A Cellular UDP-glucose Deficiency Causes Overexpression of Glucose/Oxygen-regulated Proteins Independent of the Endoplasmic Reticulum Stress Elements*

A low level of UDP-Glc occurs in cells exposed to hypoxia or glucose starvation. This work reveals that a 65% reduction in the cellular UDP-Glc level causes up-regulation of the mitochondrial chaperone GRP75 and the endoplasmic reticulum (ER) resident chaperones GRP58, ERp72, GRP78, GRP94, GRP170, and calreticulin. Conditions that cause misfolding of proteins within the ER activate the transcription factors ATF6α/β and induce translation of the transcription factors XBP-1/TREB5 and ATF4/CREB2. These transcription factors induce the overexpression of ER chaperones and CHOP/GADD153. However, the 65% decrease in the cellular UDP-Glc level does not cause activation of ATF6α, splicing of XBP-1/TREB5, induction of ATF4/CREB2, or expression of CHOP/GADD153. The activity of the promoters of the ER chaperones is increased in UDP-Glcdeficient cells, but the activity of the CHOP/GADD153 promoter is not affected, in comparison with their respective activities in cells having compensated for the UDP-Glc deficiency. The results demonstrate that the unfolded protein response remains functionally intact in cells with a 65% decrease in the cellular UDP-Glc level and provide evidence that this decrease is a stress signal in mammalian cells, which triggers the coordinate overexpression of mitochondrial and ER chaperones, independently of the ER stress elements.

The cellular response to stressful conditions includes metabolic adjustments and changes in gene expression. In mammalian cells the proteins most commonly overproduced in response to stress are molecular chaperones, which during normal cellular growth play essential roles in folding, oligomerization, and membrane translocation of newly synthesized polypeptides (1,2). During stress, chaperones bind to misfolded polypeptides hence avoiding their aggregation and facilitating their repair (1,2). Chaperones induced by heat shock, heavy metals, or exposure to amino acid analogues are termed heat shock proteins (HSPs) 1 (1). Another set of chaperones, overproduced upon exposure to glucose starvation or hypoxia, are termed glucose/oxygen-regulated proteins (GRPs/ORPs) (2,3). Several of the GRPs/ORPs are residents of the endoplasmic reticulum (ER), and their overproduction is induced through a signal transduction pathway referred to as the unfolded protein response (UPR) (4,5). This pathway is activated by exposure to conditions that cause accumulation of misfolded proteins within the ER, such as treatment with glycosylation inhibitors or substances that perturb the oxidative milieu or calcium homeostasis in the ER (4,5). The promoter region of the GRP genes possesses a cis-regulatory element, known as the ERSE (ER stress response element) (6), with the consensus sequence CCAATN 9 CCACG. This is recognized by the general transcription factor NF-Y/CBF on its CCAAT part and by the basic leucine zipper-type transcription factors ATF6␣/␤ and XBP-1/TREB5 on its CCACG part (6 -8). ATF6␣/␤ are expressed constitutively as glycoproteins anchored in the ER membrane (9,10). Upon ER stress they undergo proteolysis, which releases their N-terminal fragments (p50ATF6␣ and p60ATF6␤) that enter into the nucleus, dimerize, and bind to the promoters of the ER-resident GRP/ORP genes inducing their transcription (9,10). Proteolysis of ATF6 also induces transcription of the genes encoding the transcription factors XBP-1/TREB5 and CHOP/GADD153, the promoters of which contain ERSE sequences (7,11). Another response to the accumulation of misfolded proteins within the ER lumen is the dimerization and autophosphorylation of the ER transmembrane protein kinases IRE1␣/␤, which leads to activation of their endoribonuclease function (12,13). The endoribonuclease activity of IRE1␣/␤ is required to splice XBP-1/TREB5 mRNA resulting in the production of a potent transcriptional activator that enhances further the activity of its own promoter as well as those of the ER-resident GRPs/ORPs and CHOP/GADD153 (11,14,15). Besides binding ERSE, the product encoded by the spliced XBP-1/TREB5 mRNA broadens the spectrum of genes induced by ER stress, as it also binds with high affinity to another regulatory site known as the unfolded protein response element (UPRE) with consensus sequence TGACGTG(G/A) (11,(15)(16)(17). The activation of Ire1␣/␤ and the splicing of the XBP-1/TREB5 mRNA are absolutely required for signaling through this transcription factor because overexpression of the product encoded by the unspliced XBP-1 mRNA cannot activate UPRE (15). Accumulation of misfolded proteins within the ER lumen can also lead to a prompt activation of the ER transmembrane protein kinase PERK, which phosphorylates the ␣-subunit of eukaryotic translation initiation factor 2 (18). This results in a transient and general translational attenuation, which, however, is accompanied by an increased synthesis of ATF4/CREB2 (5,19,20). This transcription factor directly contributes to the transcriptional induction of the GRP78 and CHOP/GADD153 genes by binding to their promoters in a site different from ERSEs and UPRE (21,22).
UDP-Glc is a required precursor in the synthesis of the carbohydrate moiety of N-glycoproteins. Furthermore, it is required in the quality control of newly synthesized glycoproteins within the ER (23). A low level of UDP-Glc occurs in mammalian cells exposed to hypoxia or glucose starvation (for references see Ref. 23). In Escherichia coli a low UDP-Glc level seems to be a signal that induces the expression of a set of stress proteins required for survival under adverse conditions (24). Similarly, in plant cells a low UDP-Glc level due to hypoxia or sugar starvation induced the synthesis of at least one stress protein (25). However, whether a low UDP-Glc level is a stress signal in mammalian cells has not yet been determined.
We previously isolated a Chinese hamster mutant cell line, Don Q, having a permanent low level of UDP-Glc because of a recessive point mutation affecting the UDP-Glc pyrophosphorylase gene (UDPG:PP) (23). From Don Q we isolated a spontaneous revertant cell, Don QR, in which the mutation was reverted in one allele, partially compensating for the UDP-Glc deficiency (23). These cells provide a unique opportunity to investigate how mammalian cells are affected by fluctuations in the cellular UDP-Glc content. In this work we used them as a model system to determine whether a decrease in the cellular UDP-Glc level induces up-regulation of stress proteins.
Two-dimensional Polyacrylamide Gel Electrophoresis, N-terminal Sequencing, and Western Blot Analyses-Cells were cultivated at 37°C in Eagle's minimum essential medium supplemented with 10% fetal bovine serum, 5 mM L-glutamine, penicillin (100 units/ml), and streptomycin (100 g/ml) in a humid atmosphere containing 5% CO 2 . Proteins in cell lysates were analyzed by two-dimensional gel electrophoresis (isoelectric focusing using pH 3-10 Pharmalytes® followed by SDS-PAGE in 7.5-20% (w/v) gradient gels) and visualized by staining with Coomassie Brilliant Blue. Each cell line was analyzed at least seven times. N-terminal sequence analysis of proteins separated on twodimensional gels (28,29) was performed using Applied Biosystems 470A or 477A instruments. Homology searches in SwissProt were done using BLAST.
For induction of the UPR, cells were treated with 7 g/ml tunicamycin for 3-24 h before harvesting. Cell lysates (30 g of protein/lane) were subjected to electrophoresis under reducing conditions on a 10% (w/v) SDS-polyacrylamide gel, electroblotted onto 0.45-m nitrocellulose membranes, and subjected to Western blot analyses as described (23). Immunoreactive bands were detected and quantified with a densitometer using ImageQuant (Amersham Biosciences). Each immunoblot analysis was repeated 2-4 times with similar results.
1 H NMR Spectroscopy-Cell metabolites were extracted from confluent monolayers of 75-cm 2 culture flasks of each cell line using a dualphase extraction method (20). The methanol/water phase was freezedried and resuspended in a buffer with 8 mM Na 2 HPO 4 and 2 mM KH 2 PO 4 , pH 7.2 (prepared with D 2 O). Spectra were acquired, and adenine nucleotides and UDP-Glc were quantified as described (23,30). The measurements were performed at least three times for each cell line.
Plasmids and Luciferase Assays-A 311-bp fragment of the human GRP78 promoter (Ϫ304 to ϩ7 region; numbers indicate nucleotide positions relative to the transcription start site), a 397-bp fragment of the human GRP94 promoter (Ϫ363 to ϩ34 region), a 1099-bp fragment of the human GRP58 promoter (Ϫ1081 to ϩ18 region), a 511-bp fragment of the human calreticulin promoter (Ϫ459 to ϩ52 region), a 668-bp fragment of the murine ERp72 promoter (Ϫ647 to ϩ21 region), a 887-bp fragment of the human CHOP promoter (Ϫ870 to ϩ17 region), and a 459-bp fragment of the XBP-1 promoter (Ϫ330 to ϩ129 region) were amplified by polymerase chain reaction and cloned immediately upstream of the luciferase coding sequence of the pGL3-basic vector (Promega), as described (6,7). Mutants of the GRP78, GRP94, and calreticulin promoters with disrupted ERSE sequences were prepared by site-directed mutagenesis as described (6). ERSE1, ERSE2, and ERSE3 of the GRP78 promoter, ERSE1 and ARSE3 of the GRP94 promoter, and ERSE2 and ERSE3 of the calreticulin promoter were disrupted by mutating their sequences to gatcTN 9 aacat, CtcgaN 9 aacac, gagcTN 9 aacgc, atgttN 9 Agctc, gatcTN 9 aacat, agctcN 9 aactc, and atgttN 9 Agatc, respectively (where lowercase letters indicate mutated nucleotides). Mutated promoters were cloned in the pGL3-basic vector (Promega), as described (6). A reporter plasmid containing a 1288-bp fragment of the human ORP150 gene (Ϫ332 to ϩ956 region) cloned in the pGL3-basic vector (31) was kindly provided by Dr. H. Yanagi (HSP Research Institute, Kyoto, Japan). The reporter plasmid p5XUPREGL3 containing five repeats of the oligonucleotide CTCGAGACAGGTGCTGACGTG-GCATTC cloned into pOFLuc-GL3 in front of the c-fos minimal promoter and the firefly luciferase gene (17) was kindly provided by Dr. R. Prywes (Columbia University, New York). The reporter plasmids Ϫ457/ LUC and Ϫ457(mut)/LUC containing the rat GRP78 wild type promoter and the modified version with the mutated ATF/CRE site, respectively (22), were kindly provided by Prof. A. Lee (University of Southern California, Los Angeles, CA). The expression plasmid encoding a dominant negative mutant of the A subunit of the transcription factor NF-Y (32) was kindly provided by Dr. M. Pitarque (Karolinska Institute). The expression plasmids pCGNATF6-(1-373), encoding the wild type cytoplasmic region of the human ATF6␣, and pCGNATF6-(1-373)m1, encoding a mutant version of this protein harboring a replacement of amino acids 315-317 from KNR to TAA and lacking DNA binding activity (17), were kindly provided by Dr. R. Prywes. The reference plasmid pRL-SV40, which carries the Renilla luciferase gene under the control of the SV40 enhancer and promoter, was from Promega.
Cells cultured in 48-well plates were cotransfected with reporter plasmids (1 g) carrying the firefly luciferase gene, the reference plasmid pRL-SV40 (0.05 g), and without or with an expression plasmid (3 g) using LipofectAMINE 2000. Cells were incubated with 100 ml of the lipid-DNA complex in Opti-MEM at 37°C for 5 h and then incubated in supplemented Eagle's minimum essential medium at 37°C for 24 h. Cells were then lysed in 200 l of passive lysis buffer (Promega) and firefly, as well as Renilla, luciferase activity was measured with 5 l of the cell lysate by using the dual luciferase reporter assay system (Promega) and a TD2020 luminometer (Turner Designs, Sunnyvale, CA).

A Cellular UDP-Glc Deficiency Is Associated with Overproduction of a Specific Set of Stress
Proteins-To determine whether the mutant Don Q has any specific changes in the cellular protein pattern, whole cell lysates of Don wt, the mutant Don Q, and the revertant QR were analyzed by twodimensional gel electrophoresis (Fig. 1). Several protein spots were consistently detected in higher amounts in Don Q than in Don wt and QR cell lysates. The molecular masses of the four most conspicuously overproduced proteins were ϳ59, 62, 70, and 74 kDa, and their isoelectric points were 7, 4.5, 6, and 5, respectively. These four overproduced proteins were identified by their N-terminal amino acid sequences as GRP58/ERp61/ ERp57 (34), calreticulin (35), GRP75/PBP74/mortalin (36), and GRP78/BiP/ORP80 (37), respectively (Table I). Western blot experiments further confirmed the identity of the overproduced proteins and showed that they occur in about 3 times higher relative amounts in Don Q than in Don wt and QR cells (Fig. 2). Because three of these proteins belong to the ER-resident stress-inducible GRPs/ORPs, additional Western blot analyses were performed with antibodies against other members of this protein family: ERp72/CaBP2 (34), GRP94/ERp99/endoplasmin (3), and GRP170/ORP150 (31). The relative amounts of these proteins were also about 3-4 times higher in Don Q than in the Don wt and QR cells (Fig. 2).
However, other stress-inducible proteins, such as the cytosolic HSP70 and HSP25 and the mitochondrial HSP60 were found in comparable amounts in the three cell lines (Fig. 2). Furthermore, the protein-disulfide isomerase, the UDP-Glc: glycoprotein glucosyltransferase, the peptidyl-prolyl cis-trans isomerase CPH2, and the antioxidant enzyme heme oxygenase 1/HSP32, which are overproduced when there is an accumulation of misfolded proteins in the ER (2), were also found in similar amounts in Don wt, Q, and QR according to Western blot analysis (Fig. 2). Thus, in Don Q only a specific set of stress proteins is induced, which includes a mitochondrial chaperone and some of the proteins known to be induced by accumulation of misfolded proteins within the ER.
A Cellular UDP-Glc Deficiency Induces Chaperones from the Mitochondria and the ER but Does Not Affect CHOP Expression-Don Q cells were transfected with bovine UDPG:PP cDNA to compensate for the missing activity of the UDP-Glcproducing enzyme, and several clones were isolated of which B9 and G3 are representatives (26). 1 H NMR measurements showed that the UDP-Glc concentration in these transfectant clones was comparable with that found in Don wt or QR cells, i.e. about 2-3 times higher than in Don Q and a control clone of Don Q (QC) transfected with the vector only (Table II). Despite these differences in the UDP-Glc level, the concentrations of ATP, ADP, and AMP, respectively, were equal in all the transfected clones as well as in the Don wt, Q, and QR cells (Table II).  The relative amounts of calreticulin and the six GRPs/ORPs in the B9 and G3 transfectants were from 2.3 to 3.4 times lower than in the UDP-Glc-deficient control QC, as compared by Western blot analysis (not shown). Furthermore, reporter assays showed that the promoters of ORP150, GRP94, GRP78, ERp72, GRP58, and calreticulin are from 2.6 to 4.1 times more active in QC than in B9 or G3 cells (Fig. 3), demonstrating that the overproduction of these chaperones is because of an increased gene transcription. Because the UDP-Glc level in the B9 clone is only 70% of that in the wild type cell (Table II), this result also indicates that a threshold of UDP-Glc decrease is necessary for the induction of the stress proteins.
The expression of CHOP/GADD153 and the ER chaperones is coordinately regulated during UPR activation (2). Interestingly, CHOP/GADD153 was not detectable under normal conditions in Q/QC cells (Fig. 4, lane 7). Accordingly, the CHOP promoter showed the same activity in QC, B9, and G3 cells (Fig. 3). However, it could be induced after exposure to the glycosylation inhibitor tunicamycin, as detected by Western blot (Fig. 4, lanes 8 and 9). The results therefore suggested that the cellular UDP-Glc deficiency causes the overproduction of calreticulin and the GRPs/ORPs by a pathway distinct from that triggered by tunicamycin.
A Cellular UDP-Glc Deficiency Does Not Cause ATF6␣ Activation, XBP-1/TREB5 Splicing, or ATF4/CREB2 Induction-Under normal growth conditions Don wt, QR, or G3 cells produce GRP78 at a basal level, express p90ATF6 but not p50ATF6, and do not express CHOP/GADD153 (Fig. 4, lanes 1,  6, and 10). Exposure of these cells to tunicamycin triggers the UPR as evidenced by the conversion of p90ATF6 to p50ATF6, the 3-fold increase in the amount of GRP78 after 24 h, and the induction of CHOP/GADD153 (Fig. 4, lanes 2-5, 11, and 12). Remarkably, p50ATF6 was not detectable under normal conditions in Don Q/QC (Fig. 4, lane 7), suggesting that the upregulation of ER-resident chaperones in these cells is independent of ATF6 activation. Accordingly, the activity of the GRP78 promoter in UDP-Glc-deficient cells was not reduced by cotransfection with dominant negative mutants of ATF6␣ or NF-Y/CBF, which reduce promoter activation upon tunicamycin exposure ( Fig. 5A and not shown). Furthermore, the activity of the GRP78 promoter increased by co-transfection with the cytoplasmic domain of ATF6 similarly in QC and G3 cells (Fig. 5,  A and B). In addition, exposure of Don Q/QC cells to tunicamycin was found to induce the conversion of p90ATF6 to p50ATF6 and to cause a 3-fold increment in the amount of endogenous GRP78 after 24 h (Fig. 4, lanes 8 and 9).
Because XBP-1/TREB5 is induced in UDP-Glc-deficient cells (Fig. 3), the activity of this transcription factor was evaluated in reporter assays using the plasmid 5XUPREGL3. Interestingly, the transcriptional activity of XBP-1/TREB5 is not higher in UDP-Glc-deficient cells than in cells with a normalized UDP-Glc level (Fig. 6A). Accordingly, the spliced form of   3. A cellular UDP-Glc deficiency activates the promoters of the GRPs/ORPs, CRT and XBP-1, without affecting the activity of the CHOP promoter. Wt, QC, G3, or B9 cells (hatched, black, gray, and white bars, respectively) were transiently co-transfected with the plasmid pSV40R and a reporter plasmid having the wild type promoters of ORP150, GRP94, GRP78, ERp72, GRP58, CRT, CHOP, or XBP-1 in the vector pGL3-basic, as described under "Experimental Procedures." Firefly and Renilla luciferase activity in cell lysates was measured 24 h later using the dual luciferase assay system. Fold expression was calculated by dividing the average value of relative luciferase activity in lysates of each cell line with the relative luciferase activity in B9.
XBP-1/TREB5 is not present in UDP-Glc-deficient cells, although it can be induced upon tunicamycin exposure (Fig. 6B). Furthermore, the UDP-Glc deficiency does not affect the amount of ATF4/CREB2, the production of which was also induced upon tunicamycin exposure regardless of the cellular UDP-Glc level (Fig. 6C). Taken together the results indicate that the classic UPR is not activated in QC cells and remains functionally active despite the UDP-Glc deficiency.

The Induction of ER Chaperones by a UDP-Glc Deficiency Is
Not Mediated by ERSEs-To evaluate the role of ERSEs in the induction of the ER chaperones GRP78, GRP94, and calreticulin by the UDP-Glc deficiency, we compared the activity of reporter plasmids in which the ERSEs were disrupted in cells with different levels of UDP-Glc (Fig. 7A). The results showed that despite the disruption of the ERSEs the promoter activity of the genes encoding these stress proteins is higher in UDP-Glc-deficient cells.
We also evaluated the role of the ATF/CRE binding site located upstream of the ERSEs of the rat GRP78 promoter in the induction of this chaperone by UDP-Glc deficiency. The activity in the QC cells of a reporter plasmid with the wild type promoter was compared with that of a modified version of this promoter in which that site was mutated (Fig. 7B). The results showed that disruption of the ATF/CRE site in the GRP78 promoter did not affect its activity in the UDP-Glc-deficient cells. DISCUSSION This work revealed that a permanent 65% reduction in the cellular UDP-Glc level is a triggering factor that induces the overexpression of several stress-inducible chaperones, one of the mitochondria (GRP75) and six of the ER (GRP170, GRP94, GRP78, ERp72, GRP58, and calreticulin). GRP75 is required for the import and folding of mitochondrial proteins, but little FIG. 4. Exposure to tunicamycin induces ATF6␣ activation and CHOP expression and enhances GRP78 overexpression in cells with different levels of UDP-Glc. Cells were cultured in the presence or absence of tunicamycin (TM) (7 g/ml) for the indicated periods of time. Lysates were prepared and submitted to Western blot analysis using antibodies against CHOP, ATF6␣, or GRP78 as described under "Experimental Procedures." p90ATF6␣ and p50ATF6␣ are indicated by the open and closed arrowheads, respectively. The band marked as # denotes the unglycosylated form of p90ATF6␣, whereas the identity of the band marked by * is currently unknown.

FIG. 5. The activity of the GRP78 promoter in UDP-Glc-deficient cells is not affected by co-transfection with a dominant negative mutant of ATF6␣ and increases by co-transfection with the cytoplasmic domain of ATF6␣, as in cells with a normalized UDP-Glc level. QC cells (A) or G3 cells (B)
were transiently co-transfected with a reporter plasmid containing the firefly luciferase gene fused with the promoter of the GRP78, the plasmid pSV40R and the plasmid pCGN (vector), a dominant negative mutant of ATF6␣ cloned in pCGN (ATF6 M1), or the wild type cytoplasmic domain of ATF6␣ cloned in pCGN (ATF6 wt) and maintained in the absence or presence of tunicamycin (TM) (7 g/ml, 20 h). Firefly and Renilla luciferase activities in cell lysates were measured 24 h later using the dual luciferase assay system. The firefly luciferase activity is presented relative to the Renilla luciferase activity in the corresponding sample.

FIG. 6. A cellular UDP-Glc deficiency does not cause XBP-1 activation or ATF4/CREB2 induction.
A, QC, B9, and G3 cells were transiently co-transfected with the plasmid pSV40R and the 5XUPRE site luciferase reporter plasmid (p5XUPREGL3) as described under "Experimental Procedures." Firefly and Renilla luciferase activities in cell lysates were measured 24 h later using the dual luciferase assay system. The firefly luciferase activity is presented relative to the Renilla luciferase activity. B, QC or G3 cells were grown in the presence or absence of tunicamycin (TM) (10 g/ml, 20 h), and reverse transcriptase-PCR analysis of XBP-1 was performed as described under "Experimental Procedures." C, cell lysates of QC and G3 cells grown in the presence or absence of tunicamycin (10 g/ml, 20 h) were submitted to Western blot analysis using antibodies against ATF4/CREB2 as described under "Experimental Procedures." is known about the control of its expression (36). GRP78 and GRP170 are part of the ER machinery that imports proteins of the secretory pathway into the ER and together with GRP94 assist in the folding and assembly of secretory and membrane proteins (38). ERp72 and GRP58 are members of the thioredoxin superfamily that participate in disulfide isomerization during protein folding (34), whereas calreticulin prevents unfolded glycoproteins from leaving the ER and facilitates their interaction with other chaperones (35). Furthermore, calreticulin and GRP78 account for most of the Ca 2ϩ buffering capacity of the ER (39,40). These conserved proteins are ubiquitously expressed at a basal level but overproduced in cells of ischemic tissues, as well as in cultured cells exposed to hypoxia or glucose starvation (41,42), two conditions known to induce a reduction in the cellular UDP-Glc level (23).
UDP-Glc Deficiency and UPR Signaling-Exposure of cells to ER stressors, such as tunicamycin, induces the UPR by activation of several basic region leucine zipper transcription factors (5). Thus, a tight linkage exists among ER stress, the activation of ATF6␣/␤, the splicing of XBP-1/TREB5, the induc-tion of ATF4/CREB2, and the expression of ER chaperones, co-chaperones, and other ER-associated proteins (5,21,22). Triggering the UPR in mammalian cells induces the up-regulation of several ER-resident folding enzymes including the protein-disulfide isomerase, peptidyl-prolyl cis-trans isomerases, and the UDP-Glc:glycoprotein glucosyltransferase (2,19). However, we found in this work that these folding enzymes are not up-regulated upon a partial deficiency of UDP-Glc. Moreover, the results showed that the increased transcription of ER chaperone genes in response to a permanent UDP-Glc deficiency is independent of ERSEs (Fig. 7A), and the chaperone overexpression is not dependent on ATF6␣ activation, XBP-1/ TREB5 splicing, or ATF4/CREB2 induction (Figs. 4 and 6). Nevertheless, in UDP-Glc-deficient cells the UPR remains functional because exposure to tunicamycin leads to ATF6␣ activation, XBP-1 splicing, ATF4/CREB2 induction, a further increase in the GRP78 level, and the induction of CHOP/ GADD153 (Fig. 4).
It has been recently demonstrated that an ATF/CRE binding site located upstream of the ERSEs in the GRP78 promoter regulates its induction upon exposure to thapsigargin (22). The GRP78 promoter activation via this site is independent of ATF6 processing and does not require the ERSE (22). However, we found no difference in the activity of the wild type GRP78 promoter and a modified version of this promoter in which the ATF/CRE site is mutated (Fig. 7B), excluding its involvement in response to a low UDP-Glc level. Thus, the actual promoter element(s) and transcription factor(s) activated by UDP-Glc deficiency have yet to be defined. It has been recently concluded that additional cis-acting elements different from UPRE and ERSE in the promoters of GRP78 and other ER-resident chaperones control their induction in mouse embryo fibroblasts lacking XBP-1/TREF, ATF6␣, and ATF6␤ (43). We are currently investigating the possibility that various basic region leucine zipper transcription factors like Zhangfei, which was recently shown to interact with ATF6 and ATF4 among others (44), could control the overexpression of the ER chaperones in UDP-Glc-deficient cells.
UDP-Glc Deficiency and Cell Survival-The classic UPR induced upon acute ER stress also involves expression of the non-ER-localized transcription factor CHOP/GADD153, which has been implicated in growth arrest and cell death (45). Moreover, cellular treatments that cause accumulation of misfolded proteins within the ER trigger activation of the c-Jun N-terminal kinase pathway and the activation of caspases 3, 7, and 12, inducing an apoptotic response (46 -48). Thus, if the ER chaperone up-regulation is not sufficient to relieve ER stress the cells will ultimately die. The experimental system used in the present study differs from those of most UPR studies in one essential aspect: generally the treatments triggering the UPR are acute and lead to cell death whereas in the model cell line used in this study, the stress, i.e. the UDP-Glc deficiency, is permanent (49). These cells survive and thrive, implying that they have adapted to coping with this cellular stress.
The survival of the UDP-Glc-deficient cells could be explained in part because CHOP/GADD153 is not induced. Furthermore, we have considered the possibility that this metabolic deficiency activates some survival signal(s) to balance their stress condition. Indeed, it has been found that NFB is activated in the Q cell line and that this activation is needed for its survival. 2 In addition, the chaperones overproduced at chronic UDP-Glc deficiency may have some cytoprotective effect in themselves. Cytoprotective properties have been previ- FIG. 7. The ERSEs are not required for the activation of the GRP94, GRP78, or CRT promoters, neither is the ATF/CRE site required for the activation of the GRP78 promoter, induced by the cellular UDP-Glc deficiency. A, QC, G3, or B9 cells (black, gray, and white bars, respectively) were transiently co-transfected with the plasmid pSV40R and a reporter plasmid containing the firefly luciferase gene fused with mutated versions of the promoters of the GRP94, GRP78, or CRT in which the ERSEs were disrupted, as described under "Experimental Procedures." Firefly and Renilla luciferase activities in cell lysates were measured 24 h later using the dual luciferase assay system. Fold expression was calculated by dividing the average value of relative luciferase activity in lysates of each cell line with the relative luciferase activity in B9. B, QC cells were transiently co-transfected with the plasmid pSV40R and a reporter plasmid containing the luciferase gene fused with the wild type rat GRP78 promoter or a mutated version in which the ATF/CRE site was disrupted, as described under "Experimental Procedures." Firefly and Renilla luciferase activities in cell lysates were measured 24 h later using the dual luciferase assay system. The firefly luciferase activity is presented relative to the Renilla luciferase activity in the corresponding sample. ously ascribed to GRP78 and GRP170, because they were shown to be required for cell survival under hypoxia (50,51). GRP78 and calreticulin were also found to protect cells from Ca 2ϩ overload (52)(53)(54)(55), and recently it was reported that GRP78 prevents cell death by interacting directly with caspases 7 and 12 hence inhibiting their activation (56).
A UDP-Glc Deficiency as a Stress Signal-There are at least two different ways in which a UDP-Glc deficiency could induce the overexpression of stress proteins: it could cause protein misfolding within the ER, hence triggering the classical UPR, or alternatively, UDP-Glc may have a role as a signaling molecule such that its loss would lead to increased transcription of a selective set of chaperones. We favor the latter alternative because several of the hallmarks of the classic UPR signaling pathway are absent in the UDP-Glc-deficient cells. Considering that UDP-Glc is actively transported to the lumen of the ER (57) and that its amount in the model cell used in this study is 35% of normal, this could suffice to allow a proper quality control of N-linked glycoproteins. Accordingly, it has been shown that reducing the cellular level of UDP-GlcNAc to 5% of the normal level did not affect the synthesis or secretion of N-linked glycoproteins (58). Moreover, a G protein-coupled specific cell surface receptor for UDP-Glc, P2Y 14 , was recently identified (59,60), and the release of small amounts of UDP-Glc from a variety of cultured cells into the medium was demonstrated (61). UDP-Glc was postulated as a novel autocrine and paracrine signaling molecule, although the physiological processes regulated by its widely distributed receptor remain to be clarified. Whether such a UDP-Glc receptor or a similar one could exist and function intracellularly is not known. Anyway, a lack of cytosolic (or extracellular) activating signals mediated via a UDP-Glc receptor might explain the particular stress response observed in UDP-Glc-deficient cells. Further studies are warranted to clarify the pathway by which a low UDP-Glc level triggers a stress response in mammalian cells.
UDP-Glc-deficient Cells as Model for Tumor Cell Adaptations-Solid tumors contain regions of low glucose concentration and low oxygen tension because of their abnormal vasculature (62), and it is likely that a UDP-Glc deficiency occurs in these regions (23). The UDP-Glc-deficient cell line used in this study has features resembling those of tumor cells from the ischemic regions. Mitochondrial and ER chaperones are often up-regulated in solid tumors both in experimental models and clinical samples (63,64). As in UDP-Glc-deficient cells, NFB is activated in many tumor cells, and the NFB activation is believed to have an important survival function (65)(66)(67). The overproduction of calreticulin and the GRPs/ORPs correlates with the resistance of tumor cells to radiation, tumor necrosis factor, cytotoxic lymphocytes, several chemotherapeutic drugs, and photodynamic therapy (68 -71). The overproduction of GRP78 and GRP94 is also known to increase the tumorogenic potential of malignant cells and to favor tumor progression in experimental models (72,73). Thus, the UDP-Glc-deficient cells used in this study might serve as a unique model system for characterizing the adaptive processes that tumor cells undergo at ischemic conditions. Understanding the signals and pathways by which tumor cells overproduce stress proteins, at conditions that do not trigger cell death, is of more than basic cell biological interest because it might help to identify molecular targets for rational therapeutic intervention strategies in malignant diseases.