Structural elucidation of full-length Pfs48/45 in complex with potent mAbs isolated from a naturally exposed individual

doi:10.21203/rs.3.rs-3605634/v1

Download PDF

Article

Structural elucidation of full-length Pfs48/45 in complex with potent mAbs isolated from a naturally exposed individual

https://doi.org/10.21203/rs.3.rs-3605634/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Biomedical interventions capable of preventing the transmission of malaria-causing Plasmodium falciparum (Pf) between the human host and mosquito vector could prove a valuable tool in malaria elimination efforts. Pfs48/45, a gamete-surface protein essential for Pf development in the mosquito midgut, is a key component of clinical-stage transmission-blocking vaccines. Antibodies against this antigen have been demonstrated to efficiently reduce Pf transmission from humans to mosquitoes. Potent human monoclonal antibodies (mAbs) against Domain 3 (D3) of Pfs48/45 have been structurally and functionally described; however, in-depth information about other inhibitory epitopes on Pfs48/45 is currently limited. Here, we present a 3.3 Å resolution cryo-electron microscopy structure of full-length Pfs48/45 in complex with potent mAbs targeting Domain 1 (D1) and D3, and a moderately potent mAb targeting Domain 2 (D2). Our data indicate that while Pfs48/45 D1 and D2 are rigidly coupled, there is substantial conformational flexibility between D2 and D3. Characterization of mAbs against D1 revealed the presence of a conformational epitope class that is largely conserved across Pf field isolates and is associated with recognition by highly potent antibodies. Our study provides comprehensive insights into epitopes across full-length Pfs48/45 and has implications for the design of next-generation malaria transmission-blocking vaccines and antibodies.

Biological sciences/Structural biology/Electron microscopy/Cryoelectron microscopy

Biological sciences/Microbiology/Parasitology

The severe disease burden of malaria is evident from the estimated 619,000 deaths worldwide in 2021, primarily occurring in children under the age of 5 in sub-Saharan Africa¹. The deadliest form of malaria is caused by Plasmodium falciparum (Pf) parasites that are transmitted to humans by Anopheles mosquitoes². The development of effective vaccines and monoclonal antibodies (mAbs) to combat malaria is challenged in part by the transient presentation of immunogenic epitopes throughout the parasite’s complex lifecycle^3,4. Efforts spanning more than three decades in targeting epitopes on the pre-erythrocytic antigen Pf circumsporozoite protein (PfCSP) have culminated in the approvals of the RTS,S/AS01 and R21/Matrix-M™ subunit vaccines by the World Health Organization (WHO)^5,6. While encouraging, suboptimal levels of sterile protection in inoculated populations despite high observed antibody titers necessitates orthogonal strategies^7,8. Highly potent PfCSP-directed mAbs have been identified through different strategies, including vaccination with radiation-attenuated sporozoites in humans^9,10. Extensive preclinical structure-function relationship studies have facilitated the development of the PfCSP-directed mAbs, L9LS and CIS43LS, which are currently advancing through clinicals trials for malaria prophylaxis^11,12.

Beyond the pre-erythrocytic stage, targeting antigens expressed by parasites during their asexual blood stage¹³ and sexual mosquito stage³ provides additional opportunities for biomedical interventions. For example, vaccines preventing the transmission of sexual stage gametocytes from humans back to mosquitoes exploit a bottleneck that presents a critical opportunity to end the chain of transmission. TBVs function by eliciting a humoral response against sexual stage antigens that are required for parasite development in the mosquito midgut and represent a complementary approach to anti-infection stage vaccines towards malaria elimination^14,15.

Pfs48/45 is expressed during gametocytogenesis and its orthologue from Plasmodium berghei, Pbs48/45, is essential for male gamete fertility¹⁶. Pfs48/45 is attached to the surface of the parasite membrane via a glycophosphatidylinositol (GPI)-linked anchor^17,18. Pfs48/45-directed TB31F, which was derived from humanization of rat mAb 85RF45.1, is a highly potent clinical-stage mAb¹⁹. The structural and functional elucidation of the Pfs48/45-TB31F interaction has provided the molecular rationale for clinical evaluation of TB31F and further improved biochemical characterization of the Pfs48/45 protein²⁰.

Clinical development of Pfs48/45-based TBVs has been hindered by the inherently unstable nature of this protein^21,22. To overcome this challenge, the leading Pfs48/45-based TBV candidate is composed of Domain 3 (D3) genetically conjugated to a stabilizing carrier protein (R0), together known as R0.6C, or the Pro-domain of Pfs230, together known as ProC6C. Such promising TBV candidates comprising Pfs48/45 continue to enter Phase I clinical trials for evaluation of humoral immunogenicity (NCT04862416, PACTR202201848463189, ISRCTN13649456; as well as NCT05400746)²³. Recent preclinical studies have also demonstrated additional opportunities for improved immune responses by utilizing other designs. For example, two studies have reported the rational design of stabilizing mutations in D3 that conferred enhanced protein expression and stability leading to increased potency of the humoral response in animal studies compared to the wild-type protein^24,25. Importantly, these studies leveraged structure-function relationship information of the most potent mAb TB31F²⁰ against Pfs48/45 D3, thus highlighting the value of characterizing mAb epitopes at the molecular level to enable protein engineering of next-generation vaccine candidates.

Antibodies targeting all the domains of Pfs48/45 have been identified with varying degrees of transmission-reducing activity (TRA), yet the relationship between epitope and antibody potency remains poorly understood, particularly for Domains 1 and 2 (D1 and D2)^26,27. For example, from a panel of 16 mAbs derived from mice immunized with recombinant full-length Pfs48/45 expressed from Drosophila melanogaster S2 cells, only one mAb was detected to possess significant TRA when assessed at 350 µg/ml²³. In a follow-up study, the D2-targeting mAb 10D8, which possesses weak TRA, was observed to bind competitively with low-potency mAb 6A10 and more potent mAb 1F10²⁸. Crystal structures of full-length Pfs48/45 bound by 10D8 and two D3-directed Fabs unveiled the 10D8 epitope to an interface opposite from the other two more potent mAbs targeting D3, but it remains unclear how D2-directed mAbs interact with Pfs48/45 more broadly^23,28. Moreover, while we and others have reported that mAbs targeting D1 or D2 can also possess TRA, the structural elucidation of any D1-targeting mAbs has not been previously reported^20,27,29.

In this study, we present functional and structural insights into human mAbs that target D1, D2, and D3 of Pfs48/45. Using protein engineering and cryoEM, we determined the structure of full-length Pfs48/45 bound by Fabs RUPA-58 (D1) and RUPA-154 (D2), and previously described D3-targeting Fabs TB31F and RUPA-44²⁷. We report an extended conformation for full-length Pfs48/45, with a rigid D1-D2 interface but discernible flexibility between D2 and D3. Focused refinement of the D1-D2 segment of Pfs48/45 resulted in a 3.3 Å map and enabled delineation of yet undescribed highly-potent and moderately-potent epitopes on D1 and D2, respectively. Collectively, our study unveils structure-function relationships of transmission-reducing activity by antibodies against Pfs48/45 across all its domains.

Selection of D1-, D2- and D3-targeting human antibodies

We previously characterized a panel of human mAbs²⁷ directed against full-length Pfs48/45 and a potent humanized mAb against D3²⁰. From these, we selected for further characterization a set of mAbs with significant TRA that bind with high affinity to conformational epitopes on different domains of Pfs48/45: D1 (RUPA-58), D2 (RUPA-154), and D3 (RUPA-44 and TB31F) (Extended Data Fig. 1).

mAbs against D1 and D3 showed potent TRA in standard membrane feeding assay (SMFA), with IC₈₀ values of 7.5, 11.0 and 0.86 µg/mL for RUPA-58, RUPA-44 and TB31F, respectively (Fig. 1a). D2-targeting RUPA-154 showed more moderate transmission reducing activity with an IC₈₀ of 103.1 µg/mL (Fig. 1a). All mAbs showed high apparent affinity for female gametes in surface immunofluorescence assays (SIFA), with EC₅₀ values ranging from 0.009 to 0.56 µg/mL (Fig. 1b). Strong binding of RUPA-154 to female gametes (EC₅₀ = 0.14 µg/mL) contrasts its moderate transmission reducing activity and suggests that TRA and binding to female gametes are not absolutely correlated²⁷. Importantly, all four functionally-active antibodies bind non-overlapping epitope bins²⁷ and could therefore be used to assemble a co-complex with full-length Pfs48/45.

Engineering of a full-length Pfs48/45 construct suitable for structural determination

Recombinant expression in HEK293 cells of full-length Pfs48/45 modified to remove its native GPI-anchor (amino-acids (aa) 28–428) was challenging, as it expressed poorly and tended to aggregate (Fig. 1c). Recently, we demonstrated that stabilizing mutations introduced within D3 can improve expression of properly folded D3 when compared to wildtype protein²⁴. To obtain well-behaved full-length Pfs48/45 in sufficient quantity, we introduced these five D3 stabilizing mutations (mAgE2: G397L, H308Y, I402V, S361R and D373S²⁴) in the full-length construct. This approach improved protein yield considerably, but still resulted in heterogeneity. To further ameliorate the construct, we fused the C terminus of the stabilized full-length Pfs48/45 with a 21-aa flexible linker to the heavy chain (HC) of the single chain (sc) fragment antigen-binding (Fab) of TB31F. The resulting construct, Pfs48/45-scTB31F, showed ~ 30% higher yield compared to stabilized Pfs48/45 without TB31F, and contained a lower proportion of misfolded oligomers (Fig. 1c). Moreover, Pfs48/45-scTB31F was recognized with nanomolar or sub-nanomolar binding affinity by RUPA-44, RUPA-154 and RUPA-58 Fabs as measured by bio-layer interferometry (BLI), confirming proper folding of all three domains of Pfs48/45 in this engineered construct (Fig. 1d).

CryoEM structure determination of full-length Pfs48/45

To prepare a complex suitable for cryo-electron microscopy (cryoEM) studies, Pfs48/45-scTB31F was complexed with D1-binding RUPA-58 Fab and D3-binding RUPA-44 Fab. Initial cryoEM analysis of the collected dataset revealed a highly preferred orientation (Extended Data Fig. 2a, Extended Data Fig. 3a-e and Extended Data Fig. 4a, Table 1). Addition of D2-binding RUPA-154 Fab to the quaternary complex to form a quinary complex did not initially improve the distribution of views (Extended Data Fig. 3i). To further enrich the diversity of particle orientations and improve particle alignment rendered challenging by the extended, narrow shape and flexibility of the complex (Extended Data Fig. 3f-j, Extended Data Fig. 4b), we subsequently collected a dataset on the Pfs48/45-scTB31F:RUPA-44:RUPA-58:RUPA-154 complex with the specimen stage tilted at 40°. This approach resulted in significantly improved distribution of views and alignment of particles (Extended Data Fig. 3k-o). Combining particles from the datasets collected at 0° and 40° tilt and subsequent refinement allowed a 3.6 Å resolution map of the Pfs48/45-scTB31F:RUPA-44:RUPA-58:RUPA-154 complex to be obtained (Fig. 2a, Extended Data Fig. 3k-o and Extended Data Fig. 4c, Table 1). The 3.6 Å resolution map of Pfs48/45-scTB31F:RUPA-44:RUPA-58:RUPA-154 allowed fitting of the X-ray crystal structure of D3 bound to RUPA-44 Fab (PDB ID: 7UXL²⁷), as well as TB31F Fab (PDB ID: 6E63²⁰, Fig. 2a, Extended Data Fig. 5). To further enhance the resolution of the less-characterized D1-D2 segment of Pfs48/45, local refinement was performed on Pfs48/45 D1-D2 and RUPA-58 and RUPA-154 variable regions, resulting in a 3.3 Å resolution map for this region of the complex (Fig. 2b-c, Extended Data Fig. 3p-q and Extended Data Fig. 4d, Table 1). The locally refined 3.3 Å resolution map of the Pfs48/45-scTB31F:RUPA-44:RUPA-58:RUPA-154 complex was of sufficient quality to unambiguously build a molecular model for the majority of the RUPA-58 and RUPA-154 variable regions and ~ 80% of Pfs48/45 D1 and D2 residues (Fig. 2c). Together, the cryoEM data allows for detailed molecular insights into the Pfs48/45 structure and the epitopes it contains across its three domains.

Table 1

CryoEM data collection and refinement statistics.
Data Collection	Pfs48/45-scTB31F:RUPA-44:RUPA-58:RUPA-154
Electron microscope	Titan Krios G3
Camera	Falcon 4
Voltage (kV)	300
Nominal magnification	75,000
Calibrated physical pixel size (Å)	1.03
Total exposure (e- /Å²)	48.69
Number of frames	30
Image Processing
Motion correction software	cryoSPARC v3
CTF estimation software	cryoSPARC v3
Particle selection software	cryoSPARC v3
3D map classification and refinement software	cryoSPARC v3
Micrographs used	5964
Particles selected	2.824,624
Global resolution (Å)	3.3
Particles contributing to final maps	168,235
Model Building
Modeling software	Coot, phenix.real_space_refine
Number of residues built	677
RMS (bonds)	0.003
RMS (angles)	0.931
Ramachandran outliers (%)	0.5
Ramachandran allowed (%)	5.4
Ramachandran favoured (%)	94.1
Rotamer outliers (%)	0.8
Clashscore	7
MolProbity score	1.8
EMRinger Score	1.6

Flexibility within full-length Pfs48/45

While D1 and D2 appear close together and form a well-defined interface, D3 is considerably more distanced and only connected to D2 via a five amino acid flexible linker (Fig. 3a). 3D variability analysis of the Pfs48/45-scTB31F:RUPA-44:RUPA-58:RUPA-154 map reveals continuous rotation of D3 vs D1 and D2 by ~ 10^o (Fig. 3b, Supplementary Video 1). Interestingly, comparison of the quaternary complex Pfs48/45-scTB31F:RUPA-44:RUPA-58 revealed a larger angle between D2 and D3 compared to the quinary complex Pfs48/45-scTB31F:RUPA-44:RUPA-58:RUPA-154 (~ 120^o and ~ 80^o, respectively), even though the angle between D1 and D2 remained unchanged (Fig. 3c, left and middle panels). In the previously reported crystal structure of Pfs48/45 in complex with Fabs 10D8 and 85RF45.1²⁸, the three domains of Pfs48/45 adopted a compact, disc-like conformation, with an angle between D2 and D3 of ~ 65^o (Fig. 3c, right panels). Alignment of the Pfs48/45-scTB31F:RUPA-44:RUPA-58:RUPA-154 and Pfs48/45:10D8:85RF45.1 models reveals an almost identical fold of individual domains, with the major difference coming from the position of D3, which is displaced by ~ 40 Å between models (Fig. 3c and d). Combined, these structural data reveal a large range of motion possible at the D2-D3 hinge intrinsic to the full-length Pfs48/45 structure, and considerably beyond what had been previously described²⁸.

Delineation of the D2 epitope recognized by mAb RUPA-154

RUPA-154 interacts predominantly with three adjacent D2 β-strands (212–217, 271–275, and 282–288) and several residues of loop 226–247. RUPA-154 contacts with Pfs48/45 are mediated mainly through lambda chain complementarity-determining region (LCDR) 1 and 2 and heavy chain complementarity-determining region 3 (HCDR3), with a buried surface area (BSA) of 547 Å² and 352 Å² by the LC and HC, respectively (Fig. 4a). Three HCDR3 tyrosine residues (H.Y100B, H.Y100C and H.Y100D) contribute 25% of the overall BSA, and are optimally positioned across the D2 surface to form multiple contacts with Pfs48/45 residues Ser210, Val212, Lys239, Glu240 and Asn270 (Fig. 4b). Despite a significant number of somatic hypermutations (SHMs) in RUPA-154 (16 in the HC and 5 in the LC, Extended Data Fig. 6a), only three of these residues appear to be involved in antigen recognition (L.26D, L.51N and L.52I). The L.52I SHM results in the optimal positioning of the Ile side chain to form contacts with the sidechains of Pfs48/45 Asn211, Val212, Asn281, and Lys283 (Fig. 4b).

A superimposition of RUPA-154 to the previously-determined structure of 10D8 bound to Pfs48/45²⁸ reveals that they bind adjacent epitopes (Extended Data Fig. 7). Pfs48/45 residues Val209, Ser210, Glu214, Val285, and Lys287 contact both antibodies, and there is a significant overlap between 10D8 and the heavy chain of RUPA-154 (Extended Data Fig. 7). This result is consistent with RUPA-154 and 10D8 competing with each other in epitope binning assays (Fig. 4c). An overlay of 10D8 on the Pfs48/45-scTB31F:RUPA-44:RUPA-58:RUPA-154 cryo-EM map revealed that 10D8 binding would be occluded when Pfs48/45 is in this conformation due to steric clashes with D3 (Extended Data Fig. 7). Differences in accessibility across conformations could be one explanation for why 10D8 is less potent than RUPA-154 despite binding to a similar region of D2.

We assessed the prevalence of non-synonymous single nucleotide polymorphisms (SNPs) occurring in Pfs48/45, using genomic sequence data from 7,113 Pf field isolates collected from 73 different locations in Africa, Asia, America, and Oceania (MalariaGen³⁰) (Extended Data Table 2). We identified twelve non-synonymous SNPs in D2 across Pf field isolates (Extended Data Table 2). Of these polymorphisms, only two are present in the RUPA-154 epitope, V212I (f = 0.02%) and K239T (f = 0.03%) (Extended Data Fig. 8a). Binding kinetics experiments with RUPA-154 Fab show that while it retains nanomolar binding affinity to the V212I mutant, the other low frequency SNP, K239T, does prevent RUPA-154 binding (Extended Data Fig. 8c-f). The drastic impact of this mutation is likely due to K239 forming a salt-bridge at the antibody-antigen interface and having a high BSA (> 100 A²). Together, these results indicate that the RUPA-154 epitope is highly conserved, with rare polymorphisms that can affect binding by an inhibitory mAb.

To further elucidate the human humoral response to D2, we selected for biophysical studies several D2-targeting antibodies with varying potencies. In addition to RUPA-154 (IC₈₀ = 103.1 µg/mL) and 10D8 (TRA of ~ 56% at 375 µg/mL)¹¹, we investigated the highest potency D2 antibody, RUPA-160, and two other low potency antibodies, RUPA-111 and RUPA-43. RUPA-160 has a TRA of ~ 80% at 10 µg/mL, while RUPA-111 and RUPA-43 did not demonstrate transmission-reducing activity at 100 µg/mL²⁷. Binding kinetics experiments with a recombinant D1-D2 construct revealed that these antibodies bind with nanomolar affinities ranging from 0.5 nM to 7.3 nM (Fig. 4d). These results indicate that binding affinity to the recombinant antigen is not directly correlated with transmission-reducing activity. Epitope binning assays revealed that all selected antibodies compete except 10D8 and RUPA-111 (Fig. 4c). These two mAbs compete with RUPA-154, RUPA-160, and RUPA-43, but not with each other (Fig. 4c). Given that antibodies that range from low to moderate TRA bind to overlapping positions on D2, we propose that potency variations for this epitope class may be due to slight differences in angles of approach or contacts with specific Pfs48/45 residues.

Delineation of the D1 epitope recognized by potent mAb RUPA-58

An initial model for the RUPA-58 variable domain was created by manually docking the 3.09 Å-resolution structure of unliganded RUPA-58 solved by X-ray crystallography (Extended Data Table 1) into the Pfs48/45-scTB31F:RUPA-44:RUPA-58:RUPA-154 locally refined cryo-EM map. The potent RUPA-58 epitope includes Pfs48/45 D1 loop 86–92, β-strand 93–97, as well as loops 99–115 and 131–139. Kappa chain complementarity-determining region (KCDR) 1 and 3 and HCDR1-3 form extensive interactions with D1, with a total BSA of 992 Å² (532 and 461 Å² from HC and KC, respectively) (Fig. 5a). Interestingly, the HCDR3 of this antibody contains a disulfide bond between cysteines H.C97 and H.C100B, which helps optimally position the HCDR3 to interact with loop 86–92 of D1. H.Y100C is central to the HCDR3-D1 interaction contributing 197 Å² of BSA and forming numerous contacts with residues at the base of the 86–92 D1 loop, including the hydrophobic side chains of Ile86, Phe95, Ile96 and Ile97 (Fig. 5b). Comparison of RUPA-58 to its inferred germline precursor, IGHV1-2/IGKV1-39, indicates the presence of 11 SHMs in the HC and 10 in the KC (Extended Data Fig. 6b). Interestingly, only one SHM in the HC forms direct contact with the antigen (H.N58K). From six somatically hypermutated residues in the KC interacting with the antigen, the side chains of K.T31 and K.S50 are optimally positioned to interact with Glu100 and contribute 43 Å² and 25 Å² of BSA to the interaction, respectively (Fig. 5b).

18 non-synonymous SNPs were identified in D1, possessing exceedingly rare allele frequencies which ranged from 0.007–0.044%. Of them, only S90R, K101T, and N138I occur at positions that are involved in RUPA-58 binding (Extended Data Fig. 8b). To determine if these mutations could impact RUPA-58 binding, we conducted binding kinetics experiments with recombinant D1-D2 constructs containing these SNPs. These experiments revealed that RUPA-58 retains nanomolar affinity to all three constructs, with K_D values of 1.1 nM, 1.4 nM, and 2.8 nM for D1D2-K101T, D1D2-S90R, and D1D2-N138I, respectively (Extended Data Fig. 8g-j). One minor but notable effect was a higher off-rate for RUPA-58 binding to the N138I mutant, likely due to a loss of hydrogen bonding with the HCDR3 residue Y100C. This analysis demonstrates that the RUPA-58 epitope is well conserved across different strains of Pf and that any SNPs documented to date would minimally impact potent recognition mediated by this mAb.

Molecular description of epitopes on Pfs48/45 D1 targeted by human mAbs

Several other D1-targeting antibodies have demonstrated TRAs of over 80% at 10 µg/mL²⁷ but have yet to have their epitopes mapped at molecular level. To continue our delineation of antibody-mediated inhibition mediated by recognition of Pfs48/45 D1, we selected additional potent D1 antibodies to analyze: RUPA-119, RUPA-46, RUPA-116, RUPA-75, RUPA-71, RUPA-94, and RUPA-155. We also picked a range of low-potency antibodies that did not demonstrate TRA at 100 µg/mL, including RUPA-122, RUPA-114, RUPA-77, and RUPA-78²⁷. To determine if there was any variation in binding affinities between these different classes of mAbs, we measured the K_Ds of each Fab using BLI. These results revealed a wide spread of binding affinities, ranging from < 1 pM to > 100 nM (Fig. 5c). K_D values did not directly associate with potency; for example, the non-inhibitory RUPA-122 mAb had significantly higher affinity to the recombinant antigen than the potently inhibitory RUPA-46 mAb.

Through a series of competition assays, we found that RUPA-58, for which we obtained the cryoEM structure, competes directly with RUPA-94, outlining a first epitope bin on Pfs48/45 D1 (Fig. 5d). RUPA-119, RUPA-46, RUPA-116, RUPA-75, RUPA-71, and RUPA-155 all compete with each other, revealing a second well-defined bin associated with high potency mAbs (Fig. 5d). Given that all of these antibodies in the second bin, except RUPA-46, bind to linear epitopes²⁷, these mAbs likely bind to a similar stretch of Pfs48/45 residues. Additionally, RUPA-46 and RUPA-71 compete with antibodies in this bin as well as RUPA-58, signifying that the second potent D1 epitope bin is likely adjacent to the RUPA-58 epitope (Fig. 5d). The lower potency antibodies, RUPA-114 and RUPA-122, did not extensively compete with any other antibody, suggesting a potential third epitope on Pfs48/45 D1 (Fig. 5d).

To identify the highly potent linear bin 2 epitope associated with the high-potency D1 antibodies, we carried out negative stain electron-microscopy (nsEM) experiments using RUPA-71 (Fig. 5e; Extended Data Fig. 9). A 3D reconstruction of the RUPA-71:RUPA-154:D1D2 complex revealed that the RUPA-71 binding site maps to the region of D1D2 containing residues 85–92 and 131–139 (Fig. 5e). The epitope of RUPA-71 was further delineated using hydrogen/deuterium exchange mass spectrometry (HDX-MS). A comparison of deuterium uptake in the D1D2-RUPA-71 complex and unliganded D1D2 revealed significant reductions in deuterium uptake in the “45–51”, “85–102”, “124–141”, “188–195”, and “196–208” peptides upon RUPA-71 binding, with “124–141” showing the largest change (Fig. 5e; Extended Data Fig. 10). A comparison of RUPA-71 with the structure of Pfs48/45-bound RUPA-58 indicated that these antibodies bind to near-by, partially overlapping epitopes, consistent with our BLI competition data (Extended Data Fig. 11; Fig. 5d). BLI experiments with the point mutant D1D2 N131Q construct further validated the nsEM and HDX-MS epitope mapping, wherein the N131Q mutation resulted in a faster off-rate for RUPA-71 (Fig. 5e, Extended Data Fig. 12). This effect was even more pronounced for RUPA-116, RUPA-119, RUPA-75, RUPA-46, and RUPA-155, indicating that the binding sites of these antibodies that compete with one another are likely in proximity of N131 (Extended Data Fig. 12).

Interestingly, affinity measurements to the Plasmodium vivax (Pv) orthologue of D1D2 revealed that RUPA-71 is capable of cross-reactive binding across the two Plasmodium species, binding to PvD1D2 with an affinity of 2.2 µM as a Fab and with an avidity-boosted apparent affinity of 10.5 nM as an IgG (Fig. 5f). A comparison of residues showcased good conservation between the PvD1D2 and PfD1D2 proteins in the putative RUPA-71 binding site (Fig. 5g, Extended Data Fig. 13). Together, these results highlight a high degree of conservation for the RUPA-71 epitope, not only across Pf field isolates, but also within the Pv ortholog.

Pfs48/45 is a promising TBV candidate and a target for mAbs with potent TRA, including mAb TB31F, which is to-date the transmission-blocking mAb with the highest potency and has demonstrated to be well-tolerated in humans¹⁹. However, understanding the global molecular architecture of Pfs48/45 has proven challenging due to its inherent instability³¹. Recent progress has been made to achieve better antigen expression by introducing stabilizing mutations to D3^24,25 or by expression in Drosophila S2 cells^23,28. To enable the molecular characterization of inhibitory epitopes outside of D3, here we designed a full-length Pfs48/45 construct that incorporates previously described stabilizing mutations within D3 and combines these with the genetic fusion to a single-chain Fab fragment of the potent D3 targeting mAb, TB31F. Resulting from this construct engineering, we derived a high-resolution cryoEM structure of an antibody-antigen complex formed by full-length Pfs48/45 bound simultaneously by four inhibitory antibodies directed to all three Pfs48/45 domains.

Previously reported crystal structures of full-length Pfs48/45 bound by either D2- or D3-directed antibodies demonstrated that the three domains of Pfs48/45 could adopt a disc-like conformation that is highly similar to a model predicted by Alphafold²⁸ (Fig. 3c and Fig. 6a). In contrast, we observed that the position of D3 relative to D2 is significantly different in our cryoEM structures compared to the crystal structures of full-length Pfs48/45, inferring the capacity for around 40° range in motion in the hinge region between D2 and D3 (Fig. 6b-c). In addition to intrinsic flexibility, differences in the interdomain dispositions may be related to experimental conditions from which the structures were determined (i.e. crystal or vitreous ice) or may also be influenced by the binding of antibodies at specific interfaces. Furthermore, while D3 possesses an overall electronegative surface potential, both D1 and D2 possess a partially electropositive surface (Extended Data Fig. 14). The electronegative plasma membrane might be an additional factor that influences Pfs48/45 domain orientation on the parasite surface, which is absent in the present molecular description. Future studies examining the conformational dynamics of Pfs48/45 on the parasite surface and the interplay with native binding partners such as Pfs230 may clarify the functional determinants and implications of this flexibility.

In previous work, we reported the functional and genetic details of naturally acquired human mAbs targeting D1, D2, and D3 of Pfs48/45 derived from two donors²⁷. Notably, the D1-directed mAbs were found to possess TRA that was comparable to those directed to D3, while D2-directed mAbs possessed markedly lower TRA compared to D3 or D1, as previously described^21,23,28. Here, we extend the functional and biophysical characterization of several of these mAbs against D1 and D2. Our data revealed that binding affinity to recombinant antigens or binding to female gametes does not correlate directly with transmission reducing activity, indicating that the molecular characteristics of epitopes targeted may be a better indicator of anti-Pfs48/45 antibody potency.

Local refinement of the D1D2:RUPA-58:RUPA-154 region in the cryoEM map produced a 3.3 Å reconstruction that revealed the epitopes of RUPA-58 and RUPA-154 at atomic level. This data provides molecular insights into human antibody-mediated inhibition of gametes by binding outside the well-established potent epitopes on Pfs48/45 D3. All D2-directed antibodies assessed in this study were found to compete with RUPA-154, indicating that they all bind to overlapping epitopes. Prior to the present study, the only structurally characterized D2-directed mAb was the murine antibody 10D8, which like RUPA-154 also possesses low TRA. Although RUPA-154 and murine antibody 10D8 compete for binding to Pfs48/45, a comparison of their structures revealed that their epitopes and angles of approach are slightly different, and might or might not be compatible with binding Pfs48/45 in different domain conformations. From this observation emerges the hypothesis that certain Pfs48/45-D2 epitopes may be transiently occluded by D3, which could impart lower levels of inhibitory potency. However, since RUPA-154 binds strongly to female gametes in SIFA, it is also possible that modest potency of this antibody stems from reasons different than low accessibility of its epitope. An alternate hypothesis is that cognate binding partners, such as Pfs230³² or PfCCp^33,34, may compete for access to Pfs48/45 interfaces. Indeed, this phenomenon has also been proposed to explain why antibodies to the blood-stage vaccine antigen, PfRH5, which target the CyRPA interface, lack growth inhibitory activity³⁵.

Our cryoEM work also provides molecular insights into Pfs48/45 D1-directed human antibodies that can possess comparably high TRA as those directed to D3. RUPA-58 competed with RUPA-94, RUPA-78, and RUPA-77, demonstrating a range of TRAs within this first epitope bin. A nsEM 3D reconstruction of the RUPA-71-RUPA-154-D1D2 complex combined with binding competition assays revealed the putative location of a second inhibitory binding site on D1, and the location of this potent epitope was further delineated by HDX-MS mapping experiments. In addition, mutagenesis experiments supported this binding site, showing that an N131Q mutation negatively impacts the binding of all the antibodies in this epitope bin. Interestingly, the N131Q mutation eliminated a putative N-linked glycosylation sequon and consequently, removed a N-linked glycan at this position when the construct is expressed in HEK293 cells, which was the case in our studies. Presence and extent of N-glycosylation in Plasmodium is still heavily debated³⁶. Based on genomic analysis and metabolic labelling experiments, it appears that parasites may be capable of producing short N-glycans of no more than two GlcNAc moieties³⁷. Given that the human antibodies studied here derive from natural exposure, are highly potent and can bind to the gamete surface, this data suggests that glycan at N131 may be naturally occurring in Pf, although an alternate hypothesis is that mutation of this glycan distorts the local conformation of this epitope bin. Further studies characterizing possible post-translational modifications on Pfs48/45 derived directly from the pathogen will be necessary to fully explore the nature of this highly potent epitope. Recent advances in potent chromatographic and mass spectrometric techniques have made investigation of carbohydrate repertoires more feasible, as shown by significant progress made in characterization of heavily glycosylated HIV-1 envelope spike protein (Env)³⁸.

Intriguingly, BLI experiments demonstrated that RUPA-71 can bind both Pf D1D2 and Pv D1D2, supporting the potential for interventions seeking cross-species Plasmodium transmission-blocking activity against the two major species circulating worldwide³⁹. Overall, our data support incorporation of highly conserved Pfs48/45 D1 epitopes, now resolved at molecular level, into next-generation transmission-blocking immunogen designs. Human-derived D1 directed mAbs may also become valuable additions to TB31F mAb-based interventions and put additional pressure on the parasite at the bottleneck of transmission by conveying further resilience against any mutational escape potential. Future studies on larger cohorts of mAbs will continue to shed light on relationships between Pfs48/45 epitopes and antibody potency. Moreover, the application of novel techniques like cryoEMPEM⁴⁰ holds promise for future structural studies to provide molecular insights into epitope recognition by polyclonal antibody responses elicited by infection or vaccination trials, and should be considered as part of the evaluation of immunological outcomes.

Domain-specificity by ELISA

The domain specificity of mAbs was tested in ELISA as previously described²⁷. Nunc MaxiSorpTM 96-wells plates (ThermoFisher) were coated with 0.5 µg/mL recombinant full-length Pfs48/45 and fragments containing D1-2, D2-3 and D3, overnight at 4˚C. Plates were blocked with 5% skimmed milk in PBS for 1 h at room temperature, washed and then incubated with 10 µg/mL mAb in 1% milk/PBST for 3 h at room temperature. Plates were washed and then incubated with 1:60,000 Goat anti-Human IgG/HRP-conjugated antibody (Pierce, Cat. No.31412) in 1% milk/PBST for 1 h. Plates were washed again, 100 µl 3,3',5,5'-Tetramethylbenzidine (TMB) was added and after approximately 20 min the reaction was stopped with 0.2M H₂SO₄. Absorbance was measured at 450 nm. mAbs were considered positive if the absorbance was higher than the mean absorbance plus three standard deviations of six negative mAbs.

Surface immunofluorescence assay (SIFA)

N-acetyl-glucosamine-treated 16-day-old Pf NF54 gametocytes were harvested by centrifugation for 10 min at 2,000 x g. Pelleted gametocytes were resuspended in fetal bovine serum (FBS) and activated on a roller bank for 45 min at room temperature. The cells were then pelleted by centrifugation for 10 min at 2,000 x g at 4˚C, resuspended in PBS, loaded on an 11% Accudenz (Accurate Chemical) cushion, followed by centrifugation for 30 min at 7,000 x g at 4˚C. The top layer, containing activated female gametes, was collected and gametes were then washed with PBS prior to incubation with mAbs. mAbs diluted in PBS + 2% FBS + 0.02 % sodium azide were mixed with 50,000 gametes and incubated for 1 h at room temperature. Thereafter, gametes were washed three times with PBS, followed by 30 min incubation at room temperature with 1:200 Alexa Fluor^TM 488 anti-human IgG (Invitrogen, cat no. A-11013) and 1:1,000 eBioscience™ Fixable Viability Dye eFluor™ 780 (Invitrogen, Cat. No. 65-0865-14). Cells were finally washed three times with PBS and resuspended in PBS. mAb binding to gametes was assessed by measuring fluorescence of at least 2,000 single live gametes on a Gallios^TM 10-color system (Beckman Coulter) followed by analyses in FlowJo (BD Biosciences).

Standard membrane feeding assay (SMFA)

SMFA experiments used cultured Pf NF54 gametocytes and oocyst counts as readout, as previously described⁴¹. mAbs were diluted in FBS and mixed with cultured gametocytes and normal human serum as a source of active complement. Given mAb concentrations are final concentrations in the total blood meal volume. Blood meals were fed to A. stephensi mosquitoes (Nijmegen colony). For each condition 20 fully-fed mosquitoes were analyzed. Transmission-reducing activity was calculated as the reduction in oocysts compared to a negative control. IC₈₀ values were calculated by linear regression analysis on the log₁₀ transformed ratio of mean oocyst count in control and test sample, and square root of antibody concentration⁴². 95% confidence intervals for the IC₈₀ values were calculated with the delta method¹⁹. SMFA data analyses were done in R (version 4.1.2).

Expression and purification of Fabs

Variable light (VL) and heavy (VH) chains of Fabs used in these studies were gene synthesized and cloned (GeneArt) into custom pcDNA3.4 expression vectors located immediately upstream of constant domains. Fab heavy chain and Fab light chain plasmids were co-transfected at a 2:1 ratio into FreeStyle 293-F or 293-S cells (Thermo Fisher Scientific) at a cell density of 0.8 × 10⁶ cells/mL for transient expression using FectoPRO DNA transfection reagent (Polyplus). Cells were cultured in GIBCO FreeStyle 293 Expression Medium for 7 days, and supernatants were isolated by centrifugation and filtered through a 0.22 μm membrane. Secreted Fabs were purified using affinity chromatography with either a HiTrap KappaSelect or LambdaSelect column (Cytiva) with 1X PBS and 100 mM glycine, pH 2.2 as wash and elution buffers, respectively. Fabs were further purified by cation-exchange chromatography using a MonoS column (Cytiva) in a buffer of 20 mM NaOAc, pH 5.6 across a 0-1 M KCl gradient for elution. Fabs expressed in 293-S cells underwent subsequent treatment with EndoH.

Expression and purification Pfs48/45 proteins

A full-length Pfs48/45 construct (aa 28-428) integrating stabilizing mutations to D3 (G397L, H308Y, I402V, S361R and D373S²⁴), was fused at the C terminus using a 7xGGS linker to the HC portion of a single chain Fab of TB31F with a 21 aa long flexible linker. The scFab of TB31F contained a 14xGGGGS linker between the HC and lambda chain (LC). A Pfs48/45 construct containing only domains 1 and 2 (aa 27-294) was also generated containing the native sequences or any SNPs located with the RUPA-58 or RUPA-154 epitopes to interrogate binding kinetics. To all constructs, a signal peptide sequence was added to the N terminus and a 6xHis tag was added to the C terminus, followed by gene synthesis and cloning into a pcDNA3.4 expression vector (GeneArt). Stabilized Pfs48/45 linked scTB31F was recombinantly expressed in 293-F or 293-S cells and purified with a HisTrap FF column (GE Healthcare), using TBS pH 7.0 with a linear elution gradient of imidazole up to 500 mM, and further purified using size exclusion chromatography with a Superdex 200 Increase column (GE Healthcare) in TBS pH 7.0. The Pfs48/45-scTB31F:RUPA-44:RUPA-58 and Pfs48/45-scTB31F:RUPA-44:RUPA-58:RUPA-154 complexes were formed by incubating proteins together for 30 min to overnight at 4^oC and subsequently purifying the complex using size exclusion chromatography with a Superose 6 Increase column (GE Healthcare) in TBS pH 7.0.

Biolayer interferometry (BLI) binding studies

Direct binding kinetics measurements were conducted at 25°C using an Octet RED96 instrument (Sartorius ForteBio). Fabs and purified recombinant Pfs48/45 were diluted in kinetics buffer (PBS, pH 7.4, 0.01% (w/v) BSA, and 0.002% (v/v) Tween-20) as previously described²⁷. Binding kinetics parameters were determined using Ni-NTA biosensors loaded with 15 μg/mL of purified His-tagged Pfs48/45 protein followed by a 30 s baseline and an association phase in Fab protein serially diluted from 500 nM to 16 nM. Biosensors were then dipped into wells containing kinetics buffer for a dissociation step. Data analysis was performed using the Octet software (Sartorius ForteBio, version 8.2.0.7) and the sensograms were fit using a 1:1 binding model. Binding competition assays were performed by loading 15 μg/mL of 6xHis-tagged antigen to reach a BLI signal response of 1.0 nm. Following by a 30 s baseline step, the antigen-loaded biosensors were dipped into wells containing the first antibody (Fab 1) at 50 μg/mL for 300 s and then sensors carrying Fab 1-antigen complexes were dipped into wells containing the second antibody (Fab 2) at 50 μg/mL for an additional 300 s. Data analysis was performed using the Octet software (Sartorius ForteBio, version 8.2.0.7) and the binding competition was quantified as percent ratio of each antibody’s BLI signal at each stage of the two-step binding sequence (i.e. Fab2/Fab1 measured in nm).

Single nucleotide polymorphism detection

Single nucleotide polymorphisms were obtained from the MalariaGEN Catalogue of Genetic Variation in Pf version 6.0³⁰. Genotype calls for chromosome 13 (Pf3D7_13_v3) were downloaded (ftp://ngs.sanger.ac.uk/production/malaria/pfcommunityproject/Pf6/Pf_6_vcf/Pf_60_public_Pf3D7_13_v3.final.vcf.gz) and Bcftools⁴³ was used to subset calls between nucleotide positions 1,875,452 and 1,878,087(-) that coincide with Pfs48/45 and calculate the allele frequency of all polymorphism occurrences. SNPs calls below MalariaGEN’s quality filter (Low_VQSLOD) and all silent mutations were removed.

CryoEM data collection and image processing

The Pfs48/45-scTB31F:RUPA-44:RUPA-58 and Pfs48/45-scTB31F:RUPA-44:RUPA-58:RUPA-154 complexes were concentrated to 0.7 mg/mL and 3.0 µL of the sample was deposited on homemade holey gold grids⁴⁴, which were glow-discharged in air for 15 s before use. All samples were blotted for 3.0 s, and subsequently plunge-frozen in liquid ethane using a Leica EM GP2 Automatic Plunge Freezer (maintained at 4°C and 100% humidity). Data collection was performed on a Thermo Fisher Scientific Titan Krios G3 operated at 300 kV with a Falcon 4i camera automated with the EPU software. A nominal magnification of 75,000× (calibrated pixel size of 1.03 Å) and defocus range between 0.8 and 1.5 μm were used for data collection.

Exposures were collected for 8.8 s (Pfs48/45-scTB31F:RUPA-44:RUPA-58:RUPA-154 complex) or 8.3 s (Pfs48/45-scTB31F:RUPA-44:RUPA-58 complex) as movies of 30 frames with a camera exposure rate of ∼6 e⁻ per pixel per second, and total exposure of ~49 electrons/Å². A total of 6,850 raw movies were obtained for the Pfs48/45-scTB31F:RUPA-44:RUPA-58 complex and 4964 for the Pfs48/45-scTB31F:RUPA-44:RUPA-58:RUPA-154 complex (1529 raw movies were obtained with no stage tilt and 3435 with 40^o stage tilt).

Image processing was carried out in cryoSPARC v3⁴⁵. Initial specimen movement correction, exposure weighting, and CTF parameters estimation were done using patch-based algorithms. Micrographs were sorted based on CTF fit resolution, and only micrographs with score lower than 5.0 Å were accepted for further processing. Manual particle selection was performed on 30 micrographs to create templates for template-based picking. For the Pfs48/45-scTB31F:RUPA-44:RUPA-58 complex, 3,904,349 particle images were selected by template picking and individual particle images were corrected for beam-induced motion with the local motion algorithm⁴⁶. Particle images of intact Pfs48/45-scTB31F:RUPA-44:RUPA-58 complex were separated from dissociated components via 2D classification and several rounds of heterogeneous refinement. 133,106 particle images were selected for non-uniform refinement⁴⁷ with no symmetry applied, which resulted in a 4.3 Å resolution map of the Pfs48/45-scTB31F:RUPA-44:RUPA-58 complex estimated from the gold-standard refinement with correction of the Fourier shell correlation (FSC) for masking effects.

For the Pfs48/45-scTB31F:RUPA-44:RUPA-58:RUPA-154 complex, 2,824,624 particle images were selected by template picking and individual particle images were corrected for beam-induced motion with the local motion algorithm⁴⁶. Particle images of intact Pfs48/45-scTB31F:RUPA-44:RUPA-58:RUPA-154 complex were separated from dissociated components via one round of 2D classification and several rounds of heterogeneous refinement. 504,058 particle images were selected for non-uniform refinement⁴⁷ with no symmetry applied, which resulted in a 3.6 Å resolution map of the Pfs48/45-scTB31F:RUPA-44:RUPA-58:RUPA-154 complex. Local refinement was performed on all particle images with a mask including domains D1 and D2 domains of Pfs48/45 and variable regions of RUPA-58 and RUPA-154 Fabs, which resulted in 3.3 Å resolution map of this region. To visualize flexibility in the Pfs48/45-scTB31F:RUPA-44:RUPA-58:RUPA-154 cryoEM map, 3D variability analysis was performed⁴⁸.

An initial model for Pfs48/45:RUPA-58:RUPA-154 was created by manually docking crystal structures of Pfs48/45 D1 and 2 (PDB ID:7ZXF²⁸), the variable domain of RUPA-58 Fab derived from the crystal structure (Extended Data Table 1)²⁷ and the predicted variable domain of RUPA-154 Fab⁴⁹ into the cryoEM map with UCSF Chimera⁵⁰, followed by manual building with Coot⁵¹. All models were refined using phenix.real_space_refine⁵² with secondary structure and geometry restraints. The final models were evaluated by MolProbity⁵³. The figures were prepared with UCSF Chimera⁵⁰ and UCSF ChimeraX⁵⁴. Interactions in the Pfs48/45:RUPA-58:RUPA-154 model were identified by PDBePISA⁵⁵. Electrostatic surface potentials were calculated using the Adaptive Poisson-Boltzmann Solver (APBS) plug-in in PyMOL⁵⁶ and images were generated using PyMOL (The PyMOL Molecular Graphics System, v2.3.4, Schrödinger, LLC)⁵⁷.

X-ray crystallography

A sample of recombinant RUPA-58 Fab at 10 mg/mL in TBS pH 7.0 was added to sitting-drop crystallization trays and mixed with reservoir solution at a 1:1 ratio. RUPA-58 crystals grew in 0.2 M sodium acetate, 30% PEG 8000, 0.1 M sodium cacodylate, pH 6.5, 20% ethylene glycol. Data was collected at the 23-ID-D beamline at the Advanced Photon Source (APS). Data processing was done using XDS⁵⁸ and Xprep⁵⁹, while Phaser⁶⁰ was used for molecular replacement using a Fab from PDB: 4YK4 as a search model⁶¹. The crystal structure was then built and refined using phenix.refine⁶² and Coot⁵¹ accessed through SBGrid⁵⁸.

Negative-stain electron microscopy and processing

Purified RUPA-71 Fab, RUPA-154 Fab, and Pfs48/45-D1D2 were complexed at a ratio of 1.5:1.5:1, respectively. Intact complex was purified from excess Fabs using a Superdex 200 column in TBS pH 7.0. 5 μg/mL of complex was applied onto homemade carbon film-coated grids (previously glow-discharged in air for 15 s) and stained with 2% uranyl formate. Micrographs were collected using a Hitachi HT7800 transmission electron microscope operating at 120 kV, 80,000x magnification with a pixel size of 1.83 Å/pix. Image processing, particle extraction, 2D classes, and 3D reconstruction were done using Relion 3.1⁶³ and cryoSPARC v3⁴⁵.

HDX-MS

Protein samples were EndoH treated, further purified as described above, then concentrated to 100 μM. Hydrogen/deuterium exchange was initiated by mixing 2.5 μL of protein with 37.5 μL of D₂O buffer (10 mM phosphate in D₂O, pD 7.0), and incubated for various timepoints (10 s, 100 s, 1,000 s, and 10,000 s) at 4°C. The exchange was quenched by adding 40 μL of chilled (4°C) quench buffer (100 mM phosphate, pH 2.5, 2 M urea, 1 M TCEP) and snap frozen on dry ice at the indicated time points. For non-deuterated (ND) samples, 2.5 μL of purified protein was mixed with 37.5 μL of H₂O buffer (10 mM phosphate in H₂O, pH 7.0), and quenched and frozen in the same manner. Quenched samples were thawed and mixed with pepsin in 1:1 (w/w) ratio for 5 min for offline digestion and passed through an immobilized pepsin column (2.1 × 30 mm; Waters, Milford, MA, USA) at a flow rate of 100 μl/min with 0.1% formic acid and 2% acetonitrile in H₂O at 20°C for online peptic digestion. Peptide fragments were subsequently collected on a C18 VanGuard trap column (2.1 mm × 5 mm; Waters, Milford, MA, USA) for desalting with 0.1% formic acid and 2% acetonitrile in H₂O and then separated by ultra-pressure liquid chromatography using an Acquity UPLC C18 column (1.7 μm, 1.0 × 100 mm; Waters) at a flow rate of 40 μl/min with an acetonitrile gradient from 5% B to 90% B over 10 min. The mobile phase A consisted of 0.1% formic acid in H₂O, and solvent B consisted of 0.1% formic acid in acetonitrile. To minimize the back-exchange of deuterium to hydrogen, the entire system was maintained at pH 2.5 and 0°C (except pepsin digestion was performed at 20°C) during the analysis. Mass spectra were analyzed by SYNAPT G2-Si quadrupole-time of flight (Q-TOF) equipped with a standard electrospray ionization (ESI) source in MSE mode (Waters) with positive ion mode. The capillary, cone, and extraction cone voltages were set to 3 kV, 30 V, and 4 V, respectively. Source and desolvation temperatures were set to 80°C and 150°C, respectively. Trap and transfer collision energies were set to 6 V, and the trap gas flow was set to 2.0 mL/min. Sodium iodide solution (2 µg/µL) was utilized to calibrate the mass spectrometer. [Glu1]-Fibrinopeptide B solution (200 fg/µL) in MeOH:water (50:50 (v/v) + 1% acetic acid) was utilized for the lock-mass correction of the spectra and the ions at mass-to-charge ration (m/z) 785.8427 with a mass window of ±0.5 Da were monitored at scan time 0.1 s. Ar gas was used for collision induced dissociation (CID). For HDMSE and UDMSE measurements, the instrument was operated in ion mobility mode. The T-wave was operated with a wave height of 40 V and a wave velocity ramp from 850 to 350 m/s.

Peptide identification and HDX-MS data processing

Mass spectra were acquired in the range of m/z 50–2000 for 10 min. Methionine oxidation and delta mass of N-Acetylglucosamine (GlcNAc) were assigned as variable modification to identify peptic peptides from the non-deuterated samples using ProteinLynx Global Server 3.0.3 (Waters). Protein identification criteria were set as the detection of at least one fragment per protein, three fragments per peptide, and at least one peptide per protein. PLGS search results and deuterium exchange measurements were imported into DynamX ver. 3.0 (Waters, Manchester, UK). A minimum of 0.03 products per amino acid, and a precursor mass error of less than 25 ppm as defined as the settings in DynamX and the centroid of isotopic distributions in deuterium uptake were analyzed curated as required. Because proteins aggregated in the full deuteration condition, the analyses are represented as the comparison between different states of protein without back exchange correction. All the data were derived from at least three independent experiments. The significant threshold was set as the 95% confidence interval, calculated from the pooled standard deviations for all peptides across all time points for both states. The detailed HDX-MS results are summarized in the supplementary data, following Masson et al.’s guidelines⁶⁴.

Acknowledgements

We are grateful to Drs. J. Di Trani and S. Benlekbir for help and advice regarding cryoEM specimen preparation. We thank Laura Pelser, Astrid Pouwelsen, Jacqueline Kuhnen, Jolanda Klaassen and Saskia Mulder for assistance with mosquito dissections. This work was undertaken, in part, thanks to funding from the Canadian Institutes for Health Research and was supported by the CIFAR Azrieli Global Scholar program (JPJ), the Ontario Early Researcher Award program (JPJ), and the Canada Research Chair program (JLR, JPJ). IK and DI were supported by SickKids Restracomp Fellowships. SH was supported by the Canada Graduate Scholarships Doctoral Award. MRI and MMJ are supported by the Netherlands Organisation for Scientific Research (Vidi fellowship number 192.061). X-ray diffraction experiments were performed at GM/CA@APS, which has been funded in whole or in part with federal funds from the National Cancer Institute (ACB-12002) and the National Institute of General Medical Sciences (AGM-12006). The Eiger 16M detector was funded by an NIH–Office of Research Infrastructure Programs High-End Instrumentation grant (1S10OD012289-01A1). This research used resources of the Advanced Photon Source, a U.S. Department of Energy (DOE) Office of Science user facility operated for the DOE Office of Science by Argonne National Laboratory under contract DE-AC02-06CH11357. Cryo-EM data were collected at the Toronto High-Resolution High-Throughput cryo-EM facility and biophysical data were acquired at the Structural & Biophysical Core facility at SickKids, supported by the Canada Foundation for Innovation and Ontario Research Fund. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. X-ray crystallography and cryoEM structures are accessible from the Protein Data Bank and Electron Microscopy Data Bank under IDs: EMD-41821 (Pfs48/45-scTB31F:RUPA-44:RUPA-58:RUPA-154 complex), EMD-41822 (locally refined map of Pfs48/45-scTB31F:RUPA-44:RUPA-58:RUPA-154 complex), 8U1P (Pfs48/45:RUPA-58:RUPA-154 model), and 8U70 (RUPA-58 Fab crystal structure).

Competing interests

The authors declare no competing interests related to this work.

World Health Organization. World malaria report 2021. https://www.who.int/publications-detail-redirect/9789240040496.
Hall, B. F. & Fauci, A. S. Malaria Control, Elimination, and Eradication: The Role of the Evolving Biomedical Research Agenda. J. Infect. Dis. 200, 1639–1643 (2009).
Duffy, P. E. Transmission-Blocking Vaccines: Harnessing Herd Immunity for Malaria Elimination. Expert Rev. Vaccines 20, 185–198 (2021).
Duffy, P. E. Current approaches to malaria vaccines. Curr. Opin. Microbiol. 70, 102227 (2022).
Mahase, E. WHO recommends second vaccine for malaria prevention in children. BMJ 383, 2291 (2023).
Laurens, M. B. RTS,S/AS01 vaccine (Mosquirix^TM): an overview. Hum. Vaccines Immunother. 16, 480–489 (2020).
Datoo, M. S. et al. Efficacy of a low-dose candidate malaria vaccine, R21 in adjuvant Matrix-M, with seasonal administration to children in Burkina Faso: a randomised controlled trial. The Lancet 397, 1809–1818 (2021).
Tinto, H. et al. Long-term incidence of severe malaria following RTS,S/AS01 vaccination in children and infants in Africa: an open-label 3-year extension study of a phase 3 randomised controlled trial. Lancet Infect. Dis. 19, 821–832 (2019).
Wang, L. T. et al. A Potent Anti-Malarial Human Monoclonal Antibody Targets Circumsporozoite Protein Minor Repeats and Neutralizes Sporozoites in the Liver. Immunity 53, 733-744.e8 (2020).
Seder, R. A. et al. Protection against malaria by intravenous immunization with a nonreplicating sporozoite vaccine. Science 341, 1359–1365 (2013).
Wu, R. L. et al. Low-Dose Subcutaneous or Intravenous Monoclonal Antibody to Prevent Malaria. N. Engl. J. Med. 387, 397–407 (2022).
Kayentao, K. et al. Safety and Efficacy of a Monoclonal Antibody against Malaria in Mali. N. Engl. J. Med. 387, 1833–1842 (2022).
Silk, S. E. et al. Superior antibody immunogenicity of a viral-vectored RH5 blood-stage malaria vaccine in Tanzanian infants as compared to adults. Med N. Y. N 4, 668-686.e7 (2023).
Wu, Y. et al. Phase 1 trial of malaria transmission blocking vaccine candidates Pfs25 and Pvs25 formulated with montanide ISA 51. PloS One 3, e2636 (2008).
Healy, S. A. et al. Pfs230 yields higher malaria transmission-blocking vaccine activity than Pfs25 in humans but not mice. J. Clin. Invest. 131, e146221, 146221 (2021).
van Dijk, M. R. et al. A central role for P48/45 in malaria parasite male gamete fertility. Cell 104, 153–164 (2001).
Kocken, C. H. M. et al. Cloning and expression of the gene coding for the transmission blocking target antigen Pfs48/45 of Plasmodium falciparum. Mol. Biochem. Parasitol. 61, 59–68 (1993).
Gerloff, D. L., Creasey, A., Maslau, S. & Carter, R. Structural models for the protein family characterized by gamete surface protein Pfs230 of Plasmodium falciparum. Proc. Natl. Acad. Sci. 102, 13598–13603 (2005).
van der Boor, S. C. et al. Safety, tolerability, and Plasmodium falciparum transmission-reducing activity of monoclonal antibody TB31F: a single-centre, open-label, first-in-human, dose-escalation, phase 1 trial in healthy malaria-naive adults. Lancet Infect. Dis. 22, 1596–1605 (2022).
Kundu, P. et al. Structural delineation of potent transmission-blocking epitope I on malaria antigen Pfs48/45. Nat. Commun. 9, 4458 (2018).
Outchkourov, N. S. et al. Correctly folded Pfs48/45 protein of Plasmodium falciparum elicits malaria transmission-blocking immunity in mice. Proc. Natl. Acad. Sci. U. S. A. 105, 4301–4305 (2008).
Theisen, M., Jore, M. M. & Sauerwein, R. Towards clinical development of a Pfs48/45-based transmission blocking malaria vaccine. Expert Rev. Vaccines 16, 329–336 (2017).
Lennartz, F. et al. Structural basis for recognition of the malaria vaccine candidate Pfs48/45 by a transmission blocking antibody. Nat. Commun. 9, 3822 (2018).
McLeod, B. et al. Vaccination with a structure-based stabilized version of malarial antigen Pfs48/45 elicits ultra-potent transmission-blocking antibody responses. Immunity 55, 1680-1692.e8 (2022).
Dickey, T. H. et al. Design of a stabilized non-glycosylated Pfs48/45 antigen enables a potent malaria transmission-blocking nanoparticle vaccine. NPJ Vaccines 8, 20 (2023).
Roeffen, W. et al. Plasmodium falciparum: Production and Characterization of Rat Monoclonal Antibodies Specific for the Sexual-Stage Pfs48/45 Antigen. Exp. Parasitol. 97, 45–49 (2001).
Fabra-García, A. et al. Highly potent, naturally acquired human monoclonal antibodies against Pfs48/45 block Plasmodium falciparum transmission to mosquitoes. Immunity 56, 406-419.e7 (2023).
Ko, K.-T. et al. Structure of the malaria vaccine candidate Pfs48/45 and its recognition by transmission blocking antibodies. Nat. Commun. 13, 5603 (2022).
Stone, W. J. R. et al. Unravelling the immune signature of Plasmodium falciparum transmission-reducing immunity. Nat. Commun. 9, 558 (2018).
MalariaGEN et al. An open dataset of Plasmodium falciparum genome variation in 7,000 worldwide samples. Wellcome Open Res. 6, 42 (2021).
Singh, S. K. et al. A Plasmodium falciparum 48/45 single epitope R0.6C subunit protein elicits high levels of transmission blocking antibodies. Vaccine 33, 1981–1986 (2015).
Kumar, N. Target antigens of malaria transmission blocking immunity exist as a stable membrane bound complex. Parasite Immunol. 9, 321–335 (1987).
Pradel, G. et al. A multidomain adhesion protein family expressed in Plasmodium falciparum is essential for transmission to the mosquito. J. Exp. Med. 199, 1533–1544 (2004).
Simon, N., Kuehn, A., Williamson, K. C. & Pradel, G. Adhesion protein complexes of malaria gametocytes assemble following parasite transmission to the mosquito. Parasitol. Int. 65, 27–30 (2016).
Ragotte, R. J. et al. Heterotypic interactions drive antibody synergy against a malaria vaccine candidate. Nat. Commun. 13, 933 (2022).
Goerdeler, F., Seeberger, P. H. & Moscovitz, O. Unveiling the Sugary Secrets of Plasmodium Parasites. Front. Microbiol. 12, (2021).
Bushkin, G. G. et al. Suggestive Evidence for Darwinian Selection against Asparagine-Linked Glycans of Plasmodium falciparum and Toxoplasma gondii. Eukaryot. Cell 9, 228–241 (2010).
Crispin, M., Ward, A. B. & Wilson, I. A. Structure and Immune Recognition of the HIV Glycan Shield. Annu. Rev. Biophys. 47, 499–523 (2018).
Cao, Y., Bansal, G. P., Merino, K. & Kumar, N. Immunological Cross-Reactivity between Malaria Vaccine Target Antigen P48/45 in Plasmodium vivax and P. falciparum and Cross–Boosting of Immune Responses. PLOS ONE 11, e0158212 (2016).
Antanasijevic, A. et al. Polyclonal antibody responses to HIV Env immunogens resolved using cryoEM. Nat. Commun. 12, 4817 (2021).
Ponnudurai, T. et al. Infectivity of cultured Plasmodium falciparum gametocytes to mosquitoes. Parasitology 98 Pt 2, 165–173 (1989).
Miura, K. et al. Qualification of standard membrane-feeding assay with Plasmodium falciparum malaria and potential improvements for future assays. PloS One 8, e57909 (2013).
Li, H. A statistical framework for SNP calling, mutation discovery, association mapping and population genetical parameter estimation from sequencing data. Bioinforma. Oxf. Engl. 27, 2987–2993 (2011).
Marr, C. R., Benlekbir, S. & Rubinstein, J. L. Fabrication of carbon films with ∼ 500nm holes for cryo-EM with a direct detector device. J. Struct. Biol. 185, 42–47 (2014).
Punjani, A., Rubinstein, J. L., Fleet, D. J. & Brubaker, M. A. cryoSPARC: algorithms for rapid unsupervised cryo-EM structure determination. Nat. Methods 14, 290–296 (2017).
Rubinstein, J. L. & Brubaker, M. A. Alignment of cryo-EM movies of individual particles by optimization of image translations. J. Struct. Biol. 192, 188–195 (2015).
Punjani, A., Zhang, H. & Fleet, D. J. Non-uniform refinement: adaptive regularization improves single-particle cryo-EM reconstruction. Nat. Methods 17, 1214–1221 (2020).
Punjani, A. & Fleet, D. J. 3D variability analysis: Resolving continuous flexibility and discrete heterogeneity from single particle cryo-EM. J. Struct. Biol. 213, 107702 (2021).
Dunbar, J. et al. SAbPred: a structure-based antibody prediction server. Nucleic Acids Res. 44, W474–W478 (2016).
Pettersen, E. F. et al. UCSF Chimera--a visualization system for exploratory research and analysis. J. Comput. Chem. 25, 1605–1612 (2004).
Emsley, P., Lohkamp, B., Scott, W. G. & Cowtan, K. Features and development of Coot. Acta Crystallogr. D Biol. Crystallogr. 66, 486–501 (2010).
Afonine, P. V. et al. Real-space refinement in PHENIX for cryo-EM and crystallography. Acta Crystallogr. Sect. Struct. Biol. 74, 531–544 (2018).
Chen, V. B. et al. MolProbity: all-atom structure validation for macromolecular crystallography. Acta Crystallogr. D Biol. Crystallogr. 66, 12–21 (2010).
Goddard, T. D. et al. UCSF ChimeraX: Meeting modern challenges in visualization and analysis. Protein Sci. Publ. Protein Soc. 27, 14–25 (2018).
Krissinel, E. & Henrick, K. Inference of macromolecular assemblies from crystalline state. J. Mol. Biol. 372, 774–797 (2007).
Baker, N. A., Sept, D., Joseph, S., Holst, M. J. & McCammon, J. A. Electrostatics of nanosystems: application to microtubules and the ribosome. Proc. Natl. Acad. Sci. U. S. A. 98, 10037–10041 (2001).
The PyMOL Molecular Graphics System. (2015).
Kabsch, W. XDS. Acta Crystallogr. D Biol. Crystallogr. 66, 125–132 (2010).
XPREP. X-ray data Preparation and Reciprocal space Exploration Program. (2014).
McCoy, A. J. et al. Phaser crystallographic software. J. Appl. Crystallogr. 40, 658–674 (2007).
Schmidt, A. G. et al. Viral receptor-binding site antibodies with diverse germline origins. Cell 161, 1026–1034 (2015).
Adams, P. D. et al. PHENIX: a comprehensive Python-based system for macromolecular structure solution. Acta Crystallogr. D Biol. Crystallogr. 66, 213–221 (2010).
Scheres, S. H. W. RELION: Implementation of a Bayesian approach to cryo-EM structure determination. J. Struct. Biol. 180, 519–530 (2012).
Masson, G. R. et al. Recommendations for performing, interpreting and reporting hydrogen deuterium exchange mass spectrometry (HDX-MS) experiments. Nat. Methods 16, 595–602 (2019).
Abanades, B., Georges, G., Bujotzek, A. & Deane, C. M. ABlooper: fast accurate antibody CDR loop structure prediction with accuracy estimation. Bioinforma. Oxf. Engl. 38, 1877–1880 (2022).
Baek, M. et al. Accurate prediction of protein structures and interactions using a three-track neural network. Science 373, 871–876 (2021).

There is NO Competing Interest.

SupplementaryRawSMFAdata.xlsx
SupplementarydataHDXMS.xlsx
Supplementarymovie1.mpg
ExtendedDataFigs.pdf
Extended Data Fig. 1. Domain specificity of mAbs as determined by ELISA with full-length Pfs48/45 and fragments thereof. Values are means of three technical replicates and error bars indicate S.E.M. D3-specific mAbs TB31F and RUPA-44 only bind to D3-containing fragments, D2-specific mAb RUPA-154 only to D2-containing fragments and RUPA-58 only to D1-containing fragments. The threshold for positivity is set as the mean of six negative controls + 3 * SD and is indicated by a dashed line. Extended Data Fig. 2. CryoEM processing workflow of the Pfs48/45-scTB31F:RUPA-44:RUPA-58 complex (a) and the Pfs48/45-scTB31F:RUPA-44:RUPA-58:RUPA-154 complex (b) data in cryoSPARC v3. Extended Data Fig. 3. CryoEM analysis of the Pfs48/45-scTB31F:RUPA-44:RUPA-58 (a-e) and Pfs48/45-scTB31F:RUPA-44:RUPA-58:RUPA-154 complexes (f-q). (a) Representative cryoEM micrograph of the Pfs48/45-scTB31F:RUPA-44:RUPA-58 complex. Scale bar is 50 nm. (b) Selected 2D class averages of the Pfs48/45-scTB31F:RUPA-44:RUPA-58 complex. (c) Gold standard Fourier shell correlation (GSFSC) curves for the final 3D non-uniform refinement of the Pfs48/45-scTB31F:RUPA-44:RUPA-58 complex in cryoSPARC v3 without symmetry imposed. (d) Viewing direction distribution of the Pfs48/45-scTB31F:RUPA-44:RUPA-58 complex dataset. (e) Local resolution (Å) plotted on the surface of the cryoEM map of the Pfs48/45-scTB31F:RUPA-44:RUPA-58 complex. (f) Representative cryoEM micrograph of the Pfs48/45-scTB31F:RUPA-44:RUPA-58:RUPA-154 complex acquired with no tilt applied. Scale bar is 50 nm. (g) Selected 2D class averages of the Pfs48/45-scTB31F:RUPA-44:RUPA-58:RUPA-154 complex. (h) Gold standard Fourier shell correlation (GSFSC) curves for the final 3D non-uniform refinement of the Pfs48/45-scTB31F:RUPA-44:RUPA-58:RUPA-154 in cryoSPARC v3 without symmetry imposed. (i) Viewing direction distribution of the Pfs48/45-scTB31F:RUPA-44:RUPA-58:RUPA-154 complex dataset. (j) Local resolution (Å) plotted on the surface of the Pfs48/45-scTB31F:RUPA-44:RUPA-58:RUPA-154 complex cryoEM map. (k) Representative cryoEM micrograph of the Pfs48/45-scTB31F:RUPA-44:RUPA-58:RUPA-154 complex collected under 40^o tilt. Scale bar is 50 nm. (l) Selected 2D class averages of the Pfs48/45-scTB31F:RUPA-44:RUPA-58:RUPA-154 complex. (m) GSFSC curves for the final 3D non-uniform refinement of the Pfs48/45-scTB31F:RUPA-44:RUPA-58:RUPA-154 complex in cryoSPARC v3 without symmetry imposed. (n) Viewing direction distribution of the Pfs48/45-scTB31F:RUPA-44:RUPA-58 complex dataset. (o) Local resolution (Å) plotted on the surface of the Pfs48/45-scTB31F:RUPA-44:RUPA-58:RUPA-154 complex cryoEM map. (p) GSFSC curves for the final locally refined map of the D1D2:RUPA-58:RUPA-154 region in cryoSPARC v3 without symmetry imposed. (q) Local resolution (Å) plotted on the surface of the locally refined cryoEM map of D1D2:RUPA-58:RUPA-154 region. Extended Data Fig. 4. Increasing the molecular weight of the Pfs48/45-scTB31F:RUPA-44:RUPA-58 complex (a) by adding RUPA-154 (b), applying tilted data collection (c) and local refinement on the D1D2:RUPA-58:RUPA-154 region (d) resulted in significant improvement of resolution and distribution of views. Extended Data Fig. 5. Comparison of the D3-scTB31F:RUPA-44 region of the Pfs48/45-scTB31F:RUPA-44:RUPA-58:RUPA-154 cryoEM map with X-ray structures of D3-TB31F (PDB ID: 6E63) and D3-RUPA-44 (PDB ID: 7UXL). Models of D3-TB31F and D3-RUPA-44 were fitted individually into cryoEM map of the Pfs48/45-scTB31F:RUPA-44:RUPA-58:RUPA-154 complex using the “fit in map” command in USCF Chimera. TB31F, RUPA-44 and D3 models are depicted in red, orange and navy, respectively, while the D1-D2-RUPA-58:RUPA-154 region of cryoEM map is shown as a white outline. Extended Data Fig. 6. Comparison of D1D2 targeting antibodies with their germline precursors. Alignment of the heavy and light chains of RUPA-154 (a) and RUPA-58 (b) with their inferred germline precursor. Fab residues that are mutated from the germline are shown in red. Somatic hypermutation sites that contact D1D2 are indicated with a red asterisk, while non-mutated residue contacts are indicated by a black asterisk. Extended Data Fig. 7. Comparison of the RUPA-154 and 10D8 binding sites on full-length Pfs48/45 from the cryo-EM structure. (a) The variable domains of 10D8 and RUPA-154 bound to D1D2. (b-c) Fabs RUPA-154 (b) and 10D8 (c) bound to full-length Pfs48/45. RUPA-154 (yellow) and 10D8 (grey) are depicted as ribbons and Pfs48/45 (blue) is represented as a surface. Extended Data Fig. 8. Conservation of RUPA-58 and RUPA-154 binding sites across Pf isolates. (a-b) Interfaces between Pfs48/45 (white) and RUPA-58 (a, pink) and RUPA-154 (b, yellow/ red) shown as ribbons. Key contacts are depicted as sticks and Pfs48/45 residues with polymorphisms are labelled. (c-j) Biolayer interferometry binding curves of RUPA-58 bound to D1D2 WT (c) and mutants S90R (d), K101T (e), and N138I (f) and RUPA-154 bound to WT (g) and mutants V212I (h) and K293T (i). RUPA-58 binding to the D1D2 K239T mutant was used as a control (j). Extended Data Fig. 9. Negative stain electron microscopy epitope mapping of the RUPA-71 binding site on Pfs48/45 D1D2. 2D classes and a 3D reconstruction of the RUPA-154:RUPA-71:Pfs48/45-D1D2 complex generated using negative stain electron microscopy (nsEM). A predicted structure of RUPA-71 generated from Abodybuilder2⁶⁶ and the RUPA-154-D1D2 portion of the full-length Pfs48/45-Fab cryo-EM structure were used to model this complex. RUPA-71 (dark violet), RUPA-154 (yellow), and D1D2 (blue) are depicted as ribbons within the nsEM map. The putative epitope of RUPA-71 is shown in coral. Extended Data Fig. 10. HDX analysis and epitope mapping on Pfs48/45 D1D2. (a) Differential uptake plot representing the differences in deuterium uptake between free and RUPA71 bound states of Pfs48/45 D1D2. Each bar represents a peptic peptide of Pfs48/45 D1D2 (ordered from N- to C-terminus based on the first amino acid of each peptic peptides). The deuterium uptake differences for each peptide at different time points are shown in orange (10 s), red (100 s), blue (1000 s) and black (10,000 s). The height of the vertical gray bar represents the sum of the mass differences for each peptide between the two states at each time point. The gray shades represent the differential uptake errors for each peptide, and the dashed horizontal lines represent the threshold which the uptake differences between the two states are considered significant if exceeded. This threshold corresponds to the 95% confidence interval, calculated from the pooled standard deviations for all peptides across all time points which is ±0.3 Da. The red and blue boxes highlight the regions with decreased and increased deuterium uptake in the bound state, respectively. (b) Deuterium uptake plots of the selected peptides of the regions that show significant HDX changes. Error bars represent mean ± S.E.M of three independent experiments. Raw HDX-MS data is available in Supplementary Table 2. Extended Data Fig. 11. Comparison of the RUPA-58 and putative RUPA-71 binding sites on D1D2. RUPA-58 (light gray) and RUPA-71 (dark violet) are depicted with ribbons and surface bound to Pfs48/45-D1D2 (blue) which is represented as ribbons. Extended Data Fig. 12. Mutation N131Q reduces binding of potent Pfs48/45 domain 1 antibodies. Biolayer interferometry data of RUPA-71 (a), RUPA-116 (b), RUPA-119 (c), and RUPA-75 (d), RUPA-155 (e), and RUPA-46 (f) binding to wild-type Pfs48/45-D1D2 and a D1D2 construct containing a N131Q point mutation. This mutation removes a putative N-linked glycosylation sequon. RUPA-58 was used as a control for proper folding of the D1D2 N131Q mutant (g). (h) A model of RUPA-71 (dark violet) binding to D1D2 (blue) generated using negative stain EM with a glycan at position N131 modeled in coral, which would result from expression of the D1D2 construct in mammalian cells. Extended Data Fig. 13. Conservation of P48/45 across Pf and Pv. Sequence alignment with identical, highly conserved, semi-conserved, and non-conserved residues indicated in dark magenta, orchid, pink, and light gray, respectively. The putative epitope of RUPA-71 is indicated with a coral bar above the sequence. Extended Data Fig. 14. Electrostatic surface potential representations of the three observed Pfs48/45 interdomain conformers, as depicted in Fig. 6. Surface color indicates electrostatic potential ranging from −5kT/e (red) to +5kT/e (blue). In general, D3 possesses an electronegative surface potential, while D1 and D2 possess a mixed electronegative and electropositive surface.
ExtendedDataTables.docx

Download PDF

Version 1

posted

You are reading this latest preprint version

Structural elucidation of full-length Pfs48/45 in complex with potent mAbs isolated from a naturally exposed individual

Status:

Version 1

Abstract

Figures

Introduction

Results

Selection of D1-, D2- and D3-targeting human antibodies

Engineering of a full-length Pfs48/45 construct suitable for structural determination

CryoEM structure determination of full-length Pfs48/45

Flexibility within full-length Pfs48/45

Delineation of the D2 epitope recognized by mAb RUPA-154

Delineation of the D1 epitope recognized by potent mAb RUPA-58

Molecular description of epitopes on Pfs48/45 D1 targeted by human mAbs

Discussion

Methods

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1