ALBA4 modulates its stage-specific interactions and specific mRNA fates during Plasmodium growth and transmission

Summary

Transmission of the malaria parasite occurs in an unpredictable moment, when a mosquito takes a blood meal. Plasmodium has therefore evolved strategies to prepare for transmission, including translationally repressing and protecting mRNAs needed to establish the infection. However, mechanisms underlying these critical controls are not well understood, including whether Plasmodium changes its translationally repressive complexes and mRNA targets in different stages. Efforts to understand this have been stymied by severe technical limitations due to substantial mosquito contamination of samples.

Here using P. yoelii, for the first time we provide a proteomic comparison of a protein complex across asexual blood, sexual, and sporozoite stages, along with a transcriptomic comparison of the mRNAs that are affected in these stages. We find that the Apicomplexan-specific ALBA4 RNA-binding protein acts to regulate development of the parasite’s transmission stages, and that ALBA4 associates with both stage-specific and stage-independent partners to produce opposing mRNA fates. These efforts expand our understanding and ability to interrogate both sexual and sporozoite transmission stages and the molecular preparations they evolved to perpetuate their infectious cycle.

Introduction

Malaria remains a major global health issue, with nearly half a million deaths and 200 million new infections occurring annually (1). This disease is caused by Plasmodium parasites, which are transmitted by female Anopheles mosquitoes. The two transmission points between the human host and mosquito vector are population bottlenecks and thus are prime targets for intervention by vaccines and drugs (2, 3). Thus, extensive efforts to understand targetable processes preceding and during transmission have yielded important, but incomplete advances in parasite control.

Of the two parasite transmission stages, sexual stage gametocytes have been studied in far greater depth due to technical considerations, as they can be produced and purified in larger numbers. In contrast, the sporozoite transmission stage has not been as well characterized, as they must be produced in and manually dissected from mosquitoes. Moreover, the yield (10,000–100,000 sporozoites/mosquito) and the substantial contamination of proteins and nucleic acids from the mosquito and its microbiota have hampered transcriptomic and proteomic studies. We have recently described a purification strategy that overcomes this problem, and yields fully infectious, minimally perturbed sporozoites and greatly reduces mosquito and microbial contaminants (4). This approach has now allowed a total transcriptome and proteome of both human infectious (P. falciparum) and rodent-infectious (P. yoelii) parasites to be conducted (5, 6). Moreover, it allowed the identification and validation of the putative surface proteome of sporozoites, and serendipitously revealed prominent post-translational modifications of key vaccine antigens (CSP, TRAP) that may affect vaccine efficacy (6, 7). These proteomic studies, now conducted across multiple Plasmodium species, have allowed us to define parameters that are predictive of the surface exposure status of Plasmodium proteins (8). Together, these experiments have paved the way to finally achieve targeted proteomics of the sporozoite stage, and to allow a comparison of critical regulatory complexes across transmission stages of the parasite.

We and others have appreciated that both gametocyte and sporozoite stages are critically reliant upon translational repression to prepare for the unpredictable moment of transmission between host and vector (9). Translational repression strategies are employed by many eukaryotes by stabilizing and sequestering transcripts in cytosolic, membrane-less granules called stress granules, processing bodies, or storage granules. These messenger ribonucleoprotein (mRNP) complexes were first recognized to form in response to stress, including nutrient deprivation, heat shock, and mitochondrial compromise (10). The composition of these granule types can vary depending on their intended function. Stress granules classically contain mRNAs, many translational initiation factors (eIF2, eIF3, eIF4A, eIF4B, eIF4E, eIF4G), mRNA-binding proteins (Poly-A Binding Protein), and some components of the ribosome, indicating that these complexes have stalled at various stages of translation (10–12). Processing Bodies (P-Bodies) contain similar factors, but also typically contain mRNAs that are not being translated and may contain mRNA decay factors (XRN1, DCP1, DCP2) (13). Recent evidence indicates that these granules may exchange protein effectors as needed, and that instead of being categorically different, that a spectrum of granules may instead exist that allows different granules to be physically adjacent and yet remain discrete in their functions (11, 14–16). This approach would allow the cell to rapidly respond to the resolution of the stressor to resume translation, or to remain in a poised state until the stress is resolved (17).

In preparation for transmission, Plasmodium parasites use translational repression in coordination with the only specific transcription factor family that is currently appreciated, ApiAP2s (18). ApiAP2s bind to specific DNA sequences and are essential/important for specific developmental programs, including gametocytogenesis (19–22). Ultimately, the gametocyte and sporozoite transmitted stages are loaded with specific mRNAs that are needed following transmission, so that the parasite can proactively prepare and presumably establish the new infection more readily and successfully (23–27). Strikingly, these preserved transcripts are some of the most abundant present in the gametocyte (p25, p28) and sporozoite (UIS3, UIS4), and the extent of their translational repression is striking when compared to other transcripts that are equally abundant (e.g. CSP, TRAP) (5, 24, 28, 29). Translational repression may also be employed to ensure that deleterious, premature translation is avoided, which may include the expression of antigens that could be detected and acted upon by the host’s immune response. The specific translational repression of these transcripts is relieved upon receiving cues that transmission has occurred, which can in part be mimicked in cell culture conditions for both gametocytes and sporozoites (24, 30–32). This process of translational repression, and the resulting proteins from affected transcripts, are critical to the successful infection of the next vector or host. Obviously understanding these molecular players, their functions in both transmission stages, and their weaknesses that can be exploited are of high interest in our attempts to disrupt parasite transmission cycles.

Efforts to understand the mechanisms of translational repression in Plasmodium have defined several key players: DOZI, CITH, PUF2, IK2/UIS1, UIS2 and the ALBA family proteins (11, 23–27, 33–39). As transmission studies and reverse genetics are much more readily conducted with rodent-infectious species, we and others have used P. yoelii and P. berghei to determine their roles in these processes (reviewed in (9, 40)), while limited studies have been done using human infectious parasites (37, 41). In sporozoites, we and others have shown that PUF2 is crucial for maintaining the infectivity of sporozoites while residing in the mosquito salivary gland, and that its genetic disruption leads to premature dedifferentiation of the parasite into a liver stage trophic form while still in the mosquito (24, 27, 34). Presumably, this puf2⁻ parasite can no longer remain in a poised state and loses its preferred RNA homeostasis. As in other eukaryotes, PUF2 traffics to distinct cytosolic granules, but in contrast does not co-localize with known stress granule or P-body markers. Evidence of a role for PUF2’s in translational repression in gametocytes have been reported, where it associates with both p25 and p28 mRNAs and is necessary and sufficient to translationally repress them (41).

However, due to the technical restraints described above, the events and players behind translational repression in Plasmodium parasites have been best characterized in gametocytes. DOZI, the orthologue of DDX6 (human) and DHH1 (yeast), acts as a translational repressor in Plasmodium to greatly inhibit the degradation and translation of specific transcripts (23, 25, 26). Genetic deletion of DOZI leads to a complete developmental block at the zygote stage following fertilization. Similarly, the genetic loss of CITH (a homolog of CAR-I and Trailer Hitch), which is orthologous to SCD6 and LSM14A proteins, also leads to a developmental block slightly after that caused by the disruption of DOZI (26). Both of these proteins associate with one another in a cytosolic mRNP in Plasmodium berghei, with their Trypanosome orthologues also being found at the nuclear periphery (42–44). Interestingly, while PUF2 has also been shown to be an important translational repressor in gametocytes (41), it does not associate with DOZI or CITH. Further work is needed to clarify differences in the roles of PUF2 in human- and rodent-infectious parasites.

Targeted mass spectrometry of DOZI/CITH mRNPs revealed the presence of the ALBA family of proteins as consistent members of the complex in gametocytes (26). The involvement of ALBA proteins with DOZI and CITH appears to be conserved in other protozoans as well, including Leishmania and Trypansoma species (43–45). ALBA proteins are evolutionarily conserved from Archaea and consistently have been found to bind to nucleic acids. Structurally, the ALBA domain resembles the IF3 C-terminal domain, which binds to the small subunit of the prokaryotic ribosome (46). Together, this strategy provides a straightforward mechanism for ALBA proteins to tether nucleic acids to the ribosome. This is further corroborated by recent evidence that the yeast ortholog of DOZI (DHH1) directly binds the ribosome (47).

Plasmodium spp. contain at least four ALBA-domain containing proteins, and have been shown to bind both DNA and RNA (36, 37, 48). However, they are found most prominently in cytosolic granules in asexual blood stages that are present in all locations from the nuclear envelope to the cellular periphery (33, 35, 37). Recently, ALBA1 has been shown to be a specific translational repressor during the asexual blood stage of P. falciparum (37). Of these ALBA proteins, ALBA4 is specific to the Apicomplexan lineage and its Chromerid ancestor (40–50% identity / 60–65% similarity) (36, 49). This conservation may indicate that ALBA4 has retained an important, evolved role for this specific lineage which is based upon ALBA’s ancestral functions.

However, our understanding of how ALBA proteins participate in the development and transmission of Plasmodium parasites remains limited to its roles in blood stages, and many important questions remain. For instance, is the DOZI/CITH/ALBA (DCA) complex used to impose translational repression across the life cycle? Does this complex maintain the same composition across stages, or does it recruit different effector proteins and RNA-binding proteins to influence different transcripts? Additionally, does the Apicomplexan-specific ALBA protein, ALBA4, contribute specific functions to the growth and transmission of the parasite? These questions have remained largely intractable due to our inability to properly interrogate sporozoites, which are difficult to produce in large numbers and until recently could not be substantially purified while remaining in an unperturbed state.

Here, we address these important questions not only in asexual and sexual blood stages, but also for the first time in sporozoites. Using rodent-infectious Plasmodium yoelii parasites, we find that ALBA4 functions in both of the transmission stages by first suppressing the activation of male gametocytes, and then allowing for the semi-synchronous development of sporozoites. This is in contrast to the functions of other members of its complex, which specifically affect female gametocytes. Comparative RNA-seq further shows that ALBA4 contributes to opposing modes-of-action upon mRNAs across the life cycle, and may result in the observed phenotypes by affecting transcripts implicated in these processes. Finally, targeted proteomics reveals that ALBA4 accomplishes these functions through both stage-independent and stage-specific interactions across asexual blood stage schizonts, gametocytes, and sporozoites.

Taken together, these studies reveal the dynamics of an mRNP complex across key points of the Plasmodium life cycle, which enables the parasite to adjust its RNA homeostasis to promote its growth and transmission. Excitingly, we now conclude that experimental questions of sporozoite functions dependent upon specific protein complexes are now accessible, and permit a greater understanding of the commonalities and differences between the transmitted stages of the malaria parasite.

Results

PyALBA4 dampens the activation of male gametocytes

Plasmodium parasites have evolved specific RNA-binding proteins to impose translational repression, which they have become critically reliant upon for their gametocyte and sporozoite transmission stages. Because of this, we sought to determine the importance of the Apicomplexan-specific ALBA4 protein, which has been shown to associate with DOZI and CITH in gametocytes (26). To assess this, we generated a clonal knockout parasite line by double homologous recombination in the rodent-infectious Plasmodium yoelii 17XNL strain (pyalba4⁻). Replacement of the PyALBA4 open reading frame with both a drug resistance cassette (HsDHFR) and GFP expression cassette (GFPmut2) enabled rapid selection and isolation by limited dilution cloning, which was confirmed by genotyping PCR (Supp Figure 1). In contrast to previous reports of ALBA proteins (including PbALBA4), no significant growth defect was observed in asexual blood stage (37, 50) when compared to a wild-type line with a GFPmut2 expression cassette integrated in the p230p dispensable locus (WT-GFP) (Supp Figure 2).

However, in sexual stage gametocytes, we observed that genetic disruption of pyalba4 causes a significant, two-fold increase in the number of activated male gametes (Fig 1A). As this could arise by PyALBA4 affecting gametocytogenesis to either increase gametocyte numbers or affect the male-female sex ratio, we compared gametocytemia of WT-GFP and pyalba4⁻ parasite lines by Giemsa-stained thin blood smears. As rodent-infectious Plasmodium species produce gametocytes that are sexual dimorphic for much of their development, we assigned parasites into three categories as has been done for P. berghei previously: sexual dimorphic (immature), mature males, and mature females (scoring key provided in Supp Figure 3) (51). Based upon this, we did not observe differences in total gametocytemia, sex ratio, or mature-immature ratio (Supp Table 1), and therefore conclude the PyALBA4 contributes more directly to the activation program of male gametocytes into gametes. In addition, there were no significant differences in the distribution of the time to gametocyte activation between WT-GFP and alba4⁻ parasite lines when observing between 5–23 minutes post-blood sampling. This phenotype of an increased number of activatable male gametocytes is in stark contrast to the effects of other members of the DOZI/CITH/ALBA complex, where only effects upon female gametocytes have been noted.

Figure 1. pyalba4⁻ parasites are adversely affected in gametocyte and sporozoite stages.

The entire life cycle was characterized and compared to a WT-GFP line known to behave as WT. A) Exflagellating male gametes were identified by centers of movement (COM) / exflagellation centers and quantified per 40X field of vision. pyalba4⁻ parasites exhibit a 2–3-fold increase in the number of COMs compared to a WT-GFP control. To quantify this, 10 fields-of-view per parasite line per biological replicate were analyzed, and six biological replicates were used. B) The fraction of total sporozoites found in the salivary glands was determined by performing double dissections of the midguts and salivary glands of each mosquito. In WT-GFP parasites, a majority of the sporozoites are typical found in the salivary glands by Day 14 post blood meal. In contrast, most pyalba4⁻ sporozoites remain associated with the midgut of the mosquito throughout the infection of the mosquito. Three biological replicates per parasite line were assessed, with 10–25 mosquitoes per parasite line per time point. Student’s t-test, *p-value < 0.05, **p-value < 0.01.

PyALBA4 promotes the semi-synchronous development of sporozoites

We and others have noted that translational repression also occurs in sporozoites, but the protein regulators of this stage are largely unknown. Genetic deletion of the pypuf2 translational repressor led to the dysregulation of many transcripts related to RNA metabolism in sporozoites, including pyalba4 (24). Because of this, we also determined whether PyALBA4 plays an important role in sporozoite biology by comparing the pyalba4⁻ and WT-GFP parasite lines in mosquito and liver stages. While we observed no significant differences in the prevalence of mosquito infection, oocyst numbers, nor in the total number of sporozoites per infected mosquito (the sum of midgut and salivary gland sporozoites) (Supp Table 2, Supp Fig 4, Supp Fig 5), we observed a significant effect upon the semi-synchronous egress of sporozoites from oocysts in the midgut and their arrival in the salivary gland (Fig 1B). This coordinated development typically leads to the arrival of most of the sporozoites at the salivary gland 14 days post-blood meal for P. yoelii, while P. berghei requires several extra days to do so. Instead, pyalba4⁻ parasites continue to egress from the oocysts over time and are capable of invading the salivary gland and are highly infectious to mice (Supp Table 3). These data indicate that PyALBA4 participates in the coordination of sporozoite development while in the oocyst.

PyALBA4 localizes to cytosolic puncta throughout the Plasmodium life cycle

As ALBA proteins across species are known to bind to RNA and to localize to cytosolic granules, we hypothesized that PyALBA4’s effects to suppress male gamete activation and to coordinate sporozoite development would similarly require trafficking to granules throughout the Plasmodium life cycle. To this end, we produced a transgenic line where GFPmut2 was appended to the C-terminus of PyALBA4 by double homologous recombination into its native locus. This genetic modification was confirmed by genotyping PCR (Supp Fig 6), as well as by live fluorescence and immunofluorescence assays (Fig 2, Supp Fig 7, Supp Fig 8). Fusion of GFPmut2 to PyALBA4 did not result in any detectable morphological or phenotypical effects compared to the WT-GFP parasite line at any point of the life cycle.

Figure 2. PyALBA4 is expressed throughout the life cycle and is found in multiple subcellular locations.

PyALBA4::GFP localizes to punctate cytoplasmic foci, the nuclear periphery, and the cell periphery. A) In asexual blood stage, PyALBA4::GFP expression is exclusively nuclear-adjacent in rings, but includes cytoplasmic punctate foci in schizonts, with an intermediate expression pattern in trophozoites. In trophozoites, we also see localization along the plasma membrane. Asexual stage IFAs were performed with anti-GFP and counter-stained with anti-ACP. Nuclei were stained with DAPI. Scale bars are 5 microns. B) In male and female gametocytes expression is cytoplasmic and diffuse, with areas of greater intensity. In female gametocytes these areas of greater intensity overlap with DDX6 (DOZI) expression. Female gametocyte IFAs were performed with anti-GFP and counter-stained with anti-DDX6, while males were counter-stained with anti-alpha tubulin II. Nuclei were stained with DAPI. Scale bars are 5 microns. C) In mosquito stages, both oocyst- and salivary gland sporozoites exhibit punctate cytoplasmic expression, as confirmed by both immunofluorescence assay and live fluorescence. Both stages were stained with anti-GFP, counterstained with anti-CSP, and nuclei were stained with DAPI for IFAs. Scale bars are 5 microns.

As has been seen previously for PfALBA4, we observed that PyALBA4 is expressed throughout asexual blood stage development (Fig 2A). PyALBA4 traffics to cytosolic puncta in ring, trophozoite, and schizont stages, and can be found as nuclear adjacent foci, near the cell periphery, and at all points between. However, the expression and localization of ALBA proteins beyond asexual blood stage development has not been assessed to date. In sexual blood stage parasites, we observe that PyALBA4 is abundantly expressed, and can be found both diffusely and in puncta in the cytoplasm in both male and female gametocytes (Fig 2B). As some of its known binding partners in gametocytes are present in cytosolic foci, this may indicate that PyALBA4 also may participate in other regulatory processes and complexes beyond its appreciated role in DOZI/CITH/ALBA complexes.

In mosquito stages, PyALBA4 is first expressed diffusely throughout early oocysts (Day 3) but then much of it is directed to puncta within developing sporozoites (Supp Fig 7), with the remainder persisting within sporoblasts/cytoplasmic islands. Sporozoites liberated from the oocyst (Day 10) retain only a few puncta of PyALBA4, which by IFA and live fluorescence are largely nuclear adjacent (Fig 2C). Salivary gland sporozoites (Day 14) contain far more PyALBA4 granules, which extend from near the nucleus out to both ends of the parasite (Fig 2C). The timing of the appearance of PyALBA4 puncta within the sporozoite matches the timing of the phenotype, and thus they may be functionally linked. Finally, PyALBA4 follows a similar pattern in liver stage parasites, shifting from a largely diffuse expression pattern in mid-liver stage (24 hours) to increasingly smaller puncta during late (48 hours) and very late (52 hours) liver stage (Supp Fig 8, Supp Table 4). These patterns match the development and packaging of exoerythrocytic merozoites during schizogony for release into the vasculature to infect RBCs.

PyALBA4 affects mRNA regulation in opposing manners across the Plasmodium life cycle

As PyALBA4 largely is found in cytosolic granules reminiscent of stress granules, p-bodies, and storage granules, we hypothesized that PyALBA4 would play a consistent role in mRNA homeostasis throughout the parasite life cycle. To assess this and whether effects upon specific transcripts could explain the observed phenotypes, we conducted comparative RNA-seq with WT-GFP and pyalba4⁻ parasites at the gametocyte and oocyst sporozoite stages. Moreover, as PyALBA4 is found in prominent cytosolic puncta adjacent to the nuclear envelope and at the cellular periphery in asexual blood stage schizonts, we also conducted comparative RNA-seq in this stage as well.

In order to ensure that sufficient reads could be mapped to the parasite’s reference genome to conduct differential expression analyses, 20–30 million single end, stranded reads were used per sample. This was in part necessary due to contaminating RNAs from the mouse and mosquito despite extensive purification efforts. Moreover, we have limited our interpretations of these data to those transcripts that are sufficiently abundant to provide robust comparisons, especially for gametocytes and sporozoites where a lower fraction of reads mapped to the P. yoelii 17XNL reference genome (2–10% and 1–5%, respectively). In all cases, the pyalba4 transcript is among the most affected (Figure 3ABC) when the open reading frame is replaced, indicating that the gene was disrupted. Surprisingly, these data show that PyALBA4 affects mRNAs in opposing ways and in a stage-dependent manner (Figure 3ABC), and that the mRNAs affected are largely stage specific as well (Figure 3D).

Figure 3. PyALBA4 affects transcript levels in both stage-specific and stage-independent manners.

Comparative total RNA-seq was performed with WT-GFP and pyalba4⁻ parasite lines from purified gametocytes (A), schizonts (B), and oocyst sporozoites (C). Transcript abundance increases (green) and decreases (red) greater than a 2-fold change (dotted horizontal lines, y-axis) are indicated, with genes arranged sequentially by their gene ID number (PlasmoDB, v27) on the x-axis. Arrows indicate pyalba4 transcript. Three (schizont, gametocyte) or two (oocyst sporozoite) biological replicates were performed. A) In gametocytes, transcript abundance decreases included transcripts that are transcribed early in gametocytes, and those that encode for proteins important for ookinete development. B) In schizonts, transcript abundance increases belonged largely to transcripts that encode for proteins important for invasion and cytoskeletal processes. C) In oocyst sporozoites, there are many more transcripts where their abundances increase or decreases, with transcripts encoding RNA-binding proteins being overrepresented. D) Across the stages investigated, there are very few transcripts that are affected in multiple stages, indicating that PyALBA4 interacts with mRNAs in both a stage-independent and a stage-specific manner.

In gametocytes, we observed that PyALBA4 plays a protective role, with most transcripts that are affected by the loss of pyalba4 decreasing in abundance (Fig 3A). The majority of transcripts that are affected at least 2-fold in abundance (309 of 535) encode for currently uncharacterized proteins, and while they may be important for processes related to PyALBA4’s functions, we have not pursued further analysis due to lack of experimental information about them. Those that remain with greater than a 2-fold change in abundance (36 increase, Supp Table 5; 190 decrease, Supp Table 6) include many transcripts that encode for functions important to the gametocyte (calcium-dependent activities; inner membrane complex proteins: GAPM2, MAP45, IMC1b-IMC1k, GAP50) or to early mosquito stage infections (LCCL-domain containing proteins: CCp1, CCp2, CCp4, LAP5; CPW-WPC protein family members; WARP; PSOP proteins 6, 13, and 20). Together, these data further support a role for PyALBA4 in the protection and translational repression of specific mRNAs integral to parasite transmission and the establishment of its infection of the mosquito. This is further supported when these datasets are compared to transcripts found to associate with DOZI/CITH by RIP-chip, with 108 of 226 transcripts overlapping (Supp Fig 9A, Supp Table 7) (23).

We also conducted the first comparative RNA-seq of the oocyst sporozoite stage, where PyALBA4 participates in the semi-synchronous development of sporozoites (Fig 1B). Here, loss of pyalba4 contributes to a mixed effect but upon twice as many mRNAs (1131 transcripts with > 2-fold change in abundance), with some notable increases and decreases in transcript abundances compared to the WT-GFP control line that are distinct from other parasite stages (Figure 3CD, Supp Table 8). In pyalba4⁻ sporozoites, two gene ontology (GO) terms were significantly attributed to transcripts that increased in abundance: RNA-binding proteins and translation factor activity (Supp Table 9). This is congruent with the parasite using PyALBA4 as a translational repressor, and/or attempting to compensate for the loss of PyALBA4 by increasing the expression of other RNA-binding proteins and its capacity to translate proteins generally. In addition, there are increases in transcript abundance of several MCM proteins, which were recently identified to be regulators of stress granules in addition to their classic role in DNA replication (42). Of those genes that experienced a decrease in transcript abundance in pyalba4⁻ sporozoites were those associated with heat shock protein binding (Supp Table 10). Again, several of these factors act as chaperones and have been implicated in stress granule dynamics (e.g. DNAJ, DNAJ-like proteins, MCM, and CCT proteins). Lastly, we did not note any significant difference in transcript abundance for the few known translational repressors of sporozoites, namely ik2/uis1, uis2, sap1/slarp, and puf2. This indicates that ALBA4 is not acting at a higher level to regulate the regulators.

Finally, we also assessed the effect of PyALBA4 upon transcripts in asexual blood stage schizonts due to its prominent expression and localization patterns as puncta throughout the cytoplasm (Figure 3BD). In contrast to its roles in gametocyte and sporozoite stages, PyALBA4 exerts a pronounced effect in asexual blood stage schizonts to decrease the abundance of specific transcripts that are largely distinct from the transcripts affected in other stages (Supp Table 11, in alba4⁻ parasites where ALBA4 is missing, only four transcripts decrease in abundance >2-fold while 251 increase in abundance >2-fold). Transcripts that were affected included those that encode functions related to microtubule motor activity and binding (Supp Table 12), including several dynein heavy and light chain proteins, actin-related proteins, and kinesins. These effects are consistent with changes observed when another ALBA family member, PfALBA1, was overexpressed in asexual blood stages and may reflect invasion-related roles for ALBA family proteins (Supp Figure 9B) (37). In addition, the abundance of transcripts encoding RNA-binding proteins (NOB1, Musashi, PUF1) were also affected, indicating that the parasite may be attempting to compensate for the loss of PyALBA4’s RNA-binding activities, or may provide another layer of regulation over proteins that are important at future points in the life cycle. This is further supported by effects upon transcripts important to gametocytes (MDV1, GEST) and ookinetes (CelTOS, PSOP2, PSOP17).

Taken together, we observe that PyALBA4 plays opposing and large scale roles in regulation distinct transcripts that are relevant to specific stages of development. This is also evident from how little overlap there is between the differentially affected transcripts in gametocytes, sporozoites, and asexual blood stage schizonts, with only six transcripts being affected across all three stages (Figure 3D, Supp Table 13). Based upon this, we propose that PyALBA4 is used to specifically tune relevant transcripts in a stage-specific manner.

The composition of the PyALBA4 complex is dynamic, and includes stage-specific as well as stage-independent partners

Because PyALBA4 is expressed throughout the parasite life cycle, has two prominent phenotypes during the two transmission stages, and can specifically affect the abundance of transcripts in a stage-specific manner, we sought to determine mechanisms underlying how PyALBA4 functions differently in these key stages. We hypothesized that PyALBA4 can exert this control by adjusting the effector proteins that associate with it across the Plasmodium life cycle. To this end, we have conducted comparative, targeted proteomics of the PyALBA4 complex in asexual blood stage schizonts, sexual stage gametocytes, and excitingly also in sporozoites (Figure 4). This has allowed a first glimpse into how a critical regulatory process, translational repression, is either similarly or differently employed in both transmission stages and beyond.

Figure 4. PyALBA4::GFP binds proteins in both stage-specific and stage-independent manners.

A comparison of the proteins that associate with PyALBA4 in asexual blood stage schizonts, gametocytes and oocyst sporozoites demonstrates that stage specific and stage independent interactions occur. Selected interacting partners from paired stages (blood stages, transmissions stages, asexual stages) are indicated in offset boxes.

For these experiments, we have immunoprecipitated GFPmut2 (control from WT-GFP parasite line) or PyALBA4::GFP and its complex using anti-GFP antibodies. In all three stages, this was done with formaldehyde-crosslinked material to stabilize weak/transient interactions and to enable more stringent washing conditions to reduce contaminants. Immunoprecipitations were confirmed by western blotting prior to mass spectroscopic analyses (Supp Figures 10 and 11), and the resulting mass spectra (nano LC/MS/MS) were analyzed using the Trans-Proteomic Pipeline (TPP, Institute for Systems Biology) to identify peptides and to infer protein identities with a stringent <1% false discovery rate (FDR). Comparison across three biological replicates were made by use of the SAINT algorithm, which compares normalized spectral count and protein length data and provides a mixture model to estimate FDR with a unitless score (52). This also allows the exclusion of proteins that may be captured due to interactions with GFPmut2, the antibody, or the beads themselves. Following this analysis, we considered protein identifications with SAINT scores below 0.1 as being highly stringent hits (6), and those between 0.1 and 0.35 as being modestly stringent in these datasets. These two stringent thresholds will likely result in the underreporting of some true positives that are present in lower abundances or are not present in all replicates, but we report these hits in two tiers of confidence levels to provide a more comprehensive view of this complex. Confident inclusion of all of these proteins will require future experimental validation through reciprocal pull-downs or use of complementary identification methods.

First, using gametocytes enriched by sulfadiazine selection and Accudenz purification, we confirmed that PyALBA4 indeed associates with members of the DOZI and CITH complexes (Table 1, complete list provided in Supp Table 14) (26). These include DOZI, CITH, PyALBA1-4, PABP, eukaryotic initiation factors eIF4E and eIF4A, and ribosomal proteins. Additionally, factors that associate with RNA and shuttle between the nucleus and cytosol were also identified, including RNA-binding proteins (Bruno/CELF2, CELF4 and Mushashi), stress granule components (GBP2), and splicing factors (SR1, PREBP) (Supp Table 15) (36, 53, 54). The presence of these factors is consistent with the localization of PyALBA4 in both cytosolic granules and adjacent to the nucleus (Fig 2).

Table 1.

PyALBA4 Interactions of Special Note in Gametocytes.

Gene ID	Product Description	# Spectra	AvgP	Found in PbDOZI	Found in PbCITH
PY17X_1366000	ALBA4	55|116|101	--	Y	Y
Stress granule components
PY17X_1425300	ALBA1	74|45|67	1.00	Y	Y
PY17X_1364900	ALBA2	23|24|24	1.00	Y	Y
PY17X_1207600	ALBA3	32|20|27	1.00	Y	Y
PY17X_1441700	PABP	22|38|15	1.00	Y	Y
PY17X_1304900	CITH	27|19|11	1.00	Y	Y
PY17X_1208200	GBP2*	29|19|24	0.99	Y	Y
PY17X_1116400	conserved Plasmodium protein, unknown function	15|24|9	0.99	Y	Y
PY17X_1035100	Bruno/HoBo/CELF2*	9|12|7	0.99	Y	Y
PY17X_0705000	conserved Plasmodium protein, unknown function	8|14|8	0.99	Y	Y
PY17X_1220900	DOZI	5|9|6	0.98	Y	Y
PY17X_0821000	Musashi/HoMu	5|6|3	0.96	Y	Y
PY17X_1336600	eIF4A	4|4|4	0.94	Y
PY17X_0808800	HSP90	8|11|3	0.94	Y
PY17X_1334000	40S ribosomal protein S3	6|3|4	0.94	Y	Y
PY17X_0817500	EF1beta	2|2|2	0.88
PY17X_1318600	EF2	2|6|4	0.86
PY17X_0415700	eIF4E	2|1|3	0.81	Y	Y
PY17X_1134900	EF1alpha	2|24|12	0.78		Y
PY17X_0521000	conserved Plasmodium protein, unknown function	5|2|3	0.72		Y
mRNA export
PY17X_1213400	PREBP	8|12|11	0.99
PY17X_1035100	Bruno/HoBo/CELF2*	9|12|7	0.99	Y	Y
PY17X_1235500	SR1	4|9|7	0.97
PY17X_0932300	RAN/TC4	4|6|6	0.96
Housekeeping
PY17X_0410900	phosphoglycerate mutase	36|55|50	1.00	Y	Y
PY17X_1217500	ENO	12|27|9	1.00	Y	Y
PY17X_1330200	GAPDH	3|4|5	0.95	Y
PY17X_1312400	ALDO2	6|13|8	0.69	Y
PyALBA4-specific unknowns
PY17X_1117900	conserved Plasmodium protein, unknown function	4|7|2	0.92
PY17X_0806800	conserved Plasmodium protein, unknown function	6|1|2	0.59

The number of spectra identified per protein per biological replicate of sexual blood stage gametocytes are listed in ‘# Spectra’. The average probability of a bona fide PyALBA4 interaction is indicated in ‘AvgP’, and starred proteins (*) were identified by previous bioinformatic analysis (36). It is also indicated whether these proteins are members of the PbDOZI and/or PbCITH complexes (Y=yes).

In contrast, we did not detect eIF4G, which typically serves to non-covalently circularize mRNAs and thus increase translational efficiency, and may indicate that these complexes are in a translationally repressed state (55). Moreover, this is further corroborated by the presence of Mushashi, which can bind with PABP and block eIF4G binding and mRNA circularization (56). Lastly, only small subunit ribosomal proteins are detected in the most stringent tier, with only one large subunit protein detected in the lower tier. This is consistent with the ALBA domain resembling the C-terminal domain of IF3, a selective binding partner of the ribosomal small subunit in prokaryotes, which may suggest a mechanism by which ALBA proteins bind both mRNA and the ribosome. That small ribosomal subunits are detected is consistent with their role in storage granules in Plasmodium and in model eukaryotes (26, 57). Their presence also supports the model that these granules, upon receiving the proper stimuli, are able to immediately resume translation, which is consistent with the prevailing models of studies with model eukaryotes.

As Plasmodium undergoes two transmission events in its life cycle (host-to-vector as a gametocyte, vector-to-host as a sporozoite) and expresses the same proteins involved in translational repression in both stages, it would be reasonable for the parasite to use stage-independent complexes to be effectively transmitted in both directions. However, because the parasite will encounter and respond to significantly different conditions in these two different events, it is also plausible that stage-specific proteins or complexes will also be required. In order to determine if Plasmodium parasites use stage-specific and/or stage-independent protein complexes, we have carried out the first IP/MS of a protein complex from Plasmodium sporozoites. Historically these (and other) experiments have failed due to the technical limitations and difficulties associated with producing and purifying large numbers of minimally perturbed sporozoites. Sporozoites must be manually dissected from mosquitoes, and soluble and insoluble material carried along from the mosquito (predominantly proteins, nucleic acids, and lipids) is in extreme excess compared to that of the parasite. This situation produces weak signal-to-noise ratios and other logistical blocks for many assays. For these reasons, most of the studies of translational repression in Plasmodium have focused upon gametocytes. We have recently overcome the obstacles of conducting these experiments with sporozoites by using a discontinuous gradient centrifugation, which we have shown is scalable and produces well-purified and fully infectious sporozoites from both human- infectious (P. falciparum) and rodent-infectious (P. yoelii) species (4). This approach has allowed for unprecedented total and surface proteomics and transcriptomics of sporozoites (5, 24), and we have now used it to conduct the first IP/MS experiment with protein complexes in sporozoites. We chose to focus on the oocyst sporozoite stage, as pyalba4⁻ sporozoites remain associated with the midgut instead of invading the salivary gland (Figure 1B). Therefore, we hypothesized that defects in this stage are responsible for this phenotype, and determining PyALBA4’s binding partners may elucidate its function at this stage.

To identify proteins that associate with PyALBA4 in oocyst sporozoites, immunoprecipitations from WT-GFP and PyALBA4::GFP transgenic parasites were conducted using formaldehyde to stabilize transient interactions through crosslinking. However, because quenching the crosslinker with excess glycine or Tris led to disruption of sporozoites and sample loss, extensive washing of the crosslinked sporozoites was done instead. These experiments with sporozoites were only feasible upon extensive purification with an Accudenz gradient (two sequential gradients with pelleting of parasites between gradients to remove released, soluble material) as we have previously described (4). However, even starting with comparatively large numbers of sporozoites (2–5 million doubly purified sporozoites), these experiments yielded very little mass following immunoprecipitation compared to blood stage parasites. This is perhaps seen most readily by the resulting average total spectral counts for PyALBA4::GFP for each sample type (Schizont: 90.7, Gametocyte: 193.7, and Oocyst Sporozoite: 37.7). Because of this, we again present the data using two statistical cutoffs. These cutoffs include the use of the same highly stringent and moderately stringent SAINT score thresholds as with other samples (0.1 and 0.35), that similarly led to the inclusion of known interaction partners of ALBA4 from other stages (Table 1, 2). Complete datasets for all three stages with both thresholds noted are provided (Supp Table 14, Supp Table 16, and Supp Table 17).

Table 2.

PyALBA4 Interactions of Special Note in Oocyst Sporozoites.

Gene ID	Product Description	# Spectra	AvgP
PY17X_1366000	ALBA4	38|24|51	--
Stress Granule Components
PY17X_0821000	RNA-binding protein musashi, putative (HoMu)	7|5|8	0.9993
PY17X_1207600	DNA/RNA-binding protein Alba 3, putative (ALBA3)	8|7|18	0.9944
PY17X_1425300	DNA/RNA-binding protein Alba 1, putative (ALBA1)	4|12|15	0.9393
PY17X_1336600	helicase 45, putative (eIF4A)	9|6|8	0.8688
PY17X_0808800	heat shock protein 90, putative (HSP90)	6|11|20	0.7986
PY17X_1441700	polyadenylate-binding protein, putative (PABP)	2|2|7	0.7958
PY17X_0823500	thioredoxin-like protein 1, putative (TrxL1)	1|3|4	0.7371
PY17X_1312400	fructose-bisphosphate aldolase 2 (ALDO2)	2|1|4	0.6554
PY17X_0913600	casein kinase 1 (CK1)	0|3|3	0.6151
PY17X_1357200	elongation factor 1-gamma, putative	0|2|9	0.5389
PY17X_1364900	DNA/RNA-binding protein Alba 2, putative (ALBA2)	1|0|3	0.5169
PY17X_1035100	CUGBP Elav-like family member 2, putative (CELF2)	0|1|6	0.4351
PY17X_0712100	heat shock protein 70, putative (HSP70)	8|14|20	0.3762
PY17X_1134900	elongation factor 1-alpha, putative	17|21|49	0.3265
Cell Periphery
PY17X_1037800	glideosome associated protein with multiple membrane spans 3, putative (GAPM3)	5|4|6	0.9847
PY17X_1315100	inner membrane complex protein, putative	3|2|2	0.942
PY17X_0404800	inner membrane complex protein 1a (IMC1a)	7|13|17	0.8034
PY17X_1243800	inner membrane complex protein 1g, putative (IMC1g)	4|8|11	0.789
PY17X_1439000	inner membrane complex protein 1h, putative (IMC1h)	1|3|3	0.6999
PY17X_1205100	inner membrane complex protein 1c, putative (IMC1c)	2|7|7	0.6713
PY17X_0404900	inner membrane complex protein 1e, putative (IMC1e)	3|11|15	0.6587
PY17X_1212600	inner membrane complex sub-compartment protein 1, putative (ISP1)	3|0|2	0.5925
PY17X_1121700	inner membrane complex protein 1j, putative (ALV7)	2|3|11	0.5893
PY17X_1360600	inner membrane complex protein 1k, putative (IMC1k)	1|1|6	0.5658
PY17X_0822300	secreted acid phosphatase, putative (GAP50)	0|2|7	0.4572
mRNA Export
PY17X_0932300	GTP-binding nuclear protein RAN/TC4, putative	3|3|6	0.9028
PY17X_1235500	serine/arginine-rich splicing factor 1, putative (SR1)	0|2|2	0.5731
PY17X_1242000	karyopherin beta, putative	0|3|6	0.5195
PY17X_0307400	ATP-dependent RNA helicase UAP56, putative (UAP56)	0|1|5	0.4506
RNA-Binding Proteins/Unknown function
PY17X_1139200	conserved Plasmodium protein, unknown function	7|8|15	0.9607
PY17X_1330200	glyceraldehyde-3-phosphate dehydrogenase, putative (GAPDH)	6|9|17	0.9026
PY17X_1424900	conserved Plasmodium protein, unknown function	4|5|13	0.734
PY17X_1208200	RNA-binding protein, putative	6|0|7	0.6534
PY17X_1411000	RNA-binding protein, putative	1|2|3	0.6408
PY17X_0307200	conserved Plasmodium protein, unknown function	0|2|3	0.5532
PY17X_1369100	conserved Plasmodium protein, unknown function	4|5|7	0.3854
PY17X_1121400	conserved Plasmodium protein, unknown function	0|2|2	0.3777

The number of spectra identified per protein per biological replicate of oocyst sporozoites are listed in ‘# Spectra’. The average probability of a bona fide PyALBA4 interaction is indicated in ‘AvgP’, and starred proteins (*) were identified by previous bioinformatic analysis (36).

In sporozoites using the highly stringent (0.1) threshold, we have identified several proteins in addition to ALBA4 that are in common in all three pulldowns, including core components of the this complex from other stages. These include ALBA1, ALBA3, Musashi, GAPDH, HSP90, the ribosome, and PABP (Table 3). Upon expanding this threshold (0.35), many additional members of the DOZI/CITH/ALBA complex from blood stages are then included, such as ALBA2, Bruno/CELF2, GBP2, Aldolase, UAP56, and ribosomal proteins from both small and large subunits (Table 2, Supp Table 16). In supplement to these findings, additional experiments conducted without the use of formaldehyde crosslinker and less stringent washing conditions revealed that a core complex of ALBA1, ALBA2, and ALBA4 associate. PyALBA3 is conspicuously absent, suggesting that the interaction between PyALBA3 and PyALBA4 may be weak, transient, or dependent upon other proteins or RNA for stabilization. Finally, we also note that ALBA4 undergoes proteolysis, which may be biologically relevant or an artifact of sample preparation despite how rapidly they can be processed (Figure S11).

Table 3.

PyALBA4 Interactions of Special Note in Schizonts.

Gene ID	Product Description	# Spectra	AvgP
PY17X_1366000	ALBA4	138|213|230	--
Stress Granule Components
PY17X_1441700	PABP	98|38|78	1.00
PY17X_1208200	GBP2*	34|34|35	1.00
PY17X_1116400	conserved Plasmodium protein unknown function	53|13|64	1.00
PY17X_1220900	DOZI	52|15|36	1.00
PY17X_1364900	ALBA2	72|26|32	1.00
PY17X_1035100	Bruno/HoBo/CELF2*	44|14|35	1.00
PY17X_0705000	conserved Plasmodium protein unknown function	36|17|29	1.00
PY17X_1318600	EF2	52|8|31	1.00
PY17X_1304900	CITH	55|18|29	1.00
PY17X_0524100	alpha tubulin 2	16|27|45	0.99
PY17X_0415700	eIF4E	17|16|22	0.99
PY17X_1357200	EF1gamma	16|13|42	0.99
PY17X_1425300	ALBA1	71|47|23	0.99
PY17X_1119700	CDC48	12|3|13	0.99
PY17X_0512000	40S ribosomal protein S2 putative RPS2	12|9|11	0.98
PY17X_0822200	HSP70-2	15|6|12	0.98
PY17X_0821000	Musashi, HoMu	12|14|18	0.98
PY17X_0817500	EF1beta	7|6|16	0.98
PY17X_0407400	CCT2	5|2|8	0.98
PY17X_0521000	conserved Plasmodium protein unknown function	8|4|4	0.98
PY17X_1437600	conserved Plasmodium protein unknown function	5|5|7	0.97
PY17X_1361400	MyoA	3|3|3	0.96
PY17X_0311400	CCT8	5|1|5	0.92
PY17X_1207600	ALBA3	31|57|40	0.74
PY17X_1210100	tubulin, beta chain	40|31|69	0.72
PY17X_1134900	EF1alpha	142|47|113	0.62
PY17X_1446600	CCT3	11|0|3	0.60
Active Translation Machinery
PY17X_0942100	PAIP1	42 9|18	1.00
PY17X_0407800	60S ribosomal protein L7	19 8|16	0.99
PY17X_0605700	eIF5A	3 2|4	0.94
PY17X_0715400	eIF3G	4 1|4	0.92
PY17X_1246200	eIF3E	9 1|5	0.90
PY17X_1209300	eIF3D	10 1|12	0.86
PY17X_1034300	eIF2gamma	1|2|11	0.74
mRNA Export
PY17X_1213400	PREBP	61|20|54	1.00
PY17X_1035100	Bruno/HoBo/CELF2*	44|14|35	1.00
PY17X_1242000	Karyopherin beta	13|20|38	1.00
PY17X_0524100	alpha tubulin 2	16|27|45	0.99
PY17X_0939200	asparagine-rich antigen putative	30|7|10	0.99
PY17X_0507200	conserved Plasmodium protein, unknown function (PHAX domain)	18|3|11	0.99
PY17X_1137200	CELF1*	13|5|9	0.98
PY17X_0407400	CCT2	5|2|8	0.98
PY17X_1313500	DDX5	12|5|10	0.98
PY17X_1453200	CELF4*	9|14|11	0.98
PY17X_0913800	TSN	17|2|6	0.97
PY17X_1235500	SR1	6|7|5	0.97
PY17X_0917700	TCP1	5|2|9	0.96
PY17X_0941100	CK2alpha	4|4|7	0.96
PY17X_0311400	CCT8	5|1|5	0.92
PY17X_0707700	CELF3*	12|1|7	0.88
PY17X_1210100	tubulin beta chain	40|31|69	0.72
PY17X_0913600	CK1	4|0|12	0.63
PY17X_0307400	UAP56	3|0|10	0.60
PY17X_1446600	CCT3	11|0|3	0.60
NPG/P-Body Components
PY17X_0828100	PABP2*	11|2|4	0.96
PY17X_0521700	enhancer of rudimentary homolog	3|6|7	0.95
Phenotype-Related Components
PY17X_0617900	CDPK4	9|2|11	0.98
PY17X_0619400	NEK4	3|4|4	0.94
Housekeeping/RNA-Binding Proteins/Unknown Function
PY17X_0410900	phosphoglycerate mutase	69|24|53	1.00
PY17X_1217500	ENO	50|38|67	1.00
PY17X_0806800	conserved Plasmodium protein, unknown function	42|12|13	1.00
PY17X_0104900	conserved Plasmodium protein, unknown function	17|6|8	0.99
PY17X_1217900	conserved Plasmodium protein, unknown function	22|14|22	0.99
PY17X_1144100	conserved Plasmodium protein, unknown function	21|5|10	0.99
PY17X_1330200	GAPDH	57|42|74	0.98
PY17X_1117900	conserved Plasmodium protein, unknown function	15|11|13	0.98
PY17X_1426200	conserved Plasmodium protein, unknown function	2|5|10	0.89
PY17X_1426300	conserved Plasmodium protein, unknown function	5|1|1	0.82
PY17X_1312400	ALDO2	36|36|53	0.77
PY17X_1139200	conserved Plasmodium protein, unknown function	20|1|43	0.63

The number of spectra identified per protein per biological replicate of asexual blood stage schizonts are listed in ‘# Spectra’. The average probability of a bona fide PyALBA4 interaction is indicated in ‘AvgP’, and starred proteins (*) were identified by previous bioinformatic analysis (36).

Finally, we also identified the interacting partners of PyALBA4 in asexual blood stage schizonts because of the strong degradation effect that it has upon specific transcripts (Figure 3B). Because rodent infectious malaria species dedicate a much higher proportion of blood stage parasites to sexual differentiation (10–15%) (58), these data were assessed knowing that some interactions may exist in this sample due to gametocyte contamination. Consistent with this, we observe substantial overlap in interactions between asexual blood stage schizonts and purified gametocytes, much of which is attributable to the DOZI/CITH/ALBA complex (Figure 4, Supp Table 17). In addition to its roles in storage granules in schizonts, PyALBA4 also associates with far more proteins involved in nuclear/cytosolic shuttling and mRNA export, such as DDX5, UAP56, SR1, CELF family proteins, and a tudor staph nuclease (TSN). Moreover, PyALBA4 also associates with the nuclear PABP2 in addition to the cytosolic PABP1, which reinforces a role for PyALBA4 in receiving new mRNAs near the nuclear pore complex (NPC) and is consistent with its localization (Figure 2A). In further support of this model, PyALBA4 also associates with generalist kinase regulators casein kinase 1 and 2-alpha, which also shuttle between the nuclear and cytosolic compartments, can directly phosphorylate ALBA1 and ALBA2 in vitro, and regulate stress granule dynamics (59, 60). Comparison of the PyALBA4 and CK2-beta interactomes revealed significant overlap (Supp Table 18), whereas those proteins from the PyALBA4 dataset that did not overlap largely have functions consistent with stress granule functions. Moreover, we also searched our proteomic data for clear evidence of phosphorylation of members of the PyALBA4 complex in schizonts and gametocytes, and detected modification of PyALBA4, ALBA3, Bruno/CELF2, GBP2, EF1g, and eIF3A even without enrichment of phosphorylated proteins (Supp Table 19). Lastly, in contrast to the PyALBA4 complex found in gametocytes, the complex in schizonts included a far larger number of both small (17) and large (24) ribosomal protein subunits, indicating that PyALBA4 may be participating in active translational complexes in this stage. This is further supported by the presence of other translation factors, including elongation factors (EF1a, EF1b, EF1g, EF2), eukaryotic initiation factors (eIF2g, eIF3D, eIF3E, eIF3G, eIF5a), and PABP-Interacting Protein 1 (PAIP1) which stimulates eIF3 subunits during active translation (Table 2, Supp Table 17). Finally, we also observed a significant interaction in asexual blood stage schizonts/gametocytes between PyALBA4 and two crucial gametocyte factors: CDPK4, which is essential for male exflagellation, and NEK4, which is essential for female gametocytes in the mosquito (61, 62). As PyALBA4 dampens male gametocyte activation, the interaction with CDPK4 may serve to modulate this process.

Taken together, these proteomic datasets indicate that many of these interacting partners are conserved across all three stages and likely form a core complex: Musashi, Bruno, GBP2, SR1, GAPDH, Aldolase, PABP, a thioredoxin peroxidase, EF1alpha, ribosomal proteins and ALBA1, 2, 3, and 4 (Figure 4; Supp Table 14, Supp Table 16, and Supp Table 17). These interactions with ribosomes and ribosome-associated factors align with previous work that showed that the ALBA domain resembles the IF3-C domain that interacts with the small subunit of the ribosome (46), and other work that showed that a a member of this complex in yeast (ScDHH1) associates directly with the ribosome (47). Excitingly however, other interactions with ALBA4 appear to be specific to individual stages, or to paired stages. For instance, blood stages (schizonts, gametocytes) also include DOZI, CITH, and eIF4E (the mRNA cap-binding protein) in this complex, but these are conspicuously absent in sporozoites. Transmission stages (gametocyte, sporozoites) both include a HSP90 and eIF4A in the complex. As these stages experience large swings in temperature when transiting between the host (37C) and vector (ambient), these may be important for these events. Asexual stages (schizonts, sporozoites) also include proteins affiliated with the nuclear periphery, such as the UAP56 splicing factor, a generalist kinase regulator CK1, and components of the nuclear pore complex. Only in schizonts are many recently defined members of stress granules confidently found, including members of the CCT and MCM families (42). Additionally, uncharacterized RNA-binding proteins, nuclear shuttling factors (DDX5) nucleases (TSN), translational modulators (PAIP1), and stress granule regulators (CK2) associate with ALBA4 (60). As seen in other model eukaryotes, many small and large ribosomal subunits also associate, further implicating the DOZI/CITH/ALBA complex in translational roles in the blood stage of Plasmodium parasites. Only in the sporozoite stage are IMC and glideosome-associated proteins detected, indicating that like in other stages, ALBA4 associates with whatever proteins are found in the cell periphery. Moreover, an RNA-binding protein (PY17X_1411000) implicated in PABP’s functions is only found associated in sporozoites and may contribute to specific functions in this stage (36). Together, these data indicate that ALBA4 associates with some of the same core complex members in a stage-independent manner to perform core functions, but diversifies its interaction portfolio as needed throughout the life cycle through stage-specific interactions.

Discussion

The association of PyALBA4 with proteins involved in many aspects of mRNA metabolism across multiple stages, including mRNA storage and repression, mRNA export, active translation, NPGs, and mRNA decay, underscores the importance of this Apicomplexan-specific RNA-binding protein. A plausible explanation for its interactions with a diverse group of proteins can be simply explained by a single interaction that links them all together: mRNA. We therefore propose a model where PyALBA4 binds mRNA in a sequence-independent and stage-dependent manner to potentially stabilize/destabilize these transcripts using different collections of effector proteins. Due to PyALBA4’s nuclear adjacent expression and interactions with mRNA export proteins, it is plausible that it is available to interact with nascent mRNAs and subsequently help to direct them to mRNPs for translation or storage/protection, whereby PyALBA4 may remain associated with the transcripts regardless of its fate. It is possible that the post-translational status of PyALBA4 affects the fate of the bound transcript, either directly or indirectly by associated effector proteins. In stages where mRNA homeostasis is of tantamount importance, gametocytes and sporozoites, we most clearly see the importance of PyALBA4’s action. While the phenotypes observed in PyALBA4’s absence do not necessarily hinder the progression of parasite development or successful establishment of infection, they underscore the temporal importance of mRNA homeostasis.

Several of our observations are especially noteworthy for understanding how this regulator affects the parasite across these life stages. First, although we see an expansive effect of ALBA4 to promote the degradation of transcripts in the asexual blood stage schizont, it was striking that there is no gross phenotype in this stage when alba4 was genetically deleted. This may be accounted for in that the extent of its effect upon most affected genes was not insurmountable for the parasite to retain normal function (Figure 3). Second, we conclude that the preservation function of ALBA4 upon transcripts in the gametocyte is far more critical than its role in asexual blood stage schizonts, and thus why activation of male gametocytes was affected. In contrast, we did not detect any effect of the deletion of alba4 upon female gametocytes, which to date is the only stage affected by genetic deletion of ALBA4’s binding partners DOZI and CITH. That members of the same complex could have gender-specific roles speaks to the organization of how this parasite can orchestrate a regulatory complex across stages. Third, it was also striking that other known regulators were not affected at the mRNA level by the genetic deletion of alba4, including well-studied sporozoite regulators IK2/UIS1, UIS2, SAP1, and PUF2. As we and others have not observed colocalization of these proteins with the ALBA complex, it is perhaps not surprising that these would be operating independently and do not serve to “regulate the regulators.” In further support of this separation of regulatory roles, we found that transcripts that are affected by the genetic deletion of alba4 are nearly exclusively absent in the lists of proteins that associate with ALBA4. Only by using our most relaxed thresholds, we find that only a few ribosome proteins, an oxidoreductase, EF1 alpha, and notably CELF2/Bruno were present in both categories in individual life cycle stages. Therefore, the identification of proteins that may serve in higher level regulatory roles that may govern these regulators would certainly be exciting and warranted. Lastly, we can only hypothesize as to how ALBA4 is affecting the egress of sporozoites from the oocyst, but suspect that ALBA4’s role in affecting mRNAs is revealing (Figure 3). We propose that here the dysregulation of transcripts is causing enough of a disturbance to delay parasite development until it can be rectified and downstream regulatory events permit sporozoite egress. Clearly there is much to examine in this challenging part of the parasite’s life cycle, and our advances in sporozoite purification and characterization presented here will enable these to be done.

Taken together, we believe that PyALBA4 acts as a key regulator of mRNA homeostasis that may remain associated with the mRNA regardless of its fate. This study demonstrates how Plasmodium parasites can utilize a single regulatory protein differently across its life cycle, simply by modifying the composition of its protein complex. Presumably this is done to affect different mRNAs as needed for the successful growth and transmission of the parasite between the host and vector. However, many interesting questions persist. How are these mRNAs selected? How do the mRNPs sense and respond to external transmission cues? Previous catalogues of data and our collective understanding of these events do not yet provide a mechanism for how Plasmodium has evolved to promote its transmission through these. Efforts to uncover these answers will no doubt offer interesting insights into parasite biology, and perhaps will identify weaknesses that can be exploited therapeutically.

Experimental Procedures

An expanded version of all experimental procedures is available in Supplemental Text 1.

Experimental animals and parasite lines

Six to eight week old female Swiss Webster (SW) mice were obtained from Harlan Laboratories (Harland, IN) and were used for all experiments described. All experiments were conducted with the Plasmodium yoelii 17XNL non-lethal strain.

Ethics

All animal care strictly followed the Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC) guidelines and was approved by the Pennsylvania State University Institutional Animal Care and Use Committee (IACUC# 42678-01). All procedures involving vertebrate animals were conducted in strict accordance with the recommendations in the Guide for Care and Use of Laboratory Animals of the National Institutes of Health with approved Office for Laboratory Animal Welfare (OLAW) assurance.

Reverse genetics of Plasmodium yoelii parasites

Transgenic parasites (pyalba4⁻ and PyALBA4::GFPmut2) were generated by producing a gene-targeting construct as previously described using pyrimethamine drug cycling and conventional reverse genetics approaches (24). Oligonucleotides used in their construction are provided in Supp Table 20.

Live fluorescence microscopy and indirect immunofluorescence assays (IFA)

PyALBA4::GFP expression in blood stages, oocyst sporozoites, salivary gland sporozoites and liver stages was visualized by live fluorescence microscopy and/or indirect immunofluorescence assay as previously described (63). Fluorescent 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.

Exflagellation assays

Two mice were infected per parasite line per biological replicate and infections were monitored daily in peripheral blood by tail snip. The number of exflagellation events was determined daily 10–15 minutes after blood collection on the peak day of exflagellation (typically day five) by counting the number of centers of movement/exflagellation centers in ten 40x phase contrast fields on a monolayer of blood cells. Six biological replicates were completed with all clones.

Quantification of gametocytemia

Thin blood smears were taken from infected mice on the days of peak number of centers of movement/exflagellation centers, and were fixed with methanol and stained with a buffered 6.7mM Giemsa solution for 7–10 minutes. Smears were viewed under 100X oil objective and the numbers of mature female, mature male, and immature gametocytes were counted per ~10,000 red blood cells (see Supp Fig 3 for a key for these gametocyte populations). The fields selected for viewing followed the St. Andrew’s cross pattern across the smear, as gametocytes tend to collect at the leading edge of the smear (64).

Measurements of mosquito infection and sporozoite development

Mosquitoes infected with either WT-GFP or transgenic parasites were dissected under a stereomicroscope in Schneider’s medium to monitor oocysts on Days 5, 7, 10, oocyst sporozoites on Days 10 and 12, and salivary gland sporozoites on Days 14, 16, 18, and 20 post-blood meal to assess aspects of these infections. Details of scoring for infection prevalence and intensity are noted in Supplemental Text 1.

Immunoprecipitations and western blotting

For immunoprecipitations from asexual and sexual blood stages, parasites were Accudenz purified, fixed with formaldehyde, quenched with glycine, lysed and incubated with Dynabeads MyOne Streptavidin T1 (Life Technologies, Cat# 65601) coated with a biotin-conjugated GFP antibody (Abcam, Cat# ab6658). Details of this protocol are found in Supplemental Text 1.

For immunoprecipitations from oocyst sporozoites, approximately 600 mosquitoes were dissected per parasite line (WT-GFP and PyALBA4::GFP) in Schneider’s medium or incomplete RMPI, and were purified on an Accudenz gradient, which ultimately yielded 3-to-10 million oocyst sporozoites. Experiments were conducted in the presence or the absence of formaldehyde fixation, but were extensively washed in media instead of being subjected to a quenching solution (which leads to substantial sporozoite loss). Details of these immunoprecipitations from sporozoites are described in Supplemental Text 1.

Mass spectroscopy and protein inference

The elution samples from the immunoprecipitations were prepared for liquid chromatography-mass spectrometry (LC/MS-MS) as previously described (5, 6) and are described at length in Supplemental Text 1. The raw and fully analyzed data files for all mass spectrometry-based experiments has been deposited in PRIDE (Accession #PXD004183 and #PXD006276).

For reviewer access of these datasets, the following login information provides access to the private raw and processed datasets:

Proteomics: http://www.ebi.ac.uk/pride/archive/

PXD004183: Username: reviewer62709@ebi.ac.uk; Password: 5KDycfjV

PXD006276: Username: reviewer17170@ebi.ac.uk; Password: x53PopxP

Total RNA-seq of schizonts, gametocytes, and oocyst sporozoites

Asexual blood stage, gametocyte, and oocyst sporozoite samples were collected as described above and in detail in Supplemental Text 1. RNA from all sample types was isolated by the QIAgen RNeasy Kit (QIAgen, Cat No. 74104) using the manufacturer’s protocol with the additional on-column DNaseI digestion. RNA yields and quality were assessed spectrophotometrically (NanoDrop 2000c, Thermo Scientific) using the Agilent Bioanalzyer. A barcoded library was made from each sample by using the Illumina TruSeq Stranded mRNA Library Prep Kit (Illumina, Cat# RS-122-2101) according to the manufacturer’s protocol. An equimolar pool was made of all libraries and was sequenced on an Illumina HiSeq 2500 in Rapid Run mode according to the manufacturer’s protocol. For schizont and oocyst sporozoite samples, 100nt × 100nt paired-end sequencing was performed, and for gametocyte samples 100 nt single read sequencing was performed. The resulting data was mapped to the P. yoelii 17XNL strain reference genome (plasmodb.org, v27) using Tophat2 in a local Galaxy instance (version .9) (65). Gene and transcript expression profiles for both WT and pyalba4⁻ assemblies were generated using Cufflinks (Galaxy version 2.2.1.0), merged into a separate large transcript assemblies with Cuffmerge (Galaxy version 2.2.1.0), and compared using Cuffdiff (Galaxy version 2.2.1.3) (66). For the purified schizont and gametocyte samples, three biological replicates were merged for WT and pyalba4⁻ profiles. Oocyst sporozoite samples used two biological replicates for this analysis.

Statistical analyses carried out by CuffDiff include a Student’s T-test generating a p-value, as well as a multiple test correction generating a q-value. Statistical cutoffs for datasets from each life cycle stage were determined based on overall coverage. For schizonts, entries with a q-value < 0.05 were considered significant. For gametocytes and sporozoites, there were no significant q-values for transcripts other than those encoding YIR proteins. This is likely due to the low percentage of reads that map to the Plasmodium yoelii genome due to contamination of host or vector material. Thus, transcripts with a log₂ fold change > 1 and < −1 were also investigated as transcripts-of-interest. These datasets were also compared to the DOZI/CITH RIP-ChIP experiment datasets to conduct targeted analyses (23). Overlapping transcripts that had a significant p-value (<0.01) were considered as hits-of-interest. RNA-seq data reported here is available through the GEO depository (Accession #GSE81834).

For reviewer access of these datasets, the following login information provides access to the private raw and processed datasets:

Transcriptomics: http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?token=olibuimsjrgflqx&acc=GSE81834

Supplementary Material

Supp Figs. Supplemental Figures.

Supp Figure 1. pyalba4⁻ clonal populations were generated by double homologous crossover recombination and confirmed by genotyping PCR. A) The entire PyALBA4 ORF was deleted by introducing linear plasmid DNA containing two homology repair sequences flanking a GFP cassette and HsDHFR cassette. B) The successful disruption of the PyALBA4 ORF was confirmed by genotyping PCR, and two clonal parasite lines were isolated by limited dilution cloning. Primer sequences are provided in Supplemental Table 20.

Supp Figure 2. pyalba4⁻ parasites do not exhibit any defect in blood stage growth kinetics. One thousand blood stage parasites were IV injected into naïve SW mice, and daily parasitemia checks were done by thin blood smear followed by Giemsa stain. There were no significant differences in parasitemia, peak parasitemia, or timing of clearance. Two biological replicates are shown, represented as the average with standard error around the mean.

Supp Figure 3. The scoring key used to classify immature, mature male, and mature female Plasmodium yoelii gametocytes. Plasmodium yoelii parasites were binned into three categories: immature gametocytes, mature females, and mature males. Mature females were identified by condensed nuclear staining, and development within the entire RBC. Additionally, pigment bodies are also present in mature female gametocytes. Mature male parasites were characterized by a slightly more pink stain, less condensed nuclear staining, and an amorphous shape. Immature gametocytes were characterized by intermediate nuclear staining, and did not occupy most or all of the RBC.

Supp Figure 4. pyalba4⁻ parasites do not exhibit significant changes in the number of oocysts per infected mosquito. A) Oocyst numbers per infected mosquito were counted on Day 7 post blood meal in both the WT-GFP line and pyalba4⁻ lines. Mosquito midguts from 20 mosquitoes were dissected and placed on a slide and gently covered with a coverslip. Oocysts were counted by fluorescence microscopy. There is no significant change in the number of oocysts per infected mosquito. B) Oocysts were also quantified Day 10 post blood meal for each parasite line. Using 20 mosquitoes per group, there was no significant change in the number of oocysts per infected mosquito. Three biological replicates are shown, presented as the average with the associated standard error.

Supp Figure 5. The total number of sporozoites per infected mosquito remains unchanged in pyalba4⁻ parasites. The total number of sporozoites per mosquito was calculated for each parasite line. There was no significant change in the total number of sporozoites found per mosquito. Three biological replicates are shown, with 10–25 mosquitoes per group per replicate, and presented as the average with the associated standard error.

Supp Figure 6. PyALBA4::GFP parasites were generated by double homologous crossover recombination and confirmed by genotyping PCR. A) A PyALBA4::GFP transgenic line was generated by introducing linear plasmid DNA containing two homology repair sequences directed to the C-terminal end of the protein and its 3’ UTR flanking a GFP-tag and HsDHFR cassette. Two populations of transgenic parasites were generated (Pop. 1 and Pop. 2). B) The successful addition of the GFP-tag was confirmed by genotyping PCR. Primer sequences are provided in Supplemental Table 20.

Supp Figure 7. PyALBA4::GFP is expressed throughout mosquito stage development, and exhibits two distinct expression patterns in oocysts. A) Midguts were dissected from infected mosquitoes at several time points post blood meal, and oocysts were assessed by fluorescence microscopy. Early oocysts (Day 3) exhibit a single PyALBA4::GFP diffuse expression pattern. Throughout oocyst development (Day 5-Day 10), two distinct expression patterns emerge: diffuse, and punctate foci. These occur in a roughly 1:1 ratio, and punctate foci appear to segregate into budding sporozoites.

Supp Figure 8. PyALBA4::GFP is expressed throughout mid- to very-late liver stage, appearing to be packaged into merozoites. In liver stage, expression progresses from diffuse in mid-liver stages (LS 24 hr) to punctate in late and very-late liver stages (LS 48 and LS 52 hr). Mid-liver stage PyALBA4::GFP expression is solely diffuse, however in late and very-late liver stages, PyALBA4::GFP localization is restricted to the forming daughter merozoites. One hundred parasites were assessed at each time point, and scale bars are ten microns.

Supp Figure 9. Comparison of pyalba4⁻ transcript abundance changes with related published datasets. We compared our pyalba4⁻ comparative total RNA-seq datasets from schizont and gametocytes to transcripts that (A) interact with DOZI/CITH bound transcripts (B) or that are regulated by PfALBA1. For the latter case PfALBA1 overexpression in late trophozoite stages was compared to pyalba4− schizont transcript abundance changes. There is very little overlap between these groups, but they affect similar categories of mRNA transcripts. These include transcripts important in invasion, such as microtubule and cytoskeletal elements (37).

Supp Figure 10. PyALBA4::GFP associates with translational repression machinery and active translational machinery in a largely asexual population. Following immunoprecipitation of PyALBA4::GFP, cross-links were reversed, and samples were used for immunoblotting to confirm capture of PyALBA4::GFP and assess efficiency of the crosslink reversal. A) Immunoblotting with anti-GFP of the 2.5% Input, Flow Through (FT), and Elution samples of the WT-GFP and PyALBA4::GFP gametocytes was done. PyALBA4::GFP is ~69 kDa and GFP is ~26kDa, and both run as expected and are clearly seen in the elutions. GFP is notably absent from the WT-GFP Input, as this is a very dilute sample. High weight molecular bands were seen, which indicate incomplete cross-link reversal. A recombinant GFP with a 6xHis-GST tag (GST-6xHis-GFPmut2, ~55kDa) protein expressed from a modified pGEX in E. coli and purified using glutathione-agarose and Ni-NTA resin was used as a positive control for detection with anti-GFP antibodies. B) Immunoblotting was performed as previously mentioned with WT-GFP and PyALBA4::GFP schizont samples. GFP is notably absent from the WT-GFP Input, as this is a very dilute sample. High weight molecular bands were seen, which indicate incomplete cross-link reversal. A recombinant GFP with a 6XHis-GST tag (~55kDa) was used as a positive control.

Supp Figure 11. Immunoprecipitation from oocyst sporozoites is feasible with purification and affinity-based techniques. Parasite lysates from five million purified oocyst sporozoites from either WT-GFP or ALBA4::GFP parasites were treated as above, except they were not chemically cross-linked. The full length ALBA4::GFP protein was recovered in the elution, as well as N- and C-terminally proteolyzed species (indicated by arrows) as these species are still recognizable by anti-GFP. Based on approximate band sizes, it is predicted that cleavage is occurring within the ALBA domain, and perhaps the GFP and the C-terminus is being proteolyzed.

Supp Tables. Supplemental Tables.

Table S1: Gametocyte comparison between parasite lines.

Table S2: pyalba4⁻ parasites show no defect in prevalence of mosquito infection.

Table S3: pyalba4⁻ parasites show no defect in time to blood stage patency.

Table S4: Distribution of PyALBA4::GFP Localization Patterns in Liver Stages.

Table S5: Transcript abundance increases in pyalba4⁻ gametocytes.

Table S6: Transcript abundance decreases in pyalba4⁻ gametocytes.

Table S7: Comparison between the DOZI and CITH RIP-ChiP targets and pyalba4⁻ transcript abundance changes in gametocytes.

Table S8: Transcript abundance changes in pyalba4 oocyst-sporozoites.

Table S9: GO term analysis of transcript abundance increases in pyalba4⁻ oocyst sporozoites.

Table S10: GO term analysis of transcript abundance decreases in pyalba4⁻ oocyst sporozoites.

Table S11: Transcript abundance changes in pyalba4⁻ schizonts.

Table S12: GO Term Analysis of transcript abundance increases in pyalba4⁻ schizonts.

Table S13: Comparison of transcripts with abundance changes in pyalba4⁻ parasites across stages.

Table S14: Full List of Detected Interactions with PyALBA4 in Gametocytes Sorted by AvgP.

Table S15: Comparison of PyALBA4 interactions with PbDOZI and PbCITH interactions in gametocytes.

Table S16: Full List of Detected Interactions with PyALBA4 in Oocyst Sporozoites Sorted by AvgP.

Table S17: Full List of Detected Interactions with PyALBA4 in Schizonts Sorted by AvgP.

Table S18: Comparison between PyALBA4 schizont and PfCK2beta Interactions.

Table S19: Phosphorylation events detected in IP/MS experiments.

Table S20: Oligonucleotides used in this study.