gondii, which causes toxoplasmosis in immunocompromised people and congenitally infected neonates; Plasmodium spp., which causes malaria; and Cryptosporidium spp., which causes diarrheal disease in children (2)(3)(4). Important animal pathogens include Neospora spp., Eimeria spp., Theileria spp., and Babesia spp., which together account for enormous economic losses in the poultry and cattle industries (5)(6)(7). These apicomplexan parasites require specialized machinery to actively invade their mammalian host cells, establish an intracellular niche, and cause disease. The alveoli are one such structure and are formed from a series of flattened membranous vesicles that underlies the plasma membrane. The alveoli represent a hallmark of the broader superphylum Alveolata that includes ciliates, dinoflagellates, and apicomplexan parasites (8).

In apicomplexans, the alveoli are called the inner membrane complex (IMC). The IMC is a peripheral membrane system with two well-described roles: a platform to anchor the glideosome, the actin-myosin motor complex that interacts with micronemal adhesins secreted onto the parasite surface for gliding motility, and a scaffold for endodyogeny, an internal budding process of replication (9,10). The IMC is situated between the plasma membrane and cortical microtubules at the periphery of the cell and consists of a series of flattened membrane vesicles and an underlying cytoskeletal network of intermediate filament-like proteins called the alveolins (11,12). The membrane vesicles are organized into rectangular plates along the body of the parasite, culminating in a single cone-shaped plate at the apex called the apical cap (13,14). Because both the apical cap and body sections of the IMC are composed of similar membrane and cytoskeletal components, they were previously believed to be one unified structure. However, the discovery of an array of new IMC proteins revealed that the apical cap contains a unique cohort of proteins, suggesting a specialized function for this region (12,(15)(16)(17)(18)(19). Recent analyses of a group of these proteins revealed a third IMC function: regulating the biogenesis and stability of the apical complex (20)(21)(22).

The apical complex is a group of cytoskeletal structures at the apex of the parasite that includes the microtubule-based conoid, the flanking apical polar ring (APR), and two preconoidal rings (18,23,24). The striking basket-shaped ultrastructure of the conoid allowed it to be readily described in the tissue cyst-forming coccidian subgroup of the Apicomplexa (e.g., Toxoplasma, Sarcocystis, and Eimeria). Remarkably, the apical complex, including the conoid, has been described in early-branching alveolates that are not members of the Apicomplexa, suggesting the structure is more ancient than originally appreciated (25,26). Indeed, while the conoid was originally presumed to be missing from Haemosporidia (1,27), recent studies have identified a reduced conoid complex in multiple stages of Plasmodium, suggesting that this structure is conserved throughout the Apicomplexa (28)(29)(30). Moreover, the apical complex contains orthologs of cilium-associated proteins, leading to a potential link between the apical complex of apicomplexan parasites and more typical eukaryotic cilia (28,(31)(32)(33)(34). Numerous studies have demonstrated that the apical complex regulates the secretion of specialized organelles called micronemes and rhoptries, which govern parasite motility, attachment, invasion, and egress (35). While the trigger for rhoptry secretion at the apical complex is unknown, calcium signaling cascades have been shown to coordinate both microneme secretion and conoid extrusion, suggesting a connection between the two activities (36). The conoid has also been implicated in initiating motility via several calmodulin-like proteins, the myosin motor protein MyoH, and the essential formin protein FRM1 (37)(38)(39). In addition, several APR-localizing proteins were shown to be important in controlling microneme release, indicating that these flanking cytoskeletal structures also contribute to the function of the apical complex (18,40,41).

While the molecular composition and function of the apical complex is becoming clearer, how it is formed and maintained is largely a mystery. Recently, three apical cap proteins (AC9, AC10, and ERK7) were identified as essential for the maturation of the apical complex (20)(21)(22). Depleting any one of these proteins eliminates the conoid in mature parasites, resulting in a complete block in motility, invasion, and egress. Importantly, AC9 was shown to accomplish this by recruiting the conserved mitogen-activated protein (MAP) kinase ERK7 to the apical cap and regulating its kinase activity (22). Thus, it is evident that AC9, AC10, and ERK7 work in conjunction to facilitate the apical complex maturation and function. However, how these proteins interact and coordinate at the apical cap to confer their functions remains unknown. In this study, we explore the organization and mechanism of this essential protein complex. We show that AC10 recruits both AC9 and ERK7 to the apical cap, suggesting it is the anchor for the complex. We combine yeast two-hybrid (Y2H) experiments to examine direct pairwise interactions with deletion analyses in parasites to assess the functional importance of these interactions. Through these experiments, we reveal multiple domains in AC9 and AC10 that are critical for assembling the complex at the apical cap and for the maturation of the conoid. Importantly, we show that these domains mediate independent pairwise interactions between AC9, AC10, and ERK7. Thus, we propose that these multimeric interactions drive the oligomerization of the AC9:AC10: ERK7 complex into the apical cap cytoskeleton, which tethers ERK7 to the site of its essential function in coordinating the proper biogenesis of the apical complex.

AC10 is essential for recruitment of the AC9:AC10:ERK7 complex to the apical cap. While AC9, AC10, and ERK7 were recently shown to be essential for apical complex assembly and stabilization (20)(21)(22), the interactions between the three proteins and how they are organized in the apical cap remain poorly understood (an overview of these proteins is shown in Fig. 1). To explore their interactions, we generated parasites with AC10 tagged with an auxin-inducible degron fused to 3ÂHA, AC9 tagged with 3ÂMyc, and ERK7 tagged with 3ÂTy (triple-tagged: AC10 AID-3ÂHA /AC9 3ÂMyc /ERK7 3ÂTy ). As shown previously, the AC10 AID-3ÂHA fusion protein targets correctly to the apical cap, degrades efficiently upon addition of auxin (IAA), and results in the loss of AC9 from the apical cap (Fig. 2A andB) (21). Our triple-tagged parasites allowed us to additionally demonstrate that AC10 AID-3ÂHA knockdown removes ERK7 from the apical cap, although its cytoplasmic staining is retained (Fig. 2B). We used line intensity scans to quantify the levels of ERK7 at the apical cap versus the bulk cytosol, which clearly demonstrated a loss in concentrated apical cap signal upon AC10 knockdown (Fig. S2). Consistent with the AC9 and ERK7 staining patterns, Western blot analyses showed that AC9 is predominantly degraded while ERK7 levels appear to remain stable (Fig. 2C) (21). In agreement with previous studies (21), depletion of AC10 results in the elimination of the conoid (Fig. 2D), which is lethal -3ÂHA j AC9 3xMyc j ERK7 3xTy ) shows that endogenous AC10 AID-3ÂHA colocalizes with the apical cap marker ISP1 and is efficiently depleted upon addition of IAA (3-indoleacetic acid). Green, rabbit anti-HA; magenta, mouse anti-ISP1. (B) IFA showing that the depletion of AC10 AID-3ÂHA results in the absence of AC9 and the loss of ERK7 from the apical cap. Green, rabbit anti-Myc; magenta, mouse anti-Ty. (C) Western blot analysis confirms efficient degradation of AC10 AID-3ÂHA and the concomitant nearly complete degradation of AC9 upon AC10 AID-3ÂHA knockdown. ERK7 levels are not substantially affected. AC10 AID-3ÂHA , mouse anti-HA; AC9, mouse anti-Myc; ERK7, mouse anti-Ty. Rabbit anti-IMC12 was used as a loading control, and validation of this antibody is shown in Fig. S1. (D) AC10 AID-3ÂHA knockdown results in the elimination of the conoid, detected by SAS6L. Green, rabbit anti-HA; magenta, mouse anti-SAS6L. (E) Representative plaque assay images and quantification of plaque numbers illustrate a complete loss of plaque formation upon AC10 AID-3ÂHA depletion. (F) Using parasites tagged with AC9 AID-3ÂHA and AC10 3xMyc ,I F A shows that conditional knockdown of AC9 (1IAA) does not affect the localization of AC10. Green, mouse anti-Myc; magenta, rabbit anti-HA. All scale bars are 2 mm.

for the parasites (Fig. 2E), as it renders them immotile and noninvasive. In addition, we confirmed that the knockdown of AC9 does not affect the localization of AC10 (Fig. 2F) (21), indicating that AC10 does not rely on AC9 for apical cap localization. These results demonstrate that AC10 is essential for recruiting both AC9 and ERK7 to the apical cap and suggest that AC10 is the foundational component of the AC9:AC10:ERK7 complex.

AC9 is recruited to the apical cap through a direct interaction with AC10. Like most IMC components, AC9 and AC10 lack significant homology to other proteins. Both proteins contain large stretches of predicted intrinsic disorder as well as predicted coiled-coil (CC) domains toward their N termini (Fig. 1A andB). In addition, we previously identified a well-conserved sequence in the AC9 C terminus that is required to recruit ERK7 to the apical cap and acts as a competitive inhibitor of ERK7 kinase activity by occupying both the kinase scaffolding and active sites (22). Since AC10 likely recruits AC9 to the apical cap, we reasoned that the AC9 CC domain is required for this interaction. In the background of our AC9 AID-3ÂHA strain (22), we expressed a second copy of AC9 driven by the ISC6 promoter and targeted to the UPRT locus (AC9 wt ) (Fig. 3A andB) (42). As expected, expression of AC9 wt rescued the AC9 AID-3ÂHA knockdown phenotype, as assessed by SAS6L staining of the conoid and plaque assay (Fig. 3C to E). We also created a strain expressing AC9 in which the core of the predicted CC domain had been deleted (residues D75 to 113, AC9 DCC ) (Fig. 3F). Consistent with the high conservation of this region (Fig. 1A), AC9 DCC was not correctly targeted to the apical cap and, thus, it was unable to rescue the effects of AC9 AID-3ÂHA degradation (Fig. 3G to I). Because AC9 DCC staining was faint, we assessed its stability by Western blotting and found that it is expressed at the appropriate size, but its protein level is greatly diminished (Fig. S3A). This low level of AC9 DCC is likely the result of turnover upon loss of binding to its partner, AC10, as loss of AC9 is also seen following AC10 AID knockdown (Fig. 2C). While we and others have demonstrated a potential interaction between AC9 and AC10 through proximity biotinylation (21,22), this interaction may either be direct or through an intermediate protein. To test whether AC9 directly binds AC10, we used a Y2H system in which stable interactions drive the expression of the HIS3 marker. Full-length AC9 was expressed as an N-terminal fusion with the LexA DNA binding domain, and AC10 was expressed as an N-terminal fusion with the GAL4 activating domain. As AC10 is a large protein of 1,979 residues, we split the protein into thirds and tested each portion for activation: AC10 A containing residues 2 to 650, AC10 B containing residues 651 to 1300, and AC10 C containing residues 1301 to 1979 (Fig. 1B). Intriguingly, we found that AC9 interacts with two independent regions of AC10, robustly binding both AC10 A and AC10 B ; however, we observed no growth under restrictive conditions with the C-terminal AC10 C region (Fig. 3J; all Y2H data are shown in Table 1). These data suggest AC10 C does not bind AC9, although we cannot rule out that AC10 C is not stable in yeast and is therefore unavailable for binding.

To test whether the AC9 CC domain was required for this interaction, we deleted this region from the full-length Y2H construct (AC9 DCC ). Consistent with its inability to rescue the AC9 AID-3ÂHA knockdown phenotype in parasites, AC9 DCC was unable to bind either AC10 A or AC10 B (Fig. 3K). Moreover, the AC9 CC domain alone was sufficient to bind AC10 A in the Y2H assay, although it could not interact with AC10 B . The a-helical region of AC9 C terminal to the predicted CC is one of the more highly conserved areas in the protein (Fig. 1A). We therefore extended our Y2H construct to include this region (AC9 70-157 ), which now robustly interacted with both AC10 A and AC10 B (Fig. 3K). Taken together, these data demonstrate that the conserved a-helical sequence containing the predicted AC9 CC domain is driving interaction with at least two independent sites on AC10, and these interactions are required for forming the functional ternary complex in the apical cap.

The N-terminal third of AC10 binds both AC9 and ERK7 and is required for efficient recruitment of ERK7 to the apical cap. As AC9 CC binds AC10 at multiple distinct sites within the first two-thirds of the protein (Fig. 3J), we sought to further delineate which regions of AC10 are required for this interaction. Since AC10 A encompasses the most conserved stretch of residues in AC10 and includes a predicted CC domain (Fig. 1B), we generated a Y2H construct in which CC1 was deleted from this region [residues D422 to 513, AC10 A(DCC1) ]. The Y2H assay showed that AC10 A(DCC1) was unable to interact with full-length AC9, demonstrating that CC1 is necessary for binding (Fig. 4A). AC10 CC1 alone was not, however, sufficient to bind AC9, suggesting that this region does not form a simple coiled-coil interaction with AC9 (Fig. 4A).

To interrogate the functional domains of AC10 in parasites, we expressed full-length AC10 fused to a V5 epitope tag driven by its endogenous promoter and targeted to the UPRT locus (AC10 wt ) (Fig. 4B). As expected, the AC10 wt complementation construct correctly localized to the apical cap (Fig. 4C), fully rescued the plaque defect (Fig. 4D), properly recruited both AC9 and ERK7 (Fig. 4E), and restored SAS6L staining to the conoid upon AC10 AID-3ÂHA degradation (Fig. 4F). Thus, this complementation system serves as a platform to assess the functional domains of AC10.

To assess the role of AC10 CC1 in parasites, we deleted CC1 from the full-length construct (AC10 DCC1 ) and expressed it in the AC10 AID-3ÂHA strain (Fig. 4G). While AC10 DCC1 targeted correctly (Fig. 4H), this complemented strain was unable to form plaques upon AC10 AID-3ÂHA degradation, demonstrating that CC1 is essential for AC10 function (Fig. 4I). Consistent with the lack of plaque formation, AC10 DCC1 did not recruit ERK7 to the apical cap upon AC10 AID-3ÂHA degradation (Fig. 4J), resulting in the loss of SAS6L signal (Fig. 4K). However, we still observed AC9 recruitment in AC10 DCC1 parasites upon AC10 AID-3ÂHA degradation (Fig. 4J). This observation was surprising, as we have

a Bait and prey constructs and their relative growth on selective media are noted. previously shown that the AC9 C terminus forms a tight interaction with ERK7 and is required for its recruitment to the apical cap (22). These data suggest that AC10 CC1 also directly binds ERK7 independently of the AC10 recruitment of AC9 to the apical cap.

We tested this hypothesis using our Y2H assay and found that AC10 A was indeed able to bind the ERK7 kinase domain (Fig. 5). In contrast to the interaction with AC9, in which AC10 CC1 was required, we found that AC10 A(DCC1) was still able to bind ERK7 in the Y2H assay, although the interaction was attenuated. In addition to AC10 A interacting with the ERK7 kinase domain, we were surprised to find that AC10 B also interacted with the intrinsically disordered C terminus of ERK7, suggesting that ERK7 forms multivalent interactions with AC10. Thus, the Y2H and functional data indicate that multiple AC10 regions mediate interactions with both AC9 (Fig. 3 and4) and ERK7 (Fig. 5). Among these interactions, AC10 CC1 is required for the efficient recruitment of ERK7 to the apical cap independently of AC9, and this interaction is essential for the formation of the mature conoid.

A short, conserved sequence in AC10 is essential to bind and recruit AC9 to the apical cap. Because the AC10 DCC1 strain was still able to recruit AC9 to the apical cap, we sought to identify additional regions in AC10 that are required for AC9 recruitment. Our Y2H experiments identified regions in AC10 B that independently bound AC9 (Fig. 3J). To identify a minimal region that was sufficient for AC9 binding, we focused on a short, conserved sequence within AC10 B that is predicted to form an a-helix (Fig. 1B) and has a heptad repeat similar to that seen in coiled-coil domains (Fig. 6A). Y2H analysis showed that residues 651 to 683 were sufficient to robustly interact with AC9 (Fig. 6B), leading us to label this region the AC9 binding domain (AC10 AC9-BD ). To test the importance of this region for AC10 function in parasites, we complemented the AC10 AID-3ÂHA strain with a construct in which AC9-BD had been deleted [AC10 D(AC9-

] (Fig. 6C). We found that while the truncated protein localized properly to the apical cap (Fig. 6D), it was unable to rescue the plaque defect upon AC10 AID-3ÂHA knockdown (Fig. 6E). We also observed that both AC9 and ERK7 were absent from the apical cap upon AC10 AID-3ÂHA degradation (Fig. 6F), resulting in the loss of the conoid (Fig. 6G). These results suggest that AC10 AC9-BD forms a short coiled-coil with AC9 CC , and this interaction is absolutely required for recruitment of the AC9:ERK7 complex to the apical cap in parasites.

A third AC9 binding site on AC10 is required for full parasite fitness. While AC10 AC9-BD was sufficient to bind AC9 in our Y2H assay (Fig. 6B), AC10 B also contains the second predicted CC domain spanning residues 781 to 830 (Fig. 1B and7A). To assess the importance of CC2, we first generated a construct with AC9-BD deleted from AC10 B (AC10 684-1300 ) and found that this region still interacted with AC9 (Fig. 7A). We then deleted CC2 from AC10 684-1300 (AC10 684-1300,DCC2 ), which resulted in a somewhat attenuated interaction with AC9 in our Y2H assay. We additionally found that a FIG 4 Legend (Continued) apical cap regardless of AC10 AID-3ÂHA knockdown. Green, rabbit anti-HA; magenta, mouse anti-V5. (I) Plaque assays demonstrate that AC10 DCC1 cannot rescue the parasite's lytic ability. (J) IFA shows that AC9 is present in the apical cap, while ERK7 is mislocalized to the cytoplasm upon knockdown of AC10 AID-3ÂHA . Green, rabbit anti-Myc; magenta, mouse anti-Ty. (K) IFA illustrates that AC10 DCC1 does not rescue SAS6L localization, indicating the absence of the conoid. Green, rabbit anti-V5; magenta, mouse anti-SAS6L. All scale bars are 2 mm. portion of AC10 B containing CC2 (AC10 684-913 ) is not sufficient for interacting with AC9. These Y2H results suggest that CC2 contains minor AC9 binding regions and that the remaining residues in AC10 B provide additional binding sites, further supporting the hypothesis that AC9 and AC10 interact via multiple contact points.

We then asked whether deletion of CC2 in the context of an otherwise full-length protein would affect AC10 function in parasites. We generated AC10 DCC2 (residues D781 to 830) and expressed it in the triple-tagged AC10 AID-3ÂHA line (Fig. 7B). As with our other deletion constructs, AC10 DCC2 protein localized correctly to the apical cap (Fig. 7C). Upon degradation of AC10 AID-3ÂHA , AC10 DCC2 mostly rescued parasite fitness in a plaque assay, with a small but reproducible 15% reduction in plaque size (Fig. 7D). Consistent with this minor impact on the lytic cycle, both AC9 and ERK7 localizations were unaffected (Fig. 7E) and the conoid appeared intact (Fig. 7F). These data suggest that binding of AC9 and other potential interactors at this site, while not required for full parasite fitness, is still functionally relevant.

AC10 N-and C-terminal deletions reveal additional domains for full apical cap function. The functional regions of AC10 described above only occupy about half of the 1,979-residue protein. Notably, AC10 orthologs in other Sarcocystidae are of various lengths and display low sequence identity through the majority of the protein (Fig. 1B). To determine if the remainder of the protein harbored any additional regions important for function, we first deleted the N-terminal region of AC10 up to 36 residues N terminal to AC10 CC1 (residues 387 to 1979, AC10 DN-term ) (Fig. 8A). The AC10 DN-term protein localized properly to the apical cap independently of AC10 AID-3ÂHA degradation (Fig. 8B). Upon AC10 AID-3ÂHA depletion, parasites with AC10 DN-term displayed a substantial fitness defect by to 830 deleted from the AC10 wt construct. (C) IFA shows that AC10 DCC2 localizes to the apical cap and is not affected by AC10 AID-3ÂHA knockdown. Green, rabbit anti-HA; magenta, mouse anti-V5. (D) Plaque assays indicate that AC10 DCC2 complementation does not fully rescue the growth defect (15% reduction). Statistical significance was calculated using two-sample two-tailed t tests, and P values are noted on the graph. (E) AC9 and ERK7 staining with or without IAA shows that AC10 DCC2 can still recruit members of the complex to the apical cap. Green, rabbit anti-Myc; magenta, mouse anti-Ty. (F) IFA illustrates that AC10 DCC2 restores SAS6L staining at the conoid. Green, rabbit anti-V5; magenta, mouse anti-SAS6L. All scale bars are 2 mm. plaque assay (48% reduction in plaque size) (Fig. 8C). However, AC10 DN-term appears to be sufficient for recruiting both AC9 and ERK7 to the apical cap (Fig. 8D), resulting in the presence of a conoid as demonstrated by apical SAS6L staining (Fig. 8E). Thus, while this N-terminal region is not strictly required for recruiting AC9:ERK7 and maturation of the conoid, its deletion reduces parasite fitness, indicating that this region is important for full AC10 function.

We next focused on the C-terminal region of AC10. Due to the lack of identifiable features in this region, we deleted the C-terminal half of the protein, which includes AC10 C plus the portion of AC10 B C terminal to the CC domains (residues D914 to 1979, AC10 DC-term ) (Fig. 9A). Upon examining the localization of AC10 DC-term , we noticed striking, cell cycle-dependent variation. In mature parasites, AC10 DC-term localized to the apical cap regardless of AC10 AID-3ÂHA depletion (Fig. 9B). However, in budding parasites, AC10 DC-term was largely absent from the maternal apical cap while remaining intact in the daughter buds (Fig. 9C). Thus, we assessed the localization of AC9 and ERK7 in mature parasites expressing AC10 DC-term and found that only a small amount of AC9 could be detected in the apical cap upon AC10 AID-3ÂHA knockdown (Fig. 9D). ERK7 also appeared to be dramatically diminished from the apical cap in mature parasites (Fig. 9D). In budding parasites, while both AC9 and ERK7 were drastically reduced in mature apical caps, the signal appeared largely intact in daughter buds, similar to the localization of AC10 DC-term (Fig. 9E). Somewhat surprisingly, despite these substantial localization defects, the conoid still appeared to be intact by SAS6L staining, suggesting that the amounts of AC9, AC10 DC-term , and ERK7 in the apical cap are sufficient to stabilize the conoid (Fig. 9F). Nevertheless, plaque assays revealed that parasites expressing AC10 DC-term suffered a severe defect in parasite fitness upon AC10 AID-3ÂHA degradation (85% reduction in plaque size) (Fig. 9G).

We next sought to determine whether the C-terminal half of AC10 described above binds directly to AC9. We created a Y2H construct spanning AC10 residues 914 to 1300 to interrogate the C-terminal portion of AC10 B (AC10 914-1300 ) (Fig. 9H). Despite the defects in AC9 and ERK7 recruitment in AC10 DC-term parasites, we found that neither AC10 914-1300 (Fig. 9I) nor the remainder of the AC10 C terminus (AC10 C ) interacts with AC9 (Fig. 3J). Together, these results suggest that while the AC10 C terminus does not directly interact with AC9, it contains important regions for maintaining the integrity of the AC9:AC10:ERK7 complex. Since deletion of either the N or C termini of AC10 only partially disrupted function, we assessed whether the combination of these regions is essential by deleting both regions simultaneously (residues D2 to 337 and D914 to 1979, AC10 DN/C ) (Fig. 10A). As with AC10 DC-term , AC10 DN/C localized properly in mature parasites (Fig. 10B), and during replication, the signal was diminished specifically in maternal apical caps upon addition of auxin (Fig. 10C). Unlike AC10 DC-term , however, this construct could not rescue the plaque defect at all (Fig. 10D). Western blot analysis demonstrated that the difference between AC10 DC-term and AC10 DN/C does not appear to be due to expression levels (Fig. S3B). Consistent with the complete loss of function of AC10 DN/C , both AC9 and ERK7 were absent from the maternal apical caps of both mature and budding parasites (Fig. 10E andF). In addition, we observed reduced AC9 and ERK7 signal in the apical caps of daughter buds (Fig. 10F). In agreement with the lack of ability to form plaques, AC10 DN/C parasites were completely missing apical SAS6L staining upon AC10 AID-3ÂHA /C combining the AC10 DN-term (residues 2 to 337) and AC10 DC-term (residues 914 to 1979) deletions from the AC10 wt construct. (B) IFAs illustrate that AC10 DN/C localizes properly to the apical caps in mature parasites with or without IAA. Green, rabbit anti-HA; magenta, mouse anti-V5. (C) IFAs show that AC10 DN/C appears to be almost completely absent from the maternal apical cap of budding parasites upon depletion of AC10 AID-3ÂHA (insets, yellow arrows). However, in daughter apical caps, AC10 DN/C remains intact even upon depletion of endogenous AC10 AID-3ÂHA (insets, white arrowheads). Green, rabbit anti-HA; magenta, mouse anti-V5. (D) Plaque assays show that deleting both N-and C-terminal regions from AC10 eliminates plaque formation. (E) IFAs display the absence of AC9 and ERK7 from mature apical caps. Green, rabbit anti-Myc; magenta, mouse anti-Ty. (F) IFAs show that AC9 and ERK7 remain intact in daughter apical caps (insets, white arrowheads) but appear completely eliminated from maternal apical caps upon knockdown of AC10 AID-3ÂHA (insets, yellow arrows). Green, rabbit anti-Myc; magenta, mouse anti-Ty. (G) IFAs display absence of SAS6L upon AC10 AID-3ÂHA knockdown. Green, rabbit anti-V5; magenta, mouse anti-SAS6L. All scale bars are 2 mm. depletion (Fig. 10G). Together, these results demonstrate that the cumulative effect of deleting both N-and C-terminal regions renders AC10 nonfunctional.

AC10 effectively competes with AC9 as an ERK7 substrate. Because AC10 binds both AC9 and ERK7 (Fig. 4 to 6), and ERK7 localization (22) and kinase activity (20) are both essential for a functional conoid, we asked whether AC10 is phosphorylated by ERK7. Notably, AC10 has 396 phosphorylatable residues (Ser/Thr), and 57 of these residues have been identified as phosphorylated in parasites in published phosphoproteomics data sets (43), including 10 high-probability MAP kinase sites spread throughout the AC10 sequence. We created a bacterial expression construct of the N-terminal region of AC10 that is bound by both AC9 and ERK7. We found that this recombinantly expressed and purified AC10 was robustly phosphorylated by ERK7 (Fig. 11). Remarkably, the AC10 protein was phosphorylated to a much greater degree than myelin basic protein (MBP), a typical generic substrate used to test MAPK activity (44).

We previously demonstrated that AC9 binds ERK7 with an approximate dissociation constant (K D ) of 20 nM and robustly inhibits ERK7 activity (22). This led us to propose a model by which AC9 increases the specificity of ERK7 for its substrates, as true substrates must not only bind the active site but also compete with AC9 for scaffolding interaction. We therefore tested whether the AC10 interaction with ERK7 can overcome inhibition by the AC9 418-452 peptide (Fig. 11). As expected, addition of equimolar AC9 418-452 to the kinase reaction completely blocks MBP phosphorylation by ERK7. We found, however, that AC10 phosphorylation is undiminished by the addition of AC9.

Model for AC9:AC10:ERK7 complex oligomerization in the apical cap. AC9, AC10, and ERK7 oligomerize with the IMC cytoskeleton filaments that are associated with the cytosolic leaflet of the IMC membrane. AC10 recruits the other two proteins to the IMC, possibly through interaction with an undescribed adaptor protein. Because AC10 has multiple binding sites for both AC9 and ERK7, which also interact with one another, the three proteins likely form an irregular oligomer. These interactions concentrate ERK7 at the apical cap while allowing it to bind and phosphorylate its substrates and thereby facilitate the stability of the apical complex.

Essential Interactions Govern Apical Cap Function Furthermore, when we included equimolar AC9, AC10, and MBP in the kinase reaction, we saw that MBP phosphorylation was still fully inhibited while AC10 was still robustly phosphorylated. These data strongly suggest AC10 is a legitimate substrate of ERK7 and that one function of ERK7 kinase activity is to regulate the conformation and assembly of the AC10 complex.

In this study, we explore the organization and function of the AC9:AC10:ERK7 ternary complex. We demonstrated that both AC9 and ERK7 are dependent on AC10 to be recruited to the apical cap, suggesting that AC10 is an anchor for the complex. However, it remains unclear how AC10 itself is targeted to the apical cap. One possibility is that other apical cap proteins recruit AC10. Similar to AC10, six of the known apical cap proteins (AC2, AC3, AC4, AC5, AC7, and AC8) are associated with the IMC cytoskeletal network (15). Unlike AC9 and AC10, these other apical cap proteins were predicted to be dispensable based on a genome-wide CRISPR screen (45). Thus, it is possible that these apical cap proteins play redundant roles in organizing the AC9: AC10:ERK7 complex. It is also possible that there are undiscovered components of this protein complex or ones that serve to tether AC10 to the apical cap.

To determine how AC9, AC10, and ERK7 interact, we focused on identifiable domains using a combination of pairwise Y2H (Table 1) and complementation assays to assess direct binding and functional relevance. AC10 appears to recruit AC9 (Fig. 2) (21), which in turn recruits ERK7 through a conserved C-terminal motif that serves to both concentrate ERK7 at the apical cap and regulate its kinase activity (22). Our Y2H and complementation assays revealed a conserved helical sequence at the AC9 N terminus that was both necessary and sufficient to bind AC10 and was required for AC9's localization at the apical cap (Fig. 2). Remarkably, this single region of AC9 was able to bind multiple sites on AC10 (Fig. 3 and7). In addition, AC10 can independently interact with both the kinase domain and C-terminal regions of ERK7 (Fig. 5). AC10 therefore seems to act as a large scaffolding molecule that recruits multiple copies of each AC9 and ERK7. Furthermore, combined with the multiple binding sites on AC10 for both AC9 and ERK7, because each component of the AC9:AC10:ERK7 complex can interact with the other, it seems likely that AC10 functions to nucleate oligomerization of this complex (Fig. 12). Importantly, AC9, AC10, and ERK7 each has been demonstrated to fractionate with the detergent-insoluble parasite cytoskeleton (20,21), and their oligomerization is consistent with the characteristic meshwork of the IMC cytoskeleton. The AC10 binding region of AC9 is a predicted coiledcoil (AC9 CC ), and we identified two regions of AC10 (AC10 CC1 and AC10 AC9-BD ) with coiledcoil-like properties that are required for AC9 interaction and essential for AC10 function in parasites. Notably, predicted coiled-coil domains have also been shown to be essential in other IMC proteins (46)(47)(48), suggesting this is a general theme of IMC cytoskeleton assembly.

Deletion of the short AC10 AC9-BD sequence blocks AC9 recruitment to the apical cap in parasites (Fig. 6). However, AC9 localization was largely unperturbed in AC10 DCC1 parasites while ERK7 was unable to be recruited to the apical cap (Fig. 4). Remarkably, Y2H revealed that the N-terminal third of AC10 was able to physically interact with both AC9 and the ERK7 kinase domain, although the AC10 CC1 region itself was only required for AC9 binding (Fig. 4A and5). This differential effect of AC10 DCC1 on AC9 and ERK7 binding to this region suggests that the binding interfaces occupy different surfaces of a folded domain. We also found that this N-terminal region of AC10 was robustly phosphorylated by ERK7 in vitro and was unaffected by AC9 inhibition (Fig. 11). Together, these data indicate that AC10 is an ERK7 substrate in parasites and that its phosphorylation functions in regulating the assembly of the AC9:AC10:ERK7 complex into the apical cap cytoskeleton.

While AC10 is found throughout coccidia, its length and much of its sequence are not well conserved (Fig. 1B). Nevertheless, there are stretches of conserved sequence in the N-and C-terminal regions that are outside those we identified as critical for interacting with AC9 and ERK7. We found that neither of these regions of AC10 was essential to function, although deletion of either reduced parasite fitness (Fig. 8C and9G). Notably, AC10 DC-term parasites showed the fragility of the AC9:AC10:ERK7 complex, in which the initial recruitment to the apical cap was largely unaffected in daughter cells (Fig. 9C), but the complex appeared disrupted in mature parasites (Fig. 9D andE). While AC10 DC-term parasites showed a substantial loss of function, the complex was still able to function in facilitating maturation of the conoid (Fig. 9G). In contrast, deletion of both the N-and C-terminal regions of AC10 rendered the AC9:AC10:ERK7 complex nonfunctional, as the daughter conoids were lost (Fig. 10G) and parasites were nonviable (Fig. 10D). Therefore, it appears that these regions of AC10 either recruit other, undescribed components of the apical cap cytoskeleton or form nonessential interactions that facilitate AC9:AC10:ERK7 oligomerization.

This study builds on an increasingly robust body of evidence that the apical cap acts as an essential platform to facilitate the assembly and maintenance of the apical complex (20)(21)(22). A previously proposed model suggested that AC9 and AC10 act primarily to stabilize the Toxoplasma subpellicular microtubules due to the distribution of AC9 and AC10 proteins along the longitudinal rows of the microtubules (21). While our data support the idea that AC9 and AC10 form filaments in the apical cap cytoskeleton, this model was developed prior to establishing a connection with the MAP kinase ERK7 and its essential role in apical complex maturation (20). We have previously shown that an AC9 mutant that is unable to recruit ERK7 to the apical cap cannot rescue the AC9 knockdown (22). We have built upon that finding here, demonstrating a loss of the conoid in mutant AC10 parasites that can recruit AC9, but not ERK7, to the apical cap (Fig. 4J andK). Taken together, our data suggest a different model in which the ERK7-dependent phosphorylation of AC10 promotes functional assembly of the AC9:AC10:ERK7 complex at the apical cap (Fig. 12). It is likely that ERK7 then phosphorylates other substrates after being recruited at this site, which may include critical components of the apical complex.

T. gondii and host cell culture. T. gondii RHDku80Dhxgprt (parental) and subsequent strains were grown on confluent monolayers of human foreskin fibroblasts (HFFs; ATCC) at 37°C and 5% CO 2 in Dulbecco's modified Eagle medium (DMEM) supplemented with 5% fetal bovine serum (Gibco), 5% cosmic calf serum (HyClone), and 1Â penicillin-streptomycin-L-glutamine (Gibco). Constructs containing selectable markers were selected using 1 mM pyrimethamine (dihydrofolate reductase-thymidylate synthase [DHFR-TS]), 50 mg/ml mycophenolic acid-xanthine (HXGPRT), or 40 mM chloramphenicol (CAT) (49)(50)(51). Removal of HXGPRT was negatively selected using 350 mg/ml 6-thioxanthine (6-TX), and homologous recombination to the UPRT locus was negatively selected using 5 mM5 -fluorodeoxyuridine (FUDR) (42).

Antibodies. The hemagglutinin epitope was detected with mouse monoclonal antibody (MAb) HA.11 (diluted 1:1,000) (item no. 901515; BioLegend) or rabbit polyclonal antibody (pAb) anti-HA (diluted 1:1,000) (catalog no. PI715500; Invitrogen). The Ty1 epitope was detected with mouse MAb BB2 (diluted 1:1,000) (52). The c-Myc epitope was detected with mouse MAb 9E10 (diluted 1:1,000) (53) or rabbit pAb anti-Myc (diluted 1:1,000) (catalog no. PA1981; Invitrogen). The V5 epitope was detected with mouse MAb anti-V5 (diluted 1:1,000) (catalong no. R96025; Invitrogen). Toxoplasma-specific antibodies include mouse MAb m-IMC1 (diluted 1:500) (54), mouse MAb anti-ISP1 (diluted 1:1,000) (55), and rabbit pAb anti-IMC6 (diluted 1:2,000) (46).

Production of IMC12 antibody. The IMC12 coding sequence was cloned into the pET His6 TEV LIC bacterial expression vector (Addgene plasmid number 29653; Scott Gradia) using primers P32 to P35. The construct was transformed into BL21(DE3) Escherichia coli, and protein was induced with 1 mM isopropyl-b-D-thiogalactopyranoside (IPTG) and purified using nickel-nitrilotriacetic acid (Ni-NTA) agarose under denaturing conditions as described previously (56). The sample was then dialyzed into PBS to remove the urea, and rabbit antisera were produced by Cocalico Biologicals.

Immunofluorescence assay and Western blotting. Confluent HFF cells were grown on glass coverslips and infected with T. gondii. After 18 to 24 h, the coverslips were fixed with 3.7% formaldehyde in PBS and processed for immunofluorescence assay (IFA) as described previously (56). Primary antibodies were detected by species-specific secondary antibodies conjugated to Alexa Fluor 488/594 (ThermoFisher). Coverslips were mounted in Vectashield (Vector Labs, Burlingame, CA), viewed with an Axio Imager.Z1 fluorescence microscope (Zeiss), and processed with ZEN 2.3 software (Zeiss). Processing with the ZEN software included deconvolution as well as adaptation of the magenta pseudocolor from the 594 fluorophore.

For Western blotting, parasites were lysed in 1Â Laemmli sample buffer with 100 mM dithiothreitol Essential Interactions Govern Apical Cap Function stopped by adding 5 ml6Â SDS buffer; 10 ml of each reaction was then separated by SDS-PAGE. Gels were fixed and Coomassie stained, and the extent of phosphorylation was assessed by phosphorimager (GE Typhoon).