Impaired O-Linked N-Acetylglucosaminylation in the Endoplasmic Reticulum by Mutated Epidermal Growth Factor (EGF) Domain-specific O-Linked N-Acetylglucosamine Transferase Found in Adams-Oliver Syndrome*

Background: EOGT (epidermal growth factor (EGF) domain-specific O-linked N-acetylglucosamine) mutations have been identified in patients with Adams-Oliver syndrome (AOS). Results: O-Linked N-acetylglucosaminylation (O-GlcNAcylation) of EGF domains in the endoplasmic reticulum (ER) is impaired in all EOGT variants associated with AOS. Conclusion: AOS-causative EOGT mutations affect ER O-GlcNAcylation. Significance: Impaired EOGT glycosyltransferase activity and a consequent reduction in O-GlcNAcylation underlie the etiology of EOGT-related AOS. Epidermal growth factor (EGF) domain-specific O-linked N-acetylglucosamine (EOGT) is an endoplasmic reticulum (ER)-resident O-linked N-acetylglucosamine (O-GlcNAc) transferase that acts on EGF domain-containing proteins such as Notch receptors. Recently, mutations in EOGT have been reported in patients with Adams-Oliver syndrome (AOS). Here, we have characterized enzymatic properties of mouse EOGT and EOGT mutants associated with AOS. Simultaneous expression of EOGT with Notch1 EGF repeats in human embryonic kidney 293T (HEK293T) cells led to immunoreactivity with the CTD110.6 antibody in the ER. Consistent with the GlcNAc modification in the ER, the enzymatic properties of EOGT are distinct from those of Golgi-resident GlcNAc transferases; the pH optimum of EOGT ranges from 7.0 to 7.5, and the Km value for UDP N-acetylglucosamine (UDP-GlcNAc) is 25 μm. Despite the relatively low Km value for UDP-GlcNAc, EOGT-catalyzed GlcNAcylation depends on the hexosamine pathway, as revealed by the increased O-GlcNAcylation of Notch1 EGF repeats upon supplementation with hexosamine, suggesting differential regulation of the luminal UDP-GlcNAc concentration in the ER and Golgi. As compared with wild-type EOGT, O-GlcNAcylation in the ER is nearly abolished in HEK293T cells exogenously expressing EOGT variants associated with AOS. Introduction of the W207S mutation resulted in degradation of the protein via the ubiquitin-proteasome pathway, although the stability and ER localization of EOGTR377Q were not affected. Importantly, the interaction between UDP-GlcNAc and EOGTR377Q was impaired without adversely affecting the acceptor substrate interaction. These results suggest that impaired glycosyltransferase activity in mutant EOGT proteins and the consequent defective O-GlcNAcylation in the ER constitute the molecular basis for AOS.

extracellular domains of Notch receptors (10). Moreover, in vivo studies revealed that O-GlcNAc is abundantly expressed in the Drosophila cuticle (11). Among cuticle proteins, Dumpy, a giant membrane-anchored cuticle protein containing a very large number of EGF-like domains, was identified as a major O-GlcNAcylated protein. In addition to Notch and Dumpy, Delta, and Serrate, ligands for Notch receptors, have been shown to be O-GlcNAcylated. The addition of O-GlcNAc onto extracellular proteins is catalyzed by EGF domain-specific O-GlcNAc transferase (EOGT).
The biological function of extracellular O-GlcNAc was first suggested by the phenotype of the Eogt mutant in Drosophila (11). Although the Eogt mutant does not exhibit the classical Notch phenotype, it shows the defects in the wings, notum, and cuticle (i.e. wing blistering, vortex, and cuticle detachment), similar to the dumpy mutant. A genetic interaction was observed between Eogt and dumpy or Eogt and pyrimidine metabolism pathway components in the wing blister phenotype (12). These studies suggested that Eogt regulates Dumpy-mediated processes directly or indirectly through regulation of pyrimidine metabolism.
EOGT is evolutionarily conserved from Caenorhabditis elegans to humans. Accordingly, extracellular O-GlcNAc is found on five extracellular O-GlcNAcylated proteins (Hspg2, Nell1, Lama5, Pamr1, and Notch2) present in the mouse cerebral cortex (13). It was also suggested that thrombospondin-1 is O-GlcNAcylated in platelets (14). Sequence alignment of O-GlcNAcylated proteins suggested that the predictive consensus sequence for the modification is CXXGX(T/S)GXXC, which is located between the fifth and sixth conserved cysteines of the EGF domain. In contrast to the pivotal roles of extracellular O-GlcNAc in Drosophila development, much less is known regarding the biological requirement of EOGT in mammals (15). The first implication came from a recent report of an EOGT mutation in a rare human congenital disorder, Adams-Oliver syndrome (AOS), which is characterized by aplasia cutis congenital and terminal transverse limb defects (16,17). The aim of this study was to biochemically analyze EOGT mutations linked to AOS and demonstrate that a defect in O-GlcNAcylation in the endoplasmic reticulum (ER) caused by EOGT mutations constitutes the molecular basis for AOS.
Preparation of Recombinant EGF20 -To prepare the unglycosylated recombinant EGF domain, the V5-hexahistidinetagged 20th EGF domain of Drosophila Notch (EGF20) was prepared using the Kluyveromyces lactis protein expression kit (New England Biolabs). In brief, the EGF20:V5His fragment was cloned into the pKLAC1 expression vector, and the resultant plasmid was transformed into K. lactis GG799 cells (20). The transformed K. lactis cells were streaked on yeast carbon base (YCB) agar plates containing 5 mM acetamide; then, individual colonies were incubated in 1 ml of YPGal (1% yeast extract, 2% peptone, 2% galactose) medium for 2 days. The cell suspension was transferred to 500 ml of fresh YPGal medium

Impaired ER O-GlcNAcylation in Adams-Oliver Syndrome
and incubated for an additional 4 days to obtain saturated culture. The culture medium was collected by centrifugation at 4,160 ϫ g for 15 min, filtered through a 0.45-m filter, and concentrated 10-fold using Spin-X UF concentrators (5-kDa cut-offs; Corning). Concentrated medium was supplemented with one-tenth volume of FastBreak Cell Lysis Reagent (Promega), incubated for 30 min at 4°C, and passed through the column packed with 3 ml of Complete His-Tag purification resin (Roche Applied Science). After washing with 100 mM HEPES, 10 mM imidazole, pH 7.5, bound proteins were eluted with 100 mM HEPES, 500 mM imidazole, pH 7.5. The eluate was concentrated and buffer exchanged into 100 mM HEPES, pH 7.0, using Spin-X UF concentrators (5-kDa cut-offs).
O-GlcNAc Transferase Assay-EOGT assay was performed in 20 l of the glycosylation buffer (250 mM HEPES, pH 7.5, and 1 mM MnCl 2 ) containing 5 M UDP-[ 3 H]GlcNAc (2 Ci/mmol), 2.0 g of EGF20-V5His, and 1.0 g of FLAG-tagged EOGT at 37°C for 60 min unless otherwise noted. The reaction was applied to a LC-18 SPE tube (Supelco), washed with 5 ml of MilliQ, and eluted with 1 ml of 80% acetonitrile and 0.052% trifluoroacetic acid. Radioactivity in the elution was measured using a liquid scintillation counter (Aloka).
Purification of N1EGF1-36-MycHis and Western Blotting-For the purification of N1EGF1-36-MycHis from cell lysates, cells were lysed with lysis buffer, and then the lysates were combined with 0.2 g of anti-myc antibody (4A6, Upstate). After incubation for 2 h at 4°C, 5 l of Protein G Mag-Sepharose (GE Healthcare) was added to the lysates, which were then incubated for 1 h at 4°C. The immunoprecipitates were washed extensively with 50 mM Tris-HCl, pH 7.4, 150 mM NaCl, and 0.25% Triton X-100 and then eluted with SDS-PAGE sample buffer containing 2% SDS and 70 mM 2-mercaptoethanol. For Western blotting, each sample was separated by SDS-polyacrylamide gel and transferred onto a PVDF membrane (Millipore). The membrane was probed with the appropriate primary antibody (1:1000 dilution) and HRP-conjugated secondary antibody (1:5000 dilution); this was followed by detection with Immobilon Western Chemiluminescent HRP Substrate (Millipore) and exposure to x-ray film. Quantification of ECL-exposed films was performed using a flatbed scanner, GT-X900 (Epson), and ImageJ 1.48 software.
Immunostaining-HEK293T cells were fixed with 4% paraformaldehyde in PBS for 15 min and cold methanol for 5 min. Cells were permeabilized with 0.5% Triton X-100 in PBS for 15 min and blocked with PBS containing 5% FBS for 30 min. Immunostaining was performed by incubating first with the appropriate primary antibody (1:100 dilution) for 2 h and then with a fluorescent dye-conjugated secondary antibody (1:1000 dilution) for 1 h. All samples were analyzed using a FV10 confocal microscopy (Olympus).
Hexosamine Treatment-HEK293T cells cultured in lowglucose DMEM/FBS were transiently transfected to express N1EGF1-36-MycHis with or without EOGT in Opti-MEM cell culture medium (Invitrogen) using Lipofectamine 2000 transfection reagent (Invitrogen). After 6 h, transfected cells were incubated for 36 h in low-glucose DMEM or low-glucose DMEM supplemented with 25 mM glucose, 10 mM glucosamine, 10 mM GlcNAc, or various concentrations of mannitol.
Glucosamine was prepared as a 1 M stock solution dissolved in 500 mM HEPES, pH 7.0.
Ultracentrifugation-HEK293T cells expressing wild-type or mutant forms of EOGT were lysed in lysis buffer (Cell Signaling Technology), and cell debris was removed by centrifugation at 17,860 ϫ g for 15 min. The supernatants were further ultracentrifuged at 135,000 ϫ g for 17 h. The supernatants and pellets were subjected to immunoblot analysis.
Binding Assay for EOGT-To detect binding between EOGT and EGF20, HEK293T cells were transfected with pSegtag2/ EOGT-IRES-GFP, pSegtag2/EOGT R377Q -IRES-GFP, or pSectag2-IRES-GFP. Transfected cells were cultured for 2 days and then lysed in lysis buffer. Next, 0.1 g of EGF20-V5His was added to 500 l of the cell lysates, which were then incubated for 2 h at 4°C. The mixtures were then purified using MagneHis nickel particles (Promega) to isolate EGF20-V5His-binding proteins, which were then analyzed by immunoblotting. Alternatively, 5 g of purified EOGT or EOGT R377Q were incubated with 2.0 g of EGF20-V5His in 500 l of lysis buffer in the presence or absence of 5 M UDP-GlcNAc. After incubation for 2 h at 4°C, the mixtures were purified to isolate the EGF20⅐EOGT complex using MagneHis nickel particles, which were then analyzed by immunoblotting.
To detect binding between EOGT and EGF20, CHO cells stably transfected with pEF/EOGT-FLAG or pEF/EOGT R377Q -FLAG were lysed in lysis buffer and 500 l of the cell lysates were immunoprecipitated using anti-FLAG M2 affinity gel beads (Sigma). After washing with PBS containing 0.05% Triton X-100, 0.1 g of EGF20-V5His was added to the beads and then incubated for 2 h at 4°C. The mixtures were washed with PBS supplemented with 0.05% Triton X-100 to isolate the EGF20⅐ EOGT complex and analyzed by immunoblotting.
To detect binding between EOGT and UDP-GlcNAc, HEK293T cells were transfected with pEF/EOGT-FLAG, pEF/ EOGT R377Q -FLAG, or pEF/mycHis control vector, cultured for 2 days, and lysed in lysis buffer. Five hundred microliters of cell lysates were then supplemented with 5 M UDP-[ 3 H]GlcNAc and 2.0 g of EGF20. After incubation for 30 min on ice, the mixtures were immunoprecipitated using anti-FLAG M2 affinity gel beads (Sigma) and radioactivity was measured.

RESULTS
O-GlcNAcylation in the ER-EOGT contains a hydrophobic region corresponding to a signal peptide at the amino terminus, and a KDEL-like ER-retrieval sequence at the carboxyl terminus ( Fig. 1A) (11). ER localization of EOGT was detected in HEK293T cells transiently transfected to express EOGT (Fig.  1B). N1EGF1-36-MycHis was used to monitor EOGT-catalyzed O-GlcNAcylation in the cells, as the Notch1 EGF repeats possess an array of potential O-GlcNAcylation sites (10).
It was recently reported that CTD110.6 antibody binds terminal ␤-GlcNAc on complex N-glycans (21). To exclude the possibility that EOGT generates a CTD110.6-reactive epitope on N-glycan, we analyzed the immunoreactivity of N1EGF1-36-MycHis toward CTD110.6 antibody after removal of N-glycans. PNGase treatment of affinity-purified N1EGF1-36-MycHis revealed that the CTD110.6 reactivity of N1EGF1-36-MycHis was not altered, whereas N-glycans on the anti-Myc tag antibody were barely observed (Fig. 1D). These results confirmed the specificity of CTD110.6 for O-GlcNAc under our experimental conditions.
Except for a few proteins such as Notch receptors, most O-GlcNAcylation occurs by the action of intracellular OGT. We thus extended our analysis by examining the effect of reducing intracellular OGT activity on O-GlcNAcylation of N1EGF1-36-MycHis. As expected, OGT esiRNA treatment did not affect the O-GlcNAcylation level of N1EGF1-36-MycHis as compared with control esiRNA-treated cells (Fig. 1E). These results demonstrate that EOGT is the sole enzyme that catalyzes O-GlcNAcylation of Notch1 EGF repeats.
EOGT Enzymatic Properties-Initial experiments examining the enzymatic properties of EOGT showed that EOGT exhibits O-GlcNAc transferase activity specific to folded EGF domains (11,19). However, there were no kinetic parameters available for EOGT, and thus it was not possible to compare the kinetics of ER-resident EOGT to intracellular OGT and other GlcNAc transferases that localize in the Golgi. Therefore, we performed a detailed characterization of EOGT activity. To obtain purified enzyme, FLAG-EOGT was stably expressed in CHO cells and isolated from cell lysates using anti-M2-agarose. EOGT O-GlcNAc transferase activity was detected in a time-or dosedependent manner in a buffer containing UDP-[ 3 H]GlcNAc and EGF20 as donor and acceptor substrates, respectively ( Fig.  2A and data not shown). To determine the optimum pH for EOGT, O-GlcNAc transferase activities were measured at pH values ranging from 5.5 to 8.5 (Fig. 2B). Pronounced activity was observed at pH 7.0, as compared with that seen at pH 6.5. Moreover, only marginal activity was detected at pH 5.5 or 8.5. These data show that EOGT is more active at a neutral pH, corresponding to the estimated pH in the ER lumen (22). Intracellular OGT has been shown to have extremely high affinity for UDP-GlcNAc compared with other GlcNAc transferases in the Golgi (23,24). To explore whether the kinetic parameters of EOGT are distinct from those of intracellular OGT and Golgi-resident GlcNAc transferases, an activity assay was performed using increasing amounts of UDP-GlcNAc. Lineweaver-Burk analysis of the resulting data yields a K m value of 25 M and a V max of 92 nmol/min/mg (Fig. 2C). EOGT also exhibits a linear increase in activity when increasing amounts of EGF20 were used as an acceptor substrate. However, kinetic constants for EGF20 were not determined in our experiments (Fig. 2D).
EOGT-catalyzed O-GlcNAcylation Responds to Stimulation of the Hexosamine Biosynthesis Pathway-Glucose and glucosamine are metabolically converted into UDP-GlcNAc via the hexosamine biosynthesis pathway (HBP) (9). Previous research revealed that subsets of GlcNAc transferases with low affinity for UDP-GlcNAc respond to HBP stimulation to elevate the GlcNAc branch on N-glycans (25,26). In Drosophila S2 cells, O-GlcNAcylation by Eogt increases in response to supplementation with glucosamine and GlcNAc. However, the relatively low K m value for UDP-GlcNAc in mouse EOGT led us re-examine whether O-GlcNAcylation of N1EGF1-36-MycHis could similarly be responsive to supplementation with glucose, glucosamine, or N-acetylglucosamine in HEK293T cells. Under culture conditions using low-glucose DMEM, the level of N1EGF1-36-MycHis O-GlcNAcylation was increased by HBP stimulation using glucose, glucosamine, or N-acetylglucosamine (Fig. 3A) in EOGT-transfected cells. The change in O-GlcNAcylation level was not due to the osmotic stress because the supplementation of mannitol did not affect the O-GlcNAcylation of N1EGF1-36-MycHis (Fig. 3B). To confirm the elevation of O-GlcNAcylation levels in the ER, immunostaining with CTD110.6 antibody was performed (Fig. 3C). Upon HBP stimulation, the O-GlcNAc signal in the ER was elevated in EOGT-transfected cells. These results indicate that the luminal UDP-GlcNAc concentration in the ER is controlled in a distinct manner by unique UDP-GlcNAc transporters.
Impaired O-GlcNAcylation in EOGT Mutants Associated with AOS-Recently, the homozygous mutations in EOGT were identified in some patients with AOS (16,17). To date, three homozygous mutations for EOGT have been reported (Fig. 4). Of these, W207S and R377Q are missense mutations that change amino acid residues absolutely conserved across species from C. elegans to Homo sapiens (16). In contrast, the G359D fs * 28 mutation is a 1-bp deletion that causes a frameshift at Gly-359 and subsequent truncation. However, a molecular characterization of these mutations has not yet been performed.
To biochemically characterize the EOGT variants found in AOS, we generated mouse EOGT constructs harboring the same W207S or R377Q mutation found in the AOS patients (Fig. 4A). To characterize the G359D fs * 28 mutation, we generated a ⌬360 -527 construct that mimics the truncated form of EOGT produced by the G359D fs * 28 mutation. To analyze O-GlcNAcylation in these mutants, N1EGF1-36-MycHis and the respective EOGT constructs were co-expressed in HEK293T cells. N1EGF1-36-MycHis from the transfected cells was then immunoprecipitated and subsequently analyzed by immunoblotting with CTD110.6 and anti-Myc antibodies (Fig. 5A). The O-GlcNAcylation levels for all EOGT mutants were comparable with that obtained from mock transfectants, whereas wildtype EOGT produced a strong O-GlcNAcylation signal (Fig.  5B). Anti-O-GlcNAc antibody immunoreactivity was specifically competed away in the presence of 0.1 M GlcNAc (Fig. 5A). These results demonstrated that the EOGT mutations reduce enzyme function. Consistent with this immunoblot data, HEK293T cells cotransfected with the EOGT mutants did not have detectable O-GlcNAc staining in the ER (Fig. 5C). These results indicate that defective O-GlcNAcylation in the ER is the molecular basis for AOS.

EOGT Mutant Protein Levels and Subcellular Localization-
To analyze the precise mechanisms for defective O-GlcNAcylation seen with EOGT mutants, we quantified the expression level of each EOGT construct in HEK293T cells. In the experimental plasmids, the cDNA encoding the respective EOGT mutants is followed by an internal ribosome entry site (termed IRES) and a GFP sequence. Therefore, GFP expression was used as the internal control reference (Fig. 6A). Normalizing the expression levels for the EOGT variants revealed that the W207S mutation decreased expression by ϳ70%, whereas the R377Q mutation did not affect the level of protein expression (Fig. 6B). Deletion of the 360 -527 sequence also decreased the expression level of EOGT by ϳ60%. This decrease is not attributable to the missing ER-retrieval sequence at the carboxyl terminus, because the addition of an HDEL sequence to the ⌬360 -527 construct failed to recover the protein expression level. These results suggest that protein stability appears to be affected in EOGT W207S and EOGT ⌬360 -527 .
To further analyze the molecular mechanisms underlying the defective O-GlcNAcylation seen with the EOGT mutants, we examined the subcellular localization of the EOGT variants. Similar to EOGT WT , EOGT R377Q mainly localized in the ER, as detected by the ER marker calnexin (Fig. 6C). In contrast, the expression level of EOGT W207S in the ER was lower than that of EOGT WT , and immunostaining for EOGT ⌬360 -527 is hardly detectable in the ER. Moreover, EOGT W207S and EOGT ⌬360 -527 exhibited a punctate signal that only partially overlapped that of calnexin. Importantly, the addition of an HDEL sequence to EOGT ⌬360 -527 failed to restore ER localization, suggesting that the altered subcellular localization of EOGT ⌬360 -527 was not due to lack of the ER-retrieval sequence.
The mislocalization of EOGT W207S and EOGT ⌬360 -527 mutants prompted us to examine whether AOS-related mutations might affect the solubility of EOGT. Ultracentrifugation of Triton X-100-solubilized EOGT revealed that wild-type EOGT and EOGT R377Q were recovered in the supernatant fraction, whereas the majority of EOGT W207S and EOGT ⌬360 -527 were found in the pellet (Fig. 6D). These data suggest that W207S and ⌬360 -527 mutations, but not the R377Q mutation, affect the solubility of EOGT. Taken together, these results indicate that W207S and ⌬360 -527 mutations affect the stability and/or structure of EOGT, leading to its aberrant subcellular localization.
EOGT W207S Is Degraded through the Ubiquitin-Proteasome Pathway-Next, we attempted to analyze the stability of the EOGT W207S mutant using MG132, an ubiquitin-proteasome inhibitor. As shown in Fig. 7A, the level of EOGT W207S expression was clearly increased after 1 h of MG132 treatment. In contrast, MG132 had no effect on EOGT WT . The sensitivity of EOGT W207S to the ubiquitin-proteasome inhibitor prompted us to ask whether EOGT W207S is modified by ubiquitin (Fig.  7B). In the absence of MG132, ubiquitination of EOGT WT and EOGT W207S was barely observed. In contrast, the ubiquitination level of EOGT W207S was dramatically increased upon MG132 treatment. However, the diminished O-GlcNAcylation of N1EGF1-36-MycHis by EOGT W207S was not recovered by the MG132 treatment (Fig. 7C). These data suggested that the W207S mutation adversely affects protein folding in EOGT, leading to protein degradation and a consequent absence of O-GlcNAcylation in the ER.
We then asked whether EOGT ⌬360 -527 is similarly degraded through the ubiquitin-proteasome pathway. Although EOGT ⌬360 -527 was modified by ubiquitin in the presence of MG132, the expression level of EOGT ⌬360 -527 appeared to be unaffected (Fig. 7D), suggesting that molecular mechanisms unrelated to the ubiquitin-proteasome pathway would contribute to the decreased expression level of EOGT ⌬360 -527 .
Characterization of the Enzymatic Properties of EOGT R377Q -To characterize the enzyme activity of EOGT R377Q , FLAG-EOGT R377Q was stably expressed in CHO cells and purified to homogeneity. As shown in Fig. 8A, enzyme activity was not detectable in the R377Q mutants.
To address the effect of the R377Q mutation on the enzymatic properties of EOGT, binding capacities of EOGT R377Q to acceptor substrates and UDP-GlcNAc were analyzed. For the assay examining the binding of EOGT to EGF20, lysates from HEK293T cells transfected to express either EOGT WT or EOGT R377Q were mixed with recombinant EGF20, and the subsequent EOGT⅐EGF20 complexes were recovered from the mixtures using Ni ϩ beads. Unexpectedly, the amount of EGF20 coprecipitated with EOGT R377Q was higher than that coprecipitated with EOGT WT (Fig. 8B).
To confirm this result, purified EOGT WT or EOGT R377Q was mixed with recombinant EGF20 in the presence or absence of UDP-GlcNAc, affinity purified, and analyzed as above. The amount of EGF20 that copurified with EOGT R377Q was higher than the amount that copurified with EOGT WT . Inclusion of UDP-GlcNAc in the mixture markedly decreased the EGF20-EOGT WT interaction, but did not obviously affect the EGF20-EOGT R377Q interaction (Fig. 8C). These results demonstrate that the R377Q mutation does not affect acceptor substrate binding in EOGT.
Based on these results, a reverse-binding experiment was conducted using lysates from CHO cells stably expressing either FLAG-EOGT WT or FLAG-EOGT R377Q . After purification on the M2 anti-FLAG resin, recombinant EGF20 was supplemented in the presence of UDP-GlcNAc, and the EOGT⅐EGF20 complexes were recovered from the mixtures. As expected, physical interaction between EOGT R377Q and EGF20 was detected (Fig. 8D). We further performed a binding analysis at the cellular level (Fig. 8E). Co-immunoprecipitation of N1EGF1-36-MycHis with wild-type or mutant forms of EOGT revealed that EOGT R377Q and, to a lesser degree, wild-type EOGT were bound to N1EGF1-36-MycHis. No evident binding between N1EGF1-36-MycHis and EOGT W207S or EOGT ⌬360 -527 was observed.
The apparent increase in the binding affinity of EOGT R377Q for acceptor substrates could be due to a decreased affinity for UDP-GlcNAc, which would result in the stabilization of an acceptor substrate-enzyme complex. However, the lack of enzymatic activity for EOGT R377Q hampered our attempt to directly measure the K m value for UDP-GlcNAc. As an alternative approach, either FLAG-EOGT WT or FLAG-EOGT R377Q was incubated with tritium-labeled UDP-GlcNAc (Fig. 8F). After incubation for 30 min on ice, the proteins were immunoprecipitated from the mixtures using anti-FLAG beads. Although complexes with UDP-[ 3 H]GlcNAc were detected in both cases, the signal obtained with FLAG-EOGT R377Q was lower than that obtained with FLAG-EOGT WT . These results indicate that EOGT R377Q is defective in binding with UDP- GlcNAc; this defect may contribute, at least in part, to the lack of enzymatic activity seen in the mutant EOGT.

DISCUSSION
AOS is a rare congenital disorder characterized by aplasia cutis congenital and terminal transverse limb defects. AOS is genetically heterogeneous, and the molecular pathology appears complex. Until now, DOCK6 (27), ARHGAP31 (28), RBPJ (29), EOGT, and Notch1 (30) have recently been reported as AOS-causative genes. However, the molecular consequences of EOGT mutations have not been characterized. In this study, we demonstrated that EOGT mutations associated with AOS resulted in the loss of the ability of EOGT to O-GlcNAcylate EGF domains in the ER. Importantly, even though EOGT R377Q is catalytically inactive, it retains the ability to bind an acceptor substrate. The unique enzymatic properties of EOGT R377Q exclude the possibility of a glycosyltransferase-independent function of EOGT in the pathogenesis of AOS. Our findings show that impaired EOGT glycosyltransferase activity and defective O-GlcNAcylation in the ER constitutes the molecular basis for AOS.
Although EOGT and intracellular OGT both catalyze protein O-GlcNAcylation, EOGT has no sequence similarity with intracellular OGT. The EOGT-related gene GTDC2 encodes O-mannose ␤1,4-N-acetylglucosaminyltransferase, which acts on ␣-dystroglycan (31)(32)(33)(34). Interestingly, arginine at position 377 is conserved between EOGT and GTDC2 (Fig. 4B). Given that both GTDC2 and EOGT are involved in GlcNAc modification in the ER and that Arg-377 in EOGT is conserved in GTDC2, the arginine residue may  t test). C, HEK293T cells were transfected to express wild-type EOGT or the indicated EOGT mutants. Immunostaining was performed using anti-EOGT (magenta) and anti-calnexin (green) antibodies. Bar, 10 m. D, ultracentrifugation of Triton X-100-solubilized EOGT. Note that wild-type EOGT and EOGT R377Q were recovered in the supernatant fraction (S), whereas the majority of EOGT W207S and EOGT ⌬360 -527 were found in the pellet (P).  . The R377Q mutant exhibits lack of enzymatic activity and impaired UDP-GlcNAc binding. A, in vitro O-GlcNAc transferase assay was used to measure the enzymatic activity of the R377Q mutant. B, recombinant Notch EGF20:His was incubated with cell lysates from HEK293T mock transfectants or transfectants expressing EOGT WT or EOGT R377Q , affinity purified from the mixtures using Ni ϩ beads, and subjected to immunoblotting with indicated antibodies. C, purified EOGT WT or EOGT R377Q was incubated with recombinant Notch EGF20:His with or without UDP-GlcNAc, affinity purified, and subjected to immunoblotting with the indicated antibodies. D, Notch EGF20:His was incubated with cell lysates from CHO cells (Cont) or stable transfectants expressing FLAG-EOGT WT or FLAG-EOGT R377Q , immunoprecipitated (IP) with anti-FLAG antibody-conjugated beads, and subjected to immunoblotting as indicated. E, HEK293T cells were cotransfected to express N1EGF1-36-MycHis and the indicated EOGT constructs. Cell lysates were immunoprecipitated with anti-Myc tag antibody, and N1EGF1-36-MycHis⅐EOGT complexes were analyzed by immunoblotting with the indicated antibodies. F, lysates from HEK293T cells expressing FLAG-EOGT WT or FLAG-EOGT R377Q were incubated with UDP-[ 3 H]GlcNAc and EGF20 on ice and immunoprecipitated using anti-FLAG antibody, after which the level of radioactivity was measured. Lysates from mock transfectants were similarly processed for use as a control. Vertical bars represent the range of values obtained from duplicate samples. be important in the recognition of UDP-GlcNAc by EOGT and GTDC2.
It has been reported that the respective GlcNAc transferases exhibit different affinity for UDP-GlcNAc. Intracellular OGT has been shown to exhibit the highest affinity for UDP-GlcNAc among all GlcNAc transferases, which have K m values ranging from 545 nM to 8.5 M (23, 24). In contrast, Golgi-resident GlcNAc transferases exhibit much higher K m values than OGT. For example, the K m value of UDP-GlcNAc for Mgat1 is 40 M, and for Mgat2 is 0.96 mM (35). Even higher K m values for UDP-GlcNAc have been reported for Mgat4, with 5.0 mM, and Mgat5, with 11.0 mM (25,35). EGF repeats in Notch receptors are also modified with O-fucose, which is further modified with ␤1,3-linked GlcNAc by Lfng, Rfng, and Mfng, exhibiting K m values of 78, 53.6, and 196 M, respectively (36). Our finding of a K m value of EOGT for UDP-GlcNAc (25 M) suggested that the affinity of EOGT for UDP-GlcNAc is higher than that of most GlcNAc transferases except OGT. In the Golgi, Mgat4 and Mgat5 show lower affinities for UDP-GlcNAc, and their activities are limited by UDP-GlcNAc concentrations (26,35). Thus, branching of N-glycans by Mgat4 and Mgat5 is stimulated by a hexosamine flux that leads to UDP-GlcNAc production (26). Paradoxically, our data showed that O-GlcNAcylation of Notch 1 EGF repeats by EOGT is also stimulated by hexosamine treatment. The Golgi concentration of UDP-GlcNAc was calculated to be ϳ1.5 mM, 15-fold higher than that of cytoplasmic UDP-GlcNAc (37). Given that EOGT has a high affinity for UDP-GlcNAc but is still responsive to hexosamine flux, the concentration of UDP-GlcNAc in the ER would be maintained at a lower level than that in the Golgi.
Nucleotide sugars are transported from the cytosol to the ER and the Golgi by the nucleotide-sugar transporters, highly conserved hydrophobic proteins with multiple transmembrane domains that are part of the solute carrier family 35 (SLC35) (38,39). Of these, SLC35A3 (40 -42) and SLC35D2 (43,44) are reported to be Golgi-resident UDP-GlcNAc transporters. In contrast, SLC35B4 has been shown to transport UDP-Xyl and UDP-GlcNAc (45), but the reports of the subcellular localization of this protein are somewhat inconsistent. Although an initial report suggested that SLC35B4 localizes in the Golgi (45), a more recent report suggests that it localizes in the ER (46). These conflicting results could be due to the different cell types used in the two studies. SLC35D1 is also reported to be an ER nucleotide-sugar transporter that transports UDP-GlcNAc, UDP-GalNAc, and UDP-GlcUA (47). It would be interesting to determine whether SLC35B4, SLC35D1, and/or a yet unidentified transporter are responsible for O-GlcNAcylation by EOGT in the ER.
Currently, the molecular mechanisms of how impaired O-GlcNAcylation leads to the anomalies observed in AOS remain unknown. Given that loss of function mutation for DOCK6 (27) and gain of function mutation for ARHGAP31 (28) cause AOS, O-GlcNAcylation might be involved in the regulation of the actin cytoskeleton (16). Alternatively, EOGT might be required for optimal Notch signaling, as a heterozygous mutation for RBPJ, which encodes a transcriptional factor involved in Notch signaling, also causes AOS (29). In this regard, it would be interesting to hypothesize that O-GlcNAc might regulate Notch receptor function, as it does with other O-glycans, including O-fucose and O-glucose glycans (48 -55). However, no evidence for such a conclusion is provided by the genetic studies in Drosophila (11,12). Eogt mutants in Drosophila show phenotypes related to dumpy (dp). Dumpy is O-GlcNAcylated by EOGT, and Eogt and dp genetically interact to create the dp phenotype, which includes wing blistering. Although Eogt similarly interacts with the components for the Notch signaling pathway in wing blistering (12), the effect of EOGT on the Notch signaling pathway could be indirect, because Notch signaling per se is involved in the regulation of pyrimidine synthesis, which is likely to promote wing blistering (12). As multiple substrates may be modified by EOGT, a careful examination of EOGT function with respect to Notch receptors and other glycoproteins should be conducted to elucidate the biological function of O-GlcNAcylation in the ER and its contribution to pathological processes.