Non-lysosomal Degradation of Singly Phosphorylated Oligosaccharides Initiated by the Action of a Cytosolic Endo-β-N-acetylglucosaminidase*

Phosphorylated oligosaccharides (POSs) are produced by the degradation of dolichol-linked oligosaccharides (DLOs) by an unclarified mechanism in mammalian cells. Although POSs are exclusively found in the cytosol, their intracellular fates remain unclear. Our findings indicate that POSs are catabolized via a non-lysosomal glycan degradation pathway that involves a cytosolic endo-β-N-acetylglucosaminidase (ENGase). Quantitative and structural analyses of POSs revealed that ablation of the ENGase results in the significant accumulation of POSs with a hexasaccharide structure composed of Manα1,2Manα1,3(Manα1,6)Manβ1,4GlcNAcβ1,4GlcNAc. In vitro ENGase assays revealed that the presence of an α1,2-linked mannose residue facilitates the hydrolysis of POSs by the ENGase. Liquid chromatography-mass spectrometric analyses and fluorescent labeling experiments show that such POSs contain one phosphate group at the reducing end. These results indicate that ENGase efficiently hydrolyzes POSs that are larger than Man4GlcNAc2-P, generating GlcNAc-1-P and neutral Gn1-type free oligosaccharides. These results provide insight into important aspects of the generation and degradation of POSs.

Dolichol-linked oligosaccharides (DLOs) 3 are glycan donor substrates for asparagine (N)-linked glycosylation in mammals (1,2). The biosynthesis of DLOs starts on the cytosolic side of the endoplasmic reticulum (ER) membrane with the assembly of Man 5 GlcNAc 2 -PP-Dol (1). This biosynthetic intermediate is flipped into the ER lumen and is converted to Glc 3 Man 9 GlcNAc 2 -PP-Dol. The fully assembled DLO is then transferred onto nascent polypeptides by the oligosaccharyltransferase (3).
In mammalian cells, the biosynthesis of DLOs is regulated by the supply of glucose. When cells are deprived of glucose, incompletely assembled DLOs are produced (4 -9), which are rapidly degraded by a currently unclarified degradation system (2,10,11). The enzyme involved in the degradation has long been believed to hydrolyze the pyrophosphate bond of DLOs, releasing phosphorylated oligosaccharides (POSs) into the cytosol (Fig. 1, step i) (10 -14). Although previous biochemical studies using paper electrophoresis (11) or anion-exchange chromatography (10,14) suggest the presence of one phosphate group on POSs, the precise number has not been unequivocally determined, which has hampered the identification of the precise action of the DLO-degrading enzyme.
In addition to POSs, a protein-unbound form of neutral high mannose-type N-glycans (free N-glycans; FNGs), which bare N,NЈ-diacetylchitobiose at the reducing end (Gn2-type), is produced in the cytosol via two distinct pathways as follows: the degradation of DLOs, presumably by oligosaccharyltransferase in the ER lumen (2,15,16), and their transport to the cytosol (Fig. 1, step ii); and the deglycosylation of misfolded glycoproteins by a cytosolic peptide:N-glycanase (PNGase) (Fig. 1, step iii) (2,17,18). In the cytosol, the Gn2-type FNGs are processed by an endo-␤-N-acetylglucosaminidase (ENGase) (19 -21) that hydrolyzes the innermost GlcNAc residue to generate Gn1type FNGs (Fig. 1, step iv). The action of ENGase is prerequisite for the subsequent trimming of mannose from the FNGs by the ␣-mannosidase Man2C1 (Fig. 1, step v) (21,22). This process, which is referred to as non-lysosomal glycan degradation, accelerates the transport of FNGs into lysosomes and their subsequent degradation (Fig. 1, step vi) (23). Although the molecular mechanism responsible for how POSs are catabolized remains unclear, it has been suggested that POSs predominantly accumulate in the cytosol (10,13,14).
In this study, we investigated the issue of whether cytosolic ENGase catalyzes the hydrolysis of POSs (Fig. 1, step vii), because an N,NЈ-diacetylchitobiose core structure is still intact at their reducing ends. Structural analyses of POSs by liquid chromatography-mass spectrometry (LC-MS) showed that the reducing end of the POSs is mono-phosphorylated. Comparisons of POSs produced in wild-type and Engase knock-out (Engase Ϫ/Ϫ ) mouse embryonic fibroblasts (MEFs) revealed that POSs that carry Man␣1,2Man␣1,3(Man␣1,6)Man␤1, 4GlcNAc␤1,4GlcNAc accumulate to a significant extent in Engase Ϫ/Ϫ MEFs. Consistent with this observation, in vitro ENGase assays for POSs with various structures clearly indi-cated that this enzyme preferentially hydrolyzes POSs that contain an ␣1,2-linked mannose residue. These results strongly indicate that ENGase efficiently catabolizes POSs that are larger than Man 4 GlcNAc 2 -P and generates Gn1-type FNGs and GlcNAc-1-P in the cytosol.
Preparation of POSs from MEFs-POSs were prepared essentially as described in a previous report (10). Briefly, MEFs were extracted with 70% (v/v) ethanol, and the soluble fraction was evaporated to dryness. The dried pellet was dissolved in 2.5 ml of water and desalted by passage through a PD-10 column (GE Healthcare) equilibrated with 5% (v/v) ethanol. The desalted sample was adjusted to 10 mM Tris-HCl, pH 7.4, and subjected to anion-exchange column chromatography using a Sep-Pak Accell QMA column (Waters). POSs were eluted from the anion-exchange column with 5 ml of an elution buffer containing 10 mM Tris-HCl, pH 7.4, and 70 mM NaCl.
For the analysis of POSs by LC-MS, the eluates from the anion-exchange column were desalted on PD-10 columns equilibrated with 5% (v/v) ethanol, followed by passing through a 250-l CM-Sepharose (GE Healthcare) equilibrated with water. The flow-through fraction, which contains POSs, was evaporated to dryness and dissolved in a small volume of water and then subjected to LC-MS analysis.
Pyridylamination-POSs eluted from the anion-exchange chromatography column were dephosphorylated by treatment with recombinant calf intestine alkaline phosphatase (CIP; 1 unit, Roche Applied Science) for 16   and labeled with 2-aminopyridine (PA) as described previously (15). High Performance Liquid Chromatography (HPLC)-Size fractionation HPLC (15) and dual-gradient reversed-phase HPLC (25) were carried out as described previously. PA-glycans were quantitated using PA-Glc 6 that is included in PA-glucose oligomer (Takara; 2 pmol/l) as an external standard.
LC-MS of POSs-POSs (17 pmol as Man 2 GlcNAc 2 -P) obtained from the glucose-deprived wild-type MEFs (10) were analyzed by LC-MS using a quadrupole ion trap time-of-flight mass spectrometer (LCMS-IT-TOF, Shimadzu Corp.) in electrospray ionization (ESI) in the negative ion mode. For analysis, 10-l aliquots (maintained at 4°C in the auto sampler) were injected for separation by HPLC, using gradient elution, with a Nexera XR LC system (Shimadzu Corp.) on a Mastro C18 column (100 ϫ 2.1 mm; particle size 3.0 m) at a flow rate of 0.3 ml min Ϫ1 , with the column maintained at 40°C. The chromatographic system used a binary solvent system delivered as a gradient of Solvent A (15 mM ammonium acetate, 10 mM tributylamine solution) and Solvent B (methanol). The initial gradient conditions were 100% A, 0% B for 0.5 min followed by a linear gradient up to 25% B in 8 min, and up to 98% B over the next 4 min. The solvent composition was then held at 98% B for 3 min after which the column was returned to 0% B over the next 5 min, making a total cycle time of 20 min per sample. The following method parameters were used for sample analysis: mass range of m/z 200 -1500 in MS; ion source temperature of 200°C; heated capillary temperature of 200°C; electrospray voltage of Ϫ3.5 kV; electrospray nebulization gas flow 1.5 liters/ min; detector voltage 1.7 kV; ion accumulation time 30 ms.
The negative ion mode was used. Mass calibration was carried out using a trifluoroacetic acid sodium solution (2.5 mmol liter Ϫ1 ) from 200 to 1500 Da. Data acquisition and processing used software LCMS Solution 3.80.
To verify component identification, the Formula Predictor software was used (Shimadzu Corp.). Predicting a candidate list based on MS data takes into account a number of variables, including isotopic profile analysis, mass accuracy, and mass resolution of the experimentally derived pseudomolecular ion peak and related fragment ion data.
Large Scale Preparation of the DLOs-Saccharomyces cerevisiae strains used to prepare DLOs are listed in Table 1. Wildtype, alg6⌬, and alg3⌬ cells were transformed with pRS425-GPD-ALG7 to overproduce DLOs (15). The transformants were grown in complete synthetic medium without leucine (26). In contrast, alg11⌬ cells were grown in YPD medium (1% yeast extract, 2% polypeptone, 2% glucose). The DLOs were prepared from those strains, and their amounts were quantitated as described previously (15). Preparation of Detergent-solubilized Mouse Liver Microsomes-Microsomes were prepared from livers of C57BL/6 mice as described previously (27), except that 30% (v/v) glycerol was used instead of 1.3 M sucrose as a cushion to sediment the rough ER by ultracentrifugation. Mouse liver microsomes obtained were homogenized in a homogenization buffer containing 20 mM Hepes-KOH, pH 7.4, 150 mM KCl, 5 mM CoCl 2 , and 1% (w/v) Triton X-100 and ultracentrifuged at 100,000 ϫ g for 20 min. The supernatant was recovered as detergent-solubilized microsomes. All experimental protocols and procedures were approved by the Ethical Committees on Animal Research of RIKEN.
In Vitro DLO-Pyrophosphatase Assay-DLOs (4 nmol) were dissolved in 25 l of homogenization buffer and incubated for 16 h at 37°C with 25 l of detergent-solubilized microsomes supplemented with glycosidase inhibitors (10 M swainsonine, 100 M kifunensine, and 1 mM castanospermine). The reaction was terminated by the addition of 3 volumes of ethanol. After centrifugation at 15,000 ϫ g for 10 min, the supernatant was evaporated to dryness. POSs were purified from the dried pellet and desalted as described above under "Preparation of POSs from MEFs." Transfection-Ngly1 Ϫ/Ϫ Engase Ϫ/Ϫ MEFs (1.0 ϫ 10 6 cells) were seeded to a 10-cm dish and incubated in 10 ml of complete DMEM without antibiotics for 24 h before transfection. The cells were transfected with 1 g of pcDNA3.1-mouse ENGase-FLAG (24) using 5 l of Lipofectamine 2000 (Thermo Fischer Scientific). At 16 h after transfection, the cells were harvested, and the cytosolic fraction was prepared.
Preparation of the Cytosolic Fraction-Ngly1 Ϫ/Ϫ Engase Ϫ/Ϫ MEFs overexpressing mouse ENGase were washed three times with ice-cold phosphate-buffered saline. The cells obtained from a 10-cm dish were homogenized in 300 l of cytosol extraction buffer containing 10 mM Hepes-NaOH, pH 7.4, 250 mM mannitol, 5 mM dithiothreitol, 1 mM EDTA, 1ϫ complete protease inhibitor mixture (EDTA-free, Roche Applied Science), and 1 mM Pefabloc (Roche Applied Science). The homogenate was centrifuged at 10,000 ϫ g for 10 min at 4°C. The supernatant was further ultracentrifuged at 100,000 ϫ g for 60 min at 4°C to remove the microsomal fraction, and the supernatant was recovered as the cytosol fraction.
To remove endogenous FNGs, the cytosol fraction (400 l) was concentrated to ϳ50 l using an Amicon Ultra MWCO 50-kDa filter (Millipore) by centrifugation at 10,000 ϫ g for 6 min at 4°C and diluted with 450 l of cytosol extraction buffer. After the dilution-concentration step was carried out twice, the final concentrate (ϳ 100 l) was recovered, aliquoted, flashfrozen in liquid nitrogen, and stored at Ϫ80°C until used.
Endoglycosidase Digestion of POSs-To verify the elution positions of Gn1-type FNGs in dual-gradient reversed-phase HPLC (25), POSs (50 pmol) that were obtained as described above under "In Vitro DLO-Pyrophosphatase Assay" were digested with either Endo-H (for Glc 3 Man 9 GlcNAc 2 -P and Man 9 GlcNAc 2 -P; Roche Applied Science, 5 milliunits) or Endo-M (for Man 5 GlcNAc 2 -P and Man 3 GlcNAc 2 -P; Tokyo Chemical Industry, 1 milliunit) in 30 l of ENGase reaction buffer at 37°C for 4 h. The reaction was terminated by the addition of 300 l of 75% (v/v) ethanol, desalted, and fluorescently labeled as described above under "ENGase Activity Assay."

Results
POSs Are Mono-phosphorylated-To unequivocally determine the number of phosphate groups of the POSs, POSs that were produced in glucose-deprived wild-type MEFs (10) (Fig. 2). However, no doubly charged deprotonated molecules [M Ϫ 2H] 2Ϫ corresponding to diphosphorylated POSs were detected. These results clearly show that the POSs were mainly monophosphorylated, consistent with the predictions made by previous biochemical studies (10,11,14).
Ablation of Engase Results in the Accumulation of POSs That Are Larger than Man␣1,2Man␣1,3(Man␣1,6)Man␤1, 4GlcNAc␤1,4GlcNAc-P-Because ENGase catalyzes the hydrolysis of ␤1,4-linkage of the N,NЈ-diacetylchitobiose core of protein-bound (24) and unbound (19,21) forms of high mannosetype glycans in the cytosol of mammalian cells, we speculated that POSs, which are exclusively localized in the cytosol (10,13,14), may also serve as physiological substrates for ENGase. To test this hypothesis, the amounts and glycan structures of the POSs were compared between wild-type and Engase Ϫ/Ϫ MEFs. The rationale behind this experiment is that if ENGase hydrolyzes POSs, the knock-out of Engase would result in the accumulation of POSs. To detect POSs by a conventional fluorescent labeling method for the reducing end of glycans (29), POSs prepared from glucose-deprived MEFs were treated with or without CIP, fluorescently labeled, and separated by size-fractionation HPLC (Fig. 3A) (10). Fluorescently labeled glycans that appeared in a dephosphorylation-dependent manner (Fig.  3A, peaks a-h) were regarded as POSs. The fluorescently labeled POS-derived glycans thus obtained were fractionated   and further analyzed by dual-gradient reversed-phase HPLC (25) to determine their isomeric structures. In both glucosedeprived wild-type and Engase Ϫ/Ϫ MEFs, Man 0 -7 GlcNAc 2 -P (Fig. 3B) were detected, and their glycan structures were identical to biosynthetic intermediates of DLOs (Table 2). Quantitative analyses of POSs showed that, in both glucose-deprived wild-type and Engase Ϫ/Ϫ MEFs, Man 2 GlcNAc 2 -P predominantly accumulated, and the knock-out of Engase did not markedly affect the amounts (Fig. 3B). Unexpectedly, however, the amounts of GlcNAc 2 -P were significantly decreased in Engase Ϫ/Ϫ MEFs by an unclarified reason (Fig. 3B). By contrast, the amounts of Man 4 -7 GlcNAc 2 -P were significantly increased in Engase Ϫ/Ϫ MEFs (Fig. 3B). A similar accumulation of Man 4 -6 GlcNAc 2 -P was also observed in Engase Ϫ/Ϫ MEFs that were cultivated under normal glucose conditions (Fig. 3C), indicating that the availability of glucose is not a key factor in regulating the catabolism of Man 4 -6 GlcNAc 2 -P. The glycan structures of POSs that significantly accumulated in Engase Ϫ/Ϫ MEFs were found to contain a hexasaccharide structure com-posedofMan␣1,2Man␣1,3(Man␣1,6)Man␤1,4GlcNAc␤1,4GlcNAc  ( Fig. 3D). To assess whether the accumulation of these POSs in Engase Ϫ/Ϫ MEFs is caused by the lack of ENGase, Engase Ϫ/Ϫ MEFs were transfected with a plasmid-encoding mouse Engase, and POSs that were accumulated in the transfected cells were quantitated. The overexpression of ENGase in Engase Ϫ/Ϫ MEFs resulted in a significant decrease in Man 4,5 GlcNAc 2 -P and showed a propensity for the reduction in Man 6,7 GlcNAc 2 -P (Fig. 3E). The partial reduction of these POSs by ENGase overexpression was most likely due to low transfection efficiencies in MEFs (ϳ50% cells positive in the expression of green fluorescent protein used as a technical control; data not shown). All these results strongly indicate that ENGase is involved in the catabolism of POSs. ENGase Hydrolyzes POSs in Vitro-Although our data clearly show that the lack of ENGase causes the accumulation of Man 4 -7 GlcNAc 2 -P, the issue of whether ENGase catalyzes the hydrolysis of these POSs remained unknown.
In vitro ENGase assays using the POS preparations and the cytosolic fraction obtained from Ngly1 Ϫ/Ϫ Engase Ϫ/Ϫ MEFs overexpressing ENGase showed that the order of hydrolytic rates of the POSs was Glc 3 Man 9 GlcNAc 2 -P Ϸ Man 9 GlcNAc 2 -  Fig. 3A. b Elution positions of dephosphorylated, PA-labeled POSs were expressed as glucose units (GU) in dual-gradient reversed-phase HPLC (25). c Glycan structures were deduced by dual-gradient reversed-phase HPLC using authentic glycan standards (25).   (36 -41) and further confirmed by the product analyses in the ENGase assay ( Table 3). The DLOs were digested with mouse liver microsomes (microsomal digestion) to obtain POSs (G3M9P, M9P, M5P, and M3P, right panels). The resulting POSs were treated with (ϩ) or without (Ϫ) CIP before fluorescent labeling. B, time course analysis of the DLO-pyrophosphatase assay in the presence (ϩ) or absence (Ϫ) of microsomes using M5P as a substrate. C, time course analyses of the ENGase assay using M3P, M5P, M9P. and G3M9P as substrates. The amounts of Gn1-type FNGs were quantitated by sizefractionation HPLC. D, product analyses for the ENGase assay by dual-gradient reversed-phase HPLC. White arrowheads indicate the elution positions of authentic Gn1-type FNGs generated by Endo-H or Endo-M digestion of POSs. Black arrowheads indicate peaks that are generated by ENGase and co-eluted with the authentic standard glycans. Unassigned peaks generated by endoglycosidase digestions are probably derived from POSs that were co-purified with G3M9P, M9P, M5P, and M3P (see also A, right panels). E, ENGase assay using M5P as a substrate and the cytosolic fraction obtained from either ENGase-transfected (ϩ) or non-transfected (Ϫ) Ngly1 Ϫ/Ϫ Engase Ϫ/Ϫ MEFs as an enzyme source. The reaction mixtures were analyzed by sizefractionation HPLC. F, schematic representation of Glc 3 Man 9 GlcNAc 2 glycan. The position of the A, B, and C arms of the glycan are indicated. Asterisks in A and E indicate peaks derived from labeling reagents. PA-glucose oligomers (degree of polymerization ϭ 3-15) was used as a reference, and the glucose units (GU) are represented on the HPLC chart.
p Ͼ Man 5 GlcNAc 2 -P Ͼ Ͼ Man 3 GlcNAc 2 -P ( Fig. 4C and Table  3). Dual-gradient reversed-phase HPLC analysis (25) of the reaction mixtures confirmed the generation of Gn1-type FNGs, i.e. Glc 3 Man 9 GlcNAc 1 , Man 9 GlcNAc 1 , Man 5 GlcNAc 1 , and Man 3 GlcNAc 1 (Fig. 4D and Table 3). When the cytosolic fraction obtained from non-transfected Ngly1 Ϫ/Ϫ Engase Ϫ/Ϫ MEFs was used, no hydrolysis of Man 5 GlcNAc 2 -P was detected (Fig.  4E), verifying that the cytosolic fraction of the non-transfected Ngly1 Ϫ/Ϫ Engase Ϫ/Ϫ MEFs is devoid of degradation activity for POSs. Therefore, our assay system using the cytosol fraction of the ENGase-overexpressing Ngly1 Ϫ/Ϫ Engase Ϫ/Ϫ MEFs is a reliable one, especially when appropriate glycosidase inhibitors (swainsonine, kifunensine, and castanospermine) are included in the reaction mixture to prevent the glycan trimming of reaction products during the incubation (see "Experimental Procedures" for the detailed conditions). All these results indicate that, although ENGase has a broad substrate specificity for POSs, the presence of ␣1,2-linked mannose residues at the A arm (Fig. 4F), as well as a branching structure at the B and C arms, facilitates the action of this enzyme.

Discussion
A putative DLO-degrading enzyme has been implicated in the production of POSs (11), whereas the precise cleavage site of DLOs by this enzyme has not been unequivocally determined. Our LC-MS analyses of intact POSs showed that one phosphate group is present (Fig. 2) at the reducing end of POSs (Fig. 3A) (10,13), providing conclusive evidence that the putative DLO-degrading enzyme is a DLO-pyrophosphatase that splits the pyrophosphate bond of DLOs producing mono-phosphorylated POSs. Although mono-phosphorylated POSs have been identified by MS in bacteria (30) or in milk oligosaccharides (31), to the best of our knowledge this is the first example of the use of MS for the identification of DLO-derived POSs of mammalian origin.
In vitro DLO-pyrophosphatase assays revealed that the enzyme exhibits broad substrate specificities for glycan struc-tures (Fig. 4A, right panels), and even the fully assembled DLO (Glc 3 Man 9 GlcNAc 2 -PP-Dol) can be a substrate in vitro. This result raises the issue of why only limited POS structures, i.e. Man 0 -7 GlcNAc 2 -P, were detected in MEFs (Fig. 3, B and C) (10,13). These POSs are produced from both the cytosolic and luminal species of DLOs, which negligibly accumulate in MEFs (10). By contrast, Glc 3 Man 9 GlcNAc 2 -PP-Dol is the predominant DLO at the steady state in MEFs (10), whereas little Glc 3 Man 9 GlcNAc 2 -P is detected even in Engase Ϫ/Ϫ MEFs (Fig.  3, B and C). Our findings thus imply that the accessibility of the DLO-pyrophosphatase to the fully assembled DLO is somehow limited in vivo via unknown mechanisms. Although it is also possible that the glycan trimming of large POSs by glycosidases may generate small POSs from large POSs, such as Glc 3 Man 9 GlcNAc 2 -P, previous studies have shown that glycan structures of POSs are barely altered by treatment with various glycosidase inhibitors, such as castanospermine (13), kifunensine (13), and swainsonine (10,13), suggesting that such a possibility is unlikely. Irrespective of mechanisms involved, the identification of the membrane topology of the catalytic site(s) of the DLO-pyrophosphatase will provide deeper insights into important aspects of the regulation of the action of DLO-pyrophosphatase in vivo.
Our data showed that large POSs, i.e. Man 4 -7 GlcNAc 2 -P, carrying the Man␣1,2Man␣1,3(Man␣1,6)Man␤1,4GlcNAc␤1, 4GlcNAc structure tend to accumulate in Engase Ϫ/Ϫ MEFs (Fig. 3, B and C). The accumulation of these POSs in Engase Ϫ/Ϫ MEFs is consistent with the finding that, in vitro, ENGase efficiently hydrolyzes POSs bearing ␣1,2-linked mannose residues located at the A arm (Table 3). In contrast, ENGase hydrolyzes Man 3 GlcNAc 2 -P at much slower rates ( Table 3). The substrate specificity of ENGase is consistent with the previous report showing that glycans that contain a Man␣1,2Man␣1,3Man␤1, 4GlcNAc␤1,4GlcNAc structure serve as good substrates of ENGase (32). These findings can explain, at least in part, the reason why Man 0 -3 GlcNAc 2 -P accumulates to a lesser extent TABLE 3 In vitro ENGase activity assay using POSs with various glycan structures a Glycan structures were determined by the elution positions of both size-fractionation and dual-gradient reversed-phase HPLC and comparison with authentic standard glycans (25). To verify the position of Gn1-type FNGs under the HPLC conditions used, the reaction products were analyzed in parallel with POSs that were digested with Endo-H (for Glc 3 Man 9 GlcNAc 2 -P and Man 9 GlcNAc 2 -P) or Endo-M (Man 5 GlcNAc 2 -P and Man 3 GlcNAc 2 -P). b The amounts of reaction products were estimated from peak area of PA-Glc 6 in size-fractionation HPLC. The hydrolytic rates were estimated based on the amounts of Gn1-type FNGs at 1 h (Glc 3 Man 9 GlcNAc 1 , Man 9 GlcNAc 1 , and Man 5 GlcNAc 1 ) and 2 h (Man 3 GlcNAc 1 ) after incubation. c Hydrolytic rates of Man 9 GlcNAc 2 -P were set to 1.0. when compared with the larger POSs in Engase Ϫ/Ϫ MEFs. The glycan processing of POSs by ENGase is of biological relevance, because the putative FNG transporter that resides in the lysosomal membrane does not directly transport POSs into lysosomes (33). Therefore, the ENGase-mediated processing of POSs plays a crucial role in initiating their cytosolic clearance.
We noted that the levels of GlcNAc 2 -P were significantly reduced in Engase Ϫ/Ϫ MEFs, raising the possibility that the biosynthesis of GlcNAc 2 -PP-Dol is also compromised in glucosedeprived Engase Ϫ/Ϫ MEFs. The action of ENGase produces free GlcNAc and GlcNAc-1-P from Gn2-type FNGs and POSs, respectively (Fig. 1). These sugars may be recycled for the biosynthesis of UDP-GlcNAc, which is an essential sugar donor substrate for the assembly of GlcNAc 2 -PP-Dol. It is also possible that the catabolism of GlcNAc 2 -P is enhanced in the absence of ENGase. In this regard, it remains to be clarified how small POSs, which are poor substrates of ENGase, are catabolized. A putative POS-phosphatase activity was reported in the microsome fraction from human liver (14), but the molecular nature of this phosphatase is currently unknown. Moreover, we previously found that the induction of bulk autophagy by amino acid starvation reduces the levels of FNGs in the cytosol (34). Further studies will be needed to elucidate the molecular mechanism involved in the catabolism of small POSs.
In summary, the above findings demonstrate that POSs are mono-phosphorylated, thereby implicating the action of a DLO-pyrophosphatase in their production. Moreover, we identified the cytosolic ENGase as a POS-catabolizing enzyme that preferentially hydrolyzes POSs carrying the Man␣1, 2Man␣1,3(Man␣1,6)Man␤1,4GlcNAc␤1,4GlcNAc structure. This study provides mechanistic insights into the generation and the catabolism of POSs in mammalian cells.  Table 3. S. Y. and N. D. performed the analyses shown in Fig. 2. Y. H. and T. S. analyzed all the data and wrote the manuscript. All authors reviewed the results and approved the final version of the manuscript.