A single master regulator controls asexual cell cycle division patterns in Toxoplasma gondii

All apicomplexan parasites have complex life cycles exhibiting division characterized by a tightly regulated cell cycle control, resulting in the emergence of daughter parasites in possession of a single nucleus and a complete set of organelles. Apicomplexa have evolved efficient and distinctive strategies for intracellular replication where the timing of emergence of the daughter cells, a process termed “budding”, is a decisive element. However, the molecular mechanisms that provide the proper timing of parasite budding remain unknown. Using Toxoplasma gondii as a model Apicomplexa, we identified a master regulator that controls the timing of the budding process. We show that an ApiAP2 transcription factor, TgAP2IX-5, controls cell cycle events downstream of centrosome duplication including organelle division and segregation. TgAP2IX-5 binds to the promoter of hundreds of genes and controls the activation of the budding-specific cell cycle expression program. We show that TgAP2IX-5 regulates the expression of specific transcription factors that are necessary for the completion of the budding cycle. TgAP2IX-5 acts as a licensing factor that ensures that asexual proliferation continues by promoting the inhibition of the differentiation pathway at each round of the cell cycle. Therefore, TgAP2IX-5 is a master regulator that controls both cell cycle and developmental pathways.


INTRODUCTION
Apicomplexa is a phylum consisting of unicellular, obligate, intracellular protozoan parasites of high transmissibility, which includes various human pathogen species such as Plasmodium spp. (causative agent of malaria), Toxoplasma spp. (cause of toxoplasmosis) and Cryptosporidium spp. (cause of cryptosporidiosis).
Apicomplexan parasites trigger disease associated with an uncontrollable expansion of parasite biomass resulting in inflammation and host cell destruction [1]. Although, apicomplexan parasites usually present a sexual cycle within the definitive host, the pathogenesis of these parasites results from the ongoing asexual replication cycles within the host's cells. All Apicomplexa have complex life cycles exhibiting a proliferation characterized by a tightly regulated cell cycle control and resulting in the emergence of daughter parasites with a single nucleus and a complete set of organelles [2]. Apicomplexa have evolved the ability to independently divide their nucleus (nuclear cycle) and produce the parasite body through a process that is termed budding (budding cycle). This allows flexibility in order to produce a suitable number of offspring in a single cell cycle while allowing the parasite to cope with the different host-cell environments [3]. Apicomplexa have evolved efficient and distinctive strategies for intracellular replication [1]. How the division pattern is chosen to ensure parasite expansion during host-cell infection remains unanswered. However, it has been clear that the mode of division is dependent on the timing and coordination of the nuclear and budding cycles [4,5].
Plasmodium falciparum and Toxoplasma gondii represent two modern branch points of the Apicomplexa phylum, a divergence that occurred several hundred million years ago [6]. This divergence has led to changes in the usage of different cell division patterns. For instance, in its intermediate host, Plasmodium falciparum replicates through schizogony, a division pattern which produces a multinucleated intermediate (schizont) where the daughter parasites bud from the periphery at the closure of the division process to produce infective merozoites. By contrast, in its intermediate hosts, Toxoplasma gondii undergoes endodyogeny, a division pattern that consists of the formation of two daughter parasites within the mother cell. In this case, the formation of new daughter cells occurs within the cytoplasm rather than from the periphery [1] and is coordinated to arise simultaneously with nuclear division. In addition to endodyogeny, which is the simplest form of internal budding, T. gondii asexually divides within its definitive host through a division scheme that closely resembles schizogony. Endopolygeny consists of the production of multiple nuclei and a final step of daughter cell formation where parasites bud from the cytoplasm [7]. These three distinct replication patterns (endodyogeny, schizogony and endopolygeny) rely on the coordination of the timing of the budding and nuclear cycles.
Apicomplexa, such as T. gondii which exhibit endodyogeny and endopolygeny, are able to employ several division patterns depending on the environment and host. Such a flexibility suggests overlapping mechanisms in control of the cell cycle independent of the division pattern used.
Apicomplexa show a peculiar cell cycle with a closed mitosis that is divided into three phases (G1, S and M) while the G2 phase is apparently absent [8]. Centrosomes play a central role in controlling the cell cycle in Apicomplexa. They serve as the microtubule organizing center (MTOC) for the mitotic spindle and are key to position the site of daughter cell assembly [5,9].
In T. gondii, the bipartite nature of the centrosome allows an independent control of both the nuclear and budding cycles [9]. Centrosome division happens between the end of G1 and the start of S phase [2] and marks the beginning of the budding cycle. Division and maturation of the centrosome controls the progression and the coordination of the nuclear and budding cycle [5,[9][10][11][12]. Evidence suggests that soon after centrosome division, centrosome maturation through kinases is key to the activation of daughter cell formation [9,11,12]. Cell cycle dependent expression of these kinases may be important to control their activity [11,12].
During the Apicomplexa cell cycle, division and segregation of organelles is a highly ordinated process in order to ensure that each daughter parasite acquires the proper complement of organelles. During T. gondii endodyogeny, Golgi widening and replication is the first step to occur during the G1 phase [13]. This is followed by apicoplast duplication, endoplasmic reticulum division and is concluded with the division of the mitochondria [14]. Specialized secretory organelles such as the rhoptries and micronemes are generated de-novo in a process that begins after nuclear division [14]. This highly controlled process implies a tight regulation of the gene expression during the cell cycle. This was illustrated by the transcriptomic analysis of synchronous cell cycle populations indicating that T. gondii may have adapted a "just in time" mode of expression whereby transcripts and proteins are produced right when their function is needed [15]. This suggests that transcriptional regulation of gene expression may exert a centralized control on both centrosome activation and the expression of proteins needed for daughter cell formation.
However, little is known on the potential regulators that may control this transcriptional switch. Apicomplexan genomes encode a family of putative transcription factors that are characterized by the possession of one or more DNA binding AP2 domains [16]. ApiAP2 transcription factors were shown to control expression profiles during T. gondii differentiation [17][18][19]. ApiAP2s were also shown to cooperatively control the expression of virulence factors [20,21] in T. gondii indicating that this family of transcription factors may control cell cycle dependent expression profiles as also suggested for P. berghei [22]. In this study, we functionally characterize a cell cycle dependent ApiAP2 transcription factor, TgAP2IX-5. We show that it controls cell cycle events downstream of centrosome duplication including organelle division and segregation. TgAP2IX-5 binds to the promoter of hundreds of genes and controls the activation of the S/M-specific cell cycle expression program. Reversible destabilization of the TgAP2IX-5 protein showed that its expression is sufficient for activation of the budding cycle in multinucleated parasites. Therefore, controlled TgAP2IX-5 expression allows for the switching from endodyogeny to endopolygeny division patterns. TgAP2IX-5 acts as a master regulator controlling the budding cycle and therefore, asexual cell cycle division patterns in this parasite.

RESULTS
TgAP2IX-5 is a cell-cycle regulated transcription factor essential for parasite growth and proliferation To find specific regulators of the cell cycle dependent expression program, we identified potential transcriptional regulators whose expression was cell cycle regulated [15]. Among these regulators, we focused on the ApiAP2 family of transcription factors and discovered that the tgap2ix-5 transcript was dynamically expressed with a peak of expression within the S phase (toxodb.org). To confirm that the TgAP2IX-5 protein was also cell cycle regulated, we produced a T. gondii strain that had an epitope-tagged version of this gene at its endogenous locus. Immuno-fluorescence assays using cell-cycle markers demonstrated that TgAP2IX-5 is a nuclear protein mostly expressed during the early S phase ( Figure 1A). TgAP2IX-5 expression was detected at the G1/S transition and early S phase when centrosomes are divided but remained in close proximity of each other (as identified by TgCentrin1; a marker of the outer core of the centrosome, Figure 1A upper panel, Figure S1A). TgAP2IX-5 was detected before centromere division (as identified by TgChromo1; a marker of pericentromeric chromatin [23], Figure 1A middle panel, Figure S1B-C) but not after (late S phase), indicating that TgAP2IX-5 was only present during the G1/S transition and early S phase. TgAP2IX-5 was no longer detected when parasites enter the S/M phase (as indicated by an early budding marker TgISP1, ( Figure 1A, lower panel). Therefore, TgAP2IX-5 protein expression is tightly controlled during the cell cycle.
To identify the biological role of TgAP2IX-5, an inducible knock down (iKD) mutant of TgAP2IX-5 was generated using the AID system which allowed for the conditional depletion of the TgAP2IX-5 protein ( Figure 1B). For this purpose, we produced a strain that presented an AID-HA insert at the 3' end of the TgAP2IX-5 gene using a CrispR/Cas9 strategy ( Figure   S2A). The correct locus of integration of this insert was confirmed by PCR ( Figure S2B). The AID system allows for the conditional depletion of the target protein after addition of auxin (indoleacetic acid) ( Figure 1B). A single detectable band at the expected protein size (251 kDa) was present in absence of auxin in the iKD TgAP2IX-5 strain as verified by Western-Blot ( Figure 1C). Rapid depletion of the TgAP2IX-5 protein (as early as 1 hour) was obtained in iKD TgAP2IX-5 mutant strain grown in media containing auxin ( Figure 1C).
iKD TgAP2IX-5 mutant displays a defect in daughter parasite formation To phenotypically characterize the iKD TgAP2IX-5 mutant, a standard growth assay was carried out. The growth capacity of the iKD TgAP2IX-5 mutant in the presence of auxin was tested and we observed that the iKD TgAP2IX-5 mutant growth ability was drastically decreased in the presence of auxin ( Figure 2A). In fact, growth was completely abrogated in the iKD TgAP2IX-5 mutant in presence of auxin with a calculated mean of 1 parasite per vacuole while the parental strain (in presence and absence of auxin) and the iKD TgAP2IX-5 mutant in absence of auxin grew at a mean of 2.5 parasites per vacuole. This demonstrated a complete blockage of the proliferation capacity of the parasite in the absence of TgAP2IX-5 ( Figure 2A). A plaque assay, that measured the ability of the parasite to grow and invade over a period of seven days confirmed the phenotype with the absence of lysis plaques in the iKD TgAP2IX-5 strain treated with auxin contrary to the control used (parental strain in presence of auxin) where lysis plaques were observed ( Figure S3A). These results suggest that TgAP2IX-5 expression is essential for the parasite's growth and proliferation.
To better assess the growth phenotype, we examined the parasite by IFA using a nucleus marker (TgENO2) and an inner membrane complex (IMC) marker (TgIMC1), a network of flattened vesicles lying beneath the plasma membrane ( Figure 2B). It was apparent that the parasite accumulated nuclei and did not form daughter cells in absence of TgAP2IX-5 while exhibiting normal daughter cell formation in presence of the protein. To confirm the inability of the parasite to form daughter cells in presence of auxin, we performed transmission electron microscopy and observed that in absence of auxin the iKD TgAP2IX-5 mutant was able to produce daughter cells while in presence of auxin the parasite accumulated nuclei associated with lack of apparent daughter cell formation ( Figure 2C). To confirm these observations, we measured the number of parasites per nuclei ( Figure 2D and S3B-C). Depending on the timing during the cell cycle, the parasites exhibit either one nucleus (80 % of the total parasite population) or two nuclei (20 % of the total parasite population) (before or after cytokinesis) for the parental strain (in absence or in presence of auxin) and for the iKD TgAP2IX-5 strain in absence of auxin ( Figure 2D and S3B-C). By contrast, the percentage of parasite exhibiting more than 2 nuclei increased overtime in the iKD TgAP2IX-5 strain in presence of auxin ( Figure 2D and S3B-C). The observation of multi-nucleated parasites in presence of auxin suggests that the iKD TgAP2IX-5 mutant displays a defect in daughter parasite formation but not in the ability of the parasite to multiply and segregate its nuclear material. To assess the capacity of the iKD TgAP2IX-5 mutant to produce daughter cells, we recorded the number of parasite cells that were in the process of producing daughter cells through internal budding ( Figure 2B and E and Figure S4A-B), therefore measuring the parasite's ability to produce daughter cells at an early stage of internal budding. Using two different markers, we were able to show a drastic decrease of the number of parasites undergoing budding in the iKD TgAP2IX-5 strain after 6h of auxin treatment ( Figure 2E and S4A-B). These results indicate that TgAP2IX-5 plays an important role during the early stages of internal budding and is necessary for daughter cell formation.
TgAP2IX-5 is required at a precise time point of the cell cycle.
The T. gondii cell cycle has been described precisely and presents a well-organized time line for the division of the subcellular structures of the parasite [14] establishing the following sequence of organelle duplication and segregation: first the centrosome is duplicated and divided and then the Golgi complex, the apicoplast, the nucleus, the cytoskeleton (e.g. the IMC), the endoplasmic reticulum and eventually the mitochondrion ( Figure 3A). To identify the exact timepoint at which TgAP2IX-5 affects daughter parasite formation, a study of the effect of TgAP2IX-5 on organelle duplication and segregation was carried out. During division, the centrosome (outer core) divides first and then the centromere (as well as the inner core centrosome) follows. We measured the ability of the centrosome and centromere to divide within the iKD TgAP2IX-5 strain using TgCentrin1 and TgChromo1 as a centrosome and centromere marker, respectively. For the parental and the iKD TgAP2IX-5 strains, the centrosome to nucleus and centromere to nucleus ratio were recorded after 6 hours of auxin treatment ( Figure S5A-C). The recorded ratios of centrosome to nucleus and centromere to nucleus are close to 1 in both the presence and absence of auxin ( Figure S5B Since centrosome division remained minimally affected by TgAP2IX-5, a study of the proceeding organelles to divide in the T. gondii organellar cell cycle division timeline was carried out. The Golgi complex was labelled in parasites and the ratio of Golgi to nucleus was calculated ( Figure 3B). We observed no significant difference between the ratios of Golgi to nucleus in the presence or absence of auxin, therefore suggesting that TgAP2IX-5 depletion does not impact Golgi division and segregation ( Figure 3C). We then observed plastid division and segregation ( Figure 3B) and measured the plastid to nucleus ratio. This ratio is significantly lower in the iKD TgAP2IX-5 strain in presence of auxin ( Figure 3D), suggesting that plastid division is blocked in the mutant. The plastid division is a multistep process where it first elongates before the completion of scission and segregation [24]. To determine the exact timepoint at which plastid division is affected, the number of parasites with an elongated plastid was recorded in the presence and absence of TgAP2IX-5. We observed a significantly high number of elongated plastid in the absence of TgAP2IX-5 ( Figure 3E), suggesting a critical role for TgAP2IX-5 after elongation and before plastid division. This short time frame corresponds to TgAP2IX-5 protein cell cycle dependent expression. As expected, the mitochondria division (the last step during the cell cycle, before cytokinesis) is also affected in the iKD TgAP2IX-5 mutant ( Figure S6C), confirming that the blockage during the cell cycle precedes mitochondria division.

TgAP2IX-5 impacts the expression of cell-cycle regulated genes
Since TgAP2IX-5 is a potential transcription factor, we examined the changes in the transcriptome after depletion of the TgAP2IX-5 protein using RNA-seq. Total RNA was purified from tachyzoites of the iKD TgAP2IX-5 strain grown with or without auxin for 6 hours (three biological triplicates). Data analysis using Deseq2 allowed us to identify significant changes in the iKD TgAP2IX-5 transcriptome with an adjusted p-value cutoff of 0.05 and a minimum fold change of 2 ( Figure 4A). We identified more than 600 transcripts that were downregulated and around 300 transcripts (Table S2) that were upregulated in the iKD TgAP2IX-5 mutant when treated with auxin ( Figure 4A). We examined the cell-cycle expression of the downregulated genes and represented their expression using a heat-map ( Figure 4B). We discovered that most of them showed an expression peak during the late S and M phase with a few of them exhibiting a peak during the cytokinesis phase ( Figure 4B) while the upregulated genes showed mostly an inverse expression profile showing an expression peak during the G1 phase ( Figure S7A). In the list of downregulated transcripts, we were struck by the number of annotated genes corresponding to proteins targeted to the IMC and to the apical complex both structures that are the first to appear when the daughters bud within the mother cell. To better assess the potential localization of the proteins that correspond to downregulated transcripts, we used a HyperLopit proteomic dataset that predicts with high confidence the localization of proteins in the parasite [25]. We showed that a high proportion of the downregulated transcripts present in the dataset (331 genes) encode proteins predicted to localize to the IMC or to the apical complex (a total of 30% of the downregulated transcripts present in the HyperLopit dataset; Figure S7B). Moreover, these two localizations are overrepresented (16 and 14 %, respectively) when compared to the whole proteome localization (3% for each; 2487 genes, Figure S7C). Based on this dataset, the downregulated genes represent 64 % of the IMC proteome (52/81) and more than 70 % (45/63) of the predicted apical complex proteome. These results suggest that TgAP2IX-5 may act mainly as an activator of genes whose expression shows a cell cycle regulated profile with a peak at the late S and M phases. These genes represent a high proportion of the IMC and apical complex proteome.
By RNA-seq, we were able to identify that TgAP2IX-5 either activates directly or indirectly the expression of genes. In order to identify what promoters are directly targeted by TgAP2IX-5, we carried out a ChIP-seq analysis. Biological duplicates were produced and processed for sequencing ( Figure S8A). The MACS2 software was used to identify the significant peaks (pvalue < 0.05) that were in intergenic regions close to an annotated gene. This analysis revealed that TgAP2IX-5 directly binds to 696 gene promoters among which key genes are involved in parasite formation ( Figure 4C Figure   4Ciii). Interestingly, we also identified that TgAP2IX-5 was able to bind to the promoters of other ApiAP2 encoding genes (Figure 4Civ and S8C-D). Surprisingly, we found that TgAP2IX-5 was enriched at its own promoter ( Figure S8D, boxed) suggesting a possible negative feedback loop for the control of the TgAP2IX-5 transcript expression. We examined the cell cycle expression of the genes whose promoter was targeted by TgAP2IX-5 and found that a majority of these genes were showing an expression peak during the S and M phase, although a cluster of genes showed a strong expression during C and early G1 phase ( Figure   S8E). These data confirm the ability of TgAP2IX-5 to act as a bona-fide transcription factor by directly binding to promoters of genes encoding essential proteins for the establishment of the daughter cells.
Since RNA-seq does not enable the identification of the genes that are solely directly controlled by TgAP2IX-5, we overlapped the RNA-seq and ChIP-seq dataset. For that, we identified within the differentially regulated genes list whether they were downregulated or upregulated (as initially identified from RNA-seq), and compared this list with the genes identified to be targeted by TgAP2IX-5 from ChIP-seq analysis using the MACS software. Overall, 117 genes were recorded from the overlap of RNA-seq and ChIP-seq data representing a 17% overlap ( Figure 4D). A closer study of the overlap between RNA-seq and ChIP-seq genes identified that 10% of upregulated genes are directly targeted by TgAP2IX-5 whereas 14% of downregulated genes are directly targeted by TgAP2IX-5. We examined the cell cycle regulation of the transcripts directly controlled by TgAP2IX-5 and found that these genes exhibit a cell-cycle regulated expression peak during the late S and early M phases ( Figure 4E).
These results suggest that TgAP2IX-5 directly controls genes expressed during the S phase and beginning of the M phase. Examples of genes that were found to be directly activated by TgAP2IX-5 included a number of IMC proteins such as TgIMC1, TgIMC4, TgIMC3, TgIMC29 and TgGAPM3 (Table S3). Since TgAP2IX-5 was shown to directly activate genes that are mainly involved in daughter parasite formation, TgIMC29 was myc-tagged within the iKD TgAP2IX-5 strain and we recorded the percentage of TgIMC29 present at site of daughter cell formation. We observed a significant decrease in the TgIMC29 presence at the daughter cell budding site in the absence of TgAP2IX-5 ( Figure S9A).
Considering the difference between the RNA-seq and ChIP-seq dataset, we searched the RNAseq dataset for potential regulators that might be directly regulated by TgAP2IX-5 and therefore may in turn perturb the expression of genes not directly targeted by TgAP2IX-5. Eight ApiAP2 transcription factors were downregulated following the depletion of TgAP2IX-5 as measured by RNA-seq ( Figure S9B) and showed an expression peak during the S and M phase ( Figure   S9B). Among these genes, 4 (TgAP2IV-4, TgAP2III-2, TgAP2XII-9 and TgAP2XII-2) had their promoters directly bound by TgAP2IX-5 ( Figure S9B, underlined). Interestingly, much like other genes directly regulated by TgAP2IX-5 ( Figure 4E), these genes showed an expression peak during the late S phase ( Figure S9B). TgAP2IV-4, a known repressor of developmentally regulated genes, is expressed during the S/M phase [18]. These data suggest that TgAP2IX-5 directly activates other TF during the S phase that may in turn control the late S and M expression program.
Interestingly enough, TgAP2IX-5 which was found to be upregulated in the presence of auxin based on RNA-seq, demonstrated a ChIP-seq peak ( Figure S8D) corresponding to its direct targeting by TgAP2IX-5 suggesting that TgAP2IX-5 may be under the direct regulation of a possible negative feedback loop.
Complementation demonstrates TgAP2IX-5 is responsible for the phenotypes observed.
The iKD TgAP2IX-5 mutant was shown to exhibit a multi-nucleated phenotype after depletion of the TgAP2IX-5 protein. The reversibility of this phenotype was studied by generating a complemented strain (iKDc TgAP2IX-5) by inserting a myc-tagged version of the TgAP2IX-5 gene (under the control of its own promoter) into an exogenous locus (uprt; Figure S10A).
The expression and localization of the exogenous TgAP2IX-5-myc in the complemented strain was verified by immunofluorescence and the percentage of positively labelled parasite with myc-tag was compared to the expression of the endogeneous HA-tagged copy. A similar number of parasites (30 % of the asynchronous parasite population) were shown to express the myc-tagged copy in the complemented strain than the iKD TgAP2IX-5 strain ( Figure 5A and S10B). In order to determine whether the multi-nucleated phenotype of iKD TgAP2IX-5 is reversible, the number of nuclei per parasite was recorded in the iKDc TgAP2IX-5 strain in the absence and presence of auxin. We observed that the maximum number of nuclei per parasite did not exceed 2 nuclei per parasite ( Figure 5B). These results demonstrated that the multinucleated phenotype of iKD TgAP2IX-5 is due to the absence of the TgAP2IX-5 protein. In order to study the effect of reversing the phenotype in iKDc TgAP2IX-5 strain on subcellular components of the cell, Golgi: nucleus and plastid: nucleus ratios were recorded. Calculated ratios of around 1:1 were recorded and we therefore inferred that each parasite contains one Golgi as well as one plastid ( Figure 5C-D). These results demonstrate that the TgAP2IX-5 protein was indeed responsible for the phenotypes observed.

TgAP2IX-5 regulates cell cycle pattern flexibility from endodyogeny to endopolygeny
The timing of daughter cell formation is key to define the cell division pattern employed by the parasite at any given time of its life cycle. Since we established that TgAP2IX-5 is the master regulator controlling the production of daughter cells during endodyogeny, we reasoned that controlling the expression of TgAP2IX-5 may be sufficient to switch from one division pattern to another. T. gondii undergoes endopolygeny, during is asexual reproduction in the definitive host, where multiple nuclei are formed before a final phase of internal budding. We created a strain of the iKD TgAP2IX-5 mutant expressing a marker of the IMC (TgIMC3-mCherry) to be able to follow the daughter cell formation using live imaging. We treated this strain with auxin to deplete the TgAP2IX-5 protein and obtain parasites that presented around 4 nuclei per parasite (Over-night treatment with auxin). By time-lapse microscopy, we were able to visualize the fate of these parasites after washing auxin out from cell culture media and inducing the re-expression of the TgAP2IX-5 protein. At T0 we observed enlarged multinucleated parasites. At T1 (43 mins after T0), we started to observe multiple daughter cells emerging from within the initial multinucleated mother parasite labeled with TgIMC3. The emergence of parasites continued throughout T2 (1.5 hrs after T0) and T3 (1.7 hours after T0).
At T4 (3.9 hours after T0), we observed individual parasites each separately labelled with TgIMC3 mCherry that had completely emerged from the mother parasite ( Figure 6A). After removing auxin from the media, the multinucleated parasites where daughter cell budding was inhibited by the absence of TgAP2IX-5, continued their division in a similar way to endopolygeny. Multiple daughter parasites were observed to internally bud within the cytoplasm of the mother cell leading to the release of multiple daughters from an initial multinucleated parasite (Video S1). This forced endopolygeny was due to the re-expression of TgAP2IX-5 in these parasites. These data indicate that expression of TgAP2IX-5 controls the entire budding process and that parasites initially depleted from TgAP2IX-5 were still division competent.
To quantify the ability of the parasite to restart division after an initial depletion of TgAP2IX-5, we designed a protocol producing parasites with around 4 nuclei and then observed the effect of the re-expression of TgAP2IX-5 on the number of nuclei per parasite. For that, we counted the number of nuclei per parasite by immunofluorescence at different time points after auxin washout. While this number decreased with longer auxin treatment ( Figure 6B), we observed that the number of nuclei per parasite decreases as the duration of auxin washout increases ( Figure 6C) indicating that the parasites formed by the "forced" endopolygeny are competent for the next cycles of division. However, we noticed a small percentage of parasites with multiple nuclei after overnight auxin washout, indicating that some parasites did not recover from the original depletion of the TgAP2IX-5 protein. In order to study the effect of auxin removal on daughter parasite formation, we recorded the ability of the initial budding capability of the parasite by labelling TgISP1 and TgIMC1 during auxin washout and compared it with the parasite's budding capability when continuing the auxin treatment. We observed a significant increase in the daughter parasite's budding ability when removing auxin that increased as the duration of auxin washout increased ( Figure 6D-6G and S11). To assess if the parasites were able to divide and segregate the apicoplast after depletion and re-expression of TgAP2IX-5, we monitored the fate of the plastid during auxin washout by recording the ratio of plastid to nucleus. While we observed a steady decrease of the number of plastid per nucleus in presence of continuing auxin treatment ( Figure 6H), we recorded an increase in the ratio of plastid to nucleus per parasite as the duration of auxin washout increased ( Figure 6I) demonstrating that the plastid was competent for replication. These results demonstrate that the timing of TgAP2IX-5 expression is the sole determinant of the creation of daughter cells in the parasite. It may therefore determine the cell division pattern flexibility from endodyogeny to endopolygeny observed in this study.

Discussion
TgAP2IX-5 depletion was found to completely halt the formation of daughter cells (budding cycle) while nuclear division seemed unaffected. When auxin was added at the time of parasite invasion, daughter cell budding was stopped in the first cell cycle, indicating a direct relationship between budding and the expression of TgAP2IX-5. This led to the accumulation of multinucleated parasites without any signs of daughter cell formation. To our knowledge, such a stark phenotype was observed before in a collection of temperature sensitive cell cycle mutants [10], overexpressing a dominant negative version of TgRAb11b [26] and when mutating the specific fibers (SFA2 and 3) that connect the centrosome (outer core) to the forming daughter cells [5]. However, this is the first report demonstrating the importance of transcriptional control in the timing and assembly of daughter cell formation.
After TgAP2IX-5 depletion, the cell cycle stops at a specific timepoint after plastid elongation and before its segregation. This timepoint corresponds to the onset of daughter cell formation.
Surprisingly, the nuclear cycle proceeds while the apicoplast remains elongated and is not segregated. This underlines the absence of a checkpoint to ensure proper apicoplast segregation as was observed for other mutants [27].
Flexibility between the nuclear and budding cycle is thought to be controlled in this parasite by the dual core centrosome with the inner core controlling the nuclear cycle and the outer core controlling the budding cycle [9]. A detailed study of the effect of TgAP2IX-5 on the subcellular structures of the parasite revealed that nuclear and bi-partite centrosome division remain mainly unaffected by TgAP2IX-5 depletion. Given the lack of daughter parasite formation observed after TgAP2IX-5 depletion, it is surprising that the outer core centrosome (as represented by TgCentrin1) would remain unaffected. Other components of the outer core, such as the SFA fibers that were shown to be involved in the emergence of daughter cells [5], but were present at the centrosome after TgAP2IX-5 depletion. Indeed, SFA fiber expression was also unchanged in the mutant as measured by RNA-seq. This indicates that the outer core centrosome functionality is probably intact in the mutant but centrosome activation and maturation is lacking to proceed with the budding cycle. Therefore, it is most likely that depletion of TgAP2IX-5 hinders normal centrosomal activity despite its successful division.
The experiments inducing re-expression of TgAP2IX-5 after its depletion further illustrate the undamaged function of the centrosome in absence of the TF, since daughter cell formation restarted after several cycles of unproductive budding. TgAP2IX-5 is therefore the determinant factor for centrosome activation and the control of the budding cycle.
The phenotype observed after TgAP2IX-5 depletion illustrates the independence of the nuclear and budding cycles in this parasite. Using inducible degradation of the TgAP2IX-5 protein, we have established that the tachyzoites are able to divide by endopolygeny in vitro. These experiments also underline the ability of the centrosome to duplicate and maintain its functionality in absence of daughter cell formation. This is a key aspect of the parasite centrosome biology that allows to accumulate nuclei and then activate the budding cycle as seen for division patterns such as endopolygeny or schizogony. Although tachyzoites divide by endodyogeny, they retained this ability. This simple mechanism of control which allows multiple nuclear cycles to happen before cytokinesis is also dependent on a single protein, We have established that TgAP2IX-5 controls the timing of activation of the centrosome and therefore creation of daughter cells. However, the molecular mechanisms leading to centrosome activation remain unknown. This key event permits the production of daughter cells at the right timing irrespective of the division pattern used. We reasoned that TgAP2IX-5 may directly control the expression of the proteins in charge of centrosome activation.
Although expression of the SFA fibers or other known centrosome markers were unaffected in the mutant, we noticed that the expression of a key centrosomal protein, TgCep530, was directly under the control of TgAP2IX-5 (downregulated in RNA-seq and its promoter bound in ChIP-seq). This protein was shown to be targeted to a centrosomal region situated between the inner and outer core [29]. It is essential for the coordination of the nuclear and budding cycle although budding seemed to occur [29]. However, its depletion leads to the accumulation of outer core centrosome [29], a phenotype that is not observed with TgAP2IX-5 expression.
This protein may therefore have other functions in the maturation or activation of the outer core centrosome that were not identified previously. Another possibility leading to lack of centrosome activation is a plausible downstream effect of non-transcriptional origin, targeting either a kinase or phosphatase with an important role in the centrosome activation cascade. Key serine/threonine kinases such as MAPK-like protein kinases have been identified within T.
gondii and have been shown to ensure proper formation of new daughter parasites [9][10][11]. We found that expression of kinases and phosphatases were downregulated after TgAP2IX-5 depletion, but in our ChIP-seq analysis, TgAP2IX-5 was not found to bind their promoters.
One obvious candidate, that is directly regulated by TgAP2IX-5, was the TgCDC48AP protein but it was shown to be involved in apicoplast protein import [30]. TgAP2IX-5 directly activates the expression of several proteins of unidentified function in T. gondii; these hypothetical proteins may serve as direct centrosome activators.
We have also established that TgAP2IX-5 directly activates key components of the daughter cell scaffold. In particular, it regulates the expression of components to be targeted early to the forming daughter buds, such as TgISP1 [31], TgIMC1, TgIMC3, TgIMC4 and TgIMC10 [32].
Apical cap component, such as TgAC2 and TgAC7 may also be loaded early on the forming buds. Similarly, TgGAPM3 expression is directly dependent on TgAP2IX-5 and its invalidation provokes IMC collapse [33]. Another example is the TgIMC29 protein that has been shown to have an important role in the formation of daughter cells (Bradley P., personal communication). However, TgIMC15, a protein known to be loaded early on the centrosome before reaching the forming buds is not affected by TgAP2IX-5 depletion. TgIMC15 may have a limited role in daughter cell formation or centrosome activation as suggested by the mild phenotype exerted by its mutant [34]. This suggests that TgAP2IX-5 may activate the centrosome but also directly regulates the expression of key components of the daughter cell scaffold that will be needed in a short time frame to proceed with daughter cell formation ( Figure 7).
We observed that RNA-seq and ChIP-seq datasets only partially overlapped. We reasoned that indirect effects on gene expression may be identified using RNA-seq even at a short timing after TgAP2IX-5 depletion (6 hours). The cell-cycle expression profile of the genes that are directly controlled by this TF showed that regulated transcripts are expressed during the S phase before the budding occurs. Among these genes, we identified 4 other ApiAP2 transcription factors (TgAP2III-2, TgAP2IV-4, TgAP2XII-2 and TgAP2XII-9) that were directly controlled by TgAP2IX-5. These TF may be responsible for the expression of the transcripts that were identified as downregulated in RNA-seq and that peak during the late S and M phase. Although no data is available for TgAP2III-2, TgAP2XII-2 and TgAP2XII-9, TgAP2IV-4 was shown to be a repressor of the bradyzoite differentiation expression program [18]. The control of the expression of a repressor of differentiation by a TF that also controls the continuation of cell cycle may provide the missing link between cell cycle and differentiation. It has been shown that the cell cycle is linked to bradyzoite differentiation [35].
In particular, the developmental switch toward the latent bradyzoite is made during S phase and/or mitosis [35]. It seems convenient for the parasite to link the choice of continuing the tachyzoite cell cycle (by starting the budding cycle) with the expression of a repressor of differentiation. Therefore, expression of TgAP2IX-5 may serve as a molecular check point for the choice between proliferation or differentiation. TgAP2IX-5 may act as a licensing factor for this developmental choice at each round of the cell cycle (Figure 7).
We also observed that TgAP2IX-5 may act as a repressor since it was found at the promoter of genes whose transcript was up-regulated after depletion of this protein. Although, the repressor role is limited to a small number of genes, it was striking to see that the TgAP2IX-5 transcript was overexpressed after TgAP2IX-5 depletion. We also found that TgAP2IX-5 was bound to its own promoter indicating that a direct negative feedback loop may be present to limit the expression of TgAP2IX-5. This might explain the short timeframe when TgAP2IX-5 is expressed during the cell cycle since its expression might be self-limiting. In other eukaryotes, TF associate in complex to either activate or repress gene expression [36]. ApiAP2s are known to associate with each other [21] and differential association may impact their activity.
TgAP2IX-5 is therefore an essential regulator during the T. gondii tachyzoite cell cycle.
TgAP2IX-5 controls the activation of the outer core centrosome to induce the budding cycle.
It also controls the expression of key proteins for formation of the daughter cell scaffold, ensuring that the proper continuation of the cell cycle is achieved. TgAP2IX-5 is also a master regulator that controls the expression of the TF that will be further needed for the completion of the budding cycle. It also acts as a licensing factor that ensures that asexual proliferation continues by promoting the inhibition of the differentiation pathway at each round of the cell cycle. TgAP2IX-5 is therefore a master regulator that controls cell cycle and developmental pathways.

Parasite tissue culture and manipulation
Toxoplasma gondii tachyzoites were propagated in vitro in human foreskin fibroblasts (HFF) using Dulbeccos's modified Eagles medium supplemented with 10% fetal calf serum (FCS), 2mM glutamine, and 1% penicillin-streptomycin. Tachyzoites were grown in ventilated tissue culture flasks at 37°C and 5% CO2. Transgenes were introduced by electroporation into tachyzoites of T. gondii strains and stable transformants were selected by culture in the presence of either 25 µg/ml mycophenolic acid (MPA) and 50 µg/ml xanthine, pyrimethamine (2 µM) or FUDR (5 µg/ml). Clonal lines were obtained by limiting dilution. Prior to total RNA, genomic DNA or protein purification, intracellular parasites were purified by sequential syringe passage with 17-gauge and 26-gauge needles and filtration through a 3-µm polycarbonate membrane filter.

Generation of transgenic T. gondii strains
The iKD TgAP2IX-5 line was generated using RH ∆ku80 Tir1 strain as described previously [37]. To produced Myc-tagged TgSFA2 (TGME49_205670) or TgIMC29 (TGME49_243200) parasite lines, a DNA fragment representing the tag and the selection cassette was amplified using the pLIC-Myc-DHFR plasmid as a template. The PCR product (10µg) and a pSAG1::Cas9-U6 targeting the 3' UTR of the respective genes were transfected in the iKD

Organelle Labelling
Parental and iKD TgAP2IX-5 strains were inoculated on HFF cells grown on coverslips in a 24 well-plate for 24 hours. This was then followed by the addition of auxin for 3 hours, 6 hours and 12 hours prior to fixation with 4% PFA. The nucleus was labelled with anti-TgEno2 and the number of nuclei per parasite was counted. For Inner Membrane Complex labelling, parental and iKD TgAP2IX-5 strains were left to grow for 24 hours on HFF grown on coverslips for 24 hours. This was then followed by the addition of auxin for 6 hours.
Intracellular parasites were labelled using anti-TgISP1 and anti-TgIMC1. Components of the centrosome were labelled after leaving parental and iKD TgAP2IX-5 strains to grow overnight followed by auxin treatment for 6 hours and TgCentrin1 and TgChromo1 labelling. Golgi, plastid and mitochondrion were labelled after leaving parental and iKD TgAP2IX-5 strains to grow overnight in auxin with anti-TgCT Sort, anti-TgPlastid and anti-TgTom40, respectively.

Electron Microscopy
HFF monolayer was inoculated with iKD TgAP2IX-5 strain. iKD strain was grown in either normal media or media treated with auxin for 24 hours. Transmission electron microscopy samples were fixed with 1 % glutaraldehyde in 0.1 M sodium cacodylate pH 6.8 buffer, at 4°C overnight. They were post-fixed with 1 % osmium tetroxide and 1.5 % potassium ferricyanide then with 1 % uranyl acetate, both in distilled water at room temperature in the dark, for 1 hour.
After washing, samples were dehydrated with increasing ethanol-concentration solutions.
Samples were finally infiltrated with epoxy resin and cured at 60°C for 24 hours. Sections of 70-80 nm thickness deposited on formvar-coated grids were observed at 80 kV with a Hitachi H7500 TEM (Milexia, France), and images were acquired with a 1 Mpixel digital camera from AMT (Milexia, France).

Western Blotting
Total protein extracts of 2 x 10 6 of parental and iKD TgAP2IX-5 parasites grown for 24 hours in either normal media or media treated with auxin were resuspended in 1X SDS buffer The protein samples were then fractionated on a 6% SDS-polyacrylamide electrophoresis gel and then transferred onto a nitrocellulose membrane. Chemiluminescent detection of bands was carried out by using Super Signal West Femto Maximum Sensitivity Substrate.

Live imaging
The transgenic parasite line iKD TgAP2IX-5 expressing TgIMC3-mCherry was directly inoculated on HFF monolayers grown in an 8 well-chamber. The parasites were left to grow for 7 hours before adding 0.5mM Auxin for 16 hours. Auxin was washed out from the medium by changing normal cell culture media three times. Fluorescent signals were collected sequentially every 13 mins, with an average of 4 lines, a zoom factor (varying between 2 and 4) and resulting in images that were 512x512 pixels in size, and 8 bits in resolution (256 gray levels). The images were treated with ImageJ (NIH). Z-stack acquisitions enabled to visualize the 3D localization of fluorescent signals. Movies were created at a rate of 2 frames per second.

TgAP2IX-5 re-expression experiments
Parasites of the iKD TgAP2IX-5 strain were inoculated on HFF cells grown on coverslips in a 24-well plate for 16 hours in the presence of auxin. Auxin was washed out and parasites were left to grow in auxin-free medium for different durations of time for 3 hours, 6 hours, and 12 hours before fixation with 4% PFA. Parasite nuclei were labelled with TgEno2 and the number of nuclei per parasite were counted. In similar experiments, the parasite's inner membrane complex (IMC) was labelled with TgISP1 and TgIMC1 and the percentage of daughter parasite formation was measured. In a third comparable experiment, plastid was labelled using anti-TgACP and the ratio of plastid to nuclei was measured. Three independent experiments were performed for each experiment.

RNA sample preparation and extraction
RNA samples were prepared by infecting T175 flasks containing HFF cells monolayer with iKD TgAP2IX-5 parasite for 24 hours in either normal media or media containing auxin. T175 flasks containing intracellular parasite grown in normal media were treated with auxin for 6 hours before sample collection and addition of Trizol (Invitrogen). RNA was extracted as per manufacturer instruction. This was then followed by genomic DNA removal and cleaning using the RNase-free DNase I Amplification Grade Kit (Sigma). All RNA samples were assessed for quality using an Agilent 2100 Bioanalyzer. RNA samples with an integrity score greater than or equal to 8 were included in the RNA library preparation. Triplicate (biological replicate) were produced for each conditions.

RNA library preparation
The TruSeq Stranded mRNA Sample Preparation kit (Illumina) was used to prepare the RNA libraries according to the manufacturer's protocol. Library validation was carried out by using DNA high-sensitivity chips passed on an Agilent 2100 Bioanalyzer. Library quantification was carried out by quantitative PCR (12K QuantStudio).

Chromatin Immunoprecipitation and library preparation.
ChIP was performed using a protocol previously described [21]. Briefly, intracellular parasites were grown 40h and fixed for 10 min at room temperature using 1% formaldehyde. DNA was processed by sonication using the Bioruptor device for 10min at 4°C with a 30sec on/off cycle.
Purified DNA (1 ng) was used as a template for library preparation using the Nugen Ovation® Ultralow System V2 kit according to the manufacturer's instructions. Libraries were validated using a Fragment Analyzer and quantified by qPCR (ROCHE LightCycler 480).
Sequencing, pre-processing and dataset cleaning.

Statistical analysis
All data were analyzed with Graph Pad Prism software (San Diego, California, USA).
Differences in the means were assessed by Student's t-test. In all cases, P < 0.05 was considered as significant.