Rapid activation of glycogen phosphorylase by the endoplasmic reticulum unfolded protein response.

Endoplasmic reticulum (ER) stress is associated with misfolding of ER proteins and triggers the unfolded protein response (UPR). The UPR, in turn, helps restore normal ER function. Since fastidious N-linked glycosylation is critical for folding of most ER proteins, this study examined whether metabolic interconversions of precursors used for glycan assembly were controlled by the UPR. Thus, eight enzymes and factors with key roles in hexose phosphate metabolism were assayed in cytoplasmic extracts from primary dermal fibroblasts treated with UPR inducers. Stimulation of only one activity by the UPR was detected, AMP-independent glycogen phosphorylase (GP). GP activation required only 20 min of ER stress, with concurrent decreases in cellular glycogen and elevations of its metabolites Glc-1-P and Glc-6-P. Addition of phosphatase inhibitors to enzyme extracts from unstressed cells mimicked the effect of ER stress on GP activity, suggesting that phosphorylation of GP or a regulatory factor was involved. These data show that the UPR can modulate hexose metabolism in a manner beneficial for protein glycosylation. Since activation of GP appears to occur by a rapid post-translational process, it may be part of a general strategy of ER damage control, preceding the well-known transcription-dependent processes of the UPR that are manifested hours after the occurrence of ER stress.

Endoplasmic reticulum (ER) 1 stress initiates signals that emerge from the ER lumen, and activate cytoplasmic and/or nuclear responses, which in turn alter ER function. The general paradigm of ER stress signaling is termed the unfolded protein response (UPR) (1)(2)(3), which can be triggered by agents that cause misfolded proteins to accumulate within the ER lumen, such as dithiothreitol (DTT), tunicamycin (TN), thapsigargin, castanospermine (CSN), and azetidine-2-carboxylic acid (AZC) (4). Typically, ER stress activates the stress sensors Ire1p and ATF6 (3) to cause transcription of genes encoding chaperones, folding enzymes, and other proteins that enhance ER function. Appearance of the respective gene products occurs several hours after initiation of ER stress. For such ER proteins, there is no evidence that regulation by the UPR involves post-translational import or activation. Additionally, ER stress can temporarily inhibit cellular protein synthesis, lessening the load of misfolded protein entering the lumenal space. This occurs within 10 -20 min of the application of ER stress by rapid post-translational phosphorylation of eIF2␣ by the PKRlike ER kinase or PERK (5). Translation arrest is reversible (for example, lasting 2 h in fibroblasts, Ref. 4), allowing the subsequent translation of mRNAs encoded by UPR-responsive genes.
Previous studies from this laboratory with primary dermal fibroblasts identified another rapid effect of the UPR, involving N-linked glycosylation of ER proteins (4,6). ER quality control is highly dependent upon covalent attachment of the oligosaccharide Glc 3 Man 9 GlcNAc 2 to specific asparaginyl residues of nascent ER proteins. This requires the synthesis of a lipidlinked oligosaccharide (LLO), Glc 3 Man 9 GlcNAc 2 -P-P-dolichol. TN is a specific inhibitor of the synthesis of this LLO and, as a result, its application with cells causes ER protein misfolding and ER stress. Conditions that cause accumulation of premature LLO intermediates also result in ER stress (7). However, it was found that the UPR can compensate by promoting extension of such premature LLOs to Glc 3 Man 9 GlcNAc 2 -P-P-dolichol (6). Of particular interest is DTT-induced stress, which stimulates LLO synthesis within 20 min, and is attenuated by adaptation of cells to ER stress (4). This suggests a form of stress relief in which one or more components of the preexisting pathway for generation of Glc 3 Man 9 GlcNAc 2 -P-P-dolichol is activated post-translationally, rather than being synthesized de novo. No evidence was obtained for UPR activation of hexose transport or transferases involved in synthesis of Glc 3 Man 9 GlcNAc 2 -P-P-dolichol (6). Since multiple undermannosylated intermediates (Man 2-5 GlcNAc 2 -P-P-dolichol) were all extended by the UPR, it is feasible that the UPR acts by increasing synthesis of one or more precursors of the mannosyl residues.
In this study we present evidence that AMP-independent glycogen phosphorylase (GP) can mediate enhanced LLO extension by the UPR. Activation of GP is rapid, and may be part of a general program of ER damage control that precedes transcriptional events regulated by the UPR.

EXPERIMENTAL PROCEDURES
Cell Culture and Related Procedures-Human dermal fibroblasts, obtained from various sources (see below, Table II), were cultured and subjected to ER stress with DTT, TN, CSN, or AZC as described (6). Cells were cultured continuously in medium containing 5 mM glucose in all experiments presented. To verify the effects of UPR inducing treatments (except TN) on LLO synthesis, controls (not shown) were performed with cells labeled with 40 Ci/ml [2-3 H]mannose, in which case a 20-min incubation with medium containing 0.5 mM glucose was used.
Cytoplasmic Enzyme Extracts-Cell monolayers were treated with streptolysin-O (SLO (Murex brand, distributed by Corgenix, UK)) to selectively permeabilize the plasma membranes as described (8,9), allowing collection of diffusable cytoplasmic components with minimal physical perturbation or organelle breakage. Each 100-mm dish contained ϳ1 ϫ 10 6 cells. After treatment with SLO on ice for 4 min, unbound SLO was removed by washing the monolayers with ice-cold phosphate-buffered saline. 2 ml of modified transport buffer (containing 78 mM KCl, 4 mM MgCl 2 , and 50 mM Na-HEPES, pH 7.4, but lacking DTT as originally formulated, Ref. 8) prewarmed to 37°C were added. After 15 min at 37 degrees, the dishes were placed on ice for an additional 5 min. The buffer was collected and used for measurements of hexose-metabolizing enzymes. Protein was measured with a dye binding assay (Bio-Rad) with bovine serum albumin as a standard. Comparisons of extracts made this way with extracts made with 1% Triton X-100 showed that Triton X-100 recovered 5-6 times as much protein, but a similar amount of total GP (not shown). The various stress treatments described in this study did not influence the amount of total protein recovered in cytoplasmic extracts.
Assays for Hexose Phosphate-metabolizing Enzymes-Assays were linear over the incubation periods indicated. Except for PFK, all enzymes in 0.2 ml of cytoplasmic extract were assayed in 1 ml of buffer containing 50 mM K-HEPES, pH 7.2, 25 mM KCl, and 5 mM MgCl 2 , reactions were performed in the presence of 1.5 units of glucose-6-P dehydrogenase (Sigma, catalog number G8529) and 0.2 mM NADP ϩ (Sigma, catalog number N0505), and product was assessed spectrophotometrically at 340 nm for the reduction of NADP ϩ to NADPH, as described (10). Except for PFK, the incubation times at 37°C, and other reaction components, are indicated below for each enzyme. In all cases, assay values with blank reactions lacking the enzyme substrate were subtracted out. All data were normalized to protein content. Assay for Cytoplasmic Activators of Phosphofructokinase and Preparation of Alkaline Extracts-Fructose 2,6-bisphosphate (or similar activators) was measured by activation of PFK as described (12,13). Rabbit muscle PFK (Sigma cat. F2129) and PFK in extracts from control and stressed cells were tested. Activators were obtained by alkaline extraction as follows. 100-mm dishes were placed on ice and washed with ice-cold phosphate-buffered saline twice. 1 ml of 0.1 M NaOH (ice-cold) was added, and cells were scraped, transferred to a glass tube, and sonicated for 20 s. The mixture was heated at 80 degrees for 5 min. Samples were cooled on ice and centrifuged for 20 min at 14,900 rpm in a microcentrifuge. The supernatants were neutralized using 1 M HEPES (free acid) and added to PFK assays. In our hands, PFK was activated 2-3-fold by extracts from control cells.
Measurement of Glycogen Phosphorylase (GP)-Extracts prepared by SLO permeabilization (see above) were assayed for GP activity (14). A 0.2-ml extract was diluted into 1 ml of assay buffer containing 20 mM sodium phosphate (pH 7.2), 2 mM MgSO 4 , 1-2 mM NADP ϩ , 2 g/ml glucose-6-phosphate dehydrogenase, 3 units/ml rabbit muscle PGM, and 3 M glucose 1,6-bisphosphate. When indicated, 5 mM 5Ј-AMP was included. The reaction was initiated by adding 100 l of 10 mg/ml glycogen (Sigma cat. G1508). GP activity was determined spectrophotometrically at 340 nm over a period of 5 h at 37°C, and values for blank reactions lacking extract were routinely subtracted out. Over this period assays were linear for both time and amount of extract. The reaction was strictly dependent upon the presence of glycogen, PGM, and orthophosphate (not shown). Data presented for GP activity in the presence of AMP represent the total activity measured, without subtraction of activity obtained in the absence of AMP.
Measurement of Glucose 6-Phosphate by Enzyme Assay-Cellular alkaline extracts containing hexose phosphates (see above) were used. Since some hexose phosphates can be modified by alkali, key results were corroborated (not shown) with 70% ethanol extracts. Assays (1 ml) included 50 mM Tris-Cl (pH 7.5), 10 mM MgCl 2 , 0.25 mM NADP ϩ , and 0.2 ml of extract. 2 g/ml glucose-6-phosphate dehydrogenase (Sigma G7877) was added, and absorbance at 340 nm was measured after 15 min. Controls performed in the absence of glucose-6-phosphate dehydrogenase were subtracted out.
Measurement of Hexose Phosphates by Fluorophore-assisted Carbohydrate Electrophoresis (FACE)-Cell monolayers were extracted by scraping into 70% ethanol. The extract was clarified by centrifugation, dried under N 2 (g), dissolved in water, and applied to a 1-ml column of Dowex AG1-X2 (formate form). After washing with 20 ml of water, which removed essentially all neutral hexose, hexose phosphates were eluted with 15 ml of 4 M formic acid (15) and dried under N 2 (g). Recovery of a Glc-6-P standard was greater than 90% (not shown). However, hexose 1-phosphates were highly sensitive to hydrolysis, and under these conditions more than 95% of Glc-1-P standard was recovered as glucose. Thus, glucose appearing in the 4 M formic acid eluate represented Glc-1-P. Hexoses and hexose phosphates were modified with 2-aminoacridone (AMAC) and separated with a modified monosaccharide composition gel (16) composed of a 20% acrylamide separating FIG. 1. Metabolic steps examined as effectors of the UPR. In this study, these eight enzymes and factors involved in hexose-P metabolism that were candidates for the effects of the UPR on LLO extension were examined: PMM, phosphomannomutase; PMI, phosphomannose isomerase; HK, hexokinase; PGM, phosphoglucomutase; F-1,6-BPase, fructose-1,6-bisphosphatase; PFK, phosphofructokinase; Fru-2,6-bisP, fructose 2,6bisphosphate. The dotted line indicates the plasma membrane.
gel with 125 mM Tris borate, pH 8.3 and a 6% acrylamide stacking gel with 63 mM Tris borate, pH 6.8. The running buffer was 0.1 M Tris borate (pH 8.3). All other aspects of the electrophoresis were similar to those described for commercial monosaccharide composition gels (Glyko). AMAC-modified sugars were quantified with a BioScan fluorescence scanner as described (16).
Measurement of Glycogen-Glycogen was extracted from cell monolayers in 100-mm dishes, pre-chilled on ice, and washed with ice-cold PBS, by scraping with 1 ml of ice-cold 70% perchloric acid followed by sonication at 0 degrees (17). After removal of insoluble material by centrifugation, aliquots were used for determination of glucose with or without prior enzymatic degradation of glycogen. Glycogen degradation was achieved by combining 0.4 ml of extract with 0.2 ml of 1 M KHCO 3 and 2 ml of 1 mg/ml amyloglucosidase (Sigma A3514) solution, followed by shaking at 40°C for 2 h. The incubation was stopped by adding 1 ml of 70% perchloric acid, neutralized with 2-3 mg solid KHCO 3 to a pH of 6.9 -7.3, and centrifuged at 4000 rpm for 15 min. Glucose was assayed by mixing up to 0.1 ml of the supernatant with 1 ml of buffer containing 0.3 M triethanolamine chloride, pH 7.5, 1 mM ATP, 0.1 mM NADP ϩ , 4 mM MgSO 4 , and 5 g/ml glucose-6-phosphate dehydrogenase. After incubation at 37°C for 5 min, absorbance at 340 nm was determined. Hexokinase (Sigma H5625) was added to a final concentration of 1.4 units/ml. After an additional 5 min at 37°C, absorbance at 340 nm was again determined, from which the first reading was subtracted to determine the glucose-specific change in absorbance. After subtraction of values obtained without amyloglucosidase treatment and comparison with standard glucose samples, the quantity of glycogen-derived glucose was determined.

RESULTS
Many Activities Associated with Hexose Metabolism Are Unaffected by ER Stress-Compared with tissues such as liver, brain, and muscle, little is known about regulation of glycolysis, gluconeogenesis, or glycogenolysis in dermal fibroblasts. A number of regulatory points in hexose metabolism with the potential to modulate LLO mannosylation were therefore evaluated to determine whether they might be regulated by the UPR (Fig. 1). ER stress conditions shown previously to stimu-  late LLO extension were chosen. These conditions did not cause cytoplasmic stress, itself a potential regulator of hexose metabolism. 2 DTT was particularly useful because it acted within 20 min of addition to cells, and its effects on LLO extension were due to an ER stress response rather than a direct chemical reduction of a regulatory enzyme (4). However, for the current study two significant modifications were made. First, cells were maintained continuously in medium with 5 mM glucose. In prior studies, a 20-min period in 0.5 mM glucose medium was necessary to provide a pool of truncated LLOs. Second, TN could not be used in prior studies since it directly inhibits LLO synthesis. TN was used in the present study since any UPR effectors should be activated by it. Cytoplasmic extracts were prepared by gentle permeabilization of the plasma membrane with streptolysin-O (SLO) to minimize release of material from damaged intracellular organelles (8). Thus, our assays were limited to readily diffusable components. No changes were observed for seven key enzymes and co-factors (Table I). There were no apparent stress effects on enzymes (phosphomannose isomerase, phosphomannomutase) that might divert Fru-6-P, a key intermediate in glycolysis, to mannosyl phosphates. No changes were detected in the enzymes (phosphofructokinase, fructose-1,6-bisphosphatase) or cytoplasmic activators of phosphofructokinase 2 J. Shang and M. A. Lehrman, unpublished observations.

TABLE II
Effects of DTT-and TN-induced ER stress on 5Ј-AMP-dependent and -independent glycogen phosphorylase activities in primary dermal fibroblasts Glycogen phosphorylase was assayed without or with 5 mM 5Ј-AMP in cytoplasmic extracts from cells treated in the absence (control) or presence of 2 mM DTT for 15 min, or 5 g/ml TN for 60 min, and normalized to protein content, as described under "Experimental Procedures." Cultures with CRL designations were from the American Type Culture Collection. F-12 was a gift of Dr. H. Freeze, Burnham Institute, and F21-3 was obtained from National Psoriasis Foundation Tissue Bank. Adult donors were clinically normal. Pediatric donors were normal or had clinical abnormalities not expected to affect the parameters measured here. For reasons that are unclear, the average GP activities among the 5 adult cultures (ϪAMP, 0.072; ϩAMP; 0.21) were higher than those for four pediatric cultures (ϪAMP, 0.034; ϩAMP, 0.095). GP activity in only one of 9 cultures, pediatric CRL-1474, did not respond to ER stress. Since this culture also had the highest GP activity among the four pediatric cultures, there is a possibility of constitutive activation of GP.

Stoichiometry of hexose flux resulting from ER stress
Estimates are based upon 10 6 cells. Quantities of glycogen-derived Glc-1-P were calculated from the amount of glycogen consumed, as indicated in the text. Cytoplasmic hexose phosphates due to ER stress were determined by FACE and enzyme assay as described under "Experimental Procedures," with values from unstressed cells subtracted out. The total amount of glucose plus mannose in Glc 3 Man 9 GlcNAc 2 -P-P-dolichol detected by FACE in dermal fibroblasts grown in 5 mM glucose (12) was used to estimate the maximum amount of hexose that might be needed for complete extension of lipid-linked oligosaccharides due to ER stress. (such as Fru-2,6-bisP) that regulate interconversion of Fru-6-P and Fru-1,6-bisP, and thus possibly shift hexose units from glycolysis to glycoprotein synthesis. Consistent with the prior lack of evidence for enhanced uptake glucose and mannose from culture medium (6), no measurable changes were detected for hexokinase. Phosphoglucomutatase, inhibition of which could prevent storage of imported hexose as glycogen, was also unaffected. Considering these and earlier (6) results, it is clear that ER stress does not cause widespread changes in activities involved in hexose metabolism. Fig. 2, panel A, GP activity was detectable in fibroblasts in both its more active (AMP-independent) and less active (AMP-dependent) forms. Although detailed information about this enzyme in fibroblasts is lacking, by extension from tissues such as liver and muscle it is possible that phosphorylation converts the less active form, GPb, to the more active form, GPa (18). Caffeine, which blocks AMP binding to GPb, inhibited the AMPdependent activity but not the AMP-independent activity (Fig.  2, panel B).

AMP-independent Glycogen Phosphorylase Activity in Dermal Fibroblasts Is Increased by ER Stress-As shown in
When cells were treated with DTT, TN (Fig. 2, panel C), or AZC (not shown) at concentrations known to cause ER stress (4), AMP-independent GP activity in extracts increased, but AMP-dependent activity was unaffected. Activation of GP occurred within 20 min of DTT addition. Direct addition of DTT to assays, or its inclusion during extract preparation, did not affect GP activity (not shown). Thus, stimulation by DTT was not due to a direct chemical reduction of GP or a modulator of GP, as expected since the cytoplasmic environment in which GP exists is highly reducing. In addition to the fibroblast culture used in the experiments displayed in the figures, ER stress with DTT or TN increased GPa but not GPb activities in seven other adult and pediatric dermal fibroblast cultures (Table II). The stimulation of GP by ER stress did not exceed 2-fold, but the following sections provide evidence that this stimulation was significant.
Glycogenolysis Is Stimulated by ER Stress-Dermal fibroblasts contained, on average, 230 nmol of glycogen-derived glucose per 10 6 cells. In the presence of orthophosphate, GP converts glycogen into Glc-1-P. Small but consistent losses of glycogen were observed with DTT (23 Ϯ 2%, n ϭ 13) and TN (12 Ϯ 2%, n ϭ 3). For DTT, the decrease occurred after 15 min of addition. Since cellular glycogen loss may be counterbalanced by its synthesis, the apparently weaker effect of TN may be related to the fact that it acted more slowly (within 1 h). Thus, the resulting glycogenolysis would be more easily compensated by glycogen synthesis. These losses of glycogen, though small, were more than adequate to account for downstream effects on hexose phosphates and LLOs (Table III).
ER Stress Elevates Glucosyl Phosphates-Fluorophoreassisted carbohydrate electrophoresis (FACE) measurements of Glc-1-P showed an increase of 30% above control after 15 min of DTT treatment and 60% after 30 min (data not shown). The net effect appeared substantial (Table III). Glc-1-P can be condensed with UTP by UDP-Glc pyrophosphorylase, generating UDP-Glc plus pyrophosphate. Alternatively, Glc-1-P can be converted to Glc-6-P by phosphoglucomutase. Thus, it was likely that a portion of any Glc-1-P generated would have been converted to these other metabolites. As shown with an enzyme-linked assay (Fig. 3), DTT, TN, and CSN-induced ER stresses consistently increased cytoplasmic Glc-6-P (panel A). CSN, a weak UPR inducer (4), gave the weakest response. The effect of DTT was observed within 5 min of addition to cell cultures (panel B), and was maximal after 15 min. Concentrations as small as 0.2 mM were effective (panel C). These results were in accord with the previous demonstration that LLO extension was strongly stimulated by 0.4 mM DTT and occurred within 20 min of addition to cultures (4).
DTT treatment also resulted in elevation of Glc-6-P when measured by FACE (Fig. 4, panel A). This increase (ϳ5-fold, panel B) was greater than that measured by enzyme assay (ϳ2-fold, Fig. 3). Yet, the net increases were similar, in the range of 3 to 4.5 nmol per 10 6 cells (Table III). This suggests that an interfering component in the extract may have contributed to the background in the enzymatic assay for Glc-6-P.
Quantitative Assessment of Glycogenolysis as the Basis for LLO Extension-Taken together these results indicate that ER stress activates GP, associated with glycogenolysis and generation of Glc-1-P and Glc-6-P. As shown in Table III, ER stress generates at least 100-fold greater equivalents of hexose-P than needed to support synthesis of Glc 3 Man 9 GlcNAc 2 -P-Pdolichol in these cells as measured by FACE (16). Presumably, the bulk of the glycogen-derived hexose-P not used for LLO extension is available for glycolysis.
Phosphatase Inhibitors Mimic the Effects of ER Stress on GP Activity-In all systems previously examined, AMP-dependent GPb is converted to AMP-independent GPa by phosphorylation (18). Thus, a commercial mixture of phosphatase inhibitors was included during the preparation of extracts to optimize the determination of ER stress activation of GP. The results (Fig. 5) were unexpected. Phosphatase inhibitors had no effect on AMP-independent GP activity in extracts from stressed cells, but they increased the activity from unstressed cells to levels recovered from stressed cells. Thus, the inclusion of phosphatase inhibitors during extract preparation had the same effect on GP activity as treating intact cells with UPR inducers, but these were not additive. Presumably, the relevant phosphorylation event occurs not only in the cell, but also in the extract and/or under assay conditions. Otherwise, the inhibitor mixture should have been without effect. The implications of these data for regulation by phosphorylation are considered under the "Discussion" and in Fig. 6 below. DISCUSSION This study identified AMP-independent glycogen phosphorylase as a candidate for mediating the ER stress-dependent extension of LLO precursors and increase of Glc 3 Man 9 -FIG. 5. Phosphatase inhibitors mimic the effects of ER stress on GP activity. A proprietary aqueous mixture of phosphatase inhibitors (Sigma cat. P5726) including sodium orthovanadate, sodium molybdate, sodium tartrate, and imidazole was used. Cells were treated with DTT or TN as in Fig. 3, and cytoplasmic extracts were prepared with or without 1% (v/v) inhibitor mixture added to the transport buffer. AMP-independent GP activity was measured as units/min/mg protein (mean Ϯ S.E., n ϭ 6 for DTT and n ϭ 2 for TN).
FIG. 6. Model for stimulation of LLO extension by UPR regulation of glycogenolysis. Accumulation of LLO intermediates, such as by glucose depletion, results in ER stress. As shown in this study ER stress activates glycogen phosphorylase (GP). This, in turn, mobilizes hexose-P from glycogen, resulting in increased concentrations of precursor substrates needs for extension of LLO intermediates to mature LLOs. In this figure we speculate that the UPR may prevent dephosphorylation of GP to an AMP-dependent state. Other potential modes of regulation suggested by these data include production of a phosphatase inhibitor that protects GP from dephosphorylation, activation of a kinase that compensates for GP dephosphorylation, and/or phosphorylation of a separate regulator of GP. GlcNAc 2 -P-P-dolichol concentration. A number of trivial explanations for these changes had already been ruled out (Refs. 4 and 6; see the Introduction). Seven other candidate enzymes and factors were unchanged by ER stress. Experiments with phosphatase inhibitors were consistent with activation of AMP-independent GP activity by a phosphorylation-dependent mechanism. At present, it is not known whether ER stress controls one of the previously identified modulators of GP phosphorylation, such as phosphorylase kinase or protein phosphatase 1 or a novel phosphorylation-dependent modulator of GP. Although no change in GP activity was detected in assays in the presence of AMP, suggesting that the total amount of enzyme was unaffected, it remains possible that GP was also activated intracellularly by AMP or another cytoplasmic component that dissociated during analysis.
ER stress increased GP activity by no more than a factor of 2, and only small amounts of glycogen were consumed. However, activation of GP was significant. First, as summarized in Table III, the amount of hexose-P derived from glycogen is over 100 times the amount theoretically needed to produce substrates for LLO extension. Second, GP is activated, glycogenolysis is stimulated, and both Glc-1-P and Glc-6-P are produced by treatment times and concentrations of UPR inducers that are highly consistent with those that stimulate LLO extension (4). Third, both biochemical (6) and genetic (19) studies have shown that Glc-6-P is an important source of hexose for LLO synthesis. Fourth, in 3T3-L1 adipocytes, glycogen appears to act as a buffer against accumulation of LLO intermediates due to glucose deprivation (20). Formal proof that GP is the UPR effector responsible for extension of LLOs will require techniques that permit specific inhibition of GP activation.
The effects of GP activation by ER stress on the pathway leading from glycogen to UDP-Glc, which is the donor substrate for glucose-P-dolichol synthase, remain to be examined. However, two observations suggest that this is also enhanced. In cells that accumulate more Man 2-5 GlcNAc 2 -P-P-dolichol than Glc 3 Man 9 GlcNAc 2 -P-P-dolichol due to glucose deprivation, ER stress increased the radiochemical amount of [ 3 H] Glc 3 Man 9 GlcNAc 2 -P-P-dolichol in cells when [ 3 H]mannose was used to label mannosyl residues (6). Yet, the same stress treatments eliminated the appearance of [ 3 H]Glc 3 Man 9 -GlcNAc 2 -P-P-dolichol when [ 3 H]galactose was used to label glucosyl residues (data not shown). The latter process requires intracellular conversion of galactose to Gal-1-P, UDP-Gal, UDP-Glc, and glucose-P-dolichol. Since UDP-Glc is also formed from Glc-1-P resulting from glycogenolysis, these results could be explained if ER stress elevated UDP-Glc concentrations and diluted the UDP-[ 3 H]glucose formed from [ 3 H]galactose. In agreement with this idea, ER stress enhanced the glucosylation of LLOs in Congenital Disorder of Glycosylation Type-Ic cells, in which glucosylation of LLOs is deficient due to a mutation that reduces the activity of glucose-P-dolichol:Man 9 GlcNAc 2 -P-P-dolichol glucosyltransferase (4). Thus, elevation of UDP-Glc might stimulate glucose-P-dolichol production and help compensate for the enzymatic defect.
The rapid activation of GP, combined with the evidence for a role of phosphorylation, strongly suggests regulation by post-translational modification rather than transcription. Among the UPR signaling pathways that have been identified, this is most reminiscent of eIF2␣ phosphorylation by the ER stress sensing kinase PERK. In fibroblasts, both GP activation (this study) and eIF2␣ phosphorylation (4) occur within 15-20 min of application of ER stress. It will therefore be interesting to determine whether GP can be activated in a PERK-deficient cell, such as PERK (Ϫ/Ϫ) embryonic fibroblasts (21). Unlike transcriptional mechanisms, some post-translational modifications can be easily reversed. GADD34 has been identified as promoting dephosphorylation of eIF2␣ (22), so a similar mechanism may be responsible for deactivating GP. If ER stress generates a phosphatase inhibitor (Fig. 6), the UPR may control glycogenolysis by modulating cycling between phosphorylated and dephosphorylated GP.
Given these results and the information we have recently reported on the high priority of LLO extension in the UPR (4), the UPR appears to proceed in two phases. The first phase, which takes place within minutes of the generation of ER stress, may be responsible for ER damage control. This phase is characterized by temporary adjustments that can quickly counteract the causes of ER stress, such as by increasing the supply of hexose-P for LLO synthesis and energy production, and by decreasing the amount of protein entering the ER. However, long-term solutions would occur hours later during the second phase, characterized by remodeling of the ER in a manner requiring gene transcription.