The nucleocytosolic O-fucosyltransferase Spindly affects protein expression and virulence in Toxoplasma gondii

Once considered unusual, nucleocytoplasmic glycosylation is now recognized as a conserved feature of eukaryotes. While in animals O-GlcNAc transferase (OGT) modifies thousands of intracellular proteins, the human pathogen Toxoplasma gondii transfers a different sugar, fucose, to proteins involved in transcription, mRNA processing and signaling. Knockout experiments showed that TgSPY, an ortholog of plant SPINDLY and paralog of host OGT, is required for nuclear O-fucosylation. Here we verify that TgSPY is the nucleocytoplasmic O-fucosyltransferase (OFT) by 1) complementation with TgSPY-MYC3, 2) its functional dependence on amino acids critical for OGT activity, and 3) its ability to O-fucosylate itself and a model substrate and to specifically hydrolyze GDP-Fuc. While many of the endogenous proteins modified by O-Fuc are important for tachyzoite fitness, O-fucosylation by TgSPY is not essential. Growth of Δspy tachyzoites in fibroblasts is modestly affected, despite marked reductions in the levels of ectopically-expressed proteins normally modified with O-fucose. Intact TgSPY-MYC3 localizes to the nucleus and cytoplasm, whereas catalytic mutants often displayed reduced abundance. Δspy tachyzoites of a luciferase-expressing type II strain exhibited infection kinetics in mice similar to wild type but increased persistence in the chronic brain phase, potentially due to an imbalance of regulatory protein levels. The modest changes in parasite fitness in vitro and in mice, despite profound effects on reporter protein accumulation, and the characteristic punctate localization of O-fucosylated proteins, suggest that TgSPY controls the levels of proteins to be held in reserve for response to novel stresses.


Introduction
Toxoplasma gondii is an obligate intracellular protist classified as a Category B pathogen on NIAID's list of emerging infectious diseases (1). Based on serological evidence, the CDC estimates that T.
gondii has infected ~11% of the United States population, while infection rates in other countries may exceed 50% (2,3). Humans can become infected with T. gondii when they ingest food contaminated with oocysts shed in the feces of cats (where the sexual cycle occurs) or tissue cysts present in undercooked meat (4). T. gondii infection is generally transient and presents with no or mild symptoms, except in immunocompromised individuals and in fetuses; in these cases severe ocular and neurologic disease may occur (5,6). Because it is difficult to prevent T. gondii infection, congenital transmission remains a significant problem in the US and around the world, together with encephalitis caused by reactivation of the parasite in immunosuppressed individuals (7,8). Even though prior exposure is believed to lead to immunity against reinfection, significant challenges persist: there is no vaccine against T. gondii (9), current drug treatments are effective only in acute infections, and development of drug resistance remains a concern (10).
T. gondii goes through a series of differentiation events in both its asexual and sexual cycles (1).
During asexual development in humans and other intermediate hosts, tachyzoites (the fast replicative form of the parasite) differentiate to bradyzoites (the slow replicative form) when exposed to unfavorable conditions, e.g. immune system response or reduced availability of nutrients (11). T. gondii bradyzoites have a slower metabolic rate and organize in cysts, which are characterized by a wall that is rich in glycoconjugates and are found predominantly in muscles and in the brain (12,13). Importantly, both drug treatment failure and latency are linked to parasite persistence in tissue cysts (11).
Transcriptomics and proteomic studies of developmental stages of T. gondii have identified many stage-specific genes and proteins, indicating the importance of tightly regulated expression patterns (14,15). In support of this idea, master regulators of bradyzoites differentiation (BDF1) and sexual commitment (MORC) have been recently identified (13,16). Many of the downstream transcriptional and a few translational factors in the parasite have been shown to be regulated by post-translational modifications (PTMs), including phosphorylation, acetylation and ubiquitination (17)(18)(19)(20). While transcriptional regulation of protein expression has been the subject of many studies in T. gondii, the role that PTMs might play in regulating protein stability, concentration, and/or localization (protein homeostasis) are less understood.
Our recent studies identified O-linked fucose (O-Fuc) as an additional PTM that modifies T. gondii proteins that are involved in regulation of protein expression and cell signaling. We showed that the Lfucose-specific lectin from Aleuria aurantia (AAL) highlights the nucleoplasm and in particular the nuclear periphery of RH, a T. gondii type I strain, but not the secretory system, in contrast to what is observed in most eukaryotic cells (21,22). Super-resolution microscopy showed that O-fucosylated proteins localize in a punctate pattern in the vicinity of nuclear pore complexes (NPCs), suggesting they might be forming assemblies. Mass spectrometry of parasite AAL-enriched proteins revealed a high incidence of one or more fucose residues O-linked to Ser/Thr on peptides from intrinsically disordered regions (IDRs) on 69 proteins, including Phe/Gly-repeats nucleoporins (FG-Nups), mRNA processing enzymes such as members of the CCR4-NOT complex, transcriptional regulators such as a subunit of the transcription initiation complex, and cell-signaling proteins such as kinases and ubiquitination factors (23). A yellow fluorescent proteinfusion containing a Ser-rich domain (SRD-YFP) from a T. gondii GPN-loop GTPase (TgGPN), which is 79-amino acids long and contains 37 consecutive Ser residues, was found to be modified with O-Fuc and partially associated with AAL-labeled assemblies, while fucosylation was not detected on YFP with a nuclear localization signal (NLS) (22). The proteins and sequences modified with O-Fuc in T. gondii closely resemble the substrates of the human OGT (HsOGT). O-GlcNAcylation is important in disease-relevant signaling and enzyme regulation and is the best-studied example of intracellular glycosylation (24,25).
Subsequent to the discovery of O-fucosylated proteins in the nucleus of T. gondii, the Arabidopsis thaliana SPINDLY (AtSPY), a paralog of animal nucleocytoplasmic O-GlcNAc transferases (OGTs), was shown to be an O-fucosyltransferase (OFT) that activates the nuclear growth repressor DELLA (26)(27)(28).
OGTs, AtSPY, and its T. gondii orthologue (TgSPY) each have a C-terminal Carbohydrate Active EnZyme (CAZy) glycosyltransferase 41 (GT41) catalytic domain (29) and N-terminal tetratricopeptide repeats (TPRs), 13.5 repeats in HsOGT and 11 in both AtSPY and TgSPY (30). CRISPR/Cas9-mediated knockout of TgSPY resulted in a modest growth phenotype, despite complete loss of AAL labeling of nuclei and parasite extracts (31). These results, pointing to OFT activity of both TgSPY and AtSPY, were quite unexpected because of their sequence homology to OGTs and because members of the same CAZy GT family usually utilize activated sugar donors with a conserved nucleotide moiety, e.g. either UDP-or GDPsugars (29).
To better understand the cell and molecular biology of O-fucosylation in T. gondii, we tested the activity of TgSPY either produced as a recombinant enzyme in bacteria or expressed in the Δspy strain, with and without point mutations known to affect either host OGT or AtSPY activities. We determined the effect of the TgSPY knockout on the stability and localization of proteins bearing domains normally modified with O-Fuc and on the growth and differentiation of Δspy strains in culture and in mice.

Results
The OFT activity of His-tagged TgSPY (TgHis6SPY) purified from the cytosol of bacteria was shown in four ways. First, using GDP-Glo and UDP-Glo assays, in which hydrolysis of the nucleotidesugar produces a luminescent signal, we showed that TgHis6SPY (Fig. S1), in the absence of an acceptor substrate, hydrolyzed GDP-Fuc but not UDP-GlcNAc (donor for OGT), GDP-Man, or UDP-Gal ( Fig. 1A) (32). As a positive control, UDP-Gal was hydrolyzed by a T. gondii galactosyltransferase (TgGat1), which adds a terminal galactose to the pentasaccharide attached to the T. gondii Skp1 E3 ubiquitin ligase subunit (33). Second, in the presence of GDP-Fuc, TgHis6SPY fucosylated itself (OFT activity in cis), as shown by Western blotting with AAL (Fig. 1B). As a control, AAL binding to TgHis6SPY was inhibited by α-methyl L-fucopyranoside (Me-αFuc). To confirm that Fuc became O-linked to the side chain of Ser or Thr, we took advantage of knowledge that such linkages are uniquely susceptible to mild base with release of the Olinked Fuc and its conversion to fucitol. To detect the presence of fucitol, the product mixture from an autofucosylation reaction conducted in the presence of GDP-[ 3 H]Fuc was subjected to β-elimination and the released material was chromatographically analyzed by HPAEC. As shown in Fig. 1C, all detectable radioactivity co-eluted with the fucitol standard, which was detected by pulsed amperometry. This finding confirmed that radioactivity was incorporated as Fuc and strongly supported its linkage to Ser or Thr.
Third, to test for OFT activity in trans, we transferred the SRD of TgGPN, which is O-fucosylated when fused to YFP in T. gondii tachyzoites (22), to glutathione-S-transferase (GST) to generate GST-SRD.
The in vitro activity of TgHis6SPY strongly supports the idea that TgSPY is the OFT that directly mediates assembly of the O-Fuc linkage in cells, as shown previously for AtSPY (26).

Complementation of ∆spy tachyzoites with myc-tagged TgSPY restored O-Fuc modification
of nucleocytosolic proteins. To further demonstrate that TgSPY is the OFT, we complemented the Δspy RH strain (31) with a C-terminally MYC3-tagged copy of TgSPY cDNA, which was randomly integrated into the genome. The strategy is illustrated in Fig. 2A, and PCR validation is shown in Fig. S2A. Expression of TgSPY-MYC3 in ∆spy tachyzoites cultured in human foreskin fibroblasts (HFFs) restored AAL binding to O-Fuc proteins of tachyzoites, as detected by structured illumination microscopy (SIM) (Fig. 2B) and lectin blotting (Fig. 2C). While there was no increase in the number of AAL-labeled assemblies in the complemented SPY knockout strain (∆spy::TgSPY-MYC3) versus wild type, there was an increase in the overall intensity of the AAL labeling (Fig. 2D). Most likely, this result can be explained by the use of the highly active tubulin promoter to express TgSPY-MYC3 ( Fig. 2A). Consistent with the unchanged number of assemblies, AAL-labeled punctae closely overlapped with YFP-tagged Nup67, which is an FG-Nup that is not modified by O-Fuc (Fig. 2E). Previously, we observed that AAL labeling partially overlapped with YFP-tagged Nup68, an FG-Nup that is extensively modified with O-Fuc (22).
TgSPY-MYC3 was present in the cytosol, nucleoplasm, and residual body (arrow) of G1 and dividing stages of ∆spy::TgSPY-MYC3 parasites (Fig. 2F). Similar to the O-Fuc-modified proteins visualized with AAL, TgSPY-MYC3 was excluded from the nucleolus (asterisks). Western blot analysis of the total cell lysate from ∆spy::TgSPY-MYC3 showed TgSPY-MYC3 was full-length with the expected apparent Mr of 112,000 (Fig. 3B). Based on clonal plaque formation in fibroblast monolayers (Fig. 2G), there was a modest decrease in growth in ∆spy versus wild-type (RH) tachyzoites, confirming a previous report (31), and a modest increase in the ∆spy::TgSPY-MYC3 strain, both of which were statistically significant.
These complementation studies confirm that TgSPY is the OFT that modifies nucleocytosolic proteins recognized by AAL and provide additional evidence that AAL-labeled assemblies co-localize with NPCs (22).

Mutations in
TgSPY affected its OFT activity, its abundance, and its location. The viability of ∆spy parasites allowed us to investigate the biochemistry of the OFT directly in T. gondii tachyzoites. We generated a series of point mutations in the C-terminal GT41 catalytic domain of TgSPY-MYC3 (D619, H623 and K793) that correspond to D554, H558 and K842 in HsOGT (Fig. 3A). Mutation of either H558 or K842 to Ala abolishes glycosyltransferase activity in human and bacterial OGTs (34,35). Structural studies show D554 is linked to its peptide substrate through a chain of water molecules, suggesting that this residue might be involved in catalysis (30). The point mutations E692K and G695D correspond to mutations in two A. thaliana spy hypomorphic alleles, spy12 (G570D) and spy15 (E567K), that have each been shown to result in complete loss or a strong reduction in OFT activity, respectively (26). Because biochemical data on AtSPY have been obtained from experiments performed with truncated enzymes containing the C-terminal 3 TPRs (26), we also tested the behavior of a truncated TgSPY-MYC3 containing 8 In Western blots of lysates of tachyzoites, all TgSPY-MYC3 mutants were detected by anti-MYC antibodies, although the expression levels varied markedly (Fig. 3B). While D619A and H623A were present at levels comparable to the wild type, the other mutants (K793A, E692K, G695D, and the SPY-3TPRs) were present at much lower levels than wild type. By SIM, we observed that all of the point mutants of TgSPY-MYC3, even those with very low expression levels, were located in both the cytoplasm and the nucleus, as is the case for wild-type TgSPY-MYC3. In contrast, no truncated TgSPY-3TPRs was observed in the nucleus, suggesting that the N-terminal TPRs might be involved in import of TgSPY, which lacks a canonical NLS (36). AAL labeling by SIM was used to estimate TgSPY activity. D619A and H623A mutants showed markedly reduced AAL labeling while only residual staining was observed in E692K and G695D. AAL labeling was absent from both the K793A mutant and the truncated SPY-3TPRs (Fig. 3C).
To confirm the occurrence of the expected negative effects of the point mutations on SPY OFT activity, N-terminally His6-tagged versions of the mutant isoforms were expressed in and partially purified from E. coli. The E692K and G695D mutants expressed with very low yield of soluble protein, and the K793A expressed at a low but better yield, which paralleled their expression levels in the parasite. A new assay that measured transfer of [ 3 H]Fuc from GDP-[ 3 H]Fuc to a synthetic peptide from TgRINGF1 (TGGT1_285190), previously shown to be O-fucosylated in vivo (22), was developed. Relative to the wildtype protein prepared in parallel, the same concentration of each of the five point-mutants was found to exhibit no activity within the sensitivity of the assay (Fig. 3D), which is ~10% of full activity.
We conclude that residues important for catalytic activity in HsOGT and AtSPY are also important in TgSPY. Further, these point mutations appear to result in loss of protein stability in both the parasite and in E. coli, an effect that may be compounded by loss of potentially stabilizing auto-fucosylation. Finally, nuclear targeting of SPY appears to depend on its N-terminal TPR domains, which are also known to contribute to acceptor substrate selection and dimerization for OGT (37).

Endogenous and exogenous proteins normally modified with O-Fuc were less abundant in the
absence of SPY or the fucosylated domain. The goal here was to determine whether the absence of O-Fuc in SRDs of nucleocytosolic proteins affects their abundance and/or location. In our first experiment, we attempted to express our previously described SRD-YFP construct, which contains the 79-amino acidlong SRD from the N-terminus of TgGPN that is O-fucosylated in wild type cells, in ∆spy tachyzoites. For comparison, we also expressed in ∆spy tachyzoites NLS-YFP, which contains a nuclear localization sequence and is not O-fucosylated (22). Although we were able to generate stable parasite lines expressing NLS-YFP, we could not obtain stable lines expressing SRD-YFP in ∆spy tachyzoites. SIM analysis of SRD-YFP was, therefore, performed by fixing fibroblast monolayers containing parental, ∆spy and ∆spy::TgSPY parasites 24 h after electroporation (Fig. 4A). SRD-YFP fluorescence was strongly reduced in ∆spy parasites and restored in ∆spy::TgSPY parasites. The residual labeling in ∆spy parasites occurred in the cytoplasm and in perinuclear puncta, in contrast to the normal enrichment in the nucleus. NLS-YFP fluorescence was comparable between parental and ∆spy parasites by SIM (Fig. 4B) and Western blotting (not shown). These results suggested that the SRD destabilizes YFP unless it is modified with O-Fuc.
In our second experiment, we stably expressed YFP-tagged Nup68 under the tubulin promoter by random integration. In wild type RH, Nup68-YFP, which is normally heavily modified with O-Fuc (22), localized mainly in a punctate pattern at the nuclear periphery ( Fig. 4C) and was detected by Western blot as a single band consistent with the full length fusion protein (Fig. 4D). In contrast, in ∆spy parasites, Nup68-YFP was barely detectable by IFA (Fig. 4C), and the protein was extensively degraded (Fig. 4D).
These observations suggest that Nup68-YFP undergoes degradation in the absence of O-Fuc, as shown for SRD-YFP. The paucity of Nup68-YFP in ∆spy parasites made it impossible to judge whether the absence of O-Fuc affected its localization.
In a third experiment, the full-length TgGPN, as well as a truncated protein missing the N-terminal SRD (TgGPN∆SRD), were each tagged at the C-terminus with c-MYC3 and stably expressed under their own promoter after random integration into the genome (Fig. S2). In TgGPN∆SRD translation was initiated at an unconserved Met residue at the beginning of the G1 motif near the start of the yeast and bacterial versions of the protein (Fig. S2D-E). Full-length TgGPN accumulated at a higher Mr than expected (49,000), based on Western blotting of two independent clones in the RH background (Fig. 4E). In contrast, TgGPN∆SRD was not detected by Western blotting, even after AAL enrichment (Fig. 4F). RT-PCR demonstrated similar stable expression of the transcripts of TgGPN and TgGPN∆SRD (Fig. 4G), suggesting potential to express the protein. However, only a low level of expression could be detected by flow cytometry targeting the c-MYC tag (Fig. 4I). These findings are consistent with a model in which the addition of the N-terminal SRD evolved as a mechanism to allow increased translation or slower degradation kinetics of TgGPN when O-fucosylated. It should be noted that misfolding of TgGPN∆SRD due to the absence of the SRD cannot be ruled out.
SPY modulates the growth of the type II CZ1 strain in culture and in mice. Because type I RH strain does not efficiently differentiate to bradyzoites in culture and is too virulent for mouse infection studies, we generated a spy knockout in a type II CZ1 ∆ku80 strain, which was also engineered to constitutively express a firefly Re9 luciferase under control of the gra1 promoter (see When cloned on fibroblast monolayers, plaques formed by ∆spy CZ1 were on average 50% smaller than those of the parental Re9 strain (Fig. 5B), pointing to a stronger growth defect in this type II strain than observed for the type I RH strain (Fig. 2F). Using SAG2Y as a bradyzoite marker, we found no difference in spontaneous differentiation of ∆spy and Re9 strains (Fig. 5C). However, when parasites were cultured in alkaline medium to induce differentiation and analyzed 72 or 96 h post incubation, a modest reduction in the number of bradyzoites was observed for ∆spy, based on either SAG2Y or DBA labelling ( Fig. 5C). C57BL/6 mice infected by intraperitoneal injection with 10 4 tachyzoites from Re9 parental and ∆spy CZ1 strains showed similar patterns of weight loss and recovery on days 3, 5, 7, 12, 15 and 25 post infection (Fig. 5D). While there was no statistical difference in the luminescence signal from the abdominal cavity during the acute phase of infection (Figs. 5E and 5F), ∆spy parasites displayed consistently higher signals than the parental CZ1 Re9 strain from the brain during infection of the brain (Figs. 5G and 5H). As these were sublethal doses of infection, there was no difference in mortality between Re9 and ∆spy parasites (data not shown).
In summary, deletion of TgSPY in type II CZ1 strain caused modest decreases in growth and differentiation in culture and a moderate increase in luminescence of parasites in mouse brains.

Discussion
SPY is conserved in plants and algae, where it contributes to signaling and growth (26,27). This report shows that SPY also occurs and plays key roles in a wide range of other organisms. Arguing for an important role for the TgSPY OFT is the large number of proteins with SRDs modified by O-Fuc, many of which are predicted to be involved in gene transcription, protein transport, and signaling (22). Indeed, in a CRISPR/Cas9 screen examining growth of RH strain tachyzoites in culture, 69 AAL-enriched proteins, which include 33 with O-Fuc peptides identified directly by mass spectrometry, have average phenotype scores that are similar to those of essential proteins (Fig. S3) (41). Even so, TgSPY is not essential. This is shown by the viability of the ∆spy in both type I RH and type II CZ1 strains and is consistent with the fitness score of -0.11 (41). The conclusion that the O-Fuc modification is dispensable for the function and/or stability of its nucleocytosolic protein substrates is in contrast to two molecular observations. First, that addition of O-Fuc to the SRDs appears to direct proteins to assemblies within the nucleus and, secondly, loss of an SRD from GPN or the absence of O-Fuc on Nup68-YFP (an SRD-YFP reporter-construct), and possibly even on SPY itself, markedly decreases protein abundance.
Knockout of TgSPY slows growth and differentiation of type II CZ1 strain in culture and increases the luminescence from parasites in mouse brains. We have not yet determined the effect of TgSPY knockout on cat infections, but we have observed that AAL-labeled assemblies are absent in the periphery of meiotic nuclei of sporulating oocysts (22). Finally, addition of O-Fuc is the second cytosolic glycosylation system present in T. gondii, which also adds a pentasaccharide to a hydroxyproline residue on Skp1, an E3 ubiquitin ligase subunit (33).
TgSPY and AtSPY share a GT41 catalytic domain with HsOGT, have many of the same residues important for catalytic function, but have 11 rather than 13.5 TPRs (26)(27)(28). Our mutagenesis studies show that the conserved OGT/OFT residues in the catalytic domain are also important for TgSPY function (26,34,35). However, two of the positions targeted for mutagenesis (D619 and E692) are not strictly conserved in all OGT/OFT sequences (27), suggesting the possibility of intrinsic tolerance to amino acid changes. The residual activity observed in vivo for several of the point mutants is consistent with other studies that have noted greater activity of mutants in cell systems (e.g. (42)), likely because in vitro assays cannot comprehensively reproduce the conditions found in the cell environment (43). Additionally, while AtSPY-3TPRs was active in vitro (26), we could not detect any AAL labelling in our T. gondii tachyzoites assay.
This could be at least partially due to the difference in substrates: a peptide from Arabidopsis DELLA protein was used in the AtSPY in vitro assay (26), while TgSPY in the cell assay is acting on native proteins.
As is the case with TgSPY, HsOGT modifies proteins with IDRs including FG-Nups, and many enzymes involved with gene transcription and mRNA processing (24,25,44). While DELLA and PPR5 (a core circadian clock component) have been shown to be modified by AtSPY, the vast majority of nucleocytosolic proteins with O-Fuc of plants are uncharacterized, as secreted proteins, which have Nglycans abundantly decorated with fucose, dominate AAL pulldowns (26,45).
While HsOGT uses UDP-GlcNAc, which is responsive to the metabolic state of the cell (24), TgSPY and AtSPY use GDP-fucose, whose potential dependence on nutritional status is unclear. Although TgGPN∆SRD open-reading frame (ORF) was generated by gene synthesis (GenScript) and cloned into pgpnGoI3xMYC3/sagCAT using BglII and AvrII restriction sites. The plasmid is designed to encode a C-

terminal extension beyond the native C-terminus: PR[MYC-tag]NG[MYC-tag]NGARAE[MYC-Tag] (50).
The presence of the transgene was verified by PCR using primers P233 and P8. Parasites were electroporated with 20 μg DNA, selected with chloramphenicol and cloned by limiting dilution.
To remove the LoxP-flanked DHFR cassette, RH ∆spy parasites (31) were electroporated with p5RT70DiCre-DHFR, (a kind gift of Markus Meissner, Ludwig-Maximilians-Universität München) (51), recovered in a T25, and cloned by limiting dilution. Clones were randomly selected and tested for loss of resistance to pyrimethamine. The genotypes of the pyrimethamine-sensitive clones were verified by PCR using primers P206, P207, and P208.
The E. coli codon-optimized ORF of Toxoplasma SPY (Fig. S1) was cloned into ptubGoIMYC3/sagCAT (50) using BglII and AvrII restriction sites. The resulting plasmid (ptubTgSPYMYC3/sagCAT) was used to complement RH ∆spy. Parasites were electroporated, selected and cloned as above. The presence of the ectopic copy of TgSPY was verified by PCR using primers P9 and P198. The same protocol was used to complement RH ∆spy parasites with TgSPY point mutant and truncated constructs. Additionally, the presence of expected point mutants in the complemented cell lines was verified by amplifying the ORF fragment by PCR from gDNA followed by Sanger sequencing. Primers sequences are listed in Table S1.

CRISPR/Cas9-mediated disruption of spy in CZ1
The approach to generate the spy knockout is described in Figure S3C. Based on Me49 genome sequence, there is a SNP four bases upstream of the C-terminal PAM site for disrupting spy suggesting that the RH gRNA directing to this site might not work on a Me49-like sequence (31). For this reason, we designed an additional pU6 plasmid expressing an additional gRNA to target the Cz1 spy C-terminus. The protospacer was cloned using oligonucleotides P127-128 (spy_t2). Oligos were annealed and phosphorylated, and the resulting dsDNA was cloned in the BsaI-digested pU6Universal plasmid (38) to generate the plasmid pU6_spyt. To replace the gene of interest, the gra1 5'UTRs-LoxP-HXGPRT-LoxP-mGFP-gra2 3'UTRs expressing cassette was amplified by PCR using primers containing about 30-40 bp homology sequence to the double stranded break sites (P129 and P130, Table S1). The cassette was derived from (52) (Table S1). All constructs were verified by Sanger sequencing.
The same protocol was used for site-directed mutagenesis of pET-TEV_Tgspy, using primers that included silent restriction sites as an aid in screening (Table S1).
Semiquantitative RT-PCR About 10 7 tachyzoites from RH parental, TgGPN, or TgGPN∆SRD-expressing strains were harvested and lysed in TriReagent (Invitrogen). RNA was extracted according to the manufacturer's instructions and further cleaned with PureLink Mini RNA and the concentration was calculated on a Nanodrop (Thermo).
Different amounts of total RNA (4, 16, 32, or 40 ng) were used for first strand synthesis using SmartScribe RT (Clontech) and Oligo(dT)23 VN primer (NEB). A minus RT control was also performed using the highest total RNA amount (40 ng). Primers (P233-P8) were used to amplify a 500-bp gene-specific fragment from cDNA, common to both the full length and truncated TgGPN-MYC3 transgene, using OneTaq polymerase (NEB). RH cDNA was used as control for primer specificity, and the T. gondii GDPmannose 4,6-dehydratase (TgGMD) (P101-P102) served as positive control for cDNA synthesis. All reactions were analyzed by gel electrophoresis. Primers are listed in Table S1.

Southern blot
Tachyzoites from CZ1 parental strain and ∆spy clones were lysed in DNAzol (Invitrogen) and genomic DNA was isolated according to the manufacturer's instructions. Samples were treated with 0.1 mg/ml RNaseA (Sigma Aldrich) and genomic DNA quality was assessed by gel electrophoresis. gDNA (5 μg/lane) was digested with EcoRV, separated on a 1% agarose gel in Tris-acetate-EDTA buffer and blotted on a nylon membrane. A probe for mGFP ORF was generated using the PCR DIG Synthesis Probe Kit (Roche) and primers P35-P36 (Table S1) and the blot was processed and developed according to the manufacturer's instructions.

TgSPY recombinant expression in E. coli
The predicted ORF for full-length TgSPY from TGGT1_273500, annotated as an O-linked Nacetylglucosamine transferase in ToxoDB v46, was used to generate a cDNA that was codon-optimized (see Fig. S1) and synthesized by GenScript (Piscataway, NJ) for expression in E. coli. The spy cDNA was cloned into pET15-TEV, using the indicated restriction enzymes (Fig. S1), resulting in a version of TgSPY that was N-terminally tagged with a His6 sequence followed by a TEV-protease site, fused to the authentic N-terminus of TgSPY, including its Met. pET-TEV_Tgspy was electroporated into E. coli Gold and its expression was induced in the presence of 0.5 mM IPTG for 18 h at 22 °C. Cells were collected and protein was extracted and purified over a Ni +2 -column and a Superdex 200 column, essentially as described previously for His6Skp1 (53). A highly enriched sample from an included volume fraction was analyzed.
For improved expression of the mutant proteins, an autoinduction method was used following electroporation into E. coli (54). After 42 h of growth at 18 °C (no IPTG induction), extracts were prepared as above, and the SPY protein was captured on a Co ++ -Talon column. After washing the columns with 30 mM imidazole in 50 mM HEPES-NaOH (pH 7.4), 0.5 M NaCl, the SPY fraction was eluted with 250 mM imidazole in the same buffer, immediately dialyzed at 4° C against 50 mM HEPES-NaOH (pH 7.4), concentrated by centrifugal ultrafiltration, and frozen at -80 °C in aliquots. Samples were analyzed by SDS-PAGE and staining with Coomassie blue, or Western blotted and p probed with anti-His Ab, as described (33).

GST-SRD expression in E. coli
The sequence encoding for the 79 aa at the N-terminus of TgGPN (TGGT1_285720) was used to generate a cDNA that was codon-optimized for E. coli and synthesized by GenScript to include BamHI and XhoI restriction sites. The SRD cDNA was cloned via BamHI/XhoI into pGEX6P-1, resulting in a N-terminal His6TgSPY and unlabeled GDP-Fuc. Fucosylation was assayed by SDS-PAGE and Western blotting with biotinylated AAL as described (22). Scout assays indicated that activity was highest at pH 6.5, insensitive to NaCl from 20 to 150 mM, not affected by addition of 2 mM MnCl2, and not significantly affected by 0.2% Tween-20 or 2 mg/ml BSA. To assay transfer to a second protein, the candidate acceptor protein GST-SRD was expressed in E. coli and purified using Q-Sepharose chromatography as described above.
The reaction was conducted in the presence of 2 µM GDP-[ 3 H]Fuc (22,400 dpm/pmol; American Radiochemical Corporation), and incorporation was assayed by resolving the reaction products by SDS-PAGE and counting gel slices for radioactivity in a liquid scintillation counter ,as described (55). To test incorporation into native Toxoplasma proteins, desalted cytosolic extracts of tachyzoites were prepared as before (55) and used to assay incorporation into native acceptor substrates, using the same SDS-PAGE assay.
To assay transfer to a peptide, the tryptic peptide ZN190 (SSSSSASSSSSSFPSSSSSDSVPPR), which derives from TgRINGF1 (TGGT1_285190) and has been reported to be O-fucosylated in cells (22)

20
The cartridge was eluted with 2 × 1 ml MeOH into a 20-ml scintillation vial and mixed with 15 ml BioSafe-II, referred to as E1. A second elution with MeOH yielded negligible signal (dpm). All samples were assayed for radioactivity in a Beckman LS6500 liquid scintillation counter. Data was initially recorded as Super-resolution structured illumination microscopy (SIM) was performed on a Zeiss ELYRA microscope.
Images were acquired with a 63x/1.4 oil immersion objective, 0.089 µm z sections at RT and processed for SIM using ZEN software. For single optical sections, images were further processed with Fiji (57).

Quantification of AAL staining and AAL-positive puncta
To quantify and compare the intensity of AAL staining between wild type and complemented parasites, images were taken at the ELYRA microscope as detailed above and processed by SIM. A maximum intensity projection was then generated using the ZEN software. In Fiji, the epichromatin staining was used to define the nuclear region (NRoI) and the mean intensity of the NRoI in the AAL channel was measured.
The ratio to wild type staining from an average of three biological repeats is shown. A total of 111 tachyzoites from the parental strain and 106 from ∆spy::TgSPY were quantitated.
The number of AAL-positive puncta was counted using the 'Analyze Particle' function in Fiji (57). The puncta from a total of 86 tachyzoites from the parental strain and 104 from ∆spy::TgSPY were quantitated.
Box plots and statistical analyses (Student t Test) were performed using R Studio.

Plaque assay
Host cell monolayers on 6-well plates were infected with 250 parasites/well (RH) or 1000 parasites/well (CZ1) of either parental, ∆spy or complemented strains. RH and CZ1 parasites were allowed to grow at 37 °C, 5% CO2 for 5 or 8 days, respectively. After extensive washing in PBS, cells were fixed in ice-cold methanol at -20 °C for at least 20 min and stained with 2% crystal violet for 15 min. Wells were then washed with PBS, air-dried and imaged on an ImageQuant LAS400. Plaque areas were measured with Fiji and the box plot shows the average of three biological repeats ± SD. A two-tailed, homoscedastic Student t test was used to calculate the p values (R Studio).
In vitro differentiation HHF were grown to confluence on glass coverslips in 12-well plates. CZ1 parental and ∆spy strains were inoculated with a multiplicity of infection (MOI) of 0.5. When differentiation was evaluated under normal culture conditions, parasites were allowed to replicate for 72 h before fixation. Alternatively, parasites were allowed to replicate for 24 h before exchanging to alkaline stress medium: RPMI 1640 (Gibco) supplemented with 1% FBS, 100 units/ml penicillin, and 100 μg/ml streptomycin and buffered with 50 mM HEPES (pH 8.1). Parasites were incubated at 37 °C in the absence of CO2 for 72 or 96 h before fixation.
Fixation and labelling were performed as described above.

Mouse infections
Female C57BL/6 5 week-old mice were purchased from Charles River Laboratories (Wilmington, MA) and housed 5/cage with ad libitum access to food and water. Infections were performed after one week of acclimation as described (Agop-Nersesian et al., manuscript in preparation). Briefly, CZ1 parental and ∆spy strains were harvested immediately before infection from one T25 by scraping HFF monolayers.
Tachyzoites were released from PSVs by passing twice through a 27G needle attached to a 5-ml syringe.
Host cell debris was removed by filtering through a 3-μm polycarbonate membrane (Millipore). Parasites were counted using a haemocytometer and diluted to be 5x10 4 /ml. Each mouse was injected intraperitoneally (I.P.) with 200 μl of the suspension (10 4 tachyzoites). Additionally, for each experiment a plaque control was run: the same parasite suspension was used to infect the 6-well plate with 1000 and 2000 parasites of parental and knockout strains. Plaques were allowed to form and monolayers were fixed, labelled and quantitated as above. Mice were weighed on the day of infection and of IVIS measurements.
All mice were sacrificed 5 weeks post infections.

In Vivo Imaging
In vivo luminescence measurements were performed as described in Agop-Nersesian et al., (manuscript in preparation). C57BL/6 black mice were shaved the day before measurement on day 3 (abdomen) and day 12 (back of the head). If necessary, additional shaving was performed at later time points. Measurements were taken on days 3, 5, 7, 12, 15 and 25 post infection. A sterile solution of D-luciferin (PerkinElmer, Waltham MA) in Mg 2+ -and Ca 2+ -free DPBS was injected I.P. at 300 mg/kg, Mice were then sedated in a XGI-8 Anaesthesia System with 1.5% isofluorane and moved to the chamber of a IVIS Spectrum In vivo Imaging System (PerkinElmer) under 0.25% isofluorane. Abdomens were imaged 10 min post luciferin injection with exposures between 30 sec and 2 min. For brain imaging, a second luciferin injection was administered 20 min after the first one and, following the same sedation procedure, mice were imaged 30 min after the second injection with a 5-min exposure. Images were processed and quantified on Living Image (PerkinElmer) as total flux (photons/seconds) and adjusted to the numbers from the plaque assays to account for discrepancies in the number of parasites used to infect the mice. Bioluminescence measurements are presented as flux relative to the parental strain. SIM showed localization and AAL staining for all point mutants and 3TPRs truncation. Based on AAL, no 37 activity was detected for either K793A or the truncated construct, while all other mutants showed a reduced but detectable AAL pattern. Most mutant TgSPY proteins appear to be less abundant than wild type TgSPY. D. The mutant isoforms were expressed as N-terminally His6-tagged proteins in E. coli and partially purified on Talon-Co ++ columns. Equivalent amounts of protein were assayed for transfer of 3 H from GDP-[ 3 H]Fuc to a Ser-rich 25-mer peptide from TgRINGF1. SDS-PAGE followed by Coomassie blue staining (lower right) show the calibration for the protein amounts in each reaction and was confirmed by Western blotting using anti-His6 (upper right). Data are from a single representative trial conducted in triplicate, and error bars represent ± standard deviation.