The MYST Family Histone Acetyltransferase Regulates Gene Expression and Cell Cycle in Malaria Parasite Plasmodium falciparum

Summary

Histone lysine acetylation, normally associated with euchromatin and active genes, is regulated by different families of histone acetyltransferases (HATs). A single Plasmodium falciparum MYST (PfMYST) HAT was expressed as a long and a short version in intraerythrocytic stages. Whereas the recombinant PfMYST expressed in prokaryotes and insect cells did not show HAT activity, recombinant PfMYST purified from the parasites exhibited a predilection to acetylate histone H4 in vitro at K5, K8, K12, and K16. Tagging PfMYST with the green fluorescent protein at the C-terminus showed that PfMYST protein was localized in both the nucleus and cytoplasm. Consistent with the importance of H4 acetylation in var gene expression, PfMYST was recruited to the active var promoter. Attempts to disrupt PfMYST were not successful, suggesting that PfMYST is essential for asexual intraerythrocytic growth. However, overexpression of the long, active or a truncated, non-active version of PfMYST by stable integration of the expression cassette in the parasite genome resulted in changes of H4 acetylation and cell cycle progression. Furthermore, parasites with PfMYST over-expression showed changes in sensitivity to DNA damaging agents. Collectively, this study showed that PfMYST plays important roles in cellular processes such as gene activation, cell cycle control, and DNA repair.

Introduction

In the past few decades, Plasmodium falciparum, the causative agent of the most virulent form of malaria, has become increasingly resistant to most commonly used antimalarial drugs. The spread of drug resistant parasite strains is at least partially responsible for the global resurgence of malaria. For better treatment of malaria, artemisinin-based combination therapies are being adopted in most falciparum malaria-endemic countries. Meanwhile, great efforts have been undertaken in drug discovery and development (Gelb, 2007). The genome sequences of several malaria parasite species have revealed a wealth of new molecular targets, against which novel therapeutics may be developed (Winzeler, 2008). Among them, enzymes involved in post-translational modifications (PTMs) of histones are promising candidate targets.

Histones as the major components of the chromatin are subject to a myriad of PTMs, which affect many chromatin-based events such as transcription, DNA replication and repair (Kouzarides, 2007). Most of these PTMs are reversible and dynamic, catalyzed by a large number of enzymes. Among the covalent modifications of histones, lysine acetylation is probably the best understood. Histone acetylation occurs at multiple lysines and is regulated by different families of histone acetyltransferases (HATs) and histone deacetylases (HDACs). Because of their roles in cancer and other genetic disorders, both HATs and HDACs have been the subjects of intensive research, and inhibitors for these enzymes have great potentials for chemotherapy (Yang, 2004, Cole, 2008).

Based on structural characteristics, HATs are divided into at least five families; GNATs (Gcn5 N-acetyltransferases), MYSTs, p300/CBP (CREB-binding protein), general transcription factor HATs, and nuclear hormone-related HATs (Lee & Workman, 2007, Carrozza et al., 2003). MYST is named for its founding members MOZ, Ybf1/Sas3, Sas2, and Tip60, which are involved in a wide range of cell functions (Utley & Cote, 2003, Thomas & Voss, 2007). MOZ (monocytic leukemia zinc finger protein), when translocated and fused to CBP, causes acute myeloid leukemia (Borrow et al., 1996). Sas2 and Sas3 (something about silencing) proteins in Saccharomyces cerevisiae are involved in transcriptional silencing (Reifsnyder et al., 1996). Tip60 (Tat interacting protein of 60 kDa) is a HAT that interacts with the HIV Tat protein (Kamine et al., 1996). Other well-characterized MYST members include the yeast ESA1 (essential Sas2-related acetyltransferase 1), human HBO1 (HAT bound to replication origin recognition complex 1) and MORF (MOZ-related factor), mouse Querkopf, and Drosophila and human MOF (male absent on the first). MYST is the largest HAT family with members identified in all eukaryotes (Yang, 2004). While the importance of MYST proteins in transcription regulation is well established, recent studies have implicated their roles in DNA damage repair and cell cycle progression (Thomas & Voss, 2007, Lafon et al., 2007). For instance, yeast Esa1 preferentially acetylates H4K5, K8, K12, and K16, and ESA1 mutants lacking HAT activity could complete DNA replication but failed to proceed normally through mitosis and cytokinesis (Clarke et al., 1999). Similarly, deletion of amino-terminal domain of histone H4, or substitution of the ESA1-targeted lysines (H4K5, K8, K12, K16) caused a marked delay during the G₂/M phases of the cell cycle (Megee et al., 1995). Furthermore, ESA1-mediated acetylation of histone H4 is required for the repair of DNA double-strand breaks (DSBs), and ESA1 and H4 mutants are hypersensitive to DNA-damaging agents (Bird et al., 2002). Likewise, expression of a Tip60 mutant lacking HAT activity (dominant negative) also resulted in cells with defects in DSB repair (Ikura et al., 2000). Compared with mammals, single-cell protozoan parasites appear to have fewer MYST proteins and their functions are poorly understood. In the trypanosomatid parasite Trypanosoma brucei, three MYST proteins have been found, and they play distinct roles in mediating histone H4 acetylation (Kawahara et al., 2008, Siegel et al., 2008). In comparison, the apicomplexan parasite Toxoplasma gondii has two MSYT family proteins (Smith et al., 2005), and TgMYST-B appears to be important in facilitating ataxia telangiectasia mutated (ATM) kinase-mediated DNA damage response (Vonlaufen et al., 2010).

In P. falciparum, the genome encodes at least four HATs and four HDACs, and a number of lysine acetylation sites have been identified on histone tails (Miao et al., 2006, Trelle et al., 2009). So far, only a few members of these enzyme families have been characterized. The GNAT family HAT PfGCN5, which preferentially acetylates histone H3 (Fan et al., 2004), plays an important role in genome-wide gene activation in malaria parasites, and disturbance of its activity inhibits parasite growth (Cui et al., 2007b, Cui et al., 2008b, Cui et al., 2007a). The P. falciparum Sir2 (silencing information regulator 2) protein has intrinsic histone deacetylase and ADP-ribosylation activities, and is involved in the epigenetic control of a large family of telomeric virulence genes (Duraisingh et al., 2005, Freitas-Junior et al., 2005, Merrick & Duraisingh, 2007, Tonkin et al., 2009). The demonstrated parasiticidal effects of HAT and HDAC inhibitors on malaria parasites attest to the potential of these histone acetylation enzymes as antimalarial drug targets (Darkin-Rattray, 1996, Andrews et al., 2000, Andrews et al., 2008, Cui et al., 2007a, Cui et al., 2008b). In this study, we report a functional study of the single MYST homologue PfMYST from P. falciparum. We determined its dynamic expression, enzymatic activity and subcellular localization, and studied its function through genetic disruption and overexpression.

Results

P. falciparum has a single MYST member

BLASTP analysis of the P. falciparum genome with three S. ceravisiae MYST members, Esa1, Sas2 and Sas3, identified only one gene, PF11_0192, which encodes a putative MYST family HAT, PfMYST. The homology is restricted to the MYST catalytic domain. PfMYST is predicted to encode a protein of 608 amino acids (aa) with the MYST domain located at the C-terminus (Fig. S1A, C). There is a low-complexity region of ~260 aa in the N-terminus, which does not have apparent homology to any proteins or domains in the databases, and is devoid of any known protein motifs or signal sequence. PfMYST lacks a recognizable nuclear localization signal. Phylogenetic comparison using only the MYST domain showed that PfMYST is most closely related to other apicomplexan MYST members (Fig. S1B). Other recognized features of PfMYST include the C2HC-type zinc finger (Takechi & Nakayama, 1999, Akhtar & Becker, 2001), a similar “ER motif” found in ESA1 and HDAC (Adachi et al., 2002) and a chromodomain (CD) which binds single-stranded nucleic acids and methylated histone peptides (Akhtar et al., 2000, Lachner et al., 2001, Min et al., 2003) (Fig. S1C, D). P. falciparum genome has at least three additional proteins with CDs (Fig. S1E) and they display limited sequence homology between each other. Recently, the CD in the P. falciparum heterochromatin protein 1 (HP1) has been found to bind to trimethylated H3K9 (Perez-Toledo et al., 2009, Flueck et al., 2009).

PfMYST is expressed in erythrocytic stages as a long and a short version

Real-time reverse transcriptase-polymerase chain reaction (RT-PCR) analysis using RNA from different developmental stages confirmed PfMYST expression in both asexual erythrocytic cycle and gametocytes, similar to the results from the microarray study (Le Roch et al., 2003). During the asexual erythrocytic cycle, PfMYST mRNA level was higher in rings and schizonts, but lower in trophozoites (Fig. 1A). To determine the size of the PfMYST mRNA, Northern blot was performed with total RNA from mixed asexual stages, which detected two transcripts of ~2.3 and ~1.6 Kb (Fig. 1B). To map transcription start sites (TSSs) of the two PfMYST transcripts, RNA-ligase mediated rapid amplification of cDNA ends (RLM-RACE) was performed and two PCR products of approximately 1.0 and 0.2 Kb were generated (Fig. 1C). Sequencing of 13 clones from the larger fragment identified a major TSS at 177 bp upstream of the first putative ATG codon, whereas sequencing of 17 clones from the 0.2 Kb fragment identified a major TSS from 647 bp downstream of the first ATG codon (Fig. 1D).3′ RACE revealed polyadenylation sites at 287 and 309 bp downstream of the stop codon (Fig. 1D). Together, the RACE results predicted the sizes of the two PfMSYT mRNAs as ~2.3 and 1.5 Kb, which are consistent with the result from Northern analysis. Based on the predicted ATG codons shown in Fig. 1D, the long and short versions of PfMYST encode proteins of 608 and 369 aa, respectively.

Fig. 1. Transcription of PfMYST gene during parasite intraerythrocytic development.

(A) Relative level of PfMYST in parasite by real-time RT-PCR. Constitutively expressed Seryl-tRNA synthetase gene was included as an internal control. Values denote the fold increase in transcription relative to that in the ring stage. R –rings, ET – early trophozoites, LT – late trophozoites, S – schizonts, G – gametocytes. (B) Northern blot analysis of PfMYST expression using total RNA from mixed intraerythrocytic stages. (C) RLM-RACE products showing that the two PfMYST transcripts were initiated at two major positions. (D) Nucleotide sequence at the 5′ and 3′ end of PfMYST. The putative translational start codons (ATGs) of the longer and shorter transcripts are marked as boldface letters. TSSs and polyadenylation sites are indicated by boldface letters and marked with small arrows. Numbering is with respect to the first ATG and stop codon TAA, respectively. Numbers in parentheses indicate the numbers of clones sequenced. A possible polyadenylation signal (AATAAA) is underlined.

To study PfMYST protein expression, PfMSYT HAT domain F4 (aa 385–608) was expressed in E. coli and the His-tagged recombinant protein was purified (Fig. 2A). Polyclonal rabbit antibodies raised against the recombinant PfMYST F4 protein specifically recognized two bands of ~70 and ~50 kDa from parasite lysates, while the pre-immune serum did not react with these bands (Fig. 2B). The upper band was consistent with the predicted molecular weight of 71 kDa for the long version of PfMYST, while the lower band was slightly larger than the predicted molecular weight of 45 kDa for the short version. To confirm that the two protein bands detected by the antibodies were indeed PfMYST proteins, we fused the C-terminal end of the endogenous PfMYST with a tandem affinity purification (TAP) tag. Correct integration of the transfected plasmid at the PfMYST locus was verified by integration-specific PCR and Southern blotting (Fig. S2A, B). Western blot of parasite proteins from a TAP-tagged line and 3D7 control with horse radish peroxidase (HRP)-conjugated goat anti-rabbit IgG antibodies that bind to the TAP tag detected two protein bands of ~90 and ~70 kDa (Fig. 2C). The upward shift in mobility is consistent with the fusion of the 20.5 kDa TAP tag to the two PfMYST bands, confirming that two bands detected were PfMYST proteins. This strongly suggests that the two PfMYST proteins were derived from the two PfMYST transcripts, respectively, although we cannot rule out proteolysis of the long version. A time-course analysis of PfMYST expression detected both the long and short versions of PfMYST throughout the intraerythrocytic developmental cycle (IDC) (Fig. 2D). Similar to the pattern of the PfMYST mRNA level, PfMYST protein level was lower in late trophozoites and peaked in schizonts.

Fig. 2. Expression of PfMYST during parasite development.

(A) Purification of 6X His-tagged PfMYST HAT domain (F4) from E. coli. Coomassie-stained SDS-PAGE gel shows the purified PfMYST protein as a 35 kDa band. (B) Western blots of protein extracts from asynchronous 3D7 parasite probed with rabbit preimmune and anti-PfMYST sera. The anti-PfMYST antibodies detected two bands of ~50 and ~70 kDa. (C) Western blot with HRP-conjugated goat anti-rabbit IgG antibodies that bind to the TAP tag. Two bands of 70 and 90 kDa were detected in the TAP-tagged PfMYST clones. (D) Western blot to detect PfMYST protein levels at different stages of the IDC. Equal amounts of proteins (30 μg) from synchronized parasites at ring (R), early trophozoite (ET), late trophozoite (LT) and schizont (S) stages were separated by 12% SDS-PAGE and probed with anti-PfMYST antibodies (upper panel), while protein loading was monitored with the anti-HSP70 antibody (lower panel).

Green fluorescent protein (GFP)-tagged PfMYST was generated in order to determine the localization of PfMYST in the parasite. For this purpose, 3D7 parasite was transfected with pHD/F2-GFPint, clones were selected, and integration of the plasmid at the PfMYST locus was verified using integration-specific PCR and Southern blotting (data not shown). GFP signal was detected throughout the IDC, and distributed in the whole parasite with increased intensity in the nuclear area (Fig. 3). Indirect immunofluorescence assay (IFA) with anti-PfMYST antibodies also detected PfMYST localization in the nucleus and cytoplasm (data not shown), confirming that PfMYST is both nuclear and cytoplasmic.

Fig. 3. Localization of PfMYST.

Representative GFP fluorescent images of parasites with GFP fused at the C terminus of the endogenous PfMYST showing the localization of PfMYST-GFP at ring, trophozoite and schizont stages. Nuclei were counterstained with Hoechst 33342. Triple merge indicates the merging of the light, GFP and Hoechst images of the same cells.

PfMYST has HAT activity

To test whether PfMYST has HAT activity, recombinant PfMYST proteins tagged with 6 × His, glutathione S-transferase (GST), maltose-binding protein (MBP) were expressed in E. coli and purified under native conditions (data not shown). These purified proteins included both the long (F1) and truncated versions of PfMYST (F2: amino acids 230–608; F3: 322–608; and F4). In an in vitro HAT assay, these recombinant PfMYST proteins showed no detectable HAT activity towards bovine core histones or the recombinant P. falciparum core histones (data not shown). The F1 and F2 PfMYST fragments were also expressed in insect cells and in a cell-free wheat germ expression system, but neither sources of the recombinant PfMYST displayed detectable HAT activity (data not shown). It was surprising that PfMYST expressed in eukaryotic systems lacked HAT activity. Therefore, we wanted to determine whether PfMYST expressed in the parasite has HAT activity. The TAP-tagged PfMYST F2 fragment (C-terminal 378 residues) was episomally expressed in P. falciparum, and purified using IgG Sepharose beads as predominantly one band of ~70 kDa (Fig. 4A). Compared with the recombinant PfGCN5 protein that preferentially acetylates histone H3, the PfMYST F2 fragment preferentially acetylated H4 in a mixture of P. falciparum recombinant core histones (Fig. 4B). When nucleosomal histones from the HeLa cells were used, the purified PfMYST F2 could effectively acetylate both H4 and H2A (Fig. 4B). Furthermore, we have also purified TAP-tagged endogenous PfMYST (Fig. 2C) and used it in vitro HAT assay, which produced a similar result as with the episomally expressed F2 fragment (data not shown). Since certain CD-containing MYST proteins such as MOF exclusively acetylate H4K16, while others target H4K5, K8, K12, and K16 (Utley & Cote, 2003), we wanted to determine the target lysine residues in H4 for PfMYST. HAT assays were performed with PfH4 and acetylated lysines were detected using antibodies specific for acetylated H4K5, K8, K12, or K16. Western blots on histones isolated from control and sodium butyrate-treated HeLa cells to inhibit HDAC activity confirmed the specificities of these commercial antibodies for acetylated histones (Fig. 4C). Western blots with recombinant PfH4 showed that the recombinant PfMYST F2 fragment effectively acetylated K5, K8, and K12, and to a lesser extent, K16 (Fig. 4C). To evaluate whether multiple lysines in could be acetylated in the same H4 molecule, we performed in vitro HAT assay using the N-terminal 24-residue H4 peptide. Mass spectrometry analysis showed that the H4 peptide had either mono- or di-acetylation (with an increase of a molecular mass of 42 and 84, respectively), suggesting that two separate lysine residues could both be acetylated on a single H4 peptide by recombinant PfMYST F2 in vitro (Fig. 4D).

Fig. 4. In vitro acetylation of histones by recombinant PfMYST.

(A) Coomassie blue-stained SDS-PAGE gel showing the purification of episomally expressed, TAP-tagged PfMYST F2 fragment (indicated by an asterisk) from transfected parasites (F2-TAP). Wild-type 3D7 parasite serves as a control. (B) In vitro HAT assay showing recombinant PfMYST F2-mediated H4 acetylation of mixed core and nucleosomal histones. Upper panel: Coomassie blue-stained gel; middle panel: fluorography with mixed recombinant P. falciparum core histones as substrates; lower panel: fluorography with histones from HeLa cell mononucleosomes as substrates. (C) Detection of substrate lysine specificity of PfMYST. Left panel represents verification of specificities of the commercial antibodies. Histones isolated from control HeLa cells (Control) and sodium butyrate-treated HeLa cells (Treated) were analyzed by Western blots with antibodies for acetylated lysines H4K5, H4K8, H4K12, and H4K16, respectively. In the right panel, HAT assay was performed with the PfH4 and PfMYST F2-TAP, and acetylated lysines were detected by Western blots with antibodies for acetylated H4K5, K8, K12 and K16. (D) Acetylation of the 24-residue N-terminal H4 peptide by PfMYST F2-TAP was analyzed by mass spectrometry to show mono- (H4-Ac1) and di-acetylated (H4-Ac2) peptides.

Besides the MYST domain, PfMYST contains several additional domains (Fig. S1). To evaluate the importance of different domains for HAT activity, a series of deletions were made for episomal expression of the TAP-tagged PfMYST fragments under the hsp86 promoter (Fig. S2C). Recombinant PfMYST-TAP proteins were purified from parasite whole cell extracts and equimolar amounts of purified PfMYST proteins were used to measure HAT activity using the liquid HAT assay. As shown in Fig. 5, deletion of the N-terminal 229 aa low-complexity domain in F3, which approximates the native, short form of PfMYST resulted in over 4-fold increase in HAT activity. Deletion of the CD (F3) only led to a slight decrease in HAT activity as compared to F2, whereas deletion of the zinc finger (F4) resulted in a complete loss of HAT activity. Unlike the yeast ESA1 protein which requires the intact C-terminal end for HAT activity, the C-terminal 56 aa of PfMYST (F1-C2 and F2-C2) was dispensable for HAT activity. Further truncation of the C-terminus to aa 518 (F1-C3 and F2-C3) abolished HAT activity (Fig. 5).

Fig. 5. Deletion analysis to define the active domain of PfMYST.

Full-length (F1) PfMYST and seven deletion fragments were episomally expressed in 3D7 parasite as TAP-tagged proteins. The recombinant PfMYST-TAP fragments were purified and used to determine HAT activity in a filter HAT assay. HAT activity was expressed as average amount (CPM) ± standard deviation of ³H incorporated into histones after background correction with wild-type 3D7. The HAT activity of the F1 fragment of PfMYST was set as 100%.

PfMYST does not complement the yeast ESA1 mutants and is refractory to genetic deletion

Among the MYST family members in yeast, PfMYST is most similar to ESA1. We wanted to determine whether the yeast Esa1 mutants could be rescued with three PfMYST fragments. Three temperature-sensitive Esa1 mutants were successfully transformed with the PG1(HA)2 plasmids at 28°C, and approximately similar amounts of PfMYST expression in these transformed mutants were detected by Western blots (data not shown). However, when these cells were grown at 37°C, no obvious differences were observed between the cells transformed with the control and PfMYST constructs (Fig. S3). Furthermore, the PfMYST fragments did not complement the slow growth phenotype at the permissive temperature of 28°C either. This indicated that despite a high degree of sequence conservation within the HAT domain, PfMYST could not fully complement the yeast ESA1 gene.

To investigate the function of PfMYST in P. falciparum, we first sought to determine whether it is essential for intraerythrocytic growth of the parasite. For genetic deletion of PfMYST, 3D7 parasite was transfected with pCC-1/PfMYST (Fig. S4), and resistant parasites were obtained 3–4 weeks later under the positive selection drug WR99210. When the negative selection drug (5-fluorocytosine) was applied, no viable parasites could be seen within 4 weeks. Parasites that appeared after a longer period of culture in the presence of the two drugs were cloned and analyzed for the integration events (Miao et al., 2010). Southern blot showed no sign of genetic deletion of PfMYST. We only identified parasites with the plasmid integrated at the 3′ arm of the homologous region of PfMYST (Fig. S4), which did not compromise PfMYST expression. Repeated attempts of transfection with this plasmid produced similar results. Since PfMYST locus is accessible for 3′ integration with TAP or GFP, these results suggest that PfMYST is essential for intraerythrocytic growth.

PfMYST participates in var gene activation

It has been shown that histone modifications are strongly linked to the maintenance of mono-allelic var gene expression pattern during parasite IDC. Specifically, H4 acetylation is enriched at active var gene promoter, suggesting that PfMYST might be involved in its activation (Freitas-Junior et al., 2005, Lopez-Rubio et al., 2007). To test this hypothesis, we used var2csa as the model. In 3D7 parasite, var2csa expression was not detected by real-time RT-PCR, whereas panning with chondroitin sulfate A (CSA) selected parasites with high-level var2csa expression (data not shown). To detect relative enrichment of PfMYST, chromatin immunoprecipitation (ChIP) with anti-PfMYST antibodies and real-time PCR analysis were performed. ChIP with anti-H4 antibodies was used as a control to normalize nucleosomal occupancy, which has been shown to change greatly during var2csa activation (Westenberger et al., 2009). The results showed that PfMYST was enriched around the TSS of var2csa in ring stage when this gene was active (Fig. 6B). In comparison, PfMYST was only slightly enriched at the var2csa TSS at mature stage when this gene was poised for transcription (Fig. 6B). In both cases, PfMYST enrichment at the var2csa TSS was specific and was not detected using the var2csa exon primers. When the relative PfMYST enrichment levels in ring and mature stages were compared between CSA-selected and unselected parasites, PfMYST enrichment levels at the var2csa TSS in ring stage were much higher than those at the mature stages (Fig. 6C). These results suggest that PfMYST is recruited to var gene promoter to activate its transcription in ring stage.

Fig. 6. Enrichment of PfMYST at the var2csa locus.

(A) A simplified scheme of var2csa gene organization showing the TSS and PCR fragment positions (a–h). (B) Distribution of PfMYST along the var2csa gene in CSA selected (var2csa^ON and var2csa^POISED) and unselected parasites (var2csa^OFF) in at ring and mature stages. The relative enrichment of PfMYST at the selected promoter regions (a – h) were determined by ChIP and real-time PCR analysis shown as 2^−ΔΔCt value. The results were normalized against ChIP with anti-H4 antibodies for changes in nucleosomal occupancy. Columns labeled with different letters indicate significant difference (P<0.01, Student’s t-test). (C) Ratios of PfMYST enrichment levels along the var2csa gene in CSA-selected versus unselected parasites at ring (filled bars) and mature stages (open bars). Ratios were calculated using relative enrichment levels from 6B. Results are the average of three independent experiments. Error bars denote standard deviations.

We next mapped PfMYST enrichment levels at the promoters of a constitutively expressed gene (calmodulin) and a schizont stage-specific gene (EBA175). Similarly, PfMYST was found enriched at the TSS of these genes (Fig. S5). Whereas the levels of PfMYST at the calmodulin promoter were not significantly different between ring and schizont stage, its levels at the EBA175 promoter was significantly higher in schizont stage when this gene is active. As expected, CSA selection had no significant impact on the PfMYST levels at the promoters of these two genes. These results suggest that PfMSYT may also participate in the regulation of other genes.

Overexpression of PfMYST disturbs in vivo histone acetylation

To further explore the function of PfMYST, we investigated the effect of PfMYST overexpression. To distinguish endogenous PfMYST expression from the overexpressed versions, the exogenous PfMYST was tagged with GFP. The PfMYST-GFP expression cassette under the hsp86 promoter was integrated at the attB locus of the 3D7^attB strain using the Bxb1 integrase system (Fig. S2D) (Nkrumah et al., 2006). Three parasite lines were obtained, which expressed GFP only, the long version PfMYST-GFP (F1-GFP), or C-terminal truncated PfMYST-GFP (F1C3-GFP). F1C3-GFP lacks HAT activity but has an intact CD, which may be required for effective chromatin targeting. Western blot analysis detected the expression of both endogenous PfMYST and GFP-tagged PfMYST fragments (Fig. 7A). In addition, we found a similar cellular distribution (predominantly nuclear) of the F1-GFP and F1C3-GFP derivatives (data not shown), suggesting that the truncated PfMYST retained the ability to localize to the nucleus. From densitometry analysis of the Western blot, it was estimated that there were 81% and 75% increases in the PfMYST protein level in parasites expressing PfMYST F1-GFP and F1C3-GFP, respectively.

Fig. 7. Overexpression of PfMYST and H4 acetylation.

(A) Western blots of PfMYST expression in 3D7-attB parasite with integration of an expression cassette for GFP only (GFP Control), full-length PfMYST-GFP (F1-GFP) and C-terminal truncated PfMYST-GFP (F1C3-GFP) at the cg6 locus. Parasite lysates were probed with anti-PfMYST antibodies and GFP-tagged PfMYST proteins are marked with asterisks. Approximately equal protein loading is indicated with the antibody against HSP70 (lower panel). (B) Effect of PfMYST overexpression on in vivo histone acetylation. Histones were extracted from synchronized parasites at different stages from each parasite line expressing GFP control, F1-GFP and F1C3-GFP. Equal amounts of histone extracts (5 μg/lane) were indicated by Western blot with antibodies against H4. Lysine acetylation was detected by Western blots with antibodies for H4K5ac, H4K8ac, H4K12ac, H4K18ac, and H3K9ac.

To determine whether overexpression of PfMYST was correlated with changes in histone acetylation in vivo, we analyzed histone H4 acetylation in each parasite line by immunoblotting. We used antibodies specific for the acetylated forms of H3K9, H4K5, K8, K12, and K16. In GFP-control parasites, increased levels of acetylation at these four lysines were evident at late trophozoite and schizont stages, consistent with our earlier finding with the 3D7 parasite (Miao et al., 2006). As expected, overexpression of the F1-GFP resulted in higher levels of H4K5, K8 and K12 acetylation, especially at the late trophozoite stage (Fig. 7B). However, H4K16 acetylation was reduced in late trophozoites. In contrast, overexpression of F1C3-GFP fragment lacking HAT activity led to reduced acetylation at the four H4 lysines during the late trophozoite and schizont stages (Fig. 7B), suggesting that this inactive PfMYST fragment was efficiently recruited to nucleosomes and competed with the endogenous PfMYST for H4 acetylation and/or interfered with the ability of wild-type PfMYST to incorporate into macromolecular complexes (a dominant negative effect). Furthermore, a very strong reduction in H4K12 and H4K16 acetylation at both late trophozoite and schizont stages was observed, when these acetylated lysines were hardly detectable. In comparison, over-expression of the active and inactive versions of PfMYST did not significantly affect the acetylation pattern of H3K9, the preferred target of PfGCN5 (Fig. 7B).

Overexpression of PfMYST is deleterious for cell cycle progression

Since MYST family proteins participate in multiple cellular processes, we wanted to determine whether overexpression of PfMYST causes phenotypic changes in parasite growth. When parasite proliferation rates were measured, parasites expressing the F1-GFP and F1C3-GFP both displayed significant reductions in daily parasitemia compared with the GFP-control parasites (Fig. 8A) (linear mixed-effects models and Tukey’s pairwise comparison, P < 0.05). The difference in cell proliferation was more pronounced after the first cycle. This suggests that altering histone H4 acetylation is deleterious to the proliferation of the parasite. For a more detailed analysis of the proliferation defects resulting from PfMYST overexpression, the duration of IDC was determined in highly synchronous cultures. The GFP-control and parasites expressing the truncated PfMYST completed the IDC at an interval of approximately 47 h, whereas parasites overexpressing the full-length PfMYST (F1-GFP) had a significantly faster IDC of 43 h (ANOVA P < 0.01) (Fig. 8B). To determine whether the variations in IDC led to changes in the number of progeny merozoites, we compared the merozoite number in mature schizonts. The GFP-control, F1-GFP and F1C3-GFP lines produced 18.53±0.55, 13.97±1.19 and 17.97±0.76 merozoites per schizont, respectively (Fig. 8C). The shorter schizont duration in the F1-GFP line resulted in significantly fewer merozoites than the control (χ² goodness of fit test, P < 0.01). Further comparison showed that F1-GFP parasite line had significantly fewer numbers of schizonts containing 17 to 22 merozoites than GFP control or F1C3-GFP parasite line (ANOVA, P<0.01).

Fig. 8. PfMYST overexpression disturbs parasite cell cycle.

(A) Comparison of asexual growth of parasite 3D7-attB, and parasite expressing GFP-control, F1-GFP and F1C3-GFP. Cultures were started at 0.1% parasitemia at ring stage and parasitemia was determined every 48 h for three IDCs. Linear mixed-effects models and Tukey’s pairwise comparison showed that parasitemia between the GFP-control and other PfMYST overexpressing lines were significant at P < 0.05. (B) Comparison of cycle duration of each parasite line: GFP control (46.6 h), F1-GFP (42.8 h), F1C3-GFP (47.3 h). ANOVA analysis of cycle duration showed F1-GFP had significantly shorter cell cycle comparing with other two lines (P < 0.01). (C) Distribution of schizonts with different merozoite numbers for three parasite lines. (D) IDC profile of each parasite line showing the prevalence (percentage) of the ring, trophozoite and schizont stages through a 50 h time period. Note the shortened and prolonged schizont stage for parasite overexpressing F1-GFP and F1C3-GFP, respectively. The star line in F1C3-GFP indicates the presence of schizonts with poorly separated nuclei, which continued to exist in culture at 50–60 h when normal parasites have developed into ring stage.

To determine the stages at which cell cycle progression was affected, the dynamics of different stages of synchronized parasites was followed at a 2 h resolution (Fig. 8D). This experiment showed that the dynamics of ring and trophozoite stages were very similar among these parasite lines, whereas the most noteworthy point of divergence was at the schizont stage. The F1-GFP line had a faster transition from schizonts to rings with the duration of schizont stage at ~22 h as compared to ~25 h in GFP-control parasites. This was also reflected in the peak percentage of the parasites at the schizont stages, which was ~40% in F1-GFP expressing line as compared to ~62% in the GFP-control (Fig. 8D). In sharp contrast, expression of the F1C3 fragment resulted in a more extended schizont stage (Fig. 8D).

Overexpression of PfMYST affects nuclear division and cytokinesis

We next examined the effect of PfMYST overexpression on cell division by light microscopy. Compared with GFP-control line, schizonts in the F1-GFP line appeared to divide normally, although schizogony was faster and produced fewer merozoites (Fig. S6). However, in the F1C3-GFP line, ~5% of the parasites were arrested at the schizont stage, which were easily visible in culture between 50 and 60 h when most of the parasites had become rings. These abnormal parasites showed poorly segregated nuclei, suggesting a defect in nuclear division and cytokinesis (Fig. S6). Electron microscopy (EM) further confirmed that these F1C3-GFP cells had apparent defects in nuclear division and cytokinesis, and at a later time they became degraded without completing cell division (Fig. 9). Observation of the F1C3-GFP line over several generations verified the persistence of this phenotype in each generation.

Fig. 9. EM comparison of schizonts from control and PfMYST-overexpressing parasites.

Representative EM images of 42 h schizonts from GFP-control parasite showing multiple well-separated merozoites, schizonts from F1-GFP parasite with well-separated but fewer merozoites, and abnormal schizonts from F1C3-GFP parasite with less well-separated merozoites (~5% of schizonts). These abnormal schizonts could not complete schizogony and appeared to undergo degradation at later times (at 50 h). Scale bar = 1000 nM.

To test whether the observed changes in cell division from overexpressing PfMYST was the result of defective DNA replication, we measured the DNA content in these parasite lines at different time points. The results showed that DNA replication in all tested lines were similar up to ~30 h (Fig. 10). However, at late trophozoite stage, the number of parasites with higher genome copy numbers (12–24) was much lower in F1-GFP than in GFP-control, whereas it was significantly higher in F1C3-GFP than in GFP-control (Fig. 10). While the difference in DNA content became less evident between GFP-control and F1C3-GFP at schizont stage (42 h), this difference was more obvious between the GFP-control and F1-GFP line (Fig. 10). The peak number of schizonts contained 12–14 genome copies in GFP-control and F1C3-GFP parasite lines, whereas the majority of schizonts in F1-GFP line contained fewer genome copies (6–10) (Fig. 10). This result is consistent with the distribution pattern of merozoite numbers in well-separated schizonts (Fig. 8C).

Fig. 10. The effect of PfMYST overexpression on parasite DNA replication.

DNA content of the parasites was determined using the DNA dye Vybrant Orange and calculated using ring-stage parasite as one genome copy (c). The graphs show the counts of parasites (y axis) with different genome copy numbers (fluorescence intensities) (x-axis). Parasites were collected at trophozoite (30 h), late trophozoite (36 h) and schizont (42 h) stages.

Overexpression of PfMYST affects parasite sensitivity to genotoxic agents

There is increasing evidence demonstrating the involvement of MYST members in the repair of DNA DSBs (Bird et al., 2002, Ikura et al., 2000). To determine whether PfMYST overexpression sensitized parasites to DNA DSBs, we evaluated the sensitivity of these three parasite lines to two DNA-damaging agents, campthothecin (CPT) and methyl methanesulfonate (MMS). Compared with the GFP-control line, the F1-GFP line had significantly elevated EC₅₀ to CPT, whereas F1C3-GFP displayed significantly lower EC₅₀ (Table 1, Fig. S7, Tukey’s pairwise comparison, P<0,001). MMS treatment produced a similar trend, but the difference between the GFP-control and F1-GFP line was not significant (P>0.05, Table 1).

Table 1.

EC₅₀s of three parasite lines to two DNA damaging agents CPT and MMS

Agents	F1C3-GFP	GFP control	F1-GFP
CPT (μg/ml)	1.247±0.087a	3.087±0.116b	20.367±0.902c
MMS (μg/ml)	78.667±5.132A	83.214±4.583A	180.147±10.251B

Same letters in the same row indicate no significant difference between the parasite lines (P > 0.05), whereas different letters indicate significant difference between the parasite lines (P < 0.01 among a, b, and c; P < 0.05 between A and B).

Discussion

Like higher eukaryotes, the early-branching protozoan parasites Plasmodium also have a large array of histone PTMs to make up the histone code (Trelle et al., 2009, Miao et al., 2006). Among these modifications, abundant histone lysine acetylation is mediated by the HAT proteins, of which PfGCN5 has been shown to play an important role in regulating global gene expression in P. falciparum. In this study, we characterized PfMYST and demonstrated its distinction from the PfGCN5 HAT in substrate specificity and its essence in asexual erythrocytic growth of the parasites. More importantly, in addition to participation in transcription control, PfMYST plays important roles in cell cycle regulation and resistance to DSB-inducing agents.

Many eukaryotes have multiple MYST family members, which often exhibit different substrate preferences and carry out distinct functions. The yeast Esa1 protein preferentially acetylates H4 at K5, 8, 12 and 16 (Smith et al., 1998). Sas2 in the SAS complex targets H4 K16 (Shia et al., 2005), whereas Sas3 in the NuA3 complex acetylates histone H3 K14 (John et al., 2000). Unlike the closely-related apicomplexan parasite T. gondii that has two MYST members (Smith et al., 2005), the Plasmodium lineage has retained only one MYST homologue. Like the TgMYST-A, recombinant PfMYST expressed from bacteria and the in vitro translation system lacked detectable HAT activity on core histones. The difference in activity could be due to improper folding of the recombinant protein or that its HAT activity requires association with other proteins in a complex as shown for the Sas2 protein (Sutton et al., 2003). PfMYST purified from the parasite preferentially acetylated H4 in mixed P. falciparum core histones. In addition, it could also acetylate histone H2A in nucleosomes, suggesting that it might have broader histone substrates in vivo. Furthermore, like Esa1 and TgMYST-A, PfMYST acetylated all four lysines in the H4 tail in vitro (Fig. 4). Although we did not directly determine the in vivo histone substrates of PfMYST, overexpression of active or inactive PfMYST in the parasite disturbed the levels of acetylation at all four lysines of H4 (Fig. 7), implying that PfMYST may target all these lysines in vivo. Of the 12 or so proteins in the NuA4 complex, P. falciparum only contains homologues of five of them, suggesting that the malaria parasite may have a divergent PfMYST complex.

The domain organization, sequence homology, and substrate specificity of PfMYST suggest that it may be an orthologue of yeast Esa1 and human Tip60. Both RT-PCR and western blotting detected expression of PfMYST throughout the IDC. Interestingly, like the TgMYST-A, which is expressed as a long and short form (Smith et al., 2005), western blot also detected two PfMYST versions, which are likely the products of the two PfMYST transcripts (Fig. 1). The less abundant short form only lacked the N-terminal low-complexity sequence, while it retained the CD and MSYT domains. Tagging of both versions with GFP showed similar patterns of subcellular localizations, indicating that nuclear targeting signal is not located in the low-complexity domain. The N-terminal domain is not needed for HAT activity; deletion of this domain even enhanced PfMYST in vitro activity on core histones. The function of the N-terminus is not clear, but it may be required to assemble into protein complexes, or because deleting it increases the in vitro HAT activity, it may have a negative regulatory function. In addition, we observed no changes in the ratio of the long and short form during the parasite’s life cycle, suggesting that they are regulated similarly. We have also generated a F2-GFP overexpression clone in 3D7^attB and it showed similar phenotypes to the F1-GFP overexpression clone in terms of subcellular localization and acceleration of the IDC (data not shown), suggesting that oeverexpressing the short PfMYST isoform alone is enough to cause the phenotypic change in cell cycle. There is also a possibility that the two PfMYST isoforms may be involved in different protein complexes and have different cellular functions. With regard to the function of CD, domain deletion analysis showed that it is not required for HAT activity on free histones. This might not be surprising because CDs bind to nucleosomal DNA and histone H3 peptides. Hence, the CD residues may be more important for nucleosomal HAT activity. In fact, studies of the Esa1 CD indicate that mutations in the CD weakly affect histone HAT activity, but very strongly reduce nucleosomal HAT activity (Selleck et al., 2005). In PfMYST, a C2HC zinc finger is found adjacent to the HAT domain and it is essential for the enzyme activity of PfMYST. Since Zn-finger is absent in yeast Esa1 but found in metazoan enzymes, there is an intriguing possibility that Zn-finger has metazoan-specific functions. However, the presence of an essential Zn-finger in PfMYST suggests that this motif might be an ancestral feature in the MYST family proteins.

Esa1, the only essential HAT in S. cerevisiae, is present in two HAT complexes and has multiple roles in vivo (Pillus, 2008). Despite that PfMYST appears orthologous to Esa1, it could not rescue the yeast esa1 mutants. Similarly, our earlier work showed that only the most conserved HAT domain in PfGCN5 was functionally equivalent to the yeast GCN5 HAT domain (Fan et al., 2004). It is not uncommon to find that metazoan orthologues fail to complement yeast mutants, however. We speculate that these results might be due to significant divergence between P. falciparum HATs and yeast HATs. To investigate the functions of PfMYST, multiple attempts have been made to date in order to genetically disrupt this gene using both single and double crossover strategies but without success. Since the PfMYST locus is accessible to tagging with GFP and TAP, the failure to knock out PfMYST most likely means that this gene is essential for asexual erythrocytic growth of the parasite. This is expected for the single MYST member in P. falciparum, given that multiple MYST members in other single-cell parasites such as T. gondii and T. brucei are also essential (Smith et al., 2005, Kawahara et al., 2008).

One prominent function of the MYST family proteins is transcription regulation through histone acetylation (Carrozza et al., 2003). H4 acetylation has been associated with active var gene expression (Freitas-Junior et al., 2005). We have shown that PfMYST is specifically recruited to the active var gene promoter, which probably facilitates transcription initiation. In addition to participation in var gene regulation, PfMYST may also regulate other crucial genes that are involved in parasite development such as invasion of red blood cells (RBCs), cell cycle progression and DNA repair. To determine whether PfMYST is involved in other cellular processes, we tried to overexpress either the active PfMYST or its dominant negative version in the parasites. This has led to the detection of two distinct growth phenotypes, suggesting that like ESA1which is essential for G2/M cell cycle progression in S. cerevisiae (Clarke et al., 1999), PfMYST is also required for the control of cell cycle in P. falciparum. Overexpression of the active PfMYST led to increased acetylation at H4K5, K8 and K12, and significantly shortened the IDC by almost 4 h. This is compatible with the observation of increased levels of H4K8ac and tetra-acetylated H4 during treatment of the parasite with a HDAC inhibitor apicidin (Chaal et al., 2010). These cells could divide normally, but as a result of the shortened schizogony, they produced fewer merozoites (Fig. 8). In contrast, overexpressing the dominant negative PfMYST that lacks HAT activity led to a general decrease in acetylation at the four lysines of the H4 tail, suggesting that non-active PfMYST could target histones and compete for acetylation with the endogenous PfMYST. The reduced growth rate of the parasites in this case was most likely due to defective schizogony in ~5% of the parasites, which could not complete nuclear division and cytokinesis. This observation is rather intriguing, but continued follow-ups of this parasite line confirmed the consistency of this phenotype. We thus speculate that the presence of this phenotype in a subpopulation of the parasite in each generation resembles the epigenetic phenotypes in other organisms such as the position-effect variegation in Drosophila. In the case of DNA replication, parasite expressing the dominant negative PfMYST always had more parasites with higher number of DNA content. Since the malaria parasite does not have a clearly defined S/G2/M phases and DNA synthesis continues through schizogony (Inselburg & Banyal, 1984), it is unknown whether the increased DNA content in the F1C3-GFP line was associated with the defect in cytokinesis. Similarly, the parasite line overexpressing the active PfMYST had much fewer parasites with higher DNA content (12–24c), which could also be due to fast progression of schizogony. Therefore, although we could not conclude whether overexpression of PfMYST also affected the DNA replication process, these results indicate that PfMYST is involved in cell cycle regulation.

MYST proteins Esa1 and Tip60 are required for DNA damage response and DSB repair (Bird et al., 2002, Murr et al., 2006, Gomez et al., 2008). To determine whether PfMYST-mediated H4 acetylation also participates in DNA damage repair, we analyzed the sensitivity of the parasite lines overexpressing PfMYST to MMS (an alkylating agent that delays replication) and CPT (a topoisomerase inhibitor that causes DSBs). Parasites overexpressing the active PfMYST were more resistant to the effects of these two DNA damaging agents than control parasites overexpressing GFP, whereas parasites overexpressing the dominant negative PfMYST had increased sensitivity to these DNA damaging agents (Fig. 10). This result is consistent with an expected role of PfMYST-mediated H4 acetylation in DNA damage repair, highlighting a conserved function of MYST in evolutionarily divergent eukaryotes. Recently, it has been shown that the second T. gondii MYST member TgMYST-B also confers protection from a DNA-alkylating agent, and the molecular mechanism appears to be via increased levels of ATM kinase and phosphorylated H2AX (Vonlaufen et al., 2010). The mechanism of PfMYST-mediated protection from DSBs remains to be determined, but may be different from that in T. gondii since Plasmodium lacks the histone H2AX.

In summary, we have determined the dynamic expression and cellular localization of PfMYST, and provided experimental evidence about its role in transcription regulation, cell cycle progression and DNA damage repair. We have so far shown that the two types of HATs in malaria parasites are both essential for IDC, suggesting that in addition to HDACs, HAT proteins are potential drug targets. Data gathered from model organisms indicate HAT proteins are present in large protein complexes to execute their in vivo functions (Lee & Workman, 2007). To date, we have very limited information about such protein complexes in malaria parasites, although the parasites share a number of homologues with the yeast and mammalian complexes. Future studies are needed to determine whether these chromatin modification complexes are unique and how to work together to orchestrate the developmental program in the malaria parasite.

Experimental procedures

Sequence analysis

DNA and protein sequences of the yeast MYST members were used for BLAST search of the GenBank and PlasmoDB in order to identify Plasmodium MYST family proteins. Protein mapping and motif searches were performed with Pfam (http://www.sanger.ac.uk/Software/Pfam/) and SMART (http://smart.embl-heidelberg.de). Multiple sequence alignments were compiled using CLUSTALW program. Phylogenetic tree was constructed using the UPGMA program in MEGA4 with pairwise deletion of gaps (Tamura et al., 2007).

Parasite culture

The 3D7 clone of P. falciparum was cultured in type O⁺ human RBCs at 5% hematocrit in RPMI 1640 medium supplemented with 25 mM HEPES, 50 mg/l hypoxanthine, 25 mM NaHCO₃, 0.5% Albumax II, and 40 μg/ml gentamicin sulfate. The culture was maintained at 37°C in a gas mixture of 5% CO₂, 3% O₂ and 92% N₂ (Trager & Jensen, 1976). For synchronization, the culture was initiated with purified schizonts using a Percoll step gradient and the ring stage parasites were treated with 5% sorbitol (Lambros & Vanderberg, 1979). To isolate the parasites, infected erythrocytes were treated with 0.05% saponin to lyse the RBC membrane, and the released parasites were washed twice with cold phosphate-buffered saline (PBS).

RNA isolation, Northern blot, real-time RT-PCR and RACE

Total RNA was isolated using TRIzol Reagent (Invitrogen, Carlsbad, CA) from 3D7 parasites at 12, 24, 36, and 46 h post-infection of the RBCs to represent ring, early trophozoite, late trophozoite, and schizont stage, respectively. The RNA was treated with RNAase-free DNase I (Promega, Madison, WI) to remove contaminating genomic DNA. For Northern blot analysis, 5 μg of total RNA from mixed asexual stages were electrophoresed in 1% agarose/formaldehyde gels. RNA was transferred to nylon membrane and UV cross-linked using Stratalinker (Stratagene, La Jolla, CA). A 360 bp PCR product of Pfmyst F2 × Pfmyst R3 (Table S1) was labeled using the DIG PCR labeling kit (Roche Applied Science, Indianapolis, IN). Hybridization and washes were performed as described previously (Kyes et al., 2000). For RT-PCR analysis, cDNA was synthesized from 1 μg of total RNA using SuperScript III RT (Invitrogen) and oligo- (dT)_12–17 primer, and the reaction was diluted to 100 μl. Real-time RT-PCR was performed using the SYBR Green PCR kit (Roche Applied Science) with 1 μl of the cDNA and primers 5′-CGATTGTTTATGCTCGACC-3′ and 5′-GCTAACCAACAATCTAATC-3′. Relative expression levels of PfMYST at different stages were determined using the 2^−ΔΔCt method with the seryl-tRNA synthetase (STS) gene (PF07_0073) as the internal reference. The ΔC_t value of the ring stage was used as the calibrator (Livak & Schmittgen, 2001). Data analysis and determination of threshold cycle C_t value were as described previously (Miao et al., 2006). The transcription start sites (TSSs) of PfMYST were determined using the FirstChoice® RNA-ligase mediated (RLM)-RACE Kit (Ambion, Austin, TX) with PfMYST-specific outer primer raceF3 and inner primer raceF4 (Table S1) (Fan et al., 2004). The PfMYST 3′ polyadenylation sites were determined by 3′RACE using oligo-dT primer and PfMYST-specific primers raceR1 and raceR2 (Table S1).

Expression of recombinant PfMYST

To determine the enzymatic activity of PfMYST, several PfMYST fragments (F1, F2, F3 and F4) were amplified from genomic DNA using primers listed in Table S1. PCR fragments were cloned into pET28a (Novagen, Gibbstown, NJ) at the BamHI and XhoI sites. His-tagged recombinant proteins were expressed in E. coli strain BL21 (DE3) with 2 mM isopropyl-β-D-thiogalactopyranoside induction for 5 h at 30°C and purified on Ni-nitrilotriacetic acid resin (Qiagen, Valencia, CA) under native conditions. Purified protein from F4 was also used for immunization in rabbits for antibodies (PTG, Chicago, IL).

The aforementioned PfMYST fragments (except F1) were also cloned into pGEX-6P-1 (GE healthcare, Piscataway, NJ) and pMAL-c2X vectors (NEB, Ipswich, MA) for expression with GST and MBP fusions. Fusion proteins were purified with Glutathione Sepharose^™ 4 Fast Flow (GE Healthcare, Piscataway, NJ) and amylose resin(NEB, Ipswich, MA), respectively. F2 and F3 fragments were cloned into pFastBac HT-B vector in order to express PfMYST in insect cells according to the manufacturer’s instruction (Invitrogen) and the expressed protein was purified on Ni-nitrilotriacetic acid resin (Qiagen). The F2 fragment with either His- or GST-tag was cloned into pF3A WG Flexi vector to express PfMYST in a cell-free wheat germ expression system, the TNT SP6 High-Yield protein Expression System (Promega).

HAT assays

HAT assays were performed with ³H acetyl-CoA as previously described using 2 μg of recombinant P. falciparum core histones or HeLa cell mono-nucleosomes as the substrates (Fan et al., 2004, Cui et al., 2008a). The acetylation reactions were analyzed by 15% SDS-polyacrylamide gel electrophoresis (PAGE) followed by fluorography. For quantification, the reactions were spotted on P-81 cation exchange filters, washed, and quantified in a liquid scintillation counter (LSC). To determine the sites of acetylation on histone H4, HAT assays were performed with non-radioactive acetyl-CoA and the recombinant PfH4. PfH4 was separated in 15% SDS-PAGE gels and the acetylation of lysines 5, 8, 12 and 16 on PfH4 was determined using antibodies specific for acetylated lysine residues (H4K5ac, H4K8ac, H4K12ac, and H4K16ac) (Millipore, Billerica, MA). The specificities of the antibodies were verified by using histones from control and sodium butyrate-treated HeLa cells as negative and positive controls, respectively. To further determine the number of acetylated lysine residues, HAT assay was performed with PfMYST and a biotinylated H4 N-terminal peptide (SGRGKGGKGLGKGGAKVLRGSGSK-biotin). The acetylation status of the H4 peptide was determined by mass spectrometry using the Micromass Matrix Assisted Laser Desorption Ionization-Time of Flight (MALDI-TOF) mass spectrometer (Waters, Milford, MA).

Growth complementation of the yeast ESA1 mutants

To test whether PfMYST could complement the function of ESA1 in S. cerevisiae, three fragments of PfMYST (F2, F3 and F4) were constructed in pG1(HA)2. The wild-type yeast strain (LPY3498 MATa his3D200 leu2-3,112 trp1D1 ura3-52 ESA1) and three temperature-sensitive Esa1 mutants, esa1-L254P (LPY3500 MATa his3D200 leu2-3,112 trp1D1 ura3-52 esa1D::HIS3 esa1-L254P::URA3), esa1-L327S (LPY3430 MATa his3D200 leu2-3,112 trp1D1 ura3-52 esa1D::HIS3 esa1-L327S::URA3), and esa1-414 (LPY3291 MATa his3D200 leu2-3,112 trp1D1 ura3-52 esa1D::HIS3+pLP863) were used for complementation study (Clarke et al., 1999). Yeast transformation and confirmation of recombinant PfMYST expression in yeast by Western blots were as described previously (Fan et al., 2004). Transformed cells were plated in fully supplemented synthetic dropout (SD) medium and incubated at 28°C for 3 days. Single colonies were selected, grown in the SD medium at 28°C, and 3 μl of ten-fold serial dilutions beginning with 6 × 10⁶ cells/ml were spotted on plates and incubated at 37°C.

Participation of PfMYST in the regulation of gene expression

To determine the involvement of PfMYST in the regulation of gene expression in the parasite, we selected var2csa (PFL0030c), EBA175 (MAL7P1.176) and calmodulin (PF14_0323) genes for analysis. To select a parasite population with var2csa expression, 3D7 parasites were subjected to three rounds of panning with CSA (Alkhalil et al., 2000). The relative expression level of each var gene was determined by real-time RT-PCR as described (Salanti et al., 2003). To determine whether PfMYST is associated with the active var2csa locus, ChIP was performed using anti-PfMYST antibodies at ring and schizont stages (Cui et al., 2007b). A pair of primers for the STS gene was used as an internal control. ChIP with anti-H4 antibodies served as the control for nucleosomal occupancy (Westenberger et al., 2009). Real-time PCR analysis was performed to compare the relative enrichment of PfMYST in the var2csa promoter and exon 1 (a – h) between CSA-selected and unselected parasites using eight pairs of primers modified following previously published primer sequences of the FCR3 strain (Lopez-Rubio et al., 2007) (Table S1). The relative levels of PfMYST in the promoter regions of EBA175 and calmodulin were mapped using the previously described primers (Cui et al., 2007b). The relative enrichment level was calculated using 2^ΔΔCt=2^[(Ct_MYST−Ct_input)−(Ct_STS−Ct_input)], and normalized with 2^(Ct_H4−Ct_input). The differences of at each site between CSA-selected and unselected parasites were compared using Student’s t-test analysis.

Construction of transfection vectors

Constructs were made to tag PfMYST with either TAP (Rigaut et al., 1999) or GFP. For cloning purpose, a plasmid pBlue-GFP was constructed in pBluescript to contain an expression cassette (hsp86 promoter, GFP gene and pDT3 terminator) (Fan et al., 2009). The plasmid pBlue-TAP was constructed by replacing the GFP tag with TAP. For episomal expression, the full-length long version (F1) and different domain deletions of PfMYST (F2, F3, F4, F1C2, F2C2, F2C3) were amplified (Table S1) and cloned into pBlue-TAP at BglII and AvrII sites to create PfMYST-TAP fusions. To obtain the transfection plasmid, the hsp86-PfMYST-TAP-pDT3 fragment was cloned into pHD22Y at SpeI and NotI sites (Fidock & Wellems, 1997). These constructs were used to episomally express different PfMYST fragments tagged with TAP (Fig. S2C). To tag the endogenous PfMYST with TAP and GFP, the hsp86 promoter was removed from pBlue-TAP and pBlue-GFP and the C-terminal F2 fragment of PfMYST was cloned upstream of the respective tag. The F2-TAP-pDT3 and F2-GFP-pDT3 fragments were cloned in the transfection vector pHD22Y to form pHD/F2-TAPint and F2-GFPint, respectively (Fig. S2A).

For genetic deletion of PfMYST, two DNA fragments of 1067 bp (from 674 bp upstream to 392 bp downstream of the first start codon) and 975 bp (from 1490 bp downstream of the first start codon to 638 bp downstream of the stop codon) in PfMYST were amplified (Table S1) and cloned at the multiple cloning sites 1 (SpeI and SacII) and 2 (EcoR1 and AvrII) of pCC-1, respectively, to form pCC-1/PfMYST (Fig. 4) (Maier et al., 2006).

In order to integrate a single copy of the PfMYST expression cassette into the genome of the parasite, a site-specific integration system using mycobacteriophage Bxb1 integrase was employed to insert the expression cassette at the nonessential cg6 gene locus (Nkrumah et al., 2006). Full-length (F1) and truncated PfMYST (F1C3) were cloned into pLN-ENR-GFP to form pLN-F1-GFP and pLN-F1C3-GFP, respectively (Fig. S2D). A control plasmid pLN-GFP was also constructed for expressing GFP only. All the transgenes were directed by the hsp86 promoter for constitutive expression.

Parasite transfection

For parasite transfection, plasmids were purified using a DNA Maxiprep kit (Qiagen) and ~100 μg of plasmid DNA were introduced into fresh RBCs by electroporation (Deitsch et al., 2001). Parasites transfected with vectors for episomal expression and tagging the endogenous PfMYST (pHD/F2-TAPint and F2-GFPint) were selected for by growth in the presence of 2.5 nM of WR99210 for approximately three weeks with weekly replenishment of fresh RBCs until resistant parasites appeared. To obtain chromosomal integration by pHD/F2-TAPint and F2-GFPint, resistant parasites were subjected to three cycles of drug on-off selection (Crabb & Cowman, 1996). Single parasite clones were obtained by cell sorting (Miao et al., 2010). Correct integration of the plasmid at the PfMYST locus was screened by PCR and Southern blots with specific primers and probes (Table S1).

For single copy integration using a site-specific integration system, the 3D7^attB parasite line, which has attB site integrated at the cg6 locus (Nkrumah et al., 2006), was co-transfected with mycobacteriophage Bxb1 integrase-expressing plasmid pINT and one of attP-containing plasmids (control pLN-GFP, pLN-F1-GFP, pLN-F2-GFP, and pLN-F1C3-GFP). Transfected parasites were selected for using WR99210 (2.5 nM), BSD(2.5 μg/ml) and G418 (125 mg/ml) until resistant parasites appeared. To eliminate episomal plasmids, BSD and G418 were subsequently withdrawn from the culture after three weeks. Single clones were obtained by cell sorting and integration was confirmed by PCR (Table S1) and Western blot with anti-PfMYST (1:1000) and anti-GFP antibodies (1:1000) (Roche Applied Science, Indianapolis, IN). For genetic deletion of PfMYST, parasites transfected with pCC-1/PfMYST were first selected with 2.5 nM of WR99210 and subsequently with 0.4 μM of the negative selection drug 5-fluorocytosine (Valeant, Aliso Viejo, CA).

Purification of tagged PfMYST

To purify TAP-tagged, full-length and truncated PfMYST proteins (F1, F2, F3, F4 and F1C2, F1C3, F2C2 and F2C3) from the parasites, 2 × 10⁹ schizonts were harvested and cell extracts were prepared for TAP (Forler et al., 2003). TAP-tagged PfMYST was purified by IgG Sepharose beads (GE Healthcare) and bound proteins were released by cleavage with the TEV protease (Invitrogen) (Puig et al., 2001). The concentrations of the purified proteins were determined using the MicroBCA Protein Assay kit (Pierce, Rockford, IL) and Western blot was performed with the anti-PfMYST antibodies. Equi-molar purified proteins were used for HAT assays.

Expression and localization of PfMYST protein in parasites

Time-course expression of PfMYST during the erythrocytic cycle was measured by Western blot with PfMYST antibodies. Lysates were prepared from synchronized parasites at 12, 24, 36, and 46 h to represent ring, early trophozoite, late trophozoite and schizont stages, respectively. Equal amounts of the lysate (30 μg) were separated by 12% SDS-PAGE. Western blot was performed using Novex ECL kit (Invitrogen) with anti-PfMYST as the primary antibodies and HRP-conjugated anti-rabbit IgG as the secondary antibodies. Mouse monoclonal antibody 4C9 against HSP70 was used to indicate approximately equal loading (Tsuji et al., 1994). Subcellular locations of PfMYST were visualized using parasite clones with GFP-tagged PfMYST by fluorescence microscopy. Parasite nuclei were counterstained with Hoechst 33342. Images were merged using Adobe Photoshop. IFA was also performed with anti-PfMYST antibodies (1:2000) in 3D7 parasites as previously described (Tonkin et al., 2004).

Overexpression of PfMYST and in vivo histone acetylation

Parasite clones with integration of a single-copy F1-GFP, F1C3-GFP or GFP control at the cg6 locus were synchronized and harvested at 36 h of the IDC. The relative levels of protein expression were determined by Western blots using equal amounts (30 μg) of parasite lysate with anti-PfMYST, anti-GFP and anti-HSP70 antibodies. The level of GFP-tagged PfMYST versus endogenous PfMYST was estimated by densitometry with the Quantity One 1-D Analysis Software (BioRad, Hercules, CA). To detect changes in histone H4 acetylation, histones were purified from PfMYST overexpression lines at the ring, early trophozoite, late trophozoite and schizont stages as described (Miao et al., 2006). For each sample, 5 μg of histones were separated in 15% SDS-PAGE and probed with antibodies specific for acetylated H4K5, K8, K12 and K16, and H3K9 (Millipore catalogue #06-759, 06-760, 06-761, 07-329, and 07-352), respectively. Western blot with anti-H4 antibodies was used to show equal histone loading.

Growth phenotype analysis

Phenotypic changes during the IDC in the PfMYST overexpression lines were precisely measured as described with modifications (Reilly et al., 2007). To measure parasite proliferation, cultures of synchronized parasites were initiated with 0.1% rings, and parasitemia was monitored daily for seven days. To determine cell cycle progression, highly synchronous cultures after two rounds of consecutive synchronization were initiated at 1% parasitemia. Progression of parasites through the IDC was monitored using Giemsa-stained smears every 2 h. Cycle time was determined as the duration between the peak schizont parasitemias of two cycles. The number of merozoites produced per schizont was determined from mature schizonts (segmenters). For easier counts of merozoites, nuclei in schizonts were counter-stained with Hoechst 33342 (20 μM) for 5 min and the smears were observed by light and fluorescence microscopy. Three independent biological replications were done for each parasite line.

EM

To observe ultrastructural changes in the PfMYST-overexpressing lines, schizonts were purified on Percoll gradients, washed with PBS, resuspended in a fixative containing 2% glutaraldehyde and 4% sucrose in 0.05 M phosphate buffer (pH 7.4) for 2 h at room temperature, and post-fixed in 1% OsO4 for 2 h. The parasites were dehydrated in a series of graded ethanol and embedded in epoxy resin. Thin sections of 70 nm thick were cut with a Leica UC6 ultramicrotome (Deerfield, IL) and contrasted with uranyl acetate and lead citrate. The parasite sections were examined under a JEOL JEM 1200 EXII transmission election microscope (Peabody, MA) at 80 kV.

Flow cytometry

To determine the DNA contents of the parasites, parasites at 30, 36, and 42 h were purified on Percoll gradients, incubated with 1 μM Vybrant® DyeCycle^™ Orange (Molecular Probes, Eugene, OR) at 37°C for 30 min, and applied to a flow cytometer. The intensity of fluorescence emitted from the dye-bound DNA was quantified for each parasitized erythrocyte, and the genome copy was estimated using the ring stage parasites as the single-copy genome standard. At least 5,000 parasitized erythrocytes were differentiated for each parasite line and the distribution of parasite numbers with different genome copies was plotted.

Sensitivity to genotoxic agents

To analyze whether overexpression of PfMYST could alter the parasite’s sensitivity to DNA-damaging agents, parasites were cultured in the presence of two-fold dilutions of CPT (0.625–2560 μg/ml) and MMS (0.02–81.92 μg/μl) for 48 h. Parasite growth at different drug concentrations was measured using the SYBR Green I method (Smilkstein et al., 2004). Data were analyzed by using the GraphPad Prism software (San Diego, CA) and half maximal effective concentrations (EC₅₀) were calculated.

Statistical analysis

To compare the growth rates of different parasite lines (F1-GFP, F1C3-GFP or GFP-control), a quadratic function of time was fitted for each parasite line using linear mixed-effects models. For data at a specific time point, ANOVA analysis was carried out to compare the cycle duration for these parasite lines. Bonferroni correction was applied to control the family-wise error rate. To analyze the schizont numbers containing different numbers of merozoites, a χ² goodness of fit test was first used to evaluate if the number of schizonts that contain a certain number of merozoites was independent of the parasite lines. Then the proportions of schizonts with a certain number of merozoites were compared among these cell lines based on ANOVA for each merozoite number ranging from 17 to 22. Tukey’s pairwise comparison was conducted to analyze the EC₅₀ results of each pair of parasite lines under different chemical agents (CPT and MMS).