IntroductionMalaria is a significant global public health challenge with approximately 249 million cases and an estimated 608,000 deaths in 20221. The tropical disease is caused by Plasmodium parasites of which falciparum is the cause of most malaria-related mortality2. Following transmission of parasites to humans after a bite of a female Anopheles mosquito, symptoms such as fever, chills, diarrhea, and convulsions develop within 10 days to 4 weeks3. Insecticide-treated nets and anti-malarial drugs have contributed to a decline in malaria cases, but increasing drug resistance by malaria parasites has stymied eradication efforts, highlighting the need for effective malaria vaccines4.Plasmodium parasites undergo a complex life cycle between the human host and mosquito vector. There has been considerable vaccine development focused on disrupting the parasite life cycle at various stages, specifically aimed at preventing hepatocyte invasion, reducing symptomatic forms in the human bloodstream, and blocking transmission to mosquitoes5. The first vaccine for malaria approved for use in select African countries by the WHO is the RTS,S/AS01 subunit vaccine, which aims to disrupt the malaria life cycle at the sporozoite stage prior to liver infection. Plasmodium falciparum sporozoites have a surface coating of circumsporozoite protein (PfCSP), which consists of amino-terminal and carboxyl-terminal domains flanking a central region of highly conserved NANP and NANP-like motifs repeated approximately 37–70 times depending on the strain. PfCSP is anchored to the sporozoite surface by myristoylated glycosylphosphatidylinositol (GPI) glycolipids, a post-translational replacement of a C-terminal hydrophobic tail6. RTS,S consists of 19 NANP major repeats and the C-terminal region displayed on a virus-like particle with hepatitis B surface antigen. Clinical trials with the RTS,S/AS01 vaccine showed variable levels of efficacy against infection that waned over time7,8,9. The second malaria vaccine to be approved by the WHO is the R21/Matrix-M vaccine that displays the same PfCSP antigenic regions as RTS,S; R21 was found to reach 75% efficacy against infection in African children, surpassing the efficacy of RTS,S10. Clinical trials studying vaccination with radiation-attenuated whole sporozoites have shown lower vaccine efficacy against natural Plasmodium falciparum infection compared to subunit vaccine counterparts in children and adults11,12,13. The mechanism of vaccine-induced antibody-mediated protection is still unclear, although some studies have shown sporozoite inhibition through Fc-dependent effector cell recruitment14, complement activation15, and other humoral mechanisms16. While previous studies dissected antibody responses to full-length PfCSP to determine the relative protection conferred by antibodies to each epitope region17, we hypothesized that immunodominant epitopes might be suppressing responses to other domains. Therefore, we tested each epitope region independently.Recent studies have revealed that potent anti-PfCSP antibodies induced by sporozoite immunization bind to minor repeats of NPDP and NVDP found at the junction of the N-terminal domain and the central NANP-repeat region, and these minor repeats are termed junctional epitopes18. In controlled human malaria infection trials, passive administration of monoclonal antibodies (mAbs) that bind strongly to junctional epitopes was found to provide protection against malaria19,20,21,22. Several groups have made multimeric vaccine candidates targeting these junctional epitopes that show protection in challenge mouse models23,24,25, highlighting the potential importance of these epitopes despite their absence in RTS,S.Other studies have highlighted the C-terminus as a significant contributor to protection in RTS,S26,27. The C-terminus, which has been shown to mediate hepatocyte invasion28, was found to fold into a structure termed the α-thrombospondin repeat (αTSR) domain29. The first structural study to explore anti-αTSR domain antibodies found that the nonprotective mAb 1710 utilized its CDRH3 to insert into a hydrophobic pocket after contacting the alpha-helix on one face of the molecule30. A recent study resolved the PfCSP C-terminus into two non-competing epitope regions: the strain-specific “alpha” epitopes targeted by 1710-like mAbs and the more conserved “beta” epitopes on the opposite side of the domain targeted by several new high-affinity mAbs31. Another study found that RTS,S was significantly less effective against malaria strains that utilize different C-terminus PfCSP sequences than the vaccine strain32. The observed reduction in efficacy against infection from diverse strains may be due to length polymorphism, as the number of NANP motifs ranges from 37 to 44 in different strains, but could also be attributed to sequence polymorphisms found in the C-terminus alpha epitope33. Therefore, we set out to further evaluate this C-terminus domain to find motifs that may elicit protective antibodies regardless of strain polymorphisms.At the other terminus of PfCSP, the N-terminal domain is used by sporozoites during host cell traversal to mask the C-terminal domain28. Antibody responses to this domain have only recently been explored. 5D5 is a nonprotective murine antibody with high affinity to the N-terminal domain34. MAD2-6 is a human antibody that has protective function when presented as IgA but not as IgG35. Both target the same region of the N-terminal domain, raising the question of whether other N-terminal epitopes are immunogenic and potentially protective.In the past decade, major advancements in mRNA delivery technology have resulted in the development of mRNA preclinical and clinical vaccines with favorable immunogenicity and safety profiles36. The nucleoside-modified mRNA immunogens delivered with lipid nanoparticles (LNPs) used by Pfizer/BioNTech and Moderna in SARS-CoV-2 vaccines have been proven to be safe and highly effective37,38,39. mRNA delivery is a newcomer to the field of malaria vaccinology, with only a few but promising studies targeting liver-resident memory T-cells40 or malaria antigens PfCSP, Pfs25, and PfRH541,42,43. Thus, additional studies of mRNA delivery of malaria antigens in comparison to protein delivery are needed to identify potential next-generation malaria vaccines.Here, we displayed potentially protective epitopes of PfCSP on two different glycosylated nanoparticles, ferritin (24mer) and lumazine synthase (60mer), for protein delivery. To assess the effectiveness of these immunogens, we immunized C57BL/6 J mice and evaluated the reduction in liver burden against live sporozoite challenge, which has predictive correlation with human protection19,44. These nanoparticles displayed one or more of the following: peptide (NANP)6, peptides covering the junctional region, wild-type CSP C-terminus, a C-terminal domain variant that was modified to abrogate alpha-site mAb binding without impacting binding to the novel beta-site epitope, and novel N-terminal epitopes targeted by antibodies elicited by KLH-N-terminal peptide mouse immunizations45. We selected a construct bringing together the most protective epitopes, namely a peptide covering the junctional region with additional NANPs, for delivery by mRNA using several different construct presentations, including two nanoparticle systems and two membrane-tethered systems: GPI-anchoring and a vesicular stomatitis virus glycoprotein G transmembrane (TM) domain. Furthermore, we evaluated PfCSP epitopes on immunogenic platforms in an unbiased manner to downselect the most protective epitopes and found, using mRNA delivery, promising sterilizing immunogens suggesting prospective platforms for future development.ResultsImmunogen designTo develop immunogenic platforms for different PfCSP epitopes, we constructed ferritin- and lumazine synthase-based self-assembling nanoparticles displaying one or more different epitopes (Fig. 1b, c). We modified H. pylori ferritin (PDB:3bve) to engineer additional glycosylation sites. We also engineered additional glycosylation sites on A. aeolicus lumazine synthase (PDB:1hqk), a construct that had been previously modified to remove an unpaired cysteine and a buried glycosylation site46, add disulfide bridges, and resurface the active site47. The newly added glycosylation sites provide several potential advantages: (i) masking of non-CSP surfaces might reduce off-target responses; (ii) additional glycans might lead to improved in vivo trafficking to lymph nodes to augment immune responses48; and (iii) additional glycan allows for purification of nanoparticle immunogens without the use of potentially immunogenic affinity tags. In addition, we developed membrane-tethered platforms utilizing the VSV TM49 or the GPI signal sequence (Accession No. AL844502) natively used in PfCSP, to display junctional repeat peptides delivered by genetic immunization (Fig. 1d, e). We hypothesized that the membrane-tethered platforms would lack T-help in comparison to the nanoparticle platforms, and therefore we incorporated exogenous CD4 + T-cell epitopes, including PADRE50 and others from lumazine synthase51, into the linker.Fig. 1: Circumsporozoite Epitope Regions.a N-terminal domain divided into five linear epitopes, termed P1, P2, P8, P9, and P15. Repeat epitopes divided into four peptides, J2, J3, J3R3, and (NANP)6. C-terminal domain sequence shown. b Model of J3R3 (magenta) displayed on 60mer lumazine synthase nanoparticle (grey). c Model of J3R3 (magenta) displayed on 24mer ferritin nanoparticle (grey). d Schematic of junctional peptide (magenta) displayed on GPI anchor with T-help linker. e Schematic of junctional peptide (magenta) displayed on transmembrane domain with T-help linker.Full size imageWe constructed immunogens displaying subunits from PfCSP strain 3D7 (Accession No. AL844502), the strain used in RTS,S/AS01, as shown in Fig. 1a. First, we fused (NANP)6 as a 24-residue peptide to the N-terminus of the nanoparticle protomers as a control for the NANP epitope of RTS,S in the context of our nanoparticles; we matched the number of repeat motifs found in the junctional region (NANPNVDP)3 as a higher number of repeat motifs interfered with nanoparticle formation. We displayed three additional peptides on nanoparticles to encompass the junctional epitope with the sequences beginning with Region I, the cleavage site important for parasite invasion28. We constructed a minimal junction peptide termed J2, a full-length junctional peptide termed J3, and a peptide termed J3R3 that added three NANP repeats onto the junctional peptide to combine junctional and NANP repeat epitopes. To focus the C-terminal immune response to the conserved epitope region, we used Rosetta to design a modified αTSR termed MD1 that minimized binding of known strain-specific antibodies without impacting antibody binding to the conserved epitope (Supplemental Fig. 1). We displayed both wild-type (WT) αTSR and MD1 αTSR on nanoparticles. We did not include the C-terminal linker, because no protective antibodies have been identified for this region52,53. We were unable to express the full N-terminal domain on nanoparticles. Instead, novel linear epitopes were identified through binding of novel antibodies to overlapping peptide sequences45, and these peptides, termed P1, P2, P7, P8, and P15, were selected for nanoparticle display as a representative set of N-terminal domain immunogenic epitopes. Nanoparticle formation was confirmed by size exclusion chromatography (SEC), SEC with multi-angle light scattering (SEC-MALS), and negative stain electron microscopy (nsEM) (Supplemental Figs. 2–10). Prospective epitopes for mRNA development were displayed on membrane-bound platforms for FACS characterization before production.In vitro characterizationWe tested the nanoparticles for binding in ELISA against a select panel of mAbs. For the nanoparticles displaying junctional and NANP repeat epitopes, we confirmed antigenicity by testing against L9, a protective mAb targeting junctional repeat epitopes; 4493 and 2541, mAbs with cross reactivity for both junctional and NANP repeat epitopes19; and 239 and 317, mAbs initially found to target NANP repeat epitopes (Fig. 2a)52,54. We tested membrane-anchored immunogens displaying junctional and NANP repeat epitopes in FACS after transient DNA transfection against the same panel of mAbs (Supplemental Fig. 11) and found all immunogens bound these anti-repeat region antibodies. Related to the C-terminal αTSR domain, which contains conformational epitopes53, we confirmed appropriate epitope presentation following the modifications we had introduced by testing binding to conformation-dependent antibodies including several highly strain-specific antibodies (1710, 1488, 234, and 236) and two broadly cross-reactive antibodies (1512 and 1550) that bind a wide panel of αTSR domains from different strains (Fig. 2b)31. For the nanoparticles displaying N-terminal epitopes, we tested against a panel of mouse antibodies originally used to identify these peptides. P1 and P2 nanoparticles bound to mNCSP27. P8 and P9 nanoparticles bound to mNCSP10, and P15 nanoparticles bound to 5D5 (Fig. 2c). Thus, all immunogens were confirmed to display their respective epitopes.Fig. 2: Nanoparticle Characterization by ELISA.a Binding, area under the curve (AUC), of anti-repeat antibodies 4493, 2541, 239, and 317 to directly coated repeat-displaying 24mer and 60mer immunogens. b Binding AUC of anti-C-terminal antibodies 1512, 1550, 1710, 1488, 236, and 234 to directly coated wild-type and modified-1 C-term displaying 24mer and 60mer immunogens. c Binding AUC of anti-N-terminal antibodies mNCSP-27, mNCSP-10, and 5D5 to directly coated N-term displaying 24mer and 60mer immunogens. Anti-repeat antibody 317 was used as a negative control.Full size imageSerological response to vaccinationTo evaluate the immunogenicity of the nanoparticles, we vaccinated mice twice with proteins delivered with SMNP adjuvant55 (Fig. 3a). Because mouse age, strain, and sex can confound the raw fluorescence readouts in down-stream parasite challenge assays in naïve mice, we consistently utilized female C57BL/6 J mice that were 7–8 weeks old at the time of first immunization; C57BL/6 J mice are the most susceptible strain of mice to parasite infection and the most difficult mice to protect against challenge56. As a positive control immunogen, we used full-length monomeric PfCSP produced in Lactococcus lactis57, adjuvanted with SMNP.Fig. 3: Serological response to vaccination by ELISA.a Schematic illustrating immunizations at weeks 0 and 5. Sera was taken from 1 week after last injection and challenge with live sporozoites was done at either 2 weeks or 5 weeks after last injection. Liver fluorescence was imaged at 42 h after challenge time. b Binding of sera to J2 peptide coated plates. Symbols represent area under the curve (AUC) values for individual mice (n = 6). When comparing serological responses between 24mers and 60mers displaying same peptides, significance testing was performed using a two-tailed Mann-Whitney U test; ns not significant; **p