The chromatin bound proteome of the human malaria parasite

Abstract

Proteins interacting with DNA are fundamental for mediating processes such as gene expression, DNA replication and maintenance of genome integrity. Accumulating evidence suggests that the chromatin of apicomplexan parasites, such as Plasmodium falciparum, is highly organized, and this structure provides an epigenetic mechanism for transcriptional regulation. To investigate how parasite chromatin structure is being regulated, we undertook comparative genomics analysis using 12 distinct eukaryotic genomes. We identified conserved and parasite-specific chromatin-associated domains (CADs) and proteins (CAPs). We then used the chromatin enrichment for proteomics (ChEP) approach to experimentally capture CAPs in P. falciparum. A topological scoring analysis of the proteomics dataset revealed stage-specific enrichments of CADs and CAPs. Finally, we characterized, two candidate CAPs: a conserved homologue of the structural maintenance of chromosome 3 protein and a homologue of the crowded-like nuclei protein, a plant-like protein functionally analogous to animal nuclear lamina proteins. Collectively, our results provide a comprehensive overview of CAPs in apicomplexans, and contribute to our understanding of the complex molecular components regulating chromatin structure and genome architecture in these deadly parasites.

Keywords: Plasmodium falciparum; chromatin structure; chromatin-associated proteins; proteome; topological data analysis.

Fig. 1.

In silico comparative analysis of CADs across 12 organisms. (a) k-means clustering of the relative abundance of CADs (Table S1a) among 12 organisms (Fig. S2a, Table S2b). CAD abundance was first normalized for each organism by proteome size and then scaled to the CAD frequency with the highest relative abundance of that CAD, which was given an arbitrary abundance value of 1 (Table S1c). For each cluster, a subset of the GO enriched terms associated with the Pfam domains (FDR <0.01) are shown on the right (Table S1d). Cluster 1 (n=145 CADs); cluster 2 (n=274); cluster 3 (n=183); cluster 4 (n=452); cluster 5 (n=168); cluster 6 (n=64); cluster 7 (n=223); cluster 8 (n=282); cluster 9 (n=168); cluster 10 (n=331); cluster 11 (n=298); cluster 12 (n=166). (b) Relative abundance among 12 organisms of CADs selected from k-means apicomplexan-specific cluster 1 and conserved cluster 4 (Table S1c). Pf, Plasmodium falciparum; Pv, Plasmodium vivax; Tg, Toxoplasma gondii; Tb, Trypanosoma brucei; Tc, Trypanosoma cruzi; Lm, Leishmania major; Sc, Saccharomyces cerevisiae; Sp, Schizosaccharomyces pombe; Hs, Homo sapiens; Ce, Caenorhabditis elegans; Dm, Drosophila melanogaster; At, Arabidopsis thaliana.

Fig. 2.

Overview of CAPs in P. falciparum. (a) In silico methodology. A total of 1190 proteins (20.1 % of proteome) in P. falciparum were predicted to be CAPs, covering 1629 unique CADs (Table S2a). (b) Structural organization of the domains that were found in eight or more candidate CAPs (Table S2b).

Fig. 3.

Stage-specific enrichment of CAPs and CADs. (a) Outline of the ChEP procedure. (b) Validation of protein enrichment in the nuclear fraction from ChEP by Western blotting. Nuclear (NUC) and cytoplasmic (CYT) fractions isolated from trophozoite and schizont were probed with anti-histone H3 antibody (top gel) and with anti-aldolase antibody (bottom gel). (c) Significance plots for all proteins detected in each of the three Plasmodium stages analysed. qspec-derived log₂(fold change) and Z-statistic values between NUC:CYT (Table S3c) are shown as scatter plots with grey dots. Green dots highlight proteins with a Z score ≥2 and/or P value ≤0.05. Dots circled in black and red denote putative PfCAPs (Table S2a) that were not-significantly or significantly enriched in the nuclear ChEP fractions, respectively. Protein overlaps between the nuclear and cytoplasmic fractions at ring, trophozoite and schizont stages are shown as a Venn diagram (insets). The proteomics dataset was processed and analysed as described in Fig. S1(a). (d) Plasmodium proteins that were significantly enriched in the nuclear ChEP fractions (Table S3c) were compared to all putative PfCAPs (Table S2a) and to proteins detected in a previous analysis of the P. falciparum nuclear proteome [83] (Table S2d). Stage-specific overlaps between these proteins are shown in Fig. S1(a). (e) TDA was applied to the TopS scores calculated for 466 PfChEP proteins that passed two statistical criteria (i.e. TopS score and Z-statistic from qspec; Table S3c). TDA generated three main groups of protein nodes corresponding to ring, trophozoite and schizont stages. Protein nodes are coloured such that red nodes denote high PCA1 values and blue nodes denote low PCA1 values (see inset histogram scale). Node size is proportional with the number of proteins within the node. Heat maps for three groups of proteins are provided on the outside of the TDA map to illustrate high TopS values in one stage and not the other two (TopS >10 in one stage and <3 in the other two). To plot these heat maps, the TopS values were normalized (‘TopS_norm’) to the highest and lowest values (see scale from −1 to 1). (f) Stage-specific enrichment of domains. In the bar graph on the left, for each of the main CAD families (defined in Figs 2b and S3b), the number of CADs predicted in silico are compared to the numbers detected when merging all ChEP fractions analysed by MudPIT (Table S3a). CADs present in proteins that were deemed significantly enriched in one or two stages by TopS and qspec analyses were assessed for enrichment using Fisher’s exact test (Table S3c). On the right, bubble plots for each stage report on the number of significant proteins bearing these domains (bubble size) and on their specific enrichment or depletion (green and red bubbles, respectively). (g) A subset of the heat maps in (a) is shown for PfCAPs with their associated domains. CADs statistically enriched or depleted in each stage are marked with green or red boxes, respectively (Table S3c). Hpi, hours post infection.

Fig. 4.

Nuclear localization of candidate CAPs. Western blots show enrichment of SMC3 (a) and CRWN-like (b) proteins in the nuclear fraction [lane 1, marker; lane 2, cytoplasmic protein lysate (CYT); lane 3, nuclear protein lysate (NUC)]. (c) Subcellular localization of the SMC3 protein (PF3D7_0414000) during the asexual life stages of the parasite. (d) Subcellular localization of the CRWN-like protein (PF3D7_1325400) during the asexual life stages of the parasite. (e) Immunofluorescence analysis showing the localization of SMC3 at the periphery of the parasite nucleus away from the heterochromatin regions marked by H3K9me3 (Millipore; 07–442-AF488). (f) Colocalization of CRWN-like proteins to the heterochromatin regions of the parasite nucleus marked by H3K9me3. Bars indicate 2 µm in (c–f). (g) Significance plot of Plasmodium proteins interacting with SMC3. For all proteins co-immunoprecipitated with SMC3, qprot-derived log₂(fold change) and Z-statistic values between SMC3-IP and control-IP (Table S4) are shown as scatter plots with grey dots. Coloured dots highlight proteins with a log₂(fold change) ≥2 and Z score ≥1.645 (bright green) or Z score ≥1.45 (light green). Dots circled in black and red denote putative CAPs that were not-significantly or significantly enriched in the SMC3-IPs, respectively (Table S4). (h) Stage-specific enrichment of SCM3 interaction partners. TopS values calculated from the nuclear ChEP dataset (Table S4) are plotted for candidate PfCAPs potentially interacting with SMC3. Hpi, hours post infection.

Fig. 5.

ChIP-seq analysis showing genome-wide distribution of SMC3 in trophozoites. (a) SMC3 distribution across all 14 chromosomes. The blue box indicates the location of the centromere on each chromosome. The regions depicted in (b) are indicated with black boxes. (b) Zoomed in regions on chromosomes 2, 5 and 7 depicting SMC3 distribution.