A universal vaccine candidate against Plasmodium vivax malaria confers protective immunity against the three PvCSP alleles

Malaria is a highly prevalent parasitic disease in regions with tropical and subtropical climates worldwide. Among the species of Plasmodium causing human malaria, P. vivax is the second most prevalent and the most geographically widespread species. A major target of a pre-erythrocytic vaccine is the P. vivax circumsporozoite protein (PvCSP). In previous studies, we fused two recombinant proteins representing three allelic variants of PvCSP (VK210, VK247 and P. vivax-like) to the mumps virus nucleocapsid protein to enhance immune responses against PvCSP. The objective of the present study was to evaluate the protective efficacy of these recombinants in mice challenged with transgenic P. berghei parasites expressing PvCSP allelic variants. Formulations containing Poly (I:C) or Montanide ISA720 as adjuvants elicited high and long-lasting IgG antibody titers specific to each PvCSP allelic variant. Immunized mice were challenged with two existing chimeric P. berghei parasite lines expressing PvCSP-VK210 and PvCSP-VK247. We also developed a novel chimeric line expressing the third allelic variant, PvCSP-P. vivax-like, as a new murine immunization-challenge model. Our formulations conferred partial protection (significant delay in the time to reach 1% parasitemia) against challenge with the three chimeric parasites. Our results provide insights into the development of a vaccine targeting multiple strains of P. vivax.

genicity and protection after challenge. Tuck-ordinary (TO) outbred mice were used for parasite production and transmission. Mice were purchased from Harlan (UK). Female OF1 mice (6-7 weeks; Charles River, NL) were used to generate chimeric P. berghei lines. Immunogenicity and protection studies were performed in accordance with the recommendations of the UK Home Office Animals Act Project License. Procedures were approved by the University of Oxford Animal Care and Ethical Review Committee (PPL P9804B4F1).
Longevity assays were performed in accordance with the terms of the Guide for the Care and Use of Laboratory Animals of the Brazilian National Council of Animal Experimentation (http:// www. cobea. org. br/). The protocol (CEUA/FCF no. 74.2016-P53) was approved by the Research Committee on Animal Experimentation of the School of Pharmaceutical Sciences of the University of São Paulo, Brazil.
Experiments for the generation of the chimeric P. berghei lines were granted with a license by the Competent Authority after advice on the ethical evaluation by the Animal Experiments Committee Leiden (AVD1160020171625). All experiments were performed in accordance with the Experiments on Animals Act, the applicable legislation in the Netherlands in accordance with the European guidelines (EU directive no. 2010/63/EU) regarding the protection of animals used for scientific purposes. The experiments were executed in a licensed establishment for the use of experimental animals (LUMC). Mice were housed in individually ventilated cages furnished with autoclaved aspen woodchip, fun tunnel, wood chew block and Nestlets at 21 ± 2 °C under a 12:12 h light-dark cycle with a relative humidity of 55 ± 10%.
This study was carried out in compliance with the ARRIVE (Animal Research: Reporting of In Vivo Experiments) guidelines for animals.

Generation and genotyping of chimeric P. berghei lines expressing the PvCSP-P. vivax-like protein.
To generate the chimeric P. berghei replacement line, we replaced the P. berghei csp coding sequence (CDS; Pbcsp; PBANKA_0403200) with the PvCSP-P. vivax-like CDS (Locus PVU09738, Accession U09738) using a 2-step GIMO transfection protocol 25,26,28 . In the first step, we deleted the P. berghei csp CDS and replaced it with a positive-negative selectable marker to create a P. berghei csp deletion GIMO line (PbANKA-CSP GIMO). The construct (pL1929) used and the generation of the PbANKA-CSP GIMO line (line 2251cl1) have been described previously 29 . This construct contains the positive-negative (hdhfr::yfcu) SM cassette and was used to insert both the Pbcsp 5′ and 3′ gene targeting regions (TRs), encompassing the full-length promoter and transcription terminator sequences, respectively, and was transfected into PbGFP-Luc con parasites (676m1cl1) using standard transfection methods 30 . Transfected parasites were selected in mice by applying positive selection by providing pyrimethamine in the drinking water 30 . Transfected parasites were cloned using the limiting dilution method 31 , resulting in the PbANKA-CSP GIMO line (line 2251 cl1). In the second step, we replaced the positive-negative SM in the PbANKA-CSP GIMO genome with the PvCSP-P. vivax-like CDS by GIMO transfection to create the P. berghei chimeric Pb-PvG10 replacement line Pb-PvG10(r). This line was obtained by modifying the construct used in the first step (pL1929); specifically, the hdfhr::yfcu SM cassette was removed and replaced with the PvCSP-P. vivax-like CDS, generating plasmid pL2161. The PvCSP-P. vivax-like CDS was ordered from GeneArt Gene Synthesis-Thermo Fisher Scientific. The pL2161 construct was sequenced to ensure that no mutations were present in the PvCSP-P. vivax-like CDS during the cloning process. The construct was linearized using AflII and SacI restriction enzymes outside of the 5' and 3' TRs before transfection. The construct was used to transfect parasites of the PbANKA-CSP GIMO line (line 2251 cl1) 26 using standard methods of GIMO transfection to generate a single replacement gene chimeric parasite 25,[32][33][34] . Transfected parasites were selected in mice by applying negative selection by providing 5-fluorocytosine (5-FC) in the drinking water 35 . Negative selection results in the selection of chimeric parasites where the hdhfr::yfcu SM in the csp locus on chromosome 4 of the PbANKA-CSP GIMO line is replaced with the PvCSP-P. vivax-like CDS. Selected chimeric parasites were cloned using the limiting dilution method 31 . Correct integration of the constructs into the genome of chimeric parasites was analyzed by performing a gDNA Southern analysis of pulsed field gel (PFG)-separated chromosomes, as previously described 36 . This method creates chimeric 'gene replacement' P. berghei parasites that lack the Pbcsp CDS but express the PvCSP-P. vivax-like protein (Pb-PvG10(r); line 2710 cl1) under the control of the P. berghei csp regulatory sequences.
The PvCSP-P. vivax-like Gabon Clone G10 CDS (Locus PVU09738, Accession U09738) gene was introduced into the genome as an additional copy of the gene in the neutral 230p locus using the previously described 'gene insertion/marker out' (GIMO) technology 26,28,37 and the standard GIMO DNA construct pL0043 to generate the chimeric P. berghei additional copy line. This construct contains 5' and 3' targeting sequences for the 230p locus, as well as a multiple-cloning site for the integration of transgene expression cassettes. This construct integrates transgenes by double crossover homologous recombination and replaces the positive-negative SM (human dihydrofolate reductase:: yeast cytosine deaminase and uridyl phosphoribosyl transferase (hdhfr::yfcu)) cassette with the transgene expression cassette. The expression cassette contained the PvCSP-P. vivax-like CDS flanked by the 5' and 3' promoter and transcription terminator sequences of the P. berghei uis4 gene (PBANKA_0501200), which were amplified from P. berghei ANKA WT genomic DNA 36 . The coding sequence of the P. vivax CSP-like G10 gene (Locus PVU09738, Accession U09738) was ordered from GeneArt Gene Synthesis-Thermo Fisher Scientific. In addition, a reporter cassette containing GFP::luciferase 30 driven by the constitutive P. berghei elongation factor 1 alpha (ef1α) promoter was also cloned into the transgene construct to generate the gene insertion construct pL2163 (PvCSP-Like G10@ Pbuis4 + GFP::Luc@Pbeef1a_230p) targeting the neutral 230p locus on chromosome 3. The coding sequence and promoter region of the construct were confirmed by sequencing.
The pL2163 construct was linearized by SacII restriction digestion and introduced into parasites of the GIMO motherline 1596cl1 using standard methods of GIMO transfection 26 . Transfected parasites were selected in mice through the addition of 5-fluorocytosine (5-FC) to the drinking water 35 , resulting in negative selection of parasites in which the SM in the 230p locus was replaced by the PvCSP-P. vivax-like expression/reporter cassette. The selected chimeric parasites were cloned using the limiting dilution method 31 . Primer sequences are listed in Table S1. Correct integration of the PvCSP-P. vivax-like coding sequence (under control of the Pbuis4 promoter) into the genome of clones of the chimeric line (Pb-PvCSP-like G10, 2700 cl1) was analyzed by performing a diagnostic PCR analysis of gDNA and Southern analysis of pulsed field gel (PFG)-separated chromosomes 36 . Phenotype and fitness assessment of the chimeric P. berghei lines expressing PvCSP-P. vivax-like protein. Multiplication of blood stages in mice was determined during the cloning period as previously described , 38 . Feeding of Anopheles stephensi mosquitoes, determination of oocyst production, and sporozoite (spz) collection were performed as described elsewhere 36,38 . The infectivity of chimeric spz was assessed by determining the T1% period (i.e., the time to reach 1% parasitemia) after an intravenous injection of 1000 spz in the tail vein of inbred BALB/c mice (Harlan, UK).
The expression of the PvCSP-P. vivax-like protein in spz was analyzed by performing an immunofluorescence assay (IFA) using sera from mice immunized with the recombinant proteins (diluted 1:100). As a control, the 3D11 antibody 39  www.nature.com/scientificreports/ 20-25 µL of primary monoclonal antibody in PBS containing 10% FCS + 1% BSA (blocking medium) for 1-2 h at room temperature or overnight at 4 °C. After the incubation, the primary antibody was removed, and the slides were washed three times with PBS, followed by staining with the secondary antibody (Alexa Fluor-488 goat anti-mouse IgG from Life Technologies, Cat# A-11001) diluted 1:800 in PBS containing 10% FCS + 1% BSA (blocking medium) for 1 h at room temperature. After three washes with PBS, nuclei were stained with 2% Hoechst-33342 (Cell Signaling Technology #4082S) in PBS for 10 min at room temperature, washed twice with PBS and air-dried, followed by the addition of fluorescence mounting medium (Dako, code S3023). Cover slips were mounted onto the slides, and the slides were sealed with nail polish and allowed to dry overnight in the dark as described in a previous study 37 . The spz were analyzed using a DMI-300B Leica fluorescence microscope in both blue and green channels, and images were processed using ImageJ software.
Purified recombinant proteins obtained after expression in P. pastoris. Expression and purification of recombinant CSP proteins from P. pastoris was carried out as described previously 23  Antibody measurements. Twelve days after each immunization, blood was collected from the tail vein, and sera were analyzed for the presence of antibodies recognizing each recombinant protein. Antibodies were detected by enzyme-linked immunosorbent assay (ELISA), essentially as described in a previous study 22 . The recombinant proteins NLP-CSP CT , NLP-CSP R

23
, yPvCSP-VK210 CT , yPvCSP-VK247 CT, and yPvCSP-P. vivax-like CT 22 were employed as solid phase-bound antigens (200 ng/well). After an overnight incubation at RT, plates were washed with a solution of PBS containing 0.05% Tween-20 (PBS-T) and blocked with a blocking solution (PBS, 5% (w/v) skimmed milk) for 2 h at 37 °C. Serial dilutions of murine polyclonal sera were added to the wells and incubated for 1 h at RT; after washes with PBS-T, peroxidase-labeled goat anti-mouse IgG (Sigma, St. Louis, USA), diluted 1:3000, was added to each well. Reactions were developed with the OPD/acid stop system. Anti-IgG titers were determined based on the highest dilution of sera yielding an A 492 greater than 0.1.

Challenge of mice with chimeric sporozoites. Spz of the chimeric parasite lines Pb-PvCSP-210
(2196cl1), Pb-PvCSP247 (2199cl1) 25 and Pb-PvCSP-like G10 (2700cl1) were used to challenge immunized mice. Female A. stephensi mosquitoes were used to produce chimeric spz. After 21 days of incubation in a humidified incubator at 19-21 °C on a 12-h day-night cycle and feeding on a fructose-p-aminobenzoic acid (PABA) solution, the mosquitoes were dissected, salivary glands were isolated, and spz were extracted. The total number of spz was determined using a hemocytometer, and 2000 spz were intravenously (i.v.) injected in 100 µL 14 days after the second booster immunization.
Parasitemia analyses. Thin blood smears were prepared daily from day 4 to day 12 after challenge or until the day after mice reached 1% parasitemia. The smears were prepared on glass slides with a drop of blood obtained from mouse tail veins, fixed with methanol and stained for 15 min using 10% Giemsa. The glass slides were observed under a light microscope, and the percentages of parasitized red blood cells were determined. The time required to reach 1% parasitemia (T1%) is a variable calculated by a linear regression equation using the percentage of parasites detected in blood on the first three consecutive days with positive parasitemia. Protection analysis using T1% period as index is a useful tool to assess vaccine efficacy, as the comparative time to reach a determined level of parasitemia reflects the prepatent period and the number of parasites erupting from the liver 40 .
Statistical analyses. All analyses and graphics were performed/generated using GraphPad Prism version 8.0 (GraphPad Software Inc., La Jolla, CA, USA). IgG Ab titers were compared using one-way analysis of variance (ANOVA). One-way ANOVA was also used to compare normally distributed log-transformed means for the different animal groups. Multiple comparisons were assessed using Tukey's posttest with a significance level of p < 0.05. Survival curves were compared using a log-rank Mantel-Cox test with a significance level of p < 0.05.

PvCSP-specific antibody responses in mice immunized with the recombinant proteins NLP-CSP CT and NLP-CSP R .
In previous studies, we generated two chimeric recombinant proteins, NLP-CSP CT and NLP-CSP R , fusing domains of the three PvCSP variants (VK210, VK247, and P. vivax-like) to the mumps virus nucleocapsid protein (NP) 23 . A schematic representation of the mumps virus, the NP protein and the new recombinants is depicted in Fig. 1. Briefly, the strategy used was fusing the malaria antigens to a coreviral protein rather than the surface proteins, thus avoiding the possible interference of immunological memory against mumps virus in the general population 23 . First, we compared the humoral immune response induced by immunizing mice with different vaccine formulations. Groups of six C57BL/6 female mice were immunized with 10 μg of the two recombinant proteins, NLP-CSP CT or NLP-CSP R , mixed with either Poly (I:C) (50 μg/dose) adjuvant or emulsified in Montanide ISA720 (7:3), an oil-water emulsion. Each animal received three immunizations 14 days apart in a homologous prime-boost vaccination regimen ( Table 1). The antibody titers against each recombinant PvCSP variant (yPvCSP-VK210, yPvCSP-VK247, yPvCSP-P. vivax-like) were measured by ELISA twelve days after the administration of each dose. As shown in Fig. 2, no statistically significant difference was observed in the antibody titers elicited by the two recombinant proteins (NLP-CSP CT or NLP-CSP R ) when combined with the same adjuvant (NLP-CSP CT or NLP-CSP R in the presence of Poly (I:C) for yPvCSP-VK210 p = 0.9795, yPvCSP-VK247 p = 0.9608 and yPvCSP-P. vivax-like p = 0.9994, and NLP-CSP CT or NLP-CSP R in the presence of Montanide ISA 720 for yPvCSP-VK210 p > 0.9999, yPvCSP-VK247 p = 0.9925 and yPvCSP-P. vivax-like p = 0.9994).
In contrast, significant differences in IgG titers were observed when we compared the effects of the different adjuvants. With the Poly (I:C) adjuvant, the titers of IgG specific for all three variants were detectable (~ 10 4 ) in both immunized groups only after the administration of two doses. With the administration of the third dose, IgG titers were greater than 10 5 in all immunized mice. When the same analysis was performed for groups immunized with Montanide ISA 720, antibody titers reaching 10 4 levels were detected after the administration of a single dose. Moreover, after the administration of the second and third doses, even higher IgG titers (10 6 ) against all three PvCSP variants were detected (Fig. 2). www.nature.com/scientificreports/ When antibody titers were compared after the administration of three doses using the same recombinant protein but with a different adjuvant, all groups of mice immunized with Montanide ISA 720 had significantly higher specific IgG titers than mice immunized with Poly (I:C) (NLP-CSP CT to yPvCSP-VK210 p > 0.0001, yPvCSP-VK247 p > 0.0001 and yPvCSP-P. vivax-like p = 0.0004, and NLP-CSP R to yPvCSP-VK210 p > 0.0001, yPvCSP-VK247 p > 0.0001 and yPvCSP-P. vivax-like p = 0.0004).
We also analyzed the longevity of the IgG antibodies. As shown in Suppl. Figure 1, PvCSP-specific IgG titers remained higher than 10 4 in the Poly (I:C) adjuvant-treated groups and on the order of 10 5 in the Montanide adjuvant-treated groups for at least 102 days after priming. These responses were antigen-specific, since mice immunized only with either adjuvant did not elicit detectable PvCSP-specific IgG antibodies at any time point analyzed (Suppl. Figure 1).
It is worth mentioning that any potential interference of yeast components in the specificity of the produced antibodies was discarded in previous studies, by using recombinant proteins produced in bacteria as solid phasebound antigens 22,23 . Preclinical immunization-challenge model: generation of a chimeric P. berghei line expressing P. vivax-like CSP in sporozoites (Pb-PvCSP-like G10). In previous studies, we generated two chimeric P. berghei parasites (Pb-PvCSP210 and Pb-PvCSP247) that express the full-length VK210 and VK247 variants of P. vivax CSP on the spz surface 25 . These parasites have been used to analyze protective efficacy in mice immu-  Table 1. IgG antibody titers were determined using ELISA assay at 12 days after the administration of each dose using the individual PvCS proteins (yPvCSP-VK210, yPvCSP-VK247 and yPvCSP-P. vivax-like) as solid-phase bound antigens. www.nature.com/scientificreports/ nized with different vaccine candidates targeting PvCSP with an immunization-challenge mouse model [41][42][43] .
To analyze protective immune responses induced by the two recombinant proteins NLP-CSP CT or NLP-CSP R against not only the VK210 and VK247 repeats but also against the P. vivax-like variant, we generated a third chimeric P. berghei parasite that expressed the PvCSP-P. vivax-like protein in spz. We first generated a chimeric P. berghei ANKA line in which the endogenous P. berghei csp gene was replaced with the PvCSP-P. vivax-like gene using the same GIMO transfection approach that was used to generate the chimeric Pb-PvCSP210 and Pb-PvCSP247 lines (Suppl. Figure 2 and Fig. 4a) 25 . Two independent clones of this Pb-PvG10(r) line (2710cl1, 2710cl2) produced normal numbers of oocysts. However, in contrast to the other two chimeric lines that express PvCSP-VK210 and PvCSP-VK247, the parasites expressing the PvCSP-P. vivax-like protein did not form visible spz inside oocysts, and only very few salivary gland spz were observed (Table S2). The cause of the lack of sporozoite formation is unclear, but the CSP of different Plasmodium species do not always fully complement the function of the CSP of other species. For example, chimeric P. falciparum replacement lines expressing P. vivax CSP (VK210 and VK247) also have a defect in the formation of spz 44 . An alternative strategy is the generation of rodent-infectious P. berghei and P. falciparum spz which can be engineered to express CSP proteins on the spz surface from two different Plasmodium species [45][46][47] . Based on this observation, we decided to generate chimeric P. berghei spz expressing both the PbCSP and the PvCSP-P. vivax-like protein. In this chimeric line (Pb-PvCSP-like G10, 2700 cl1), the PvCSP-P. vivax-like gene is introduced into the genome as an additional copy of the csp gene in the neutral 230p locus using GIMO transfection (Figs. 3, 4a,b). The PvCSP-P. vivax-like gene is flanked by the 5′ and 3′ promoter and transcription terminator sequences of the P. berghei uis4 gene, which is specifically expressed in spz and liver stages 48 . Pb-PvCSP-like G10 parasites showed normal asexual blood stage multiplication in mice (data not shown), and both oocyst and sporozoite production in A. stephensi mosquitoes was comparable to wild-type P. berghei parasites (Table S2). The infectivity of chimeric spz, as determined by the length of the T1% period after an intravenous injection of 1000 spz in BALB/c mice, was similar to the T1% period in mice infected with wild-type P. berghei spz (Table S2). These results demonstrate that the chimeric Pb-PvCSP-like G10 parasites produce fully infectious spz that are able to complete liver stage development in mice. The expression of the PvCSP-P. vivax-like protein in Pb-PvCSP-like G10 spz was determined by immunofluorescence analysis using sera from mice immunized with the recombinant proteins. The 3D11 antibody recognizing P. berghei CSP was used as a control. Pb-PvCSP-like G10 spz stained both with the antiserum and the 3D11 antibody (Suppl. Figure 3), demonstrating the expression of both P. berghei CSP and PvCSP-P. vivax-like protein in spz of the Pb-PvCSP-like G10 line.
Protective efficacy in mice immunized with different vaccine candidates targeting PvCSP variants using the immunization-challenge mouse model. To analyze protective immune responses induced in the mice by immunization with the recombinant proteins NLP-CSP R and NLP-CSP CT , we challenged mice with spz of the three chimeric P. berghei lines, Pb-PvCSP210, Pb-PvCSP247 and Pb-PvCSP-like G10. The results are shown in Fig. 5; Pb-PvCSP210 challenge in the upper panel (Fig. 5a-d), Pb-PvCSP247 challenge in the middle panel (Fig. 5e-h) and Pb-PvCSP-like G10 challenge in the lower panel (Fig. 5i-l).   Table S1, while the expected PCR product sizes and the primer numbers are listed in the table below the PCR analysis. As an additional control (ctrl), parasite line 2117cl1 (right panel) was used to validate the primers used to amplify the hdhfr::yfcu SM that was integrated into chr. 3 but has been removed from 2700cl1. www.nature.com/scientificreports/ Groups of six C57BL/6 mice were immunized as described in Table 1 and challenged 2 weeks after the last immunization by administering an intravenous (i.v.) injection of 2000 spz Pb-PvCSP210. The protective efficacy was assessed by determining the T1% period after challenge (Fig. 5a, b) by following parasitemia (Fig. 5c, d); p values denoting significant differences in the T1% period between control (adjuvant) and immunized mice are depicted in Table 2. With either adjuvant, a statistically significant delay in the T1% was observed in mice immunized with NLP-CSP CT and NLP-CSP R compared to the adjuvant-only groups. However, a significant difference between the protective efficacy of mice immunized with NLP-CSP CT or NLP-CSP R was only observed in the Montanide adjuvant-treated groups (p = 0.0192, Table 2), showing higher protective efficacy of the NLP-CSP CT protein.
Other groups of C57BL/6 mice immunized as described in Table 1 were challenged 2 weeks after the last immunization by administering an i.v. injection of 2000 spz of Pb-PvCSP247. Percent protection to reach T1% and parasitemia are shown in Fig. 5 (e,f and g,h, respectively). Mice immunized with formulations containing the Montanide adjuvant presented a significant delay in the T1% period compared to the adjuvant-only groups ( Fig. 5f and Table 2, p = 0.0005). Additionally, a significant delay in the T1% period (p = 0.0049) was observed in mice immunized with NLP-CSP CT protein formulated with the Poly (I:C) adjuvant ( Fig. 5e and Table 2).
Finally, a third set of C57BL/6 mice immunized as described in Table 1 were challenged 2 weeks after the last immunization by administering an i.v. injection of 2,000 spz Pb-PvCSP-like G10. Percent protection to reach T1% and parasitemia are shown in Fig. 5 (i, j and k, l, respectively). With either adjuvant, a statistically significant delay in the T1% period in mice immunized with either protein compared to the adjuvant-only groups was observed (p = 0.0007 for Montanide adjuvant-treated groups and p < 0.0015 for Poly (I:C) adjuvant-treated groups, Table 2). As in previous experiments using Pb-PvCSP210 and Pb-PvCSP247 parasites, no significant differences were observed between the naïve (not immunized, infected mice) and adjuvant-only groups (data not shown).

Discussion
The development of an effective vaccine would be an important tool against malaria, as it would provide a costeffective form of prevention and would help circumvent adaptive strategies both from the vector and parasite.
In addition to the general obstacles to overcome when developing vaccines against parasitic diseases, research groups developing vaccines against P. vivax face other issues. One of them is the formation of hypnozoites, which can cause relapses within months and even years after the primary infection 49 . Our vaccine formulations were able to elicit high and long-lasting titers of PvCSP-specific antibodies in mice, providing partial protection against challenge with chimeric P. berghei spz expressing the P. vivax CSP variants. Since a significant proportion of sporozoites released in the challenge were prevented from causing an infection, our formulation could hypothetically contribute to the reduction in cases of relapse, as it would prevent the formation of new hypnozoites in the liver 49 .
Another important obstacle to overcome is that P. vivax does not infect rodents. For this reason, the preclinical evaluation of the protective efficacy of vaccine formulations is mainly restricted to the use of monkeys. In addition to the ethical conflict associated with the use of NHP in such early stage of the vaccine development, these animals must undergo a splenectomy to facilitate the development of parasitemia. This procedure may not provide robust data since organ removal causes immunological changes 50 . Thus, a strategy that enables the preclinical determination of protective immune responses induced by vaccine formulations against P. vivax malaria is based on the use of chimeric parasites expressing P. vivax proteins that are the targets of the vaccine formulations 51 . In particular, the use of transgenic parasites in the study of CSP-based vaccine formulations for the pre-erythrocytic phase of infection has allowed the analysis of functional inhibition of the exogenous CSP, expressed in replacement of the endogenous protein. Following this strategy, chimeric P. berghei parasites expressing the P. vivax CSP variants VK210 and VK247 were used to determine the protective efficacy of vaccine formulations consisting of viral vectors carrying P. vivax CSP alleles 25,[41][42][43] . In this study, we generated for the first time a chimeric P. berghei parasite line expressing the third PvCSP variant, PvCSP-P. vivax-like.
We first attempted to develop a chimeric parasite in which the endogenous P. berghei csp gene was replaced with the P. vivax-like csp gene. Unfortunately, this transgenic parasite failed to produce visible spz inside oocysts, and only very few in the salivary glands of An. stephensi mosquitoes. This failure to complement the function of the endogenous CSP was also observed in transgenic P. falciparum expressing P. vivax CSP as replacement lines 44 . In addition, the chimeric P. berghei Pb-PvCSP247 line produces significantly less salivary gland spz than the Pb-PvCSP210 line 25 . To overcome this concern, we generated an "additional copy" chimeric parasite in which the endogenous P. berghei csp gene is maintained and the P. vivax-like csp gene is expressed under the control of the promoter region of the sporozoite-and liver-specific P. berghei gene uis4. This strategy was successfully applied in previous studies, in which chimeric P. berghei spz have been generated by introducing a P. falciparum csp gene as an additional copy into the P. berghei genome. These chimeric P. berghei spz expressed both PbCSP and PfCSP at their surface [45][46][47] . Additionally, chimeric P. falciparum spz have been generated by introducing a P. vivax csp gene as an additional copy into the P. falciparum genome. These chimeric P. falciparum spz also expressed both PvCSP and PfCSP at their surface 52 , similarly to our results.
The development of the first chimeric P. berghei parasite expressing the PvCSP-P. vivax-like protein allows the preclinical determination of protective immunity of vaccines targeting this mostly neglected CSP variant. The genome of P. vivax-like as a malaria-transmitting parasite in apes has been recently published 53 . Similar to P. knowlesi and P. simium infections in humans, P. vivax-like malaria could currently be considered a zoonotic disease, probably with continuing cross-species exchange of P. vivax between humans and apes in tropical Africa 16 . This hypothesis is based on not only the shared vector species (An. vinckei, An. moucheti, and An. marshallii) but also their low host specificity and high longevity 54 , l) is shown. Significant differences in T1% periods (see Table 2) were analyzed by applying the log-rank (Mantel-Cox) test. www.nature.com/scientificreports/  55 . These results prompted us to propose that human-to-human transmission is very likely. The aim of this work is not to elucidate whether human-infective P. vivax-like is a P. vivax allelic variant or a different species causing zoonosis; however, regardless of the origin and classification, we cannot continue to neglect actions to combat it. Therefore, in this work, we developed the first chimeric P. berghei parasite expressing the PvCSP-P. vivax-like protein, and used these parasites to analyze protective immunity in mice immunized with recombinant proteins representing all three P. vivax CSP variants. Our recombinant proteins include the repeats of the PvCSP VK210 and VK247 variants and a sequence representing P. vivax-like CSP. Thus, our formulations would be predicted to be effective against a broad spectrum of cases of vivax malaria, caused not only by P. vivax, but P. simium and P. vivax-like as well. This could represent an improvement when compared to formulations such as VMP001 56 and Rv21 25 , which contain only sequences of PvCSP VK210 and VK247, as would confer higher protection against P. vivax-like infections, than that expected to be achieved through cross-reactivity from other CSP allelic variants. Moreover, our immunization data show that our vaccine formulations stimulated the production of high titers of specific antibodies against each of the variants. In previous studies, it was demonstrated the absence of significative cross-reaction or antigenic interference among the PvCSP-repeat sequences in animals immunized with individual (VK210, VK247 and P. vivaxlike) recombinant proteins, produced in bacteria 57 or yeast 22 . Besides it, formulations containing the chimeric fusion protein, comprising epitopes of all three different allelic forms, were as immunogenic as the mixture of three individual PvCSP proteins. Thus, the specific response to PvCSP RI, repeats and C-terminal regions combined in our formulations might contribute substantially to enhancing protective efficacy, since specific antibodies against these regions are highly neutralizing 41 , thus indicating the importance of a universal formulation.
Although our vaccine formulations did not confer sterile protection after challenge with chimeric spz, the significant delay in T1% periods was noteworthy for several reasons, as described below.
i. Each day of delay in blood-stage parasitemia is representative of ~ tenfold fewer sporozoites reaching the liver 58 , and as previously discussed, this decrease would impact hypnozoite formation, thus preventing relapses 49 . ii. Due to technical reasons, we used our previously established i.v. challenge system, which does not allow us to consider the effect of specific CSP antibodies that potentially act in the skin. In the case of Plasmodium infections, these types of antibodies were recently shown to contribute significantly to protective effects 59,60 . iii. In natural infections, most mosquitoes inoculate only ~ 1% of the sporozoites in their salivary glands, with median inocula ranging between ~ 40 and 100 sporozoites 61 . Supporting this, it was demonstrated that a natural infection with 8 P. berghei-infected mosquitoes is equivalent to i.v. inoculation of 250-500 sporozoites 60 . Therefore, our challenge system (i.v. inoculation of 2000 spz) is stronger than other strategies using intradermal or s.c. challenge. The variation in the challenge system also explains the apparent discrepancies in protective efficacy comparing our results with previous studies 23  www.nature.com/scientificreports/ in the i.v. system). In agreement, a similar situation was observed comparing the protective performance of previous PvCSP-based formulations using s.c. 22 and i.v. 43 challenges, respectively. Taking into account all these facts, the effectiveness of the formulations developed in this study might be even greater when considered in the case of natural infection. Consistent with a previous study 43 , a clear positive relation was observed between high titers of CSP-specific antibodies and protection. In a comparison of both recombinant proteins, the delay in the T1% period obtained with NLP-CSP CT was longer than with NLP-CSP R when administered with the same adjuvant. Most likely, antibodies against the C-terminal domain of PvCSP, which is absent in NLP-CSP R , are responsible for the differences, as this region might be important for protection 41 . The lack of anti-C-terminal antibodies and the lower titers against VK247 repeats in vaccines using Poly (I:C) as an adjuvant reported in previous work 23 would explain the lack of protection provided by NLP-CSP R /Poly (I:C) against challenge with Pb-PvCSP247 parasites. By the other hand, the C-terminal region of PvCSP contains predicted T cell epitopes, which could also contribute to protective effect of pre-erythrocytic vaccines 40 . However, in previous studies, very low levels of PvCSP-specific CD4 + and CD8 + T cell responses were elicited by our PvCSP recombinant proteins 22,57 . Results from the VMP001 clinical trial also indicate the low contribution of repeats and C-terminal region to induce T cell responses, as only 17% of vaccinated subjects responded to these antigens whereas 90% showed strong cellular responses to the N-terminal region 56 (absent in NLP-CSP proteins). Moreover, it was shown that PvCSP short repeat-region peptides, when presented on a VLP, can induce antibodies mediated protection 41 . Nonetheless, we do not exclude the participation of T cell-mediated immune responses in the protection observed in this study.
Finally, Montanide ISA720 was overall a better adjuvant in terms of IgG titers and protective efficacy. However, phase I clinical trials showed some concerns regarding the safety of Montanide ISA720-adjuvanted vaccines against malaria, particularly high reactogenicity [62][63][64] .
For all these reasons, we aim to analyze the mechanisms underlying the observed protection before moving into clinical testing of safety and toxicology. Ongoing research analyzing the specificity of humoral and cellular responses and performing transcriptomic analysis of lymphocytes from immunized mice will provide insights into the pathways that are selectively activated by these formulations and will provide valuable information about the type of immune response that a protective vaccine against vivax malaria should elicit.  Table 2. Prepatent (T1%) periods and statistical significance of protective efficacy. Significant differences (p < 0.05) were analyzed by applying the log-rank (Mantel-Cox) test. n = 6 mice/group. n.s.: not significant differences.