Biosynthesis of lipid-linked oligosaccharides in Saccharomyces cerevisiae: Alg13p and Alg14p form a complex required for the formation of GlcNAc(2)-PP-dolichol.

N-Glycosylation in the endoplasmic reticulum is an essential protein modification and highly conserved in evolution from yeast to man. Here we identify and characterize two essential yeast proteins having homology to bacterial glycosyltransferases, designated Alg13p and Alg14p, as being required for the formation of GlcNAc(2)-PP-dolichol (Dol), the second step in the biosynthesis of the unique lipid-linked core oligosaccharide. Down-regulation of each gene led to a defect in protein N-glycosylation and an accumulation of GlcNAc(1)-PP-Dol in vivo as revealed by metabolic labeling with [(3)H]glucosamine. Microsomal membranes from cells repressed for ALG13 or ALG14, as well as detergent-solubilized extracts thereof, were unable to catalyze the transfer of N-acetylglucosamine from UDP-GlcNAc to [(14)C]GlcNAc(1)-PP-Dol, but did not impair the formation of GlcNAc(1)-PP-Dol or GlcNAc-GPI. Immunoprecipitating Alg13p from solubilized extracts resulted in the formation of GlcNAc(2)-PP-Dol but required Alg14p for activity, because an Alg13p immunoprecipitate obtained from cells in which ALG14 was down-regulated lacked this activity. In Western blot analysis it was demonstrated that Alg13p, for which no well defined transmembrane segment has been predicted, localizes both to the membrane and cytosol; the latter form, however, is enzymatically inactive. In contrast, Alg14p is exclusively membrane-bound. Repression of the ALG14 gene causes a depletion of Alg13p from the membrane. By affinity chromatography on IgG-Sepharose using Alg14-ZZ as bait, we demonstrate that Alg13-myc co-fractionates with Alg14-ZZ. The data suggest that Alg13p associates with Alg14p to a complex forming the active transferase catalyzing the biosynthesis of GlcNAc(2)-PP-Dol.

lowed by the addition of one GlcNAc and five mannose residues from UDP-GlcNAc and GDP-Man, respectively. The resulting Man 5 GlcNAc 2 -PP-Dol, thus generated, is translocated into the lumen of the ER (4,7,8) and extended by four mannose and three glucose residues deriving from Dol-P-Man and Dol-P-Glc, respectively. Some of the reactions of this pathway have been studied in vitro using microsomal membranes, solubilized extracts, or partially purified enzymes from different sources (9 -24). However a detailed enzymology and especially the regulation and coordination of their activities remain to be elucidated. In the yeast Saccharomyces cerevisiae, alg mutants (for asparagine-linked glycosylation) defective in lipid-linked oligosaccharide (LLO) assembly have been isolated (25)(26)(27)(28), which was extremely helpful in defining the pathway and isolating the corresponding genes. Similarly, different mutant cell lines from mouse lymphoma cells (29) or Chinese hamster ovary cells have been described that produce truncated lipid-linked oligosaccharides (30 -32). It seems that the individual substrate specificity of the glycosyltransferases ensures an ordered assembly of the oligosaccharide (33).
For the second step in LLO synthesis, which is the formation of Glc-NAc 2 -PP-Dol from GlcNAc 1 -PP-Dol and UDP-GlcNAc, thus far no alg mutant has been isolated, and the corresponding gene has not been identified yet. We therefore searched for a potential GlcNAc-transferase in silico. While this study was in its final stage, circumstantial evidence was presented that two unknown open reading frames (ORFs), YGL047w and YBR070c, designated below as ALG13 and ALG14, respectively, are involved in this reaction step (34). We extended these observations by establishing a specific assay to directly measure the enzyme activity. We have demonstrated that yeast membranes depleted of Alg13p or Alg14p lack GlcNAc-transferase activity in vitro and accumulate GlcNAc 1 -PP-Dol in vivo. Furthermore we have detected the proteins and shown that both are required for enzymatic activity, by acting in a protein complex. We found that an immunoprecipitate from detergent extracts of membranes of wild-type cells using Alg13p as target is able to catalyze the transfer of GlcNAc to GlcNAc 1 -PP-Dol, whereas the immunoprecipitate obtained from extracts of cells repressed for ALG14, or of a cytosolic form of Alg13p, is inactive. By affinity chromatography we show that Alg13 and Alg14 co-fractionate, indicative that the proteins form a complex.
To construct the expression plasmids pVT100-Alg13-ZZ and pVT100-Alg14-ZZ, both ORFs with engineered HindIII and BamHI restriction sites at the 5Ј and 3Ј ends, respectively, for cloning purpose were amplified by PCR and ligated into the HindIII/BamHI cut pVT100-ZZ vector under the control of the ADH1 promotor. The constructs were sequenced, and the expression of the genes containing two protein A epitopes (ZZ) in-frame at the C terminus was verified by immunoblotting. For PCR amplification the following primers were used: Alg13 forward (5Ј-CCCAAGCTTGGGATGGGTATTATTG-3Ј), Alg13 reverse (5Ј-CGGGATCCGCTGTATATAGTTTAACTAG-CAATC-3Ј), Alg14 forward (5Ј-CCCAAGCTTAT GAAAACGGCCT-ACTTGGCGTCATTGGTG-3Ј), Alg14 reverse (5ЈCGCGGATCCAA-CAAGGATGCCGAACCACTTGGATCTTGGT-3Ј). To generate C-terminal ZZ (protein A) and myc fusions, respectively, of ALG13 at the chromosomal locus, an integrative cassette, containing either the ZZ epitope or the myc epitope and kanMX4 as the selection marker, was created by PCR on template DNA pZZ-kanMX or pMyc-kanMX. The correct integration was confirmed by PCR amplification. ALG13-Kan forward (5Ј-CCCAGT TTCTCATAACCCGTCATTTGAGCGATT-GCTAGTTGAAACTATATACAGCGGAGCAGGGGCGGGTGC-3Ј), ALG13-Kan reverse (5Ј-GATCAAAAACGTCTAAGGCATTACG CGGATACTGCTAACTAACCAAAGAAACCTTCCAGGTCGAC-GGTATCG-3Ј). Transformation into yeast and Escherichia coli was carried out using standard techniques.

Isolation of Microsomal Membranes and Preparation of Solubilized Enzyme Extract
Rough microsomal membranes were isolated as described (37). The membranes were resuspended in 30 mM Tris-HCl buffer, pH 7.5, containing 3 mM MgCl 2 , 1 mM dithiothreitol, 35% glycerol (v/v) at a concentration of 10 mg/ml protein. For solubilization, membranes were diluted to 6.5 mg/ml with 10 mM Tris-HCl, pH 7.2, 3 mM MgCl 2 , and 150 mM NaCl, and Nonidet P-40 was added to a final concentration of 1.5%. After 20 min incubation on ice the solubilized extract was separated from insoluble material by centrifugation at 150,000 ϫ g for 40 min. Unless otherwise stated, all steps were carried out at 4°C.

GlcNAc-transferase Assays
Assay I-To measure the transfer of GlcNAc from UDP-GlcNAc to endogenous lipid acceptors in microsomal membranes, the reaction contained 7 mM Tris-HCl, pH 7.5, 0.05 Ci of UDP-GlcNAc (227 Ci/ mmol), 13 mM MgCl 2 , and a 60 l membrane fraction in a final volume of 75 l. After 15 min at 24°C the reaction was stopped with chloroform/methanol 3:2 (v/v). Labeled lipids were extracted and washed as described (11) and separated by TLC.
Assay II-GlcNAc-transferase activity in solubilized extracts was determined with GlcNAc 1 -PP-Dol as acceptor and UDP-GlcNAc as donor. The reaction contained the following in a final volume of 0.06 ml: 50 mM Tris-HCl, pH 7.5, 10 mM MgCl 2, 1 mM dithiothreitol, 25% glycerol, [ 14 C]GlcNAc 1 -PP-Dol (3000 cpm), 0.1 mM UDP-GlcNAc, and solubilized enzyme (equivalent to 30 g of membrane protein). After incubation at 24°C for the times indicated, GlcNAc 1/2 -PP-Dol was extracted and analyzed by TLC as described above. Radioactivity was detected by phosphorimaging or x-ray film.
Assay III-Determination of enzyme activity by immunoprecipitation of Alg13-ZZ or Alg14-ZZ was carried out as follows. 0.2 ml of solubilized extract was incubated on ice for 30 min with 0.05 ml of IgG-Sepharose 6 Fast Flow (Amersham Biosciences) by gentle tumbling. IgG-Sepharose 6 was equilibrated before use with 15 mM Tris-HCl, pH 7.5, 1.5 mM MgCl 2 150 mM NaCl, and 17.5% glycerol. The affinity matrix was centrifuged and washed three times with 0.2 ml of 1.5 mM Tris-HCl, pH 7.5, containing 150 mM NaCl and 0.1% Nonidet P-40, 15 mM MgCl 2 , and 17.5% glycerol. The pellet was resuspended in a 40-l assay mixture: 17 mM Tris-HCl, pH 7.5, 13 mM MgCl 2 , 0.8 mM dithiothreitol, 26% glycerol, 0.08% Nonidet P-40, [ 14 C]GlcNAc 1 -PP-Dol (3000 cpm), and 0.1 mM UDP-GlcNAc. The reaction was incubated at 24°C, The C-terminal region of MurG has been shown to bind UDP-GlcNAc. Homology of Alg13p exists also to the CspF galactosyltransferase and of Alg14p to its enhancer CspG. In the E. histolytica an unknown ORF (NCBI accession no. EAL50504) can be identified with Alg13p/Alg14p modules, which, however, are in opposite arrangement to MurG. stopped with chloroform/methanol (3:2) after the times indicated, and processed as for assay II.

IgG-Sepharose Affinity Chromatography
Solubilized enzyme extract was prepared as described above and subjected to a IgG-Sepharose 6 Fast Flow column (0.5 ml) chromatography, prepared according to the manufacturer, and equilibrated in 15 mM Tris-HCl, pH 7.5, containing 1.5 mM MgCl 2 , 150 mM NaCl, 17.5% glycerol, and 0.1% Nonidet P-40. 0.5 ml solubilized extract was loaded, and the column was washed with 10 column volumes of equilibration buffer to remove unbound material. Elution was carried out with 100 mM glycine-HCl buffer, pH 3.3, and fractions of 0.3 ml were collected and analyzed by Western blotting using antibodies against the ZZ (for Alg14p) or myc (for Alg13p) epitopes.

Identification in Silico of the Yeast ORFs YGL047w and YBR070c as
Potential GlcNAc-transferases-Early steps in biosynthesis of the lipidlinked oligosaccharide are essential for the viability of yeast. To identify candidate genes encoding the not yet identified GlcNAc-transferase, catalyzing the transfer of the second GlcNAc residue in LLO biosynthesis to give rise to GlcNAc 2 -PP-Dol, the S. cerevisiae data base was searched for essential genes with a potential glycosyltransferase motif. The survey revealed the unknown ORF YGL047w as a protein annotated to the glycosyltransferase family 28 C-terminal domain (pfam04101). Because of structural analysis this domain has been suggested to be involved in UDP-GlcNAc binding. We reasoned that this gene might be a potential candidate for the not yet identified GlcNActransferase. Ygl047p also shares homology with the C-terminal half of the E. coli MurG enzyme, which belongs to the glycosyltransferase family 28 as well, and catalyzes the transfer of the GlcNAc residue from UDP-GlcNAc to lipid-linked N-aycetylmuramoyl pentapeptide, an intermediate of peptidoglycan biosynthesis. The YGL047w gene codes for a predicted protein of 22.6 kDa that seems to have no signal sequence and only a weak candidate membrane-spanning segment.
We also realized that in a genome-wide protein-protein interaction project (Yeast Resource Center, Seattle, WA) the essential Ybr070c protein was detected among other proteins as a potential candidate interacting with Ygl047p. Interestingly Ybr070p has homology to the Cps14f gene product involved in the formation of capsular polysaccharide of Streptococcus pneumoniae. cps14f was reported to stimulate cps14g encoding the ␤-1,4-galactosyltransferase. This enzyme catalyzes the transfer of the galactose unit, the second sugar moiety of the capsular  polysaccharide (38). In addition, cps14f has homology to the N-terminal half of murG. Therefore we reasoned that Ybr070p may also be involved in the formation of GlcNAc 2 -PP-Dol. In contrast to Ygl047p, the amino acid sequence of Ybr070p predicts a protein with an N-terminal signal peptide and one to three trans-membrane segments. As we have demonstrated (see below) that these two proteins are indeed involved in the GlcNAc transfer onto GlcNAc 1 -PP-Dol, we designated the YGL047w locus as ALG13 and the YBR070c locus as ALG14.
To extend our understanding of how Alg13 and Alg14 proteins may be organized, we used the two amino acid sequences to search protein data bases for homologous proteins using the PHI (pattern-hit initiated) BLAST algorithm (39). We found Alg13 and Alg14 homologous proteins in virtually all eukaryotic cells including yeast, man, mammals, insects, worms, and plants. In general, Alg13 or Alg14 proteins are shorter than members of the bacterial MurG proteins. In almost all cases the Alg13p and Alg14p homologs encode separate polypeptides. An exception is the Entamoeba histolytica sequence, encoding a protein of 345 amino acids in length, that clearly contains both an Alg13 and an Alg14 protein domain. The domains are fused to each other but in an inverse order as compared with bacterial MurG proteins (Fig. 1).
Repression of ALG13 and ALG14 Causes a Defect in N-Glycosylation and Accumulation of GlcNAc 1 -PP-Dol-To analyze the function of ALG13 and ALG14 genes in N-glycosylation, we examined the glycosylation pattern of carboxypeptidase Y (CPY), a well characterized glycoprotein, in strains in which chromosomal ALG13 and ALG14 are under the control of the tetracycline-regulated tetO 7 promoter (40). The addi-tion of tetracycline, which causes repression of the respective gene, led to an under-glycosylation of CPY in both strains (Fig. 2). Mature CPY contains four N-linked glycan chains, whereas a glycosylation defect gives rise to CPY species lacking one or more chains and thus migrating faster on SDS-gels.
Under-glycosylation could be caused by a reduced transfer of truncated, non-optimal oligosaccharides, a reduced pool of full-length core oligosaccharides, or a defect in the glycan transfer onto the nascent polypeptide chain. We therefore investigated the formation of LLO by metabolic labeling of yeast cells with [ 3 H]mannose and [ 3 H]glucosamine, respectively, under conditions in which ALG13 was expressed or turned off by tetracycline. Fig. 3A shows the mannoselabeled LLO fraction after release of the glycan moieties by mild acid and subsequent size fractionation of the oligosaccharides by HPLC. In the presence or absence of tetracycline, the whole spectrum of LLOs was visible up to the full-length glucose-containing core, but a dramatic decrease in the total amount occurred upon down-regulation of Alg13p (Fig. 3A, traces A and B). We reasoned that a defect in the initiating steps of LLO biosynthesis might be the cause of this decrease. To verify this interpretation, we labeled cells with [ 3 H]glucosamine and analyzed by TLC the chloroform/methanol (3:2) extract containing the more hydrophobic dolichol-linked saccharides with short sugar chains (Fig. 3B). Upon down-regulation of ALG13 a radioactive peak accumulated that comigrated with GlcNAc 1 -PP-Dol at the expense of GlcNAc 2 -PP-Dol (Fig. 3B, traces A and B). The minor, shorter migrating peaks than Glc-NAc 2 -PP-Dol are supposed to be Man 1-2 GlcNAc 2 -PP-Dol. These results indicate that Alg13p is required for the formation of GlcNAc 2 -PP-Dol, and the longer LLOs during mannose labeling, under conditions in which ALG13 is turned off (Fig. 3A, trace B), are the result of residual Alg13p (see also below). Essentially similar results were obtained, when a tetO 7 -ALG14 strain was labeled with either [ 3 H]mannose (Fig. 3A, traces C and D) or [ 3 H]glucosamine (Fig. 3B, traces C and D) in the absence or presence of tetracycline.
In Vitro Assay of UDP-GlcNAc: GlcNAc 1 -PP-Dol GlcNAc-transferase in Microsomal Membranes and Detergent-solubilized Extracts-The above results indicate that both Alg13p and Alg14p are involved in the formation of GlcNAc 2 -PP-Dol. Therefore we tried to prove this more directly by measuring this enzymatic reaction in vitro. Microsomal membranes were prepared from cells expressing ALG13 and ALG14, respectively, or from cells in which the genes had been "turned off." As shown in Fig. 4, membranes from cells depleted of Alg13p or Alg14p were still able to form GlcNAc 1 -PP-Dol and also GPI-linked GlcNAc when incubated with UDP-[ 14 C]GlcNAc, but the formation of Glc-NAc 2 -PP-Dol was abolished. This finding was in clear contrast to incubations in which microsomes were used that were prepared from cells  expressing ALG13 or ALG14 genes. To further this, we established an assay to measure specifically the formation of GlcNAc 2 -PP-Dol, instead of using the endogenous lipid acceptors of the membrane. Detergentsolubilized extracts were incubated with [ 14 C]GlcNAc 1 -PP-Dol as acceptor and unlabeled UDP-GlcNAc as donor, and the formation of [ 14 C]GlcNAc 2 -PP-Dol was followed. As shown in Fig. 5, the extension of GlcNAc 1 -PP-Dol occurred only with extracts from membranes expressing Alg13p (panel A) and Alg14p (panel B) but not in extracts from membranes depleted for these proteins by the addition of tetracycline.
Alg13p and Alg14p Are Required for Enzyme Activity-Because the ALG13 and ALG14 genes encoded proteins that have not been detected thus far, we introduced a protein A epitope (ZZ) at the C terminus of ALG13 and ALG14, respectively, and cloned these variants into the constitutive expression vector pVT100. We used this epitope to immunoprecipitate Alg13-ZZp and Alg14-ZZp from solubilized extracts from wild-type cells with IgG-Sepharose and to measure the formation of GlcNAc 2 -PP-Dol in the immunoprecipitate. In the experiment depicted in Fig. 6 [ 14 C]GlcNAc 1 -PP-Dol was almost completely converted to [ 14 C]GlcNAc 2 -PP-Dol (lane 1) during a 25-min incubation. No elongation was measurable in a control experiment in which Sepharose without covalently linked IgG was used as the affinity matrix for precipitation (Fig. 6, lane 2; the asterisk indicates a small contamination of Dol-PP-GlcNAc 2 in the synthesized substrate). In contrast, an immunoprecipitate of Alg14-ZZp solubilized from wild-type membranes under the same conditions was enzymatically inactive (Fig. 6, lanes 3  and 4). We also determined the GlcNAc-transferase activity in an Alg13-ZZp immunoprecipitate from a tetO 7 -ALG14 strain expressing ALG14 (Fig. 6, lane 5) or after ALG14 has been shut off (lane 6). We observed that Alg13-ZZp was no longer active under the latter conditions. This could be because Alg13p needs Alg14p for activity or/and because Alg13p is depleted when ALG14 is repressed (see also below).
Localization of Alg13p to the Membrane Is Dependent on ALG14-As mentioned above, Alg13p, in contrast to Alg14p, does not have a pronounced predicted membrane spanning domain. During the analysis of expression of these proteins we realized that Alg13p localizes both to FIGURE 6. GlcNAc-transferase activity in Alg13 and Alg14 immunoprecipitates. Membranes from wild-type SS328 cells (lanes 1-4) or tetO 7 ALG14 cells (lanes 5 and 6) harboring ALG13-ZZ and ALG14-ZZ, respectively, were isolated and solubilized, and the extracts were immunoprecipitated with IgG-Sepharose beads (lanes 1, 3, 5, and 6) or with Sepharose beads as a control (lanes 2 and 4). The precipitate was assayed for transfer of GlcNAc from UDP-GlcNAc to [ 14 C]GlcNAc 1 -PP-Dol (assay III). Incubation time was 45 min.
The asterisks indicate a small contamination of Dol-PP-GlcNAc 2 in the synthesized [ 14 C]GlcNAc 1 -PP-Dol substrate. FIGURE 7. Localization and activity of Alg13p. A, GlcNAc-transferase activity in Alg13p immunoprecipitates from the membrane and cytosolic fraction. Solubilized extract (0.4 ml) from membranes and cytosol isolated from wild-type cells harboring ALG13-ZZ were subjected to immunoprecipitation with IgG-Sepharose, and the activity was determined with assay III; incubation was for 45 min. B, Western blot analysis. Membranes and cytosol of wild-type cells harboring ALG13-ZZ, ALG14-ZZ, or ALG3-ZZ were prepared and subjected to SDS-PAGE followed by transfer to nitrocellulose. Blots were probed with anti-mouse IgG peroxidase conjugate whole molecule (Sigma). C, Western blot of Alg13p and Wbp1p in membranes from tetO 7 -ALG14 cells after repression by tetracycline for 15 and 27 h, respectively. Alg13-ZZ and Alg3-ZZ was probed with anti-mouse IgG peroxidase conjugate whole molecule (Sigma), and Wbp1p was decorated with an anti-Wbp1 antiserum (35).
the membrane and to the cytosolic fraction (Fig. 7B). By quantitative Western blot analysis we calculated that slightly more than 50% of Alg13p was in a soluble form in the cytosol. In contrast, Alg14p was exclusively associated with the membrane fraction. As a control, another glycosyltransferase of the LLO pathway, the Alg3-mannosyltransferase, catalyzing the formation of Man 6 GlcNAc 2 -PP-Dol from Man 5 GlcNAc 2 -PP-Dol, is shown. This enzyme is exclusively membrane-bound as well. Next we asked whether the cytosolic Alg13p is enzymatically active as well. As depicted in Fig. 7A, an immunoprecipitate of cytosolic Alg13p was not able to catalyze the formation of [ 14 C]GlcNAc 2 -PP-Dol, compared with the Alg13p form solubilized from microsomal membranes.
In Fig. 7C it is shown that localization of Alg13p to the membrane is dependent on expression of ALG14. By Western blot analysis we observed that down-regulation of ALG14 by tetracycline in a tetO 7 -ALG14 strain, harboring a chromosomal ALG13-ZZ-tagged variant, Alg13p, decreased in the microsomal membrane fraction in a time-dependent manner. In contrast, as a control, the amount of membranebound Wbp1p, a subunit of the oligosaccharyltransferase, did not change. But as expected, an under-glycosylation of Wbp1p occurs as a consequence of the glycosylation defect and is reflected by a faster mobility on the SDS-gel. Wbp1p contains two N-linked glycan chains. Furthermore, we investigated the level of membrane-bound Alg13p in comparison with the GlcNAc-transferase activity in the tetO 7 -ALG13 strain expressing chromosomal ALG13-ZZ. Membranes were isolated from cells repressed by tetracycline at the times indicated, and Alg13p was quantified in Western blots. The GlcNAc-transferase activity was determined in solubilized extracts with [ 14 C]GlcNAc 1 -PP-Dol as acceptor and UDP-GlcNAc as donor. As depicted in Fig. 8 the protein level and the enzyme activity decrease in a similar manner when ALG13 expression is turned off.
Physical Interaction between Alg13p and Alg14p-The results obtained thus far suggest that both proteins may interact physically to form the active enzyme. To demonstrate such an interaction, we constructed a yeast strain co-expressing an ALG14-ZZ and an ALG13 myctagged variant. A detergent-solubilized extract was isolated from microsomal membranes of this strain and applied to an IgG-Sepharose column that recognizes the ZZ epitope, and the eluate was analyzed by Western blot analysis for Alg14p and Alg13p. As shown in Fig. 9, A and B, both proteins co-fractionated. In contrast, when the solubilized extract was chromatographed on Sepharose material without IgG coupling, neither Alg14p-ZZ nor Alg13p-myc was detectable (Fig. 9, C and  D). This indicates that both proteins associate in a complex and that the binding to the affinity matrix is specific. The somewhat earlier elution of the more hydrophilic Alg13p may indicate that the interaction is prone to some dissociation.

DISCUSSION
In this article we have identified and characterized two essential genes, ALG13 and ALG14, necessary for the formation of GlcNAc 2 -PP-Dol both in vitro and in vivo. The participation of two genes in a reaction of lipid-linked oligosaccharide biosynthesis is novel and was unexpected, because in yeast all other glycosyltransferases of this pathway known so far are catalyzed by only a single polypeptide. We show that down-regulation of each gene causes hypoglycosylation of glycoproteins (CPY and Wbp1), accumulation of GlcNAc 1 -PP-Dol during metabolic labeling with [ 3 H]glucosamine, and a defect in the transfer of GlcNAc from UDP-GlcNAc onto GlcNAc 1 -PP-Dol in membranes as well as in solubilized extracts.
Alg13p and Alg14p, identified in this work, are evolutionarily conserved proteins in eukaryotic cells and also have a similarity to domains of glycosyltransferase involved in bacterial cell wall glycoconjugate and polysaccharide biosynthesis, respectively (41)(42)(43). For most eukaryotes, the Alg13 and Alg14 proteins are translated from two individual loci found in the genome, the ALG13 and the ALG14 loci, respectively. Interestingly, in the E. histolytica we found only one gene encoding a FIGURE 8. Expression of Alg13p and GlcNAc-transferase activity. TetO 7 -ALG13 cells containing chromosomal tagged Alg13-ZZ were repressed by tetracycline for the times indicated. The level of Alg13p in membranes was monitored by Western blot analysis, and GlcNAc-transferase activity was determined in solubilized extracts using assay II. protein that clearly contained the Alg13 and Alg14 (Fig. 1) domains. Sequence analysis of Alg13p revealed that it maps to the C-terminal region of the MurG GlcNAc-transferase that, based on crystal structure data, is thought to bind the UDP-GlcNAc donor. Alg14p, on the other hand, maps to the N-terminal domain of MurG containing a structural glycine-rich loop motif, postulated to be involved in recognition of the lipid-linked PP i acceptor, in this case MurNAc-PP-polyprenyl (44). Conserved glycine residues as part of a potential G-loop occur also in Alg14p ( 61 GSGGHTG 67 ). Analysis of the membrane topology of Alg13p indicates that it has neither a signal sequence for ER translocation nor a well defined trans-membrane domain. In contrast, topology predictions for Alg14p suggest a hydrophobic N-terminal signal peptide and one to three trans-membrane segments. In accord with these predictions, Western blot analysis showed that Alg13p is localized both to the cytosol and membrane fractions, whereas Alg14p is associated exclusively with the membrane (Fig. 7B). The membrane-bound form of Alg13p seems to be dependent on the presence of Alg14p, because shutting off ALG14 causes the disappearance of Alg13p from the membrane (Fig. 7C). It is not known whether Alg13p is released into the cytosol or is degraded in the absence of Alg14p. We postulated that Alg14p is needed to recruit Alg13p to the membrane and to give rise to the active glycosyltransferase.
In support of this hypothesis, we were able to demonstrate a physical interaction of Alg13p and Alg14p by affinity chromatography (Fig. 9). Furthermore, when the enzyme activity was determined in an Alg13p immunoprecipitate obtained from membrane detergent extracts from wild-type cells (Fig. 6, lane 1) or tetO 7 -ALG14 cells under non-repressed conditions (Fig. 6, lane 5), an efficient elongation of GlcNAc 1 -PP-Dol was observed. But when ALG14 was shut off (Fig. 6, lane 6), no enzyme activity was detectable in the Alg13p immunoprecipitate, presumably because of depletion of Alg13p (see Fig. 7C). Lack of Alg14p may also explain why cytosolic Alg13p is inactive (Fig. 7A, lane 2).
The data at present, however, do not allow us to decide what the precise function of Alg13p and Alg14p, respectively, is in the glycosyltransfer. The available x-ray crystal structures and the structural similarities among the sequences of glycosyltransferases suggest Alg14p as a candidate to be involved in acceptor binding, whereas Alg13p may contain the transferase activity. It is also not clear at the moment, while demonstrating a physical association between the two proteins, why an immunoprecipitate using Alg14p as bait (Fig. 6, lane 3) leads to an inactive enzyme, whereas the Alg13p immunoprecipitate is active (Fig.  6, lane 1). We suppose that under the stringent washing conditions of the immunoprecipitate, the somewhat water-soluble Alg13p dissociates from Alg14p, which is not or is less the case in the reverse situation when Alg13p is the target. What could be the benefit of having two separate modules in the eukaryotic enzymes as compared with the prokaryotic homologs? One possibility could be that such an organization provides a level of regulation of this early step of eukaryotic N-glycosylation to assure a cellularly coordinated and efficient formation of the unique core oligosaccharide, when needed. Independently, a further detailed structural and biochemical analysis will be required to unravel the interplay of the two proteins. Having identified the genes in the synthesis of GlcNAc 2 -PP-Dol, one should also be able to diagnose potential defects in the existing human orthologs that may cause a new subtype in the congenital disorders of glycosylation.