Dolichol-linked oligosaccharide selection by the oligosaccharyltransferase in protist and fungal organisms

The eukaryotic oligosaccharyltransferase (OST) transfers preassembled oligosaccharides onto asparagine residues as nascent polypeptides are translocated across the rough ER membrane (for review see Kelleher and Gilmore, 2006). The consensus site for N-linked glycosylation in eukaryotic organisms is conserved and corresponds to the simple tripeptide sequence N-X-T/S, where X can be any residue except proline (Gavel and Von Heijne, 1990).

The oligosaccharide donor assembled by most eukaryotes for N-linked glycosylation is the dolichol pyrophosphate–linked oligosaccharide Glc₃Man₉GlcNAc₂-PP-Dol (abbreviated here as G₃M₉GN₂-PP-Dol). Synthesis of the dolichol-linked oligosaccharide (OS-PP-Dol) donor occurs by the stepwise addition of monosaccharide residues onto the dolichol-pyrophosphate carrier by a family of ER-localized membrane bound glycosyltransferases (asparagine-linked glycosylation [ALG] gene products; for review see Burda and Aebi, 1999). Man₅GlcNAc₂-PP-Dol (M₅GN₂-PP-Dol) is assembled on the cytoplasmic face of the ER membrane, with UDP-GlcNAc and GDP-Man serving as monosaccharide donors. Man-P-Dol and Glc-P-Dol are the donors for the luminally oriented glycosyltransferases that add four mannose and three glucose residues to OS-PP-Dol assembly intermediates within the ER lumen. Depletion of the yeast Rft1 protein causes severe hypoglycosylation of proteins and accumulation of Man₅GlcNAc₂-PP-Dol (Helenius et al., 2002) even though Alg3p, not Rft1p, is the mannosyltransferase that adds the sixth mannose residue. Rft1p has been proposed to flip cytosolically oriented M₅GN₂-PP-Dol across the ER membrane (Helenius et al., 2002).

Certain kinetoplastids (Trypanosoma cruzi and Leishmania mexicana) and the ciliate Tetrahymena pyriformis assemble OS-PP-Dol compounds that lack the glucose residues (M₉GN₂-PP-Dol by T. cruzi) and/or the mannose residues (G₃M₅GN₂-PP-Dol by T. pyriformis and M₆GN₂-PP-Dol by L. mexicana) that are transferred by the luminally oriented ALG gene products (de la Canal and Parodi, 1987; Parodi, 1993). Searches of fully sequenced genomes using yeast ALG proteins as query sequences has revealed considerable diversity in OS-PP-Dol biosynthesis amongst unicellular organisms (Samuelson et al., 2005). Biochemical studies have confirmed bioinformatic predictions that Giardia lamblia synthesizes GN₂-PP-Dol, Trichomonas vaginalis and Entamoeba histolytica synthesize M₅GN₂-PP-Dol, and the pathogenic fungi Cryptococcus neoformans synthesizes M₉GN₂-PP-Dol (Fig. 1; Samuelson et al., 2005). The diversity of eukaryotic OS- PP-Dol donors was proposed to have occurred by secondary loss of ALG genes during the evolution of current eukaryotes from a last common ancestor with a complete ALG pathway (Samuelson et al., 2005).

In fungi and vertebrate organisms, the OST is an oligomer composed of seven to eight nonidentical subunits (for review see Kelleher and Gilmore, 2006). Of the eight Saccharomyces cerevisiae OST subunits (Stt3p, Ost1p, Ost2p, Ost3p or Ost6p, Ost4p, Ost5p, Wbp1p, and Swp1p), five are encoded by essential yeast genes (STT3, OST1, OST2, WBP1, and SWP1). With the exception of STT3, which contains the enzyme active site (Yan and Lennarz, 2002; Kelleher et al., 2003; Nilsson et al., 2003), relatively little is known about the roles of the essential or nonessential subunits (for review see Kelleher and Gilmore, 2006). Vertebrate, plant, and insect genomes encode two forms of the catalytic subunit that are designated as STT3A and -B (Kelleher et al., 2003; Koiwa et al., 2003). The canine STT3 homologues are assembled with a shared set of noncatalytic subunits (ribophorin I [Ost1 homologue], ribophorin II [Swp1], OST48 [Wbp1], DAD1 [Ost2], and TUSC3 or IAP [Ost3 or -6] and OST4) to generate OST isoforms with kinetically distinct properties (Kelleher et al., 2003). Protein and DNA sequence database searches of fully sequenced eukaryotic genomes using the yeast and human OST subunits as query sequences suggest that the OST in protist organisms has a simpler subunit composition (Fig. 1; Kelleher and Gilmore, 2006). The genomes of G. lamblia and the kinetoplastids T. cruzi and Trypanosoma brucei encode several different STT3 proteins (Samuelson et al., 2005), yet lack genes encoding the noncatalytic subunits. Four-subunit complexes, consisting of STT3, OST1, OST2, and WBP1, are predicted for T. vaginalis and E. histolytica. A six-subunit complex (STT3, OST1, OST2, OST3, OST4, and WBP1) is predicted for Cryptosporidium parvum. The C. neoformans genome encodes readily identifiable homologues of all S. cerevisiae OST subunits with the exception of Ost5p (Fig. 1).

The absence of glucose residues on OS-PP-Dol compounds assembled by most protists and C. neoformans is of particular interest because the terminal glucose residue on G₃M₉GN₂-PP-Dol is a critical substrate recognition determinant for the OST. OS-PP-Dol assembly intermediates that lack the terminal glucose residue are transferred less rapidly by the vertebrate and yeast OST (Turco et al., 1977; Trimble et al., 1980; Bosch et al., 1988; Karaoglu et al., 2001; Kelleher et al., 2003), thereby minimizing synthesis of glycoproteins with aberrant oligosaccharide structures. Glycosylation of proteins with an oligosaccharide assembly intermediate could interfere with glycoprotein quality-control pathways in the ER as well as subsequent oligosaccharide-processing reactions in the Golgi complex (for review see Helenius and Aebi, 2004). Cellular defects in G₃M₉GN₂-PP-Dol biosynthesis cause a family of inherited diseases (congenital disorders of glycosylation [CDG-I]) due to hypoglycosylation of nascent glycoproteins by the OST in cells that accumulate an assembly intermediate or are unable to maintain a normal concentration of fully assembled G₃M₉GN₂-PP-Dol (for review see Freeze and Aebi, 2005).

Preferential utilization of G₃M₉GN₂-PP-Dol by the yeast and vertebrate OST occurs by allosteric interactions between a regulatory OS-PP-Dol binding site and the active site subunit, as well as by oligosaccharide structure–mediated alterations in tripeptide acceptor binding affinity (Karaoglu et al., 2001; Kelleher et al., 2003). Kinetic analysis of the purified canine OST isoforms has suggested that the regulatory OS-PP-Dol binding site is not located on STT3A or -B, but is instead associated with one or more of the noncatalytic subunits (Kelleher et al., 2003).

Does the OST from organisms that synthesize nonglucosylated OS-PP-Dols transfer the in vivo donor in preference to OS-PP-Dol assembly intermediates or G₃M₉GN₂-PP-Dol? Previous studies indicate that the T. cruzi OST transfers glucosylated (G_1-3M₉GN₂-PP-Dol) and large nonglucosylated (M_7-9GN₂-PP-Dol) donors at similar rates in vitro (Bosch et al., 1988), suggesting that the T. cruzi OST is nonselective. Can biochemical analysis of the OST from primitive eukaryotes reveal properties of the higher eukaryotic OST that arose as additional subunits were added to the STT3 catalytic core? Here, we report a comparison of the OST from T. vaginalis, E. histolytica, T. cruzi, C. neoformans, and S. cerevisiae, with emphasis placed upon an analysis of donor substrate selection. Our results support the hypothesis that terminal mannose residues on the OS-PP-Dol are important for donor substrate recognition by the OST in organisms that assemble nonglucosylated OS-PP-Dol compounds. Cooperative OS-PP-Dol binding, a feature of the yeast and canine OST complex that facilitates exquisite G₃M₉GN₂-PP-Dol selection, is not a property of the predicted one- and four-subunit protist OST complexes.

Is preferential utilization of the in vivo oligosaccharide donor an OST property that is restricted to eukaryotes that assemble triglucosylated OS-PP-Dols? To address this question, the OST from selected protists and C. neoformans was assayed using a synthetic tripeptide acceptor and a heterogeneous bovine OS-PP-Dol library that consists of donors that range in size between M₃GN₂-PP-Dol and G₃M₉GN₂-PP-Dol. Enzyme concentrations were adjusted to ensure that a maximum of 3% of the total donor substrate was converted into glycopeptides. Radiolabeled glycopeptide products that were captured with an immobilized lectin (ConA Sepharose) were subsequently eluted and resolved by high-pressure liquid chromatography (HPLC) according to the number of saccharide residues (Fig. 2). As expected, G₃M₉GN₂-NYT was the most abundant product when the purified S. cerevisiae OST was assayed (Fig. 2 A). In contrast, G₃M₉GN₂-NYT was less abundant in the T. cruzi glycopeptide products (Fig. 2 B) and barely detectable in glycopeptide products derived from assays of the T. vaginalis (Fig. 2 C), C. neoformans (Fig. 2 D), or E. histolytica (not depicted) OST. The composition of the OS-PP-Dol donor library was determined as described previously (Kelleher et al., 2001) by incubating an excess of the purified yeast OST with a low quantity of the donor substrate (OST endpoint assay; Fig. S1, available). A normalized initial transfer rate (glycosylated tripeptide [OS-NYT]/ OS-PP-Dol; Fig. 2, E and F) was calculated for the eight most abundant donors by dividing the glycopeptide product composition by the composition of the OS-PP-Dol donor substrate library. A normalized initial transfer rate of 1 (Fig. 2, E and F, dashed lines) indicates nonselective utilization of a donor substrate relative to the total donor pool.

The T. vaginalis and E. histolytica OST transfer the mannosylated donors (M_4-9GN₂-PP-Dol) threefold more rapidly than G₃M₉GN₂-PP-Dol (Fig. 2, E and F). Although the T. cruzi OST utilizes compounds ranging in size between M₇GN₂-PP-Dol and G₃M₉GN₂-PP-Dol at rates similar to those reported previously (Bosch et al., 1988), OS-PP-Dol donors with fewer mannose residues (M_3-6GN₂-PP-Dol) were transferred less rapidly (Fig. 2 F). The C. neoformans OST showed preferential utilization of the in vivo donor (M₉GN₂-PP-Dol) relative to assembly intermediates (M_3-6GN₂-PP-Dol) and the glucosylated donor (Fig. 2 F).

Careful inspection of the glycopeptide elution profiles (Fig. 2, A–D) revealed that several of the smaller glycopeptide peaks (e.g., M₅GN₂-NYT) have prominent shoulders, suggesting oligosaccharide structural heterogeneity. The structural diversity of the OS-PP-Dol donor substrate library is thought to arise by exposure of G₃M₉GN₂-PP-Dol to cellular glucosidases and mannosidases during isolation (Kelleher et al., 2001). Mannosidase degradation of M₉GN₂-PP-Dol (Fig. 3 A, compound a) could yield seven M₅GN₂-PP-Dol isomers (Fig. 3 A, compounds c–i) that differ from biosynthetic M₅GN₂-PP-Dol (Fig. 3 A, compound b). Biosynthetic M₅GN₂-PP-Dol (Fig. 3 A, compound b) can be readily distinguished from these other isomers by digestion with α-1,2 mannosidase, as it is the only isomer that has two α-1,2–linked mannose residues (Fig. 3 A, red circles). M₅GN₂ glycopeptides produced in an OST endpoint assay were purified by preparative HPLC (Fig. 3 B, left) and digested to completion with α-1,2 mannosidase. Resolution of the digestion products by HPLC (Fig. 3 B, right) yielded three peaks (M3, M4, and M5) that are derived from M₅GN₂-PP-NYT isomers that contain 2, 1, or 0 α-1,2–linked mannose residues. Quantification (Fig. 3 D, black bars) of two independent experiments revealed that 22% of the M₅GN₂ glycopeptides were derived from biosynthetic M₅GN₂-PP-Dol (Fig. 3 A, compound b), 63% from compounds c–h, and 15% from compound i.

If the OST from T. vaginalis, E. histolytica, T. cruzi, or S. cerevisiae selects biosynthetic M₅GN₂-PP-Dol (Fig. 3 A, compound b) in preference to other M₅GN₂-PP-Dol isomers, the M₅GN₂ glycopeptides synthesized in the presence of excess donor substrate should be enriched in glycopeptides that contain two α-1,2–linked mannose residues. To ensure that our glycopeptide product analysis provided a reliable measure of the relative initial transfer rate, the OST assays were terminated when <10% of the total M₅GN₂-PP-Dol was converted into glycopeptides. Typical HPLC profiles of the α-1,2 mannosidase digestion products of the M₅GN₂ glycopeptides are shown in Fig. 3 C, and the results from assays of all four organisms are quantified in Fig. 3 D. We observed a very similar distribution of M₅GN₂ isomers for the donor substrate pool and the initial S. cerevisiae glycopeptide products (Fig. 3 D, compare black and white bars), thereby indicating that the S. cerevisiae OST does not discriminate between M₅GN₂-PP-Dol isomers. The M₅GN₂ glycopeptides synthesized by the T. vaginalis and E. histolytica OST also resembled the M₅GN₂-PP-Dol donor pool; hence, the OST from these organisms does not select biosynthetic M₅GN₂-PP-Dol in preference to other M₅GN₂-PP-Dol isomers (Fig. 3 D). M₅GN₂ glycopeptides synthesized by the T. cruzi OST were twofold deficient in biosynthetic M₅GN₂-NYT and enriched in one or more M₅GN₂-NYT isomers that have one α-1,2–linked mannose residue (Fig. 3 D). This result, taken together with a reduced transfer rate for M_3-6GN₂-PP-Dol relative to M_7-9GN₂-PP-Dol (Fig. 2 F) by the T. cruzi OST suggests that a terminal α-1,2–linked mannose residue on the B or C antennae of M₉GN₂-PP-Dol serves as a positive determinant for substrate selection by the T. cruzi OST.

Donor substrate competition experiments were conducted using purified biosynthetic M₅GN₂-PP-Dol (Fig. 3 A, isomer b), M₉GN₂-PP-Dol (Fig. 3 A, compound a), and G₃M₉GN₂-PP-Dol. The T. vaginalis OST will synthesize G₃M₉GN₂-NYT when G₃M₉GN₂-PP-Dol is the sole donor substrate (Fig. 4 A, profile a). The absence of the M₅GN₂-NYT product indicates that the endogenous donor substrate is not abundant in the assay mix relative to the exogenous donor substrate. Analogous results were obtained using detergent extracts prepared from T. cruzi, E. histolytica, and C. neoformans (unpublished data). When the M₅GN₂-PP-Dol/ G₃M₉GN₂-PP-Dol ratio is 2.5:1, the yeast OST primarily synthesized G₃M₉GN₂-NYT, unlike the T. vaginalis OST that synthesized M₅GN₂-NYT (Fig. 4 A, profiles b and c). Quantification of this competition experiment, as well as additional assays containing 1.5 μM M₅GN₂-PP-Dol and variable concentrations of G₃M₉GN₂-PP-Dol, showed that donor substrate selection by the T. vaginalis (Fig. 4 B, squares) and S. cerevisiae (Fig. 4 B, circles) OST occurs across a wide range of donor substrate ratios (Fig. 4 B). Additional donor substrate competition experiments were conducted using 1:1 mixtures of the three purified oligosaccharide donors (Fig. 4, C–E). The S. cerevisiae OST selects G₃M₉GN₂-PP-Dol in preference to both nonglucosylated donors (Fig. 4, C and D) but does not discriminate between M₅GN₂-PP-Dol and M₉GN₂-PP-Dol (Fig. 4 E). The OST from E. histolytica and T. vaginalis selects both nonglucosylated donors in preference to G₃M₉GN₂-PP-Dol (Fig. 4, C and D) but does not discriminate between M₅GN₂-PP-Dol and M₉GN₂-PP-Dol (Fig. 4 E). The T. cruzi OST does not discriminate between M₉GN₂-PP-Dol and G₃M₉GN₂-PP-Dol (Fig. 4 D), but both donors are selected in preference to M₅GN₂-PP-Dol (Fig. 4, C and E). In contrast, both the glucosylated donor and biosynthetic M₅GN₂-PP-Dol are nonoptimal donors for the C. neoformans OST relative to the in vivo donor (Fig. 4, D and E).

The observation that G₃M₉GN₂-PP-Dol but not M₉GN₂-PP-Dol is a nonoptimal donor for the T. vaginalis and E. histolytica OST suggests that the A antennae of M₅GN₂-PP-Dol may be recognized by the OST in these organisms. To test this hypothesis, an additional competition experiment was performed using a mixture of purified M₅GN₂-PP-Dol and an enriched sample of G₂M₉GN₂-PP-Dol. The G₂M₉GN₂-PP-Dol preparation contains G₁M₉GN₂-PP-Dol as a minor component. Glycopeptide products synthesized by the S. cerevisiae and T. vaginalis OST were resolved by HPLC (Fig. 4 F, top). The initial transfer rates of ∼1 for the S. cerevisiae OST serves as an important control for the observed lower transfer rates of G₂M₉GN₂-PP-Dol and G₁M₉GN₂-PP-Dol relative to M₅GN₂-PP-Dol by the T. vaginalis OST. Each additional glucose residue on the A branch of the oligosaccharide reduces the normalized initial transfer rate by the T. vaginalis OST.

Enzyme kinetic experiments suggest that selection of the fully assembled OS-PP-Dol by the yeast or mammalian OST occurs by allosteric communication between a regulatory OS-PP-Dol binding site and the donor substrate binding site on STT3, in addition to oligosaccharide structure–dependent alterations in tripeptide substrate binding affinity (Karaoglu et al., 2001; Kelleher et al., 2003). Nonlinear Lineweaver-Burk plots for the OS-PP-Dol substrate are diagnostic of the cooperative OS-PP-Dol binding kinetics of the yeast and mammalian OST (Karaoglu et al., 2001). Donor substrate saturation experiments for the T. vaginalis (Fig. 5 A), E. histolytica (Fig. 5 C), and T. cruzi enzymes (Fig. 5 D) were conducted using a constant concentration of tripeptide acceptor and increasing concentrations of purified OS-PP-Dols. The linear Lineweaver-Burk plots yielded K_m values in the submicromolar range for the in vivo donor substrate. The experimental data for the T. vaginalis OST was replotted as an Eadie-Hofstee plot (Fig. 5 B). The linear Eadie-Hofstee plot for the T. vaginalis OST is inconsistent with cooperative OS-PP-Dol binding kinetics. In contrast, the S. cerevisiae OST binds the same donor substrate (M₅GN₂-PP-Dol) in a cooperative manner, as revealed by a nonlinear Eadie- Hofstee plot (Fig. 5 B, inset). Additional donor substrate saturation experiments using the nonoptimal donors (M₅GN₂-PP-Dol for T. cruzi OST and G₃M₉GN₂-PP-Dol for T. vaginalis OST) did not reveal differences in the apparent K_m that could account for the lower transfer rates of the nonoptimal donor substrate (unpublished data). Donor substrate selection by the protist OST does not involve a regulatory OS-PP-Dol binding site, nor is it explained by a reduced affinity for the nonoptimal oligosaccharide donor.

Reduced transfer rates for nonoptimal donors by the yeast and mammalian OST is in part explained by a reduced binding affinity for the tripeptide acceptor in the presence of an OS-PP-Dol assembly intermediate (Breuer and Bause, 1995; Gibbs and Coward, 1999; Karaoglu et al., 2001; Kelleher et al., 2003). The T. vaginalis (Fig. 6 A) and T. cruzi (Fig. 6 B) OST were assayed in the presence of a constant concentration of the optimal and nonoptimal oligosaccharide donors and increasing concentrations of the tripeptide acceptor. The linear Lineweaver-Burk plots for the tripeptide acceptors were indicative of a single acceptor tripeptide binding site, as observed for the yeast and mammalian OST (Karaoglu et al., 2001; Kelleher et al., 2003). The nonoptimal donor substrate (G₃M₉GN₂-PP-Dol for T. vaginalis and M₅GN₂-PP-Dol for T. cruzi) reduces the binding affinity of the OST for the tripeptide acceptor. In both cases, the threefold decrease in acceptor tripeptide binding affinity is responsible for the reduction in the normalized transfer rate when the acceptor tripeptide is not saturating. The apparent V_max is not influenced by the structure of the OS-PP-Dol donor, as revealed by a shared I/V intercept, when the oligosaccharide donors are present in fourfold excess relative to the apparent K_m for the donor substrate (Fig. 6 A).

GN₂-PP-Dol is the smallest oligosaccharide donor that is an effective substrate for the yeast OST (Sharma et al., 1981; Bause et al., 1995; Gibbs and Coward, 1999). The 2′ N-acetyl modification on the first saccharide is critical for catalysis, whereas the 2′ N-acetyl modification on the second residue is important for substrate recognition (Tai and Imperiali, 2001). Efficient N-glycosylation by the OST from higher eukaryotes is also dependent on the terminal glucose residue on the A antennae of the oligosaccharide (Turco et al., 1977; Trimble et al., 1980). As the OS-PP-Dol donors synthesized by many protists and the fungi C. neoformans lack glucose residues, one might predict that the OST from these organisms would only recognize the GlcNAc₂ core of the donor substrate. However, donor substrate competition experiments demonstrate that the in vivo oligosaccharide donor for T. vaginalis, E. histolytica, T. cruzi, and C. neoformans is a preferred substrate relative to certain larger and/or smaller OS-PP-Dol compounds. To our knowledge, this is the first evidence that oligosaccharide donor substrate selection is not restricted to organisms that synthesize the triglucosylated oligosaccharide donor. In all four cases, preferential utilization of the in vivo donor is less stringent than that observed for the S. cerevisiae or mammalian OST both in terms of the size range of compounds that are optimal in vitro substrates and the fold selection of the in vivo donor substrate relative to nonoptimal donors.

The predicted one-subunit OST from T. cruzi utilizes larger OS-PP-Dol compounds, including the in vivo donor M₉GN₂-PP-Dol in preference to M₅GN₂-PP-Dol. The latter compound is one of four lumenal OS-PP-Dol assembly intermediates that could compete in vivo with M₉GN₂-PP-Dol as a donor substrate. The observed two- to threefold more rapid in vitro transfer of M₉GN₂-PP-Dol than M₅GN₂-PP-Dol appears to be sufficient to ensure that small assembly intermediates are rarely used in vivo, in part because M₉GN₂-PP-Dol is more abundant in the T. cruzi ER than the luminally oriented (M_5-8GN₂-PP-Dol) assembly intermediates (Parodi and Quesada-Allue, 1982). The presence of a terminal α-1,2–linked mannose residue on the B or C antennae appears to be important for preferential utilization of M₉GN₂-PP-Dol by the T. cruzi OST, as revealed by the relative transfer rates of M₅GN₂-PP-Dol isomer classes (Fig. 3) and by the reduced utilization of M_3-6GN₂-PP-Dol relative to M₉GN₂-PP-Dol. In vivo transfer of an assembly intermediate may be deleterious, as protein-linked high-mannose oligosaccharides that lack the terminal mannose residue on the B antennae (M8B isomer) or C antennae (M8C isomer) are less efficiently glucosylated by the UDP-glucose glycoprotein glucosyltransferase (UGGT; Trombetta and Parodi, 2003). UGGT, which was first detected in T. cruzi, serves as the folding sensor for the glycoprotein quality-control pathway in the ER (Caramelo et al., 2003).

The predicted four-subunit OSTs from T. vaginalis and E. histolytica (Fig. 1) transfer the in vivo donor (M₅GN₂-PP-Dol) at the same rate as other OS-PP-Dol compounds that lack glucose residues (M_4-9GN₂-PP-Dol), including M₅GN₂-PP-Dol isomers that lack one or more mannose residues on the A antennae. Because synthesis of the M₅GN₂-PP-Dol donor is completed on the cytoplasmic face of the rough ER, the T. vaginalis and E. histolytica OST do not need to discriminate between luminally oriented M₅GN₂-PP-Dol and cytoplasmically oriented OS-PP-Dol assembly intermediates. Consequently, M₅GN₂-NYT is the major glycopeptide synthesized in vitro when an acceptor tripeptide is incubated with intact T. vaginalis or E. histolytica membranes (Samuelson et al., 2005) despite the lack of a mechanism to discriminate against underassembled oligosaccharide donors. We propose that the STT3 active-site subunit of the OST has evolved to have a catalytic site that is optimal for the in vivo oligosaccharide. For T. vaginalis, E. histolytica, and C. neoformans, the proposed loss of genes that encode the ALG glucosyltransferases (ALG6, -8, and -10; Samuelson et al., 2005) has apparently been accompanied by compensatory alterations in the STT3 structure that are optimal for an oligosaccharide donor with an A antennae that lacks all three glucose residues.

The predicted seven-subunit C. neoformans OST transfers the larger mannosylated OS-PP-Dol donors (M_7-9GN₂-PP-Dol) more rapidly than smaller assembly intermediates or G₃M₉GN₂-PP-Dol. Utilization of the fully assembled in vivo donor in preference to luminally exposed OS-PP-Dol assembly intermediates may be a shared property of the OST in organisms that synthesize donors larger than M₅GN₂-PP-Dol. The relatively modest (∼1.5-fold) preference for M₉GN₂-PP-Dol relative to biosynthetic M₅GN₂-PP-Dol leads to selective synthesis of M₉GN₂-NYT when the acceptor tripeptide is incubated with intact C. neoformans membranes (Samuelson et al., 2005).

Kinetic analysis of the T. cruzi and T. vaginalis OST revealed that oligosaccharide structure–mediated modulation of acceptor substrate binding affinity is a conserved property of the eukaryotic OST that can be ascribed to the STT3 active site. The threefold reduction in acceptor substrate binding affinity readily accounts for the reduced transfer of nonoptimal donors when the acceptor tripeptide is present at subsaturating levels. Future studies will address the order of substrate binding to the one- and four-subunit OSTs that are predicted for T. cruzi and E. histolytica. One objective of these experiments will be to determine whether the subunit composition of protist complexes matches the bioinformatic predictions.

Cooperative OS-PP-Dol binding by the S. cerevisiae OST is not explained by dimerization of heterooctamers, as coimmunoprecipitation experiments using yeast strains that express STT3-HA₃ and STT3-His₆FLAG₁ from chromosomal loci did not reveal higher order OST oligomers (Karaoglu et al., 2001). Potential explanations for the discrepancy between a recent report describing dimeric assembly of the yeast OST complex (Chavan et al., 2006) and our previous conclusions are being explored. Cooperative OS-PP-Dol binding is not explained by separate but interacting binding sites for the chitobiose core of G₃M₉GN₂-PP-Dol and the terminal glucose residue, because cooperative binding by the yeast or canine OST is not dependent on the presence of glucose residues on the oligosaccharide donor, as confirmed here using M₅GN₂-PP-Dol as a donor substrate. Instead, our results indicate that cooperative donor substrate binding is diagnostic of a regulatory OS-PP-Dol binding site that is primarily responsible for the highly selective utilization of the G₃M₉GN₂-PP-Dol donor (Karaoglu et al., 2001; Kelleher et al., 2003).

Based on a kinetic analysis of canine OST isoforms, we proposed that the regulatory OS-PP-Dol binding site is not located on the catalytic subunit (STT3A or -B), but is instead provided by one or more of the shared noncatalytic subunits. Support for this hypothesis has now been provided by recent experiments showing that a T. cruzi STT3 can assemble with the noncatalytic yeast OST subunits and, upon doing so, mediate selective utilization of G₃M₉GN₂-PP-Dol as the donor substrate both in vitro and in vivo (Castro et al., 2006).

One objective of this study was to determine whether protist OSTs use a regulatory OS-PP-Dol binding site to select the in vivo oligosaccharide donor. Unlike the S. cerevisiae and Canis familiaris OST, the predicted one-subunit OST from T. cruzi (STT3) and the predicted four-subunit OSTs from E. histolytica and T. vaginalis (STT3-OST1-OST2-WBP1) do not bind OS-PP-Dol in a cooperative manner; hence, the OST from these organisms lacks the regulatory OS-PP-Dol binding site. The simplest interpretation of this observation is that the regulatory OS-PP-Dol binding arose as additional subunits were acquired during evolution of the eukaryotic OST. The IAP and TUSC3 (N33) proteins dissociate from the canine OST during purification, so these OST3/OST6 family members are not candidates for the regulatory OS-PP-Dol binding site. OST4 and -5 can be discounted based on structural considerations because neither of these polypeptides has more than a few residues exposed to the lumen of the ER (Fig. 1). Therefore, cooperative OS-PP-Dol binding by the yeast or vertebrate OST correlates with the presence of a Swp1p/ribophorin II subunit in the OST complex. Extensive biochemical and genetic evidence supports direct interactions between Wbp1, Swp1p, and Ost2p (te Heesen et al., 1993; Silberstein et al., 1995), as well as between their respective mammalian homologues, OST48, ribophorin II, and DAD1 (Fu et al., 1997; Kelleher and Gilmore, 1997). We hypothesize that the regulatory OS-PP-Dol binding site is located on the Swp1p–Wbp1p–Ost2p subcomplex. Interestingly, OS-PP-Dol protects a critical cysteine residue in Wbp1p from modification by a cysteine-directed protein modification reagent (Pathak et al., 1995). A role for the Swp1p–Wbp1p–Ost2p subcomplex as the regulatory OS-PP-Dol binding site might help explain why expression of each of these subunits is essential for viability of S. cerevisiae (te Heesen et al., 1992, 1993; Silberstein et al., 1995). With the exception of C. neoformans, there is a strong correlation between organisms that assemble a glucosylated oligosaccharide donor (either G₃M₅GN₂-PP-Dol or G₃M₉GN₂-PP-Dol) and organisms that express or are predicted to express a Swp1p/ribophorin II homologue (Samuelson et al., 2005; Kelleher and Gilmore, 2006).

Trophozoites of E. histolytica strain HM1:IMSS were grown axenically (in the absence of bacteria or other cells) in TYI medium supplemented with 10% heat-inactivated adult bovine serum at 37°C. Axenic cultures of T. vaginalis strain G3 were maintained in TYM medium supplemented with 10% heat-inactivated horse serum at 37°C. Axenic cultures of T. cruzi epimastigotes (strain Y) were grown in the LIT medium supplemented with hemin and 10% heat-inactivated fetal calf serum at 25°C. C. neoformans strain B3501, maintained on YPD plates, was grown in YPD broth for 20 h at 30°C.

Whole cells were collected by centrifugation and resuspended in 10 mM Hepes, pH 7.4, 25 mM NaCl, 10 mM MgCl₂, and 1× protease inhibitor cocktail (PIC; as defined by Kelleher et al., 1992). E. histolytica, T. vaginalis, or T. cruzi cells were homogenized using 50 strokes of a Teflon-glass homogenizer. The C. neoformans cell suspension was mixed with an equal volume of glass beads and vortexed extensively (200 5-s bursts). Total membrane fractions were collected by a 30-min centrifugation of the cell homogenate at 267,000 g_av using a rotor (TLA 100.4; Beckman Coulter). The membrane pellets were solubilized in 1.5% digitonin, 20 mM Tris-Cl, pH 7.5, 500 mM NaCl, 1 mM MgCl₂, 1 mM MnCl₂, 1 mM DTT, and 1× PIC at a membrane concentration of 2 eq/μl (1 eq/μl = 50 A₂₈₀ in 1% SDS). The detergent extracts were clarified by a 5-min centrifugation at 66,600 g_av using the rotor. The S. cerevisiae OST was purified from an epitope-tagged (6xHisFLAG-OST1) yeast strain as described previously (Karaoglu et al., 2001).

Detergent extracts of the E. histolytica, T. vaginalis, T. cruzi, and C. neo formans membranes were diluted fourfold with 20 mM Tris-Cl, pH 7.4, 1 mM MgCl₂, 1 mM MnCl₂, 1 mM DTT, and 1× PIC. 5-μl aliquots of the 4×-diluted soluble extracts were assayed for OST activity in a total volume of 100 μl as described previously (Kelleher and Gilmore, 1997), using Nα-Ac-Asn-[¹²⁵I]Tyr-Thr-NH₂ as the acceptor substrate and either structurally homogeneous OS-PP-Dol compounds or a previously described heterogeneous bovine pancreas OS-PP-Dol pool (Kelleher et al., 2001) as the donor substrate. OST assays were supplemented with 1.4 mM deoxynojirimycin, 1.4 mM mannojirimycin, and 1.4 mM swainsonine to inhibit glucosidases and mannosidases. Glycopeptide products from OST assays were isolated with ConA Sepharose and quantified by gamma counting.

Structurally homogeneous G₃M₉GN₂-PP-Dol, M₉GN₂-PP-Dol, M₅GN₂-PP-Dol, and an enriched G₂M₉GN₂-PP-Dol preparation were purified as described previously (Kelleher et al., 2001) from porcine pancreas (G₃M₉GN₂-PP-Dol and G₂M₉GN₂-PP-Dol), an alg5Δ yeast strain (M₉GN₂-PP-Dol), or an alg3Δ yeast strain (M₅GN₂-PP-Dol).

The concentration and composition of OS-PP-Dol samples was determined from the yield and oligosaccharide distribution of radiolabeled glycopeptides obtained in the OST endpoint assay (Kelleher et al., 2001). In brief, 12–15 pmol of OS-PP-Dol was incubated with 60 fmol of purified yeast OST for 24–48 h under OST assay conditions to quantitatively convert the donor substrate into glycopeptides. The previously isolated heterogeneous bovine pancreas OS-PP-Dol pool consists of a mixture of biosynthetic OS-PP-Dol assembly intermediates and OS-PP-Dol degradation products that were produced by exposure of the OS-PP-Dol to endogenous mannosidases and glycosidases during isolation (Kelleher et al. 2001). As shown in Fig. S1 B, the bovine OS-PP-Dol pool has the following oligosaccharide composition: 4.7% M₃GN₂, 14% M₄GN₂, 19% M₅GN₂, 23% M₆GN₂,17% M₇GN₂, 11% M₈GN₂, 3.7% M₉GN₂, 1% G₁M₉GN₂, 1.5% G₂M₉GN₂, and 5.1% G₃M₉GN₂. Because of low abundance in the donor pool, initial transfer rates are not reported for G₁M₉GN₂-PP-Dol and G₂M₉GN₂-PP-Dol.

Assays designed to analyze the donor substrate preference of the OST using the bovine OS-PP-Dol library contained 1.2 μM OS-PP-Dol and were terminated before 3% of the total donor was consumed. Donor substrate competition experiments using purified donors were terminated before 10% of the substrate was consumed. Glycopeptide products from the competition experiments were eluted from the ConA beads and resolved according to oligosaccharide size by HPLC as described previously (Mellis and Baenziger, 1981; Kelleher et al., 2001), except that the HPLC buffer A was acetonitrile/water/acetic acid/triethlyamine (73.6:23:2.4:1), whereas HPLC buffer B was water/acetic acid/triethlyamine (91:3:6). Glycopeptides prepared using the yeast OST and purified OS-PP-Dol compounds served as HPLC elution standards.

OST assays to prepare M₅GN₂ glycopeptides using the heterogeneous OS-PP-Dol library were designed to ensure that <10% of the total M₅GN₂-PP-Dol was converted to M₅GN₂-NYT. HPLC fractions corresponding to M₅GN₂-NYT were dried and resuspended in 50 μl of 1× reaction buffer supplied by the manufacturer (Prozyme) and incubated for 18 h at 37°C with 0.33 mU α-1,2 mannosidase. The Savant-dried glycopeptide digestion products were dissolved in 500 μl HPLC buffer A and resolved by HPLC as described in the preceeding paragraph.

The kinetic parameters for the tripeptide acceptor and oligosaccharide donor for the T. vaginalis, T. cruzi, and E. histolytica enzymes were determined by a nonlinear least-squares fit of the kinetic data to the Michaelis-Menten equation and by linear least-squares fits of Lineweaver-Burk plots or Eadie-Hofstee plots. The kinetic parameters for the dolichol-oligosaccharide donor for the S. cerevisiae enzyme were obtained using a nonlinear least-squares fit of the kinetic data to equations for a substrate activated enzyme as described previously (Karaoglu et al., 2001). Kaleidagraph 3.5 (Synergy Software) was used for curve fitting.

Fig. S1 shows the oligosaccharide composition analysis of the OS-PP-Dol library used for the experiments in Fig. 2 and Fig. 3.

The authors thank Phillips Robbins for helpful discussions.

This work was supported by National Institutes of Health grants GM43768 (R. Gilmore), AI44070 (J. Samuelson), and GM31318 (Phillips Robbins).