The Malaria Cell Atlas: Single parasite transcriptomes across the complete Plasmodium life cycle

Abstract

Malaria parasites adopt a remarkable variety of morphological life stages as they transition through multiple mammalian host and mosquito vector environments. We profiled the single-cell transcriptomes of thousands of individual parasites, deriving the first high-resolution transcriptional atlas of the entire Plasmodium berghei life cycle. We then used our atlas to precisely define developmental stages of single cells from three different human malaria parasite species, including parasites isolated directly from infected individuals. The Malaria Cell Atlas provides both a comprehensive view of gene usage in a eukaryotic parasite and an open-access reference dataset for the study of malaria parasites.

Single-cell RNA-seq references for the Plasmodium genus.

Left: Single-cell transcriptomes from across the life cycle of Plasmodium berghei were profiled (including liver, blood, and mosquito life stages). Center: Deep exploration of blood-stage parasites captured transcriptomic diversity at single-cell resolution across three different Plasmodium species by droplet sequencing. Right: Such datasets can serve as references to understand wild parasites isolated from clinical samples.

Fig. 1. A single-cell atlas of the P. berghei life cycle.

(A) The life cycle begins when an infected mosquito injects sporozoites into the mammalian host. From here, parasites enter the liver, where they develop, replicate, and then egress to enter the IDC. During the IDC, parasites invade erythrocytes, where they develop, replicate asexually, burst, and re-invade erythrocytes cyclically. Sexual forms are taken up by the mosquito, and if fertilization is successful, parasites invade the midgut and subsequently the salivary glands of the mosquito. In these different environments, parasites adopt different cellular strategies: replicative stages (liver stage, schizont, oocyst), invasive stages (merozoite, ookinete, and sporozoite), and sexual stages (male and female gametocytes). (B) UMAP of single-cell transcriptomes sampled from all stages of the life cycle, with cells colored according to their stage from (A). (C) The first three principal components from transcriptomes of all stages in the life cycle.

Fig. 2. Graph-based clustering of genes reveals gene usage throughout the life cycle.

(A) A k-NN force-directed graph of all 5156 detected genes. Each node represents a gene. Nodes are colored according to their graph-based spectral clustering assignment, and each cluster is labeled by cluster number. (B) A heat map of mean expression for each cluster across all cells in the dataset. Cells are ordered by their developmental progression. Shown at the right are the number of genes in each cluster (N), the most significant Gene Ontology term (biological process) associated with the cluster (from data S1, Benjamini P < 0.05), and the top cited gene in the cluster (based on PlasmoDB). In the clusters where there is a tie for lowest P value, the GO term with the greatest percentage of genes in the cluster relative to the background is shown. If terms had identical representation, the term with the lowest GO number is shown (see complete table in data S1). (C) The same graph as in (A) colored according to relative growth rate of knockout mutants in asexual blood-stage parasites (11).

Fig. 3. Alignment of datasets reveals transcriptional rates in the IDC.

(A) P. berghei 10x data mapped to Smart-seq2 data using scmap-cell. Cells are grouped according to their 10x cluster assignment (figs. S13 and S15) and the cluster of the Smart-seq2 cell it mapped to (fig. S15). A cosine similarity threshold of 0.5 led to classification of 283 cells (<6% of cells) as unassigned (UA). (B) PCA of P. berghei IDC cells from 10x. Pseudotime of each cell was measured by fitting an ellipse to the data and calculating the angle (radians) around the center of this ellipse for each cell relative to the start cell (red point). Black points represent the mean PCA coordinates of the bulk prediction for each cell (22) (fig. S14). (C) Left: PCAs of three Plasmodium species colored by their P. berghei cell assignment based on scmap. Arrows represent the relative change in transcriptional state based on RNA velocity. Right: Scaled increase in expression over the IDC. Cells are ordered according to the pseudotime of their scmap-assigned cell in the IDC P. berghei index. The top bar represents the matched time point between the P. berghei RNA velocity–derived transcription rates and P. falciparum transcription rates reported by Painter et al. (21) using Pearson correlations. Vertical gray lines mark peaks and troughs determined from the P. berghei data, as described in the methods.

Fig. 4. The Malaria Cell Atlas enables high-resolution mapping of field-derived single-cell transcriptomes of P. falciparum and P. malariae.

(A) Phylogeny of Plasmodium showing the mammalian host and the stages found in circulation for each species. P. falciparum and P. berghei sequester their late stages in deep tissue, whereas other species have all morphological forms in circulation. Species in color were profiled in the atlas. (B) P. falciparum and P. malariae field-derived cells mapped onto the P. berghei 10x reference index using scmap-cell. The field-derived samples mapped to developmental stages that were expected in circulation for each species.