Two alveolin network proteins are essential for the subpellicular microtubules assembly and conoid anchoring to the apical pole of mature Toxoplasma gondii

Toxoplasma gondii belongs to the coccidian sub-group of Apicomplexa that possess an apical complex harboring a conoid, made of unique tubulin polymer fibers. This enigmatic and dynamic organelle extrudes in extracellular invasive parasites and is associated to the apical polar ring (APR), a microtubule-organizing center for the 22 subpellicular microtubules (SPMTs). The SPMTs are linked to the Inner Membrane Complex (IMC), a patchwork of flattened vesicles, via an intricate network of small filaments composed of alveolins proteins. Here, we capitalize on super-resolution techniques including stimulated emission depletion (STED) microscopy and ultrastructure expansion microscopy (U-ExM) to localize the Apical Cap protein 9 (AC9) and its close partner AC10, identified by BioID, to the alveolin network and intercalated between the SPMTs. Conditional depletion of AC9 or AC10 using the Auxin-induced Degron (AiD) system uncovered a severe loss of fitness. Parasites lacking AC9 or AC10 replicate normally but are defective in microneme secretion and hence fail to invade and egress from infected cells. Remarkably, a series of crucial apical complex proteins (MyoH, AKMT, FRM1, CPH1, ICMAP1 and RNG2) are lost in the mature parasites although they are still present in the forming daughter cells. Electron microscopy on intracellular or deoxycholate-extracted parasites revealed that the mature parasite mutants are conoidless. Closer examination of the SPMTs by U-ExM highlighted the disassembly of the SPMTs in the apical cap region that is presumably at the origin of the catastrophic loss of APR and conoid. AC9 and AC10 are two critical components of the alveolin network that ensure the integrity of the whole apical complex in T. gondii and likely other coccidians.

. Candidates were further filtered by expression patterns similar to AC9 (S1G Fig). 156 TGGT1_292950 exhibited a very similar cyclic mRNA expression pattern compared to AC9 (Fig  157   1E). Upon C-terminal Ty-epitope tagging, the product of the gene colocalized with AC9 in mature 158 parasites and daughter cells and hence was named AC10 (Fig 1F). Like AC9, AC10 was showed to 159 be partially solubilized by sodium carbonate and completely insoluble in Triton X-100 using the 160 AC10-mAID-HA strain, indicating its alveolin network association (Fig 1G). 161 AC9 and AC10 are recruited very early during daughter cells formation 162 ISP1 was localized to the apical cap along with several apical cap proteins (ACs) (6-8). Intriguingly, 163 some ACs are recruited early during daughter cells formation while other appear later, highlighting 164 the existence of a temporal and spatial hierarchy of IMC and alveolin network formation during 165 parasite endodyogeny (6). Colocalization experiments revealed that AC9 and AC10 are recruited very 166 early to the daughter cytoskeleton, prior to ISP1 incorporation (Fig 2A and 2B); in contrast, AC2 and 167 Super-resolution imaging with a stimulated emission depletion (STED) microscope revealed that 178 AC9 and AC10 signals are not homogeneous at the apical cap but rather organized in rows with a 179 regular periodicity (Fig 3A and 3B). Remarkably, IMC1 staining is also organized in regular rows 180 reminiscent of the SPMTs arrangement and the intramembranous particle lattice observed by electron 181 microscopy (EM) (10). Further colocalization of IMC1 with tubulin showed that microtubules are 182 interspaced by two rows of IMC1 (S3A Fig). In contrast, GAP45 staining is more homogeneous along 183 parasite pellicle on the PM side of the IMC. Colocalization with GAP45 and IMC sub-compartment 184 protein 1 (ISP1) (8) confirmed that AC9 is confined to the SPMTs side of the IMC and follows a 185 periodic arrangement like SPMTs and alveolins (S3B Fig). We next applied Ultrastructure Expansion 186 Microscopy (U-ExM) (33) for the first time in T. gondii to gain further resolution. As proof of 187 concept, we first used antibodies specific to alpha/beta tubulin and to poly-glutamylation (Poly-E) to 188 detect the SPMTs. Remarkably, the shape and ultrastructure of the parasite were preserved while the 189 expansion rate was approaching 4x and the SPMTs showed a high level of poly-glutamylation all 190 along their length, except at their most distal part (Fig S3C). Interestingly the conoid fibers seem to 191 be devoid of poly-glutamylation. Of relevance, AC9 and AC10 were clearly colocalizing as a regular 192 pattern between each SPMTs just below the conoid and APR (Fig 3C-E and S3D Fig). Given the 193 close proximity to the SPMTs, AC9 and AC10 were produced recombinantly to assess binding to 194 MTs in an in vitro binding assay (Cytoskeleton ink.). AC9 did not interact directly with MTs (S3E 195 Fig) while recombinant AC10 produced either in bacteria or insect cells formed a gel resistant to 9 solubilization and hence could not be tested. Analysis of the other ACs biotinylated proteins by AC9 197 revealed that AC2 is localized on the SPMTs up to the APR in contrast to AC8, which is present 198 between the SPMTs and does not reach the APR (Fig 3F and 3G). STED microscopy confirmed that 199 AC9 and AC10 are surrounding the APR stained by RNG2 without being in direct contact with it 200 ( Fig 3F). Of interest, RNG2 adopted a regular pattern at the APR both by STED and U- ExM,201 suggesting that the APR is composed of discrete subunits possibly including microtubule plus-end 202 tracking proteins that ensure the apical docking of the SPMTs and their regular interspacing ( Fig 3H  203 and 3I). 204

AC9 and AC10 are essential for invasion and egress and for induced microneme secretion 205
Upon addition of auxin, AC9 and AC10 are tightly downregulated as shown by IFA (Fig 4A and 4D) 206 and WB (Fig 4B and 4E) in the AC9-mAID-HA and AC10-mAID-HA strains. Depletion of both 207 proteins resulted in no lysis plaques in the monolayer of human foreskin fibroblasts (HFF) after 7 208 days (Fig 4C and 4F). Further phenotyping showed that depletion of AC9 and AC10 caused a severe 209 defect in invasion ( Fig 4G) and egress ( Fig 4H) without impacting on parasite intracellular growth 210 and replication (S4A Fig). This phenotype is at least in part explained by the block in induced 211 microneme secretion observed upon stimulation by two known triggers of microneme exocytosis, 212 BIPPO (a phosphodiesterase inhibitor) (34) and ethanol (35) (Fig 4I and S4B Fig). 213

Conditional depletion of AC9 or AC10 causes severe morphological defect of the apical complex 214
In the absence of AC9 or AC10, numerous proteins crucially implicated in motility, invasion and 215 conoid stability were no longer detectable at the apical complex and notably the conoid-associated 216 motor MyoH (20) and the apical polar ring resident protein RNG2 (Fig 5A). Among other apical 217 proteins lost, we found the apical methyltransferase (AKMT) (S5A Fig) (22), the conoid ankyrin-218 repeat containing protein hub 1 (CPH1) (Fig S5B) and the apical actin nucleator FRM1 ( Fig S5C) 219 (15). Conversely, ICMAP1, a MTs binding protein localizing to the intraconoidal microtubules (36) 220 was still detectable but mispositioned ( Fig S5D). Remarkably, all the proteins lost in the conoid of 221 the mature parasite (mother) were still present in the forming daughter cells suggesting that these 222 markers are lost during the last step of daughter cell formation. 223 Extracellular parasites shortly treated with deoxycholate, a detergent which solubilizes the 224 membranes (PM and IMC), offers a better resolution of the parasite cytoskeleton by EM. Strikingly, 225 AC9 treated parasites resulted in the loss of the APR and conoid thus explaining the disappearance 226 of apical markers shown by IFA (Fig 5B).  beginning of the division process, shortly after centrosome duplication. Next, we assessed the fate of 247 APR markers such as KinA and APR1, which are incorporated very early (26) and RNG1 a late 248 marker of division (37). In the absence of AC9 or AC10, APR1 and KinA are lost in mature parasites, 249 but present in daughter cells, respectively (Fig 6D and 6E). We confirmed the late incorporation of 250 RNG1 in daughter cells (Fig 6F and 6G); however, in presence of IAA, RNG1 failed in most of the 251 case, to be properly inserted at the APR shortly before parasites emergence from the mother, resulting Interestingly, IFA performed on intracellular parasites fixed earlier post-invasion revealed that the 255 conoid is not lost in a synchronous manner, at least during the first 2 division cycles ( Fig S6F). Some 256 apical markers could still be observed at the very late stage giving some hints about a possible 257 explanation of the loss of conoid and APR occurring at the end of division and/or daughter cells 258 emergence when newly formed parasites acquire the PM from the mother (Fig S6F). 259

AC9 and AC10 are required for the assembly of the SPMTs at the apical cap 260
Given the periodicity of AC9 and AC10 arrangement between the SPMTs and the loss of the conoid 261 and APR upon depletion of two proteins, we wondered if the overall microtubular structure was also 262 impacted. Deoxycholate extraction revealed that in the AC9 and AC10 depleted parasites, the SPMTs 263 were disconnected ( In this study we have identified two apical cap proteins implicated in the stability of the APR and 273 anchoring of the conoid to the apical complex of mature parasites. AC10 was identified as the most 274 prominent partner of AC9 via proximity biotinylation of AC9-BirA. Conditional depletion of AC9 or 275 AC10 resulted in a very severe phenotype with parasites impaired in microneme secretion and 276 consequently unable to glide, invade and egress from infected cells. Mature parasites lacking AC9 277 and AC10 are conoidless and also deprived of APR but replicate normally. In consequence these 278 mutant parasites also fail to assemble the actomyosin system at the conoid that depend on FRM1 and 279 MyoH. In absence of these two factors, no actin filament is produced and delivered to the glideosome, 280 explaining the extreme severity of the phenotype. 281

ACs are likely components of the alveolin network conserved in coccidians 282
Super-resolution microscopy by STED and U-ExM unambiguously established that AC9 and AC10 283 are not distributed evenly in the apical cap region but organized in longitudinal rows intercalating 284 between the 22 spiraling SPMTs. Fractionation assays indicated that both proteins are poorly soluble 285 consistent with their implication in the alveolin network that spreads between the IMC and the 286 SPMTs. The lack of solubility of AC10 hampered biochemical demonstration that AC9 and AC10 287 form a complex. However, their perfect colocalization and the fact that AC10 depletion photocopied 288 AC9 and caused its destabilization strongly support their functional and physical association. In 289 addition to AC9 and AC10, AC5 (or TLAP3) was previously suggested to arrange in such 290 longitudinal rows in the apical cap region (38). AC9 and AC10 are the earliest markers of parasite 291 division, prior to ISP1, AC1 and AC5. Like AC9 and AC10, AC8 localizes between SPMTs without 292 being in contact with the APR. In contrast AC2 is coating the SPMTs up to the APR suggesting that 293 it is a microtubule-associated protein (MAPs). It would be informative to assess the precise 294 localization of the other ACs even if the corresponding genes are not predicted to be essential (6) to 295 determine the composition of the alveolin network. Of interest, the first described alveolin protein 296 13 IMC1 is also arranged in longitudinal rows and not homogeneously distributed around the parasite 297 (1). IMC1 appears to form two rows between each SPMT, again reminiscent of the double rows of 298 IMPs observed by freeze-fracture studies (10). IMC1 is well conserved (39) and harbors a strong 299 negative fitness score however its function in T. gondii has not been reported. Of note, AC9 and AC10 300 are conserved uniquely in the cyst forming coccidian subgroup of Apicomplexa that includes T. that both AC9 and AC10 surrounds the APR without making significant contacts with it. Surprisingly, 372 the RNG2 signal did not appear to be homogeneous at the APR but rather forming discrete dots by 373 STED and U-ExM. Scale bars = 2µm, except for 1A (bottom panel) and 1H = 0.5 µm. 374