Clinical expression and antigenic profiles of a Plasmodium vivax vaccine candidate: merozoite surface protein 7 (PvMSP-7)

Abstract

Background: Vivax malaria is the predominant form of malaria outside Africa, affecting about 14 million people worldwide, with about 2.5 billion people exposed. Development of a Plasmodium vivax vaccine is a priority, and merozoite surface protein 7 (MSP-7) has been proposed as a plausible candidate. The P. vivax genome contains 12 MSP-7 genes, which contribute to erythrocyte invasion during blood-stage infection. Previous analysis of MSP-7 sequence diversity suggested that not all paralogs are functionally equivalent. To explore MSP-7 functional diversity, and to identify the best vaccine candidate within the family, MSP-7 expression and antigenicity during bloodstream infections were examined directly from clinical isolates.

Methods: Merozoite surface protein 7 gene expression was profiled using RNA-seq data from blood samples isolated from ten human patients with vivax malaria. Differential expression analysis and co-expression cluster analysis were used to relate PvMSP-7 expression to genetic markers of life cycle stage. Plasma from vivax malaria patients was also assayed using a custom peptide microarray to measure antibody responses against the coding regions of 12 MSP-7 paralogs.

Results: Ten patients presented diverse transcriptional profiles that comprised four patient groups. Two MSP-7 paralogs, 7A and 7F, were expressed abundantly in all patients, while other MSP-7 genes were uniformly rare (e.g. 7J). MSP-7H and 7I were significantly more abundant in patient group 4 only, (two patients having experienced longer patency), and were co-expressed with a schizont-stage marker, while negatively associated with liver-stage and gametocyte-stage markers. Screening infections with a PvMSP-7 peptide array identified 13 linear B-cell epitopes in five MSP-7 paralogs that were recognized by plasma from all patients.

Conclusions: These results show that MSP-7 family members vary in expression profile during blood infections; MSP-7A and 7F are expressed throughout the intraerythrocytic development cycle, while expression of other paralogs is focused on the schizont. This may reflect developmental regulation, and potentially functional differentiation, within the gene family. The frequency of B-cell epitopes among paralogs also varies, with MSP-7A and 7L consistently the most immunogenic. Thus, MSP-7 paralogs cannot be assumed to have equal potential as vaccines. This analysis of clinical infections indicates that the most abundant and immunogenic paralog is MSP-7A.

Keywords: Antigen; Clinical isolates; Epitope; Malaria; Plasmodium vivax; PvMSP-7; Transcriptomics; Vaccine.

Fig. 1

Principal component analysis (PCA) of genome-wide expression profiles for ten clinical isolates (5838 genes). The principal components were calculated from normalized read counts implemented in DESeq2. Axes represents the first and second principal components (PC1 and PC2), accounting for 42% and 27% of total variance, respectively. Each dot represents an individual isolate, and isolates were divided into four ‘patient groups’ based on the PCA

Fig. 2

Expression profiles for 12 PvMSP-7 paralogs in ten clinical isolates. The heat-map describes log2-transformed normalized read counts from the DESeq2 analysis; shading reflects transcript abundance from low (red) to high (black) shading. Clustering in the y-axis dendrogram reflects similarity in profile between genes. Clustering in the x-axis dendrogram indicates the similarity of PvMSP-7 expression profile between patients; for instance, patient group 4 isolates display a very similar profile for PvMSP-7

Fig. 3

Expression profiles of differentially expressed genes (DEGs) in a comparison of patient groups 3 and 4 (n = 1493). The heat-map shows expression profile across five clinical isolates, generated from log2-transformed normalized read counts. Clustering in the y-axis dendrogram reflects similarity in profile between DEGs. Clustering in the x-axis dendrogram indicates the similarity of differential expression between patients. The heat-map is divided into five sectors based on y-axis clustering. For each sector, the results of enrichment analysis of Gene Ontology (GO) terms are shown on the right. GO terms with significant enrichment within a given heat-map sector are listed with their description, ontological category [i.e. biological process (BP), molecular function (MF), cellular compartment (CC)], frequency in the background gene set (‘size’) and incidence within the heat-map sector (‘count’). Significance threshold was p-value < 0.05 after Bonferroni correction

Fig. 4

Co-expression analysis of genes differentially expressed between patient groups 3 and 4. Coseq was used to generate clusters of genes with positively correlated transcription profiles. 1493 DEGs formed 14 clusters; five clusters with dynamics that clearly distinguish the two patient groups are shown here. a, b Concern transcripts that are abundant in patient group 4 and positively correlated with PvMSP-7 expression. c–e Concern transcripts that were abundant in patient group 3, and were negatively correlated with PvMSP-7l. Boxplots (left) represent expression values for all genes in a cluster (Log2-transformed, normalized read counts) from an individual patient, shaded by patient group; the black line connecting boxplots is the mean expression value of all genes in the cluster. a PvMSP-7K, 7I, 7H and 7C (only 7H is shown) were co-expressed with 91 other genes including various invasion-related genes such as ETRAMP, SEA1, and RON4. b A fifth MSP-7 gene, PvMSP-7L, was co-expressed with 125 other genes including PMV, MOP, and RON5. c A gametocyte stage-specific marker, gamete antigen 27/25, was absent in patient group 4 but positively correlated with TRAG36 and PIR genes. d A cohort of 168 genes that are abundant in patient group 3 but absent in group 4. e Another stage-specific marker, liver stage antigen 3, is abundant in patient group 3 and co-expressed with 204 other genes. Pearson’s correlation coefficient (r) for abundance values of MSP-7H and selected genes are shown

Fig. 5

Mapping of putative B-cell epitopes across 12 PvMSP-7 paralogs by peptide microarray. Plasma from naturally infected vivax malaria patients was applied to a custom peptide microarray to determine linear B-cell epitopes. The microarray consisted of 2346 spots, representing 1173 peptides (each 13 amino acids) derived from the predicted protein sequences of 12 PvMSP-7 paralogs, overlapping by one amino acid and spotted in duplicate. a–e Fluorescent response following incubation of the peptide microarray with patient plasma, pooled by age in years: 0–14 (a), 15–29 (b), 30–44 (c), 45–59 (d), 60–74 (e), and application of a Cy3 (red) conjugated, secondary goat anti-human IgG antibody. Each spot on the peptide microarray corresponds to one peptide. f Negative control consisting of plasma derived from naïve patients living in malaria-free areas. Control peptides (HA) are located around the edge of the array and are stained with Cy3 (red) conjugated anti-HA antibody. For comparison, the strongest responding spots are highlighted between the white lines. g A Venn diagram showing the overlap in predicted linear B-cell epitopes across five patient age groups. 13 peptides that gave significant responses in all experimental groups were observed to be present in all age groups. h Peptides that gave fluorescent responses significantly greater than the naïve control in one or more age groups (N = 236) are presented in order of response intensity and shaded by PvMSP-7 paralog. The predicted amino acid sequence of the epitope is given for 13 peptides that were significant in all five age groups

Fig. 6

Observed linear B-cell epitopes mapped to predicted secondary protein structures of 12 PvMSP-7 proteins. The cartoon depicts the intensity and position of linear B-cell epitopes identified by microarray within each PvMSP-7 protein. The x-axis shows the secondary structure of each paralog to scale, as predicted by JPred4 [65]: coiled-coil (grey line), alpha-helix (red lozenge), beta-strand (blue arrows). The intrinsically unstructured/disordered regions (blue zig-zag line) was predicted using the GeneSilico MetaDisorder service [66]. The positions of 236 peptides giving significant responses from the microarray are plotted; the y-axis measures the epitope coverage of a given amino acid position (maximum of four epitopes). The positions of 13 ‘universal’ epitopes observed in all assays (see Fig. 5g) are marked by black bars and their corresponding epitope sequences

Fig. 7

Comparison of MSP-7 expression profiles from published transcriptomes of synchronized cell cultures in multiple Plasmodium species. a Expression profiles for 12 PvMSP-7 genes in P. vivax, as determined by microarray analysis [25]. b Expression profiles for seven PfMSP-7 genes in P. falciparum across seven blood-stages, as determined by cDNA sequencing [56]. c Expression profiles for four PbMSP-7 genes in P. berghei, across five life stages, as determined by RNA-seq [57]. All values are log2-transformed. Shading represents transcript abundance from low (red) to high (black). An asterisk and hash indicate two sets of orthologous genes in different species respectively