The E3 Ubiquitin Ligase Adaptor Protein Skp1 Is Glycosylated by an Evolutionarily Conserved Pathway That Regulates Protist Growth and Development*♦

Toxoplasma gondii is a protist parasite of warm-blooded animals that causes disease by proliferating intracellularly in muscle and the central nervous system. Previous studies showed that a prolyl 4-hydroxylase related to animal HIFα prolyl hydroxylases is required for optimal parasite proliferation, especially at low O2. We also observed that Pro-154 of Skp1, a subunit of the Skp1/Cullin-1/F-box protein (SCF)-class of E3-ubiquitin ligases, is a natural substrate of this enzyme. In an unrelated protist, Dictyostelium discoideum, Skp1 hydroxyproline is modified by five sugars via the action of three glycosyltransferases, Gnt1, PgtA, and AgtA, which are required for optimal O2-dependent development. We show here that TgSkp1 hydroxyproline is modified by a similar pentasaccharide, based on mass spectrometry, and that assembly of the first three sugars is dependent on Toxoplasma homologs of Gnt1 and PgtA. Reconstitution of the glycosyltransferase reactions in extracts with radioactive sugar nucleotide substrates and appropriate Skp1 glycoforms, followed by chromatographic analysis of acid hydrolysates of the reaction products, confirmed the predicted sugar identities as GlcNAc, Gal, and Fuc. Disruptions of gnt1 or pgtA resulted in decreased parasite growth. Off target effects were excluded based on restoration of the normal glycan chain and growth upon genetic complementation. By analogy to Dictyostelium Skp1, the mechanism may involve regulation of assembly of the SCF complex. Understanding the mechanism of Toxoplasma Skp1 glycosylation is expected to help develop it as a drug target for control of the pathogen, as the glycosyltransferases are absent from mammalian hosts.

Toxoplasma is a worldwide obligate intracellular apicomplexan parasite that infects most nucleated cells of warm-blooded animals (1). Toxoplasmosis, the disease caused by Toxoplasma, is an opportunistic infection in AIDS and other immune-suppressed patients (2). In addition, in utero infections can cause mental retardation, blindness, and death (3). Toxoplasma is transmitted by digesting parasites from feline feces (as oocysts) or undercooked meat (as tissue cysts). Once in the host, parasites convert to the tachyzoite form that disseminates to peripheral tissues (e.g. brain, retina, and muscle). The resulting immune response and/or drugs can control tachyzoite replication, but the parasite survives by converting into slow growing bradyzoites that encyst. Cysts sporadically burst, and the released parasites convert to tachyzoites whose unabated growth, as can occur in immune suppressed hosts, results in cell and tissue damage (4). Currently, no Toxoplasma vaccine exists; anti-toxoplasmosis drugs have severe side effects, and resistance to these drugs is occurring.
Recently, disruption of the gene for PhyA, the prolyl 4-hydroxylase that hydroxylates Pro-154 in Skp1, was observed to reduce tachyzoite proliferation in cell culture and fitness in a competition assay (5). Skp1 is an adaptor in the Skp1/Cullin-1/ F-box protein (SCF) 2 class of E3 ubiquitin ligases, and its hydroxylation was hypothesized to contribute to O 2 -dependent proliferation. That study noted that loss of hydroxylation resulted in increased migration in SDS-polyacrylamide gels suggesting a decrease in M r of ϳ1000. Previous studies in an unrelated protist, the social soil amoeba Dictyostelium, had shown that the Skp1-hydroxyproline (Hyp) could be glycosylated by five glycosyltransferase activities encoded by three genes, resulting in assembly of a pentasaccharide at the equivalent Pro residue (6,7). Because two of these genes, gnt1 and pgtA, have apparent homologs in the Toxoplasma genome, we suspected that the gel shift might result from inability of the glycosyltransferases to modify Skp1 in the absence of formation of the Hyp anchor. In Dictyostelium, Ddgnt1 encodes a polypeptide ␣GlcNAc transferase that transfers GlcNAc from UDP-GlcNAc to form GlcNAc␣1-O-Skp1 (8). Ddpgta encodes a dual function diglycosyltransferase whose N-terminal domain then transfers Gal from UDP-Gal to form a Gal␤1-3GlcNAc linkage and whose C-terminal domain processively transfers Fuc from GDP-Fuc to form a Fuc␣1-2Gal linkage (9). However, the two domains are switched in the Toxoplasma version of the predicted protein (TGGT1_260650), and there is no evidence for agtA, the Dictyostelium gene that is responsible for addition of the final two sugars, both ␣Gal residues.
If the Skp1 Hyp of Toxoplasma can be glycosylated, the importance of hydroxylation for proliferation might be due to consequent loss of glycosylation rather than inability to hydroxylate per se. In Dictyostelium, hydroxylation alone partially rescues O 2 -dependent development (10). Full recovery depends, however, on full glycosylation (11), and glycosylation is required to promote efficient assembly of the Skp1/F-box protein heterodimer, based on interactome studies (12). Therefore, we sought direct evidence for Hyp-dependent glycosylation of Toxoplasma Skp1 and the role of the predicted glycosyltransferases, and we used disruption strains of the predicted glycosyltransferase genes to test their contribution to parasite proliferation. The findings implicate Skp1 as the functional target of this novel post-translational modification pathway in Toxoplasma and indicate that the Skp1 modification pathway is evolutionarily conserved among protists.
To perform cell growth plaque assays, confluent HFF monolayers in 6-well tissue culture plates were infected with freshly lysed-out (see below) parasites at 250 parasites/well, equivalent to a multiplicity of infection of 0.002. After 3 h, unattached parasites were removed by two rinses with phosphate-buffered saline (15 mM sodium phosphate, pH 7.4, 135 mM NaCl). After undisturbed incubation in Complete medium for 5.5 days, monolayers were fixed with methanol and stained with crystal violet to detect plaques. Plaques (n Ն50) from at least two wells were manually encircled, and areas were calculated by ImageJ software (National Institutes of Health). Data were presented and statistically analyzed using GraphPad Prism version 6.
TgphyA, Tggnt1, and TgpgtA Disruption Strains-DNAs for gene disruptions were generated from pminiGFP.ht (gift of Dr. Gustavo Arrizabalaga, University of Idaho), in which the hxgprt gene is flanked by multiple cloning sites. The approach was modeled after that used for the TgphyA disruption strain RHphyA⌬-1, in which exon 1 of TgphyA was replaced with hxgprt (5). To generate an independent TgphyA disruption strain, RHphyA⌬-2, the complete coding region was replaced with hxgprt by double crossover homologous recombination. First, the 5Ј-and 3Ј-flank targeting sequences of TgphyA from RH⌬⌬ were PCR-amplified with primer pairs a and aЈ and pairs b and bЈ, respectively (supplemental Table S1). The 5Ј-fragment was digested with KpnI and HindIII and inserted into pminiGFP.ht between its KpnI and HindIII sites. The resulting plasmid was digested with XbaI and NotI and ligated to the XbaI-and NotI-digested 3Ј-flank. The resulting vector was linearized with KpnI and electroporated into RH⌬⌬ strain as described (5). Drug-resistant transformants were selected in the presence of 25 g/ml MPA and 25 g/ml xanthine and cloned by limiting dilution. Genomic DNA from three clones was screened by PCR to identify TgphyA disruption strain RHPhyA⌬-2, as described under "Results." The PCRs were performed on extracts from 2 ϫ 10 6 parasites, using Taq polymerase, and primers as listed in supplemental Table S1. Standard conditions included 1.5 mM MgCl 2 , and reactions were run for 30 cycles of the following standard scheme: 94°C, 30 s; 60°C, 1 min; 68°C, 3 min. Exact conditions were adjusted for specific reactions.
To disrupt Tggnt1, the 5Ј-flank and 3Ј-flank targeting sequences were PCR-amplified with primer pairs c and cЈ and d and dЈ (supplemental Table S1), respectively, and inserted into pminiGFP.ht as above. The vector was linearized with SapI and transfected into RH⌬⌬, and drug-resistant clones were screened by PCR to generate the Tggnt1 disruption strain (RHgnt1⌬). Similarly, the TgpgtA disruption construct was generated by PCR amplification and insertion into pminiGF-P.ht of 5Ј-flank and 3Ј-flank targeting sequences using primer pairs e and eЈ and f and fЈ, respectively. After digestion with PacI, the DNA was transfected into RH⌬⌬, and RHpgtA⌬ clones were screened by PCR.
Tggnt1 and TgpgtA Complemented Strains-pminiGFP.ht was used as the backbone for constructing the Tggnt1 complementation construct after removing its HXGPRT cassette by KpnI and NotI digestion. A 7-kb DNA fragment containing the Tggnt1 genomic region was PCR-amplified using primer pairs c and dЈ (supplemental Table S1), digested with KpnI and NotI, and ligated into the similarly digested pminiGFP.ht. The resultant vector was linearized with KpnI and electroporated into RHgnt1⌬. Transformants were selected under 300 g/ml 6-thioxanthine (Matrix Scientific), and clones were screened by PCR. To complement TgpgtA knock-out, the fosmid clone Rfos01M21 (13), containing a 36-kb fragment of RH strain chromosome VIIb (2039542-2076165), which includes the TgpgtA gene (gift of Dr. Boris Striepen, University of Georgia), was linearized with ScaI and electroporated into RHpgtA⌬. Complemented clones were isolated as described for Tggnt1.
Epitope Tagging of Endogenous TgSkp1-To modify the C terminus of endogenous TgSkp1, the skp1 genomic locus was modified by the insertion of SF-tag cDNA. A 1.5-kb region upstream of the skp1 stop codon was PCR-amplified using primer pairs g and gЈ. Using a ligation-independent cloning strategy (14), the product was inserted into PacI-digested pSF-TAP-LIC-HXGPRT and pSF-TAP-LIC-CAT vectors (from Dr. Vern Carruthers, University of Michigan). 50 g of the resulting constructs were linearized with EcoRV at position 527 of the insert, and the DNAs were electroporated into RH⌬⌬ and RHphyA⌬-1, respectively. RH⌬/SF transformants were selected under 25 g/ml MPA/xanthine, and RHphyA⌬-1/SF was selected under 20 M chloramphenicol. Site-specific integration was confirmed by PCR of clones using the primer pairs h and hЈ. DNA sequencing confirmed that the 3Ј-end of the Skp1 coding sequence encoded the native C terminus ( . . . VREENKWCEDA) followed by a peptide containing two Strep-II tags and a FLAG tag (in boldface), AKIGSGGR-EFWSHPQFEKGGGSGGGSGGGSWSHPQFEKGASGEDYK-DDDDK*. Characteristics of the above strains are summarized in Table 1.
Purification of Endogenous TgSkp1-Tachyzoites from RH⌬⌬, RHphyA⌬-1, RHgnt1⌬, and RHpgtA⌬ strains were harvested from infected HFF monolayers by scraping and passage through a 27-gauge needle, centrifuged at 2000 ϫ g for 8 min at room temperature, resuspended in sterile phosphate-buffered saline, and counted on a hemacytometer chamber as described (15). 6 ϫ 10 8 tachyzoites were pelleted, frozen at Ϫ80°C, and subsequently thawed and solubilized in 8 M urea, 50 mM HEPES-NaOH, pH 7.4, supplemented with protease inhibitors (1 mM PMSF, 10 g/ml aprotinin and 10 g/ml leupeptin) at 4°C for 30 min. The lysates were centrifuged at 16,000 ϫ g for 15 min at 4°C, and supernatants (S16) were collected and diluted 8-fold in IP buffer (0.2% Nonidet P-40 (v/v) in 50 mM HEPES-NaOH, pH 7.4, protease inhibitors) and incubated with 60 l of rabbit polyclonal anti-TgSkp1 UOK75-Sepharose beads for 1 h at 4°C. The UOK75 antiserum (5) was first affinity-purified against recombinant TgSkp1-Sepharose beads, performed as described for affinity purification of anti-DdSkp1 (12), and then coupled to CNBr-activated Sepharose CL-4B. After centrifuging and resuspending the beads three times with IP buffer and three times with wash buffer (10 mM Tris-HCl, pH 7.5, 154 mM NaCl), bound material was eluted twice with 150 l of 133 mM triethylamine, pH 11.5, for 10 min and immediately neutralized with 150 l of 200 mM acetic acid, pH 2.7. The pooled eluates (ϳ400 l) were divided into two equal parts, concentrated by vacuum centrifugation to ϳ10 l, and snapfrozen in liquid nitrogen. To purify SF-tagged TgSkp1, soluble S16 fractions were prepared from RH⌬/SF and RHphyA⌬-1/SF strains as described above and incubated with 100 l of mouse anti-FLAG M2-agarose beads (Sigma) for 1 h at 4°C. The beads were washed as above and eluted with 300 l of 8 M urea in 25 mM NH 4 HCO 3 (natural pH), supplemented with 40 mg of urea, and incubated for 15 min at room temperature, and the supernatants harboring TgSkp1-SF were collected at 2400 ϫ g for 5 min in room temperature.
Mass Spectrometry of TgSkp1 Peptides-The untagged TgSkp1 samples were taken to dryness and solubilized in 100 l of 8 M urea in 25 mM NH 4 HCO 3 . The untagged and SF-tagged TgSkp1 samples were reduced by addition of 0.5 M tris(2-carboxyethyl-)phosphine to a final concentration of 5 mM for 20 min at 22°C and alkylated by the addition of 0.5 M iodoacetamide to a final concentration of 10 mM for 15 min in the dark. Tris(2-carboxyethyl)phosphine was then added to a final concentration of 10 mM, and 300 l of 50 mM NH 4 HCO 3 , pH 7.8, was added to dilute urea to 2 M. Samples were treated with 1 mg/ml mass spectrometry grade trypsin (Promega) at a final concentration 8.0 g/ml and incubated overnight at 37°C. Peptides were recovered by adsorption to a C18 Zip Tip (OMIX TIP C18 100 l) and eluted with 0.1% trifluoroacetic acid in 50% (v/v) acetonitrile.
Dry peptides were reconstituted in 15.6 l of solvent A (0.1% formic acid) and 0.4 l of solvent B (0.1% formic acid in 80% acetonitrile) and loaded onto a 75-m (inner diameter) ϫ 115-mm C18 capillary column (YMC GEL ODS-AQ120ÅS-5, Waters) packed in-house with a nitrogen bomb. Peptides were eluted into the nanospray source of an LTQ Orbitrap TM mass spectrometer (Thermo Fisher Scientific) with a 160-min linear gradient consisting of 5-100% solvent B over 100 min at a flow rate of 250 nl/min. The spray voltage was set to 2.0 kV, and the temperature of the heated capillary was set to 210°C. Full scan MS spectra were acquired from m/z 300 to 2000 at 30,000 resolution, and MS2 scans following collision-induced fragmentation were collected in the ion trap for the 12 most intense ions. The raw spectra were searched against a Toxoplasma gondii protein database (UniProt ATCC 50611/Me49, Sept. 2013) using SEQUEST (Proteome Discoverer 1.3, Thermo Fisher Scientific) with full MS peptide tolerance of 50 ppm and MS2 peptide fragment tolerance of 0.5 Da and filtered to generate a 1% target decoy peptide-spectrum-match false discovery rate for protein assignments. The spectra assigned as glycosylated TgSkp1 peptides were manually validated.
GalT activity was assayed similarly except that the donor was 1 M UDP-[ 3 H]Gal and prepared from a mixture of 1 Ci of UDP-[ 3 H]Gal (7) and unlabeled UDP-Gal, and the acceptor was recombinantly prepared GlcNAc-DdSkp1 (18). FucT activity was similarly assayed except that 2 M GDP- Radioactive Sugar Analyses-The chemical form of the radioactivity incorporated into DdSkp1 was determined by high pH anionic exchange chromatography analysis after acid hydrolysis as described (17). Briefly, Gnt1 reaction products were resolved by SDS-PAGE and electroblotted onto a 0.45-m PVDF membrane (EMD Millipore). The membrane was stained with 0.2% Ponceau S in 3% (w/v) TCA, and the Skp1 protein band was excised with a razor blade, submerged in 400 l of 6 M HCl, and incubated at 100°C for 4 h. Hydrolysates were removed into microtubes, evaporated to dryness under vacuum centrifugation, dissolved in 500 l of H 2 O, dried again twice, and reconstituted in 25 l of H 2 O. A solution containing 1 nmol of GalNH 2 and GlcNH 2 in 9 l of H 2 O was added to the hydrolysates and chromatographed on PA-1 column on a DX-600 Dionex high pH anionic exchange chromatography workstation in 16 mM NaOH at 1 ml/min with pulsed amperometric detection. Fractions were collected into EconoSafe (Research Products International) scintillation fluid and counted for 3 H incorporation on a liquid scintillation counter. PgtA assay products were similarly hydrolyzed in 4 M TFA and mixed with a standard solution containing 1.5 nmol each of Glc, Gal, Man, and Fuc.
Western Blotting-Western blotting was performed as described (5). Briefly, tachyzoite pellets were solubilized in lysis buffer containing 8 M urea, 50 mM HEPES-NaOH, pH 7.4, supplemented with protease inhibitors at 4°C for 30 min and centrifuged at 16,000 ϫ g for 15 min at 4°C to generate a soluble S16 fraction. After combining with SDS-PAGE sample buffer, proteins were resolved on a 4 -12% gradient SDS-polyacrylamide gel (NuPAGE Novex, Invitrogen) and transferred to a nitrocellulose membrane using an iBlot system (Invitrogen). After probing with a 1:500-fold dilution of the UOK75 anti-TgSkp1 antibody and a 1:10,000-fold dilution of Alexa-680-labeled goat anti-rabbit IgG secondary antibody (Invitrogen), blots were imaged on a Li-Cor Odyssey infrared scanner.

Toxoplasma Skp1 Is Modified by a Pentasaccharide-Our
previous study showed that disruption of exon 1 of the Skp1 prolyl 4-hydroxylase gene (phyA) in the parental type 1 strain RH⌬ku80⌬hxgprt (RH⌬⌬) resulted in greater mobility of Skp1 during SDS-PAGE, corresponding to an M r difference of about 1000 (5). To investigate the possibility that this was due to a loss of hydroxylation-dependent glycosylation as occurs in Dictyostelium (6), a previously described antiserum raised against recombinant TgSkp1 (5) was used to immunoprecipitate Skp1 from tachyzoite extracts, and its tryptic peptides were analyzed by conventional nano-LC/MS in an LTQ-XL Orbitrap mass spectrometer. Peptides covering 75% of the 170-amino acid sequence of Skp1, including 145 IFNIVNDFTPEEEAQVR containing unmodified Pro-154 (m/z 1011, [M ϩ 2H] 2ϩ ) eluting at 88.7 min, were observed. Potential hydroxylated glycopeptides were sought using a theoretical mass list of glycoforms of this peptide containing any combination of 1-8 residues of Hex, dHex, HexNAc, HexUA, and pentose. This search yielded, at 82.7 min elution time, a single glycopeptide ([M ϩ 2H] 2ϩ , 1437.1464; [M ϩ 3H] 3ϩ , 958.09) with an exact match (within 0.56 ppm) to a glycoform containing 1 HexNAc, 3 Hex, and 1 dHex residues (Fig. 1A). Similar results were obtained for a Skp1-SF preparation isolated by anti-FLAG immunoprecipitation from a strain in which Skp1 was C-terminally modified by an SF-tag (data not shown). The putative glycopeptide ion was absent from the RHphyA⌬ strain (summarized in Table 2), consistent with its identity as the predicted Skp1 glycopeptide.
The putative glycopeptide ion was subjected to MS/MS analysis to confirm its composition and characterize its organization. Fragmentation of the doubly charged parent ion by CID yielded a series of ions whose mass differences corresponded to loss of a combination of Hex, dHex, and/or HexNAc residues (Fig. 1B), resulting ultimately in the expected hydroxypeptide  ion. The predominant ions were consistent with the presence of a linear pentasaccharide whose composition from the non-reducing end is Hex-Hex-dHex-Hex-HexNAc-. Fragmentation also yielded a series of a singly charged peptide and glycopeptide fragment b-and y-ions that confirmed the predicted amino acid sequence of the parent ion and demonstrated the position of the additional O atom as occurring on Pro (to yield Hyp) and the attachment of all sugars via Pro-154 (Fig. 1C). These data indicate that a fraction of Skp1 in Toxoplasma is modified by a linear pentasaccharide reminiscent of the linear pentasaccharide on DdSkp1.
Predicted TgSkp1-modifying Glycosyltransferases-BLASTP and TBLASTN searches for sequences corresponding to the three GT genes that catalyze formation of the pentasaccharide on DdSkp1 in ToxoDB (Version 7.3) yielded high scoring hits for DdGnt1 and DdPgtA. No candidates for a homolog of DdAgtA were detected using these algorithms or PSI-or PHI-BLAST toward either its catalytic or WD40 repeat domains. The Toxoplasma Ddgnt1-like sequence is represented by a oneexon gene model ( Fig. 2A) in three sequenced strains of Toxoplasma (GT1, ME49, and VEG). The GT1 (type 1 strain) sequence (TGGT1_315885) exhibited 42% identity and 67% similarity to DdGnt1 over 214 amino acids of the ϳ250 amino acid catalytic domain (supplemental Fig. S1). Like DdGnt1, TGGT1_315885 is predicted to be a cytoplasmic protein because of the absence of detectable membrane or nuclear targeting motifs. However, at 1510 amino acids, TGGT1_315885 is substantially longer than DdGnt1 by 423 amino acids. As revealed by the amino acid sequence alignment (supplemental Fig. S1), and as illustrated in Fig. 2C, this results from multiple insertions throughout the length of the protein, a common occurrence in Toxoplasma genes as observed, e.g. in TgIF2Kb (19). In addition, a C-terminal sequence referred to as domain A that lies outside of the predicted catalytic domain (10) but is required for DdGnt1 activity is weakly conserved across these predicted proteins (Fig. 2). The remaining intervening sequences are poorly conserved even among coccidian apicomplexans that have PhyA-, Gnt1-, and PgtA-like sequences (supplemental Fig. S1). The coccidian Sarcosystis neurona is predicted to contain even longer insert sequences. In comparison, sequences from Chromera velia and Vitrella brassicaformis, representatives of the closest known photosynthetic relatives of apicomplexans in the alveolate superphylum (20), largely lack these inserts and resemble the length of the Dictyostelium sequence. Although the TgGnt1-coding region remains to be confirmed experimentally, two of the inserts are present in expressed sequence tags derived from mRNA, and a third was detected in a shotgun proteomics screen ( Fig. 2C; supplemental Fig. S1). Although the significance of these additional sequences is enigmatic, their low conservation suggests that they fulfill Toxoplasma-specific functions that are unlikely to be related to the proposed enzymatic activity.
The Toxoplasma DdpgtA-like sequence is represented by a 14-exon gene model (Fig. 2B). The predicted amino acid sequence of TGGT1_260650, from the type I GT1 strain, is 98 and 97% identical to that of the type II ME49 and type III VEG strain sequences, and the protein is predicted to be cytoplasmic. The Toxoplasma candidate is 1801 amino acids long, compared with the 768 amino acid length of DdPgtA, and the order of the two putative glycosyltransferase domains is reversed (Fig. 2D). The N-terminal CAZy GT2 family sequence of DdPgtA, which encodes a ␤3-GalT activity, has its sequence homolog in the C-terminal half of the Toxoplasma protein, whereas the C-terminal CAZy GT74 family sequence of DdPgtA, which encodes

MS detection of Skp1 glycopeptides in strains
Isoforms of the Skp1 peptide 145 IFNIVNDFTPEEEAQVR were detected as described in Fig. 1. The distribution of raw ion counts among the detected isoforms are shown for the strains analyzed.
an ␣2-FucT activity, has its homolog in the N-terminal half of the Toxoplasma protein. The GT2-like domain of the predicted TgPgtA is 22% identical and 60% similar to the DdPgtA-␤GalT domain over the 191 most conserved amino acids, and the GT74-like domain is 33% identical and 66% similar to the DdPgtA-␣FucT domain over 190 amino acids. The amino acid sequence alignment (supplemental Fig. S2) of the PgtA-like sequences reveal that, as for DdGnt1, the predicted apicomplexan PgtA-like sequences have numerous multiple inserts relative to the Dictyostelium prototype, as illustrated in Fig. 2D. These insert sequences, several of which occur in the protein based on expressed sequence tags and MS data (Fig. 2D), tend to diverge even within the coccidian apicomplexans and are minimal in the chromerids. Thus, they are unlikely to be critical for enzymatic activity.
Tggnt1 and TgpgtA Are Required for TgSkp1 Glycosylation-To determine whether Tggnt1 and TgpgtA are involved in TgSkp1 glycosylation, their genes were disrupted by doublecrossover homologous recombination in the RH⌬⌬ strain, as described under "Experimental Procedures" and illustrated in Fig. 3A for Gnt1. Deletion of gnt1 in recovered clones was demonstrated by loss of a PCR product for gnt1-coding DNA, and positive PCRs for the insertion of the selection marker hxgprt between gnt1-flanking sequences, as described in Fig. 3B. To control for off-target genetic modifications, a gnt1 disruption clone was complemented with a version of the original disruption DNA in which hxgprt was replaced by the deleted coding region and counter-selected for loss of hxgprt (Fig. 3A). The same set of PCRs was used to confirm the desired gene restoration in clonal isolates (Fig. 3C).
The effect of gnt1 deletion on Skp1 glycosylation was evaluated initially by SDS-PAGE and Western blotting. As shown in Fig. 4A, Skp1 from parasites lacking gnt1 (lane 3) migrated more rapidly than wild-type Skp1 (lane 1), and similarly to Skp1 from parasites whose phyA had been disrupted (lane 2). MS searches for the Skp1 glycopeptide were negative, but a novel hydroxypeptide corresponding to the hydroxylated but nonglycosylated Skp1 was obtained, in addition to the unmodified peptide ( Table 2). Analysis of the complemented strain revealed that normal mobility of Skp1 in SDS-PAGE was at least partially restored. Thus, Skp1 HexNAcylation depends on Gnt1, and by analogy with the Dictyostelium example, Gnt1 is expected to directly catalyze addition of the first sugar in ␣-linkage onto Hyp154.
A similar analysis was performed on PgtA, whose genetic locus was manipulated as described in Fig. 3, D-F. In this case, a recently prepared genomic fosmid (13) was utilized to restore the genomic locus in the pgtA-disruption clone. Genomic DNA was utilized because of the large number of predicted introns and our difficulty in isolating a full-length cDNA using RT-PCR  0K  1K  2K  3K  4K  5K  6K  7K  8K  9K  10K  11K  12K  13K  14K   Intron   Exon  1000 nt   Tggnt1 (4533 bp)  (data not shown), and the fosmid clone was used because of the predicted large size (13.2 kb) of its genomic locus. As shown in Fig. 4, lane 7, Skp1 glycosylation also appeared to be affected by the loss of pgtA based on SDS-PAGE/Western blotting, and this was confirmed by the accumulation of the HexNAc form of Skp1, the expected acceptor substrate of PgtA, based on MS analysis of tryptic peptides ( Table 2). As expected, restoration of the pgtA locus resulted in at least partial recovery of Skp1 glycosylation (Fig. 4, lane 8). At a minimum, PgtA is thus required for addition of the second sugar to the Skp1 glycan.    Table 1.  Soluble S16 fractions from equivalent numbers (3 ϫ 10 6 cells) of parental RH⌬⌬ (RH) and RHphyA⌬-1 (phyA Ϫ ), RHgnt1⌬, RHpgtA⌬, and their complemented cells were resolved by 4 -12% SDS-PAGE, electroblotted, and probed using anti-TgSkp1 (UOK75) antiserum. Changes in glycosylation inferred from altered gel mobility were confirmed by mass spectrometry (Table 2). Similar results were obtained for independently derived clones of RHgnt1⌬ and RHpgtA⌬.

Complex Glycosylation of Toxoplasma Skp1
FEBRUARY 26, 2016 • VOLUME 291 • NUMBER 9 from UDP-[ 3 H]GlcNAc to DdHis 10 -Skp1 that was recovered from an SDS-polyacrylamide gel (21). This activity depended on addition of DdPhyA indicating dependence on Hyp. Replication of an optimized form of this assay (see "Experimental Procedures") revealed time-dependent transfer of 3 H to HO-DdSkp1 that was absent from gnt1⌬ extracts (Fig. 5A). As shown in Fig. 5B, no incorporation into endogenous proteins was detected based on analysis of an entire SDS-polyacrylamide gel lane, indicating absence of activity of endogenous GTs from other sources, such as the Golgi, that modify other targets in these cytosolic preparations. The lack of other radiolabeled proteins is inconsistent with the existence of an intermediate TgGnt1 substrate that itself mediates modification of Skp1. Incorporation was reduced as expected after addition of a 10-fold excess of unlabeled UDP-GlcNAc, but not of unlabeled UDP-GalNAc (Fig. 5C), indicating that the enzyme is selective for the GlcNAc isomer. Because a homolog of Gnt1 that resides in the Golgi transfers GalNAc to proteins (22), and Toxoplasma possesses an epimerase that can interconvert UDP-GlcNAc with UDP-GalNAc, the nature of the transferred 3 H was confirmed by another method. 3 H was found to be incorporated as GlcNAc, based on co-chromatography of 3 H released by HCl hydrolysis, which de-N-acetylates GlcNAc to GlcNH 2 , with a GlcNH 2 standard (Fig. 5D). Although the evidence that TgGnt1 modifies Skp1 in this assay is indirect, its ho-mology with DdGnt1, whose purified recombinant version can ␣GlcNAcylate HO-DdSkp1 in vitro (8), suggests that TgGnt1 also directly ␣GlcNAcylates TgSkp1.
TgPgtA Is a Bifunctional Glycosyltransferase with GalT and FucT Activities-To characterize the role of pgtA in extending the Skp1 glycan, the above assay was first modified by substituting UDP-[ 3 H]Gal for UDP-[ 3 H]GlcNAc and GlcNAc-Skp1 for HO-Skp1. Time-dependent incorporation of 3 H into GlcNAc-Skp1 was observed (Fig. 6A). Incorporation into the Skp1 band on the SDS-polyacrylamide gel required the inclusion of Skp1, and Skp1 was the only protein that incorporated detectable radioactivity (Fig. 6B). No incorporation was detected in extracts of pgtA⌬ cells. Recovery and analysis of the 3 H after acid hydrolysis confirmed incorporation as Gal rather than a derivative (Fig. 6C). Similar findings were observed in a corresponding FucT assay, in which UDP-[ 3 H]Gal was replaced by GDP-[ 3 H]Fuc (Fig. 6, D-F). However, incorporation of 3 H depended on the inclusion of UDP-Gal (unlabeled), in contrast to the GalT reaction that did not require GDP-Fuc (Fig. 6A). This indicated that TgPgtA is, like DdPgtA (9), a processive diglycosyltransferase that catalyzes the sequential addition of Gal and then Fuc, an order consistent with the MS-MS data. Although Ͼ95% of incorporation of [ 3 H]Fuc into the Skp1 band depended on Skp1 and pgtA (Fig. 6, D and E), residual incorporation was observed at this position and elsewhere in the gel indicative of additional pgtA-and Skp1-independent FucT activity in the extract. These results are consistent with accumulation of GlcNAc-Skp1 and the absence of higher glycosylation states in pgtA⌬ cells ( Fig. 4 and Table 2). TgGnt1 and TgPgtA Are Important for Toxoplasma Growth in Cell Culture-Previous studies revealed that disruption of exon 1 of phyA results in a parasite growth defect, which could be detected as reduced plaque areas after 5 days of replication on a fibroblast monolayer (5). To check that no residual phyA activity was present, all nine exons were deleted (RHphyA⌬-2). A similarly reduced ability to grow on monolayers was observed, as illustrated in Fig. 7A and quantified in B. Furthermore, deletion of exon 1 in a strain in which Skp1 was C-terminally modified with an SF-epitope tag, which itself did not affect growth, also resulted in slowed growth (Fig. 7C). To examine the roles gnt1 and pgtA, the plaque-forming abilities of the disruption strains described above were analyzed. As shown in Fig. 7B, gnt1⌬ cells exhibited slow growth that was statistically indistinguishable from that of phyA⌬ cells. pgtA⌬ cells also exhibited a slow growth phenotype, which was intermediate between that of gnt1⌬ and parental (RH⌬⌬) cells. Complementation of gnt1 and pgtA by gene replacement at their original loci restored normal growth (Fig. 7D), showing that the growth differences in the original disruption strains could be attributed to the GT targets. Although these findings do not demonstrate directly that the altered modifications of Skp1 are involved in reduced growth, the finding that three independent enzymes that share Skp1 as a target substrate exhibit a similar deficiency is consistent   with a role for Skp1, a model that has both biochemical and genetic support in Dictyostelium.

Discussion
Skp1 isolated from the tachyzoite stage of Toxoplasma is partially modified by a glycan chain that consists of five sugars and is linked to the hydroxylated form of Pro-154. The monosaccharides are organized as a linear pentasaccharide with a sequence, reading from the peripheral non-reducing end, of Hex-Hex-␣Fuc-␤Gal-␣GlcNAc-, based on mass spectrometry and characterization of the core GTs. Remarkably, these properties match those of the glycan that was previously characterized on Skp1 from an unrelated protist, the social amoeba Dictyostelium (6). Genetic disruption of glycosylation rendered a growth defect in host cell monolayers, implicating a role for Toxoplasma E3 SCF ubiquitin ligases in cell proliferation.
The predicted Skp1 glycan was analyzed at the glycopeptide level because of the resistance of the glycan-Hyp linkage to known methods of cleavage. Native Skp1 from parasites was analyzed to avoid overexpression artifacts, but this necessitated application of highly sensitive methods due to the limited amount of material available because Toxoplasma can only be grown intracellularly. As is typical for the analysis of glycopeptides by mass spectrometry, their detection required manual inspection of spectra for candidate ions, because of suppression and abundance issues. The pentasaccharide peptide was detected in an exhaustive search of the primary spectra for tryptic peptides bearing Pro-154 (Fig. 1), the previously documented hydroxylation site (5), and any combination of up to eight monosaccharides. Decomposition analysis in the gas phase confirmed the glycan's location at Hyp154 and suggested that it is organized as a linear pentasaccharide with a sequence, from the non-reducing end, of Hex-Hex-dHex-Hex-HexNAc-. The only other peptide isoform detected was non-hydroxylated and therefore non-glycosylated ( Table 2), suggesting limited glycosylation microheterogeneity.
The sugar identities were investigated by highly sensitive incorporation of radioactive sugars mediated by parasite extracts that harbor the biosynthetic enzymes. Cytosolic extracts were observed to specifically incorporate radioactivity from three different radioactive sugar nucleotides into recombinantly generated isoforms of Dictyostelium Skp1, which was previously shown to be an excellent substrate for DdPhyA (5). Based on the confirmed identities of the 3 H-sugars after incorporation, the core trisaccharide is concluded to consist of Fuc-Gal-GlcNAc, which matches the dHex-Hex-HexNAcfound by MS. Incorporation of [ 3 H]GlcNAc into HO-Skp1 depended on Tggnt1 based on the absence of activity in gnt1⌬ extracts (Fig.  5). By analogy with its Dictyostelium ortholog, TgGnt1 is inferred to transfer GlcNAc in an ␣-linkage to Hyp154 of TgSkp1. Incorporation of [ 3 H]Gal depended on TgpgtA (Fig. 6), and the addition of [ 3 H]Fuc, which also depended upon TgpgtA, required prior addition of Gal (Fig. 6). These linkages require further study to determine whether they are conserved with the Dictyostelium example. Thus, although the physical order of the two GT domains is reversed relative to DdPgtA (Fig. 2), the order of addition of the two sugars is conserved. The existence of the two terminal Hex residues was unexpected because of the absence of a homolog for AgtA in the Toxoplasma genome, and their identities are under current investigation.
Both Tggnt1 and TgpgtA were important for efficient plaqueforming ability in tissue culture monolayers cultivated under standard conditions (Fig. 7). Loss of gnt1 was as severe as loss of phyA, the prolyl 4-hydroxylase that is required for Gnt1 action on Skp1. Loss of pgtA resulted in an intermediate effect, i.e. the plaque-forming ability was improved relative to cells lacking phyA. These effects were specific for the GT genes as normal plaque-forming ability was restored upon genetic complementation to the original genotype. Thus, Gnt1 appears to make a major contribution to phyA-dependent growth, but complete realization appears to depend on Gnt1-dependent glycosylation contributed by, at least on Skp1, PgtA and the additional unknown GTs. In Dictyostelium, biochemical studies indicate that Skp1 is the only substrate for the orthologs of these GTs, and disruption of either gnt1 or pgtA inhibits development in a way that is related to, but less severe than, disruption of phyA (10). Gene dosage manipulations on Skp1 expression show inverse effects on development consistent with the modification genes acting via Skp1 in this organism, a model that is supported by point mutations that remove the target Pro residue, and from a double mutant between a GT gene and one of two Skp1 genes (23). By analogy, and based on similar phenotypes of TgphyA⌬ and Tggnt1⌬, we propose that the modification genes render their effects in a common pathway that affects Skp1. Further studies on Skp1 itself are needed to evaluate this possibility.
The reduced plaque areas are consistent with a role for Skp1 modification in cell cycle progression as demonstrated in yeast and mammalian cells, where SCF-type E3 ubiquitin ligases are important for signaling proteasome-dependent turnover of cell cycle kinase inhibitors (24). However, SCF ligases represent a family of enzymes that includes many F-box proteins with distinct substrate receptor activities (25). Given the large variety of substrates known to be recognized by yeast, plant, and human F-box proteins, further studies are needed to evaluate whether other processes required for plaque formation, such as binding of the parasite to host cells, ingress and/or egress are affected. In future studies, it will be interesting to evaluate whether any of these potential mechanisms are more severely affected in low O 2 or altered metabolic states that are anticipated to influence Skp1 modification enzyme activities in cells (11).
PhyA-, Gnt1-, and PgtA-like sequences are selectively conserved across a broad spectrum of unicellular eukaryotes, including representatives of all major protist clades (6). Validation of their shared functions in both Toxoplasma and Dictyostelium, which are highly diverged, suggests that Skp1 hydroxylation and glycosylation occurred in ancestral eukaryotes prior to their loss in fungi, higher plants, animals, and select protists. The relatively rapid evolution of F-box proteins, with which Skp1 partners for many if not all of its functions, suggests that Skp1 modifications serve an outsized role in environmental regulation of unique lineage and species-specific functions in many unicellular organisms.