The ApiAP2 factor PfAP2-HC is an integral component of heterochromatin in the malaria parasite Plasmodium falciparum

Summary Malaria parasites undergo a complex life cycle in the human host and the mosquito vector. The ApiAP2 family of DNA-binding proteins plays a dominant role in parasite development and life cycle progression. Most ApiAP2 factors studied to date act as transcription factors regulating stage-specific gene expression. Here, we characterized an ApiAP2 factor in Plasmodium falciparum that we termed PfAP2-HC. We demonstrate that PfAP2-HC specifically binds to heterochromatin throughout the genome. Intriguingly, PfAP2-HC does not bind DNA in vivo and recruitment of PfAP2-HC to heterochromatin is independent of its DNA-binding domain but strictly dependent on heterochromatin protein 1. Furthermore, our results suggest that PfAP2-HC functions neither in the regulation of gene expression nor in heterochromatin formation or maintenance. In summary, our findings reveal PfAP2-HC as a core component of heterochromatin in malaria parasites and identify unexpected properties and substantial functional divergence among the members of the ApiAP2 family of regulatory proteins.


INTRODUCTION
The apicomplexan parasite Plasmodium falciparum is the main cause of severe malaria worldwide, with the majority of the estimated 405,000 malarial deaths in 2018 attributed to this pathogen (WHO, 2019). The symptoms of the disease occur owing to repeated asexual intraerythrocytic developmental cycles (IDCs), where merozoite stage parasites invade human red blood cells (RBCs) and develop through the ring stage (0-24 h post invasion [hpi]) and trophozoite stage (24-30 hpi), before undergoing schizogony to produce mature segmented schizonts containing up to 32 merozoites (30-48 hpi). Rupture of the infected RBCs (iRBCs) releases the merozoites, which in turn undergo another IDC after invading new RBCs. A small proportion of schizonts per cycle commit to the sexual development pathway and produce ring stage daughter cells that mature over a period of 10 days and through four intermediate stages (I-IV) into mature stage V gametocytes (Venugopal et al., 2020). Circulating stage V gametocytes are the only forms of the parasite able to infect the mosquito vector and are therefore essential for malaria transmission.
A key trait of P. falciparum is the ability to adapt to and evade the constantly changing environment in its human host through clonally variant gene expression, a process vital to a broad range of biological processes, including antigenic variation, RBC invasion, solute transport, and sexual conversion (Duraisingh and Skillman, 2018;Llora-Batlle et al., 2019;Rovira-Graells et al., 2012). Clonally variant gene expression in P. falciparum is regulated epigenetically, with heritable gene silencing mediated by heterochromatin (Voss et al., 2014). Heterochromatin is found at subtelomeric regions on all 14 chromosomes and in some chromosome internal islands and is characterized by the binding of heterochromatin protein 1 (PfHP1) to the histone modification histone 3 lysine 9 trimethylation (H3K9me3) (Flueck et al., 2009;Fraschka et al., 2018;Lopez-Rubio et al., 2009;Perez-Toledo et al., 2009;Salcedo-Amaya et al., 2009). These PfHP1/ H3K9me3-demarcated heterochromatic domains cover over 400 genes in total (approximately 8% of all protein-coding genes in the genome) (Flueck et al., 2009;Fraschka et al., 2018). As a core component of heterochromatin, PfHP1 plays an essential role in heterochromatic gene silencing and has a multi-faceted role in parasite biology as previously demonstrated with a conditional loss-of-function mutant (Brancucci et al., 2014). Conditional depletion of PfHP1 resulted in the de-repression of multi-copy gene families important in antigenic variation, including the well-characterized var gene family (Brancucci et al., 2014; Scherf Nuclei were stained with DAPI. DIC, differential interference contrast. Scale bar, 5 mm. (C) Log2-transformed a-PfHP1 (orange) and a-GFP (blue) ChIP-over-input ratio tracks obtained from 3D7/GFP-PfAP2-HC schizont stage parasites. a-PfHP1 and a-GFP ChIP tracks have been offset by 2 and 1, respectively, to be able to display the full scale of variation. In addition, a-GFP ChIP-seq data are mirrored on a negative scale. Dashed boxes highlight regions that are enlarged in (E)-(G).
In order to investigate the genome-wide binding profile of GFP-PfAP2-HC and to allow comparison with PfHP1 at high resolution, we performed chromatin immunoprecipitation-sequencing (ChIP-seq) using a-GFP and a-PfHP1 antibodies to compare binding profiles within the same parasite population. We found that GFP-PfAP2-HC indeed co-localizes with PfHP1 throughout the genome (Figures 1C,1E,1F and 1G). To quantify the degree of co-localization, we computed and compared PfHP1 and PfAP2-HC ChIP-over-input enrichment values in coding regions across the genome (Data S1). This confirmed a strong correlation (R 2 = 0.91) between PfAP2-HC and PfHP1 occupancies across coding regions of all heterochromatic genes (Figure 1D). In addition, we visualized on all chromosomes the locations of the putative PfAP2-HC target DNA motif (CACACA) as predicted by in vitro binding preference of the recombinant PfAP2-HC AP2 DBD (Campbell et al., 2010). The CACACA motif showed no enrichment in heterochromatic over euchromatic regions and therefore showed no positional association with the in vivo PfAP2-HC binding profile (Figure 1E). Collectively, these findings show that PfAP2-HC localizes exclusively to PfHP1-defined heterochromatic regions and seems not to bind to the predicted CACACA target motifs in vivo.

PfAP2-HC is not required for heterochromatin maintenance and inheritance
Having shown that PfAP2-HC shares the genome-wide binding profile of PfHP1, we next investigated the function of this ApiAP2 factor by creating a conditional knockdown line employing the FKBP destabilization domain (DD) system. DD-tagged proteins are stabilized in the presence of the small molecule Shield-1, and removal of this ligand leads to protein degradation (Armstrong and Goldberg, 2007;Banaszynski et al., 2006). We utilized our two-plasmid CRISPR-Cas9 approach to N-terminally tag PfAP2-HC with DDGFP to create the cell line 3D7/DDGFP-PfAP2-HC (Figures 2A and S2). Limiting dilution cloning resulted in a parasite clone containing the correctly edited locus, which we confirmed by PCR on gDNA ( Figure S2). Substantial depletion of DDGFP-PfAP2-HC expression in the absence of Shield-1 was verified by live cell fluorescence imaging ( Figure 2B) and Western blot (Figures 2C and S2). Depletion of DDGFP-PfAP2-HC expression caused no major cell cycle-or proliferation-related phenotypes nor did it have an effect on sexual conversion rates ( Figure S3).
In order to investigate the potential effect of PfAP2-HC depletion on heterochromatin, we grew parasites in the presence or absence of Shield-1 for 13 generations and compared their genome-wide PfHP1 binding profiles by ChIP-seq. The genome-wide PfHP1 coverage tracks in 3D7/DDGFP-PfAP2-HC parasites grown in the absence or presence of Shield-1 are highly similar ( Figure 2D). Likewise, the genome-wide PfHP1 coverage of coding regions in the two populations is nearly identical (R 2 = 0.99) ( Figure 2E and Data S1) showing that depletion of PfAP2-HC has no discernible effect on PfHP1 localization on chromatin. To test whether the lack of obvious loss-of-function phenotypes was due to the residual amounts of DDGFP-PfAP2-HC protein remaining after Shield-1 removal ( Figure 2C), we also generated a PfAP2-HC knockout cell line, 3D7/PfAP2-HC-KO ( Figure S4), which we confirmed by PCR on gDNA ( Figure S4). 3D7/PfAP2-HC-KO parasites did not show obvious growth-related phenotypic changes either ( Figure S4) and maintained PfHP1 occupancy at levels similar to 3D7 wildtype (3D7/WT) and 3D7/DDGFP-PfAP2-HC parasites ( Figure 2D). Changes in PfHP1 coverage of some genes were observed in 3D7/PfAP2-HC-KO parasites compared with 3D7/WT and 3D7/DDGFP-PfAP2-HC ( Figures 2F and S4 and Data S1). However, these changes are likely unrelated to the lack of PfAP2-HC expression but rather attributable to clonally variant changes in PfHP1 occupancy as similar differences are observed when comparing different PfAP2-HC-expressing clonal lines (3D7/WT and 3D7/DDGFP-PfAP2-HC) ( Figure S4). Together, these results show that PfAP2-HC is neither required for asexual proliferation nor for the maintenance and inheritance of PfHP1demarcated heterochromatin.   Figure S2. (B) Representative live cell fluorescence images of 3D7/DDGFP-PfAP2-HC schizonts (36-44 hpi) grown in the presence (+) or absence (À) of Shield-1. Nuclei were stained with Hoechst. DIC, differential interference contrast. Scale bar, 5 mm. See also Figure S3.

OPEN ACCESS
iScience 24, 102444, May 21, 2021 5 iScience Article PfAP2-HC does not act as a transcription factor in blood stage parasites To identify any possible role of PfAP2-HC in transcriptional regulation we performed a transcriptome-wide microarray time course analysis. We compared 3D7/DDGFP-PfAP2-HC parasites grown in the presence and absence of Shield-1 across five time points throughout the IDC ( Figure 3A). For each of the five time points the paired transcriptome data were strongly correlated based on Pearson correlation values, demonstrating highly comparable stage composition across the time course ( Figure 3A and Data S2). We found no significant difference in gene expression, with no transcripts showing greater than 2-fold average fold change in steady-state mRNA abundance between the +Shield-1 and -Shield-1 populations ( Figure 3B), suggesting that PfAP2-HC does not play a dominant role in transcriptional regulation in blood stage parasites and further corroborating the lack of obvious phenotypes associated with PfAP2-HC depletion.
The AP2 domain of PfAP2-HC is dispensable for targeting PfAP2-HC to heterochromatin To discern the importance of the single AP2 DBD in targeting PfAP2-HC to heterochromatin we introduced a STOP codon prior to the AP2 domain, replacing amino acid R1319 with a premature STOP codon in 3D7/ GFP-PfAP2-HC to create the parasite line 3D7/GFP-PfAP2-HC-DDBD ( Figures 4A and S5). PCR on gDNA confirmed successful editing of the locus ( Figure S5). The transgenic population consisted of a mixture of parasites either with correctly edited locus or carrying integrated donor plasmid concatemers ( Figure S5). Of importance, both recombination events introduce the desired premature STOP codon into the pfap2-hc coding sequence. Indeed, Sanger sequencing of the amplified PCR products verified successful introduction of the premature STOP codon in the entire population ( Figure S5). The localization of GFP-PfAP2-HC-DDBD is comparable with that of GFP-PfAP2-HC by IFA and similarly shares this localization pattern with PfHP1 ( Figure 4B).
For a more comprehensive analysis, we again performed ChIP-seq experiments using a-GFP and a-PfHP1 antibodies on 3D7/GFP-PfAP2-HC-DDBD parasites. As with full-length GFP-PfAP2-HC, the truncated PfAP2-HC-DDBD protein co-localized with PfHP1 throughout the genome with highly correlated enrichment on all heterochromatic genes (Figures 4C and 4D and Data S1) showing that the AP2 DBD of PfAP2-HC is dispensable for its localization to heterochromatin.

Binding of PfAP2-HC to heterochromatin is PfHP1 dependent
PfAP2-HC is targeted to heterochromatin in the absence of its only recognizable DBD, suggesting a reliance on protein-protein interactions independent of the AP2 domain. To gain insight into this interaction, we tagged PfHP1 with the fluorescent protein mScarlet. In addition, we introduced a sequence encoding the glms riboswitch element (Prommana et al., 2013) downstream of the STOP codon, such that the resulting pfhp1-mscarlet mRNA contains a functional glms ribozyme in its 3 0 untranslated region. Upon addition of glucosamine (GlcN) to the culture medium, the glms ribozyme mediates mRNA cleavage and degradation (Prommana et al., 2013;Watson and Fedor, 2011). We generated this conditional PfHP1 knockdown cassette in the background of the 3D7/GFP-PfAP2-HC clone to create the 3D7/GFP-PfAP2-HC/PfHP1-mScarlet-glmS double transgenic parasite line ( Figures 5A and S6). We confirmed correct editing of the pfhp1 locus by PCR on gDNA ( Figure S6). To investigate the effect of PfHP1 depletion on the localization of GFP-PfAP2-HC, we split 3D7/GFP-PfAP2-HC/PfHP1-mScarlet-glmS parasites at 0-8 hpi into two populations, adding GlcN to one of them to induce the knockdown of PfHP1-mScarlet expression (+GlcN) and keeping the other one under stabilizing conditions (-GlcN). Live cell fluorescence imaging and Western blot analysis of schizont stage parasites confirmed the efficient depletion of PfHP1-mScarlet expression in +GlcN conditions (Figures 5B,5C,and S6). Of interest, upon PfHP1-mScarlet depletion, GFP-PfAP2-HC localized diffusely throughout the nucleoplasm and no longer displayed a punctate perinuclear pattern ( Figure 5B), showing mis-localization in the absence of PfHP1. ChIP-over-input tracks from 3D7/DDGFP-PfAP2-HC schizont stage parasites grown in the presence (+) or absence (À) of Shield-1 (top two tracks). Log2-transformed a-PfHP1 ChIP-over-input tracks from 3D7/WT and 3D7/PfAP2-HC-KO schizonts (bottom two tracks). Coding sequences are shown as blue (sense strand) and red (antisense strand) boxes. See also Figure S4.
(E and F) Scatterplots of average log2-transformed a-PfHP1 ChIP-over-input values at all coding regions in 3D7/DDGFP-PfAP2-HC schizonts grown in the presence (+) or absence (À) of Shield-1 (E) and in 3D7/WT and 3D7/PfAP2-HC-KO schizonts (F). Depicted regression lines are based on heterochromatic genes only (log2 ratio a-PfHP1/input R0). The coefficient of determination (R 2 ) is shown in the upper left corner. See also Data S1. The ChIP-seq results presented in Figure 1 provided no evidence for direct binding of PfAP2-HC to DNA in euchromatic regions. However, this experiment did not allow us to test if PfAP2-HC binds to DNA sequences in heterochromatic regions because its association with PfHP1 would have masked such interactions. Hence, we used the 3D7/GFP-PfAP2-HC/PfHP1-mScarlet-glmS line to ask whether PfAP2-HC binds directly to DNA in the absence of PfHP1. We grew 3D7/GFP-PfAP2-HC/PfHP1-mScarlet-glmS parasites in the presence of GlcN from early ring stages (0-8 hpi) and harvested samples for ChIP-seq at 40-48 hpi within the same cycle. As expected, we observed a large reduction in PfHP1 enrichment in heterochromatic domains ( Figure 5D). GFP-PfAP2-HC occupancy was massively reduced, and in two biologically independent ChIP-seq experiments we could not detect signals over background ( Figure 5D and Data S1). Together, these results show that PfAP2-HC localization to heterochromatin is entirely dependent on PfHP1 and no evidence for direct binding of PfAP2-HC to DNA in these regions could be discerned.

PfAP2-HC is likely not involved in heterochromatin formation
We have shown that maintenance and inheritance of heterochromatin was unaffected in both the 3D7/ PfAP2-HC-KO null mutant and in the conditional 3D7/DDGFP-AP2-HC loss-of-function mutants after 13  Figure 2D). However, factors influencing the initial establishment of heterochromatin can be independent of maintenance and inheritance (Sadaie et al., 2004). Taking advantage of the fact that conditional knockdown of PfHP1 expression produces progeny consisting of approximately 50% viable heterochromatin-depleted early-stage gametocytes and 50% growth-arrested trophozoites (Brancucci et al., 2014), we investigated whether PfAP2-HC is required for the re-establishment of heterochromatin during gametocyte maturation. To achieve this, we generated a parasite line allowing for the conditional knockdown of both PfHP1 and PfAP2-HC, 3D7/DDGFP-PfAP2-HC/PfHP1-mScarlet-glmS ( Figures 6A and S6). The 3D7/DDGFP-PfAP2-HC/PfHP1-mScarlet-glmS line was obtained by tagging the pfhp1 gene in the 3D7/DDGFP-AP2-HC clone with mscarlet-glmS as described above ( Figure S6). We confirmed correct editing of the pfhp1 locus by PCR on gDNA ( Figure S6). Routine culture of this parasite line in the presence of Shield-1 and absence of GlcN stabilizes DDGFP-PfAP2-HC and PfHP1-mScarlet expression, respectively. We divided ring stage parasites into two populations at 0-8 hpi (generation 1), of which one was maintained under stabilizing conditions for both proteins and from the other one Shield-1 was removed to induce DDGFP-PfAP2-HC depletion. At 0-8 hpi in The pfap2-hc gene was tagged with gfp. The pfhp1 gene was tagged with the mscarlet sequence followed by a glmS ribozyme element to allow for detection and conditional expression of PfHP1-mScarlet, respectively. See also Figure S6. (B) Representative live cell fluorescence images of 3D7/GFP-PfAP2-HC/PfHP1-mScarlet-glmS parasites at 32-40 hpi grown in the absence of GlcN (PfHP1 expressed) or the presence of GlcN (PfHP1 depleted). Nuclei were stained with Hoechst. DIC, differential interference contrast. Scale bar, 5 mm.

OPEN ACCESS
iScience 24, 102444, May 21, 2021 9 iScience Article Figure 6. Depletion of PfAP2-HC has no marked effect on re-establishment of heterochromatin (A) Schematic map of the endogenous pfap2-hc and pfhp1 loci in 3D7/DDGFP-PfAP2-HC/PfHP1-mScarlet-glmS parasites after CRISPR-Cas9-mediated gene editing. The pfap2-hc locus was modified to introduce a ddgfp tag. The pfhp1 locus was modified to contain an mscarlet tag followed by the glmS ribozyme element. See also Figure S6.

DISCUSSION
Clonally variant gene expression is key to the survival of P. falciparum in the human host and is dependent on heterochromatin-mediated gene silencing. PfHP1, as a core component of heterochromatin, is essential for regulating processes as diverse as antigenic variation, invasion pathway switching, commitment to gametocytogenesis, and asexual proliferation (Brancucci et al., 2014;Voss et al., 2014). Our study characterizes PfAP2-HC, a member of the ApiAP2 family of putative DNA-binding proteins that specifically associates with heterochromatin throughout the genome.
Despite progress toward understanding the heterochromatic landscape of P. falciparum, a global view of the dynamic processes occurring to regulate and maintain heterochromatin in this parasite remains elusive.
Here, we describe PfAP2-HC as an integral component of heterochromatin, only the second such factor to be characterized after gametocyte development 1 (GDV1) (Filarsky et al., 2018). GDV1 is not expressed in asexual parasites but only in parasites undergoing sexual commitment. In these cells, GDV1 binds to heterochromatin throughout the genome and destabilizes heterochromatin particularly at the pfap2-g locus and early gametocyte markers thus facilitating their expression (Filarsky et al., 2018). In contrast, PfAP2-HC is expressed and binds to heterochromatin in asexual parasites. Depletion of PfAP2-HC had no effect on PfHP1 localization suggesting it is not required for heterochromatin maintenance. Factors shown to date to be involved in heterochromatin maintenance in P. falciparum consist of histone-modifying enzymes, such as the histone deacetylase PfHda2, whose absence leads to the expression of many PfHP1-associated genes including subtelomeric multi-gene families and the internally located pfap2-g locus (  . Continued horizontal arrow) to restore PfHP1-mScarlet expression during gametocytogenesis. The double vertical arrows indicate the time points of live cell fluorescence imaging experiments to assess PfHP1-mScarlet localization in DDGFP-PfAP2-HC-expressing (+Shield-1) and -depleted (ÀShield-1) parasites.

OPEN ACCESS
iScience 24, 102444, May 21, 2021 11 iScience Article reversible gene silencing in P. falciparum is an interesting and equally challenging question for future research.
We also tested whether the absence of PfAP2-HC may influence heterochromatin formation rather than maintenance. Because PfHP1 is essential for the proliferation of asexual parasites, we performed this experiment in gametocytes where PfHP1 is dispensable (Brancucci et al., 2014). To this end, we first depleted PfHP1 in 3D7/DDGFP-PfAP2-HC/PfHP1-mScarlet-glmS parasites through conditional knockdown of PfHP1 expression and then rescued PfHP1 expression in the sexual ring stage progeny and visualized the re-establishment of heterochromatic foci in stage II and V gametocytes by fluorescence microscopy based on PfHP1-mScarlet positivity. We did not observe any difference in the localization of PfHP1 between gametocytes expressing or not expressing PfAP2-HC. This result provides preliminary evidence suggesting that de novo formation of heterochromatin occurs independent of PfAP2-HC. This is in keeping with our observation that PfAP2-HC does not seem to bind chromosomal DNA in vivo and that the localization of PfAP2-HC is dependent on the presence of PfHP1, as discussed further below. However, we cannot exclude the possibility that a role for PfAP2-HC in nucleating heterochromatin may have been masked in this experiment by the spreading of heterochromatin from residual PfHP1 foci that remained bound to chromatin owing to incomplete PfHP1 knockdown in the asexual progenitors ( Figure 5D).
We showed that the AP2 DBD of PfAP2-HC is not required for correct localization of the protein to heterochromatin. Furthermore, we could not detect direct binding of PfAP2-HC to the predicted CACACA target motifs (Campbell et al., 2010) or to other sites in chromosomal DNA in vivo by ChIP-seq, neither in euchromatin nor in heterochromatin, and PfAP2-HC depletion had no effect on gene transcription during the IDC. In addition, recent pull-down experiments of native nuclear proteins binding to specific DNA probes also failed to reveal an interaction of full-length PfAP2-HC with the CACACA motif (Toenhake et al., 2018). Together, these results imply that PfAP2-HC does not bind chromosomal DNA in vivo, suggesting functional divergence of AP2 domains within the ApiAP2 family. Although DNA-binding motifs were identified for most AP2 domains in vitro (Campbell et al., 2010), two of the three AP2 domains of PfAP2-I were recently shown to be dispensable in the IDC and it is unknown if they actually bind DNA in vivo (Santos et al., 2017). It is still possible that any direct DNA binding of PfAP2-HC was below the detection limit of our ChIP-seq experiments. However, it is perhaps more likely that PfAP2-HC indeed does not bind DNA directly in vivo, given its dependence on PfHP1 for correct localization. In fact, because PfAP2-HC interacts with heterochromatin independent of its AP2 domain, PfAP2-HC may actually not be meant to bind DNA directly; PfAP2-HC would likely recruit heterochromatin to any chromosomal sites it would bind to and thus potentially silence expression of genes that are important for parasite viability.
The apparent lack of DNA-binding activity displayed by the PfAP2-HC AP2 domain and the capacity of PfAP2-HC to localize to heterochromatin in absence of the AP2 domain suggest that protein-protein interactions involving the large N terminus of the protein are responsible for targeting PfAP2-HC to heterochromatin. Multiple sequence alignments of AP2-HC orthologs across all human-infecting Plasmodium spp.
show only 30%-36% sequence identity to PfAP2-HC, and this is comparable with the AP2-HC orthologs of rodent-infecting species (31%-32%) ( Figure S7). High sequence similarity is mainly confined to the AP2 domain itself, which shares R90% identical amino acids across all species ( Figure S7). Of interest, there is a second semi-conserved region of 172 amino acids within PfAP2-HC with 64%-67% sequence identity to the orthologs of other human-infecting species and 53%-56% identity to those from rodent-infecting species ( Figure S7), which points to an evolutionarily conserved feature. One could speculate that this region may be involved in mediating interactions with PfHP1 or other chromatin-associated factors. To date, the role of the non-AP2 region of ApiAP2 proteins has not been explicitly studied. However, given the regula- iScience Article The AP2-HC factor is conserved among all Plasmodium spp., which clearly suggests an important role for this factor in the biology of malaria parasites, at least in vivo. We obtained a viable PfAP2-HC KO line that lacks any obvious phenotype in asexual blood stage parasites, but we cannot rule out functionally critical roles in other life cycle stages. Indeed, RNA-seq data show pfap2-hc expression in gametocyte and sporozoite stages (plasmodb.org) (Aurrecoechea et al., 2009;Gomez-Diaz et al., 2017;Lasonder et al., 2016). However, the orthologs of PfAP2-HC were successfully disrupted in the rodent malaria parasites P. berghei and P. yoelii, without discernible growth defects observed during the full life cycle in laboratory animals (Modrzynska et al., 2017;Zhang et al., 2017). These results suggest that functional redundancy or compensatory mechanisms may exist among the ApiAP2 family, as also proposed by Zhang and colleagues (Zhang et al., 2017). However, at least in asexual blood stage parasites, we believe mechanisms compensating for loss of PfAP2-HC function are highly unlikely given that the conditional knockdown of PfAP2-HC expression did not result in any transcriptional changes and caused not even a temporary defect on parasite growth or multiplication. Beyond this, it is also possible that PfAP2-HC is involved in more subtle processes not studied here, which may not present as immediate phenotypes in loss-of-function mutants but may be crucial for parasite fitness in the field. Examples of such processes are DNA repair/recombination within heterochromatic regions or epigenetic memory/switching frequencies of heterochromatic genes. The heterochromatic subtelomeric regions, which contain several hundred members of multi-copy gene families, recombine at a higher rate than the core genome in P. falciparum, resulting in high antigenic diversity within the parasite population (Bopp et al., 2013;Claessens et al., 2014;Frank et al., 2008). Furthermore, DNA repair mechanisms are generally less efficient in heterochromatin compared with euchromatin and thus contribute to increased mutation rates in these regions ( , PfAP2-HC joins the ranks of ApiAP2 factors that do not primarily act as transcriptional regulators. We rather characterized PfAP2-HC as a PfHP1-interacting protein and core component of heterochromatin in P. falciparum. We found no evidence for direct binding of PfAP2-HC to chromosomal DNA in vivo and show that the localization of PfAP2-HC to heterochromatin is independent of the AP2 domain but strictly dependent on the presence of PfHP1. Although our efforts failed to reveal conclusive insight into PfAP2-HC function, we discovered unexpected properties of ApiAP2 factors that highlight the functional diversity among the members of this family of putative DNAbinding proteins.

Limitations of the study
As we did not observe any PfAP2-HC loss-of-function phenotypes in P. falciparum blood stage parasites in our study, targeted experiments in other life cycle stages will be necessary to reveal insight into the function of this ApiAP2 factor. Furthermore, although our preliminary microscopy-based data presented in Figure 6 suggest that PfAP2-HC is not involved in de novo heterochromatin formation, ChIP-seq and RNA-seq experiments would be required to confirm this result at higher resolution.   (C) PCR on gDNA from a 3D7/GFP-PfAP2-HC clone and 3D7 wild-type parasites. Primers ap2-hc-5'_F and ap2-hc-5'_R bind to chromosomal sequences outside the HRs and amplify a 2104 bp or 1393 bp fragment from the edited or wild-type pfap2-hc locus, respectively. The ap2-hc-5'_F-gfp_R and gfp_F-ap2-hc-5'_R primer combinations are specific for the edited locus and amplify 650 bp and 795 bp fragments, respectively. Primer pD_F binds to the donor plasmid backbone and, when used in combination with primer ap2-hc-5'_R, will amplify a fragment of 2126 bp if a donor plasmid concatemer was integrated into the genome.
(D) Live cell fluorescence imaging of 3D7/GFP-PfAP2-HC parasites throughout the IDC. R, ring stage. LT, late trophozoite with two parasites infecting one RBC. ES, early schizont. LS, late schizont. Nuclei were stained with Hoechst. DIC, differential interference contrast. Scale bar, 5 µm. (A) Schematic maps of the pfap2-hc locus (PF3D7_1456000) in 3D7 parasites (top), the CRISPR/Cas9 transfection vectors pFDon_ddgfp-pfap2-hc and pH_gC-ap2-hc-5'-2 (centre), and the modified pfap2-hc locus after CRISPR/Cas9-based genome editing in 3D7/DDGFP-PfAP2-HC parasites (bottom). The AP2 DBD-encoding sequence, which is interrupted by an intron, is indicated (AP2, dark blue). The position of the sgt_ap2-hc-5'-1 sgRNA target sequence is indicated (chromosome 14 coordinates). The pFDon_ddgfp-pfap2-hc donor plasmid contains an FKBP destabilisation domain (dd, orange) and gfp sequence (green) flanked by homology regions (HR, yellow) for homology-directed repair. The pH_gC-ap2-hc-5'-2 plasmid contains expression cassettes for SpCas9 (dark grey), the sgRNA (pink) and the hdhfr resistance marker (brown). Successful gene editing results in the expression of an N-terminally tagged DDGFP-PfAP2-HC protein. PCR primer binding sites are indicated by half arrows and were used to confirm successful gene editing. (B) Schematic map of the modified pfap2-hc locus after CRISPR/Cas9-based genome editing in the event of donor plasmid concatemer integration into the genome. PCR primer binding sites are indicated by arrows and were used to check for donor plasmid concatemer integration. (C) PCR on gDNA from a 3D7/DDGFP-PfAP2-HC clone and 3D7 wild-type parasites. Primers ap2-hc-5'_F and ap2-hc-5'_R bind to chromosomal sequences outside the HRs and amplify a 2440 bp or 1393 bp fragment from the edited or wild-type pfap2-hc locus, respectively. The ap2-hc-5'_F-gfp_R and gfp_F-ap2-hc-5'_R primer combinations are specific for the edited locus and amplify 986 bp and 795 bp fragments, respectively. Primer pD_F binds to the donor plasmid backbone and, when used in combination with primer ap2-hc-5'_R, will amplify a fragment of 2466 bp if a donor plasmid concatemer was integrated into the genome.
(D) Full sized Western blot of the sections shown in Figure 2C showing DDGFP-PfAP2-HC expression levels in 3D7/DDGFP-PfAP2-HC parasites grown in the presence (+) or absence (-) of Shield-1. The membrane was first probed with α-GFP antibodies (top) before inactivation of horseradish peroxidase with 2 mM NaN3, followed by re-probing with the α-PfHP1 antibodies (bottom) used as a loading control. Dashed boxes show the sections presented in Figure 2C. Representative flow cytometry plots of an infected (3D7/WT,  Shield-1, panel A) and uninfected RBC control sample (panel B) measured on day 1 of the multiplication assay. The first plot shows the gate to remove debris smaller than cell size to include only the 'cells' population. The second plot shows the gate to include only single measurement events, termed 'singlets', and the third gate separates uninfected from infected RBCs based on the SYBR Green intensity of the uninfected RBC control, termed 'parasites'. The numbers are the percentage of events included within the gate, with the final gate 'parasites' reflecting the parasitaemia of the sample. This gating strategy was applied to all flow cytometry data shown in panels C and D, and in Figure S4.  (A) Schematic maps of the pfap2-hc locus (PF3D7_1456000) in 3D7 parasites (top), the p_gCH-pfap2-hc-KO transfection vector (centre), and the modified pfap2-hc locus after CRISPR/Cas9-based genome editing in 3D7/PfAP2-HC-KO parasites (bottom). The AP2 DBD-encoding sequence, which is interrupted by an intron, is indicated (AP2, dark blue). The position of the sgt_ap2-hc-KO sgRNA target sequence is indicated (chromosome 14 coordinates). The p_gCH-pfap2-hc-KO plasmid contains expression cassettes for SpCas9 (dark grey), the sgRNA (pink) and the hdhfr resistance marker (brown) flanked by two homology regions (HR, yellow) for homology-directed repair. Successful gene editing results in the hdhfr expression cassette replacing a section of the pfap2-hc gene, disrupting its expression. PCR primer binding sites are indicated by arrows and were used to confirm successful gene editing. (A) Schematic maps of the gfp-pfap2-hc locus in 3D7/GFP-PfAP2-HC parasites (top, see Figure S1), the CRISPR/Cas9 transfection vectors pD_gfp-pfap2-hc-ΔDBD and pH_gC-ap2-hc-3' (centre), and the modified gfp-pfap2-hc locus after CRISPR/Cas9-based genome editing in 3D7/GFP-PfAP2-HC-ΔDBD parasites (bottom). The AP2 DBD-encoding sequence, which is interrupted by an intron, is indicated (AP2, dark blue). The position of the sgt_ap2-hc-3' sgRNA target sequence is indicated (chromosome 14 coordinates). The pD_gfp-pfap2-hc-ΔDBD donor plasmid contains a premature TAA stop codon (red) flanked by homology regions (HR, yellow) for homology-directed repair. The pH_gC-ap2-hc-3' plasmid contains expression cassettes for SpCas9 (dark grey), the sgRNA (pink) and the hdhfr resistance marker (brown). Successful gene editing results in the expression of a truncated GFP-PfAP2-HC protein lacking the AP2 DNA-binding domain (GFP-PfAP2-HC-ΔDBD). PCR primer binding sites are indicated by arrows and were used to confirm successful gene editing. (B) Schematic map of the modified gfp-pfap2-hc locus after CRISPR/Cas9-based genome editing in the event of donor plasmid concatemer integration into the genome. PCR primer binding sites are indicated by arrows and were used to check for donor plasmid concatemer integration.
(C) PCR on gDNA from 3D7/GFP-PfAP2-HC-ΔDBD and 3D7 wild-type parasites. Primers ap2-hc-3'_F and ap2-hc-3'_R bind to chromosomal sequences outside the HRs and amplify a 1296 bp or 1549 bp fragment from the edited or wild-type pfap2-hc locus, respectively. Primer pD_R binds to the donor plasmid backbone and, when used in combination with primer ap2-hc-3'_F, amplifies a fragment of 1364 bp if a donor plasmid concatemer was integrated into the genome.
(D) Sanger sequencing of the two PCR products ap2-hc-3'_F-ap2-hc-3'_R (top) and ap2-hc-3'_F-pD_R (middle) amplified from 3D7/GFP-PfAP2-HC-ΔDBD parasites (see panel B, lanes 2 and 3) confirms the successful introduction of the AGTA double mutation creating a premature STOP codon (R1319*). The PCR product ap2-hc-3'_F-ap2-hc-3'_R (bottom) amplified from 3D7 wild-type parasites (see panel B, lane 5) shows the wild-type sequence. Additional mutations downstream of the AGTA double mutation are part of the re-codonised sequence introduced to avoid homologues recombination at an undesired location to ensure correct CRISPR/Cas9 genome editing. . The position of the sgt_guide250 sgRNA target sequence is indicated (chromosome 12 coordinates). The pD_hp1-mScarlet-glmS donor plasmid contains the mScarlet sequence (red) followed by the glmS ribozyme sequence (purple) flanked by homology regions (HR, yellow) for homology-directed repair. The pBF-gC-guide250 plasmid (Bui et al., 2019) contains expression cassettes for SpCas9 (dark grey), the sgRNA (pink) and the blasticidin deaminase (bsd) resistance marker (brown). Successful gene editing results in the expression of a Cterminally tagged PfHP1-mScarlet protein controlled by the glmS ribozyme element. PCR primer binding sites are indicated by half arrows and were used to confirm successful gene editing. (B) Schematic map of the modified pfhp1 locus after CRISPR/Cas9-based genome editing in the event of donor plasmid concatemer integration into the genome. PCR primer binding sites are indicated by arrows and were used to check for donor plasmid concatemer integration.
(D) PCR on gDNA from 3D7/GFP-PfAP2-HC/PfHP1-mScarlet-glmS and 3D7 wild-type parasites. Primers hp1_F and hp1_R bind to chromosomal sequences outside the HRs and amplify a 3158 bp or 2147 bp fragment from the edited or wild-type pfhp1 locus, respectively. The hp1_F-mScarlet_R and mScarlet_F-hp1_R primer combinations are specific for the edited locus and amplify 1925 bp and 1945 bp fragments, respectively. Primer pD_R binds to the donor plasmid backbone and, when used in combination with primer hp1_F, will amplify a fragment of 3125 bp if a donor plasmid concatemer was integrated into the genome.
(F) PCR on gDNA from 3D7/DDGFP-PfAP2-HC/PfHP1-mScarlet-glmS and 3D7 wild-type parasites. Primer explanations are as in panel D. Primer combination mScarlet_F-hp1_R results in a faint nonspecific product at ~1000 bp in all reactions (panels D and F).
(G) Full sized Western blot of the sections shown in Figure 5C showing PfHP1-mScarlet expression levels in 3D7/GFP-PfAP2-HC/PfHP1-mScarlet-glmS parasites grown in the absence () or presence (+) of GlcN. The membrane was first probed with α-PfHP1 antibodies (top) before inactivation of horseradish peroxidase with 2 mM NaN3, followed by re-probing with the α-GAPDH antibodies (bottom) used as a loading control. Dashed boxes show the sections presented in Figure 5C.

Fluorescence microscopy
Live cell fluorescence imaging was performed as previously described (Witmer et al., 2012) with the minor modification of nuclear staining with Hoechst (Merck) instead of DAPI at a final concentration of 5 µg/ml. IFAs were carried out on methanol-fixed cells using primary antibodies mouse mAb α-GFP (Roche Diagnostics #11814460001) (1:100) and rabbit α-PfHP1 (Brancucci et al., 2014) (1:100). Secondary antibodies Alexa Fluor 488-conjugated α-mouse IgG (Invitrogen #A11001) and Alexa Fluor 568-conjugated α-rabbit IgG (Invitrogen #A11011) were used, each at 1:250 dilution. Nuclei were stained during slide preparation with Vectashield containing DAPI (Vector Laboratories). Images were acquired on a Leica DM 5000B microscope with a Leica DFC 345 FX camera using the Leica application suite (LAS) software. Image processing was carried out using Fiji (Schindelin et al., 2012). For each experiment, all images were acquired and processed with identical settings.
To show the genome-wide colocalization of PfHP1 and PfAP2-HC occupancy and comparison between cell lines the average log2 Chip-over-input ratios at coding genes were calculated using bedtools genomeCoverageBed (v2.27.1) (Quinlan and Hall, 2010) (Dataset S1) and visualised with Excel 2016. Genes with PfHP1 log2 ChIP-over-input ratios of greater than or equal to zero were classified as heterochromatic genes. The position of putative AP2-HC binding sites (i.e. CACACA motifs) has been defined using the position weight matrix of the CACACA motif as described by Campbell et al. (Campbell et al., 2010) and searching the P falciparum genome for matching sequences (fdr<0.05) with the use of the gimme scan tool of GimmeMotifs (van Heeringen and Veenstra, 2011).

Flow cytometry
Synchronous 3D7/DDGFP-PfAP2-HC and 3D7/WT parasites were split at 0-8 hpi to 0.1% parasitaemia and cultured either in the presence of 700 nM Shield-1 (+ Shield-1) or absence of Shield-1 (-Shield-1) during the duration of the multiplication assay. Synchronous 3D7/PfAP2-HC-KO parasites at 0-8 hpi were diluted to 0.1% parasiteamia. After 24 hours (24-32 hpi) parasite DNA was stained with SYBR Green DNA stain (1: 10,000) (Invitrogen #S7563) for 30 min at 37 °C and the fluorescence intensity was measured using a MACS Quant Analyzer 10 (at least 200,000 RBCs were measured per sample) to determine the parasitaemia (day 1). Measurements were repeated on day 3 and day 5 at 24-32 hpi. Data were analysed using the FlowJo_v10.6.1 software. Gating was performed to remove debris smaller than cell size, to include only single measurement events and to separate uninfected from infected RBCs based on the SYBR Green intensity of an uninfected RBC control sample (the gates for 'cells', 'singlets' and 'parasites', respectively, are shown in Figure S2).

Microarray Experiments and Data Analysis
3D7/DDGFP-PfAP2-HC parasites, continuously cultured in the presence of Shield-1, were synchronised with sorbitol to obtain an eight-hour growth window (16-24 hpi) and again in the next generation after RBC invasion at 0-8 hpi. The culture was then split at 8-16 hpi and one half was maintained in the presence of Shield-1 (+Shield-1) and the other half was cultured in the absence of Shield-1 (-Shield-1) to achieve DDGFP-PfAP2-HC depletion. The parasites proceeded through the IDC and were harvested for total RNA extraction in the subsequent generation at the following time points: TP1 (8-16 hpi), TP2 (16-24 hpi), TP3 (24-32 hpi), TP4 (32-40 hpi) and TP5 (40-48 hpi). RNA isolation and cDNA synthesis were performed as previously described (Bozdech et al., 2003). Cy5labelled sample cDNAs were hybridised against a Cy3-labelled cDNA reference pool prepared from 3D7 wild-type parasites (Brancucci et al., 2014). Equal amounts of Cy5-and Cy3-labelled samples