Heterochromatin formation in bistable chromatin domains controls the epigenetic repression of clonally variant Plasmodium falciparum genes linked to erythrocyte invasion


  • Valerie M. Crowley,

    1. Institute for Research in Biomedicine (IRB), 08028 Barcelona, Catalonia, Spain.
    Search for more papers by this author
  • Núria Rovira-Graells,

    1. Institute for Research in Biomedicine (IRB), 08028 Barcelona, Catalonia, Spain.
    Search for more papers by this author
  • Lluís Ribas de Pouplana,

    1. Institute for Research in Biomedicine (IRB), 08028 Barcelona, Catalonia, Spain.
    2. Institució Catalana de Recerca i Estudis Avançats (ICREA), 08010 Barcelona, Catalonia, Spain.
    Search for more papers by this author
  • Alfred Cortés

    Corresponding author
    1. Institute for Research in Biomedicine (IRB), 08028 Barcelona, Catalonia, Spain.
    2. Institució Catalana de Recerca i Estudis Avançats (ICREA), 08010 Barcelona, Catalonia, Spain.
    Search for more papers by this author

E-mail alfred.cortes@irbbarcelona.org; Tel. +34 934034867; Fax +34 934037109.


Clonally variant gene expression is a common survival strategy used by many pathogens, including the malaria parasite Plasmodium falciparum. Among the genes that show variant expression in this parasite are several members of small gene families linked to erythrocyte invasion, including the clag and eba families. The active or repressed state of these genes is clonally transmitted by epigenetic mechanisms. Here we characterized the promoters of clag3.1, clag3.2 and eba-140, and compared nuclease accessibility and post-translational histone modifications between their active and repressed states. Activity of these promoters in an episomal context is similar between parasite subclones characterized by different patterns of expression of the endogenous genes. Variant expression is controlled by the euchromatic or heterochromatic state of bistable chromatin domains. Repression is mediated by H3K9me3-based heterochromatin, whereas the active state is characterized by H3K9ac. These marks are maintained throughout the asexual blood cycle to transmit the epigenetic memory. Furthermore, eba-140 is organized in two distinct chromatin domains, probably separated by a barrier insulator located within its ORF. The 5′ chromatin domain controls expression of the gene, whereas the 3′ domain shares the chromatin conformation with the upstream region of the neighbouring phista family gene, which also shows variant expression.


Plasmodium falciparum is an apicomplexan protozoan responsible for over 300 million clinical malaria cases every year, including the most severe forms of the disease. This parasite has a complex life cycle involving multiple well-differentiated stages in two different hosts, humans and Anopheles mosquitoes. All clinical symptoms in the human host are associated with the 48 h asexual blood cycle of the parasite. During this cycle, the parasite goes through the intraerythrocytic stages of ring, trophozoite and schizont, the latter containing multiple nuclei. Upon schizont bursting, around 16 merozoites are released from each schizont and start a new round of infection by invading new erythrocytes (Miller et al., 2002).

Progression along the asexual blood cycle is mainly controlled at the transcriptional level. The majority of genes follow a ‘just in time’ pattern of expression, such that they are expressed only at the stage when their product is needed, and repressed for the rest of the cycle. This results in a continuous cascade of gene expression along the asexual cycle (Bozdech et al., 2003; Le Roch et al., 2003). Although the transcriptional mechanisms controlling life cycle progression are not well understood, a recently discovered expanded family of 27 AP2 domain-containing transcription factors (ApiAP2) is likely to have a major role. It has been hypothesized that each ApiAP2 controls expression of a specific set of genes, and additionally may control expression of the next ApiAP2(s) along the cycle (Balaji et al., 2005; Campbell et al., 2010).

In addition to progression along the life cycle, transcriptional regulation in P. falciparum also controls clonally variant (variegated) expression of some genes expressed during asexual blood stages. Genes under variant expression can be expressed at different levels, from silenced to abundantly expressed, in different parasites at the same point of the asexual cycle. Clonally variant gene expression is a common survival strategy in many pathogens, from bacteria to fungi and protozoans (Deitsch et al., 2009). By regularly switching on and off the transcription of specific genes and thus changing the antigens that are displayed to the host immune system, pathogens can survive in the hostile environment of a host able to develop complex immune responses against them. The best studied example of variant expression in P. falciparum are var genes (Horrocks et al., 2008; Scherf et al., 2008), a family of about 60 genes encoding PfEMP-1, a virulence factor with a critical role for both antigenic variation and cytoadherence of infected erythrocytes. Expression of var genes is mutually exclusive, such that at a particular time an individual parasite expresses a single var gene and maintains all others repressed.

Expression of var genes is controlled epigenetically. The active var gene is clonally inherited, and switching to express a different var gene occurs in the absence of detectable genetic changes (Scherf et al., 1998). The transition between the active and the silenced state of a var gene is associated with modifications in its chromatin structure, including changes in the position of specific nucleosomes in the promoter (Duraisingh et al., 2005), differences in the post-translational covalent modifications of N-terminal histone tails (Freitas-Junior et al., 2005; Chookajorn et al., 2007; Lopez-Rubio et al., 2007) and recruitment of heterochromatin protein 1 (PfHP1) (Flueck et al., 2009; Perez-Toledo et al., 2009). These data indicate that var gene silencing is mediated by formation of heterochromatin. Heterochromatin is a highly condensed type of chromatin, which results in low nuclease accessibility and transcriptional repression. At the molecular level, the hallmarks of heterochromatin are low levels of histone acetylation, methylation of lysine 9 of histone H3 (H3K9) and presence of the structural protein HP1, which binds trimethylated H3K9 (H3K9me3). Heterochromatin can spread into adjacent regions by positive feedback mechanisms based on recruitment of the H3K9 methyltransferase by HP1, and is inherited by epigenetic mechanisms (Grewal and Jia, 2007).

Clonally variant gene expression in P. falciparum occurs beyond var genes, affecting other gene families predominantly located in subtelomeric regions (Scherf et al., 2008). Those include relatively large gene families such as rif, stevor and pfmc-2tm, and also several small gene families encoding proteins that participate in erythrocyte invasion, such as eba, PfRh and clag (Taylor et al., 2002; Duraisingh et al., 2003; Stubbs et al., 2005; Cortés et al., 2007; Cortés, 2008; Gomez-Escobar et al., 2010). Erythrocyte invasion is a multistep process that the parasite can complete using alternative sets of receptor–ligand interactions, termed alternative invasion pathways. Invasion pathways largely depend on eba and PfRh genes. These five- and six-member gene families encode proteins localized in apical organelles of the merozoite, which function as ligands that interact with different erythrocyte receptors during invasion (Cortés, 2008). The five-member clag family also encodes proteins (RhopH1/CLAG) localized in the apical organelles of the merozoite, but their precise function is not known. However, their localization and time of expression strongly suggest a role in erythrocyte invasion, and/or in the subsequent formation of the parasitophorus vacuole (Kaneko, 2007). Interestingly, two genes of the clag family, clag3.1 and clag3.2, constitute the only known example of mutually exclusive expression in P. falciparum apart from var genes (Cortés et al., 2007). These two genes are neighbours in the left subtelomeric region of chromosome 3, separated by a pseudogene and a non-protein coding transcript (PF03TR002) located upstream of clag3.1 (Otto et al., 2010). Variant expression of genes linked to erythrocyte invasion may play a role in immune evasion (Cortés, 2008), in addition to controlling the use of alternative invasion pathways. The capacity to use alternative invasion pathways confers P. falciparum the ability to invade essentially all types of erythrocytes and contributes to its high virulence compared with other Plasmodium species.

Genes under variant expression linked to erythrocyte invasion are regulated at two different levels, similar to other variantly expressed genes. First, these genes are only expressed in late schizonts and merozoites, and silenced for the rest of the asexual cycle (Bozdech et al., 2003; Le Roch et al., 2003). Second, in some parasites the gene is only expressed at very low levels (repressed state), even in late schizonts and merozoites. We recently found that in a clonal, genetically homogeneous parasite line, expression of these genes varies between individual parasites. The active or repressed state of these genes is stable and clonally transmitted through successive generations by epigenetic mechanisms (Cortés et al., 2007), as in the case for var genes.

In spite of the critical role of clonally variant gene expression for survival and virulence in P. falciparum, the molecular mechanisms controlling it are poorly understood. The limited mechanistic insights on the regulation of variant expression are largely restricted to var genes, but it remains unknown whether the same mechanisms apply to other genes under variant expression in P. falciparum. Here we studied the chromatin modifications responsible for the epigenetic regulation of two genes of the clag family, clag3.1 and clag3.2, one gene of the eba family, eba-140, and one gene of the unrelated phista family, which encodes proteins exported to the erythrocyte. We used isogenic unselected parasite subclones that either do or do not express these genes to compare chromatin structure between the active and the repressed states. We found that some nucleosomes are positioned differently when the genes are active or repressed, and that post-translational modifications at H3K9 play a key role in regulating variant expression of these genes and maintaining the epigenetic memory. Our data show that clonally variant expression is controlled by chromatin state, either euchromatin characterized by acetylated H3K9 (H3K9ac) or heterochromatin characterized by H3K9me3. This is the basis of a common regulatory mechanism for different P. falciparum variantly expressed gene families.


Promoter structure of clag3.1, clag3.2 and eba-140

As a first step to understand the regulation of clag3.1 (PFC0120w), clag3.2 (PFC0110w) and eba-140 (MAL13P1.60), we identified functional elements in the upstream region of these genes. The transcription start sites (TSSs) of the three genes were determined by 5′ RLM-RACE (Figs 1A and S1A). For all three genes, multiple TSSs were detected, indicating that they have broad-type promoters. Multiple TSSs are commonly observed in other P. falciparum genes (Horrocks et al., 2008) and in genes from other eukaryotic organisms (Sandelin et al., 2007). The clusters of TSSs spanned a broader region in clag3.1 and clag3.2 (positions −172 to −454 bp and −168 to −315 bp relative to the start codon, respectively) than in eba-140, where all TSSs were concentrated in a 43 bp region (positions −549 to −592 bp). To determine whether frequently used TSSs located further upstream were missed by the RACE analysis, we conducted qPCR analysis of cDNA using primers located upstream and downstream of the mapped TSSs. For the three genes, less than 1% of transcripts originated upstream of the mapped TSSs (Fig. S1B).

Figure 1.

Characterization of the regulatory regions of clag3.1, clag3.2 and eba-140. A. Position and frequency of TSSs identified by sequencing cloned 5′ RLM-RACE products. Positions are relative to the translation start codon. B. Gene reporter assays with clag3.1, clag3.2 and eba-140 upstream regions. Constructs containing the upstream regions indicated were tested for luciferase activity in parasites at the schizont stage. The locations of the TSSs are marked with arrows. The lengths of the upstream sequences tested are included in the name of the constructs (in base pairs). Values are % luciferase activity relative to the construct with the longest upstream fragment, and correspond to the average of two independent experiments (each performed in duplicate), with standard deviation.

The promoter activity of serial 5′ deletions of the upstream regions of clag3.1, clag3.2 and eba-140 was determined in parasites at the schizont stage using transient transfections and luciferase gene reporter assays. Multiple upstream fragments were found to be involved in the transcriptional regulation of the three genes. Constructs containing only the 5′UTR and the region comprising the TSSs showed minimal promoter activity, indicating that in addition to the core promoter, other cis-acting elements (proximal enhancers) are necessary for efficient promoter activity (Fig. 1B). In the case of eba-140, a construct spanning 1080 bp upstream of the start codon (≈ 500 bp upstream of the TSSs) had minimal promoter activity, but sequences located further upstream contained cis-acting regulatory elements that resulted in large increases in promoter activity. Similarly, regulatory elements located more than 500 bp upstream from the start codon of clag3.1 and clag3.2 resulted in progressive increases in promoter activity (Fig. 1B). Thus, the three regulatory regions conform to the typical bipartite structure: the core promoter, which comprises the short DNA region around the TSSs, had minimal promoter activity, and several upstream regulatory elements (proximal enhancers) were necessary for efficient transcription.

None of the 5′ fragments had promoter activity in parasites at the ring/trophozoite stage, indicating that the upstream sequences contain sufficient information to direct the stage-specific expression of clag3.1, clag3.2 and eba-140 (Fig. S2). This result indicates that the promoters contain binding sites for stage-specific activators (transcription factors). Experimental and in silico approaches have predicted regulatory motifs that may control P. falciparum gene expression along the life cycle. Multiple predicted regulatory motifs are present in regions that are important for promoter activity in the three genes (Fig. S3), but additional experiments will be necessary to determine which specific motifs are necessary for the activation of these genes.

The longest 5′ upstream fragments tested for clag3.1 and eba-140 (3.1-LH-2198 and 140-LH-1925) did not encompass their full upstream intergenic regions. Because all increases in size of the 5′ upstream fragments tested resulted in increased promoter activity, we suspected that 3.1-LH-2198 and 140-LH-1925 may not contain the full regulatory region. We tested constructs with over 3 kb of the two upstream regions, but none of the new constructs showed higher promoter activity than the corresponding longest constructs tested in Fig. 1B, and in the case of eba-140, the longer construct even had lower promoter activity (Fig. S4A). These results indicate that 3.1-LH-2198 and 140-LH-1925 probably have full promoter activity.

Activity of episomal clag3.1, clag3.2 and eba-140 promoters in parasite lines with different patterns of expression of the endogenous genes

Parasite lines 10G and 1.2B are genetically identical but transcriptionally different recent subclones of the clonal parasite line 3D7-A (Fig. 2A) (Cortés et al., 2004; 2007; Cortés, 2005). Our previous work comparing expression of clag3.1, clag3.2 and eba-140 by semi-quantitative PCR between subclones 10G and 1.2B demonstrated that clag3.1 is repressed in 10G, whereas clag3.2 and eba-140 are repressed in 1.2B, and these clonally transmitted expression patterns are stable for at least 30 asexual cycles (Cortés et al., 2007). Using qPCR analysis, we found that the differences in mRNA levels between 10G and 1.2B schizonts were 14- and 17-fold for clag3.1 and clag3.2, respectively, whereas the difference in eba-140 transcript abundance was 63-fold (Fig. 2B). The levels of transcripts of these genes at stages of the asexual life cycle when they are not expressed (20–25 h trophozoites) were even lower than in schizonts of the subclone where the gene is repressed (Fig. 2C), indicating that either repression in schizonts is not complete or a small fraction of individual parasites have switched on expression of the genes.

Figure 2.

Expression of clag3.1, clag3.2 and eba-140 in subclones 10G and 1.2B. A. Origin of the parasite lines used in this article. Black circles represent genetically identical parasites. B. qPCR analysis of transcript levels in schizonts (40–45 h postinvasion). Results are expressed in arbitrary units relative to a standard curve using genomic DNA. Values were normalized against non-variantly expressed genes that have the same timing of expression as the genes tested, rhopH2 for clag3.1 and clag3.2, and ama1 for eba-140. C. qPCR analysis comparing transcript levels between trophozoites (20–25 h postinvasion) and schizonts. Results are expressed as in panel B, but values were normalized against seryl tRNA synthetase (PF07_0073), which is expressed throughout the asexual cycle. D. Luciferase assays with plasmids 3.1-2198-LH (clag3.1), 3.2-1371-LH (clag3.2) and 140-1925-LH (eba-140) in 10G and 1.2B schizonts. A plasmid containing the upstream region of ama1 and the plasmid linker-LH were used as controls. Values correspond to average luciferase activity (in relative light units, RLU), with standard deviation, normalized to 1% parasitaemia. E. Luciferase assays with plasmids 3.2–1371-LH (clag3.2) and 3.2–1371-LH-R20 (clag3.2 + R20) in 3D7-A (left) or 10G and 1.2B (right) parasites. In the left panel data are expressed as percent luciferase activity relative to 3.2-1371-LH, whereas in the right panel they are expressed in relative light units (RLU), normalized to 1% parasitaemia.

Plasmids 3.1-2198-LH, 3.2-1371-LH and 140-1925-LH were tested in 10G and 1.2B schizonts. In spite of endogenous clag3.1 being repressed in 10G schizonts and endogenous clag3.2 and eba-140 repressed in 1.2B, all the promoters resulted in similar levels of luciferase activity between 10G and 1.2B when placed in an episome (Fig. 2D). The same holds true for constructs with longer eba-140 upstream sequences with lower promoter activity than 140-1925-LH (Fig. S4B). In our assay, parasites go through the S phase (DNA synthesis) after transfection and before promoter activity is assayed, ensuring that chromatin is assembled in the episomes. This result demonstrates that repression is not controlled by trans-acting factors expressed only in 10G or only in 1.2B. This result also indicates that, when placed in an episome, these promoters are active by default, in contrast to var gene promoters that by default are repressed and enter the mutually exclusive expression program (Voss et al., 2006), at least when coupled to a var intron (Deitsch et al., 2001; Dzikowski et al., 2006). Variantly expressed genes are commonly found near telomeric clusters at the nuclear periphery (Horrocks et al., 2008; Scherf et al., 2008). To test the possibility that this physical location is necessary for repression, we added a subtelomeric repeat sequence, rep20, into the 3.2-1371-LH plasmid, generating plasmid 3.2-1371-LH-R20. The rep20 sequence is known to tether with endogenous rep20 sequences and direct the episome to chromosomal ends clusters (O'Donnell et al., 2002). The plasmid 3.2-1371-LH-R20 had similar or even higher luciferase activity than the plasmid 3.2-1371-LH, and it had similar activity between 10G and 1.2B parasites (Fig. 2E). Thus, even in an episome directed to subtelomeric regions, this promoter is active by default.

Specific nucleosomes are positioned differently between the active and repressed states of clag3.1, clag3.2 and eba-140

Analysis of the 5′ regions of clag3.1, clag3.2 and eba-140 by micrococcal nuclease (MN) digestion revealed differences in the position of some nucleosomes between the active and the repressed states of these genes (Fig. 3). MN cleaves DNA preferentially between nucleosomes, whereas DNA wrapped around nucleosomes is relatively resistant to cleavage. In schizonts, the most striking differences between the active and repressed states were observed in the clag3.2 promoter. A major hypersensitive site in 10G schizonts at position ≈ −800 (relative to the start codon) was much less prominent in 1.2B schizonts, where the gene is repressed. In contrast, a major hypersensitive site at ≈ −1100 was associated with the repressed state of the gene (arrows in Fig. 3B). These positions are within regions important for promoter activity (Figs 1B and 3B). Another major difference between 10G and 1.2B schizonts was observed in the region that contains the majority of clag3.1 TSSs, where an hypersensitive site was present only in 1.2B schizonts, where the gene is active (arrow in Figs 3A and S5A). Examples of less pronounced differences between 10G and 1.2B in the hypersensitive sites in clag3.2, clag3.1 and eba-140 are indicated by arrowheads in Figs 3 and S5. These results demonstrate that variant expression of clag3.1, clag3.2 and eba-140 involves changes in the position of specific nucleosomes in the regulatory regions of these genes.

Figure 3.

MN sensitivity assays. Southern blot analysis (end-labelling) of MN-digested nuclei with specific probes against clag3.1 (A), clag3.2 (B) and eba-140 (C). Nuclei from 10G and 1.2B parasites at the schizont (Sch.) or trophozoite (Tr.) stage were digested with the indicated amounts of MN (0, 0.5 or 1.5 units). The position of TSSs and the relative promoter activity attributable to different parts of the upstream region (in %) are shown. Positions are relative to the start codon. Vertical red lines indicate the position of the probes. Arrows indicate the major differences in sensitivity to MN between 10G and 1.2B in schizonts, whereas black arrowheads indicate examples of less pronounced differences. Red arrowheads indicate differences maintained in trophozoites. An asterisk indicates a difference between schizonts and trophozoites independent of transcriptional state. The insert at the right of panel B corresponds to a storage phosphor imager trace of schizonts lanes 0.5 U.

We also analysed nucleosome positions in parasites at the trophozoite stage. The three genes under study are never expressed at this stage, but the clonal transmission of their expression status implies that in trophozoites they are bookmarked for expression (poised) or bookmarked for repression when the parasites reach the schizont stage. In clag3.2, nucleosomes were positioned identically between 10G and 1.2B trophozoites, and these positions were very similar to 1.2B schizonts, where the gene is repressed (Fig. 3B). A parallel situation was observed for the hypersensitive site in the TSSs region of clag3.1, which was absent in both 10G and 1.2B trophozoites, same as in 10G schizonts where the gene is repressed (Figs 3A and S5A). These results indicate that some nucleosome positions are exclusively associated with active transcription of the genes. In contrast, some hypersensitive sites associated in schizonts with the active or repressed state of clag3.1 or eba-140 were similarly associated with the bookmarked for expression or bookmarked for repression states in trophozoites (red arrowheads in Figs 3 and S5). These nucleosome positions are maintained through non-schizont stages and may be involved in the maintenance of the epigenetic memory. Last, we identified a major hypersensitive site in the 5′UTR of eba-140 that was present in schizonts but absent in trophozoites independently of the variant expression of the gene (asterisk in Figs 3C and S5B).

Restriction enzyme (RE) accessibility is reduced in the promoters of clag3.2 and eba-140 when the genes are repressed

10G and 1.2B nuclei were digested with RE with sites within the regulatory regions of clag3.1, clag3.2 and eba-140 (Fig. 4A). Cleavage of the clag3.2 promoter with NsiI and cleavage of the eba-140 promoter with BglII and XbaI were reduced in 1.2B schizonts, where the genes are repressed, compared with 10G schizonts, where the genes are active. In contrast, the gene pfg377 (PFL2405), which does not show variant expression between 10G and 1.2B, was cleaved by the three RE with similar efficiency (Fig. 4B). This indicates that clag3.2 and eba-140 upstream regions are less accessible to RE when the genes are repressed, consistent with repression being mediated by heterochromatin formation. Reduced RE accessibility of eba-140 in 1.2B was maintained in trophozoites, but not reduced accessibility of clag3.2 (Fig. 4B). In contrast, no difference was observed in cleavage of the clag3.1 promoter by NsiI between 10G and 1.2B schizonts, and cleavage was reduced in 1.2B trophozoites. Accessibility of RE depends on the euchromatic or heterochromatic state of the region but also depends on the position of nucleosomes relative to the restriction site and presence of DNA binding proteins at this site. In the case of clag3.1, the NsiI site is amid the TSSs, and binding of the preinitiation complex and RNA pol II may interfere with NsiI cleavage when the gene is active in 1.2B.

Figure 4.

Digestion of 10G and 1.2B nuclei with RE. A. Schematic of the position of NsiI, BglII and XbaI restriction sites in the genes under analysis. Positions are relative to the start codon (ATG). Purified DNA was digested with MfeI (clag3.1 and clag3.2) or NdeI (eba-140 and pfg377) before Southern blot. The position of the probe for pfg377 is indicated by a line underneath the gene, probes for the other genes were as in Fig. 3. B. Southern blot analysis of a representative RE accessibility experiment. In all cases, the upper band corresponds to undigested DNA and the lower band is the digested fragment. The bar graphics are % digestion in each lane, as quantified in a storage phosphor imager. Results were fully consistent between two independent experiments.

H3K9me3 and H3K9ac are associated with the repressed and active states, respectively, of both clag3.1 and clag3.2 in schizonts

We used chromatin immunoprecipitation (ChIP) followed by qPCR analysis (qChIP) to compare post-translational histone modifications between the active and repressed states of clag3.1 and clag3.2. The two genes showed opposite patterns of histone modifications between 10G and 1.2B schizonts, reflecting their opposite patterns of transcription in these two subclones (Figs 5 and S6). H3K9me3, a histone modification generally associated with repression and a hallmark of heterochromatin (Grewal and Jia, 2007; Trojer and Reinberg, 2007), marked clag3.1 or clag3.2 only when the gene was transcriptionally repressed (in 10G and 1.2B schizonts, respectively) (Figs 5C and S6A). The repressed gene had similar levels of H3K9me3 as silent var genes, which are well-described targets for this modification (Chookajorn et al., 2007; Lopez-Rubio et al., 2007), whereas the active gene had similar levels as euchromatic controls where this modification is absent (Lopez-Rubio et al., 2009; Salcedo-Amaya et al., 2009) (Fig. S6A). Conversely, clag3.1 or clag3.2 were enriched in H3K9ac in the parasite subclone where the gene was active, compared with the parasite subclone where the gene is repressed (Figs 5E and S6C). H3K9ac is a mark typically localized in active promoters in model organisms (Pokholok et al., 2005). The increase in H3K9me3 or H3K9ac associated with repression or activation, respectively, was distributed across the promoter, 5′UTR and coding region of clag3.1 and clag3.2 (Figs 5A,C,E and S6A,C). H3K4me3 and global H4 acetylation (H4ac), which are typically associated with the 5′ region of active genes (Pokholok et al., 2005; Li et al., 2007), showed only minor differences in their relative enrichment between the active and silenced states of clag3.1 and clag3.2. However, both marks tended to occur at a slightly higher level when the gene was active, mainly at positions immediately upstream of their start codons (Figs 5G,I and S6E,G).

Figure 5.

qChIP analysis of histone modifications associated with the variant expression of clag3.1, clag3.2, eba-140 and MAL13P1.59. A. Position of the fragments amplified by qPCR. Exons are indicated by black boxes and the position of the TSSs is indicated by an arrow. B. Expression of the genes under analysis in 10G and 1.2B schizonts and trophozoites. An arrow indicates active expression, whereas an ‘X’ indicates repression. A ‘B’ indicates bookmarked, for either expression or repression, indicated as before. C–J. Relative enrichment of histone modifications between 10G and 1.2B parasites. Raw qChIP data from Fig. S6 is represented as the log2 ratio of the % input of 10G divided by the % input of 1.2B. Values are the average of three independent experiments, with standard error. The % input is defined as the percentage of immunoprecipitated chromatin relative to the input sample. Positive values correspond to higher enrichment in 10G, whereas negative values correspond to higher enrichment in 1.2B. Values correspond to qChIP from schizonts (C,E,G,I) or trophozoites (D,F,H,J).

ChIP analysis with antibodies against H3 did not reveal large changes between 10G and 1.2B in any of the genes under study (data not shown), suggesting that nucleosome density is similar between the active and repressed states of the genes. However, we did not normalize results against H3 because ChIP with this antibody resulted in poor recovery and had high variability. The antagonistic enrichment observed with different antibodies eliminates the possibility that the differences observed between the active and repressed states of these genes are due to different nucleosome densities.

eba-140 shows a complex pattern of histone modifications associated with two independent chromatin domains

Similar to clag3.1 and clag3.2, H3K9me3 marked eba-140 in schizonts only when the gene was transcriptionally repressed (Figs 5C and S6A). However, in eba-140 the increase in H3K9me3 associated with repression peaked in the promoter region and progressively diminished midway through the first exon (Fig. 5C). Also similar to clag3.1 and clag3.2, H3K9ac marked eba-140 in the parasite line where the gene was active, but again the increase in this modification persisted only until the beginning of the first exon. Downstream from this position, the parasite subclone where eba-140 is repressed (1.2B) was enriched in H3K9ac (Figs 5E and S6C). This surprising distribution of H3K9me3 and H3K9ac along eba-140 lead us to speculate that the chromatin conformation associated with the transcriptional status of the gene located downstream of eba-140 in chromosome 13, a phista family gene (MAL13P1.59), may spread into the coding region of eba-140 (Fig. 5A). To test this hypothesis, we studied histone modifications in the intergenic region between the two genes, which presumably contains the 3′UTR of eba-140, and the promoter and 5′UTR of MAL13P1.59. While H3K9me3 occurred at similar intermediate levels between 10G and 1.2B schizonts, the intergenic region was clearly enriched in H3K9ac in 1.2B parasites, similar to the second half of the eba-140 ORF (Figs 5C,E and S6A,C). H3K4me3 and global H4ac were only moderately increased in 10G schizonts in the eba-140 upstream region and beginning of the ORF, but both showed a gradual enrichment in 1.2B starting midway through the first exon of eba-140, and clearly marked the intergenic region only in 1.2B (Figs 5G,I and S6E,G).

Next, we performed transcriptional analysis of MAL13P1.59 in 10G and 1.2B, and found that it is also variantly expressed, with an opposite pattern of expression to eba-140, as expected from the histone modifications in its promoter region. MAL13P1.59, which is active throughout the asexual cycle, was expressed at much higher levels in 1.2B than in 10G, both in schizonts and in trophozoites (Fig. 6A). All together, these results indicate that the variant expression of eba-140 is mainly controlled by the methylation/acetylation balance at H3K9 in its upstream region, similar to clag3.1 and clag3.2, but activation of MAL13P1.59 is associated with a large increase in H3K9ac, H3K4me3 and global H4ac in its upstream region. On the other hand, H3K9me3 marked MAL13P1.59 in schizonts, but repression of this gene was not associated with a large increase in this mark.

Figure 6.

Chromatin domains within eba-140. A. qPCR analysis of MAL13P1.59 transcript levels in 10G and 1.2B trophozoites (20–25 h postinvasion) and schizonts (40–45 h postinvasion). Values are relative to a gDNA standard curve, and were normalized against seryl tRNA synthetase. B. Schematic representation of chromatin domains organization in eba-140. In 1.2B parasites, where eba-140 is repressed, a heterochromatic domain (red) rich in H3K9me3 extends from the promoter region to the middle of the first exon. This is followed by a progressive transition into a neighbouring domain marked by typical euchromatic marks (green shades) that includes the second half of the eba-140 ORF and the MAL13P1.59 upstream region. In 10G parasites, the opposite situation is observed, but H3K4me3 and H4ac show only a moderate enrichment associated with the active state of eba-140. The region where the transition between the two domains occurs in the two parasite lines (striped) may have barrier insulator activity.

The gene eba-140 is organized in two distinct chromatin domains with opposite patterns of histone modifications. The chromatin conformation at the second half of the eba-140 ORF resembles that of the upstream region of MAL13P1.59 in both 10G and 1.2B, indicating that these two regions are part of the same chromatin domain. However, the variant expression of eba-140 is controlled by its 5′ chromatin domain, as only in this region do active and repressive histone marks correlate with expression and repression respectively (Fig. 6B).

Histone marks associated with the epigenetic memory of clag3.1, clag3.2 and eba-140

The pattern of histone modifications in clag3.1, clag3.2 and eba-140 in 10G and 1.2B parasites at the trophozoite stage, when these genes are not expressed, was remarkably similar to the pattern observed in schizonts (Figs 5 and S6). H3K9me3 marked clag3.1, clag3.2 and eba-140 in trophozoites of the parasite line where the genes will be transcriptionally repressed later in schizonts (bookmarked for repression) (Figs 5D and S6B). Similarly, H3K9ac persisted during the trophozoite stage in the parasite lines where the genes will be transcriptionally activated later in schizonts (bookmarked for expression) (Figs 5F and S6D). We only observed small changes in the levels of H3K4me3 and H4ac between 10G and 1.2B trophozoites, similar to the changes observed in schizonts (Figs 5H,J and S6F,H). These results indicate that the methylation/acetylation balance at H3K9, and maybe even the pattern of other modifications, persists throughout the asexual cycle and maintains the epigenetic memory of the expression status of these genes. In the upstream region of MAL13P1.59, which is expressed both in trophozoites and in schizonts (Fig. 6A), the euchromatic histone marks observed in 1.2B schizonts were also present and similarly distributed in 1.2B trophozoites (Figs 5F,H,J and S6D,F,H). The repressive mark H3K9me3, which occurred at similar intermediate levels between 10G and 1.2B schizonts, was clearly enriched in 10G trophozoites, where the gene is repressed (Figs 5D and S6B).


Here we characterized the regulatory regions of clag3.1, clag3.2 and eba-140, and also characterized the chromatin modifications associated with the clonally variant expression of these genes and another variantly expressed gene, MAL13P1.59. We demonstrate that repression of the four genes is associated with characteristic properties of heterochromatin: low levels of H3K9ac, high levels of H3K9me3, and in the case of clag3.2 and eba-140, low nuclease accessibility. Histone modifications in these genes are maintained during stages of the asexual cycle when the genes are not expressed, transmitting the epigenetic memory for their variant expression. We also observed changes in the position of specific nucleosomes in the upstream regions of clag3.1, clag3.2 and eba-140 when they were repressed, consistent with chromatin remodelling complexes participating in the transition between the active and repressed states.

Genome-wide mapping of the heterochromatin marks H3K9me3 and HP1 revealed that in P. falciparum heterochromatin is largely restricted to subtelomeric regions and a few chromosome-internal gene clusters, marking subtelomeric repeats (TAREs), large gene families that participate in antigenic variation, such as var, rif and stevor, and other gene families possibly involved in host–parasite interactions and virulence (Flueck et al., 2009; Lopez-Rubio et al., 2009; Salcedo-Amaya et al., 2009). However, H3K9me3 and HP1 are not general marks of stage-dependent silencing. In these genome-wide studies, H3K9me3 and HP1 marked clag3.1 and clag3.2, but not eba-140 and MAL13P1.59. This probably reflects that eba-140 and MAL13P1.59 were active in the parasite lines used. These studies did not compare histone marks between the active and repressed states of these genes, which we did by taking advantage of isogenic subclones with opposite transcriptional states for the four genes.

Similar to our findings, silenced var genes are characterized by typical heterochromatin properties, such as H3K9me3 and low histone acetylation (Freitas-Junior et al., 2005; Chookajorn et al., 2007; Lopez-Rubio et al., 2007). During the preparation of this manuscript, Jiang and co-workers reported the chromatin modifications associated with the epigenetic silencing of a gene from the PfRh family, PfRh4, and a neighbouring pseudogene, eba-165. Repression of these genes, which were co-regulated by a single intergenic region, was also associated with histone modifications consistent with formation of heterochromatin (Jiang et al., 2010). There are important similarities in the histone modifications associated to the active or repressed states between different variantly expressed genes, but family-specific peculiarities also exist (Table 1). Both H3K4me and global H4 acetylation are clearly associated with active transcription in var genes (Freitas-Junior et al., 2005; Lopez-Rubio et al., 2007), PfRh4/eba-165 (Jiang et al., 2010) and MAL13P1.59, although only H3K4me (both H3K4me3 and H3K4me2) enrichment in the active var gene is maintained through stages when the gene is not expressed (Table 1). We did not observe major differences in the levels of H3K4me3 or H4ac associated with active or repressed clag3.1, clag3.2 and eba-140, but we can not completely exclude the possibility that the small changes observed for these modifications have some functional relevance. In contrast to the different patterns of these typically euchromatic histone modifications between different variantly expressed gene families, H3K9 modifications play a conserved role in regulating variant expression (Table 1). Trimethylation of H3K9 is a common theme in repressing P. falciparum genes under variant expression and maintaining the epigenetic memory through stages at which the gene is not expressed. H3K9ac is also associated with activation in both var genes and the genes studied here. While the presence of H3K9ac at stages of the asexual cycle when the genes are not expressed was not assessed for var genes or PfRh4/eba-165 (Lopez-Rubio et al., 2007; Jiang et al., 2010), we clearly found that it is maintained through non-schizont stages in clag3.1, clag3.2 and eba-140, indicating that the methylation/acetylation balance at H3K9, rather than only H3K9me3, has a predominant role in maintaining the epigenetic memory. It has been proposed that H3K9 deacetylation is a necessary step for repression of variantly expressed genes in P. falciparum, as H3K9ac may prevent methylation at this position (Lopez-Rubio et al., 2009), similar to the mechanism found in other organisms (Grewal and Elgin, 2002). All together, the analysis of histone modifications in six genes under variant expression (four in this study and two from previous studies) from families as diverse as var, PfRh, phista, clag and eba, reveals that H3K9me3 and the competing mark H3K9ac are the only features that consistently correlate with the repressed or active states in all variant genes where it was studied (Table 1).

Table 1.  Histone modifications in clonally variant genes.
  1. The signs + and − indicate whether or not large changes in the abundance of the histone modification are observed between the active and repressed states of a variant gene, at stages when the gene is active (columns A) or at stages when the gene is not expressed but bookmarked for expression or bookmarked for repression (columns B). MAL13P1.59 is expressed throughout the asexual cycle. ND, not determined. Data for var genes are from ref. (Freitas-Junior et al., 2005; Chookajorn et al., 2007; Lopez-Rubio et al., 2007) and data for pfRh4/eba-165 are from ref. (Jiang et al., 2010), whereas the other data are from this article.


It is currently unknown whether additional post-translational histone modifications participate in the regulation of variant gene expression in P. falciparum, but modifications other than H3K9me3 typically associated with heterochromatin in other organisms, such as methylation at H3K27 and H4K20 (Trojer and Reinberg, 2007), are unlikely to have an important role. H3K27me3 has not been identified in P. falciparum in spite of thorough mass spectrometry analysis (Miao et al., 2006; Trelle et al., 2009), whereas H4K20me3 shows a broad distribution across the genome inconsistent with a role in heterochromatin-based silencing of variant genes (Lopez-Rubio et al., 2009). These observations, together with the consistent association of H3K9me3 and H3K9ac with the repressed or active states of variant genes (Table 1), indicate that the expression state of variantly expressed genes and its epigenetic transmission largely (and potentially exclusively) depend on H3K9 modifications. Other histone modifications such as H3K4me and H4ac may play family-specific roles, probably as effectors that facilitate transcription initiation or elongation, rather than carrying information that determines the active or repressed state of a gene. While the key role of H3K9me3 in controlling variant expression is supported by genome-wide mapping (Lopez-Rubio et al., 2009; Salcedo-Amaya et al., 2009), the participation of H3K9ac, which shows a broad genomic distribution, could not be predicted by these studies, as it probably performs other roles in P. falciparum apart from controlling variant expression (Cui et al., 2007; Salcedo-Amaya et al., 2009).

The apparent paucity of specific transcription factors in the P. falciparum genome led to the hypothesis that transcriptional regulation of life cycle progression is controlled epigenetically at the chromatin level (Aravind et al., 2003; Coulson et al., 2004; Hakimi and Deitsch, 2007), but to orchestrate transcription of specific genes, chromatin modifications must be targeted to these loci by sequence-specific DNA binding proteins such as ApiAP2 transcription factors (Campbell et al., 2010). Our data show that the upstream regulatory regions characterized here, even when truncated, are sufficient to dictate correct stage-specific transcription when placed in an episome. This is strongly suggestive of a mechanism where presence or absence of stage-specific promoter-bound activators (specific transcription factors) orchestrate changes in the expression of these genes during the life cycle, whereas changes in histone modifications along the life cycle, if any, are mere executers or markers in this process. On the other hand, chromatin state, either euchromatin or heterochromatin, maintained throughout the asexual cycle and clonally transmitted through successive generations epigenetically, would dictate the transcriptionally permissive or repressed state of genes under variant expression. Only the combination of permissive chromatin conformation and presence of activators results in active transcription (Fig. 7).

Figure 7.

Model for the regulation of variant expression and stage-specific expression of genes linked to erythrocyte invasion. Green marks represent permissive histone modifications, such as H3K9ac, whereas red marks represent repressive histone modifications, such as H3K9me3. The permissive or repressive chromatin conformation is maintained along the full asexual cycle. Insulators limit chromatin domains. TF is a stage-specific transcription factor, absent during non-schizont stages. Infrequent, stochastic transitions between the two stable chromatin conformations are triggered by random interaction (or inefficient recruitment) with enzymes acting on H3K9 (HMT, histone methyl transferases, HAT, histone acetyl transferases, HDM, histone demethylases, HDAC, histone deacetylases). The model is of general applicability to genes under variant expression, but the trans-acting factor (activator or repressor) may be present at different stages of the asexual cycle.

The next important question is: how is chromatin state regulated? The origin of the parasite subclones used in this study (Fig. 2A) provides some insight into this question. Transcriptional patterns in subclones of the clonal parasite line 3D7-A, such as 10G and 1.2B, analysed soon after subcloning, reflect the patterns of gene expression in individual parasites within the 3D7-A population. 3D7-A is a mixture of isogenic single parasites with different combinations of expressed and repressed genes (Cortés et al., 2007). This indicates that switches between the active and repressed states of the genes studied here occur in the absence of selection, and these transitions are spontaneous and stochastic, as opposed to deterministic, because these genes are in a different expression status in individual parasites maintained in the same environment. Switches in the expression of var, pfmc-2tm and stevor genes also occur spontaneously (Frank et al., 2007; Lavazec et al., 2007). Thus, regulation of the variant expression of the genes studied here (and probably also other P. falciparum genes), does not involve sensing changes in the environment or complex regulatory cascades, in contrast to the complex and tightly regulated modes of heterochromatin-mediated gene repression related to development or to responses to changing environment conditions in higher eukaryotes (Trojer and Reinberg, 2007). Instead, we propose that variant expression of P. falciparum genes is controlled by a stochastically established reversible chromatin state, either euchromatin or heterochromatin, maintained throughout the asexual cycle and clonally transmitted through successive generations. There is an intense debate on whether histone modifications carry epigenetic information. Only in cases where histone modifications influence transcriptional outcome, and are transmitted in the absence of the causal event that triggered these modifications, it is clear that these modifications carry epigenetic information (Campos and Reinberg, 2009). The genes studied here, where both chromatin states are stable and inheritable under homogeneous conditions, represent one of the few clear examples where histone modifications unquestionably qualify as epigenetic marks.

The chromatin domains controlling expression of clag3.1, clag3.2, eba-140 and MAL13P1.59, where both the active and the repressed states are stable for many generations, clearly match the emerging concept of bistable chromatin. Bistable chromatin domains can adopt two states, active or repressed, which are stably maintained and inherited, with transitions between the two states maintained at low frequency by positive feedback loops (Dodd et al., 2007; Campos and Reinberg, 2009). A theoretical framework has been proposed to explain the low transition rates at bistable chromatin domains. According to this model, transitions are controlled by the balance between two opposing forces: random interactions with chromatin modifying activities (noise) can promote transitions in single nucleosomes, whereas positive feedback loops maintain the existing state in the domain (Dodd et al., 2007). Applying this model to the genes studied here, where our gene reporter assays demonstrate that the default state is active, we predict that random interaction or inefficient recruitment of H3K9 deacetylases and/or methylases can initiate transition to the repressed state by altering the methylation/acetylation balance at H3K9 and triggering the formation of heterochromatin (Fig. 7). The model for bistable domains predicts that it occurs at low frequencies because of euchromatin positive feedback loops, consistent with the experimental observation of stable expression patterns (Cortés et al., 2007). The repressed state is also stable, as expected if random interaction with enzymes with opposite activity is confronted by heterochromatin positive feedback loops. The model can also explain the spontaneous nature of the transitions between the active and repressed states for the genes studied here. The regulation of var genes is somehow different, as their default state is repressed, suggesting that their promoters are able to recruit repressive complexes efficiently, either directly or through interaction with var gene-specific factors such as PfSIP2 (Flueck et al., 2010).

Independent regulation of variantly expressed genes located in the same chromosomal region poses an additional layer of complexity, as heterochromatin has the capacity to spread (Grewal and Jia, 2007). Two well-differentiated patterns of histone modifications were observed in the gene eba-140. In the upstream region and first half of the ORF, H3K9ac and H3K9me3 correlated with expression or repression of the gene, respectively, as expected. Opposite modifications were observed in the second half of the ORF. We found that the state of the 3′ chromatin domain coincides with the chromatin state at the intergenic region between eba-140 and the next downstream gene, MAL13P1.59 (Fig. 6B). This gene, which encodes a PHISTa family exported protein of unknown function (Sargeant et al., 2006), shows opposite patterns of expression to eba-140 in both 10G and 1.2B. In both parasite lines the transition between the two chromatin domains occurs within the first exon of eba-140, suggesting the existence of a barrier insulator within this exon (Fig. 6B). Barrier insulators are boundary elements that limit the spreading of heterochromatin (Gaszner and Felsenfeld, 2006). The predicted barrier insulator within eba-140 ORF would prevent spreading of heterochromatin from the eba-140 regulatory region into MAL13P1.59 when eba-140 is repressed, and it would also prevent spreading of heterochromatin from MAL13P1.59 into the eba-140 regulatory region when MAL13P1.59 is repressed. Consequently, the insulator would ensure independent variant expression of the two neighbouring genes. In P. falciparum there are many neighbouring genes that can have opposite patterns of variant expression, as is the case of eba-140 and MAL13P1.59, or clag3.1 and clag3.2, and activation of a var gene does not affect repression of its neighbouring var genes (Scherf et al., 2008). Furthermore, knocking out the histone deacetylase PfSIR2A deregulated specific var genes, but this was restricted to discrete genes rather than affecting large chromosomal regions (Lopez-Rubio et al., 2009). This suggests that subtelomeric regions of the P. falciparum genome, where variantly expressed genes are typically located, are organized into small chromatin domains separated by barrier insulators to prevent heterochromatin spreading. This would enable independent variant expression of genes in the same chromosomal region, increasing the number of possible combinations of expressed and repressed genes. The identification of sequences containing insulators is a first step towards the characterization of these important chromatin elements.

Experimental procedures


Plasmodium falciparum 3D7 stocks 3D7-A and 3D7-B, and 3D7-A isogenic subclones 10G and 1.2B (Cortés et al., 2004; 2007; Cortés, 2005), were cultured according to standard procedures in media containing Albumax II (Invitrogen) and no human serum. For nuclease digestions or ChIP analysis, parasites were harvested at the early trophozoite stage (no multinucleated schizonts observed by microscopic examination of Giemsa-stained smears) or tightly synchronized to a 5 h window and harvested at the late schizont stage (40–45 h postinvasion). Tight synchronization was achieved by Percoll or magnet purification of schizonts followed by sorbitol lysis after 5 h.


Transcription start sites were determined using the FirstChoice 5′ RLM-RACE kit (Ambion) according to manufacturer's instructions. First strand cDNA synthesis was carried out using random hexamers on a mixture of total RNA derived from synchronized schizont preparations of 3D7-A and 3D7-B. Nested PCR was performed using gene specific primers (Table S1) and primers specific to the RACE adaptor (supplied by the kit). Amplified PCR products were cloned into pCR-2.1 or pCR2.1-TOPO vector using either TA or TOPO-TA cloning kits (Invitrogen) and sequenced using M13 forward and reverse primers.

Transfections and luciferase assays

Plasmids derived from HLH (Deitsch et al., 2001) and HRH (Amulic et al., 2009) were used for luciferase assays. Both plasmids contain the hrp3 promoter, controlling expression of either firefly luciferase (HLH) or Renilla luciferase (HRH). Serial 5′ truncations of the upstream regions of clag3.1, clag3.2 and eba-140 were PCR-amplified from 3D7-A genomic DNA (primers described in Table S1) and subcloned into the Asp718I and NsiI sites of plasmid HLH, replacing the hrp3 promoter. A fragment of the ama1 upstream region (1569 bp) was also PCR amplified and cloned into both HLH and HRH. To generate a plasmid that contains the firefly luciferase gene but has no promoter driving its expression, the hrp3 promoter of HLH was replaced with a linker prepared by annealing primers LinkAspNsiFw and LinkAspNsiRev (Table S1), creating the plasmid linker-LH. To generate a plasmid with the luciferase cassette under the control of the clag3.2 promoter followed by rep20 subtelomeric repeats, a PstI- BglII fragment with 0.5 kb of rep20 sequence from the pHBupsCR plasmid (Voss et al., 2006) was cloned into PstI-BamHI sites of the plasmid 3.2-1371-LH (Table S1), downstream from the luciferase cassette. This plasmid was termed 3.2-1371-LH-R20.

For transient transfection assays, non-infected erythrocytes were transfected with 50 µg of test plasmid under standard conditions (Crabb et al., 2004). To control for potential problems with transfection or preparation of parasite extracts, all test plasmids were co-transfected with 10 µg of a control plasmid (HRH for experiments with trophozoites or ama1-RH for experiments with schizonts). Percoll-purified schizonts (3D7-A parasite line, unless otherwise stated) were added to plasmid-loaded erythrocytes to a final parasitaemia of 0.8% adjusted using a Countess automated cell counter (Invitrogen), with a final haematocrit of 2%. To determine promoter activity at the trophozoite and schizont stages, parasites were harvested 24 h or 48–50 h later respectively. Parasites were isolated by lysis of erythrocytes with 0.1% saponin in PBS, then washed in PBS and lysed at room temperature with Renilla lysis buffer (Promega). Supernatants were assayed for firefly and Renilla activities using luciferase assay systems from Promega and a Lumat LB 9507 luminometer (Berthold Technologies).

Nuclease digestions

Parasites obtained by saponin lysis were incubated in NP-40-containing buffer and subjected to partial DNA digestion with MN (Fermentas) for 3 min at 37°C as previously described (Duraisingh et al., 2005). For partial chromatin digestions with RE, after incubation in the same NP-40 containing buffer, parasites were spun, resuspended in a solution containing the RE in the buffer recommended by the manufacturer, and incubated at 37°C for 40 min. MN and RE digestions were processed identically from there onwards. After purification, DNA was digested with a RE (MfeI or NdeI, as indicated in the figures), resolved in 1.25% agarose gels and transferred to Amersham Hybond N+ nylon membranes (GE Healthcare) for end-labelling by Southern blot with specific probes labelled with α-32P dATP (Perkin Elmer). The primers used to PCR amplify the probes are described in Table S1. All probes were specific, including the probes for the 95% identical genes clag3.1 and clag3.2, as indicated by the presence of a single band in non-digested samples. Hybridization and washes were performed at 62°C, and membranes were exposed to X-ray film and also analysed quantitatively with a Typhoon 8600 storage phosphor imager (Molecular Dynamics).

ChIP and qPCR

ChIP experiments were carried out essentially as described (MR4 Methods in Malaria Research, 5th edition, http://www.mr4.org). Briefly, trophozoites or tightly synchronized schizonts were released from erythrocytes by saponin lysis. Cross-linked chromatin was obtained by treating parasites with formaldehyde to a final concentration of 1% for 5 min at 37°C. Parasites were washed several times in PBS and resuspended in cold lysis buffer without NP-40, transferred to a chilled Dounce homogenizer (Kimble Chase), and incubated on ice for 30 min. Nuclei were obtained by the addition of NP-40 to a final concentration of 0.25% followed by 200 strokes with a small-clearance pestle (pestle B). Nuclei pellets were resuspended in a buffer containing 1% SDS and sonicated in a Bioruptor (Diagenode) on high power, until DNA fragments ranged between 500 and 1000 bp in size. Immunoprecipitations were carried out with approximately 1 µg of DNA with 1:100 dilutions of anti-H3K9me3 (Millipore, #07-442), anti-H4ac (H4 acetylated at K5/K8/K12/K16, Millipore, #06-866), anti-H3K9ac (Millipore, #07-352), anti-H3K4me3 (Millipore, #05-745), anti-H3 (Abcam, #ab1791), or no antibody (negative control). After washing, reversal of cross-links and proteinase K treatment, immunoprecipitated DNA was purified by standard phenol/chloroform methods, ethanol-precipitated and analysed by qPCR, using the relative standard curve method. All qPCRs were performed in triplicate in a StepOnePlus Real Time PCR System (Applied Biosystems) using Power SYBR Green Master Mix (Applied Biosystems) and the primers listed in Table S1. All ChIP experiments were performed in triplicate, using independent biological samples.


We thank A. Vaquero (IDIBELL, Barcelona) for useful comments on the manuscript, K. W. Deitsch (Cornell University, NYC) for the plasmids HLH and HRH, T. S. Voss (STI, Basel) for the plasmid pHBupsCR and J. J. López-Rubio (Institut Pasteur-CNRS, Paris) for advice with qChIP. This work was supported by Plan Nacional de Investigación Científica, Desarrollo e Innovación Tecnológica (I+D+I), Instituto de Salud Carlos III – Subdirección General de Evaluación y Fomento de la Investigación (Grant PI07/0891 to A.C.); V.M.C. acknowledges receipt of a fellowship from Institute for Research in Biomedicine, Barcelona.