The Plasmodium NOT1-G Paralogue Acts as an Essential Nexus for Sexual Stage Maturation and Parasite Transmission

Productive transmission of malaria parasites hinges upon the execution of key transcriptional and post-transcriptional regulatory events. While much is now known about how specific transcription factors activate or repress sexual commitment programs, far less is known about the production of a preferred mRNA homeostasis following commitment and through the host-to-vector transmission event. Here we show that Plasmodium parasites have taken the unique approach to duplicate the NOT1 scaffold protein of the CAF1/CCR4/Not complex in order to dedicate one paralogue for essential transmission functions. Moreover, this NOT1-G paralogue is central to the sex-specific functions previously associated with its interacting partners, as deletion of not1-g in Plasmodium yoelii leads to a comparable or complete arrest phenotype for both male and female parasites. We show that, consistent with its role in other eukaryotes, PyNOT1-G localizes to cytosolic puncta throughout much of the Plasmodium life cycle. PyNOT1-G is essential to both the complete maturation of male gametes and to the continued development of the fertilized zygote originating from female parasites. Comparative transcriptomics of wild-type and pynot1-g- parasites shows that loss of PyNOT1-G leads to transcript dysregulation preceding and during gametocytogenesis, and shows that PyNOT1-G acts to preserve mRNAs that are critical to sexual and early mosquito stage development. Finally, we demonstrate that the tristetraprolin-binding domain, which acts as the typical organization platform for RNA decay (TTP) and RNA preservation (ELAV/HuR) factors is dispensable for PyNOT1-G’s essential blood stage functions but impacts host-to-vector transmission. Together, we conclude that Plasmodium has created and adapted a NOT1-G paralogue to fulfill the complex transmission requirements of both male and female parasites.

In these processes, NOT1 acts as a central nexus for the overall functions of its complex. Because NOT1 acts as an essential scaffold and yet lacks enzymatic activity of its own, recruitment of the appropriate effector proteins and their regulators is critical for the function of this complex. In addition to the proteins noted above, ELAV-family proteins and tristetraprolin (TTP) are recruited and act as antagonistic, mutually exclusive modulators to functionally toggle the complex's activity between transcript preservation vs. transcript degradation [27]. ELAV-family proteins (which include HuR and CELF proteins) and TTPs (which are C3H1 zinc finger proteins, and include ZFP36, TIS11) are RNA-binding proteins that bind with AU-rich elements in eukaryotes [28]. In addition to its roles in RNA metabolism, recently it was shown that the CAF1/CCR4/NOT complex acts to monitor for low translational efficiency. In this work, NOT5 was found to associate with ribosomes attempting to translate suboptimal codons by binding the ribosomal E-site in its posttranslocation state [29]. Those authors proposed that the association of CAF1/CCR4/NOT and Dhh1 (the orthologue of DDX6, DOZI) with these ribosomes prompts co-translational decay of these mRNAs with an abundance of suboptimal codons via deadenylation and decapping. As the localization of Dhh1 and orthologues to cytosolic processing bodies is associated with its activity, it is notable that the association of NOT1 with Dhh1 acts to inhibit the assembly of processing bodies in yeast [30]. Taken together, NOT1 ties together protein effectors that can degrade or preserve mRNAs and can selectively target mRNAs that are not optimally coded for translation. Due to key differences between Plasmodium and other eukaryotes, such as its extreme AU-rich transcriptome, it is not certain if the same activities are used in the same ways in Plasmodium parasites.
Previous work that aimed to determine whether Plasmodium uses the CAF1/CCR4/NOT complex similarly was first focused on the P. falciparum CAF1 deadenylase, which was identified through the piggyBac transposon screen [31].
While initially thought to be dispensable, further experiments demonstrated that only the conserved N-terminal CAF domain was essential and remained expressed with the piggyBac insertion [21]. We have recently shown that both the specialty CCR4-1 deadenylase and the generalist CAF1 deadenylase are critical for RNA metabolism in gametocytes and affect the synchronicity of male gametocytogenesis and transmission to mosquitoes [21]. In this work, we experimentally identified that the CCR4-1 deadenylase interacts with canonical members of the CAF1/CCR4/NOT complex (NOT2, NOT4, NOT5, CAF40), including both deadenylases (CCR4-1, CAF1), as well as the Plasmodium orthologue of DDX6 (DOZI), CITH, an ELAV-family protein (CELF2/Bruno/HoBo), and more. Moreover, we also identified ribosomal proteins, suggesting that a physical link between CAF1/CCR4/NOT via NOT5 could be present in Plasmodium as has been seen for model eukaryotes [29]. Through label-free quantification, we also identified a highly abundant, but uncharacterized protein annotated simply as "NOT Family Protein." Here we demonstrate that in contrast to other sequenced eukaryotes, Plasmodium and two related parasites have duplicated the not1 gene and has dedicated one of the paralogues for essential transmission-related functions. We have reannotated these paralogues as NOT1 and NOT1-G according to their phenotypes and functions. We show that NOT1-G is responsible first for a slight dampening of the production of gametocytes, and yet then promotes the maturation and fertility of those gametocytes that do commit. PyNOT1-G affects both sexes, as it is essential for the production of male gametes and the continued development of fertilized female gametes in early mosquito stage development. Because the CAF1/CCR4/NOT complex associates with members of the DOZI/CITH/ALBA complex, these phenotypes indicate that PyNOT1-G is a central and essential organizer of gametocyte development.

Parasites of the Aconoidasida Class have Duplicated the not1 Gene
Control of mRNA metabolism is a central feature of eukaryotic gene regulation. The CAF1/CCR4/NOT complex plays many central roles in these processes from the birth to the death of mRNAs, with its best appreciated role being in the decay of the poly(A) tail of mRNAs. Previously, we and others have demonstrated the importance of the CAF1 and CCR4-1 deadenylases of this complex to the sexual development of both rodent-infectious and human-infectious malaria parasites (Plasmodium yoelii and Plasmodium falciparum, respectively) [21,31]. In our previous efforts to experimentally determine the composition of the CAF1/CCR4/NOT complex in Plasmodium through mass spectrometry-based proteomics, we identified a protein annotated only as "NOT Family Protein" (PY17X_1027900) that was found to be associated with CCR4-1 [21] . Bioinformatic analyses using PlasmoDB and BLASTp identified that the protein encoded by PY17X_1027900 mostly closely matched eukaryotic NOT1 proteins, but lacked a bioinformatically predictable tristetraprolin (TTP)-binding domain that is present in the gene currently annotated as NOT1 (PY17X_0945600) (Figure 1) [32,33]. Because no other sequenced eukaryotes to date have an annotated duplication of not1, we used bioinformatics to explore this further. We concluded that these two not1 genes were paralogues due to the high degree of sequence conservation in specific domains at both the DNA and protein levels (DNA: CAF1-binding domain 72% identity, NOT4binding domain 66% identity; Protein: CAF1-binding domain 41% identical/63% positive, CAF40-binding domain 33%/57%, NOT4-binding domain 48%/62%, NOT1 domain 50%/73%). Moreover, these genes are highly conserved, syntenic, and only present across Plasmodium species and two other species of the Aconoidasida class (Genera: Theileria, Babesia), but not in other apicomplexans, model eukaryotes, or humans. Together, this provides evidence that these are paralogs of NOT1 that arose from a gene duplication in the common evolutionary ancestor [34]. As these parasites share some common traits for their transmission strategies between mammals and insects that are not shared with other apicomplexans, we hypothesized that this NOT1 paralogue may have evolved for specialized functions during parasite transmission.

Plasmodium's NOT1 Paralogues Localize to Cytosolic Puncta
Due to the central roles that the CAF1/CCR4/NOT complex plays in RNA metabolism in eukaryotes, this complex is typically found in the nucleus and/or in cytosolic puncta associated with complexes in the spectrum between stress granules and processing bodies (P-bodies) [22,35]. Representative members of this complex (CAF1, CCR4-1) in Plasmodium also have this expression and localization profile and are found predominantly in cytosolic puncta and associate with proteins that are commonly associated with these granules [21]. To determine if the two paralogues of NOT1 are used at different points in the Plasmodium life cycle, we used conventional reverse genetics approaches to append a C-terminal GFP tag to each NOT1 paralogue in P. yoelii (Supplemental Figure 1) and assessed their expression and localization (Figure 2, Supplemental Figure 2). Both NOT1 paralogues were expressed and localized at the same times in the life cycle, and both are predominantly found as cytosolic puncta as was seen for other members of the CAF1/CCR4/NOT complex in Plasmodium yoelii. Of note, the NOT1 paralogues were expressed in both asexual and sexual blood stage parasites with a non-uniform, cytosolic distribution with some puncta visible. Both paralogous proteins were expressed in mosquito stage parasites, with localization shifting from a cytosolic diffuse pattern in oocysts, to a more nucleus-proximal pattern in oocyst sporozoites, and finally to a more apical pattern in salivary gland sporozoites. No expression of either protein was observed in mid-or late-liver stage parasites. Together, these expression and staining patterns match what is commonly seen for members of the CAF1/CCR4/NOT complex in other eukaryotes, as well as what we observed previously in P. yoelii with two other members of this complex: CAF1 and CCR4-1 [21]. Coupled with proteomic data showing their association with CCR4-1, this strongly indicates that these NOT1 paralogues are resident members of the CAF1/CCR4/NOT complex in Plasmodium yoelii.

PyNOT1 (PY17X_1027900) is Critical for Asexual Blood Stage Growth in P. yoelii
NOT1 is essential and acts as a central scaffold to nucleate several RNA-centric functions in eukaryotes [22]. The CAF1/CCR4/NOT complex is most well-known for its role in mRNA deadenylation, but also contributes to transcriptional initiation and elongation, translational repression, mRNA export, and nucleus-based quality control activities. To determine if either Plasmodium NOT1 paralogue bore a greater functional semblance to canonical NOT1 proteins in eukaryotes or had dedicated functions in Plasmodium, we used conventional reverse genetics approaches to delete the coding sequence of each paralogue, which was confirmed by genotyping PCR (Supplemental Figure 3).
In agreement with the critical/essential role of NOT1 to eukaryotic RNA metabolism, our attempts to delete py17x_1027900 were largely unsuccessful. However, in one of two technical duplicates of six independent transfection attempts, we were able to obtain a 100% transgenic population with an extremely slow growth phenotype. It was only possible to produce this parasite line because the population was entirely transgenic, as the presence of even a small number of wild-type parasites with de novo resistance to pyrimethamine would likely have rapidly outgrown these transgenic parasites in the mouse. These data align with results from both PlasmoGEM (P. berghei) and piggyBac (P. falciparum) genetic screens that noted the importance of this gene to asexual blood stage growth [36][37][38]. Due to the severe asexual blood stage defect, which closely aligned with observations of NOT1-related phenotypes in other eukaryotes, we propose that this "NOT Family Protein" is truly the NOT1 protein of Plasmodium parasites. Finally, it is notable that in contrast to other eukaryotes, PyNOT1 can be deleted and indicates that it is not strictly essential.

PyNOT1-G (PY17X_0945600) Dampens Sexual Stage Commitment but is Essential for Sexual Maturation
In stark contrast, the paralogous gene currently annotated as NOT1 (py17x_0945600) was readily deleted and multiple independent clones were obtained by limited dilution cloning (Supplemental Figure 3B). This was unexpected given the essential role of NOT1 in all other eukaryotes and lends further credence to the assignment of the first paralogue described above as the true NOT1 protein of Plasmodium parasites. The ability to disrupt this gene was corroborated by the piggyBac screen, which identified that transposon insertions in this gene could still produce viable parasites [36].
In order to determine if py17x_0945600 plays a specialized role in Plasmodium growth, development, or transmission, mice were infected with 10,000 Py17XNL wild-type parasites or clonal py17x_0945600-null parasites and the progression of the asexual blood stage infection was monitored by daily Giemsa-stained thin blood smears in technical and biological triplicate. No significant asexual blood stage growth defects were observed for py17x_0945600-null parasites until late in the blood stage infection (day 10). At this point, there was a small, but statistically significant decrease in peak parasitemia with a concomitant earlier time to parasite clearance ( Figure 3A).
One explanation for this defect could be due to an overcommitment to gametocytogenesis in the absence of this gene product, as committed male and female parasites are differentiated cells and cannot contribute to increasing the total parasitemia further. Moreover, this asexual blood stage growth defect became pronounced approximately when gametocyte numbers begin to be abundant in the P. yoelii 17XNL strain (~1% parasitemia). To assess this, we compared the total gametocytemia of Py17XNL WT-GFP parasites and py17x_0945600-null parasites following treatment with a 2day course of sulfadiazine to selectively kill asexual blood stage parasites and to leave gametocytes in circulation. By a flow cytometric assay to count GFP-expressing parasites, we observed a slight increase in gametocytogenesis (27-35% increase) in py17x_0945600-null parasites compared to wild-type parasites ( Figure 3B, Supplemental File 2). The increase in gametocytemia paired with the decrease in asexual blood stage parasitemia, may indicate that PY17X_0945600's role in wild-type parasites is to slightly dampen commitment to gametocytogenesis . To assess this effect upon gametocytes further, the same mice that were used to assess the asexual blood stage growth kinetics were also monitored for the presence of male gametes through observation of exflagellation centers ("centers of movement") ( Figure 3C). We observed no exflagellating male gametes throughout the entire course of blood stage infection, indicating that PY17X_0945600 is essential for either gametocyte maturation and/or gametogenesis. Taken together, we have reannotated PY17X_0945600 as PyNOT1-G due to its dual and opposed functions to perhaps dampen gametocyte commitment and yet act in an essential manner to drive male gametocyte maturation.

PyNOT1-G is Essential to Zygote Development
While these data indicate the PyNOT1-G is essential for the complete maturation of male parasites, these experiments did not directly study the impact of PyNOT1-G on female gametocytes or upon host-to-vector transmission. To address these questions, we conducted a genetic cross experiment with Py17XNL wild-type parasites that do not express a fluorescent protein and pynot1-gparasites that express GFPmut2 from a constitutive promoter integrated in the pynot1-g locus. If either male or female pynot1-gparasites were viable for transmission either by self-fertilization or by crossing with wild-type parasites, GFP-expressing oocysts would be evident in the mosquito ( Figure 4A). In contrast, if only wild-type male and female parasites were fertile, only non-fluorescent oocysts would be observed. In three independent transmission experiments, donor mice infected with either Py17XNL wild-type parasites or pynot1-gparasites were used to produce parasites for IV transfer of 1) 10,000 wild-type parasites, 2) 5000 wild-type parasites + 5000 pynot1-gparasites, or 10,000 pynot1-gparasites into experimental mice. Parasitemia increased to 1% with the same timing in all mice and exflagellating male gametes were observed in mice infected with wild-type parasites or a combination of wild-type and pynot1-gparasites, but were not seen with infections with pynot1-gparasites. Mice were anesthetized and mosquitoes were allowed to blood feed for one period of 15 minutes. Seven days post-blood meal, midguts were removed and oocysts were counted and scored by DIC and fluorescence microscopy. No oocysts were detected in mosquitoes that fed upon mice infected with pynot1-gparasites, which is consistent with the absence of exflagellating male pynot1-ggametes ( Figure 3C). However, in mosquitoes that fed upon wild-type parasites or a mixture of wild-type and pynot1-gparasites, only non-fluorescent oocysts were detected thus indicating that both male and female pynot1-gparasites are unable to productively transmit to mosquitoes. Additionally, the effect upon both sexes of pynot1-gparasites was further corroborated by the presence of approximately half of the number of oocysts in mosquitoes that fed upon the mixture of wild-type and pynot1-gparasites as compared to those that fed upon only wild-type parasites. This indicates that PyNOT1-G is an essential driver of maturation for both male and female Plasmodium parasites.
As these defects in pynot1-gparasites parasite transmission could result from a block in early mosquito stage development of female parasites, we used an in vitro genetic cross approach to determine if either sex of pynot1-gparasites could produce zygotes or ookinetes. As controls, we used PyWT-mScarlet (Py1115) and PyWT-GFP (Py489) transgenic lines. As anticipated, we found that wild-type parasites could readily self-cross (GFP+/mScarlet-, GFP-/mScarlet+) and cross (GFP+/mScarlet+) to yield zygotes, retorts and ookinetes ( Figure 4B). In contrast, experiments that crossed PyWT-mScarlet with GFP-expressing pynot1-gparasites could not yield pynot1-gself-crossed zygotes but could yield crossed (GFP+/mScarlet+) zygotes that failed to develop further ( Figure 4B). This female-specific phenotype matches that seen previously for pbdoziand pbcithparasites and indicates that PyNOT1-G may play an important role in preparing the female gametocyte for further development [19,20].

Extensive Transcriptomic Dysregulation in pynot1-g -Schizonts and Gametocytes
As the CAF1/CCR4/NOT complex can contribute toward both mRNA preservation and mRNA decay activities in Plasmodium and other eukaryotes [21,22], we used total comparative RNA-seq (WT vs pynot1-gparasites) to assess differences in transcript abundances in both mixed male and female gametocyte populations and asexual blood stage schizonts depleted of rings, trophozoites, and gametocytes through a subtractive magnetic approach.
First, to further examine the effect of PyNOT1-G in gametocytogenesis, we conducted comparative RNA-seq with Py17XNL wild-type and pynot1-ggametocytes ( Figure 5A, Supplemental Table 1). In addition to the use of DEseq2, we have also imposed a Transcript Integrity Number (TIN) metric to help identify and remove reads due to run-in transcription [39,40]. We observed that pynot1-ggametocytes had lower abundances of 427 transcripts (threshold of ≤ log2 -2.5), including a substantial number of gametocyte-enriched (p25, p28, three apiap2 transcripts, imc-related transcripts, ccp family transcripts, cpw-wpc family transcripts, migs, and actin II) and early mosquito stage (ctrp, warp, psop family transcripts, and perforins) transcripts [4,41]. Notably, the transcript abundance of the NOT1 paralogue (pynot1) was not affected, whereas pyapiap2-g, the master regulator of commitment to gametocytogenesis, was down log2 -1.2 (2.3-fold reduction) in pynot1-gparasites. Enriched GO terms include cellular component terms related to the cytoskeleton, microtubules/actin, and molecular function terms related to binding to the cytoskeleton (Supplemental Table 1). In contrast, few transcripts had increases in abundance (threshold of ≥ log2 2.5), but those that were more abundant are from off-stage genes related to invasive stages and/or sporozoites. Further expansion of the threshold (≥ This transcriptional dysregulation is catastrophic for male gametocytes, as they are unable to complete gametogenesis. However, despite this substantial dysregulation of transcript abundances, it is remarkable that female gametocytes are still permissive to form female gametes, which can be fertilized by competent wild-type male gametes. However, because development arrest occurs at the zygote stage, this indicates that the RNA homeostasis promoted by the PyNOT1-G complex is essential for further development, as was seen for studies of PbDOZI and PbCITH [19,20]. In agreement with this, we compared differentially expressed transcripts from pynot1-gand pbdoziand/or pbcithparasites, and found that 98 of the 115 transcripts that are ≤ log2 -1 in pbdoziand/or pbcithare also reduced in abundance ≤ log2 -2.5 in pynot1-gparasites. By matching the reporting thresholds to ≤ log2 -1, nearly all (105/115) of the transcripts that are lower in abundance in pbdoziand/or pbcithparasites are also in lower abundance in pynot1-gparasites. This highly similar effect may be attributable to these proteins working in concert, as DOZI and CITH are known to interact with the CAF1/CCR4/NOT complex in Plasmodium and other eukaryotes [21,22]. Additionally, many of the same transcripts that are similarly dysregulated in the pyalba4line are also dysregulated in pynot1-gparasites (265 of 438 transcripts are ≤ log2 -1 in both lines) [14]. Finally, 81 transcripts are similarly dysregulated across all four transgenic parasite lines (pbdozi -, pbcith -, pynot1-g -, pyalba4 -), and include p25, lccl family, IMC-related, ap2-o, warp, and psop family gene products (Supplemental Table 2). These results also have extensive overlap with PbDOZIassociated transcripts identified by RIP-chip (409 of 1046 transcripts) [16].
However, as we also determined that PyNOT1-G has an impact on the peak and clearance of asexual blood stage growth, we also assessed changes in transcript abundance in asexual blood stage schizonts that can already be in the process of commitment. As blood stage infections of P. berghei and P. yoelii often contain a significant number of gametocytes (10% of total parasites), we developed a magnet-based subtractive separation approach to remove trophozoites, schizonts, and gametocytes from the sample through four serial passes over magnetized columns. Flow cytometric analyses demonstrated that gametocytes consisted of <1% of the remaining parasites, and that additional passes through the magnetized columns provided no additional enrichment (data not shown). The remaining ring stage parasites were cultured ex vivo until they reached a mature schizont stage based upon Giemsa staining, and then were subjected to total comparative RNA-seq.
Here we observed similar trends in the differential abundance of mRNAs that could be directly or indirectly preserved by PyNOT1-G and its complex in schizonts ( Figure 5B, Supplemental Table 3). Those 293 mRNAs that are less abundant in pynot1-gschizonts include many classic and recently defined early gametocyte transcripts [4,41]. These include many mRNAs that are translationally repressed in P. berghei and/or P. falciparum, such as p25 and p28 (49.5-fold and 74.6fold reductions) as well as several members of the psop, lccl, and ccp/lap gene families. Moreover, other transcripts that were also dysregulated in gametocytes were similarly dysregulated in schizonts, including warp, hsp20, migs, soap, p230p, p48/45, gest, gamete egress protein (gep), male development protein 1 (mdv1), plasmepsin vIII, and actin ii.
Notably, pyapiap2-g is also down log2 -2.3 (~5-fold) in schizonts. Statistically significant GO terms related to these transcripts, and to processes downstream of the schizont stage, include cellular component terms related to the crystalloid, cytoskeleton, microtubules, and dynein, molecular function terms related to motor activity, and biological process terms related to meiosis, microtubules, and movement. Together this indicates that the specific transcripts that are regulated by PyNOT1-G in gametocytes are also affected by it prior to gametocytogenesis, suggesting that the setting of a preferred RNA homeostasis for host-to-vector transmission may occur earlier than anticipated. These multifaceted roles of PyNOT1-G in RNA metabolism match and exceed the roles seen with CCR4-1 (one of the deadenylases of this complex) and the DOZI/CITH/ALBA4 that can associate with it [14,[19][20][21].

Important for Transmission
The diverse functions of the CAF1/CCR4/NOT complex rely upon the associations of diverse regulators with the NOT1 scaffold to direct its activities to specific mRNAs. One such pair of antagonizing RNA-binding regulators, ELAVs (embryonic lethal abnormal vision, Hu family of proteins (HuR)) and tristetraprolin (TTP, C3H1 RNA-binding Zinc Finger Proteins), bind to the same TTP-binding domain on NOT1 proteins and functionally toggle this complex between transcript preservation and transcript degradation respectively [28]. While ELAV-like/HuR proteins can be bioinformatically predicted and have been found associated with the CAF1/CCR4/NOT and DOZI/CITH/ALBA regulatory complexes (CELF1, CELF2/Bruno/HoBo), a Plasmodium TTP orthologue cannot similarly be predicted with confidence [14,21,42]. The only bioinformatically discernable difference between the two NOT1 paralogues was the presence of a conserved tristetraprolin (TTP)-binding domain (AA7-164) on NOT1-G, which is strongly conserved across Plasmodium species (e.g. 100% identical in P. berghei ANKA, 73% identical/88% similar in P. falciparum NF54, 80% identical/92% similar in P. vivax), but that cannot be detected on PyNOT1 or its orthologues in other Plasmodium species (Figure 1).
To this end, we used two complementary approaches to examine the importance of this putative TTP-binding domain to the function of PyNOT1-G. First, we created a transgenic parasite line using the strong, constitutive pbeef1a promoter to overexpress the TTP-binding domain (AA1-199) of PyNOT1-G fused to GFPmut2 (TTPbd::GFP) from a safe harbor locus (pyp230p) (Supplemental Figure 4A). Expression boundaries for this protein variant were chosen based upon the use of domain predictions along with the presence of an enriched region of asparagine residues in P. yoelii that is even more pronounced in P. falciparum, as these regions are often found between functional domains in Plasmodium proteins. We hypothesized that overexpression of TTPbd::GFP would act dominant negatively by binding and sequestering proteins that need to interact with PyNOT1-G to be effective. Expression of TTPbd::GFP was observed by western blotting of the immunoprecipitated protein vs control WT-GFP parasites (Supplemental Figure 4B). Despite this, we did not observe any statistically significant defects in asexual blood stage growth or the ability of these parasites to produce exflagellating male gametes ( Figure 6AB). Moreover, in stark contrast to pynot1-gparasites, the TTPbd::GFP-overexpressing parasites were competent for transmission to mosquitoes (Supplemental Table 4). This indicates that overexpression of PyNOT1-G's TTP-binding domain does not act dominant negatively, perhaps because it could be insufficient to sponge away the relevant factors and/or because this domain and its interacting partners may not be essential to PyNOT1-G's functions in gametocytes.
Second, we created a transgenic parasite lacking the TTP-binding domain of PyNOT1-G by replacing those sequences with GFPmut2 (TTPbd). Clonal lines were isolated and compared to Py17XNL WT-GFP parasites (Supplemental Figure   4C). As with the TTPbd::GFP overexpression line, no statistically significant defect in asexual blood stage growth or in the numbers of exflagellating male gametes was detected ( Figure 6CD). However, statistically significant effects upon the prevalence of mosquito infection were observed (multiple unpaired t-test, no assumptions of standard deviation, p = 0.006763 (WT vs Clone 1, p = 0.030700 (WT vs Clone 2)) but not for the number of oocysts per infected mosquito (Table   1). This indicates that the TTP-binding domain of PyNOT1-G, and by interference proteins that require it to interact with PyNOT1-G, are ultimately dispensable for PyNOT1-G's functions in male gametocytogenesis and/or gametogenesis and female parasite development beyond the zygote stage, but yet it appears to play a significant role in host-to-vector transmission. In contrast, the remaining portions of PyNOT1-G are essential for these functions and thus may provide additional interaction points for these regulatory proteins.

Discussion:
Here we describe the unique duplication and adaptation of a NOT1 paralogue for the transmission requirements of malaria parasites. Due to the high degree of DNA identity and amino acid similarity/identity, these genes likely arose due to a duplication event. It is notable that the same domain organization of two NOT1 paralogues is also present in the closely related apicomplexans Babesia and Theileria, while this duplication is not evident in Toxoplasma or other apicomplexans that have evolved other transmission strategies (Figure 1). To our knowledge, this duplication of NOT1 is unique to the Aconoidisida class. This duplication of NOT1 has been noted in preliminary findings in Plasmodium falciparum asexual blood stage parasites presented in a recent preprint [43]. In that work, the transgenic P. falciparum parasites (and presumably the parental parasites used to generate them) could not generate gametocytes and thus those studies were focused on discerning differences in their functions within the asexual blood stage. Here we also observe effects upon asexual blood stage growth, which were extreme when pynot1 is deleted, but which were more muted when pynot1-g is deleted. In contrast, here we show that the major essential roles of PyNOT1-G are in the development of male and female gametocytes/gametes, and the continuing development of the fertilized zygote in early mosquito stage.
Consistent with their domain architecture and protein interactions, we found that PyNOT1 and PyNOT1-G localize to cytosolic puncta reminiscent of regulatory mRNP granules that are observed in many eukaryotes including Plasmodium.
We and others have also observed this localization for CAF1, CCR4-1, ALBA4, DOZI, and CITH, which also interact with one another in specific contexts and life cycle stages [14,19,21]. Through extensive reverse genetic attempts, we were only able to isolate a completely transgenic parasite population where py17x_1027900 (pynot1) was deleted, as the presence of any wild-type parasites would likely outcompete it rapidly. We now propose the reannotation of this gene as the canonical NOT1 as it matches the hallmarks of NOT1 from other eukaryotes, including an extremely slow growth phenotype when its gene is deleted [22]. It is interesting that pynot1 could be deleted at all given its central and essential role in the activities of the CAF1/CCR4/NOT complex in model eukaryotes. This suggests that the organizing functions provided by the PyNOT1 scaffold can presumably be achieved by subcomplexes, or by proteins that can act in a functionally redundant manner, perhaps by PyNOT1-G although this remains to be experimentally determined.
Here we have focused on the NOT1 paralogue that does not behave like any other eukaryotic NOT1, which we have termed PyNOT1-G due to its essential roles in Plasmodium gametocytes and gametes. The timing of a slight asexual blood stage growth defect in pynot1-gcorrelates with the typical onset of a wave of commitment to gametocytogenesis and thus may correlate with the 27-35% increase in gametocyte numbers observed ( Figure 3B). Furthermore, changes in transcript abundance in asexual blood stage schizonts depleted of gametocytes indicated that several mRNAs relevant to early gametocytogenesis are preserved by PyNOT1-G quite early on, and include transcripts for male development gene 1 (mdv1), actin ii, p25, p28, gest, and several others. The ApiAP2-G specific transcription factor, which acts as a master regulator of commitment to gametocytogenesis was also dysregulated at the transcript level (~5-fold lower abundance in pynot1-gparasites). Transcriptomics of gametocytes also reflect that males are severely affected and is consistent with their inability to complete gametogenesis to produce exflagellating gametes. Many of the transcripts that are dysregulated in schizonts are also dysregulated in gametocytes and in some cases to an even greater extent. In considering female-enriched transcripts, there is a high overlap in the specific mRNAs that are dysregulated by a deletion of pynot1-g, pyalba4, pbdozi, or pbcith (Supplemental Table 2), and a common phenotype in female parasites when pynot1-g, pbdozi, or pbcith is deleted. This indicates that this level of female-enriched transcript dysregulation is permissive to allow gametogenesis and fertilization by a competent male, but it appears to prevent further development of the zygote. This aligns with the classic model of the maternal-to-zygotic transition, where pre-made mRNAs are produced, stored, and translationally repressed/silenced until fertilization occurs [44].
Here we find that PyNOT1-G primarily acts to promote male gametocytogenesis and/or gametogenesis as evidenced by the complete absence of male gametes and a transcriptomic profile that reflects severe dysregulation of male-enriched mRNAs ( Figure 5). Unsurprisingly, male pynot1-gparasites are also unable to transmit to mosquitoes ( Figure 4A).
Moreover, the inability of female parasites to produce viable zygotes that can develop into ookinetes in vitro ( Figure 4B), or oocysts in vivo ( Figure 4A) matches the phenotypes previously observed for deletions of pbdozi and pbcith. Together, it is clear that PyNOT1-G acts upon both sexes and matches or exceeds the phenotypes observed previously for the deletion of genes for its key interaction partners.
Our previous work demonstrated that both PyNOT1 and PyNOT1-G are capable of binding with the PyCCR4-1 deadenylase [21]. As NOT1 acts as a scaffold onto which other factors assemble, it is anticipated that NOT1 and NOT1-G are mutually exclusive members of discrete complexes, although this remains to be formally and robustly demonstrated.
What is clear is that canonical CAF1/CCR4/NOT proteins, as well as possible ELAV/HuR and TTP candidates, all associate with this complex. As the functions of NOT1 in other eukaryotes can be toggled based upon the antagonistic association of ELAVs/HuRs or TTPs, our finding that this predicted domain is not important for NOT1-G functions in gametocyte biology is possibly surprising. One explanation for why Plasmodium may have evolved away from the use of this toggle in this rare NOT1 paralogue could be due to its extreme AT-rich genome and resulting AU-rich transcriptome. As ELAV/HuR and TTP typically bind with AU-rich elements on target mRNAs, this strategy would not provide much regulatory specificity in Plasmodium. Thus, different or additional parameters may be employed to ensure on-target regulation occurs. Consistent with this hypothesis, we identified other zinc finger proteins associated with this complex, including one (PY17X_0417500) with a triple C3H1 ZnF configuration with predicted roles in mRNA decay.
In order to synthesize our current understanding of how proteins implicated in translational repression in Plasmodium blood stages may interact, we propose a composite network based upon formaldehyde crosslinking IP/MS datasets for PbDOZI::GFP, PbCITH::GFP, and PyALBA4::GFP from gametocytes and PyCCR4-1::GFP from schizonts [14,19,21] (Figure   7). Moreover, it is notable that the phenotypes for pynot1-gtransgenic parasites are identical to those for pbdozi -/ pbcith -(female gametocytes can activate into gametes and be fertilized, but cannot develop further) and more extreme than those for pyccr4-1or the truncation of pycaf1 (complete arrest vs. partial defects in either male gametocyte development or gametogenesis). Taken together with proteomic evidence, and consistent with the general role of NOT1 in model eukaryotes and humans, we propose that PyNOT1-G has been evolved to act as a central organizing nexus for these regulatory activities for both sexes of parasites, whereas DOZI/CITH and CCR4-1/CAF1 have sex-specific roles.
Because crosslinking was used in these studies, this permits the detection of direct, short-distance, and long-distance interactions. However, future studies with proximity proteomics approaches could define the spatial and temporal parameters of this complex. Hints as to how these granules may be organized and regulated in Plasmodium are perhaps being found in work with yeast. First, a recent study indicated that members of the CCR4/NOT complex are present in processing bodies but were found at lower concentrations and were more dynamic [45]. Additionally, two other studies indicated that the partitioning of Dhh1, the ortholog of DOZI, is adversely affected by the association with NOT1, and positively affected by association with Pat1 [30,46]. Because no bioinformatically predictable ortholog of Pat1 is present in any Plasmodium species, identifying and understanding the regulators of granule assembly and disassembly will require further experimentation.
Overall, we conclude that Plasmodium and closely related parasites have taken the novel strategy of duplicating NOT1 to allow the dedication of one paralogue for a gametocyte/gamete-specific role in host-to-vector transmission. However, key aspects of how it toggles between its dual natures remain unanswered. As there is an overrepresentation of AU-rich sequences in the Plasmodium transcriptome, the role of ELAV/HuR and TTP RNA-binding proteins that typically bind these AU-rich sequences may be diminished. Accordingly, the TTP-binding domain that typically enables their recruitment to the CAF1/CCR4/NOT complex in model eukaryotes and human cells is dispensable for PyNOT1-G's critical functions. Therefore, it is plausible that Plasmodium has adopted an alternate strategy by adapting the functions of canonical binding partners, or through the recruitment of additional Plasmodium-specific proteins to this unique, transmission-adapted NOT1 paralogue.

Ethics Statement
All animal care strictly followed the Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC)

Experimental Animals
Six-to-eight-week-old female Swiss Webster mice from Envigo were used for all experiments in this work. Anopheles stephensi mosquitoes were reared at 24 C and 70% humidity and were fed with 0.05% w/v PABA-supplemented 10% w/v sugar water.

Production and Confirmation of Plasmodium yoelii Transgenic Parasites
Transgenic Plasmodium yoelii 17XNL strain parasites were produced through conventional reverse genetics approaches using two homology arms that are specific to the targeted gene [47,48]. Homology arms were PCR amplified from wildtype genomic DNA, were combined into a single PCR amplicon by SOE PCR, were inserted into pCR-Blunt for sequencing, and were finally inserted into a pDEF plasmid for use in Plasmodium yoelii parasites. Plasmids used for gene deletion were based upon pSL0444, which replaces the gene sequences with a GFPmut2 expression cassette, a HsDHFR expression cassette, and the plasmid backbone. Plasmids used for appending a C-terminal GFPmut2 tag to a protein were based upon pSL0442, which inserts a the coding sequence for GFPmut2 with the P. berghei DHFR 3'UTR, a HsDHFR expression cassette, and the plasmid backbone into that genomic locus. The PyWT-GFP transgenic line (Py0489) has been described previously [14], and the PyWT-mScarlet line (Py1115) was created in an identical manner but substitutes mScarlet coding sequences for those of GFPmut2. Overexpression of the TTP-binding domain of PyNOT1-G was achieved by inserting these coding sequences (AA1-199) upstream of the GFPmut2 sequences of Py0489. The deletion of the TTPbinding domain of PyNOT1-G was achieved by substitution of these coding sequences with those of GFPmut2 in an Nterminal tag format. Primers used to create all amplicons are provided in Supplemental Table 6, and complete plasmid sequences used to create PyWT-GFP, PyWT-mScarlet, pynot1 -, pynot1-g -, pynot1::gfp, pynot1-g::gfp, overexpressor of the TTP-binding domain, and the deletion of the TTP-binding domain are provided in Supplemental File 1.
Schizonts used for transfections were produced via ex vivo cultures and were purified by an Accudenz discontinuous gradient as previously described [49]. Purified parasites enriched in schizonts were transfected with 10ug linearized plasmid using an Amaxa Nucleofector 2b device using either Lonza T-Cell Solution with Program U-033, or with cytomix using Program T-016. Transgenic parasites were selected by pyrimethamine drug cycling in a parental and a transfer mouse. In some cases, parasites were enriched by FACS and/or cloned by limited dilution approaches. The genotype of parasites was determined by PCR across both homology arms for the targeted genomic locus.

Production and Accudenz Purification of P. yoelii Schizonts and Gametocytes
Schizonts that were depleted of rings, trophozoites, and gametocytes were produced through a subtractive magnetic approach. Blood was collected from mice at approximately 1-2% parasitemia into RPMI1640 media, and was passed over a magnetic column (Miltenyi Biotec, LS Columns) to capture late trophozoites, schizonts, and gametocytes by virtue of the magnetic hemozoin present in them. In contrast, rings and early trophozoites flowed through the column and were collected into fresh media. This depletion process was repeated a total of four times, after which no further depletion of gametocytes in the flow through was observed. Parasites were then subjected to ex vivo culture for 12-14 hours until mature schizonts were visible by Giemsa-stained thin blood smears.
Gametocytes were produced by treatment of the mice at 1% parasitemia with 10 mg/L sulfadiazine (VWR, Cat# AAA12370-30) in their drinking water for two days before exsanguination. The blood was placed in 30 mL of pre-warmed RPMI1640 with 20% v/v FBS to prevent activation of gametocytes and was purified using an Accudenz gradient as previously described [14].
Live Fluorescence and IFA-Based Microscopy PyNOT1 and PyNOT1-G expression in blood stages, oocyst sporozoites, salivary gland sporozoites and liver stages was investigated by an indirect immunofluorescence assay (IFA), and expression in day seven oocysts was observed by live fluorescence microscopy. All samples for IFA were prepared as previously described [47].  [50]). Secondary antibodies used for all stages were Alexa Fluor-conjugated (AF488, AF594) and specific to rabbit or mouse (1:1000, Invitrogen, Cat# A11001, A11005, A11008, A11012). 4′,6diamidino-2-phenylindole (DAPI) was used to stain nucleic acids following washing away unbound secondary antibodies and samples were covered with VectaShield anti-fade reagent (Vector Laboratories, VWR, Cat# 101098-048) and a coverslip, then sealed with nail polish before visualization. Fluorescence and DIC images were taken using a Zeiss fluorescence/phase contrast microscope (Zeiss Axioscope A1 with 8-bit AxioCam ICc1 camera) using a 40X or 100X oil objective and processed by Zen imaging software.

RNA-seq Sample Production, Data Analysis and Interpretation
RNA from P. yoelii asexual blood stage schizonts or gametocytes was prepared using the Qiagen RNeasy kit with two sequential DNaseI on-column digests, with quality control analysis by BioAnalyzer. Samples were used to create barcoded libraries (Illumina TruSeq Stranded mRNA Library) and were sequenced on an Illumina HiSeq 2500 to yield 100 nt long single end reads for each of three biological replicates per sample type. These data were mapped to the P. yoelii reference genome (P. yoelii 17XNL strain, plasmodb.org v50) using hisat2 (version 2.1.0) [51] specifying --rna-strandness R and --max-intronlen 5000 parameters. Coverage files were generated and the mapped data was visualized and manually inspected in Integrative Genomics Viewer (IGV) as a quality control check [52]. Reads that mapped to annotated genes (PlasmoDB.org v50) were counted using featureCounts (version 2.0.0) [53] and specified -s 2 -t exong gene_id parameters. The raw read counts were normalized with DESeq2 (version 1.26.0) [54], which was also used to determine the differentially expressed genes between the sample types. The Transcript Integrity Number (TIN) [40] for each gene was calculated and used to further filter out the differentially expressed genes that do not have coverage evenness across the entire length of the gene as described before [39]. Briefly, TIN was calculated from the transcriptome as built from the annotation files and pseudobam files as generated using Kallisto (version 0.46.1, parameters used: specifying --rf-stranded --single -l 150 -s 20 -b 10) [55]. The samtools depth command was used to determine read depth at each position in the transcript and TIN was calculated as previously described [55]. Custom python scripts are provided as Supplemental File 3. For genes with multiple transcripts, the maximum of the transcript TINs was considered as the integrity value for the gene. Differentially expressed genes with FDR <0.05 were flagged for inspection if the read count is >20 in either of the conditions but TIN is <40 in both conditions. TIN-log2-fold change was calculated, and the flagged genes were resolved and included in the differential expression set if the absolute value of TIN log2-FC is <=1.5. For genes with read count >20 and TIN <40, transcript overlap from neighboring genes were checked using a custom python script (Supplemental File 3) and were flagged if an overlap was detected. A MakeFile describing the complete bioinformatics workflow and other scripts used are also provided in Supplemental File 3. Gene ontology terms (cellular components, molecular functions, and biological processes) were retrieved from PlasmoDB.org (v51). RNA-seq data reported here is available through the GEO depository (Accession #GSE136674).

Measurement of Asexual Growth Kinetics and Male Gametocyte Activation
Cryopreserved blood infected with either P. yoelii wild-type (Py17XNL strain) or transgenic parasites was injected intraperitoneally into Swiss Webster starter mice, and parasitemia was allowed to increase to 1%. This blood was extracted via cardiac puncture and diluted in RPMI to 1,000 parasites per 100ul microliters. One hundred microliters was injected intravenously (IV) into three mice for each of three biological replicates for each parasite line. Parasitemia was measured daily by Giemsa-stained thin blood smears, and male gametocyte activation was monitored daily by counting exflagellation centers ("centers of movement") via wet mount of a drop of blood incubated at room temperature for 8 minutes as previously described [21].

Measurement of Differences in Gametocytemia by Flow Cytometry
Comparisons of gametocyte numbers were made using flow cytometry. Briefly, three experimental mice were infected with 1,000 PyWT-GFP or pynot1-g -(GFP+) parasites by IV injection as described above. The parasitemia was monitored daily, and upon reaching 1% parasitemia, the mice were treated with sulfadiazine for 2 days to selectively kill asexual blood stage parasites. Blood was collected by cardiac puncture and the fraction of GFP+ cells were counted on a Beckman Coulter Astrios Moflo EQ (Brea, CA) in tube mode using uninfected blood and P. yoelii 17XNL wild-type parasite-infected blood as negative controls. All data was analyzed by FlowJo (v10.6.1).

Mosquito Transmission
Swiss Webster (SW) mice were infected with P. yoelii wild-type or transgenic parasites by IP or IV injection as previously described [21]. Mice were screened daily for parasitemia and the presence of male gametocyte activation (centers of movement). On the timepoints indicated, the mice were anesthetized by IP injection of a ketamine/xylazine cocktail and were exposed to mosquitoes for 15 minutes with their positions adjusted every 5 minutes to allow for more even feeding. Mosquito transmission was assessed by dissection of a minimum of 50 midguts on day seven post-blood meal, which were analyzed for the prevalence of infection and oocyst numbers by DIC and fluorescence microscopy.
Genetic Crosses WT (Colorless) vs pynot1-g -(GFP+) Genetic Cross: Blood from mice infected with either Py17XNL wild-type parasites or pynot1-gtransgenic parasites was serially diluted to 5,000 infected RBCs per 50ul, and was used to infect groups of three experimental mice with either 10,000 wild-type parasites, 5,000 wild-type parasites and 5000 pynot1-gtransgenic parasites, or 10,000 pynot1-gtransgenic parasites. Parasitemia and male gametocyte activation were monitored daily, and mice were fed to mosquitoes as above on the day indicated. Midguts were analyzed on day seven post-blood meal on a fluorescence microscope as above, and numbers of oocysts and their fluorescence were recorded (Green or colorless).
PyWT-mScarlet, PyWT-GFP, pynot1-g -(GFP+): Blood from mice infected with either Py17XNL wild-type parasites expressing mScarlet (PyWT-mScarlet, Py1115) or GFPmut2 (PyWT-GFP, Py489) or pynot1-gtransgenic parasites was collected by cardiac puncture and combined for in vitro ookinete cultures as previously described [56,57]. Briefly, blood was collected on the peak day of exflagellation for Py489 and/or Py1115 (as male pynot1-gtransgenic parasites do not complete gametogenesis and do not exflagellate) when the parasitemia was each of the mice was comparable. Blood from each infected mouse was added to 30 ml iRPMI (RPMI with HEPES and L-glutamine; no FBS), and infected red Parasites were assessed by live fluorescence microscopy for mScarlet and/or GFPmut2 expression as described above.

Statistical Analyses
Statistical tests for RNA-seq analyses and comparisons were provided by DEseq2 [54]. The statistical test assessments of GO terms (Odds Ratio, P-value, Benjamini, Bonferroni) were provided by PlasmoDB.org (v51). All other statistical tests were conducted with GraphPad Prism (v9).

Data Availability Statement
RNA-seq datasets, including the raw and processed transcriptomic files (including DESeq2 output files), have been deposited at GEO (Accession # GSE136674). The complete bioinformatics workflow (MakeFile) and the scripts used in RNA-seq data processing are provided as Supplemental File 3.

Acknowledgments
We acknowledge Andy Waters for helpful discussions in the depletion of gametocytes from mixed blood stage parasites,         Table 6 for assessing (A) PyNOT1::GFP and (B) PyNOT1-G::GFP parasites. (A) The PSU 1kb molecular weight (MW) ladder [58] or the NEB 1kb+ molecular weight (MW) ladder (B) flanks all experimental lanes.