Histone variants are important components of eukaryotic chromatin and can alter chromatin structure to confer specialized functions. H2B variant histones are rare in nature but have evolved independently in the phyla Apicomplexa and Trypanasomatida. Here, we investigate the apicomplexan-specific Plasmodium falciparum histone variant Pf H2B.Z and show that within nucleosomes Pf H2B.Z dimerizes with the H2A variant Pf H2A.Z and that Pf H2B.Z and Pf H2A.Z occupancy correlates in the subset of genes examined. These double-variant nucleosomes also carry common markers of euchromatin like H3K4me3 and histone acetylation. Pf H2B.Z levels are elevated in intergenic regions across the genome, except in the var multigene family, where Pf H2A.Z/Pf H2B.Z double-variant nucleosomes are only enriched in the promoter of the single active var copy and this enrichment is developmentally regulated. Importantly, this pattern seems to be specific for var genes and does not apply to other heterochromatic gene families involved in red blood cell invasion which are also subject to clonal expression. Thus, Pf H2A.Z/Pf H2B.Z double-variant nucleosomes appear to have a highly specific function in the regulation of P. falciparum virulence.
Eukaryotic genomes are organized into nucleosomes that consist of DNA wrapped around an octamer of histone proteins. The composition of the histones within the nucleosomes and the histones' post-translational modifications are important for determining whether the DNA is packaged into compact, transcriptionally silent heterochromatin or relaxed, transcriptionally inducible euchromatin (Campos and Reinberg, 2009). The genome of the malaria parasite Plasmodium falciparum has an unusually euchromatic composition during its intraerythrocytic lifecycle, but with islands of heterochromatin that stretch from the telomeres into subtelomeric domains and that are also scattered through the chromosome bodies (Flueck et al., 2009; Lopez-Rubio et al., 2009; Salcedo-Amaya et al., 2009). During the intraerythrocytic lifecycle the majority of the parasite's genome is transcribed in a tightly regulated transcriptional cascade (Bozdech et al., 2003), presumably facilitated by maintaining most of its genes in an inducible, euchromatic state. Conversely, the islands of heterochromatin maintain clusters of genes in a silent state and are defined by histone 3 trimethylation at lysine 9 (H3K9me3) and the heterochromatin protein PfHP1 (Flueck et al., 2009; Lopez-Rubio et al., 2009; Salcedo-Amaya et al., 2009). These heterochromatic clusters are enriched in genes that encode proteins involved in host–parasite interactions, many of which have redundant functions and/or are present as multigene families.
In addition to the four canonical histones H2A, H2B, H3 and H4, numerous variants have been described across eukaryotes. Histone variants replace their canonical histone counterparts in nucleosomes at certain sites in the genome and can create structurally and functionally specialized chromatin domains (Flaus et al., 2004; Kobor et al., 2004; Raisner and Madhani, 2006). While variants of H2A and H3 are widely distributed throughout evolution, variants of H2B and H4 are less common. Protist parasites of the trypanosomatids and apicomplexa are unique in possessing lineage-specific variants of H2B. Originally both lineages were called H2Bv (Dalmasso et al., 2011) but the apicomplexan variants were recently reclassified as H2B.Z to reflect their independent evolution (Talbert et al., 2012). Both Trypanosoma brucei H2Bv and Toxoplasma gondii H2B.Z have been shown to exclusively dimerize with the alternative histone H2A.Z (Dalmasso et al., 2009; Siegel et al., 2009). In T. brucei H2A.Z/H2Bv double-variant nucleosomes show genome-wide enrichment at Pol II transcription start sites (TSS) of polycistronic transcription units (Siegel et al., 2009), and in T. gondii H2A.Z/H2B.Z double-variant nucleosomes are associated with the TSS of several active genes (Dalmasso et al., 2009). Consistent with a role in promoter structure and gene regulation, H2A.Z/H2Bv or H2A.Z/H2B.Z nucleosomes in both parasites are also associated with active chromatin marks (Mandava et al., 2008; Dalmasso et al., 2009; Siegel et al., 2009). The P. falciparum H2B.Z (Pf H2B.Z) has not yet been studied in detail, and it is unknown how Pf H2B.Z relates to other histones and to gene expression. Pf H2B.Z differs from H2B mainly in the N-terminal tail and in the second alpha helix of the histone fold domain, while the C-terminus is highly conserved (Miao et al., 2006). Pf H2B.Z, but not H2B, was found to be acetylated at five lysine residues in the N-terminus, whereas H2B is ubiquitinated in the shared residue lysine 112 (Trelle et al., 2009), a site that is known to be required for H3K4 and H3K79 methylation in other organisms (Sun and Allis, 2002). Acetylated histone tails can affect chromatin structure and gene regulation by directly recruiting trans factors or by destabilizing the nucleosome (Campos and Reinberg, 2009) so the unique acetylations of Pf H2B.Z may confer a specialized function.
The H2A variant H2A.Z is widespread throughout evolution and is enriched in gene promoters and 5′ untranslated regions (5′ UTRs) and associated with euchromatic post-translational modifications of histones in organisms as diverse as humans and yeast (Barski et al., 2007; Schones et al., 2008; Buchanan et al., 2009). In P. falciparum Pf H2A.Z is also enriched around the TSS and in the 5′ UTRs of euchromatic genes together with the euchromatic histone modifications H3K4me3 and H3K9ac (Bartfai et al., 2010; Petter et al., 2011). H2A.Z in the promoter and around the TSS poises yeast genes for activation (Li et al., 2005) and recruits the yeast and metazoan transcription machinery (Hardy et al., 2009; Svotelis et al., 2009) but H2A.Z is then lost from yeast and human genes following activation (Zhang et al., 2005; Hardy et al., 2009). Consistent with these observations, H2A.Z enrichment in promoters and around the TSS does not correlate with active gene expression in many organisms, including P. falciparum (Allis et al., 1980; Guillemette et al., 2005; Raisner et al., 2005; Bartfai et al., 2010; Petter et al., 2011). Surprisingly, this is in contrast to the closely related T. gondii, where H2A.Z/H2B.Z nucleosomes have only been observed in active genes, whereas nucleosomes at inactive loci contain H2A.X (Dalmasso et al., 2009). Acetylation of H2A.Z may affect transcription by the decreased stability of its association with chromatin (Bruce et al., 2005; Millar et al., 2006; Thambirajah et al., 2006) and, interestingly, P. falciparum H2A.Z has eight acetylations in its N-terminal tail compared to the five acetylations in yeast H2A.Z (Trelle et al., 2009).
Histone modifications and alternative histones are both implicated in the regulation of expression of the immunodominant variant antigen and cytoadhesin P. falciparum erythrocyte membrane protein 1 (PfEMP1). PfEMP1 proteins are encoded by the var multigene family (Baruch et al., 1995; Smith et al., 1995; Su et al., 1995) which has approximately 60 members per genome (Rask et al., 2010). The var genes are located in clusters within the heterochromatic regions but a single var gene escapes silencing and is abundantly transcribed in a defined perinuclear domain in immature parasites (Duraisingh et al., 2005). In its active state, the var gene carries the euchromatic marks H3K4me3 and H3K9ac around the transcription start site (TSS) (Lopez-Rubio et al., 2007). As the parasite matures the active var gene is transiently repressed but is marked with H3K4me2 that probably poises the gene for reactivation in the progeny (Lopez-Rubio et al., 2007). The parasite employs partially understood epigenetic mechanisms to switch between the exclusively transcribed var genes (Scherf et al., 1998). These mechanisms are associated with various histone modifications and require the var gene promoter (Dzikowski et al., 2006; Voss et al., 2006), a bidirectional promoter within the var gene intron (Deitsch et al., 2001) and the PfSir2A & B histone deacetylases (Duraisingh et al., 2005; Tonkin et al., 2009). Pf H2A.Z is depleted in P. falciparum heterochromatin (Bartfai et al., 2010; Petter et al., 2011) and therefore also depleted in the 5′ intergenic region of most var genes; however, the single active var gene 5′ intergenic region is enriched in Pf H2A.Z, but only when it is expressed in immature stages. Pf H2A.Z is then lost from the active var gene as it is repressed in the maturing parasite (Petter et al., 2011). In a subset of var genes the depletion of Pf H2A.Z both from the transiently repressed active var gene and from the silent var genes is dependent on the histone deacetylase PfSir2A (Petter et al., 2011). Pf H2A.Z is also enriched at var gene introns which contain chromatin-dependent, bidirectional promoter activity important for var gene silencing (Deitsch et al., 2001). Whether Pf H2B.Z is also implicated in var gene regulation has not yet been investigated.
Plasmodium falciparum utilizes several, redundant pathways for erythrocyte invasion, some of which are also mediated by phase variably expressed parasite proteins (Stubbs et al., 2005), and some of these also seem to exhibit mutually exclusive expression (Cortés et al., 2007). Phase variable expression of some invasion genes also correlates with acetylation of the histones in their 5′ intergenic region (Jiang et al., 2010; Comeaux et al., 2011; Crowley et al., 2011) and several of these invasion genes are also located within the heterochromatin islands (Flueck et al., 2009). However, it is unknown whether alternative histones are involved in the regulation of invasion gene expression.
Here, we have investigated Pf H2B.Z and show that: (i) within nucleosomes Pf H2B.Z associates with euchromatic histone modifications and Pf H2A.Z, (ii) Pf H2A.Z/Pf H2B.Z double-variant nucleosomes are enriched in intergenic regions across the genome, (iii) Pf H2A.Z/Pf H2B.Z double-variant nucleosomes are also present at var introns and in active var promoters, but they are lost from the latter when in the poised state and (iv) Pf H2A.Z/Pf H2B.Z double-variant nucleosomes are depleted from silent var gene promoters, but not from silent promoters of heterochromatic invasion gene families that have similar patterns of variegated expression. Thus, our study reveals an important and unique function of the double-variant nucleosomes in virulence gene regulation of the malaria parasite.
Pf H2B.Z and Pf H2A.Z are components of the same nucleosomes which carry euchromatic histone modifications
The P. falciparum histone variant Pf H2B.Z is encoded by the gene PF07_0054. We created a parasite line expressing Pf H2B.Z fused to haemagglutinin (HA) under the control of the Pf H2B.Z promoter from an episomally maintained plasmid (Fig. 1A). We used the endogenous Pf H2B.Z promoter in order to avoid artefacts caused by incorrect expression timing, which can have severe functional implications in P. falciparum (Treeck et al., 2006). The total level of H2B.Z in the H2B.ZHA-expressing parasites was similar to wild type [expression relative to WT from qPCR analysis: median ring stages 0.74 (IQR 0.63, 0.83); median schizonts 1.31 (IQR 0.61, 1.45)]. This was due to the relatively low level of expression of H2B.ZHA compared to endogenous H2B.Z in the transfected parasite population; from absolute quantitative RT-PCR using plasmid standards there was 63-fold more endogenous H2B.Z than HA-tagged H2B.Z in rings and 400-fold more in schizonts. Expression of the 17.5 kDa Pf H2B.ZHA fusion protein exclusively in transfected parasites was verified by Western blot analysis (Fig. 1B).
To investigate the subcellular localization of Pf H2B.Z in relation to the variant histone Pf H2A.Z and markers of active or inactive chromatin we performed immunofluorescence assays (Fig. 2A). We found that Pf H2B.Z and Pf H2A.Z colocalized very well and thus occupy the same subnuclear region, which largely overlaps with the DAPI-stained DNA except for a small area in which alternative histone staining was less intense. Pf H2B.Z also colocalized with the euchromatin marker H3K4me3, but not with the heterochromatin marker H3K9me3. These results are consistent with our previous observations of Pf H2A.Z being enriched in the euchromatin compartment (Petter et al., 2011) and point towards a function of Pf H2B.Z related to Pf H2A.Z and euchromatin.
In order to examine the composition of Pf H2B.Z-containing nucleosomes, we performed co-immunoprecipitation experiments on mononucleosomes. The precipitated material was separated by SDS-PAGE and analysed by Western blot. Antibodies to HA co-precipitated Pf H2B.ZHA with Pf H2A.Z and Pf H3. Similarly, antibodies to Pf H2A.Z co-precipitated Pf H2B.Z and Pf H3 (Fig. 2B), indicating that the majority of nucleosomes containing Pf H2B.Z also contained Pf H2A.Z. Antibodies to Pf H2A did not co-precipitate detectable Pf H2A.Z indicating that Pf H2A.Z and Pf H2A do not occur in the same nucleosomes (Fig. 2C). Antibodies to H2A precipitated similar levels of Pf H3 as antibodies to Pf H2B.ZHA indicating that the precipitated histones were part of intact nucleosomes and that equivalent quantities of nucleosomes had been analysed by Western blot. Control immunoprecipitations with non-immune serum and IgG did not give any signal. We conclude that Pf H2B.Z and Pf H2A.Z are associated with each other within nucleosomes, as observed in T. brucei and T. gondii (Dalmasso et al., 2009; Siegel et al., 2009).
To further investigate the composition of Pf H2A.Z/Pf H2B.Z-containing nucleosomes with respect to histone modifications, we co-precipitated mononucleosomes using anti-Pf H2A.Z serum and then tested them for various previously characterized histone modifications (Fig. 2D). We found that Pf H2A.Z co-precipitates with the euchromatic modifications H3K4me3, as shown previously (Petter et al., 2011), as well as H3K9ac and H4K12ac. However, Pf H2A.Z did not co-precipitate H4K20me3, previously shown to be enriched at repressed genes. These data indicate that Pf H2A.Z/Pf H2B.Z nucleosomes are present in euchromatic regions of the genome and are largely absent from the repressed heterochromatin domains.
Pf H2A.Z/Pf H2B.Z nucleosomes are enriched in 5′ intergenic regions across most of the genome
To explore the genome-wide profile of Pf H2B.Z occupancy, we performed genome-wide chromatin immunoprecipitation in conjunction with microarray analysis (ChIP-on-chip) using antibodies against HA. Mononucleosomal chromatin of ring, trophozoite and schizont stage parasites of the Pf H2B.ZHA parasite line was analysed. A novel long-oligonucleotide microarray was used with 10 416 probes representing 5343 coding sequences and with an additional specific probe within 1.5 kb upstream of each start codon for each 5′ intergenic region (IGR). IGRs represent 40% of the P. falciparum genome and the mean IGR length is 1620 bp (median 1231 bp).
To characterize Pf H2B.Z enrichment along genes and their 5′-IGR, we binned all array probes according to their distance from the start codon into 500 bp bins. For each distance bin, the proportion of probes that exhibited high, medium or low Pf H2B.Z levels (log2 ratio Pf H2B.Z/input) was calculated (Fig. 3). This analysis showed that in all three stages, high Pf H2B.Z levels were clearly more frequent in the 5′-IGRs than elsewhere in genes, and less frequent within the first 500 bp following the start codon than elsewhere (Fig. 3, upper panel). Accordingly, very low Pf H2B.Z levels were more frequent within the first 500 bp from the start codon than elsewhere (Fig. 3, lower panel). Moderate levels of Pf H2B.Z appeared to be present throughout the open reading frames (ORFs) (Fig. 3, middle panel). A very similar pattern was observed for Pf H2A.Z when the experiment was repeated using cross-linked chromatin (Fig. S1). These findings are consistent with previous evidence that Pf H2A.Z is enriched in euchromatic IGRs (Bartfai et al., 2010; Petter et al., 2011) and our finding that Pf H2B.Z and Pf H2A.Z dimerize within the same nucleosomes.
To validate the trend towards Pf H2B.Z enrichment in IGRs observed from the genome-wide ChIP-on-chip results, we investigated the Pf H2A.Z/Pf H2B.Z occupancy of individual loci by ChIP and qPCR. Thirteen genes were analysed which represent different states of activity, including constitutively active genes (Hsp86, casein kinase), stage specifically transcribed genes (Eba140, MSP2, Eba175, Rex1, SBP1, FIKK PFL0040c, AC7 PFL0035c, phistc2) and silent genes (CSP, SSP2, hypothetical PF11_0480). qPCR reactions targeted regions in the 5′-IGR and ORF. In all genes, Pf H2B.Z showed significant enrichment in the 5′-IGR as opposed to the ORFs. This was observed in both ring and schizont stage parasites (Figs 4A and S2A). Likewise, and consistent with our previous results, Pf H2A.Z enrichment at the 5′-IGR of these genes was also high regardless of their transcriptional status (Figs 4A and S2A) (Petter et al., 2011). The enrichment of both Pf H2A.Z and Pf H2B.ZHA at individual genes is presented in Fig. S2B and S2C. In line with the cooperative nature of Pf H2A.Z and Pf H2B.Z, enrichment of the two histone variants was highly correlated (Spearman, rings R = 0.83, schizonts R = 0.91) (Fig. 4B). Thus, these data provide evidence that Pf H2B.Z and Pf H2A.Z double-variant nucleosomes are enriched in the 5′-IGR, irrespective of the transcriptional status of the associated genes.
var gene promoters contain Pf H2A.Z/Pf H2B.Z nucleosomes only when actively transcribed
Previously we and others found that Pf H2AZ is maintained at the TSS of genes regardless of their transcriptional state (Bartfai et al., 2010; Petter et al., 2011). However, we also revealed a notable exception in the TSS of var genes from which Pf H2AZ is depleted when the var genes are silent, or when the single, expressed var gene is transiently repressed in mature schizont parasites (Petter et al., 2011). Because the 60 var genes are a minor component of the genome we could not predict the association of Pf H2A.Z and Pf H2B.Z at var TSSs from the immunoprecipitations shown in Fig. 2. Indeed the atypical expression-dependent occupancy of Pf H2A.Z at the TSS of var genes was suggestive of a difference in composition of the Pf H2A.Z-containing nucleosome at the var TSS compared to nucleosomes at the TSS of other genes. To extend our analysis of the atypically regulated nucleosome at the var TSS we used ChIP qPCR to determine the association between H2B.Z occupancy and var gene expression.
Chromatin and RNA were harvested from parasite cultures which were selected for high expression of the var2csa gene, as well as from unselected parasites that do not transcribe var2csa. Expression levels of the var2csa gene in both populations were verified by q-RT-PCR of the entire var gene repertoire (Fig. S3A). ChIP and qPCR analysis of Pf H2A.Z and Pf H2B.Z complexes was performed at ring and schizont stages with primers targeting seven different regions along the 5′-IGR and the ORF of the var2csa gene. To normalize for ChIP efficiency, the enrichment of each histone variant is presented as a ratio to the level of enrichment of the same histone variant in the ORF of the constitutively expressed, euchromatic gene hsp70 (Fig. 5A). We found Pf H2B.Z to be highly enriched only near the TSS [predicted to be at about −1200 bp (Petter et al., 2011)] of the actively transcribed var2csa gene in ring stage parasites (Fig. 5A). In accordance with our previous results, Pf H2A.Z was also highly enriched exclusively at the active var2csa TSS (Fig. 5A). In contrast, both Pf H2B.Z and Pf H2A.Z were depleted from the var2csa promoter at schizont stage, when the var2csa gene is repressed but poised for transcription in the next replication cycle (Fig. 5A). Pf H2B.Z and Pf H2A.Z were also both depleted near the TSS of the silent var2csa in unselected ring and schizont parasites (Fig. 5B) and near the TSS of the silent var gene PFL0020w (var20) (mapped by RT-PCR to between −429 and −473, data not shown). Together, these data reveal that Pf H2B.Z and Pf H2A.Z are both present at the active var TSS, and that occupancy of Pf H2A.Z/Pf H2B.Z double-variant nucleosomes correlates with var gene transcription and is developmentally regulated.
To verify whether this observation is generally true for var genes, we tested the presence of both Pf H2B.Z and Pf H2A.Z in a Pf H2B.ZHA parasite population that was selected for dominant expression of a different var gene, var20, by selection for binding to ICAM-1 (Fig. S3B). ChIP analysis confirmed that Pf H2B.Z and Pf H2A.Z are both enriched near the TSS of var20 in ring stage parasites actively expressing this var gene (Fig. 5C).
Pf H2A.Z/Pf H2B.Z nucleosomes are enriched at var introns
var genes have a partially conserved intron with bidirectional promoter activity that generates sense and anti-sense transcripts which are important for silencing of the upstream var promoter (Frank et al., 2006; Epp et al., 2009). Earlier studies by ourselves and others showed that Pf H2A.Z is significantly enriched at the introns of var genes (Bartfai et al., 2010; Petter et al., 2011), thus we were interested in whether Pf H2B.Z is also associated with var introns. For both ring and schizont stage parasites ChIP and subsequent qPCR were carried out on eight silent var genes, which therefore were not enriched in H2B.Z and H2A.Z near their TSS. The primer pairs used spanned the var ORF/intron boundaries as well as regions in the ORF and within the 5′-IGR near the predicted TSS of the respective var genes. The TSSs were predicted using MAPP (Brick et al., 2008) (available on previous versions of PlasmoDB) and the proximity of the 5′-IGR primers to the TSS was previously confirmed by ChIP qPCR (Petter et al., 2011). Pf H2B.Z was significantly enriched at var introns compared to the IGR or ORF in both ring and schizont stage parasites (Figs 5D and S2D). Pf H2A.Z occupancy was also clearly higher at the var introns, consistent with our previous results (Figs 5D and S2D). All comparisons within two biological replicates were statistically significant with one exception; however, the trend in the outlier was still similar (Fig. S2D, Pf H2B.Z IGR versus intron). Therefore we conclude that Pf H2B.Z also interacts with Pf H2A.Z in var gene introns.
Invasion gene families show no fluctuation in Pf H2A.Z/Pf H2B.Z occupancy
The correlation of Pf H2A.Z/Pf H2B.Z nucleosome occupancy with var gene expression raised the question of whether other gene families that express different variants in a clonal fashion are regulated through similar mechanisms. The P. falciparum invasion-related pfRh, rhopH1/clag and eba gene families exhibit clonal expression and are regulated through epigenetic mechanisms (Cortés et al., 2007; Jiang et al., 2010; Comeaux et al., 2011; Crowley et al., 2011). Some of these genes have been shown to be located in heterochromatin domains (Flueck et al., 2009), and intergenic heterochromatin is depleted of Pf H2A.Z (Bartfai et al., 2010) suggesting that depletion of Pf H2A.Z/Pf H2B.Z from the promoters of silent var genes might be a regulatory mechanism shared with other heterochromatic genes.
To determine transcription levels of the invasion gene family members in our Pf H2B.ZHA cell line, RNA from ring and schizont stage parasites was harvested and the transcription levels assessed by q-RT-PCR. These transcript levels were then compared to the level of transcription of the hypothetical PF11_0480 gene (hyp), which is in the lowest 10th percentile of genes transcribed by the parasite at both ring and schizont stages. This analysis showed clag 2, clag3.1 and clag3.2 were significantly more highly expressed than the repressed hyp gene at schizont stages (P < 0.05) but not ring stages (Fig. 6A). In contrast, expression levels of rh2a, rh2b, eba140, eba165, rh1 and rh4 were not significantly different to the silent hyp gene in either schizont or ring stages (Fig. 6A). As a control for correct staging, transcription of the single copy, schizont stage expressed euchromatic msp1 gene was also shown to be significantly upregulated relative to the hyp gene in our schizont stage parasites. ChIP and qPCR with primers targeting areas near the TSS and within the ORF was carried out on ring and schizont stage parasites for eba140, eba165, rh1, rh2a, rh2b, rh4, clag2, clag3.1, clag3.2 and casein kinase as a control. Interestingly, in both schizont and ring stages there was no correlation between the levels of transcription and the ratio of Pf H2B.Z or Pf H2A.Z levels in the IGRs relative to their levels in the ORFs (all correlations P > 0.43, Spearman rank two-tailed). Thus Pf H2A.Z and Pf H2B.Z were enriched at the TSS of all of these genes at schizont stage, regardless of whether there was evidence for expression or not (Fig. 6B). Similarly, Pf H2A.Z/Pf H2B.Z occupancy remained high at the TSS at ring stage, when these genes assume a transiently silent state poised for transcription later on during the life cycle (Fig. 6C). This is in contrast to our observations for var genes and indicates that the involvement of Pf H2A.Z/Pf H2B.Z in var gene regulation is a highly specific regulatory mechanism of this important virulence protein family, and is not a general mechanism regulating heterochromatic gene families.
In this work we have investigated the novel P. falciparum histone variant Pf H2B.Z. Our analyses reveal that Pf H2A.Z and Pf H2B.Z are intimately linked and form double-variant nucleosomes. These Pf H2B.Z and Pf H2A.Z nucleosomes are enriched at 5′ intergenic regions throughout the genome and at introns of the var genes, which are critical for virulence. Occupancy by Pf H2A.Z and Pf H2B.Z nucleosomes generally seems not to be linked to transcription, except in the case of var gene promoters which are devoid of Pf H2A.Z/Pf H2B.Z double-variant nucleosomes, unless they are actively transcribed in ring stage parasites. This indicates that a distinct mechanism of regulation involving Pf H2A.Z/Pf H2B.Z nucleosomes is important for clonal var gene expression. Importantly, we discovered that other heterochromatic, clonally expressed P. falciparum gene families which are involved in invasion and are also subject to epigenetic regulation do not conform to this pattern. This highlights that the selective deposition of Pf H2A.Z/Pf H2B.Z nucleosomes at var promoters only when they are active is a highly specialized mechanism inherent to this gene family.
Pf H2A.Z and Pf H2B.Z – a faithful pair
Our co-immunoprecipitation results suggest that Pf H2A.Z and Pf H2B.Z frequently occur within the same nucleosomes (Fig. 2), consistent with previous studies showing that the lineage-specific H2Bv and H2B.Z variants of the trypanosomatid parasite T. brucei and the apicomplexan parasite T. gondii solely associate with H2A.Z (Lowell et al., 2005; Dalmasso et al., 2009). The canonical histone Pf H2A did not co-precipitate Pf H2A.Z; thus we found no evidence in P. falciparum of heterotypic nucleosomes containing one copy each of Pf H2A.Z and Pf H2A, although we cannot exclude the possibility that they exist as a small subpopulation as observed in Drosophila and human cells (Viens et al., 2006; Weber et al., 2010).
The existence in P. falciparum of a variant of H2B which frequently dimerizes with Pf H2A.Z is intriguing. It is possible that the pairing of Pf H2A.Z with Pf H2B.Z represents an adaptation to the unique features of Pf H2A.Z in comparison to H2A.Z variants from other organisms, such as differences in the L1 loop and N-terminus (Petter et al., 2011). In support of this theory, most differences between Pf H2B.Z and its canonical counterpart are found in the highly modified N-terminus and in the alpha 2 helix (Miao et al., 2006), which are both critical for dimerization and stability of H2A/H2B heterodimers (Placek and Gloss, 2002). The double-variant nucleosomes' function may depend on their stability: T. brucei H2Bv/H2A.Z nucleosomes are less stable than nucleosomes containing their canonical counterparts (Siegel et al., 2009) suggesting that they affect chromatin structure and gene expression through reduced stability, as has been observed for vertebrate H2A.Z/H3.3 double-variant nucleosomes (Jin and Felsenfeld, 2007). Pf H2B.Z may also aid targeting of Pf H2A.Z nucleosomes to particular regions. A precedent for this possibility exists in humans where H3.3 confers a preference for CA/TG dinucleotide repeats on H3.3/H2A.Z nucleosomes, suggesting that H3.3 may position the H2A.Z-containing nucleosomes where they can affect chromatin structure and gene expression (Le et al., 2010).
We found that Pf H2A.Z/Pf H2B.Z-containing nucleosomes also carry the euchromatic marks H3K4me3, H3K9ac and H4K12ac (Fig. 2). It is known from genome-wide studies that H3K4me3, H3K9ac and Pf H2A.Z all predominantly occupy intergenic regions in P. falciparum (Salcedo-Amaya et al., 2009; Bartfai et al., 2010), and our experiments expand this knowledge by showing that these marks indeed all occur in identical nucleosomes, rather than neighbouring ones. The level of H3K9ac correlates well with transcription throughout intraerythrocytic differentiation (Bartfai et al., 2010), thus H3K9ac in double-variant nucleosomes appears to differentiate active and inactive genes. Acetylation at promoters is critical for the binding of protein complexes that facilitate gene expression; for example, H3 acetylation recruits TFIID to promoters resulting in initiation of transcription (Matangkasombut et al., 2000). H4K12ac has not been studied in P. falciparum, but is enriched at gene promoters in human cells (Wang et al., 2008). Ubiquitination of H2B is required for methylation of H3K4 in diverse organisms (Chandrasekharan et al., 2010) but H2Bv has been proposed to replace H2B ubiquitination as the signal for H3K4 methylation in T. brucei (Mandava et al., 2008). This function might also apply to Pf H2B.Z which is present in the same nucleosomes as H3K4me3, although Pf H2B is ubiquitinated (Trelle et al., 2009). The repressive histone mark H4K20me3 was shown to be present throughout the genome in P. falciparum (Lopez-Rubio et al., 2009), but is absent from double-variant nucleosomes, indicating that H4K20me3 is unlikely to repress transcriptional activity from promoters occupied by double-variant nucleosomes. However, the function of H4K20me3 in P. falciparum is unknown.
Double-variant nucleosomes in 5′ intergenic regions
We show that Pf H2A.Z and Pf H2B.Z double-variant nucleosomes occupy 5′-IGRs and are thus likely associated with gene promoters (Figs 3 and 4), where they may serve to maintain an open chromatin structure favouring rapid gene activation. In contrast to the closely related apicomplexan parasite T. gondii (Dalmasso et al., 2009), Pf H2A.Z/Pf H2B.Z nucleosome occupancy does not generally correlate with transcriptional activity. T. gondii possesses a third H2A variant, H2A.X, which is associated with silent genes. No such variant is present in P. falciparum, identifying another major difference in the use and function of histone variants between the two apicomplexan parasites. Pf H2A.Z levels have previously been suggested to regulate the strength of a promoter but not its temporal activity, although the mechanisms remain unclear (Bartfai et al., 2010). Our genome-wide ChIP-on-chip approach clearly shows a preferential enrichment of high levels of Pf H2B.Z at 5′ intergenic sites (Fig. 3). However the higher resolution of the next generation sequencing study of Hoeijmakers et al. (2013) is required to properly query correlations between gene expression and the highly restricted enrichment of H2B.Z at the TSS of genes, such as was shown for H2A.Z in human cells (Valdes-Mora et al., 2012).
Fourteen of the 33 histone acetylations reported in P. falciparum are unique modifications of the N-terminal tails of Pf H2B.Z and Pf H2A.Z (Miao et al., 2006; Trelle et al., 2009). Clearly these acetylations are functionally important and could plausibly exert a dynamic influence on gene expression. Indeed, enrichment of acetylated H2A.Z at the promoter and around the TSS strongly correlates with transcription in yeast, chicken and human cells (Bruce et al., 2005; Millar et al., 2006; Valdes-Mora et al., 2012). In human cells, H2A.Z occupies promoters in a bimodal fashion, being spread out across inactive promoters in a deacetylated state, but highly localized at the TSS of active genes in an acetylated state (Valdes-Mora et al., 2012). Thus future studies of acetylated Pf H2A.Z and Pf H2B.Z may also reveal associations with gene regulation. Acetylation of variant histones in other organisms probably affects gene expression by direct effects on chromatin structure, as acetylation of H2A.Z decreases nucleosome stability (Thambirajah et al., 2006) and the N-terminal tails of T. brucei H2Bv and H2A.Z (Lowell et al., 2005) and the specific N-terminal tail acetylations of Tetrahymena thermophila H2A.Z (Ren and Gorovsky, 2003) are not essential and thus do not specifically recruit trans factors essential for gene expression. The divergence between the extended, acetylated N-terminal tails of both P. falciparum histone variants and human H2A.Z and H2B may provide opportunities for future drug development.
Double-variant nucleosomes in heterochromatic gene families
Interestingly, the static pattern that Pf H2A.Z/Pf H2B.Z nucleosomes exhibit throughout differentiation of the malaria parasite within many IGRs does not apply to var genes. In 5′-IGRs of var genes, Pf H2A.Z and Pf H2B.Z show enrichment only in a very defined region around the TSS which correlates with var gene activity (Fig. 5). var gene promoters present a special case of dynamic remodelling of the double-variant nucleosomes but Pf H2B.Z/Pf H2A.Z still appear to dimerize primarily with each other at var promoters. The temporary loss of Pf H2A.Z and presumably also Pf H2B.Z is dependent on the histone deacetylase PfSir2 (Petter et al., 2011) which forms silent heterochromatin and represses transcription of var genes (Duraisingh et al., 2005; Tonkin et al., 2009). The antagonistic relationship between PfSir2A and Pf H2A.Z at var promoters suggests that the acetylation state of Pf H2A.Z/Pf H2B.Z histones may be important in their unique, dynamic regulation at the var promoter and is also reminiscent of yeast H2A.Z which forms a boundary to protect subtelomeric genes from the ectopic spread of Sir-dependent, silent heterochromatin (Meneghini et al., 2003).
Transcriptional repression in the heterochromatin compartment also relies on the presence of H3K9me3 and HP1. The genes rh4 (Jiang et al., 2010), clag2, clag3.1, clag3.2, eba140 and rh2b (Crowley et al., 2011) have variegated expression and are members of small gene families involved in invasion. Silencing of the genes rh4 (Jiang et al., 2010), rh1, eba165, clag2 (Flueck et al., 2009), clag3.1, clag 3.2 and eba140 (Crowley et al., 2011) is associated with the heterochromatic mark H3K9me3 and clag2, clag3.1, clag3.2, rh1 and rh4 are all located in the same heterochromatic islands as the var genes (Flueck et al., 2009). However, none of these heterochromatic invasion genes exhibit the same activity-dependent enrichment of variant histones as var genes; instead double-variant nucleosomes are maintained at the 5′-IGR of these invasion genes independent of transcriptional activity throughout the asexual replication cycle (Fig. 6).
Another unique feature of var genes is that CpG islands, although rare in P. falciparum, are concentrated in var gene coding sequences. CpG DNA methylation in P. falciparum remains unproven but P. falciparum possesses the enzyme required for CpG methylation (Templeton et al., 2004). H2A.Z antagonizes DNA methylation in plants and humans (Zilberman et al., 2008; Yang et al., 2012). Possibly the unique fluctuations of Pf H2A.Z/Pf H2B.Z at the var TSS relate to regulation of var genes by DNA methylation.
Unlike the 5′ promoter, the var introns are consistently occupied by double-variant nucleosomes. The var intron has bidirectional promoter activity in both poised and inactive var genes in late stage parasites (Epp et al., 2009). The var intron is critical for var gene silencing (Deitsch et al., 2001; Gannoun-Zaki et al., 2005; Frank et al., 2006) and var genes are retained at the nuclear periphery via their introns in a process involving F-actin (Zhang et al., 2011). H2A.Z plays a structural role in retaining recently repressed, but poised, yeast genes at the nuclear periphery in association with nuclear pores (Brickner et al., 2007). Loss of H2A.Z is also associated with effective silencing of these yeast genes, although unlike var genes they are then removed from the nuclear periphery. H2A.Z is also implicated in structural organization of the human genome as it is enriched at the same sites as the insulator binding protein CTCF which orchestrates inter- and intrachromosomal interactions (Fu et al., 2008). The enrichment of Pf H2A.Z/Pf H2B.Z nucleosomes at var gene introns hints at a similar role in organizing these loci.
Pf H2B.Z is a H2B variant that is associated with Pf H2A.Z in nucleosomes that are enriched at gene promoters. Pf H2B.Z enrichment only correlates with promoter activity for the heterochromatic var multigene family. Interestingly, other heterochromatic gene families with similar patterns of variegated expression are not depleted of Pf H2A.Z/Pf H2B.Z when silent. Thus, it is possible that a var-specific machinery exists which is responsible for deposition of variant histones exclusively at the var promoter. The unique acetylations of the variant histones and the involvement of PfSir2A in their dynamic regulation at var promoters suggest their acetylations are critical for their function and this is a promising avenue for future investigations. Further identification of the mechanisms leading to Pf H2A.Z/Pf H2B.Z deposition will be critical for our understanding of var gene transcription and expression switching.
Parasite culture and transfection
Plasmodium falciparum parasites of the 3D7 strain were cultured following standard protocols (Trager and Jensen, 1976) with 0.5% Albumax (Invitrogen) and 0.2% sodium bicarbonate in RPMI-HEPES. Parasites were synchronized with 5% sorbitol on a regular basis. Transfection of ring stage parasites using 100 μg of the pH2BZHA plasmid was performed as described previously (Crabb et al., 2004). Parasites carrying the plasmid were initially selected on 2.5 nM WR99210 (Walter Reed Army Institute of Research) but H2B.ZHA expression could not be detected at this level of WR99210 so the drug concentration was raised to 25 nM to increase recombinant protein expression in the stably transfected cultures.
Selection for var2CSA and var20 expression
Parasites were incubated in a Petri dish coated with 50 μg ml−1 bovine trachea CSA (Sigma) to select for var2csa expression, or 8 μg ml−1 recombinant human ICAM-1 (Bender MedSystems) to select for var20 expression, as described previously (Noviyanti et al., 2001).
The Pf H2B.Z locus (PF07_0054) comprising the promoter region and entire ORF was amplified by PCR from −880 bp upstream of the start codon up until before the stop codon using the following primers: H2B.ZFor: 5′-TAcccgggTATAATAATTATGCGCATATATAT-3′ and H2B.ZRev: 5′-CAtgtacaCAGCACTTGTTGTATATTTG-3′. The 1263 bp PCR product was cloned in frame with a C-terminal triple HA tag into the pHA vector using the XmaI and AauI restriction sites. Correct cloning was verified by sequencing analysis.
Primary antibodies employed in immunoprecipitation assays, Western blot (WB) and immunofluorescence analysis (IFA) in this study were mouse anti-HA (mAB12CA5, Walter and Eliza Hall Institute monoclonal antibody facility), mouse IgG2b (Abcam Ab 18421), rabbit anti-Pf H2A.Z (Petter et al., 2011), rabbit anti-H3 (Abcam Ab1791), rabbit IgG (Abcam Ab 46540) and pre-immune rabbit serum, rabbit anti-H2A (Abcam Ab 88770), rabbit anti-H2B (Abcam Ab1790), rabbit anti-H3K4me3 (Millipore 04-745), rabbit anti-H3K9ac (Millipore 06-942), rabbit anti-H4K12ac (Millipore 07-595), rabbit-anti H4K20me3 (Abcam Ab 9053), rabbit anti-HP1 (Petter et al., 2011) and rabbit anti-H3K9me3 (Active Motive 39161). Secondary reagents used for WB were goat anti-mouse HRP (Invitrogen), goat anti-rabbit HRP (Invitrogen) and protein G HRP (Abcam). Secondary antibodies for IFA were Alexa-Fluor coupled anti-mouse and anti-rabbit antibodies (Invitrogen).
Parasite cultures were fixed with paraformaldehyde/glutaraldehyde and permeabilized and stained essentially according to published methods (Tonkin et al., 2004). The cells were allowed to settle for 30 min as a monolayer on Menzel Superfrost Plus slides, washed with PBS, mounted in Prolong Antifade (Invitrogen) and left to cure overnight. Imaging was performed using an Olympus FV1000 confocal microscope (100 × oil objective).
Mononucleosome preparation and co-immunoprecipitation
Mononucleosomes were prepared from freshly isolated nuclei by micrococcal nuclease (MNase) digestion with 20KU MNase (NEB) and subsequent extraction with salt-free buffers, essentially as described (Flueck et al., 2009), followed by an additional extraction using RIPA buffer. The chromatin samples contained 95–97% mononucleosomes as calculated by the ImageJ software (http://rsb.info.nih.gov/ij/), as described previously (Lowell et al., 2005). Mononucleosome fractions were incubated with antibodies overnight and then precipitated with 30 μl of protein G agarose beads (Millipore) (Petter et al., 2011). The immune complexes were either eluted with 2 × SDS-PAGE loading buffer and analysed by SDS-PAGE and Western blot, or with 1% SDS in TE buffer and processed for microarray analysis.
Microarray and data analysis
Genome-wide profiling of mononucleosomal ChIP DNA was performed on the oligonucleotide microarray platform containing 10 416 70-mer ORF probes representing 5343 coding genes from the P. falciparum strain 3D7 as previously described (Hu et al., 2007). Using the OligoRankPick algorithm (Hu et al., 2007), intergenic probes were designed up to 1.5 kb upstream of start codons with each upstream region being represented by one specific probe. Mononucleosomal DNA was extracted with the MinElute kit (Qiagen) and subjected to random amplification (Chaal et al., 2010). Cy5-labelled ChIP DNA samples were hybridized at 63.5°C against a Cy3-labelled pool of input DNA in equal amounts from H2B.ZHA ring, trophozoite and schizont ChIP experiments. Labelling and hybridization were carried out essentially as described (Bozdech et al., 2003). Microarrays were scanned and analysed using GenePix Pro 6.0 (Axon Instruments). Automatic spot detection was visually controlled and poor-quality probes were eliminated from the analysis. The data were normalized using the Lowess normalization method implemented in the Acuity 4.0 software (Molecular Devices) and filtered to only contain spots in which 99% of pixels had a signal intensity of greater than 2 standard deviations above background for both Cy3 and Cy5.
Pf H2B.Z enrichment was analysed as log2 ratios (Pf H2B.Z/input) and probes ranked according to decreasing log2 ratios in Excel. Probes exhibiting the top 10% (high), mid 20% (medium) and bottom 10% (low) of Pf H2B.Z signal were plotted in a histogram according to their distance from the ATG. Bins were defined in 500 bp intervals. As the number of probes represented on the microarray in each 500 bp bin differs significantly, the data were corrected for the number of probes that were present in each bin on each particular array. The data were thus expressed as the proportion of all probes present in each 500 bp bin (in %).
Cross-linked chromatin immunoprecipitation (ChIP)
ChIP experiments were performed as described (Petter et al., 2011). Briefly, chromatin was isolated at two time points during the intraerythrocytic developmental cycle (IDC), at early ring (6–14 hpi) and late schizont (36–44 hpi) stages. Parasite cultures were cross-linked with 1% formaldehyde for 10 min at 37°C and subsequently quenched with 125 mM glycine on ice for 5 min. Following saponin lysis, the cross-linked parasites were washed twice in cold PBS and then resuspended in cold lysis buffer (10 mM HEPES pH 7.9, 10 mM KCl, 0.1 mM EDTA pH 8.0, 0.1 mM EGTA pH 8.0, 1 mM DTT, 1 × EDTA free protease inhibitors) and incubated on ice for 30 min. NP-40 was added to a final concentration of 0.25% before lysing the parasites with a dounce homogenizer (Pestle B). The nuclei were pelleted at 14 000 r.p.m. for 10 min at 4°C and the chromatin was extracted by resuspending the nuclei in 1% SDS lysis buffer (1% SDS, 10 mM EDTA, 50 mM Tris pH 8.1, 1 × EDTA free protease inhibitor). Chromatin was sheared into 200–1000 bp fragments by sonication in a Bioruptor UCD-200 (Diagenode) for 2 × 8 min on high at 30 s intervals. The chromatin was diluted 1:10 in ChIP dilution buffer (0.01% SDS, 1.1% Triton X-100, 1.2 mM EDTA, 16.7 mM Tris-HCl pH 8.1, 150 mM NaCl) and used for subsequent immunoprecipitation. Immunoprecipitation (IP) was performed with the EZ ChIP Kit (Millipore) as described by the manufacturer. For each IP, approximately 1 × 109 ring or 1 × 108 schizont stage parasites were used. Cross-linking of the immune complexes was reversed overnight at 45°C after addition of 500 mM NaCl, followed by treatment with 50 μg of proteinase K for 2 h at 45°C. DNA was extracted using the MinElute® PCR purification kit (Qiagen). The DNA was further purified with ethanol precipitation prior to analysis. All ChIP experiments were performed at least twice.
Quantitative real time PCR and statistical analysis
ChIP DNA was amplified (Applied Biosystems 7900HT) with specific primers for candidate genes (Petter et al., 2011) (Table S1) in 10 μl of PCR reactions containing 5 μl of SYBR® green PCR master mix (Applied Biosystems), 2 μl of diluted DNA template and 0.30 μM or 0.90 μM of each primer. PCR was performed in duplicates and the melting curves were analysed after each run to confirm the specificity of the amplification. The mean Ct (cycle threshold) values were normalized with input signals (ΔCt) taking into account the dilution factor and then corrected for signals obtained with non-immune background controls (ΔΔCt). The assay site fold enrichment above background was calculated as 1 + E−ΔΔCt (where E = primer efficiency calculated based on gDNA titration curves). To correct for differences in nucleosomal occupancy at different sites, the enrichment ratio in relation to H3 was determined. Statistical analysis was performed using GraphPad Prism software (Version 5). The non-parametric Mann–Whitney test was applied to calculate the significance of Pf H2A.Z and Pf H2B.Z enrichment in the intergenic region (IGR) and open reading frame (ORF) of each euchromatic gene as well as in the IGR, ORF and intron of each var gene. Spearman's correlation between Pf H2B.Z and Pf H2A.Z at ring and schizont stages was also determined.
RNA was harvested in parallel with the chromatin preparation and extracted from pelleted infected erythrocytes with 10 (rings) or 20 (schizonts) pellet volumes of TRIzol (Invitrogen). RNA was purified as described previously (Kyes et al., 2000). cDNA was synthesized using Superscript III Reverse Transcriptase (Invitrogen) and quantified by reverse transcription PCR (Q-RT-PCR) as described previously (Duffy et al., 2009) using primers targeting the open reading frames of genes (Table S1). The level of each cDNA sequence was determined relative to its level in a constant quantity of 3D7 strain gDNA and the amounts of cDNA and gDNA were normalized using either fructose bisphosphate aldolase (invasion genes) or the ring stage expressed gene SBP1 (var genes) by 2−ΔΔCt analysis. Paired t-test was used to calculate whether expression levels of invasion genes were significantly different to the level of expression of the repressed hypothetical gene PF11_0480, hyp, in ring and schizont stage parasites.
We thank Fransisca Sumardy and Cameron Nowell for excellent technical assistance. We also thank the donors and the Red Cross Blood Service (Melbourne Australia) for providing erythrocytes and serum.
This work was supported by the Australian Research Council (Grant Number DP110100483 to M. F. D.), and by the University of Melbourne (Early Career Research Grant to M. P.).