Cytotoxic anti-circumsporozoite antibodies target malaria sporozoites in the host skin

The circumsporozoite protein (CSP) is the major surface protein of malaria sporozoites (SPZs), the motile and invasive parasite stage inoculated in the host skin by infected mosquitoes. Antibodies against the central CSP repeats of different plasmodial species are known to block SPZ infectivity1–5, but the precise mechanism by which these effectors operate is not completely understood. Here, using a rodent Plasmodium yoelii malaria model, we show that sterile protection mediated by anti-P. yoelii CSP humoral immunity depends on the parasite inoculation into the host skin, where antibodies inhibit motility and kill P. yoelii SPZs via a characteristic ‘dotty death’ phenotype. Passive transfer of an anti-repeat monoclonal antibody (mAb) recapitulates the skin inoculation-dependent protection, in a complement- and Fc receptor γ-independent manner. This purified mAb also decreases motility and, notably, induces the dotty death of P. yoelii SPZs in vitro. Cytotoxicity is species-transcendent since cognate anti-CSP repeat mAbs also kill Plasmodium berghei and Plasmodium falciparum SPZs. mAb cytotoxicity requires the actomyosin motor-dependent translocation and stripping of the protective CSP surface coat, rendering the parasite membrane susceptible to the SPZ pore-forming-like protein secreted to wound and traverse the host cell membrane6. The loss of SPZ fitness caused by anti-P. yoelii CSP repeat antibodies is thus a dynamic process initiated in the host skin where SPZs either stop moving7, or migrate and traverse cells to progress through the host tissues7–9 at the eventual expense of their own life. In a rodent malaria model, antibodies against the CSP protein that coats sporozoites lead to Plasmodium yoelii killing in the skin in a process that involves stripping off the CSP coat, rendering parasites susceptible to pore-forming-like proteins.

Intravascular SPZ inoculation shortcuts the skin phase of the infection, where a substantial part of the parasite population commits to invade either the lymphatics 8 or skin cells 30 , and thus cannot infect hepatocytes.Consequently, inoculation with the same number of SPZs by the iv route results in higher hepatic infection than skin inoculation.To discard a possible confounding effect of dose in protection, we adjusted the number of PySPZs inoculated iv or in the skin to achieve the same level of hepatic and blood infection (Supplementary Fig. 1a,c; 2,500 SPZ iv ~ 5,000 SPZ skin).Since no protection is observed after iv challenge with the adjusted dose (Supplementary Fig. 1b), these results indicate that in this model, protective effectors target extravascular PySPZs delivered in the skin of the immune host by mosquito bite or microinjection.To identify the nature of these protective effectors, mAb-mediated depletion of CD8 + , CD4 + and Ly6G + cells was performed in immunized hosts just prior to challenge.Despite the depletion of most targeted cells from the peripheral blood, none of these treatments revert the protection induced by the PyCSP immunization (Supplementary Fig. 2a-d).In contrast, mice lacking functional B-cells due to a mutation of the gene coding for the heavy chain-joining region (JHT -/-) completely lose PyCSP-induced sterile protection (Fig. 1e).In addition, sera transfer from immune hosts to naïve recipients protects passively immunized animals against a PySPZ skin challenge (Supplementary Fig. 2e).Altogether, these results imply a protective action of neutralizing antibodies in the skin of the sterile protected host.
To observe the protective activity of these antibodies in situ, we imaged GFP-expressing PySPZs in the skin of control (A) and immune hosts (A.CSP) using high-speed spinning disk confocal microscopy.In vivo imaging revealed two major differences between these two groups.The first is a prominent decrease in the speed of parasites microinjected in the dermis of immune hosts (Fig. 2ab).In contrast to what was previously described using irradiated PbSPZ immunization or passive transfer of anti-PbCSP mAb 7 , PySPZs were not completely immobilized in the dermis, possibly, reflecting differences in the experimental settings, or in the availability or binding capacities of cutaneous antibodies.The second difference is the striking death of PySPZs, substantiated by the sudden loss of parasite fluorescence in the dermis of immune animals (Fig. 2c).This death process produces a peculiar fluorescent dot in the posterior end of slow-moving parasites that can be short-lived or persist for several minutes (Fig. 2c, arrowheads, left or right panels, respectively).This dotty death phenotype is observed in ~50% of the population analysed during thirty minutes after skin microinjection (Fig. 2d).Despite the decrease in speed, and the high percentage of dotty death, PySPZs are still capable of invading blood vessels (Fig. 2e, intravasation).These results indicate that migration through the skin of immunized mice is a harmful process, eventually leading to the death of most PySPZs.Those that escape killing in the dermis by invading a blood vessel cannot establish a successful infection, evidencing a loss of fitness during this deleterious progression, as supported by the decrease of parasite velocity in the skin.This loss of fitness likely results in the impairment of liver invasion and/or intrahepatic development.
Using this immunization protocol, a library of anti-PyCSP mAbs was generated to identify clonal effectors capable of sterilizing infection initiated by the skin microinjection of 5,000 PySPZs, one day after the passive transfer of 100 µg of the purified antibody.Out of the six mAbs tested, three targeting the NT region and three the repetitive region of PyCSP, only the anti-repeats J6 mAb (IgG1) completely blocked PySPZ infection (Fig. 3a).The non-protective anti-NT W24 mAb (IgG1) was used as an isotype control in the following experiments.In general, anti-NT mAbs are less effective to inhibit SPZ invasion than anti-repeats mAbs 31,32 .An exception is the anti-NT 5D5 mAb that when administered iv at high dose (300 µg) and prior to intravascular SPZ challenge, protects mice better, but not sterilely, when compared to the anti-repeats 2A10 mAb 33 .The J6 mAb remarkably recapitulates the protection observed using the sterilizing PyCSPimmunization protocol.Therefore, no protection is observed after challenge using the adjusted iv dose of 2,500 PySPZs, but sterilizing protection is obtained after skin inoculation with 5,000 PySPZs (Fig. 3b; iv and skin).The same pattern of sterile protection is also observed using ten times less PySPZs in the challenge (Supplementary Fig. 3ab).Critically, J6 also completely blocks infection initiated by the bites of infected mosquitoes (Fig. 3b, bite), corroborating that protective antibodies target extravascular PySPZs delivered in the skin.J6 protective activity is independent of the presence of the C3 protein, required for the classical, lectin and alternative complement activation pathways 34 (Fig. 3c), and of the Fc Receptor gamma (FcRg), required for antibody-dependent phagocytosis and cell-mediated cytotoxicity 35 (Fig. 3d).In vitro, J6 decreases PySPZ speed in a dose-dependent manner, reaching the level of motility inhibition observed in vivo (Fig. 3e and 2b).More dramatically and unique, the purified mAb directly kills PySPZs, leading to the incorporation of the live cell-impermeant dye, propidium iodide (PI), into the nucleic acids of the parasite.The loss of selective membrane permeability starts at the anterior pole of PySPZs and the staining propagates towards the posterior end of the parasite (Fig. 3f, red arrowheads).PI incorporation is accompanied by the loss of GFP fluorescence, which notably recapitulates the dotty death phenotype observed in vivo (Fig. 3f, white arrowheads).A flow-cytometric assay was developed to detect this cytotoxic activity based on the quantification of live (GFP+/PI-) and dead (GFP-/PI+) SPZ population after 45 minutes of incubation at 37 °C (Supplementary Fig. 3cd).Fifteen anti-PyCSP mAbs were screened for their killing capacity using this method.Six of these mAbs recognize the PyCSP repeats and nine, the NT of the protein as determined by ELISA (Supplementary Table 1).However, only the two antirepeats mAbs, J6 and 15D7, which bind to the PyCSP major repeats (QGPGAP) and strongly label live PySPZs, are cytotoxic in vitro and confer sterile protection in vivo (Fig. 3a,g and Supplementary Fig. 3e).Accordingly, mAbs against the major repeats of CSP from Pb, another rodent-infecting parasite (mAb 3D11), and from Pf, the most lethal human-infecting malaria parasite (mAb 2A10), are also cytotoxic to Pb and PfSPZs, respectively (Fig. 3h), showing that this unique killing activity of anti-repeats antibodies is species-transcendent.
Despite the killing activity of anti-repeats mAbs, ~20% of Pb and Py GFP-expressing SPZs, and ~50% of PfSPZs remain alive after incubation with high concentrations of cytotoxic mAbs.To determine if GFP expression could explain this difference, we compared the killing activity of J6 mAb on wild-type and GFP-expressing PySPZ.Since both parasites are equally susceptible to J6 cytotoxicity (Supplementary Fig. 3f), a potentiating effect of GFP expression in SPZ death was discarded.Interestingly, PbSPZs expressing a hybrid PbCSP harbouring the central part of the PfCSP 36 (Fig. 3h, PbPfSPZ) are as susceptible as PfSPZs to intermediate concentrations of the cytotoxic 2A10-anti-PfCSP repeats mAb.However, at high concentration, only ~20 and 50% of PbPfSPZs and PfSPZ remain alive, respectively (Fig. 3h).Since this percentage roughly corresponds to the population of immotile parasites in similar experimental conditions 37 , we tested if the antibody cytotoxicity is dependent on SPZ motility.
As shown in the figure 4a, inhibition of actin polymerization by cytochalasin D (CD) at a concentration that blocks parasite motility 38 , strongly reduces mAb cytotoxicity.After 45 minutes of incubation, however, a progressive killing is observed in the presence of CD (Fig. 4b, green squares), suggesting that a factor other than actin-dependent motility contributes to the parasite death.CSP is known to be secreted at the apical pole of SPZs and then translocated backwards, forming a thread-like precipitate at the posterior end of the SPZ in the presence of anti-repeats mAb 38 (circumsporozoite precipitation reaction, CSPR; Fig. 4c, J6 -CD).When CSP translocation is inhibited by CD, a vesicular and membranous extrusion is formed at the SPZ apical end 38 (Fig. 4c, J6 +CD).This turgid extrusion can be easily evidenced by microscopy and differentiated from the CSPR by the anterior staining of microtubules (Fig. 4c, tubulin).The absence of GFP signal in the CSPR and apical extrusion (AE) indicates that these structures are frequently devoid of cytoplasmic content.An exception is the punctual GFP fluorescence structure observed inside the CSPR (Fig. 4c, J6 -CD), similar to the persistent fluorescent signal observed during the dotty death.AE and CSPR are observed after SPZ incubation with W24 and J6, however, while the cytotoxic anti-repeats mAb mostly induces large, well-delineated and refringent extensions at the anterior (strong AE) and posterior end of SPZs (strong CSPR), the anti-NT mAb induces amorphous and low-refringent structures (Fig. 4c, weak AE and CSPR).
Both anti-CSP mAbs are thus capable of triggering an exacerbated secretion in PySPZs, albeit with different intensities (Fig. 4c).To determine if the strong oversecretion induced by J6 could be linked to PySPZ death, the intracellular calcium chelator BAPTA-AM, a known inhibitor of micronemal vesicular secretion 39 was tested in the presence of the cytotoxic mAb.Both strong AE (Fig. 4d) and J6 cytotoxicity (Fig. 4e) were inhibited by BAPTA-AM, indicating that micronemal secretion could be involved in the PySPZ killing.The fact that J6-treated PySPZs become permeable to PI through their anterior pole, the region of secretory granule discharge, added to the presence of a pore-forming like protein involved in the wounding of the host cell plasma membrane called SPECT2 (Sporozoite microneme Protein Essential for Cell Traversal 2, SP2) 6 in these granules, prompted us to test the role of SP2 in the SPZ killing by anti-repeats mAb.The disruption of the SP2 gene in PySPZs (SP2 ko ) 40 rendered the knockout parasite resistant to J6-mediated killing despite the formation of the CSPR and the AE in CD-treated SPZs (Fig. 4f).Therefore, in the absence of SP2, the actin polymerization-dependent CSPR cannot lead alone to parasite death.Equally, apical oversecretion induced by J6 in CD-treated PySPZ SP2ko does not significantly affect parasite viability (Fig. 4f), showing that SP2 acts as a terminal effector in the J6-induced PySPZ death.
Since PySPZ death is associated with anterior secretion, actin polymerization and the presence of SP2, we hypothesize that CSP could be acting as a shield, inhibiting the insertion of this poreforming like protein in the parasite membrane.Therefore, when the continuous anterior secretion and backwards shedding of CSP elicited by J6 deplete the secretory stock of CSP, the parasite membrane becomes devoid of this protective shield, starting at the anterior pole, and thus, susceptible to the SP2 cytotoxicity.Indeed, while almost all of J6-treated live GFP + PySPZs display a J6 + CSP staining shielding the entire SPZ body (Fig. 4g, PySPZ J6 live; J6 + : 32/34; ~94%), most of dead GFP -PySPZs lack the CSP staining on the surface of their anterior end (Fig. 4g, PySPZ J6 dead; J6 -: 51/67; ~76%).Accordingly, the apical pole of SP2 ko PySPZs is predominantly J6 -following mAb-treatment, but PySPZs remain alive due to the absence of SP2 (Fig. 4g, PySP2ko J6 live; J6 -: 28/32; ~87%).Finally, the non-cytotoxic W24 mAb induces the shedding of the CSP population harbouring the NT region, but not of the processed CSP.This generates a body-labelling pattern positive for J6 and negative for W24 (Fig. 4g, PySPZ W24 live; J6 + W24 -: 39/43; ~91%), indicating that this J6 + CSP population is enough to protect W24treated PySPZs from SP2 cytotoxicity.
PbSP2 ko SPZs are also much less susceptible to the killing activity of anti-repeats 3D11 cytotoxic antibody (Fig. 4h), despite having similar motile capacity in the absence or presence of mAb, as well as the same surface amount of CSP as control SPZs (Supplementary Fig. 4a-c).To distinguish a local and individual cis-effect of SP2 on SPZs from a trans-effect of SP2 released from antibody-killed parasites on the rest of SPZ population, we mixed PbRFP-expressing control parasites with PbGFP-expressing SP2 ko SPZs in the presence of different concentrations of cytotoxic antibody.Figure 4i shows that at killing concentrations of mAbs, the SP2 released from susceptible control SPZs does not increase the killing of PbGFP-SP2 ko SPZs, indicating that SP2 released from dying/dead parasites does not kill in trans.PbGFP and PbRFP control SPZs are equally susceptible to the cytotoxic mAb (Supplementary Fig. 4d).
In addition to SP2, the disruption of two other proteins, SPECT (SP) 41 and CelTOS 42 , severely impairs the host-cell wounding capacity of PbSPZs (Fig. 4j).However, only the knockout of the SPZ-specific proteins, SP and SP2, decrease the cytotoxic activity induced by anti-repeats mAb (Fig. 4k).CelTOS is thought to target and wound the inner leaflet of the host cell plasma membrane 43 , and accordingly, its disruption does not alter the antibody cytotoxicity.
Altogether these results indicate that CSP has a protective role, shielding the parasite against membrane-targeted cytotoxic proteins.Anti-CSP repeats mAbs trigger the oversecretion of apical vesicles containing CSP 44,45 and the cell wounding proteins, SP 41 and SP2 6 .Concomitantly, CSP is translocated and shed at the posterior pole of SPZs until stripped from the parasite surface.The exposition of the parasite membrane then renders SPZ susceptible to these secreted cell-wounding proteins, which are final effectors of mAb-induced cytotoxic death (Supplementary Fig. 5).
By dissecting the determinants of protection associated with anti-PyCSP humoral immunity, we identify the skin as a critical site for protection and the mechanisms by which anti-CSP antibodies eliminate SPZs, via a singular cytotoxic activity.Our results support the notion that protective anti-CSP repeats antibodies preferentially target PySPZs in the cutaneous tissue 7,25 , the physiological site of parasite inoculation.PySPZ neutralization in the blood is thus less efficient than in the skin 25 .Accordingly, sterile protection of passively immunized hosts against iv inoculation with SPZs is only effective at high concentrations of transferred mAbs 5,21,25 and/or using a small effective inoculum of infectious SPZs 21,22,25,46 .This lower neutralizing efficacy could explain the lack of protection of immunized hosts possessing high titers of anti-CSP antibodies when iv challenged with SPZs [26][27][28] .Remarkably, anti-repeats mAbs directly kill malaria SPZs in a species-transcendent manner, clarifying why pre-incubation with anti-repeats mAbs efficiently neutralizes SPZ infectivity [1][2][3] .In contrast to other anti-plasmodial cytotoxic antibodies 47,48 , complement and FcRg are not required as downstream host effectors.Cytotoxicity is instead dependent on intrinsic SPZ factors, such as parasite motility or proteins involved in the membrane wounding and traversal of host cells.Since these two factors are both indispensable for the parasite progression and exit from the dermis [7][8][9] , cytotoxic concentrations of mAbs optimally target moving SPZs in the host skin.Strategies to improve the generation of potent immobilizing and cytotoxic antibodies might help to ameliorate the efficacy of CSP-based vaccines.

Parasites, mice and mosquitoes
We used Plasmodium yoelii 17XNL 49 ; Plasmodium yoelii 17XNL expressing GFP under the control of the pbef1αa promoter 50 or hsp70 promoter 51 ; P. yoelii 17XNL SPECT2 ko (PyΔplp1) expressing GFP under the control of hsp70 promoter 40 , Plasmodium yoelii YM expressing a GFPluciferase (GFP-LUC) fusion protein under the control of elongation factor1α promoter 52 ; P. berghei ANKA expressing GFP under the control of hsp70 promoter 53 ; P. berghei cell traversal deficient parasites SPECT ko , SPECT2 ko9 , and CelTOS ko 42 expressing GFP under the control of hsp70 promoter, P. berghei ANKA expressing RFP under the control of eef1-α 54 promoter; P. berghei NK65 expressing GFP under the control of hsp70 promoter 55  Immunization experiments were performed using five to ten animals per experiment, which gives a significance threshold of 75 to 50% of sterile protection using Fisher's Exact Test.SPZ challenge was blind, but not immunization.
Anopheles stephensi mosquitoes (Sda500 strain) were reared in the Centre for Production and Infection of Anopheles (CEPIA) at the Institut Pasteur using standard procedures.For the production of rodent Plasmodium spp.SPZs, mosquitoes were fed on infected RjOrl:SWISS mice 1-2 d after emergence and kept in a humidified chamber at 24°C (Py) or 21°C (Pb).One week after infection, Py infected mosquitoes were fed on naïve RjOrl:SWISS mice.Infected mosquitoes used for natural transmission experiments (15-18 days after the infectious blood meal) were deprived of sucrose for one day before experimentation to enhance the rate of mosquito bites.
For intravenous and footpad injections Py SPZs were collected from infected salivary glands 15-22 d after the infectious blood meal.Pb SPZs were collected from infected salivary glands 21-28 d after the infectious blood meal.
Female Anopheles gambiae or stephensi mosquitoes were fed on a membrane feeder with gametocytes resuspended in human serum AB+ and fresh Rh+ erythrocytes (v/v) 2-4 days after emergence.Mosquitoes were kept at 26°C, 70% humidity and on 10% sucrose.Seven days after the infective blood meal, mosquitoes were fed on non-infected Rh+ RBCs and AB+human serum.
Salivary glands SPZs were collected in phosphate buffered saline at days 17-21.

Recombinant PyCSP and peptides.
The PyCSP open reading frame without the signal peptide and GPI anchor signal sequence (accession number P06914; amino acids 23 to 346) was codon optimized and cloned in a bacterial expression vector pET21b to produce a recombinant protein with 6x histidine residues at the C terminal end.The recombinant protein was produced in E. coli BL21 Star (DE3) (Invitrogen) upon induction with 1mM IPTG.Purification was accomplished using Ni-NTA super flow resin under native conditions, according to the manufacturer's instructions.The several fractions containing the purified protein were joined and subjected to a PD-10 desalting column (GE Healthcare) and the imidazole was removed by the gravitational method (as per manufacturer's instructions) using PBS.Endotoxins were removed using two-phase extraction with Triton-X114 61 and their residual levels were determined using the Pierce Limulus amebocyte lysate (LAL) chromogenic quantitation assay (TermoFisher Scientific).All protein batches used for immunization had final endotoxin levels bellow 1 EU/mL.After removal of endotoxins, protein concentration was determined by semi-quantitative SDS-PAGE stained with Coomassie Blue using BSA as a standard.The N-Terminal (amino acids 23 to 138) and the C-Terminal (amino acids 260 to 346) regions of PyCSP were cloned and expressed as described above to determine by ELISA the specificity of the mAbs generated.Sequence of peptides used to determine anti-PyCSP repeats mAbs: QGPGAPQGPGAPQGPGAP and QQPPQQPPQQPP.
Animals were primed, and boosted two weeks later.Passive sera transfer was performed by iv administration of 0.5 mL of serum collected from individual mice 2 weeks after immunization (as described above).MAbs were injected ip (100-150 μg/mouse) diluted in PBS, 24 hours before the SPZ challenge 62 .

SPZ Challenge
Py SPZs collected from the salivary glands of infected A. stephensi mosquitoes were inoculated iv in the tail vein using a 30-Gx1/2" needle in a U-100 insulin syringe, or in the footpad skin using 35-to 36-G needle with a NanoFil syringe (World Precision Instruments) in naïve, control or immunized mice.Unless stated otherwise, 5,000 SPZs were microinjected in the skin.Mice challenge experiments by mosquito bites were done using A. stephensi mosquitoes infected with P. yoelii 17XNL GFP.Mice anesthetized with a mixture of ketamine (50 mg/kg body weight, Imalgene 1000, Merial) and xylazine (5 mg/kg body weight, 2% Rompun, Bayer) were bitten by 4 to 7 infected mosquitoes during 10-15 minutes.Challenges were performed two weeks after immunization boost, or the following day after sera/mAb transfer.Parasite load in the liver (using P. yoelii YM GFP-LUC) was assessed at 46 hours after infection by measuring whole mice bioluminescence upon injection of 150 mg/kg luciferin potassium salt (Perkin Elmer) and using an IVIS LUMINA II (Perkin Elmer).Parasitemia was determined by thin blood smears or flow cytometry, performed 3-12 days after SPZ challenge.Quantitative analysis of protection was determined using the log of the parasitemia at day 4 post-infection when the blood parasites are still growing exponentially.Sterile protection was defined as the absence of blood stage parasites after day 10 post-challenge.At least 20 microscopic fields (10,000-15,000 red blood cells) or 200,000 red blood cells were examined by flow cytometry for each mouse designated as protected.

SPZ Imaging in the skin
Intravital imaging was performed using high-speed spinning disk confocal microscopy 65 .Briefly, Py GFP SPZs were microinjected in the previously epilated ear pinnae of anaesthetized control or immunized BALB/cJRj mice using a microsyringe (NanoFil 10 μL syringe, World Precision Instruments).When indicated, blood vessels were previously labelled by the iv injection of 50 µg of Alexa Fluor TM 647 BSA (Molecular Probes TM ).The mouse was positioned with the ear flattened on a slide.Images were acquired in different focal planes and analysed using the software ImageJ.The speed of SPZs was calculated for one minute of analysis during the first five minutes post-injection

In vitro gliding assay
The gliding assay was performed using 18 well μ-slide ibiTreat.All materials were kept on ice.
Each test was performed individually.Each well contained 20 μL of a suspension in PBS with 5,000 GFP SPZ, 10% FCS, and different concentrations of mAbs when indicated.The slide was centrifuged for 3 min at 500 rcf, 4° C, and transferred to the microscope chamber, which was kept at 37°C, and 5% CO2.After 5 min of incubation, images were acquired every second, during 2 minutes in an inverted AxioObserver microscope (Zeiss).Two-minutes maximal projections of SPZ trajectories were analysed and circular gliding, defined as a circular motion of crescentshaped SPZ, was quantified using the software ImageJ.The speed of SPZs was determined using the plugin MTrackJ from ImageJ.Briefly, the images acquired were loaded on ImageJ, and manual tracking of moving SPZs was performed.The average speed was calculated considering the length travelled and the time of acquisition.For the in vitro analysis, only SPZs that moved in circular motion during the entire acquisition period were considered.

MAb library
MAbs were produced by the hybridoma technology based on the protocol of Köhler and Mielstein 66 .Mice were immunized using the recombinant PyCSP and the immunization protocol described above.Two weeks after the administration of the booster dose, splenocytes were fused with P3U1 (ECACC 85011417) or with P3x63Ag8.653(ATCC Number CRL-1580™) myeloma cells.Other mAbs were produced by hybridoma cell lines already established, such as 3D11 anti-Pb CSP repeats (MRA-100, BEI Resources) 67 , and 2A10, anti-Pf CSP repeats (MRA-183, BEI Resources) 68 .These cells were not tested for mycoplasma contamination.MAbs were purified from culture supernatants by affinity chromatography using protein G beads (GE healthcare).ELISA using the supernatant was performed for identification of wells containing anti-PyCSP antibody secreting cells.Wells were coated overnight at 4° C with recombinant PyCSP, or its fractions for the specificity determination.After blocking, supernatants or purified antibodies were incubated for 2 hours at 37°C, washed and detected using anti-mouse IgG coupled with peroxidase for 1 hour at 37°C followed by development with TMB.For isotyping, wells were coated with rabbit anti-mouse IgG (H+L) antibody, and detection was done using antibodies anti-IgG and light chain isotypes coupled to peroxidase.The epitope sequences recognized by two anti-NT mAbs were identified using an array containing linear 15-mer peptides with 14-mer peptide-peptide overlap of the entire PyCSP ORF (Pepperprint).The detection was done using Cy3 goat anti-mouse IgG (H+L), and the array was scanned on a motorized stage of an inverted microscope.

In vitro SPZ killing test -viability assay
Unless otherwise stated, GFP SPZs from salivary glands were incubated at 37°C for 45 min, except when otherwise indicated, with 10% FCS in PBS in the presence of different mAbs concentrations.When indicated, 0.5 μM cytochalasin D (Sigma) and/or 100 μM BAPTA-AM (Sigma) was added.J6 mAb was heat inactivated (h.i.) at 70° C for 20 min.After incubation, parasites were kept on ice and incubated with 5 μg/mL propidium iodide (Invitrogen).Viability was determined by flow cytometry as the percentage of GFP + PI -SPZs to the sum of GFP + PI -and GFP -PI + SPZs.The gating strategy is described in the legend of the Supplementary Figure 3.In the case of non-fluorescent Py or Pf SPZ, they were identified using 0.5 μg/mL of Alexa Fluor TM 647 conjugated J6, or 2A10, respectively.For Pb and Pf SPZ, the mAbs W24 and E12 were used as isotype controls, respectively.At least 1,500 events in the SPZ gate were acquired on CytoFLEX S flow cytometer (Beckman Coulter), and analysed using the softwares CytExpert 2.0 (Beckman Coulter) or FlowJo 10.2 (FlowJo LLC).

SPZ immunostaining
For the surface CSP staining, 50,000 GFP SPZs from salivary glands were incubated on ice with 5 μg/mL of different mAbs in PBS for 60 minutes.For CSP quantification in Pb or PySPZ surface, 0.1 or 1 μg/mL of mAb was used.After washing, SPZs were incubated on ice with 4 μg/mL Alexa Fluor TM 647 conjugated anti-mouse IgG (H+L) (Invitrogen) in PBS, and analysed by flow cytometry.The Alexa Fluor TM 647 median of fluorescence was determined from the GFP-positive population in a SSC-H vs GFP-H pseudocolor plot (at least 1,000 events).Data were acquired on a FACSCalibur (Becton Dickinson) or CytoFLEX S flow cytometer (Beckman Coulter), and analysed on the software CytExpert 2.0 (Beckman Coulter) or FlowJo 10.2 (FlowJo LLC).For the secretion analysis, after incubation with 100 μg/mL of mAbs ± 0.5-1 μM CD as described for the viability assay, PyGFP SPZ were fixed in 2% PFA, permeabilized with 0.3% Triton X-100, and stained with 0.25 μg/mL of anti-α-Tubulin (11H10) Alexa Fluor TM 647 (Cell Signaling Technology), 2 μg/mL of goat anti-mouse IgG (H+L) Alexa Fluor TM 350, and 2 μg/mL of Hoechst 33342.When indicated, W24-treated PyGFP SPZ were incubated with J6 conjugated with Alexa Fluor TM 647.Images were acquired using an inverted AxioObserver microscope (Zeiss).

Host cell wounding assay
The assay was performed based on Prudêncio et al. 69 .Briefly, HepG2 cells obtained from the American Type Culture Collection (ATCC Number HB-8065™) were maintained in high glucose Dulbecco's Modified Eagle Medium GlutaMAX TM (Gibco) containing 10% FCS (Biowest) and 1% MEM Non-essential Amino Acids solution (100X) (Sigma-Aldrich) on flasks coated with collagen.
They were not authenticated in our lab, nor tested for mycoplasma contamination.Cells were trypsinized and seeded on 96-well plates 24 hours before infection (50,000 cells/well coated with collagen).Pb GFP control and cell wounding deficient mutant SPZs (12,500 SPZ; MOI 1:4) in the presence of 1 mg/mL Tetramethylrhodamine Dextran (DxRed), lysine fixable, 10,000 MW (Molecular Probes) were transferred to wells with HepG2 cells.The plate was centrifuged for 5 min at 500 rcf, and incubated for 2 hours at 37 °C, 5% CO2 and 10% O2.Cytochalasin D (Sigma) at the concentration of 0.5 μM was used as negative control for cell wounding.The HepG2 population (10,000 events acquired) was identified on a FSC-A vs SSC-A pseudocolor plot, followed by singlet isolation in a FSC-A vs FSC-H plot.GFP positive events were considered as invaded cells, and DxRed positive as wounded cells (Supplementary figure 4e).The gating population of the latter was determined by mechanically wounding cells using a pipette tip, and comparing it to the control.Data were acquired on a CytoFLEX S flow cytometer (Beckman Coulter), and analysed using the softwares CytExpert 2.0 (Beckman Coulter) or FlowJo 10.2 (FlowJo LLC).

Statistical analysis
Statistical significances were determined by one-way ANOVA with Holm-Sidak correction for multiple comparisons, two-tailed Fisher's Exact Test, or two-tailed unpaired t test.Analyses were calculated using GraphPad Prism version 6 for Mac OS X.
and a hybrid CSP containing the tandem repeated region of P. falciparum CSP (PbPf-GPF) generated by crossing P. berghei NK65 expressing GFP with the parasite bearing the hybrid CSP 36 , and P. falciparum NF54 56 .BALB/cJRj, RjOrl:SWISS and C57BL/6JRj mice were purchased from Elevage Janvier.C57BL/6 JHT -/-57 mice were obtained from Instituto Gulbenkian de Ciência, kindly provided by Dr Miguel Soares.C3 ko mice B6.129S4-C3 tm1Crr /J 58 were kindly provided by Dr Pierre-Marie Lledo, and C57BL/6J-FcRg ko Fcer1g tm1Rav 59 by Dr. Pierre Bruhns, both from the Institut Pasteur.All animal experiments were approved by the Animal Care and Use committee of Institut Pasteur (CETEA Institut Pasteur 2013-0093, Ministère de l'Enseignement Supérieur e de la Recherche MESR 01324) and were performed in accordance with European guidelines and regulations (directive 2010/63/EU).For all tests, 4-8 weeks old females were used, and allocate in cages randomly.

Figure 1 :
Figure 1: Sterilizing anti-PyCSP humoral immunity is dependent on the host skin

Figure 2 .
Figure 2. Decrease of motility and killing of parasites in the skin of CSP-immunized mice.

Figure 4 .
Figure 4. Motility, secretion and SPZ-specific wounding proteins play a critical role in the