Fig 1.
α-tag sequencing in a longitudinal cohort in Kalifabougou, Mali. (A) Clinical malaria frequency in a cohort of ~600 individuals aged 3 months to 45 years measured every 2 days in 2019, diagnosed by axillary temperature ≥37.5 °C and ≥2,500 asexual parasites per μl of blood. (B) P. falciparum infection prevalence determined by qPCR at the beginning (startDry) and end (endDry) of the dry season, and midway in the wet season (midWet). Columns are sorted to show the same individual in a single row at all time points. (C) Parasitaemia in 104 RDT+ samples from startDry (n = 30), endDry (n = 28), and midWet (n = 24) asymptomatic infections, and wet season clinical cases (MAL, n = 22) by qPCR of varATS relative to a standard curve. (D) Proportion of individuals with different complexities of infection (COI) measured as number of P. falciparum ama1 haplotypes in 79 asymptomatic infections (startDry n = 20, endDry n = 23, midWet n = 15) and 21 clinical malaria cases p-value = 0.008 (E) Demographics and culture time of samples passing QC in DBLα-tag sequencing (F) DBLα-tag read abundance in samples of 7 individuals compared to abundance of the same transcript sequences measured by qPCR relative to a housekeeping gene (Fructose-bisphosphate aldolase, PF3D7_1444800), colour highlights timepoint of collection and shape indicates individual. (G) Number of different DBLα-tags observed (0.96 similarity) in clinical cases (n = 24), and asymptomatic infections (startDry n = 17, endDry n = 24, midWet n = 17). C & G Data shows median and IQR, dot size indicates participant age. C, D and G Kruskall-Wallis test with Bonferroni multiple comparison correction.
Fig 2.
Malaria cases express more var genes than dry season asymptomatic infections.
(A) Correlation between number of DBLα-tags and parasitaemia measured by varATS-qPCR. (B) Correlation between the number of DBLα tags and COI measured by ama1-amplicon sequencing. (C) Proportion of most abundant DBLα-tag at the start (startDry n = 17) and end (endDry n = 24) of the dry season, in clinical cases (MAL n = 24), and midway in the wet season (midWet n = 17). Median and IQR are shown, dot size shows participant age, Kruskall-Wallis test with Bonferroni multiple comparison correction. (D) Individual distribution of DBLα-type annotated by best Blast hit in a reference var database [56]. Each column represents a person, column sub-divisions show individual DBLα-tags.
Fig 3.
Similar var gene types and PfEMP1 binding phenotypes predicted in asymptomatic infections and clinical malaria cases.
(A) Proportion of DBLα-tags annotated to PfEMP1 binding type based on domain composition predicted by Varia [56]. Genes were classified as EPCR-binding if the N-terminal CIDR domains was predicted as a non-var1 CIDRα1 domain, as CD36-binding if a CIDRα2 – CIDRα6 domain was predicted, and as unknown, if prediction corresponded to var1 or a CIDRβ, γ or δ was predicted. (B) Proportion of DBLα-tag reads mapping to var genes predicted by Varia to encode CD36- (left) or EPCR- (right) binding PfEMP1s in the beginning (startDry n = 17) and end (endDry n = 24) of the dry season, during clinical malaria (MAL n = 24), and midway (midWet n = 17) in the wet season. (C) Proportion of DBLα-tags mapping to predicted group A (left), B (centre), and C (right) var genes using cUPS in the beginning (startDry n = 17) and end (endDry n = 24) of the dry season, in clinical malaria (MAL n = 24), and midway (midWet n = 17) in the wet season using cUSP. (D) Proportion of DBLα-tags mapping to predicted group A (left), B (centre), and C (right) var genes using cUPS in the beginning (startDry n = 17) and end (endDry n = 24) of the dry season, in clinical malaria (MAL n = 24), and midway (midWet n = 17) in the wet season using upsAI. (E) Normalized expression of var subtypes in the beginning (startDry n = 15) and end (endDry n = 24) of the dry season, malaria cases (MAL n = 15), and wet season asymptomatic infections (midWet n = 10) by qRT-PCR. DBLα2/1.1/2/4/7/9 MAL vs endDry (p-value 0.0083. Median and IQR are shown, dot size show participant age; Kruskall-Wallis with Bonferroni correction.
Fig 4.
Asymptomatic parasites show lower var transcript levels than clinical cases.
(A) varATS relative expression in asymptomatic infections (startDry n = 17, endDry n = 16, midWet n = 14) and malaria cases (MAL n = 19), measured by qRT-PCR. Samples are divided by time point (left) and clinical status (right). (B) Relative expression of ncRNA ruf6 in asymptomatic infections (startDry n = 17, endDry n = 16, midWet n = 14) and malaria cases (MAL n = 19) measured by qRT-PCR. Samples are divided by time point (left) and clinical status (right). Median and IQR are shown, Kruskall-Wallis test with Bonferroni correction. (C) Correlation of varATS and ruf6 expression. (D) Correlation of number of DBLα-tags expressed and relative expression of varATS (left) and ruf6 (right).
Fig 5.
Antibody recognition of iRBC VSAs correlates negatively with the number of expressed DBL
α-tags. (A) Plasma anti-AMA1 antibody levels in 54 samples (startDry n = 17, endDry n = 10, midWet n = 15, MAL n = 12) measured by ELISA. (B) Surface recognition of PfFCR3 iRBCs by plasma of 54 samples (startDry n = 17, endDry n = 10, midWet n = 15, MAL n = 12) measured by flow cytometry. Gray dashed line represents average signal from naïve plasmas (n = 2) (C) Proportion of individuals with antibodies specific to PfEMP1 domains from asymptomatic individuals collected at start (n = 17), end of the dry season (n = 16), clinical malaria cases (n = 11) and mid wet season asymptomatic infections (n = 15). Recognition defined as MFI greater than the level in 6 malaria-naïve controls + 2sd in Luminex. Kruskall-Wallis test with Bonferroni multiple comparison correction. (D, E) Correlation of plasma antibody levels measured by AMA1 ELISA (D) and Surface recognition assay (E) with number of DBLα-tag in the same sample. (F) Correlation of proportion of domains recognised and number of DBLα-tags in the same samples.
Fig 6.
Single Cell RNAseq data supports restricted expression of var genes in dry season asymptomatic infections.
(A) Heatmap and hierarchical clustering of z-score values of whole transcriptome nucleotide similarities between single iRBCs of one dry season sample (end-dry, n = 51) and one malaria case (MAL, n = 29), that additionally was in vitro cultured for ~16h (MAL16h, n = 31). Bars next to the heatmap indicate predicted stage, collection timepoint, var expression status and var cluster for each cell. (B) Summary of var domain coverage in var-expressing cells of var domains with >1000 normalized reads in at least one cell within the dry season sample (n = 48). (C) Summary of var domain coverage in var-expressing cells of var domains with >1000 normalized reads in at least one cell of the malaria case’ sample (n = 41). Cells with similar var domain coverage are clustered, cells sequenced ex vivo and after 16h of culture are indicated by circles and crosses, respectively.
Table 1.
PCR cycling conditions.
Table 2.
Starting samples and filtering steps.
Table 3.
Primer mix details for different var domain types.