A comprehensive epigenome map of Plasmodium falciparum reveals unique mechanisms of transcriptional regulation and identifies H3K36me2 as a global mark of gene suppression

Abstract

Background: Role of epigenetic mechanisms towards regulation of the complex life cycle/pathogenesis of Plasmodium falciparum, the causative agent of malaria, has been poorly understood. To elucidate stage-specific epigenetic regulation, we performed genome-wide mapping of multiple histone modifications of P. falciparum. Further to understand the differences in transcription regulation in P. falciparum and its host, human, we compared their histone modification profiles.

Results: Our comprehensive comparative analysis suggests distinct mode of transcriptional regulation in malaria parasite by virtue of poised genes and differential histone modifications. Furthermore, analysis of histone modification profiles predicted 562 genes producing anti-sense RNAs and 335 genes having bidirectional promoter activity, which raises the intriguing possibility of RNA-mediated regulation of transcription in P. falciparum. Interestingly, we found that H3K36me2 acts as a global repressive mark and gene regulation is fine tuned by the ratio of activation marks to H3K36me2 in P. falciparum. This novel mechanism of gene regulation is supported by the fact that knockout of SET genes (responsible for H3K36 methylation) leads to up-regulation of genes with highest occupancy of H3K36me2 in wild-type P. falciparum. Moreover, virulence (var) genes are mostly poised and marked by a unique set of activation (H4ac) and repression (H3K9me3) marks, which are mutually exclusive to other Plasmodium housekeeping genes.

Conclusions: Our study reveals unique plasticity in the epigenetic regulation in P. falciparum which can influence parasite virulence and pathogenicity. The observed differences in the histone code and transcriptional regulation in P. falciparum and its host will open new avenues for epigenetic drug development against malaria parasite.

Keywords: Chromatin; Genome-wide mapping; Histone modifications; Plasmodium; Transcription; Virulence and pathogenicity genes.

Fig. 1

Generation of comprehensive epigenomic maps of histone modifications at three different stages of intra-erythrocytic stages of P. falciparum life cycle. a Schematic representation of different stages of Plasmodium falciparum harvested at 18, 30 and 40 h post-synchronization. A single large-scale culture of highly synchronized cells was used for ChIP-sequencing of all histone modifications. b A snapshot of the representative genome browser shows the histone methylation and acetylation, and RNA sequencing reads at trophozoite stage. Representative active chromatin (presence of RNA-seq reads), inactive chromatin (presence of H3K9me3 and absence of RNA-seq reads) and poised chromatin (present of histone activation marks but no RNA-seq reads) are highlighted in green, red and blue color, respectively. Anti-IgG antibody was used as a negative control. c Heat map representing the correlation between occupancy of indicated histone modifications in the gene body for all P. falciparum genes. Active promoter marks; H3K4me2, H3K4me3, H3K9ac and H3K14ac clustered separately to form an active chromatin state (green square). Histone variant H2A.z and other histone modifications; H4K20me3, H3K79me3, H3K36me3, H3K27ac, H3K4me1 and H4ac are clustered together to form active/poised chromatin state (blue square). H3K9me3 and H3K36me2 are anti-correlated with each other suggesting mutually exclusive function for these two histone modifications

Fig. 2

Comparison of histone modification profiles of P. falciparum and human genes. Normalized mean tag density gene profile of histone modifications over 5265 P. falciparum and 41,390 human genes. Transcription start site (TSS) and transcription termination site (TTS) are highlighted in green and red color, respectively. Error bars (black) standard error of the mean. Multiple histone modifications exhibit differential profiles in P. falciparum than in human cells suggesting differential mode of gene expression in the malaria parasite

Fig. 3

Correlation of histone modifications and transcription in P. falciparum. a Genes were categorized based on increasing expression levels (based on reads assigned per kilobase of target per million mapped reads (RPKM); enrichments were collected at corresponding genes and Whisker plots were plotted for each category. Whisker plots representing the enrichment of histone modifications (H3K4me2, H3K4me3, H3K9ac, H3K14ac, H4ac, H3K79me3, H4K20me3, H3K4me1, H3K9me3, H3K36me2 and H3K36me3) at the trophozoite stage. Pearson correlation (r) was calculated between the median of gene expression and histone enrichment. Gene expression positively correlates with all histone modifications. Least correlation with transcription was observed with H3K9me3 and H3K36me3 as compared to other histone modifications. b Genes were categorized based on decreasing expression levels (based on RPKM), enrichments were collected at corresponding genes and occupancies of multiple potential repressive histone modifications (H3K9me3, H3K36me2 and H3K36me3) to that of activation marks (H3K4me3, H3K9ac and H4ac) were plotted for each category at trophozoite stage. Linear increase in the ratio of activation to repression mark observed with H3K36me2 suggesting that it is a global repressive mark in P. falciparum. c Fold change in expression was calculated in SET4KO, SET8KO and SETvsKO as compared to wild-type conditions and plotted as a function of decreasing transcription and increasing H3K36me2

Fig. 4

Clustering of P. falciparum genes based on histone modification profiles. a Heat map of the signal density using k-means clustering observed on 3750 P. falciparum genes for IgG control and histone modifications (H3K4me3, H3K9ac, H4ac, H3K9me3, H3K36me2, and H3K36me3). Genes were clustered into four distinct categories based on the distribution of histone modifications in the gene body: (1) Cluster 1 (411 genes), peaks at the 5′ end of the gene (red squared and labeled 1), Cluster 2 (562 genes), peaks at the 3′ end of the genes (green squared and labeled 2), Cluster 3 (335 genes), peaks observed at the center of the gene body (blue squared and labeled 3) and Cluster 4 (2442 genes), with very low levels of histone modifications (black squared and labeled 4). b RNA sequencing (RNA-seq) profiles of the respective cluster plotted as average gene unit. RNA-seq reads correlate with the histone modifications. c Levels of naturally anti-sense transcript calculated for cluster 1, 2 and 3 using publicly available data. Cluster 2 shows maximum anti-sense transcript levels as predicted with histone modification profiles. d Sense and anti-sense transcript were measured for PF11_0468 and PF10_0287 genes from cluster 2. Actin (from cluster 1) was used as control. Both these genes showed significant anti-sense transcript production. e UCSC snapshot of a representative gene having histone modifications at the center of the gene unit. The intragenic region was cloned in a dual promoter reporter vector in which mCherry and EGFP were cloned in sense and anti-sense orientation. f Intragenic region showed both EGFP and mCherry expression suggesting bidirectional promoter activity from the two randomly selected genes. Differential levels of mCherry and EGFP between two intragenic sequences indicate differential promoter strength of these sequences. Empty vector backbone was used a negative control

Fig. 5

Histone modification profiles over CVM genes. a Enrichment of various histone modifications and histone variant H2A.z on CVM and 500 ring-expressed genes calculated in a stage-specific manner. b Stage-specific profile of histone modifications across different IEC stages in CVM and ring-expressed genes. H3K9me3 and H4ac showed differential pattern over CVM and ring stage-expressed genes. c Average profiles of H3K9me3, H4ac and H4K20me3 histone modifications over CVM genes, which were found to be differentially distributed between CVM and ring-expressed genes. Error bars (black) standard error of the mean

Fig. 6

Model depicting differential occupancies of histone modifications regulating transcription of P. falciparum genes. Active (a) and inactive (b) genes exhibit similar profiles of histone modifications with H3K4me3 and H3K9ac as dominant marks across the gene body. On inactive genes slight increase in the levels of H3K36me2 and decrease in the levels of H3K4me3 and H3K9ac leads to alteration in the ratio of activation to repression mark, which affects gene expression. c Clonally variant multicopy (CVM) genes have different chromatin modifications. On CVM genes, most prominent activation and repression marks are H4ac and H3K9me3, respectively