All the animal studies were performed at the animal facilities of Hospital Clinic in Barcelona in accordance with guidelines and protocols approved by the Ethics Committee for Animal Experimentation of the University of Barcelona CEEA-UB.

Female BALB/c mice, 7 to 9 weeks of age, were used throughout the study. Splenectomized BALB/c mice were obtained from Charles River Laboratories.

The Plasmodium yoelii non-lethal strain 17XNL(1.1) (MRA-593) and the P. yoelii lethal strain 17XL (MRA-680) were obtained from MR4, ATCC® Manassas Virginia. Infections were maintained in Balb/c mice by intraperitoneally (i.p.) injection of 5 × 10⁵pRBCs from the tail blood of donor mice at 5–10% parasitemia. Parasitemia was monitored by Giemsa staining of blood smears.

For immunizations, mice were injected subcutaneously (s.c.) with 10 μg of exosomes and 10 μg CpG ODN-1826. Twenty days after, mice were re-immunized with 5 μg of exosomes. Twenty days after the second immunization, mice were analyzed for spleen cellular responses. In challenge experiments, mice were infected with 5 × 10⁵ P. yoelii 17XL 20 days after the second immunization. Parasitemia was followed using Giemsa-stained blood smears.

Splenocytes were obtained from the spleens of mice immunized with rexC and rexPy on day 20. Briefly, the spleens were homogenized and passed through a nylon mesh to create a single-cell suspension. Recipient mice in transfer experiments received 10⁸ splenocytes re-suspended in 500 μL of phosphate-buffered saline (PBS) by injection into the tail vein.

Reticulocytes were obtained from mice blood collected in EDTA. Blood from non-infected mice or mice infected with P. yoelii 17X strain at 20–30% parasitemia was obtained by intracardiac puncture and passed through a CF11 cellulose filter to remove the leukocyte population (Venkatesan et al., 2012). Reticulocytes were purified by layering them on top of a Percoll/NaCl gradient (1.058–1.096 g/mL). After 250 × g centrifugation for 30 min at 4°C, reticulocytes were collected from the interface of the two Percoll layers. Purified reticulocytes were washed twice and cultured for 24 h at 37°C in DMEM, supplemented with 5 mM glutamine, 5% fetal calf serum, 50 U/mL penicillin, and 50 g/mL streptomycin at 1–3% hematocrit. We obtained 7 ± 3 × 10⁷ reticulocytes from uninfected mice and 3,6 ± 0,6 × 10⁸ reticulocytes from per P. yoelii 17X-infected mice. To remove exogenous exosomes in the culture medium, the fetal calf serum was precentrifuged (100,000 × g overnight).

Exosomes from supernatant of reticulocyte cultures were purified by sequential centrifugations at 500 × g for 30 min, 20,000 × g for 45 min, and 100,000 × g for 2 h at 4°C. The final pellet was resuspended in PBS, filtered through a 0.22-μm membrane, and centrifuged at 100,000 × g for 2 h at 4°C. The pellets were resuspended in PBS, and the protein content was determined by Bradford assay. Routinely, exosomes purity was assessed by ME analysis as previously described (Martin-Jaular et al., 2011). Exosomes from plasmas of P. vivax patients were obtained with the same differential centrifugation method.

Spleen cells were obtained from non-immunized mice, self-cured mice 25 days post P. yoelii 17X infection and immunized mice with rexPy or rexC on day 20 after the second immunization, as described above. Splenocytes were analyzed for the expression of different markers with an LSRFortessa flow cytometer and data were analyzed with FlowJo software. T cells subsets were identified according to lymphocytes' SSC-A/FSC-A profile and labeling with CD4-PerCp or CD8-Alexa Fluor-700 antibodies (Biolegend). The phenotype of CD4⁺ and CD8⁺T cells in spleen was analyzed with a panel of fluorochrome-conjugated antibodies obtained from Biolegend that included: CD62L-Pacific Blue, PD1-PE/Cy7, CD127-PE and CD44-FITC.

An LTQ OrbitrapVelos (Thermo Fisher) was used for performing liquid Chromatography (nanoLCULTRA-EKSIGENT) followed by mass spectrometry (nanoLC-MS/MS). Samples of isolated vesicles (rexPy) were reduced with 10 mM DTT (Dithiothreitol), alkylated with 55 mM iodoacetamide, precipitated by 10% TCA, washed with 100% acetone and reconstituted in 2 μL of 8M urea. Samples were brought to a concentration of 1.6 M urea, 1 μg of trypsin (Sus scrofa) was added and digestion was carried overnight at 37°C. The reaction was stopped with 1% formic acid. Trypsinized samples were injected into a precolumn (C18PepMap-100-Thermoscientific-5 mm-ID300 um-5 um-100A) before being flushed into the analytical column (AcclaimPepMap100-Thermoscientific-15 cm-ID75 um-3 um-100A-C18) and eluted at a flow rate of 400 nL/min with a mobile phase gradient: 0–40% of solvent B in solvent A in the first 80–90 min and then 40–100% of solvent B in solvent A until the finish of the run at 110–120 min (A: 3% Acetonitrile 0.1% Formic acid in water, B: 97% Acetonitrile 0.1% Formic acid in water). The eluate was applied to the nanospray source of the Orbitrap spectrometer and a full scan was acquired for all spectra within the 400–1500 m/z range with a 30.000 resolution and a maximum injection time of 500 ms. The MS/MS was performed in the LTQ and the top 20 most intense peptides were isolated and fragmented by low energy CID, 35% collision energy.

Raw spectral data from Xcalibur™ (Thermo Scientific, v2.1) was searched against a custom database compiled from (i) the RefSeq mouse (Mus musculus) reference proteome with isoforms (51193 entries, downloaded from www.uniprot.org on 25th October 2013), (ii) the predicted proteome of Plasmodium yoelii (5979 entries, downloaded from www.genedb.org in its 2nd version), (iii) the RefSeq reference proteome for Bos taurus (24210 entries downloaded from www.uniprot.org on 18th February 2014), (iv) trypsin from Sus scrofa (Accession P00761 downloaded from www.uniprot.org on 14th October 2013), and (v) keratins from the RefSeq reference proteome for Homo sapiens (236 proteins downloaded from www.uniprot.org on 18th July 2013), totaling 81619 proteins. The predicted proteome for Bos taurus was added to the database to control for the presence of proteins from the culture medium, even though only exosome-depleted FBS was used in the preparation of culture media. The search was done with the Sequest HT algorithm on the Proteome Discoverer™ Software version 1.4.1.14 (Thermo Scientific). Searches were performed with the following parameters: digestion by trypsin, 2 missed cleavage sites allowed, precursor mass tolerance of 10 ppm, fragment mass tolerance of 0.6 Da, oxidation of methionine as the variable modification and carbamidomethylation of cysteine as the fixed modification. The Signal-to-Noise (S/N) threshold was set to 1.5. Percolator was used for PSM validation at 1% false discovery rate (FDR) at peptide level. High confidence peptides (1% FDR) were further filtered at ΔCn ≤ 0.1 and Xcorr greater than 1.5, 2.0, 2.25, 2.5, 2.75, 3.0, 3.2, and 3.4 for charge states 1, 2, 3, 4, 5, 6, 7, and > 7, respectively. Proteins identified by the same set of peptides were grouped under one master protein entry. Finally, proteins detected by two or more peptides and present in at least 2 (out of 4) preparations were considered as the core proteomes, for mouse or parasite, from the infected reticulocyte-derived exosomes (rexPy). “The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD005048.”

The Biological Networks Gene Ontology tool–BiNGO (Maere et al., 2005) was used to perform the analysis of over represented GO terms among the detected proteins in the core proteomics set. The reference ontology file used (go.obo, version 1.2 release-date 26/01/2015) was downloaded from www.geneontology.org. The annotation file for Mus musculus was custom made from the gene_association.mgi file (downloaded from www.geneontology.com, release-date 22/01/2015) and the annotation file for Plasmodium yoelii was custom made using an in-house script to parse GO data downloaded from for the Plasmodium yoelii 17X strain from www.plasmodb.org (version 13.0). The analyses were performed with the Hypergeometric test, using the Benjamini and Hochberg False Discovery Rate correction for multiple testing. P-values under 0.05 were considered as significant.

Human spleens were retrieved from deceased donors between 18 and 60 years old after brain death or circulatory death at the Hospital Clínic of Barcelona in accordance with a protocol approved by the Ethics Committee for Clinical Research of the University of Barcelona (No 041499). Neoplasms affecting the spleen or concomitant infection were exclusion criteria. The organ was conserved in saline solution. A spleen portion of about 4 g was homogenized with a GentleMACS™. After that, the sample was washed and resuspended in RPMI medium. The resulting volume was filtered in a 70 μm nylon cell strainer to obtain single cell suspensions.

exPv were labeled using the PKH67 labeling midi kit (Sigma) according to the manufacturer's protocol with minor modifications. Briefly, 10–50 μg of exosomes diluted in PBS were added to 0.5 ml of Diluent C. 2 μl of PKH67 dye was added to 0.5 ml of Diluent C before being added to the exosomes. The samples were mixed gently for 5 min and labeling was stopped by addition of 2 ml of exosome-free FBS for 2 min, followed by the addition of PBS to fill up the centrifuge tube. Labeled exosomes were washed by ultracentrifugation for 90 min at 100,000 × g. After two additional washes in PBS to remove residual lipid dye, the PKH67 exosomes were resuspended in PBS. Each exosome preparation was stored at 4°C and used within 2 days after labeling.

10⁶ splenocytes resuspended in DMEM containing 5% exosome-depleted FBS were seeded on a 24-well plate. Plates were incubated at 37°C, 5% CO₂, 100% humidity for 1 h. Exosomes labeled with fluorescent PKH67 (5 μg/ml) were added to spleen cells and incubated at 37 or 4°C for different times. Cells were then washed three times with PBS and analyzed on an LSRFortessa flow cytometer and data were analyzed with FlowJo software (Tree Star Inc).

5 × 10⁵ splenocytes obtained from human donors were incubated with medium alone or medium containing exosomes from plasma of P. vivax patients (10 μg/ml). Cells were cultured at 37°C in RPMI supplemented with 10% fetal calf serum, 50 U/mL penicillin, and 50 μg/mL streptomycin. After 72 h, cells were labeled with a panel of antibodies in order to discriminate different types of cells (Figure S1) and analyzed in an LSRFortessa flow cytometer.

We have previously described a rexPy plus CpG-ODN vaccination approach that induced survival after first P. yoelii 17XL infection in immunized BALB/c mice and long-lasting sterile protection in subsequent challenges (Martin-Jaular et al., 2011). To analyze whether the protection was mediated by the spleen, groups of normal (Figures 1A,B) or splenectomized BALB/c mice (Figures 1C,D) were immunized subcutaneously (s.c.) with 10 μg of reticulocyte-derived exosomes from non-infected BALB/c mice (rexC) or rexPy combined with 10 μg CpG-ODN. Twenty days after the first immunization, mice were boosted s.c. with 5 μg of exosomes. Non-immunized mice (NI) were left untreated. Twenty days after the second immunization, all mice were infected with 5 × 10⁵ P. yoelii 17XL lethal parasites and parasitemia was monitored. Protection was achieved in mice immunized with rexPy (Figures 1A,B) but did not occur in splenectomized animals (Figures 1C,D), indicating that the spleen played an essential role in the protective immune responses induced by these exosomes.

To address the role of spleen cells in protection from the lethal challenge, spleens were collected from rexC and rexPy immunized mice 2 weeks after the second immunization and cells were adoptively transferred into naïve animals. Groups of 4 naïve BALB/c mice were transferred with splenocytes as described in the methods section. Recipient animals were challenged with 5 × 10⁵ P. yoelii 17XL lethal parasites 24 h post-transfer and parasitemia monitored. A group of non-transferred mice were used as controls of experimental lethal infections. The animals treated with splenocytes obtained from rexPy-immunized animals were protected from lethal challenge (Figures 1E,F). In contrast, none of the animals receiving splenocytes from untreated animals, nor splenocytes from animals immunized with rexC+CpG, survived the lethal challenge. These data strongly suggested that protection was mediated by splenic cellular immunity.

Having demonstrated that passive transfer of splenocytes from animals immunized with rexPy+CpG conferred full protection against lethal challenge, we evaluated whether exosomes were able to promote the appearance of memory cells in the spleen. Therefore, splenocytes from immunized mice were analyzed 20 days after the second immunization to evaluate the phenotype of the cells elicited by our vaccination protocol. In order to compare the immune response to rexPy+CpG vaccination with immunity to malaria infection, a group of animals were infected with P. yoelii 17X causing a self-resolving reticulocyte-prone infection and analyzed after clearance of parasitemia. Splenocytes from immunized and self-cured mice were stained for different surface markers in order to discriminate between different populations by flow cytometry and percentages of different subsets of T cells were evaluated. We found that percentages of both CD4⁺CD62L⁻ and CD8⁺CD62L⁻cells were increased in spleens of mice immunized with rexPy when compared with mice immunized with rexC or non-immunized animals. Similar results, albeit of higher magnitude, were observed in non-immunized animals recovered from P. yoelii17X experimental infections (Figure 2A). However, after self-cured parasite infection, around 40% of CD4⁺T cells and 60% of CD8⁺T cells expressed the exhaustion marker PD-1, indicating loss of their functional capacity. Percentages of PD-1⁺cells in vaccinated animals were not statistically different from those in mice immunized with rexC or non-immunized animals (Figure 2B).

In order to get insight into T cell subsets after immunization and compared them with the ones after self-cured infection, CD62L and CD127 markers were used in order to differentiate between activated effector (CD62L⁻CD127⁻) and effector memory (CD62L⁻CD127⁺) T cells. Moreover, we investigated the relative contribution of effector memory pools by CD44 and CD62L surface expression. Percentages of all the different populations analyzed were statistically different after a self-cured infection of P. yoelii 17X when comparing with non-immunized mice and mice immunized with rexC or rexPy (Figures 3A,B), indicating differences in the role and relevance of the CD4⁺ and CD8⁺ T cell subsets between natural infection and after exosome immunization. rexPy immunization lead to a significant increase in the percentage of effector memory CD4⁺ T cells (CD62L⁻CD127⁺) compared to non-immunized animals or animals immunized with rexC (Figure 3A). An increasing trend in CD8⁺ T effector memory cells was also observed in rexPy-immunized animals (Figure 3B).

All together, these results confirmed the increase in the percentages of both CD4⁺ and CD8⁺ T CD62L⁻memory cells in the spleens from rexPy-immunized mice and after self-resolved infections. Importantly, PD-1⁺ T cells were found only after P. yoelii17X infection. Both CD4⁺ and CD8⁺ T cells after rexPy immunization had mainly an effector phenotype.

Purified P. yoelii-infected reticulocytes were obtained by removing all contaminating leukocytes followed by a Percoll gradient. Infected reticulocytes were then culture in vitro for 24 h in an exosome-depleted medium. After confirming that the viability was >98%, rexPy were obtained from supernatants following a sequential centrifugation/filtration protocol and analyzed by mass spectrometry as described in material and methods.

Four rexPy preparations from pooled infected animals (n = 3 each) were examined by mass spectrometry and both host and parasite proteins were detected. Although the total number of proteins varied among preparations (Figures 4A,C), there were sets of recurrently identified proteins shared by at least two preparations, to, which we denoted as the core rexPy proteome. From host origin, the core proteome was composed of 77 proteins (Dataset S1), among which four known exosomal markers were detected. Of interest, the transferrin receptor (TFRC) and the heat shock 70 protein (HSPA8), two proteins classically associated to reticulocyte-derived exosomes since their first description (Johnstone et al., 1987), were detected in 4/4 and 3/4 preparations, respectively (Figure 4B). A 40S ribosomal protein, three hemoglobin subunits (alpha, beta-1, and beta-2) and elongation factor 2 were also detected on all 4 preparations (Dataset S1). Several other exosomal markers (Figure 4B) and some MCH I proteins (Dataset S1) were also detected.

An overrepresentation analysis of GO terms on the rexPy core proteome supports the conclusion that the preparations were of vesicular origin as several terms related to extracellular vesicles were significantly overrepresented in the Cellular Component annotation (Dataset S2). Among parasite proteins identified, there was also a variably overlap among preparations (Figure 4C), and the core proteome was composed of 75 proteins. Noticeably, in the list of parasite proteins (Dataset S3), there were several which are exposed by the parasite and, thus, could have antigenic properties, such as rhoptry proteins, serine repeat antigens (SERA), fam-a proteins, YIR proteins, exported proteins of unknown function and the merozoite surface protein 1 (MSP1), which is a “classical” malaria vaccine candidate (Figure 4D). The overrepresentation of GO terms suggested an enzymatic activity linked to the catabolism of proteins by the proteasome complex (Dataset S4). Of interest, enolase and L-lactate dehydrogenase were detected in all four preparations and enolase has been shown to have a “moonlight” function in host-parasite interactions as antibodies against this protein are able to protect mice against infection with P. yoelii (Pal-Bhowmick et al., 2007).

Infections in BALB/c mice with the P. yoelii 17X strain resemble the natural tropism for reticulocytes observed in natural human infections of P. vivax, a major human malaria parasite. To try extrapolating the results obtained in this rodent malaria model to human infections by P. vivax, experiments with human splenocytes were performed. Worth of mentioning, these cells were obtained from transplantation donors, which limited the experiments. Moreover, due to the absence of an in vitro culture system for blood stages of P. vivax, exosomes purified directly from plasma of P. vivax patients were used. To monitor whether exosomes were captured by human splenocytes, purified exPv were labeled with a green fluorescent lipid dye (PKH67) and incubated with spleen cells to assess exPv incorporation into splenocytes by flow cytometry (Figure 5A). As expected, when purified PKH-labeled exosomes were incubated with splenocytes for 30 min to 3 h, splenocytes indeed captured exPv. Moreover, incubation at 4°C abrogated the incorporation, indicating an active uptake (Figure 5B).

To analyze changes in the percentages of different populations of spleen cells in response to exosomes, human splenocytes were incubated with exPv for 72 h and cells were labeled with a panel of antibodies to discriminate different subsets of spleen cells by flow cytometry (Figure S1). We found a statistically significant increase of both the number of CD3⁺ T cells and CD8⁺ T cells in response to exPv stimulation, while no differences in the number of B or NK cells were observed between splenocytes incubated with exosomes or medium alone (Figure 6). An increasing tendency in the number of CD4⁺ T cells was also observed. In addition, exosomes from healthy donors' plasma (exC) were isolated and used as a control of allogeneic stimulation. Incubation with exC also incremented CD3⁺ T cell numbers, but to a much lesser extent: a 22% raise compared to the 80% by exPv (data not shown).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. LM, CF, and HD are named inventors on a patent application covering reticulocyte-derived exosomes as malaria vaccines.