and animal experimental models 7,18,19 , but not following natural malaria exposure 13 . Field studies showed conflicting evidence on whether naturally acquired immunity can result in delays to patency and showed no evidence for sterile immunity completely preventing the appearance of parasites in the blood stream of exposed individuals 11,13 . We selected children aged 5-10, who have lower blood stage immune responses (a potential confounder when studying pre-erythrocytic immunity) compared to semi-immune adults, and who allow repeated blood sampling compared to toddlers. Since submicroscopic infections are prevalent in the area 20 , we used a curative dose of antimalarials to clear possible sub-patent infections and, upon confirmation that children were parasite-free at the start of the transmission season, assessed the infection incidence by sensitive molecular assays. Our finding that 98% of the cohort became infected with P. falciparum within 3 months confirms the high force of infection in the area.
Consistent with previous sero-epidemiological studies 21-24 , our results indicate that antibodies to P. falciparum sporozoite and liver-stage antigens are acquired following natural exposure. Antibody titres to the CSP antigen were on average low in our cohort and not significantly different from malaria-naive donors. These low titres could be related to the timing of the study recruitment (prior to the transmission season after ~7 months of very low malaria exposure), which would imply that responses to CSP are short lived. The antibody levels could also have been influenced by repeated blood-stage infections in our cohort, which may have suppressed immune responses against the pre-erythrocytic stages 25 , although, in theory, responses to both LSA-1 and CSP would be affected. To assess the contribution of pre-erythrocytic humoral immunity to protection, previous studies have related antibodies at baseline with time-to-infection after parasite clearance with antimalarials 12 . A significant challenge in these studies is that both protective immunity and cumulative exposure increase with age, and so it is often unclear whether measured responses mediate protection or are merely a marker of past exposure 26,27 . In an attempt to move beyond indirect epidemiological associations, we explored functional anti-sporozoite immunity by assessing the ability of plasma to inhibit sporozoite gliding motility 28 and hepatocyte invasion 29 and related these in vitro phenotypes to field findings. We observed that CSP IgG antibodies showed a strong positive association with sporozoite gliding motility inhibition. It has been demonstrated that Plasmodium parasites use the system of adhesion-based motility, gliding, to actively penetrate host cells 30 ; and that the invasive ability of sporozoites is associated with their motility 31 . In agreement with these findings, the data presented here suggest that in vitro gliding inhibition by naturally acquired antibodies may be a useful surrogate marker for in vitro hepatocyte invasion. In an immuno-epidemiological study undertaken in Indonesia, hepatocyte invasion inhibition was associated with higher anti-CSP antibody titres 32 . In our cohort, subjects with high CSP responses (i.e. higher than the study population median) and those whose plasma more efficiently inhibited sporozoite gliding movement developed P. falciparum infection later compared to children with lower responses. To our knowledge, this is the first study to show that there are functional antibodies against pre-erythrocytic malaria stages in malariaexposed children.
Our study has several limitations. We observed that high responses in the asexual stage lysate assay with unknown functionality were also associated with longer time to infection. We thus cannot rule out a supportive role for asexual antibody responses in the observed associations 33 . Indeed, as pointed out by one of the reviewers of the first version of this manuscript, 17/29 (58.6%) children with high anti-sporozoite responses in either CSPspecific or sporozoite gliding assays had above-median asexual lysate responses versus 8/22 (36.4%) children with below-median anti-sporozoite activities. We believe it is likely that the functionally important pre-erythrocytic antibody responses that we quantified here are acquired alongside anti-blood-stage antibodies. In addition, a recent study in the same geographical area demonstrated that heterogeneity in mosquito exposure contributes considerably to heterogeneity in parasite inoculation risk 34 . In the current study, we used delayed time to blood-stage infection as a simplistic indicator of partial protection, which fails to take into account variation in exposure. In an ideal approach, we would have been able to quantify malaria exposure at an individual level, which may involve linking of blood meals in householdcaught mosquitoes to household occupants and determining sporozoites in the salivary glands of these mosquitoes 35 . Measuring exposure at individual level in such an approach will help to shed further light on pre-erythrocytic immunity in naturally exposed individuals. With respect to discriminating between preerythrocytic and blood-stage immunity, a valuable but laborious approach would be to examine the observed relationships in a larger cohort that allows stratification based on similar blood stage immunity but different levels of pre-erythrocytic immunity at baseline and that provides sufficient power to detect weaker associations. Another limitation of our study is that we were only able to determine humoral responses. Cellular responses to preerythrocytic stages have been implicated in malaria protection in multiple studies 36,37 . It is conceivable that by quantifying both antibody and cellular responses we would be able to better define natural immunological phenotypes associated with differential malaria risk 38 . Immunity to mosquito saliva might have also influenced sporozoite invasion although the magnitude of such an effect compared to the established impact of anti-sporozoite responses is currently unknown and it is currently unknown whether this would require responses to antigens that are conserved between A. stephensi mosquitoes used for sporozoite production and vectors that study participants were naturally exposed to 39.
In summary, in our cohort of children, anti-CSP antibodies were strongly associated with in vitro sporozoite gliding inhibition and hepatocyte invasion inhibition. Children with functional anti-sporozoite antibody responses had a longer time to P. falciparum infection compared to children with lower functional responses, suggesting that these in vitro assays are relevant to understand natural protection. The partial protection (i.e., delay in infection) observed in our study does not prevent individuals from becoming infected during an entire transmission season, but reduces infection incidence and consequently needs to be considered in epidemiological studies aiming to understand malaria risk heterogeneity and in malaria vaccines trials. Identifying host or parasite factors linked to these functional immunological phenotypes and characterizing how these phenotypes change with cumulative exposure to malaria parasites will help the understanding of why natural immunity against pre-erythrocytic stages is incomplete.

Study design
This study was performed from June to December 2015 in the village of Balonghin in the Saponé health district, Burkina Faso, which is exposed to intense and seasonal P. falciparum transmission 20 . Written informed consent was provided by the parent or guardian of each child. The study was approved by the ethics committees of the London School of Hygiene and Tropical Medicine (reference number 9008) and the Ministry of Health in Burkina Faso (reference number 2015-3-033). Children aged 5-10 years with haemoglobin levels above 8 g/dl and no Plasmodium parasites detected by microscopy were eligible. DHA-PQ was used to clear sub-microscopic infections. At 3 weeks (20-22 days) after treatment, finger-prick blood samples were collected to ensure parasite negativity by nested PCR 40 prior to formal enrolment into the cohort. Citrated plasma samples were collected before treatment using citrated vacutainer cell preparation tubes (CPT vacutainers, Becton Dickinson), stored at -80°C and used for malaria-antigen-specific IgG ELISAs and sporozoite assays. Peripheral blood mononuclear cells (PBMC) were also collected, but were lost due to the inability to maintain liquid nitrogen supplies during civil unrest in Ouagadougou. Following enrolment, participants were examined during weekly visits, when finger-prick samples were collected for P. falciparum nested PCR that was performed within 48 hours. Following parasite detection, finger-prick blood samples were collected every day for 1 week, and every week afterwards, up to 35 days after parasite detection. Study participants were closely monitored for the development of malaria symptoms. Artemetherlumefantrine was given upon the detection of symptoms or 35 days after initial detection of infection by nested PCR, whichever came first. For the current analyses, only the time to first infection detection was used and related to baseline immunological assays.

Molecular analyses
Nucleic acids from 100 μl whole-blood samples stored in RNAprotect Cell Reagent were extracted using MagNAPure LC automatic extractor (Total Nucleic Acid Isolation Kit-High Performance, Roche Applied Science) and used for qPCR targeting 18S rRNA 41 . Genomic DNA from the same extraction was used to test for human haemoglobinopathies haemoglobin S and C 42 .
Parasite culture and generation of P. falciparum-infected mosquitoes As source of sporozoites, Anopheles stephensi mosquitoes were infected by standard membrane feeding on P. falciparum NF54 gametocyte cultures 43 . Salivary glands from infected mosquitoes were dissected, collected in Leibovitz culture medium (Lonza) without serum (supplemented with 1% penicillin-streptomycin and 1% L-glutamine for in vitro gliding motility assays), and homogenized in a homemade glass grinder. The number of sporozoites was counted in a Bürker-Türk counting chamber using phase contrast microscopy 18 . Enzyme-linked immunosorbent assays and sporozoite opsonization assays Levels of antibodies were determined to circumsporozoite protein (CSP: full-length P. falciparum NF54 CSP with repeats, produced in E. coli by Gennova Biopharmaceuticals Ltd., Pune, India), LSA-1 (LSA-NRC construct expressed in E. coli) and asexual lysate using previously reported standardized enzyme-linked immunosorbent assays (ELISAs) 44,45 in naturally exposed children (n=51) and malaria-naive European donors (n=9). Antibody levels were calculated in relation to the positive control (hyperimmune plasma pool from Tanzania) that was set at 100 arbitrary units (AU) using Auditable Data Analysis and Management System (ADAMSEL, version 1.1) 19 .
Recognition of whole sporozoites by naturally acquired IgG and IgM antibodies was determined by an in vitro flowcytometry-based antibody opsonization assay that was presented in detail elsewhere 46 . Due to limited plasma availability and available sporozoite numbers, 16 naturally exposed children with highest (8) and lowest (8) gliding activity were selected for sporozoite opsonization assays and invasion assays, allowing us to investigate potential correlations. Flow cytometric analysis was performed with a LSRII flow cytometer (BD BioSciences); data analysis by FlowJo software (version 10.0.8, Tree Star).
Malaria-naive donors are healthy malaria-naive European volunteers who participated in CPS-immunization trial (immunization of malaria-naive human volunteers under chloroquine prophylaxis with sporozoites delivered by mosquito bites) at the Radboud University Medical Center (Nijmegen, The Netherlands) 47 . Written informed consent was obtained from these individuals including for their samples to be stored and used in additional immunological experiments. Pre-immunization samples collected before the CPS-immunization were used for analysis of malaria antigen-specific antibody levels.
In vitro sporozoite gliding motility assay Prior to in vitro sporozoite assays, plasma aliquots were heat-inactivated for 30 minutes at 56°C, centrifuged at 13,000 rpm for 5 minutes at room temperature and kept at 4°C. Flat-bottom optical-bottom 96-well plates with cover glass base were incubated overnight at 4°C with an anti-CSP monoclonal antibody (produced at Radboudumc Nijmegen, Netherlands 48 ) 3SP2; 5 μg/ml in PBS). Following incubation, wells were washed twice with 150 μl/well PBS, blocked for 20 minutes at room temperature with 100 μl/well Leibovitz medium (Lonza) supplemented with 1% penicillin-streptomycin (GIBCO), 1% L-glutamine (GIBCO) and 10% heat-inactivated FBS(GIBCO). P. falciparum NF54 sporozoites (100 μl) were pre-incubated with citrated samples (70 μl; 40% final concentration) for 30 minutes at room temperature and added to each well in triplicate (50 μl/well) at a concentration of 10,000 sporozoites/well. Sporozoites were allowed to glide for 90 minutes at 37°C, 98% humidity, 93% N 2 , 4% CO 2 and 3% O 2 . Wells were washed thrice with 100 μl/well PBS and gliding trails were fixed for 15 minutes at room temperature with 4% paraformaldehyde (Affymetrix). Following fixation, wells were washed thrice with 100 μl/well PBS and blocked with 150 μl/well 10% FBS/PBS for 20 minutes at room temperature. Subsequently, gliding trails were stained for 1 hour at room temperature with 50 μl/well 5 μg/ml biotinylated anti-CSP monoclonal antibody (anti-CSP 3SP2 antibodies were produced at Radboudumc, Nijmegen, the Netherlands 48 ), followed by a wash step (thrice with PBS) and a 1 hour incubation at room temperature with 50μl/well 10 μg/ml streptavidin-Alexa Fluor-594 (Life Technologies) diluted in 10% FBS in PBS. Subsequently, wells were washed thrice with 100 μl/well PBS and stored in 150 μl/well PBS at 4°C in the dark until analysis. Gliding trails were imaged automatically with the BioTek Cytation cell imager (25 images per well at 200x magnification) and images were analysed automatically by FIJI software (under ImageJ version 2.0.0-rc-68/1.52h) with Otsu's thresholding 28 . Results were plotted in GraphPad Prism version 5.03.The number of pixels present on a stitched image made from 25 individual pictures taken per well is a measure of the amount of shed CSP in that particular well and therefore, differences in the number of pixels can be interpreted as differences in sporozoite gliding trail surface 28 .
In vitro sporozoite infectivity assay of a human hepatoma cell line Neutralization of P. falciparum sporozoite hepatocyte invasion by naturally acquired antibodies was assessed in a flow-cytometrybased in vitro invasion assay, as previously described with small adaptations 46 . Briefly, freshly dissected P. falciparum NF54 sporozoites were added to heat-inactivated plasma samples (10% final concentration) from malaria-naive or malaria-exposed individuals and pre-incubated for 30 minutes at 4°C. Subsequently, the sporozoite-plasma mixtures (5.10 4 sporozoites in the presence of 10% plasma) were added to HC-04 hepatocytes in 96-well plates. Following 3 hours of incubation at 37°C in 5% CO2, invaded and intracellular sporozoites were stained with an Alexa Fluor 488-conjugated anti-CSP antibody. Flow cytometric analysis was performed with a Gallios (Beckman Coulter) flow cytometer and data were analysed with FlowJo software (version 10.0.8, Tree Star). The percentage of CSP-positive hepatocytes was first corrected for background reactivity by subtracting the background (uninfected HC-04 cells in the presence of 3SP2-Alexa Fluor-488 antibody). The percent invasion inhibition was expressed relative to control IgG.

Statistical analysis
For analysis of in vitro sporozoite data, comparisons between two (controls versus field samples) or multiple groups were performed using Mann-Whitney U-test and Kruskal-Wallis test followed by Dunn's test between two groups, respectively. The associations between immune responses and malaria infection risk were assessed using survival analysis methods. Log-rank test was used to compare times to infection incidence for individuals with different values of immune phenotypes. Cox survival models were fit to assess the effect of pre-erythrocytic immunity after adjustment for blood-stage immunity; the proportional hazards assumption was tested using Schoenfeld residuals. In these analyses, study participants were considered to have high or low responses (binary explanatory variables) based on the study population median (see Figure S5 in Extended data 14 ). The first scheduled weekly visit or intensive follow-up visit when parasitaemia of at least 0.1 parasites per μl was detected by 18S qPCR was considered the time of infection incidence. Using this criterion, the median 18S qPCR-based parasitaemia at infection detection was 2.2 parasites per μl and the interquartile range was 0.4 -80.4. This threshold of parasitaemia was chosen to minimise false-positive results. In the supplemental material, sensitivity analyses were included that used different cut-offs of 18S qPCR-based density to determine infection positivity. Stata 14 (StataCorp LP, Texas, USA) and GraphPad Prism software (version 5, GraphPad Software Inc., California, USA) were used for statistical analysis. P<0.05 was considered statistically significant.

Data availability Underlying data
The main dataset relating to the field study contains individual level data and identifying information on participants; as such, this dataset is stored under restricted access and not available through an open-access repository. Requests from researchers to access these data for pooled or meta-analysis should be addressed to the corresponding author (teun.bousema@radboudumc.nl). However, the dataset used in the survival analysis has been de-identified and is available from the Dryad repository, along with raw ELISA results and sporozoite gliding data. DOI: https://dx.doi.org/10.5061/dryad.n1m33qq 14 .
Data are available under the terms of the Creative Commons Zero "No rights reserved" data waiver (CC0 1.0 Public domain dedication).

Extended data
The results of sensitivity analyses (see Results section) and Figure S1- Figure S5 are available from the Dryad repository.  Comparisons between multiple groups were tested by Kruskal Wallis test.   References equivalent to half of the lower limit of detection. In (D), the results of gliding assays are presented: the left plot presents the distribution of log 10 -transformed sporozoite gliding surface; the right plot shows gliding inhibition (y-axis) for each study participant (different bars; x-axis). The median gliding inhibition (59.6%) was used to define high and low inhibition in the survival analysis. The study by Barry and colleagues is important to shed more light into the efficiency of naturally-acquired sporozoite immunity to interfere with the parasite development at the pre-erythrocytic stages. Functional antibodies are assessed in plasma from exposed children for their capacity to in vitro P. falciparum interfere with sporozoite gliding and hepatocyte invasion which are required for progress with their life cycle to liver and blood stages, and these are related to titers. The outcome measured is blood stage parasitemia (delay of appearance of subpatent infections) as no direct measurement of liver parasitemia can be done in humans.
Some changes are required to clarify some aspects and make the work more scientifically sound:

ABSTRACT
In the Abstract background, specify if the statement "it is unclear whether immunity that affects the establishment of infections develops following continuous natural exposure" refers to children or in general, and if it refers to naturally acquired immunity as typically perceived (mostly against blood stages) or here if specific for pre-erythrocytic immunity. This also applies to the main Introduction. Although sterilizing immunity does not develop, parasite densities tend to decrease with age.
In the Abstract Methods it would be relevant to specify the intensity of malaria transmission and the specific area of study (location) within Burkina Faso, to know how repeatedly are they exposed to Pf. Also clarify if it is qPCR or nested PCR (also in results, it is a bit confusing when each is used). Liver stage antigen: which one? Specify also that the functional sporozoite assays are . in vitro In the results, mention if CSP antibody titres correlated with hepatocyte invasion and if the latter correlated to delayed time to qPCR +ve.

INTRODUCTION
See comments in the Abstract section.

METHODS
Can it be clarified if children did not receive antimalarial treatment, if during the weekly monitoring visits the nested PCR (done 48h after bleeding) was Pf positive but they had no symptoms (at least for 35 days)? If so, this was accepted by ethics committees easily? Were the families for 35 days)? If so, this was accepted by ethics committees easily? Were the families communicated of a positive PCR reaction in absence of symptoms? Indicate where the sporozoites were generated. Indicate if and how the patent parasitemias were also monitored and taken into account in the analysis.

RESULTS
When presenting data on median invasion inhibition, please comment on whether these % are high, moderate or low, and give some references.
The following statement does not look convincing by looking at the Fig S3A scatter plot despite the P=0.02 and Spearman's ρ=0.60 that appears driven by exteme values: "There was a positive correlation between gliding and invasion inhibition suggesting that gliding inhibition by in vitro naturally acquired antibodies might serve as a good surrogate for hepatocyte invasion in vitro inhibition." The correlation in Fig S3B is clearer. Fig 3A: interesting to see that there is not always overlap of the three metrics, expected to be discussed. Why was hepatocyte invasion inhibition not included? Explain the rationale for categorizing in high and low responders. Was the analysis also done with continuous values but did not yield significant insights? The joint analysis of the antibody levels to CSP and blood stage lysate and the functional pre-erythrocytic responses is helpful, but shouldn't there be a model with the 3-4 responses together? If not, explain why.

DISCUSSION
It is not convincing to say that children 5-10 years in a highly endemic area of Burkina Faso have limited blood stage immunity. They will have quite high antibody responses to blood stage antigens already. Better to reformulate. It is surprising the lack of difference in antibody levels for CSP between these children and naive volunteers. Even if CSP is not as highly immunogenic as many blood stage antigens, the responses are usually significant and detectable, even in children. Could there be a problem with the capture antigen? Seasonality by itself may not totally explain this, or it would imply they are very short lived. The potential inhibition by blood stage has been proposed but it would also affect LSA1. It is not well discussed how the affection of gliding and invasion before liver stage is detected as delay in positive qPCR in the blood, how do authors think this works As mentioned, the study is valuable due to the lack of data on functional pre-erythrocytic antibody responses in exposed African children.

If applicable, is the statistical analysis and its interpretation appropriate? Yes
Are all the source data underlying the results available to ensure full reproducibility? Partly

Comment "The study by Barry and colleagues is important to shed more light into the efficiency of naturally-acquired sporozoite immunity to interfere with the parasite development at the pre-erythrocytic stages. Functional antibodies are assessed in vitro in plasma from P. falciparum exposed children for their capacity to interfere with sporozoite gliding and hepatocyte invasion which are required for progress with their life cycle to liver and blood stages, and these are related to titers. The outcome measured is blood stage parasitemia (delay of appearance of subpatent infections) as no direct measurement of liver parasitemia can be done in humans.
Some changes are required to clarify some aspects and make the work more scientifically sound:

Answer
In the sentence mentioned by the reviewer, we refer to the fact that there is little or no evidence that naturally acquired immunity against liver stage parasites blocks initiation of blood-stage infection after parasite inoculation by mosquitoes, including in adults. The term "establishment of Anopheles infections" in that sentence is equivalent to "release of merozoites, from hepatocytes, in the blood".
We now modified two similar sentences in the and section:

Abstract Introduction
We now modified two similar sentences in the and section: Abstract Introduction "Individuals living in malaria-endemic regions develop naturally acquired immunity against severe malarial disease, but it is unclear whether immunity against pre-erythrocytic stages that blocks develops following continuous natural initiation of blood-stage infection after parasite inoculation exposure." (Abstract) "Individuals living in malaria-endemic regions can develop naturally acquired immunity against severe malaria disease and death, but it is unclear whether immunity that reduces, entirely or develops following partially, the probability of blood-stage infection after parasite inoculation natural exposure." (Introduction section)

In the results, mention if CSP antibody titres correlated with hepatocyte invasion and if the latter correlated to delayed time to qPCR +ve."
Answer We now mention in the the study area and clarify which liver stage antigen was studied: Abstract "We cleared schoolchildren living in an area ( The hepatocyte invasion inhibition phenotype was not included in the survival analysis as only 16 participants had samples tested with this assay.

Answer
We addressed these comments in our answer to the first comment of Reviewer 3.

Answer
In this study, we monitored children with weekly samples to identify new infections during the transmission season. We used nested PCR to detect and early, i.e. low density, infections characterize for the first time the evolution of naturally acquired acute infections in partially immune children. Children with parasites detected by nested PCR were closely monitored for development of symptoms and treated with first line therapy when symptoms developed. Burkina Faso's national malaria guidelines did not include treatment of asymptomatic parasite carriage, consequently initial detection of parasites, often at low levels, by nested PCR was not sufficient to prompt administration of antimalarials. It should be noted, and we have clarified this in the Methods section, that children received a long-lasting net at the start of the study and were guaranteed access to care 24 hours per day. Moreover, children were clinically examined on a daily basis in the first week of detection of infection. In this manner, the study adhered to the national guidelines but provided improved (access to) care for all study participants.
Parents and guardians were informed that study participation could involve intensive follow-up if parasites were detected in their child's blood samples and were oriented to contact the study staff if their child developed symptoms. The study was approved by the ethics committee of the London School of Hygiene and Tropical Medicine (reference number 9008) and the Ministry of Health in Burkina Faso (reference number 2015-3-033). Of note, previous studies with similar design were performed that described the development of malaria symptoms in children after detection of patent infections (e.g. M. A. Missinou, B. Lell, P. G. Kremsner, Uncommon asymptomatic . Clin Infect Dis 2003). In the study by Plasmodium falciparum infections in Gabonese children Missinou and colleagues, children were followed every second week for detection of infection by blood smear, and after malaria infection was diagnosed, children were clinically monitored with daily blood smears. In large cohort studies conducted in the context of the International Centers of Excellence for Malaria Research (ICEMR) program in Uganda, participants of all age groups (including much younger and with less prior malaria exposure compared to the current study) are followed every 4 weeks with asymptomatic infections sometimes remaining untreated for more than a year (Moss et al.

Malaria Epidemiology and Control Within the International Centers of
Am J Trop Med Hyg. 2015). An important difference between Excellence for Malaria Research. some of these studies and ours is that we used molecular methods to be able to detect infections from the early, sub-patent, phases and we also characterized malaria infectiousness over time, which will be described in a separate manuscript ( ). In conclusion, we appreciate the in preparation which will be described in a separate manuscript ( ). In conclusion, we appreciate the in preparation discussion about the necessity to treat asymptomatic infections and believe this is an important discussion to have, but the current study followed national guidelines and is by no means unique in following asymptomatic infections without treatment but with excellent care.
Comment "Indicate where the sporozoites were generated."

Answer
As described in the section, mosquitoes were infected by standard Methods Anopheles stephensi membrane feeding assays on NF54 gametocyte cultures. These experimental P. falciparum mosquito infections were performed in the insectarium of the Department of Medical Microbiology of the Radboud University Nijmegen Medical Centre.

Answer
The analysis presented in this manuscript focuses on the influence of immune factors on infection establishment, i.e. appearance of falciparum parasites in the blood. Consequently, all different comparisons presented relate to the first visit of each participant when parasites were detected by sensitive molecular methods. Blood smears were also performed for all study visits, before and after infection detection, but results were not immediately available. The relationship between blood smear results and parasite and gametocyte levels quantified by molecular methods after infection detection will be described in detail in a separate manuscript ( ). in preparation

Answer
The assessment of whether hepatocyte invasion inhibition was high, moderate or low in vitro requires comparisons of inhibition percentages between different studies, which might be less valid than within-study comparisons between samples from different participants tested with the same protocol. Furthermore, without an established quantitative relationship between the results of this assay and infection incidence, semi-quantitative statements of this type are difficult to in vivo interpret. Some studies that performed the same assay were cited: in vitro Behet et al.

The complement system contributes to functional antibody-mediated responses
. Infect Immun. 2018 induced by immunization with Plasmodium falciparum malaria sporozoites

Sattabongkot et al. Establishment of a human hepatocyte line that supports in vitro development of
Am J the exo-erythrocytic stages of the malaria parasites Plasmodium falciparum and P. vivax. Fig S3A scatter plot  despite the P=0.02 and Spearman's ρ=0.60

that appears driven by extreme values: "There was a positive correlation between gliding and invasion inhibition suggesting that in vitro gliding inhibition by naturally acquired antibodies might serve as a good surrogate for in vitro hepatocyte invasion inhibition." The correlation in Fig S3B is clearer."
Answer In our analysis of the correlation between hepatocyte invasion inhibition and in vitro in vitro sporozoite gliding inhibition, we used the Spearman's rank correlation method, which assessed the correlation between the ranking of the values in these two phenotypes, rather than the values themselves. As additional descriptive information, we added the following sentence: Figure S3A in Extended data14, P=0.005, Spearman's ρ=0.67), suggesting that in vitro gliding inhibition by naturally acquired " antibodies might serve as a good surrogate for in vitro hepatocyte invasion inhibition.

Answer
The hepatocyte invasion assay was only performed in a subset of the study population (16 in vitro children) and for this reason was not included in this figure and in the survival analysis. The invasion experiments are highly laborious and resource intensive. We used these as supportive experiments in our study, to confirm the relevance of gliding assays, but invasion assays did not form the core of our analysis. We therefore feel that our approach to perform these experiments on a selective number of samples is justified.

Answer
The survival analysis was only performed with binary variables. The rationale for categorizing the study population in high and low responders, based on the median, was to facilitate interpretation of the results and avoid assumptions about linearity of effect.

Answer
Since CSP-specific antibody levels are highly correlated with sporozoite gliding inhibition (see section), we decided not to include these two variables in the same model. Additionally, Results the limited sample size also influences the number of predictors that can be reasonably included in our multivariate analyses.

Answer
We agree that children in highly endemic areas will have considerable immunity by the age of 5 years. However, our statement compared immunity in children relative to adults: "We selected children aged 5-10, who have limited blood stage immunity (a potential confounder when studying .". We have now modified the pre-erythrocytic immunity) compared to semi-immune adults, … sentence to avoid misinterpretation: "We selected children aged 5-10, who have (a potential lower blood stage immune responses confounder when studying pre-erythrocytic immunity) compared to semi-immune adults, and who allow repeated blood sampling compared to toddlers."

Answer
Thank you for these suggestions. The reviewer suggested that suppression of immune responses by blood-stage parasites would, theoretically, affect responses to both antigens, LSA-1 and CSP. We agree and have now modified the section accordingly: Discussion "These low titres could be related to the timing of the study recruitment (prior to the transmission

"These low titres could be related to the timing of the study recruitment (prior to the transmission season after ~7 months of very low malaria exposure), which would imply that responses to CSP influenced by repeated blood-stage are short lived. The antibody levels could also have been infection in our cohort, which may have suppressed immune responses against the s pre-erythrocytic stages ." , although, in theory, responses to both LSA-1 and CSP would be affected
The reviewer also suggested that there might have been a problem with the antigen used in the CSP assay. The significant correlations between anti-CSP levels and functional phenotypes (i.e. sporozoite gliding inhibition or hepatocyte invasion inhibition) suggest that this is not the case (or that even if there was an issue, the interpretation of the relative values is still valid), and this is supported by the fact that a subset of children had particularly high antibody responses to CSP (Figure 2A).

Answer
We observed an association between sporozoite gliding inhibition and time to parasite positivity by qPCR. This could either reflect an ability of these immune responses to completely neutralize some sporozoite inoculums, leading to a delay in patency because this population would have no merozoites being released into the bloodstream from a first sporozoite inoculum. Alternatively, the number of sporozoites that reach hepatocytes can be reduced by immune responses targeting sporozoites and thus, the number of merozoites released upon completion of the hepatic phase of parasite development. This would result in a lower starting point of blood stage infection and plausibly a longer time to reach detectability of infections by PCR. Lastly, and we believe this is well explained in the revised manuscript, it is possible that the number of released merozoites is unaltered in the population with high sporozoite immunity, but the concurrent higher blood stage immunity dampens parasite growth and results in a longer time to reach detectability of infections by PCR.
No competing interests were disclosed. Competing Interests: 1.

6.
This ms measures antibody responses in Burkinabe children after drug clearance and subsequent infections, then attempts to explain delays in time to infection in terms of Plasmodium falciparum antibodies against circumsporozoite protein (CSP) and whole parasites. The authors demonstrate that, in individuals who show greatest delays in time to parasitemia, there were stronger functional antibody responses against sporozoites which reduced gliding motility and liver cell invasion . The major in vitro question addressed in this ms is whether these anti-sporozoite responses are affecting time to parasitemia in these children. From the data presented, the answer is "maybe". The problem the authors face is that for almost every infection that stimulates anti-sporozoite antibodies, that infection also stimulates blood stage antibodies as the infection progresses. It is these latter that show the best negative correlation in reduction in time to parasitemia, and while separating out the functional anti-sporozoite effects is informative it is not definitive. This is illustrated in Figure 3A, where of the 29 participants with anti-sporozoite functional activities in one or both of the assays, 17 also have higher than median anti-blood stage responses. I do not see a way around this at the participant level and the authors clearly acknowledge this difficulty in their discussion; one is trying to see subtle effects on infection (in this case time to infection no children we uninfected during the study) using markers that autocorrelated and cannot therefore be separated from one another in terms of cause and effect. One additional complication is not considered as far as I can tell. In order to have become infected, these children must have been bitten by mosquitoes. The mosquito bites induce antibody responses. The sporozoites used in the assays for gliding motility and hepatocyte invasion were from crude homogenates that contain salivary gland material. It is possible (though I admit not too likely) that there was some functional interaction between these two. Could antibodies against ( ) salivary glands Anopheles stephensi be present in these sera (of children bitten by African anophelines), and if so, could they have affected the functional antibody assays? in vitro Nevertheless, the ms provides important information concerning anti-sporozoite functional antibodies, even if a clinical outcome (delayed time to parasitemia) cannot be ascribed to them. :

Other points
Given the comments above and the stronger correlation of blood stage antibodies with delay to parasitemia, the authors should consider changing the abstract (and perhaps the title) to include this information. It is buried in the paper but should be more explicit.
At the start of the introduction, the authors highlight the protective efficacies of RTS,S and PfSPZ Vaccine. However, they should also note that the outcomes, efficacies are different, RTS,S being time to clinical malaria, PfSPZ Vaccine being time to parasitemia. This has direct relevance to the ms. Also in the introduction, the authors suggest that there is a discrepancy between protection with sporozoites and naturally induced protection not preventing infection. However, they do not consider the obvious reason for this -dose. Mosquitoes deliver a few (median 50 sporozoites per bite) now and again over a transmission season, while the sporozoite vaccines are delivering thousands of sporozoites per dose and defined intervals. The authors say that there are sporozoite-specific IgG and IgM antibodies in the malaria exposed children, but such specificity has not been demonstrated, only reactivity. The term should be changed.
What do the authors mean by the title "Evidence of natural risk-modifying pre-erythrocytic immunity"? The authors note an important variable that could not be accounted for in their studies, namely variability in exposure. One of the authors has worked on antibody responses to mosquito saliva as a marker for exposure risk: would it not be possible to measure antibody responses against mosquito bites here? :

Minor points
Introduction, paragraph 1. The sentence ending " but it is unclear whether immunity that prevents Introduction, paragraph 1. The sentence ending " but it is unclear whether immunity that prevents or reduces infection incidence develops following natural exposure". This is a misuse of the term incidence -prevention of infection in one person will reduce incidence in the population. I think the authors mean "that prevents infection or reduces infection intensity in an individual….". In the same paragraph, change the word 'suggests' (…suggests that sterilizing immunity…" to demonstrates; I think the evidence is pretty clear). Results, paragraph 2. Should read Parasitological and immunological data from the remaining 51 children followed intensively were…" What does "data from this child were censored after the onset of symptoms" mean? Could have any one of many interpretations. In the results section "Naturally acquired antibodies in children neutralize in vitro sporozoite gliding motility" the authors should change the term "considerably larger' for something more scientific (significantly greater"?). Authors should give sample sizes (n values) used in the curves for figure 3B,C. In the discussion, the authors say "To our knowledge, this is one of the first studies to show that there are functional antibodies against pre-erythrocytic malaria stages in malaria-exposed children." Shouldn't the other studies be referenced here? The authors should check the axis labeling of figures: for example (% infected) ought to be "Percent children infected with P. falciparum", in Figure S1 the abbreviations/units are not explained, and so on.

If applicable, is the statistical analysis and its interpretation appropriate? Yes
Are all the source data underlying the results available to ensure full reproducibility? Yes

Are the conclusions drawn adequately supported by the results? Partly
Sanaria has studies running in both Burkina Faso and in Nijmegen testing our Competing Interests: sporozoite vaccines. Sanaria's interest is in the infectivity sporozoites and protection against sporozoite infections. This ms is directly relevant to our vaccines in this regard, but complimentary rather than conflicting. Like the authors, we are trying to both protect against sporozoite infections and understand the background factor affecting infectivity.  Figure 3A,

Nevertheless, the ms provides important information concerning anti-sporozoite functional antibodies, even if a clinical outcome (delayed time to parasitemia) cannot be ascribed to them."
Answer Thank you. We agree with this assessment of the difficulty in separating pre-erythrocytic-stageand blood-stage-specific immune responses. We have now included in the section a Discussion statement that emphasizes this point: "We observed that high responses in the asexual stage lysate assay with unknown functionality were also associated with longer time to infection. We thus cannot rule out a supportive role for asexual antibody responses in the observed associations. Indeed, as pointed out by one of the were also associated with longer time to infection. We thus cannot rule out a supportive role for asexual antibody responses in the observed associations. Indeed, as pointed out by one of the reviewers of the first version of this manuscript, 17/29 (58.6%) children with high anti-sporozoite responses in either CSP-specific or sporozoite gliding assays had above-median asexual lysate responses versus 8/22 (36.4%) children with below-median anti-sporozoite activities." The reviewer also suggested that the functional activity of the plasma samples from study participants could be partially related to targeting antigens of salivary gland material rather than targeting sporozoites antigens. This is a very novel idea and an area of research that is both exciting and relatively new. There are indeed some indications that saliva responses may influence both the viability of sporozoites ( In conclusion, there is no convincing evidence that salivary gland responses could explain our current findings but it is a theoretical possibility that we cannot rule out completely. We have thus added the following underlined sentence to the paragraph where we describe some of the limitations of this study: "Another limitation of our study is that we were only able to determine humoral responses. Cellular responses to pre-erythrocytic stages have been implicated in malaria protection in multiple studies. It is conceivable that by quantifying both antibody and cellular responses we would be able to better define natural immunological phenotypes associated with differential malaria risk. Immunity to mosquito saliva might have also influenced sporozoite invasion although the magnitude of such an effect compared to the established impact of anti-sporozoite responses is currently unknown and it is currently unknown whether this would require responses to antigens that are conserved between An. stephensi mosquitoes used for sporozoite production and vectors that study " participants were naturally exposed to.

Answer Answer
To address this comment and comments from Reviewer 1, we have now modified the : Abstract "Survival analysis indicated longer time to infection in individuals displaying higher-than-median sporozoite gliding inhibition activity (P=0.01), although this association became non-significant . after adjustment for blood-stage immunity (P = 0.06) "In summary, functional antibodies against the pre-erythrocytic stages of malaria infection are acquired in children who are repeatedly exposed to Plasmodium parasites. This immune response does not prevent them from becoming infected during a malaria transmission season, but might delay the appearance of blood stage parasitaemia. Our approach could not fully separate the effects of pre-erythrocytic-specific and blood-stage-specific antibody-mediated immune responses in vivo; epidemiological studies powered and designed to address this important question should ." become a research priority pre-erythrocytic stages, which could be possibly linked to the lower number of sporozoites in . Individuals living in natural parasite inoculations or to the frequency of host-vector contacts malaria-endemic regions can develop naturally acquired immunity against severe malaria disease and death, but it is unclear whether immunity that reduces, entirely or partially, the probability of develops following natural exposure." blood-stage infection after parasite inoculation Comment "The authors say that there are sporozoite-specific IgG and IgM antibodies in the malaria exposed children, but such specificity has not been demonstrated, only reactivity. T44he term should be changed."

Answer
We agree with the reviewer that the specificity of these antibodies has not been demonstrated, and have now modified the title of the sub-section in the manuscript: " " Sporozoite-specific IgG and IgM antibodies in malaria-exposed children to: " "

Naturally acquired IgG and IgM antibodies targeting sporozoites
We also modified the following sentence: "In addition to antigen-specific assays, IgG and IgM antibodies recognizing Plasmodium sporozoites were quantified using fluorescently labelled anti-IgG and anti-IgM antibodies by flow cytometric analysis. hole sporozoite were significantly higher Levels of antibodies targeting w s in malaria-naive donors ( Figure 2D, E, P=0.01 and 0.04 for IgG the study participants compared to and IgM, respectively). Strong correlations between IgG and IgM antibodies targeting whole ( Figure S1A in Extended data 14 , P=0.0005, Spearman's ρ=0.75), and between sporozoites CSP-IgG antibody levels and whole sporozoite IgG antibodies ( Figure S1B in Extended data 14 , Spearman's ρ=0.83, P<0.0001) were also observed." We also replaced the expression " " with more accurate expressions in other sporozoite specific paragraphs of the manuscript.

Comment "What do the authors mean by the title "Evidence of natural risk-modifying pre-erythrocytic immunity"?"
Answer This sub-section title refers to the observation that functional immune responses against pre-erythrocytic stages were associated with delay in infection appearance. As discussed and addressed in answers to other comments, this evidence is based on the univariate survival analysis. "Risk" here is being used loosely, and not in the epidemiological sense of probability of an event during a specified time period.

Comment "The authors note an important variable that could not be accounted for in their studies, namely variability in exposure. One of the authors has worked on antibody responses to mosquito saliva as a marker for exposure risk: would it not be possible to measure antibody responses against mosquito bites here?"
Answer Although quantification of antibody responses to mosquito saliva antigens would provide some information on exposure to mosquitoes generally, as shown in a previous study in the Anopheles same area (Guelbéogo et al, Variation in natural exposure to anopheles mosquitoes and its effects ), only a small proportion of blood-fed mosquitoes are parasite on malaria transmission. eLife 2018 carriers, and the prevalence of infection in malaria vectors varies from household to household. For these reasons, we believe quantification of immunity against mosquito saliva antigens alone would not adequately adjust for exposure. Moreover, saliva responses are complicated to use in areas of high mosquito exposure. There is considerable evidence for a level of immune tolerance that reduces responses to mosquito saliva proteins in older individuals with high mosquito exposure. This would complicate the interpretation of mosquito salivary gland responses considerably in our cohort experiencing high but heterogeneous mosquito exposure.

Answer
We modified the sentence below for clarification: "Individuals living in malaria-endemic regions can develop naturally acquired immunity against severe malaria disease and death, but it is unclear whether immunity that reduces, entirely or develops following partially, the probability of blood-stage infection after parasite inoculation natural exposure." We prefer to keep the word 'suggests' in that same paragraph as "high incidence of blood-stage re-infection after effective anti-malarial treatment in adults" is a strong but not conclusive evidence that sterilizing immunity does not develop.

Comment
"Results, paragraph 2. Should read Parasitological and immunological data from the remaining 51 children followed intensively were…" What does "data from this child were censored after the onset of symptoms" mean? Could have any one of many interpretations."

Answer
In that sentence, we meant that any visits and samples collected after the development of symptoms were disregarded and the child was classified as uninfected during the entire valid follow-up period.
We replaced the first sentence of that paragraph with the following sentence: "Parasitological and immunological data from the remaining 51 children followed intensively were analysed."

Answer
The modified sentence is: "To characterize the effect of human plasma on sporozoite gliding motility that is independent of naturally acquired immunity, sporozoites were incubated in the presence of plasma from malaria-naive individuals (n=5) and showed an average gliding trail surface of 1,427(95% CI 646.6-2,207) pixels, than the gliding trail surface of sporozoites incubated with significantly greater 30 μg/ml of a monoclonal anti-CSP antibody, our positive control,141 (95% CI 71.9-225.4) pixels."

Answer
We have now included this information in the figure legend.

Answer
We modified the sentence: "To our knowledge, this is the first study to show that there are functional antibodies against pre-erythrocytic malaria stages in malaria-exposed children." Figure S1 the abbreviations/units are not explained, and so on."

Answer
We have now modified the legends, rather than axes' labels, of the following figures:  , which permits unrestricted use, distribution, and reproduction in any medium, provided the original Attribution Licence work is properly cited.

Silvia Portugal
Malaria Infection Biology and Immunity Section, Laboratory of Immunogenetics, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Rockville, MD, USA In this manuscript Barry, Behet and colleagues address an important question aiming to possibly appoint an effective role in protection from clinical malaria to liver-stage immunity acquired naturally in malaria endemic areas. I believe the work presented is a significant contribution to our understanding of naturally acquired immunity to pre-erythrocytic stages of and I make a few comments and P. falciparum suggestions that may improve the current version of the manuscript. The data shows that ABs developed during past malaria cases can reduce sporozoite motility and hepatocyte invasion suggesting that ABs acquired during natural infections can reduce new in vitro liver-stage infections. However, the contribution of a possible similar effect is, in my view, less clear in vivo from the data presented, and more caution may be needed to discuss the results. All the children followed throughout the study became parasite positive by qPCR, indicating that even children with strong inhibiting ABs were unable to block liver-stage infection efficiently. Furthermore, high response to asexual inhibiting ABs were unable to block liver-stage infection efficiently. Furthermore, high response to asexual might be confounding the analysis. I would add to the manuscript (or supplemental) figures stage lysate the association of CSP ABs with asexual stage lysate ABs, and also the association of whole SPZ ABs with asexual stage lysate ABs, to give the reader an idea of how close these parameters are.
I also suggest including, if that has not been done already, the in the multivariate Reported bed net use analyses, as this could also be a factor increasing time to PCR positive and clinical malaria.
The authors cite the study by Tran et al. where it was shown that, in Mali, time to PCR positive was independent of age, while time to clinical malaria increased with age, and where as stated it was concluded that there no or very limited evidence for an age dependent acquisition of immunity protecting Similarly, in the present manuscript, functional data of higher humoral response from infection.
in vitro against pre-erythrocytic stages does not (independently of blood-stage immunity) protect from infection. So, I would rephrase the last sentence in first paragraph of the discussion to add a bit more caution in interpreting what may be causing partial protection.
At the end of section the authors should, in Evidence of natural risk-modifying pre-erythrocytic immunity my view, clearly state that does not independently associate in a statistically High gliding inhibition activity significant way, with on the multivariate analyses where blood stage protection against falciparum infection immunity was included; the P value is above 0.05 (0.055) and the CI includes 1, making the relative risk not statistically significant.
I believe the manuscript could be improved by presenting the quantitative analysis of the qPCR upon 18s first parasite detection and determine if there is a negative association with the inhibitory capacity of the individuals' ABs. It would also be very interesting to question if time from first PCR positive to time of presentation of symptoms is different between poor and strong inhibitors. If the data in vitro in vitro showing gliding inhibition and reduced hepatocyte invasion are significant , one would expect a in vivo lower inoculum in the liver and thus a lower parasitaemia on the first PCR positive time-point. And then potentially a slower progression to clinical malaria. I believe with the data generated in this manuscript these analyses could be done, and would enrich the story.
It is not totally clear to me how individuals were selected for the flow cytometry assays. Survival, gliding inhibition, CSP, LSA1 and asexual lysate ELISAs were performed for the 51 participants, but flow cytometry data presented in fig2 D and E was obtained from 17 Burkinabes only; how were those selected and what is their time to PCR+ in the survival analysis. If they are the 8 poor and 8 strong inhibitors as defined by their gliding inhibition it should be stated in the methods (seems to be so, given supFig3, but there is one extra?). I also suggest to pinpoint these 8 poor and 8 strong inhibitors in fig1 so that the reader would be informed of their time to PCR+ and time to malaria symptoms.  On page 3 below table 1 there is mention to which may be a mistake. field PCR The data from the gliding inhibition by LSA IgG seems to be not shown. I think it should be clarified in vitro in the text that that is indeed the case. Likewise, if the LSA-1-specific IgG antibodies correlation with sporozoite invasion inhibition is data not shown I would clearly state it in the text. In figS2A I would specify that is IgG in the figure x axis and use the label CSP IgG titer instead of CSP antibody titer. Figure S3D is called before Figure S3C, I would call figures in ascending and alphabetical order instead. We would like to thank the reviewers for their insightful comments and suggestions. The modifications we made to address their comments improved the quality of our manuscript and clarified the interpretation of our results. modifications we made to address their comments improved the quality of our manuscript and clarified the interpretation of our results.

Reviewer 1
Comment "In this manuscript Barry, Behet and colleagues address an important question aiming to possibly appoint an effective role in protection from clinical malaria to liver-stage immunity acquired naturally in malaria endemic areas. I believe the work presented is a significant contribution to our understanding of naturally acquired immunity to pre-erythrocytic stages of P. falciparum and I make a few comments and suggestions that may improve the current version of the manuscript.

Answer
In response to this comment and to a comment from Reviewer 2, we modified the to Abstract mention that our study did not fully address the correlation between liver-stage-and blood-stage-specific immunities and that additional studies are necessary.
"Survival analysis indicated longer time to infection in individuals displaying higher-than-median sporozoite gliding inhibition activity (P=0.01), although this association became non-significant ." after adjustment for blood-stage immunity (P = 0.06) "In summary, functional antibodies against the pre-erythrocytic stages of malaria infection are acquired in children who are repeatedly exposed to Plasmodium parasites. This immune response does not prevent them from becoming infected during a malaria transmission season, but might delay the appearance of blood stage parasitaemia. Our approach could not fully separate the effects of pre-erythrocytic-specific and blood-stage-specific antibody-mediated immune responses in vivo; epidemiological studies powered and designed to address this important question should ." become a research priority We have also modified the first paragraph of the section: Discussion "Our epidemiological data together with the in vitro data provide evidence that there is partially effective immunity that influences individual level infection incidence pre-erythrocytic , although we cannot exclude the contribution of immunity against asexual blood-stage antigens to this ." observation Additionally, we agree that the inclusion of a figure showing the correlation between liver-stage immune responses and asexual stage lysate responses would be informative for the readers. We included a scatter plot of CSP-specific immune responses versus asexual stage lysate responses ( Figure S4). " Figure S4. CSP antibody levels versus immune responses against asexual stage antigens. In this figure, the x-axis shows CSP antibody levels; the y-axis represents antibody responses against asexual stage parasites lysate. Both axes are in log-scale; one child with undetectable CSP " response is not included in this graph.
We also modified the legend of Figure 3 to refer to this supplementary figure: "To assess the effect of immune responses on infection risk, children were classified based on whether the results of their assays were higher or lower than the study population median (see Results): in ( A), vertically aligned cells represent the same participant, and orange cells indicate that assay results are higher than the median (see also Figure S4, that shows a scatter plot of CSP " . antibody levels and immunity against asexual lysate)

Answer
The analysis in this manuscript concerns time to infection detection, regardless of whether symptoms are present or not. In this cohort, reported bed net use at enrollment had no effect on time to infection (hazard ratio 1.16, 95% CI [0.66 -2.03], P = 0.61). In analyses that included reported bed net use at enrollment and either CSP immune responses or sporozoite gliding inhibition phenotype as binary variables, neither the hazard ratio point estimates nor the confidence intervals changed significantly (data not shown). Of note, only data on bed net reported use were collected and it is evident that this is an imprecise indicator of personal protection.

Answer
Our multivariate analyses showed that the effects of functional phenotypes became non-significant after adjustment for blood stage immunity. However, we disagree with the reviewer that we can after adjustment for blood stage immunity. However, we disagree with the reviewer that we can conclude the opposite, i.e. that "higher humoral response against pre-erythrocytic stages does not (independently of blood-stage immunity) protect from infection". The study was not designed, and might not have been powered, to assess the separate effects of the correlated pre-erythrocytic-stage and blood-stage immunities on malaria infection establishment. A study with a larger sample size, where a sufficient number of individuals would have relatively high liver-stage immunity and relatively low asexual responses, and vice-versa, would be more suitable to address this question.
We have modified the section and the to mention that our approach did not Discussion Abstract fully separate the effects of these two types of immunity (see response to the first comment).

At the end of section Evidence of natural risk-modifying pre-erythrocytic immunity the authors should, in my view, clearly state that High gliding inhibition activity does not independently associate in a statistically significant way, with protection against falciparum infection on the multivariate analyses where blood stage immunity was included; the P value is above 0.05 (0.055) and the CI includes 1, making the relative risk not statistically significant.
Answer With the small sample size, we believe a very strong focus on the P-value threshold of 0.05 is potentially problematic. In general, the P-value and strict thresholds for interpreting significance are criticized by many Nature 2019). With our small Scientists rise up against statistical significance. sample size, we agree that all results should be interpreted with caution (also related to the limited sample size, but mostly due to the correlations between blood stage and pre-erythrocytic humoral immunity). The sentence in the section has now been modified to: Results

"
he relationship In a model that also included the results of the asexual stage lysate assay, t between anti-CSP responses and time to malaria infection was not statistically significant. Despite , inhibition of in vitro sporozoite gliding a clear trend for a protective effect the association between motility falciparum infection after adjustment and time to also did not reach statistical significance for blood-stage immunity (hazard ratio, 0.55; 95% CI, 0.29-1.01)."

Comment "I believe the manuscript could be improved by presenting the quantitative analysis of the 18s qPCR upon first parasite detection and determine if there is a negative association with the inhibitory capacity of the individuals' ABs. It would also be very interesting to question if time from first PCR positive to time of presentation of symptoms is different between poor and strong in vitro inhibitors. If the in vitro data showing gliding inhibition and reduced hepatocyte invasion are significant in vivo, one would expect a lower inoculum in the liver and thus a lower parasitaemia on the first PCR positive time-point. And then potentially a slower progression to clinical malaria. I believe with the data generated in this manuscript these analyses could be done, and would enrich the story."
all figures where a selection of the total study population of 51 individuals was used. These 16 individuals were selected based on highest gliding inhibition (84-99%) and lowest gliding inhibition (-34 -20%), as illustrated in supplementary figure 2B. To ensure consistency, we updated Figure  2, Figure S2 and Figure S3. This had no impact on any of the effects, significance levels changed marginally and have been updated. We have now included this information in the section: Methods "Recognition of whole sporozoites by naturally acquired IgG and IgM antibodies was determined by an in vitro flow-cytometry-based antibody opsonization assay that was presented in detail elsewhere. Due to limited plasma availability and available sporozoite numbers, 16 naturally exposed children with highest (8) and lowest (8) gliding activity were selected for sporozoite ." opsonization assays and invasion assays, allowing us to investigate potential correlations The median time to infection detection in the group of children classified as strong inhibitors of sporozoite gliding motility was 41 days (range 5 -64 [N = 7]; 1/8 child did not have parasites detected); poor gliding inhibitors (N = 8) had a median time to infection of 34 days (range 13 -48).

Answer
We cite the study by Stewart et al in the following sentence: "It has been demonstrated that Plasmodium parasites use the system of adhesion-based motility, gliding, to actively penetrate host cells ; and that the invasive ability of sporozoites is directly associated with their motility ." This is an accurate statement about that study, where the authors concluded that "sporozoite invasiveness is associated with sporozoite motility".

Answer
We modified Table 1 to 'Presence of parasites post-treatment'. Answer "Neutralization of P. falciparum sporozoite hepatocyte invasion by naturally acquired antibodies was assessed in a flow-cytometry-based in vitro invasion assay as previously described with small . adaptations Briefly, freshly dissected P. falciparum NF54 sporozoites were added to heat-inactivated plasma samples (10% final concentration) from malaria-naive or malaria-exposed individuals and pre-incubated for 30 minutes at 4 ˚C. Subsequently, the sporozoite-plasma mixtures (5.10 sporozoites in the presence of 10% plasma) were added to HC-04 hepatocytes in 96-well plates. Following 3 hours of incubation at 37 ˚C in 5% CO , invaded and intracellular Flow cytometric sporozoites were stained with an Alexa Fluor 488-conjugated anti-CSP antibody. analysis…."

Comment "On page 3 below table 1 there is mention to field PCR which may be a mistake."
Answer One child in the study did not develop symptoms nor had parasites detected by nested PCR performed in Ouagadougou. We have now modified the sentence to: "The median time from confirmation of the absence of parasites (i.e. 3 weeks after anti-malarial treatment) to infection detection by nested PCR or onset of symptoms was 28 days; one child who did not have parasites detected by PCR was not included in this calculation." nested

Comment "The data from the in vitro gliding inhibition by LSA IgG seems to be not shown. I think it should be clarified in the text that that is indeed the case. Likewise, if the LSA-1-specific IgG antibodies correlation with sporozoite invasion inhibition is data not shown I would clearly state it in the text."
Answer Although no correlation plot is shown for LSA-1 titers and gliding inhibition, the lack of association with gliding inhibition is included in the section: Results "In vitro gliding inhibition did not correlate with LSA-1 IgG antibody levels (P=0.11, Spearman's ρ=0.23)…" " porozoite invasion inhibition correlated with IgG ( Figure S3B in S targeting whole sporozoites Extended data14, P=0.004, Spearman's ρ=0.67) but not with whole sporozoite IgM antibody levels ( Figure S3C in Extended data 14 , P=0.13, Spearman's ρ=0.38). There was a correlation of ( Figure S3D in Extended data14, hepatocyte invasion inhibition with CSP-specific IgG antibodies P=0.002, Spearman's ρ=0.76), but not with LSA-1 specific IgG antibodies (P=0.08, Spearman's ." ρ=0.48) In response to the next comment and to comments by another reviewer, we have changed the order of the sentences and removed the word 'specific' from the section above and replaced it with 4 2 order of the sentences and removed the word 'specific' from the section above and replaced it with "targeting whole sporozoites".

Comment
"In figS2A I would specify that is IgG in the figure x axis and use the label CSP IgG titer instead of CSP antibody titer. Figure S3D is called before Figure S3C, I would call figures in ascending and alphabetical order instead."

Answer
We have now modified Figure S2 and rephrased the text so Figure S3C is mentioned before Figure  S3D (see answer to previous comment).
No competing interests were disclosed. Competing Interests: