Growth and generation of transgenic T. gondii

T. gondii tachyzoites from the strain RH and derived strains, including RH Δku80/TATi⁶⁴ and RH Δku80/Tir1⁶⁵, were maintained at 37 °C with 10% CO₂ growing in human foreskin fibroblasts (HFFs) cultured in Dulbecco’s Modified Eagle Medium supplemented with 1% heat-inactivated fetal bovine serum, 10 U ml⁻¹ penicillin, and 10 μg ml⁻¹ streptomycin, as described elsewhere⁶⁶. When appropriate for selection, chloramphenicol was used at 20 μM and pyrimethamine at 1 μM. Reporter protein-tagging of endogenous gene loci with reporters 3× HA, 3× v5 and eGFP was done according to our previous work⁶⁷. Reporters were C-terminally fused to their proteins of interest unless this was not tolerated (K13 and KAE-fusions could only be recovered when at the N-terminus). For protein function tests by gene knockdowns, endogenous promoters were replaced with an anhydrotetracycline (ATc)-regulatable t7s4 promoter⁶⁴ or the proteins where tagged with mini-Auxin-inducible degron (mAID) tag⁶⁸ using the same strategy as for the endogenous gene fusions. Oligonucleotides and DNA constructs used for all gene modifications are shown in Supplementary data 3.

Immunofluorescence microscopy

T. gondii-infected HFF monolayers grown on glass coverslips were fixed with 2% formaldehyde at room temperature for 15 min, permeabilized with 0.1% Triton X-100 for 10 min, and blocked with 20% FBS for 1 h. The coverslips were then incubated with a primary antibody for 1 h, followed by 1 h incubation with a secondary antibody. The primary antibodies used were: anti-HA High Affinity (ROAHAHA, Roche 11867423001); anti-V5 (ThermoFisher R960-25); anti-SAG1 (ThermoFisher MA5-18268), anti-GAP45 and anti-IMC1 (both gift from Dominique Soldati-Favre, University of Geneva, Switzerland), all at a dilution 1:250. The secondary antibodies used were: Goat anti-mouse Alexa Fluor 488 (ThermoFisher A-11029), Goat anti-rabbit Alexa Fluor 594 (ThermoFisher A-11012), Goat anti-rat Alexa Fluor 594 (ThermoFisher A-11007), Goat anti-rabbit Alexa Fluor 488 (ThermoFisher A-11008), and Goat anti-rabbit Alexa Fluor 405 (ThermoFisher A-31556), all at a dilution 1:1000. Coverslips were mounted using ProLong Diamond Antifade Mountant (ThermoFisher Scientific, Massachusetts, USA). Images were acquired using a Nikon Eclipse Ti wide-field microscope with a Nikon objective lens (Plan APO, 100×/1.45 oil) and a Hamamatsu C11440, ORCA Flash 4.0 camera. 3D structured illumination microscopy (3D-SIM) was implemented on a DeltaVision OMX V4 Blaze (GE Healthcare, Issaquah, California, USA) with samples prepared as for wide-field immunofluorescence assay (IFA) microscopy with the exception that High Precision coverslips (Marienfeld Superior, No1.5H with a thickness of 170 ± 5 μm) were used in cell culture, and Vectashield (Vector Laboratories, Burlingame, California, USA) was used as mounting reagent. Samples were excited using 405, 488, and 594 nm lasers and imaged with a 60× oil immersion lens (1.42 NA). The structured illumination images were reconstructed in softWoRx software version 6.1.3 (Applied Precision). All fluorescence images were processed using ImageJ version 1.53.

BioID

Sample preparation

For the proximity biotinylation assay, we generated 5 different cell lines (in parental line T. gondii tachyzoites RH Δku80) by in situ genomic N- or C-terminal tagging of one of the 5 bait proteins (K13 and KAE tagged at the N-terminus, AP-2α, AP-2µ and DrpC all tagged at their C-terminus) with the promiscuous bacterial biotin ligase, BirA*. The parental cell line was used as a negative control in biotin treatments. We followed the BioID protocol according to Chen et al.⁶⁹ and our previous work¹⁹. Briefly, the parasites were grown in the elevated biotin concentration (150 μM) for 24 prior to egress, separated from the host cell debris and washed 5× in phosphate-buffered saline. The cell pellets were lysed in RIPA buffer by sonication and the lysates containing ~5mg of total protein content were incubated with 250 µl of unsettled Pierce^TM Streptavidin Magnetic Beads (Thermo-Fisher: 88817) overnight at 4 °C with gentle agitation. The beads were then sequentially treated as follows: washed 3× in RIPA, 1× in 2 M UREA and 100 mM triethylammonium bicarbonate (TEAB; Sigma); washed 3× in RIPA; reduced in 10 mM DTT and 100 mM TAEB for 30 min at 56 °C; alkylated in 55 mM iodoacetamide 100 mM TAEB for 45 min at room temperature in the dark; and washed in 10 mM DTT 100 mM TAEB, followed by 2× 15 min in 100 mM TAEB with gentle agitation. The peptides were digested on the beads by a 1 h 37 °C incubation in 1 μg of trypsin dissolved in 100 mM TAEB, followed by an overnight 37 °C incubation after adding an extra 1 μg of trypsin.

TMT labelling

The peptide concentrations were measured using the Pierce™ Quantitative Fluorometric Peptide Assay (ThermoFisher: 23290) according to the manufacturer’s instructions and 5 μg of the digested peptides were subjected to the tandem mass tag (TMT) labelling using TMT10plex isobaric tagging reagent set (ThermoFisher: 90110). Each TMT reagent vial containing 0.8 mg of the labelling reagents was brought to room temperature and dissolved in 60 µl of LCMS-grade acetonitrile immediately before use. The TMT reagents were then split to three sets without exceeding the labelling capacity, and 20 µl of the TMT reagents were added to each peptide sample. After incubating for 1 hour at room temperature, 8 μl of 5% hydroxylamine (v/v) was added to each sample, followed by incubation for 15 min to quench the reaction. The TMT-labelled fractions were combined and dried in a vacuum centrifuge (Labconco) at 4 °C.

Analysis of TMT-labelled peptides by liquid chromatography and tandem mass spectrometry

LCMS analyses were carried out on an Orbitrap™ Fusion™ Lumos™ Tribrid™ mass spectrometer coupled on-line with a Dionex Ultimate™ 3000 RSLCnano system (Thermo Fisher Scientific) as previously described⁶⁷. The XCalibur v3.0.63 software was used to control the instrument parameters and operation, and record and manage the raw data. The LCMS system was operated in the positive-ion data-dependent acquisition mode with the SPS-MS³ acquisition method with a total run time of 120 min. The dried TMT10plex-labelled peptide samples resolubilized in an LC-MS sample loading solution (0.1% aqueous formic acid) at a concentration of approximately 1 µg/µl. Approximately 1 µg of the sample was loaded onto a micro-precolumn (C18 PepMap 100, 300 µm i.d. × 5 mm, 5 µm particle size, 100 Å pore size, Thermo Fisher Scientific) with the sample loading solution for 3 min. Following the loading step, the valve was switched to the inject position, and the peptides were fractionated on an analytical Proxeon EASY-Spray column (PepMap, RSLC C18, 50 cm × 75 µm i.d., 2 µm particle size, 100 Å pore size, Thermo Fisher Scientific) using a linear gradient of 2–40% (vol.) acetonitrile in aqueous 0.1% formic acid applied at a flow rate of 300 nl/min for 95 min, followed by a wash step (70% acetonitrile in 0.1% aqueous formic acid for 5 min) and a re-equilibration step. Peptide ions were analyzed in the Orbitrap at a resolution of 120,000 in an m/z range of 380–1500 with a maximum ion injection time of 50 ms and an AGC target of 4E5 (MS¹ scan). Precursor ions with the charge states of 2–7 and the intensity above 5000 were isolated in the quadrupole set to 0.7 m/z transmission window and fragmented in the linear ion trap via collision-induced dissociation (CID) at a 35% normalized collision energy, a maximum ion accumulation time of 50 ms and an AGC target of 1E4 (MS² scan). The selected and fragmented precursors were dynamically excluded for 70 s. Synchronous precursor selection (SPS) was applied to co-isolate ten MS²-fragments in the linear ion trap with an isolation window of 0.7 m/z in the range of m/z 400–1200, excluding the parent ion and the TMT reporter ion series. The SPS precursors were activated at a normalized collision energy of 65% to induce fragmentation via high-collision energy dissociation (HCD). The product ions were measured in the Orbitrap at a resolution of 50,000 in a detection range of m/z 100–500 with a maximum ion injection time of 86 ms and an AGC of 5E4 (MS³ scan). The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE⁷⁰ partner repository with the dataset identifier PXD034193.

Raw LCMS data processing

The processing of raw LSMS data for peptide and protein identification and quantification was performed with Proteome Discoverer v2.3 (Thermo Fisher Scientific). Raw mass spectra were filtered converted to peak lists by Proteome Discoverer and submitted to a database search using Mascot v2.6.2 search engine (Matrix Science) against the protein sequences of Homo sapience (93,609 entries retrieved from UniProt on 09.04.2018), Bos taurus (24,148 entries retrieved from UniProt on 17.04.2017), and T. gondii strain ME49 (8,322 entries retrieved from ToxoDB.org release 36 on 19.02.2018)⁷¹. Common contaminant proteins—e.g., human keratins, bovine serum albumin, porcine trypsin—from the common Repository of Adventitious Proteins (cRAP, 123 entries, adapted from https://www.thegpm.org/crap/) were added to the database, as well as the sequence of the BirA* used to generate the BioID bait proteins by gene fusion. The precursor and fragment mass tolerances were set to 10 ppm and 0.8 Da, respectively. The enzyme was set to trypsin with up to two missed cleavages allowed. Carbamidomethylation of cysteine was set as a static modification. The dynamic modifications were TMT6plex at the peptide N-terminus and side chains of lysine, serine, and threonine, oxidation of methionine, deamidation of asparagine and glutamine, and biotinylation of the peptide N-terminus or lysine side chain. The false discovery rate of peptide-to-spectrum matches (PSMs) was validated by Percolator v3.02.1⁷² and only high-confidence peptides (FDR threshold 1%) of a minimum length of 6 amino acid residues were used for protein identification. Strict parsimony was applied for protein grouping.

TMT reporter ion abundances were obtained in Proteome Discoverer using the most confident centroid method for peak integration with 20 p.p.m. tolerance window. The isotopic impurity correction as per the manufacturer’s specification was applied. For protein quantification, PSMs with precursor co-isolation interference above 50% were discarded, and the TMT reporter ion abundances determined for unique (sequence found in proteins belonging to a single protein group) and razor (if sequence is shared by protein belonging to multiple protein groups, the quantification result is attributed to the best-associated Master Protein) peptides were summed.

Statistical analysis of protein enrichment in BioID

Data analysis was performed with R v3.6⁷³ using packages tidyverse v1.2.1⁷⁴ for data import, management, and manipulation, Bioconductor⁷⁵ packages MSnbase v2.10.1⁷⁶ for managing quantitative proteomics data, biobroom v1.16.0 (https://github.com/StoreyLab/biobroom) for converting Bioconductor objects into tidy data frames, and limma v3.40.6⁷⁷ for linear modelling and statistical data evaluation.

The protein-level report generated by Proteome Discoverer was imported into R and filtered to remove non-Toxoplasma and low-confidence (protein FDR confidence level “Low”, q ≥ 0.05). Only Master Proteins with a complete set of TMT abundance values across three replicates of the BioID bait and control samples were considered for the analysis. The protein abundance values in each biological sample were corrected for the total amount using normalization factors derived from the abundances of two proteins, acetyl-CoA carboxylase ACC1 (TGME49_221320) and pyruvate carboxylase (TGME49_284190). Both proteins are highly expressed, endogenously biotinylated, and reside in the matrix of subcellular compartments, the apicoplast and mitochondrion for ACC1 and PC, respectively, where they are not accessible to the BirA*-fused BioID baits. Hence, these two proteins could serve as suitable internal standards. The normalized protein abundances were log2-transformed and modelled as a simple linear relationship between the abundances described by a constant factor (intercept), which is of no interest to us, and the condition (BirA*-tagged vs. control) using limma. If the condition parameter estimated by limma linear model was significantly different from zero we concluded that the condition (presence of the BirA*-fused bait) had a significant effect on the protein abundance. Also, limma estimated the model parameters taking into account the relationship between protein average intensities and the variance (low-abundance proteins tend to have a greater variance) by empirical Bayesian shrinking of the standard errors towards the trend. This enabled a better control of false discoveries and outliers affording more robust identification of significantly enriched proteins. The resulting p-values were adjusted for multiple testing using the Benjamini-Hochberg method (FDR < 1%). Proteins with the adjusted p-value below 0.01 were deemed significantly changing abundance in the BioID bait condition vs. the control, and those of them whose abundance in the BioID bait condition was more than two-fold greater than in the control were considered significantly enriched.

Plaque assay

To test lytic cycle competence of knockdown cell lines by plaque formation in HFF monolayers, 500 freshly lysed parasites were added to T25 flasks containing HFF monolayer. 0.5 μg/ml of ATc was added to induce the gene knockdown, or omitted for uninduced controls. After 7 days of growth, flasks were aspirated, washed once with PBS, fixed with 5 ml of 100% methanol for 5 min and stained with 5 ml of crystal violet solution for 15 min. After staining, the crystal violet solution was removed, and the flasks were washed three times with PBS, dried and imaged. All protein depletion plaque assays were performed three times independently.

Replication assay

For replication assays, the parasites were pre-treated with ATc for 48 h before the egress from the host cell and subsequent invasion of the HFF monolayer growing on coverslips in 6-well plates. Parasites were allowed to grow for further 24 h with ATc. Parasitophorous vacuoles were scored containing either 1, 2, 4, 8 or 16 parasites. A minimum of 200 parasitophorous vacuoles were scored for each of the three biological replicates. P values were calculated with the Student’s T-test using STATA v.14, and the bar graphs were drawn using GraphPad Prism, v8 (GraphPad, California USA).

Host cytosolic protein uptake assay

Inducible mCherry HeLa cells previously described⁴⁸ were seeded into 6-well plates at a density of 1.5 × 10⁵ cells per well. Cytosolic mCherry expression was induced for 4 days by adding 2 µg mL⁻¹ of doxycycline each day. Prior to infection of host cells expressing cytosolic reporter, parasites were treated with 0.5 µg mL⁻¹ anhydrotetracycline (ATc) or vehicle control for 48 h. Additionally, parasites were treated with 5 µM of the protease inhibitor LHVS for 24 h to inhibit the degradation of material delivered to the VAC. Cells were infected with 1.0 x10⁶ parasites and allowed to replicate for 24 h in the presence of ATc or vehicle control and LHVS. Parasites were harvested for analysis by scraping and syringing the infected monolayer on ice followed by filtration and centrifugation at 1500 × g for 10 min at 4 °C. Isolated parasites were then treated with a 1 mg mL⁻¹ pronase and 0.01% saponin-1× PBS solution for 1 h at 12 °C, centrifuged and washed 3× before adding to Cell-Tak coated slides. Extracellular parasites were fixed with 4% paraformaldehyde for 10 min and permeabilized with 0.1% Triton X-100 for 10 min prior to imaging. At least 200 parasites were analyzed per sample and the percentage of parasites positive for mCherry was quantified by dividing the number of parasites containing mCherry signal derived from the host cytosol divided by the total number of parasites analyzed.

Uptake of parasite surface protein assay

Internal tagging of SAG1 (SRS29B TGME49_233460) by transient CRISPR-Cas9 expression

SAG1-Halo and iΔHA‐TgK13-SAG1-Halo were generated by tagging SAG1 in the Δku80 and Δku80 iΔHA‐TgK13 line using transient Cas9 transfection targeting the following gRNA: TGCAGCCCCGGCAAACTCCAC(GGG). The strain was generated as previously described for Cas9 tagging⁷⁸. Briefly, gRNA oligos were annealed and ligated into the Cas9 vector and verified by sequencing (Eurofins Genomics). Reparation template DNA were generated by amplifying the halo with homology arm (50bp) with SAG1 by PCR using Q5 High-Fidelity DNA Polymerase (New England BioLabs). The repair template was purified using a PCR purification kit (Blirt). Parasite transfection, sorting and screening for positive mutants was done as previously⁷⁸. Briefly, newly released RHDiCreΔku80 or Tet-K13-Ha Δku80 tachyzoites were transfected with the repair template and 10–12 μg of SAG1-Cas9-YFP. The parasites were mechanically egressed 24 to 48 h after transfection, passed through a 3 μm filter, and those transiently expressing Cas9-YFP enriched via FACS 23 (FACSARIA III, BD Biosciences) into 96-well plates (a minimum of 3 events per well). Resultant clonal lines were screened by IFA for SAG1 labelling and integration was confirmed by PCR.

Specific labelling of the plasma membrane SAG1

SAG1-Halo strain was labelled with either the non-permeable Alexa 660 (1/1000) or permeable Oregon green (1/4000) dyes (Promega) in cold media for 1h. Parasites were then washed prior to transfer to Ibidi live-cell dishes (29 mm) coated with 0.1% poly-L-Lysine as previously described³¹. After letting the parasite settle for 15 min, parasites where imaged live on a Leica-DMi8 wide-field microscope attached to a Leica DFC9000 GTC camera, using a 100× objective to compare the difference between permeable and non-permeable dye labelling.

SAG1-endocytosis assays

For extracellular tachyzoites, the SAG1-Halo strain was labelled with the non-permeable Alexa 660 dye (1/1000) in cold media for 1 h. Parasites were wash 3× to remove excess ligand. Parasites were incubated at 4 °C or 37 °C on the FBS coated live-cell dishes (Ibidi live-cell dishes, 29 mm) for 1 h prior to live imaging, as described below, for evaluation of internalization of the SAG1 signal. For SAG1 recycling assays of intracellular tachyzoites, parasites were labelled with Alexa 660 as above and then inoculated onto HFF monolayers in Ibidi dishes. Parasites were allowed to replicate for 24 h prior to imaging. For endocytosis assays with K13-depletion, iΔHA‐TgK13-SAG1-Halo parasites were first induced for 48 h with or without ATC, before mechanical egress, filtering and labelling. Parasites were then allowed to reinfect new HFFs cells, grow for 24 h under the same ATc treatment and were then imaged live, as described below, for evaluation of internalization of the SAG1 signal. Endocytic activity was assessed by the presence of the PM-SAG1 vesicles (non-permeable halo-Alexa Fluor®-660) and the percentage of parasites showing the presence of vesicles was determined. Mean values of three independent experiments ± SD were determined. The mean number of PM-SAG1 vesicles/parasite was determined from the total number of vesicles inside vacuoles divided by the number of parasites in the vacuole. For membrane accumulation, we scored aggregation of PM-SAG1 as signal in the residual body or outside the typical parasite plasma membrane. For these analyses at least 25 vacuoles per replicates were used and mean values of three independent experiments ± SD were determined. All SAG1-Halo images were acquired on a Leica-DMI8, objective 100x with the LasX software (v3.7.4). Imaged were deconvolved using Huygens essential software (v18.04) and batch express processing. Fiji (v1.53c) was used to analyze the picture and all count were made manually.

Egress assay

iΔHA‐TgK13 HP03-eGFP parasites were seeded in a T25 flask and treated with ATc for 48 h or vehicle (DMSO) for the negative control. After 48 h the extracellular parasites were removed, and the intracellular parasites were mechanically egressed by needle-passed, reseeded into the ibidi live imaging chamber, and then grown for a further 24 h with ATc or DMSO. Images of 10 fields of view were taken using the transmembrane protein HP03-eGFP as a marker for the parasites. Next, 5 µM of the Ca²⁺ ionophore A23187 was added and incubated for 10 min at 37 °C before the same 10 fields of view were imaged. The number of vacuoles and parasites per vacuole were then counted in the first set of images (before adding the Ca²⁺ionophore) and the percentage of successfully egressed parasites quantified. P values were calculated with the Student T-test using STATA v.14, and the bar graphs were drawn using GraphPad Prism, v8 (GraphPad, California, USA).

Electron microscopy

Immuno-TEM

3v5-mAID-K13 egressed parasites were harvested by centrifugation, washed with phosphate buffer (0.2 M Na₂HPO₄, and 0.2 M NaH₂PO₄ with a Ph of 7.24), and then fixed for 30 min at RT (4% formaldehyde). The subsequent processing was performed at 4 °C where cells were dehydrated in 30%, 50% and 70% ethanol (2× for 5 min each), infiltrated with hard LRW resin (Agar Scientific) that was cured in an embedding oven at ~55 °C overnight. Ultrathin sections were cut using a Leica Ultracut ultramicrotome and collected on 200 mesh nickel grids covered with a carbon film (EM resolutions). The grids were blocked for 30 min at RT in 0.8% BSA/PBS/0.01% Tween 20 (blocking buffer) then incubated with the primary antibody (anti-V5 mouse, ThermoFisher R960-25) diluted in blocking buffer at 1:10 in a humid chamber at 4 °C overnight. The grids were washed 3× for 5 min each in washing buffer (PBS/0.01% Tween 20) and then incubated with the goat anti-mouse IgG H&L (10 nm Gold) secondary antibody (Abcam ab39619, 1:20 in blocking buffer) for 1h at RT. Negative controls were incubated with the secondary antibody only. Grids were washed 3x for 5 min each in a drop of washing buffer, then passed over two drops of PBS and fixed on a drop of 3% glutaraldehyde/PBS for 10 min. Then the grids were washed 3× in deionized water, allowed to air dry and were then post-stained in 2% aqueous uranyl acetate and Reynold’s lead citrate for 5 min each. Samples were viewed in a Tecnai G2 transmission electron microscope run at 200 keV with a 20 µm objective aperture for improved contrast and images were acquired using an AMT CCD camera.

Ultrastructural TEM

For iΔHA‐TgK13-SAG1-Halo parasites with or without K13 depletion, the parasites were induced for 48 h with or without ATc, released mechanically and filtered prior transfer to Ibidi μ-dishes previously seeded with HFF cells. After 24 h of replication the parasites were fixed with 2.5% glutaraldehyde in 0.1 M phosphate buffer pH 7.4. The parasites were washed three times at room temperature with PBS (137 mM NaCl₂, 2.7 mM KCl, 10 mM Na₂HPO₄, 1.8 mM KH₂PO₄, pH 7.4) and post-fixed with 1% (w/v) osmium tetroxide for 1 h. Subsequent to washing with PBS and water, the samples were stained en bloc with 1% (w/v) uranyl acetate in 20% (v/v) acetone for 30 min. Samples were dehydrated in a series of graded acetone and embedded in Epon 812 resin. Ultrathin sections (thickness, 60 nm) were cut using a diamond knife on a Reichert Ultracut-E ultramicrotome. Sections were mounted on collodium-coated copper grids, post-stained with lead citrate (80 mM, pH 13) and examined with an EM 912 transmission electron microscope (Zeiss, Oberkochen, Germany) equipped with an integrated OMEGA energy filter operated in the zero-loss mode at 80 kV. Images were acquired using a 2k × 2k slow-scan CCD camera (Tröndle Restlichtverstärkersysteme, Moorenweis, Germany).

Ultrastructural SEM

DrpC-mAID-3v5 and wildtype parasites were grown with auxin treatment (500 µM) for 48 h, mechanically egressed and then reinfected fresh host cells in 35 mm ø Ibidi dishes with plastic coverslips. A further 24 h of growth was allowed with continued auxin treatment and then cells were processed for EM in these culture dishes. Cells were fixed in 2% glutaraldehyde/2% formaldehyde in 0.05 M sodium cacodylate buffer pH 7.4 containing 2 mM calcium chloride overnight at 4 °C. After washing 5× with 0.05 M sodium cacodylate buffer pH 7.4, samples were osmicated (1% osmium tetroxide, 1.5% potassium ferricyanide, 0.05 M sodium cacodylate buffer pH 7.4) for 3 days at 4 °C, washed 5x in DIW (deionised water), then treated with 0.1% (w/v) thiocarbohydrazide/DIW for 20 min at room temperature in the dark. After washing 5× in DIW, samples were osmicated a second time for 1 hour at RT (2% osmium tetroxide/DIW), washed 5× in DIW, then blockstained with uranyl acetate (2% uranyl acetate in 0.05 M maleate buffer pH 5.5) for 3 days at 4 °C. Samples were washed 5× in DIW and then dehydrated in a graded series of ethanol (50%/70%/95%/100%/100% dry) and 100% dry acetonitrile, 3× in each for at least 5 min. Samples were infiltrated with a 50/50 mixture of 100% dry acetonitrile/Quetol resin (without Benzyldimethylamine [BDMA]) overnight, followed by 3 days in 100% Quetol (without BDMA). Then, the sample was infiltrated for 5 days in 100% Quetol resin with BDMA, exchanging the resin each day. The Quetol resin mixture is 12 g Quetol 651, 15.7 g NSA, 5.7 g MNA and 0.5 g BDMA (all from TAAB). The Ibidi dishes were filled with resin to the rim, covered with a sheet of Aclar and cured at 60 °C for 3 days. The sample blocks were then cut from with a hacksaw and ultrathin sections cut and collected on 300 mesh copper grids. Samples were imaged in a Verios 460 scanning electron microscope (FEI/Thermo Fisher Scientific) run at an accelerating voltage of 4 keV and 0.2 nA probe current using the concentric backscatter detector in field-free mode (low magnification) or immersion mode (high resolution).

Binding of α ear appendage domains to WxxF motif of TgEps15L

The construction of a GST fusion of MmAP-2α ear appendage domain has been previously described⁷⁹ and encompasses amino acids 695–938 of mouse AP2A2. Equivalent domain boundaries for TgKAE (TGME49_272600) and TgAP-2α (TGME49_221940) ear appendage domains were selected based on homology to MmAP-2α. TgKAE (amino acids 1406–1672 end) and TgAP-2α (amino acids 1015–1348 end) appendage domains were made synthetically and codon optimized for expression in Escherichia coli; adding BamH1 and Xho1 sites for TgAP-2α and BamH1 and Sal1 for TgKAE for in-frame cloning into pGEX4T-1. For fusion protein production, the DNA was transformed into BL21(DE3) competent cells for high protein expression and induced mid-log phase with 0.4 mM IPTG 16–20 h at 22 °C. Fusion proteins were recovered using Glutathione-Sepharose 4B (GE Healthcare) and eluted from the beads using 30 mM reduced glutathione. His-tagged fragment of TgEps15L (TGME49_227800) encompassing amino acids 897–1023 (952-WxxF), whose boundaries were predicted using DomPRED (URL:http://bioinf.cs.ucl.ac.uk/software.html) was made synthetically and codon optimized for expression in E. coli. The fragment was cloned using BamH1 and EcoR1 (adding a stop codon) into pTrcHisA. For fusion protein production, the DNA was transformed into BL21(DE3) competent cells (Invitrogen), induced mid-log phase with 0.4 mM IPTG, purified using Ni-nitrilotriacetic acid agarose beads (QIAGEN) and eluted with 250 mM imidazole. Point mutation W952A (952-AxxF) was made using site-directed mutagenesis.

For the far-western blot assay, 2.5 μg of the His-tagged fusion proteins were subjected to SDS-PAGE on 16% Bolt Bis-Tris gel (ThermoFisher) in MOPS running buffer gels and blotted onto a nitrocellulose membrane. The blot was blocked with 20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 0.05% Tween 20, 0.5% BSA, 3 μM reduced glutathione for 30 min and this buffer was used in all the following steps. The blot was incubated with 10 nM GST, GST-MmAP-2α, GST-TgAPale or GST-TgAP-2α for 1h, washed for 30 min, and then labelled with 1:30,000 anti-GST (Invitrogen 700775) followed by 1:10,000 anti-rabbit-HRP (Invitrogen 31460) and developed using ECL Prime (GE Healthcare) and X-ray film.

Structural predictions

The 3D structure of TgKAE ear appendage domain was modelled using SWISS-MODEL⁸⁰ by using as template the mouse Alpha-adaptin Appendage Domain (PDB 1w80; chain A, sequence identity 31%). Using the positions of the bound peptides (FxDxF and WxxF) in 1w80, a resultant TgKAE peptide complex structure was rebuilt and further energy minimized in coot⁸¹. Figures for the protein structures were drawn with Pymol⁸², and the protein-protein interaction networks were generated with Ligplot+⁸³.

Homology searching, domain detection and phylogenetic analyses

Protein homologues were searched with the iterative version of the profile hidden Markov models (HMMs) search engine (jackhammer)⁸⁴. Conserved domains were detected with InterProScan⁸⁵ and the similarity levels between the T. gondii and human BTB, Kelch, Adaptin_N and α_C2/αC (ear) domains were evaluated using a pairwise protein BLAST (https://blast.ncbi.nlm.nih.gov/Blast.cgi). For the phylogenetic analyses, sequences were aligned using Mafft v7.407 with the L-INS-i algorithm⁸⁶. Phylogenetic analyses were performed locally and using the CIPRES online portal⁸⁷. Alignments were masked and trimmed manually using Mesquite v3.6 (https://www.mesquiteproject.org) or Jalview (https://www.jalview.org). Bayesian analysis was carried out using MrBayes v3.2.7a35 hosted on CIPRES. Datasets were run under a mixed model with four independent runs of four chains each, sampling every 500 generations up to a total of either 1,000,000 or 10,000,000 MCMC generations, depending on convergence criteria. Trees were summarized discarding the first 20% of samples as burn-in; convergence was assessed by manually expecting the generation vs. log-likelihood plots for stationarity, as well checking parameter PSRF values for various parameters. Maximum-likelihood (ML) rapid boostrapping was carried out using RAxML v8.2.1232 (LG+Γ model, rapid bootstrapping) hosted on CIPRES or locally using PhyML-3.1 or IQ-TREE v1.6.12 (1000 bootstraps, with the optimal substitution model)^88,89. SH-like aLRT and aBayes branch support values were obtained in PhyML-3.1.

Statistics and reproducibility

All experiments were performed as three independent replicates. The details on statistical analyses used are provided in figure legends where applicable. To identify the differentially enriched proteins in the BioID experiments (Supplementary data 2), the normalized log2-transformed protein abundance values measured for the BirA*-tagged and negative control samples were compared using a moderated one-sided t-test in the limma linear modelling framework. To compute the moderated t-statistics, the variance estimates were moderated across proteins using an empirical Bayes method of shrinking the protein-wise residual variances towards a global trend. The computed p-values were adjusted for multiple comparisons according to the Benjamini-Hochberg method. The results of the experiments that were not statistically analyzed (immunofluorescence assays, electron microscopy analyses, western blot analyses; Figs. 1, 2B, 3A, B, 4A, B, 6D, 7C, Figs. S1, S2A,C-E, S3, S4B, S5A, B, S6A, B, S7A, S9C) were similar for all replicates and representative images are shown for each experiment. For microscopy analyses, over 100 individual parasites/vacuoles were observed for each slide (i.e. a single replicate), and the results were consistent for each replicate.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.