Structure guided mimicry of an essential P. falciparum receptor-ligand complex enhances cross neutralizing antibodies

Invasion of human red blood cells (RBCs) by Plasmodium falciparum (Pf) merozoites relies on the interaction between two parasite proteins, apical membrane antigen 1 (AMA1) and rhoptry neck protein 2 (RON2)1,2. Antibodies to AMA1 confer limited protection against P. falciparum in non-human primate malaria models3,4. However, clinical trials with recombinant AMA1 alone (apoAMA1) saw no protection, likely due to inadequate levels of functional antibodies5–8. Notably, immunization with AMA1 in its ligand bound conformation using RON2L, a 49 amino acid peptide from RON2, confers superior protection against P. falciparum malaria by enhancing the proportion of neutralizing antibodies9,10. A limitation of this approach, however, is that it requires the two vaccine components to form a complex in solution. To facilitate vaccine development, we engineered chimeric antigens by strategically replacing the AMA1 DII loop that is displaced upon ligand binding with RON2L. Structural characterization of the fusion chimera, Fusion-FD12 to 1.55 Å resolution showed that it closely mimics the binary receptor-ligand complex. Immunization studies showed that Fusion-FD12 immune sera neutralized parasites more efficiently than apoAMA1 immune sera despite having an overall lower anti-AMA1 titer, suggesting improvement in antibody quality. Furthermore, immunization with Fusion-FD12 enhanced antibodies targeting conserved epitopes on AMA1 resulting in greater neutralization of non-vaccine type parasites. Identifying epitopes of such cross-neutralizing antibodies will help in the development of an effective, strain-transcending malaria vaccine. Our fusion protein design is a robust vaccine platform that can be enhanced by incorporating polymorphisms in AMA1 to effectively neutralize all P. falciparum parasites.

demonstrated greater protection than apoAMA1 against P. falciparum in a non-human primate malaria model 9,10 . Vaccine e cacy was strongly correlated with the ability of the binary complex vaccine to increase the proportion of neutralizing antibodies targeting AMA1-RON2 interaction 9,10 . This enhancement in neutralizing antibodies was not only limited to vaccine-type parasites but also against some heterologous parasites 9,10 . Despite these encouraging results, manufacturing and deploying a vaccine that relies on producing and mixing two proteins that need to spontaneously assemble in solution, presents technical challenges. To facilitate vaccine development, we engineered a single chimeric antigen that would recapitulate the receptor-ligand complex and promote the effective development of neutralizing antibodies against P. falciparum.
Engineering a receptor-ligand fusion chimeric malaria vaccine AMA1 is comprised of three domains (Fig. 1A) with domains 1 and 2 together forming a hydrophobic groove, the binding site for RON2L 20 . In this study, we generated fusion chimeras to mimic the structure of the receptor-ligand complex in a single protein immunogen. We replaced a section of the extended PfAMA1 DII loop close to the RON2L binding site that is largely disordered in the apo structure 20 (AMA1 residues T358-K370), with RON2L (RON2 residues T2023-S2059) ( Fig. 2A, Suppl Fig. 1, Suppl Fig. 2E).
We initially generated two recombinant chimeras with the RON2L sequence positioned either in the same direction as the AMA1 primary sequence (Fusion-F D123 ) or in the reverse direction (Fusion-R D123 ) (Suppl Fig. 1A). However, recombinant production of these three-domain chimeras (AMA1 D123 ) in Sf9 cells proved unsuccessful (data not shown). Previous studies showed that binding of a Toxoplasma gondii RON2L to its AMA1 partner led to allosteric structural changes in domain 3 of TgAMA1 21 . Such conformational changes in PfAMA1 D123 may result in protein instability that is not tolerated in the Sf9 heterologous expression system. AMA1 domains 1 and 2 are the main targets of AMA1+RON2L binary complex-induced neutralizing antibodies As AMA1 domains 1 and 2 are su cient for RON2L binding 22 , we hypothesized that these regions may be su cient to generate recombinant chimeras that mimic the structure of AMA1 in the binary complex.
Since earlier vaccine studies used AMA1 D123 containing all three domains to generate the binary complex 9,10 and some AMA1 antibodies target domain 3 23, 24 , we rst tested if domain 3 is required for the enhancement in antibody quality generated by the binary complex vaccine. IgG from rats immunized with apoAMA1 (using AMA1 D123 or AMA1 D12 ) had overall higher levels of AMA1-speci c antibody titers compared to rats immunized with the corresponding binary complex (AMA1 D123 + RON2L or AMA1 D12 + RON2L) antigens (Fig. 1B). Despite this, IgG from the binary complex groups neutralized vaccine type parasites (Pf3D7 strain) to a greater degree than the corresponding apoAMA1 groups (Fig. 1C). Furthermore, comparing anti-AMA1 titer of IgG from rats immunized with apoAMA1 D12 and AMA1 + RON2L binary complex revealed that at similar antibody titer IgG from the binary complex immunized animals exhibited greater parasite neutralization than the apoAMA1 groups (Fig. 1D), indicating that the binary complex immunogen generated better quality antibodies. Importantly, these results show that a receptor-ligand binary complex containing only domains 1 and 2 of AMA1 (AMA1 D12 ) is su cient to improve vaccine quality. It is also noteworthy that at the same immunogen dose used for vaccination (10µg/rat), the AMA1 D12 + RON2L binary complex generated higher levels of neutralizing antibodies compared to AMA1 D123 + RON2L (Fig. 1D), suggesting that AMA1 domains 1 and 2 contain the key targets of neutralizing antibodies.
Fusion chimera closely mirrors the structure of the AMA1+RON2L binary complex Two recombinant chimeras, Fusion-F D12 and Fusion-R D12 , each on two different AMA1 allele backbones (3D7 and HP41) and both lacking domain 3 (Suppl Fig. 1B), were expressed in Sf9 cells and veri ed using conformation dependent mAbs 19,25 (Suppl Fig. 2A and 2B). We hypothesized that if RON2L bound to AMA1 D12 as designed, then the hydrophobic binding groove would be occupied and no longer available to bind free RON2L in solution. ELISAs showed that free RON2L readily bound to apoAMA1 D12 and Fusion-R D12 but not Fusion-F D12 chimera (Suppl Fig. 2C and 2D). This suggests that the hydrophobic binding groove in AMA1 D12 is occupied by RON2L in the forward (Fusion-F D12 ) but not in the reverse (Fusion-R D12 ) chimera. We next sought to establish a detailed molecular blueprint of the Fusion-F D12 chimera using X-ray crystallography. A fusion chimera engineered using the HP41 AMA1 sequence was ultimately selected for structural studies based on yield and stability of the recombinant protein.
The structure of the Fusion-F D12 protein incorporating residues 105 (2nd residue in the chimera) through 470 including the inserted PfRON2L sequence was determined to 1.55 Å resolution (Fig. 2B). Clear, contiguous electron density in the apical groove of Fusion-F D12 enabled unambiguous tracing of the entire PfRON2L sequence (Fig. 2C). Structural analysis revealed that the Fusion-F D12 protein adopts a conserved architecture relative to the structure of the previously observed binary complex 22 with a root mean square deviation (rmsd) of 0.61 Å over 252 Cα positions. Moreover, RON2L in the fusion and binary complex overlay with an rmsd of 0.53 Å over 34 Cα positions (Fig. 2D). The core interactions between RON2L and AMA1 are retained in Fusion-F D12 including R376 of PfRON2L (R2041 in PfRON2) that docks into a pocket at the end of the PfAMA1 apical groove (Fig. 2D, Box 1), and P368 (P2033 in PfRON2) that docks into a pocket formed by Y142, Y234, and Y236 (Fig. 2D, Box 2). A few notable differences were observed, including in the β strand of PfRON2L which is shorter in Fusion-F D12 than the structurally equivalent strand in the binary complex (~ 6.6 Å vs ~ 10.1 Å). Moreover, these β strands are displaced bỹ 1.8 Å, which may be due to the changes in the orientations of P379 and P380 in PfRON2L compared to the equivalent prolines (P2044 and P2045) of PfRON2L in the binary complex ( Fig. 2D-Box 3). Despite the reorganization of the β strand, there is no commensurate change in the AMA1 surface loop against which it packs, indicating a degree of structural permissiveness in this region of the hydrophobic groove (Fig. 2E). In fact, all the AMA1 loops (1b, c, d, and f) that pack against the C-terminal end of the RON2L sequence are highly conserved (rmsd of 0.21 Å over 76 Cα positions) relative to the binary complex. In contrast, some differences are observed in loops 1e/1a that pack against the N-terminal section of the RON2 sequence (Fig. 2E). Notably, the sidechain of N136 on loop 1a is displaced by 2.5 Å and no longer forms a hydrogen bond with the sidechain of N233 positioned on loop 1e ( Fig. 2E-Box 1). The loss of the hydrogen bond may also be responsible for the observed loop displacements. Furthermore, many of the sidechains on loop 1e adopt different rotameric states with N228, N231 and D229 showing relatively similar displacement (~ 2.9 Å, 3.1 Å, and 3.4 Å) while K230 (~ 7.2 Å) and S232 (~ 6.9 Å) show signi cantly larger shifts ( Fig. 2E-Box 2), which may alter epitopes in loop 1e. But overall the interface between PfAMA1 and the inserted PfRON2L sequence in the chimera appears to faithfully mimic that observed in the binary complex.
Antigen and adjuvant effects on immunogenicity, IgG speci city, and parasite neutralizing activity As Fusion-F D12 closely mimics the structure of the receptor-ligand binary complex, we next tested the immunogenicity of the chimera to evaluate if antibody quality will similarly be improved compared to AMA1 D12 . We used the recombinant chimeric antigen based on the 3D7AMA1 D12 backbone (Suppl Fig. 1B and Suppl Fig. 2D) as the 3D7 parasite strain is commonly used in neutralization assays. Two different adjuvants were used, AddaVax, a squalene-based oil-in-water nanoemulsion adjuvant whose composition is similar to MF59 used in certain Flu vaccines 26 , and Freund's, a water-in-oil emulsion adjuvant known to induce robust antibody responses in rodents. Antibody titers were higher in the Freund's adjuvant group compared to AddaVax group f in both puri ed IgG and serum (Suppl Fig. 3). Furthermore, puri ed IgG from Fusion-F D12 group had lower AMA1-speci c antibody titer compared to apoAMA1 D12 immunized groups in both adjuvants (Fig. 3A).
Surprisingly, in contrast to the enhancement of the neutralizing activity observed using puri ed IgG from the AMA1 D12 + RON2L binary complex (Fig. 1C, 1D and Suppl Fig. 4B), Fusion-F D12 immunized animals had lower neutralizing activity (Fig. 3B). It is important to note that the binary complex is a 2-component vaccine in which the proportion of the immunogen in the form of a receptor-ligand complex following immunization is determined by the on-off rate for RON2L binding to AMA1 (K D = ~ 20nM 22 ) and may result in a population of dissociated complex in the animal. In contrast, our ELISA results showing that free RON2L peptide is unable to bind to the Fusion-F D12 chimera (Suppl Fig. 2C and 2D) suggests that the binding groove in AMA1 is continuously occupied by RON2L. These apparent differences in RON2L occupancy between the fusion chimera and the AMA1 + RON2L binary complex following immunization may result in the observed differences in anti-AMA1 speci c IgG titer and its neutralizing activity. Interestingly, anti-AMA1 titer in the puri ed IgG of the binary complex group was intermediate to that of apoAMA1 and Fusion-F D12 groups (Suppl Fig. 4A).
The stable binding of RON2L to the hydrophobic groove of AMA1 in the Fusion-F D12 chimera could lead to the generation of new epitopes that are not present on apoAMA1 D12 resulting in the generation of fusion-speci c antibodies. A closer inspection of the surface of the apoAMA1 D12 and the Fusion-F D12 shows distinct differences in charge density particularly within close proximity to the RON2L binding pocket of AMA1 that may in uence immunogenicity differently between the two proteins ( Fig. 3C).
Therefore, while the apoAMA1 D12 induced antibodies can only target AMA1, the Fusion-F D12 immunogen may generate antibodies targeting both AMA1 as well as fusion-speci c epitopes, and the overall proportion of such antibodies may impact the neutralizing activity of the polyclonal IgG. Detection of AMA1-RON2 complex-speci c antibodies has not been possible previously due to their dynamic interaction. Having a stable fusion chimera allowed us to examine if such antibodies are generated by comparing the ratio of antibodies binding apoAMA1 and Fusion-F D12 . Animals immunized with the fusion chimera showed a signi cantly greater ratio of Fusion IgG titer to AMA1 IgG titer, suggesting a proportion of the antibodies induced by the Fusion-F D12 may target fusion-speci c epitopes (Fig. 3D). Interestingly, again the proportion of Fusion-F D12 binding IgG in the binary complex group was intermediate to that of apoAMA1 and Fusion-F D12 (Suppl Fig. 3C). As apoAMA1 lacks any fusion-speci c epitopes, the increase in the titer ratio in the fusion chimera immunized animals suggests the generation of fusion-speci c antibodies. To nd direct evidence of the presence of such antibodies we performed a competition ELISA using normalized anti-AMA1-speci c IgG from individual animals from each group (Suppl Fig. 6A). As expected, quenching of AMA1-speci c antibodies with excess free apoAMA1 abolished reactivity of IgG from the AMA1 D12 group to both AMA1 and Fusion-F D12 antigens ( Fig. 3E and 3F, blue bars). Interestingly, in the fusion chimera immunized animals, IgG binding to Fusion-F D12 antigen but not AMA1 was enhanced upon removal of AMA1-speci c antibodies ( Fig. 3E and 3F, pink bars). This suggests that AMA1 targeting IgG near the RON2L binding site may compete for binding with fusion epitope-speci c antibodies and may in uence the overall neutralizing activity of the polyclonal IgG.

Fusion-F D12 enhances the proportion of broadly neutralizing antibodies
In our previous studies, we observed that vaccination with the AMA1 + RON2L binary complex not only enhanced protection against vaccine-type parasites but also increased the proportion of certain straintranscending antibodies 9,10 . We hypothesized that this is in part due to the binary complex enhancing antibodies against more conserved regions of AMA1 such as loops 1e and 1f (Fig. 2E). Interestingly, the proportion of antibodies targeting loops 1e and 1f was higher in the Fusion-F D12 group compared to apoAMA1 D12 , while no differences were observed against loops 1bcd that contains the highly polymorphic 1d loop ( Fig. 4A-4C). The higher proportion of loop 1e targeting antibodies in the Fusion-F D12 group compared to AMA1 D12 also suggests that the conformational changes in this loop observed in the fusion chimera (Fig. 2E, Box 2) did not affect immunogenicity against this important neutralizing antibody target.
Furthermore, the fusion chimera group also had a higher proportion of IgG against AMA1 from the PfFVO strain, a heterologous parasite that differs substantially from vaccine type Pf3D7 AMA1 sequence (Fig. 4D, 4E and Suppl Fig. 3). Antibodies generated against 3D7AMA1 generally exhibit poor crossneutralization of heterologous parasites 27 . Once again, the anti-FVO titer in AMA1 + RON2L binary complex group was intermediate to that of apoAMA1 D12 and Fusion-F D12 groups (Suppl Fig. 4D). As the Fusion-F D12 group had higher levels of IgG binding to epitopes conserved between 3D7 and FVO strains of P. falciparum, we tested if this would differentially affect the invasion of PfFVO parasites. At the same anti-3D7AMA-speci c antibody titer (30,000 ELISA units), neutralizing activity of IgG from Fusion-F D12 group was on average nearly four times higher than apoAMA1 against FVO strain (Fig. 4F), suggesting that the fusion chimera enhanced antibodies targeting cross-neutralizing epitopes.
Fusion-F D12 immunized rat sera exhibit enhanced parasite neutralization compared to apoAMA1 D12 While anti-AMA1 antibody titers of puri ed IgG from the Fusion-F D12 group were lower than the apoAMA1 D12 group, anti-fusion protein titers were proportionally higher in the fusion chimera immunized animals (Suppl Fig. 3A). We reasoned that this difference may be due to the fusion protein preferentially generating antibodies to fusion-speci c epitopes that are not present in the apoAMA1 (Fig. 3C). While the serum titer against both apoAMA1 and Fusion-F D12 were proportional to the overall immunogenicity within each animal ( Fig. 5B and 5E) there was greater fusion-speci c antibodies in the Fusion-F D12 group following protein G column puri cation of IgG ( Fig. 5A and 5D). This effect was observed after accounting for immunogenicity differences between the two groups (Suppl Fig. 3). As these differences were observed using two different adjuvants, it suggested an increase in the proportion of fusion-speci c IgG in the Fusion-F D12 group after a nity puri cation. The reason for this enrichment is not clear, however, since differences in antigen speci city of the IgG may in uence their biological activity, we compared the antibody titer in the puri ed IgG and serum for their respective parasite neutralizing activity. At similar overall anti-AMA1 titer, column puri ed IgG from apoAMA1 D12 immunized animals had greater or proportional neutralizing activity than Fusion-F D12 in the AddaVax and Freund's groups respectively ( Fig. 5C and 5F). Surprisingly, however, at the same anti-AMA1 antibody titer, heated inactivated serum from the Fusion-F D12 Freund's group neutralized parasites signi cantly higher compared to the apoAMA1 D12 group (Fig. 5G, Suppl Fig. 5A and 5B), suggesting the Fusion-F D12 chimeric antigen enhanced overall antibody quality. The differences in the neutralizing activity of serum compared to puri ed IgG prompted us to examine potential differences in the immunoglobulin type in the sera of these animals. Generally, a Th1 immune response leads to the production of IgG2a isotype while a Th2 response leads to IgG1 isotype antibodies 28 . Fusion-F D12 immunized animals had proportionally higher levels of the IgG1 isotype antibodies (Suppl Fig. 6B and 6C). We also observed a greater proportion of IgM type antibodies against the fusion chimera in the Fusion-F D12 immunized serum compared to apoAMA1 D12 (Suppl Fig. 6C). Notably, IgM type antibodies against malaria antigens including AMA1 are produced robustly after malaria infection, expand with repeated exposure, and exhibit neutralizing activity 29 . Vaccination did not induce measurable IgA type antibodies (not shown). Additionally, only low levels of RON2L directed antibodies were detected in the Fusion-F D12 group (Suppl Fig. 6D), levels that are not likely to in uence neutralizing activity 2 . The impact of vaccine-induced differences in Ig type or IgG isotype targeting AMA1 on P. falciparum parasites in vivo remains to be determined.

Discussion
Stalling of malaria control efforts in recent years and the threat of a resurgence in P. falciparum malaria related deaths 11  In this study using structure-guided design, we engineered a chimeric immunogen to mimic the structure of the P. falciparum receptor-ligand (AMA1-RON2L) complex. We rst determined the minimal region of AMA1 that is su cient to enhance antibody quality upon binding to RON2L. The structure of one of the fusion designs (Fusion-F D12 ) was determined to 1.55 Å resolution and found to mirror the structure of the binary complex. Immunization studies performed in rats showed adjuvant-dependent changes in antibody quantity and quality. Importantly, the Fusion-F D12 chimera like the binary complex generated higher levels of cross neutralizing antibodies targeting conserved epitopes on AMA1. An interesting observation was the difference in activity of antibodies in the serum and following a nity puri cation of IgG. While Fusion-F D12 immunized rat serum exhibited greater parasite neutralization, a nity puri ed IgG had lower neutralizing activity compared to apoAMA1 D12 immunized animals. Such differences in neutralizing activity in the serum and puri ed IgG may explain the superior protection observed in vivo compared to moderate increases in neutralizing activity observed in vitro using a nity puri ed IgG following AMA1 + RON2L binary complex immunization 10 .
The ability of the Fusion-F D12 chimeric antigen to enhance neutralization of both vaccine type and nonvaccine type parasites highlight the potential of using structure guided antigen design to improve antibody quality against AMA1. However, this was at the cost of reduced immunogenicity against AMA1 and generation of fusion-speci c antibodies that may not be functional. Eliminating fusion-speci c epitopes through targeted mutagenesis or glycan masking could further promote antibody responses against AMA1. Additionally, identifying targets of the broadly neutralizing antibodies generated by the chimera and incorporating AMA1 polymorphisms in the fusion design can broaden the antibody repertoire while improving antibody quality to effectively neutralize all P. falciparum parasites. Other parasite antigens like RH5 that are targets of potent neutralizing antibodies also have interaction partners 34,35 . Applying structure guided antigen design to focus antibody responses against critical epitopes can help develop effective next generation malaria vaccines.

Cloning, expression in Sf9 cells, and protein puri cation
A clone encoding the full ectodomain (domains 1, 2 and 3) of PfAMA1 (PfAMA1 N25 to K546) sequences from 3D7 and HP41 strains with a section of the DII loop (T358 to K370) replaced by the PfRON2L sequence (T2023 to S2059) and a six residue in the forward (Fusion-F D123 ) or reverse (Fusion-R D123 ) direction with suitable linkers (Suppl Fig. 2) was codon optimized for insect cells and synthesized by GenScript. A second set of constructs encompassing PfAMA1 D1 and D2 (N104 to E438) corresponding to both 3D7 and HP41 AMA1 sequences with the embedded PfRON2L insertion (Fusion-F D12 and Fusion-R D12 ) was ampli ed by PCR and sub-cloned into a modi ed pAcGP67b vector with a TEV-cleavable Nterminal hexahistidine tag. Viruses for insect cell protein production were generated and ampli ed using established protocols 36 . Recombinant proteins including apoAMA1 D12 was produced and puri ed from the insect cell culture supernatant as described previously 37 . Brie y, Ni-a nity chromatography was followed by overnight TEV cleavage, and further puri cation using a Phenix.re ne 44 , and complete structural validation was performed with Molprobity 45 , including analysis of the Ramachandran plots, which showed 98% of residues in the most favored conformations. Five percent of re ections were set aside for calculation of R free . Data collection and re nement statistics are presented in Suppl Table 1. The atomic coordinates and structure factors for Fusion-F D12 chimera have been deposited in the Protein Data Bank under the following code: 8G6B.
Peptide Synthesis RON2L and AMA1 loop peptides 1bcd, 1e and 1f were synthesized by LifeTein (South Plain eld, NJ). Quality control included mass spectrometry and high-performance liquid chromatography for assessing purity. The peptides utilized were found to be over 95% pure based on the results obtained from these evaluations.

Animals, Adjuvants And Antigen Dose
Outbred female Sprague Dawley rats (Charles River Laboratory) that were 5-6 week old (n = 4-5 per group) were used in all our studies. All animal experiments were approved by the Johns Hopkins Animal Care and Use Committee (ACUC), under protocol RA22H291. The AMA1-RON2L complex was prepared immediately before immunization by mixing the two proteins at a 1:3 gram ratio as previously described 2,9,10 . The mixture was then incubated for 30 min at room temperature (RT) to spontaneously form a complex 46 . Antigens diluted in 1X phosphate buffered saline (PBS, pH 7.4) and mixed with equal volume of AddaVax (Invivogen) or Freund's (Sigma, complete or incomplete) adjuvant. AddaVax adjuvant was mixed by pipetting while antigen in Freund's (complete or incomplete) were emulsi ed by vortexing for 30 min. Antigen were administered subcutaneously three times at 2-week intervals. Animals receiving antigen in the Freund's group received their rst immunization in complete Freund's adjuvant and the two boosts in incomplete Freund's adjuvant. All data except that shown in Fig. 1 and Suppl Fig. 4 used 15µg antigen per animal per dose while data in Fig. 1 and Suppl Fig. 4 used 10µg AMA1( D123 or D12 ) or 10µg AMA1( D123 or D12 ) + 30 µgRON2L binary complex per animal per dose. Serum aliquots were mixed with 50µl of human RBCs for 1hr at RT on a rotator to reduce non-speci c binding activity in neutralization assays and stored at -80°C.

Sample Preparation For Immunogenicity Evaluation And Parasite Neutralization Assays
IgG were a nity puri ed on pre-equilibrated protein G column (GE health sciences), dialyzed against RPMI 1640 medium, concentrated on AMICON ULTRA 3k cutoff spin columns (Millipore) and sterile ltered using AMICON Ultrafree-MC Sterile 0.22 µm tubes (Millipore). Serum samples were diluted in RPMI 1640 medium and sterile ltered as described above.

ELISA
The assay was performed as described 47 with some modi cations. Immulon 4HBX at bottom 96-well plates (Thermo Fisher) were coated with indicated recombinant antigens (0.5 µg/ml) and incubated overnight at 4°C. Antigen-speci c ELISA units were rst determined by generating a standard curve using serially diluted serum made previously by mixing equal volumes of serum from rats immunized with AMA1 and AMA1 + RON2L binary complex. The dilution that produced an OD 405 of 1 was identi ed, and the reciprocal of that dilution was used to assign ELISA units to the standards. All samples were then tested against this same standard. Antibody binding was detected using immunoglobulin-speci c (IgG or IgM) or isotype-speci c (IgG1 or IgG2a) secondary antibody conjugated to HRP (Thermo sher Scienti c).
Antibody titer against various AMA1 loops (1bcd, 1e and 1f) and RON2L peptide were measured by coating plates with 2 µg/ml of the respective peptides overnight at 4°C. To assess the relative proportions of antibodies in the different groups, each sample was normalized to contain the same quantity of anti-AMA1 titer (ELISA units).
Competition ELISAs were performed as above with some modi cations. Brie y, plates were incubated with the coating antigen (0.5 µg/ml) for 2hr at 37°C. Dilutions of IgG from each group was pre-incubated with 2 µM r3D7AMA1 for 1hr at RT before adding to ELISA plate. The proportion of antibodies binding to the target antigen was measured relative to the no competition wells.
Parasite Neutralization Assay Using A nity Puri ed IgG 1-cycle assay (NIH): Infected RBCs were incubated with a nity puri ed IgG (2mg/mL) for 40h. Parasitemia was quanti ed by biochemical measurement using a Pf lactate dehydrogenase assay as described previously 49 .
2-cycle assay (JHU): The assay was setup as described above with indicated amounts of total IgG, incubated for 72 h at 37°C and stained with SYBR green dye (Invitrogen). Infected cells were counted using a AttuneNxT owcytometer connected to an autosampler. All assays were carried out in duplicate.

Parasite Neutralization Assay Using Rat Serum
Heat inactivated individual serum (up to 2% in the assay wells) were used in the 2-cycle neutralization assay as described above. Non immunized rat serum at the same dilutions were used as controls. Percent inhibition of invasion = 1-(% parasitemia in test serum well/% parasitemia in control serum well).  t-test was performed to compare differences between groups. (C) In vitro neutralization (1-cycle) assay against vaccine-type 3D7 parasites using 2mg/mL of total IgG from each animal. Data are from individual animals (n=4) done in triplicate and horizontal lines show the mean neutralizing activity in each group. Welch's t-test was performed to compare differences between groups. (D) Relationship between anti-AMA1 titer in the IgG (x-axis) and neutralizing (1-cycle) activity in 2mg/mL total IgG (y-axis).  that interacts with loop 1a. A hydrogen bond formed between N233 (loop 1e) and E136 (loop 1a) in the binary complex structure that is lost in the chimera. The curved arrows indicate the displacement of the side chains. Box 2 shows the differences in loop 1e. There is a displacement of ~2.9 Å between the loops, most likely caused by the inability to form hydrogen bond between N233-E136 in chimera. The different orientations of the side chains are indicated by curved arrows and the extent of displacement is indicated.

Figure 3
Qualitative changes in vaccine response to apoAMA1 vs Fusion-F D12 immunogens. (A) AMA1 titer in puri ed IgG (10mg/mL) from animals immunized with apoAMA1 D12 (blue) or Fusion-F D12 (purple) antigens in AddaVax ( lled squares) and Freund's (open squares) adjuvants. Data are from individual animals (n=4) done in duplicate. Horizontal line marks the mean titer in each group. Welch's t-test was performed to compare differences between groups. (B) In vitro neutralization (2-cycle) assay against vaccine-type Pf3D7 parasites. Puri ed IgG from AMA1 D12 and Fusion-F D12 groups were normalized for 3D7AMA1 titer within each adjuvant group. Data are from individual animals (n=5) done in duplicate. Horizontal line shows mean neutralizing activity in each group. Welch's t-test was performed to compare differences between groups. (C) Differences in surface charge density between apoAMA1 D12 and Fusion-F D12 . (D) Proportion of IgG in animals within each group binding to Fusion-F D12 and apoAMA1. Data shows ratio of antibody titer from IgG of individual animals (n=5) performed in duplicate. Welch's t-test was performed to compare differences between groups. (E) Competition ELISA (cELISA) to determine IgG speci city against apoAMA1 antigen in the absence (-) or presence (+) of 2µM free apoAMA1 between apoAMA1 and Fusion-F D12 immunized animals. Assays were performed in duplicate and shown as mean±SEM (n=5 per group). (F) Competition ELISA (cELISA) to determine IgG speci city against Fusion-F D12 antigen in the absence (-) or presence (+) of 2µM free apoAMA1 between apoAMA1 and Fusion-F D12 immunized animals. Assays were performed in duplicate and shown as mean±SEM (n=5 per group).