Regulation of ERGIC-53 Gene Transcription in Response to Endoplasmic Reticulum Stress*

Accumulation of unfolded proteins within the endoplasmic reticulum (ER) activates the unfolded protein response, also known as the ER stress response. We previously demonstrated that ER stress induces transcription of the ER Golgi intermediate compartment protein ERGIC-53. To investigate the molecular events that regulate unfolded protein response-mediated induction of the gene, we have analyzed the transcriptional regulation of ERGIC-53. We found that the ERGIC-53 promoter contains a single cis-acting element that mediates induction of the gene by thapsigargin and other ER stress-causing agents. This ER stress response element proved to retain a novel structure and to be highly conserved in mammalian ERGIC-53 genes. The ER stress response element identified contains a 5′-end CCAAT sequence that constitutively binds NFY/CBF and, 9 nucleotides away, a 3′-end region (5′-CCCTGTTGGCCATC-3′) that is equally important for ER stress-mediated induction of the gene. This sequence is the binding site for endogenous YY1 at the 5′-CCCTGTTGG-3′ part and for undefined factors at the CCATC 3′-end. ATF6α-YY1, but not XBP1, interacted with the ERGIC-53 regulatory region and activated ERGIC-53 ER stress response element-dependent transcription. A molecular model for the transcriptional regulation of the ERGIC-53 gene is proposed.

In eukaryotic cells, protein misfolding within the ER 3 activates a signaling pathway known as the unfolded protein response (UPR) or ER stress response (1)(2)(3). The UPR rapidly induces transcription of ER stress genes (1-3), attenuates pro-tein synthesis by activating PERK-dependent phosphorylation of the translational initiation factor eIF2␣, and intensifies the proteasomal degradation pathway (4,5). In mammalian cells, UPR stimulates the expression of proteins ERGIC-53, MCFD2, and VIP36, which exert functions in the post-ER compartments of the secretory pathway (6 -8). ERGIC-53 is a highly conserved calcium-dependent lectin that shares structural and functional homology with VIP36 (9). ERGIC-53 is mainly localized within the ERGIC; it cycles continuously between the ER and the Golgi complex and exports a defined number of glycoproteins from the ER (9 -16). The adaptor protein MCFD2 forms a complex with ERGIC-53 to transport newly synthesized clotting factors V and VIII to the Golgi complex (17)(18)(19). ERGIC-53 and MCFD2 accumulate in response to distinct signaling pathways of the cell stress response. In particular, heat shock stimulates preferential translation of ERGIC-53 and MCFD2 mRNAs (7), whereas UPR induced by either thapsigargin or nitric oxide activates the transcription of both genes (7,8).
Transcriptional activation by the UPR requires the presence of cis-acting ER stress response elements (ERSE) on the promoter region of target genes. Three types of ERSE have been identified: ERSE-I and ERSE-II (20 -22, 23) and the mammalian unfolded protein response element (UPRE) (23)(24)(25). Multiple copies of the ERSE-I control the UPR-mediated induction of the ER-resident proteins Grp78/BiP, Grp94, calreticulin, and protein-disulfide isomerases (20,21). ERSE-I contains a CCAAT site at its 5Ј-end for the constitutive transcription factor NFY/CBF (20,26,27), a 9-bp spacer containing the CGG triplet that is the TFII-I binding site (28), and a CCACG motif at the 3Ј-end that is required for ATF6␣ (activating transcription factor 6␣) recruitment (20, 26, 28 -31). ATF6␣ is an ER transmembrane protein that, during ER stress, can be recruited by transport vesicles destined for the Golgi complex, where, consequent to intramembrane proteolysis, a 50-kDa cytosolic form is generated that activates transcription of UPR genes (26 -32). A single copy of the ERSE-II type regulates the ER stress response of the Herp gene, which encodes an ER integral membrane protein that is involved in the ER-associated degradation pathway (22,33,34). ERSE-II (5Ј-ATTGGNCCAC(G/A)-3Ј) retains a reversed NFY/CBF binding site at its 5Ј-end and a flanking ATF6␣ site (22). The UPRE, which controls the expression of a subset of ER-resident chaperones (35,36), contains the ATF6␣ binding site on its complementary strand and is also the preferred binding site for the transcription factor XBP1 (X-box-binding protein 1). XBP1 is under the control of the ER membrane nuclease IRE1␣ (inositol-requiring enzyme 1␣), which in response to protein misfolding triggers the processing of XBP1 mRNA, thereby leading to the synthesis of the transcription factor (35,37,38). Interestingly, XBP1 binds both the UPRE and ERSE sequences of target genes (23,37) independently of NFY/CBF, whereas ATF6␣ binds UPRE albeit with a lower affinity compared with ERSE (23,37).
We have studied the regulation of the ERGIC-53 promoter in an attempt to shed light on the transcriptional mechanisms that control ERGIC-53 expression. The transcriptional regulation of ERGIC-53 by the UPR requires a single cis-acting element (ERGIC-53 ERSE), which is highly conserved in mammals and is different from the other types of ERSE identified.
Cell Culture and Induction of ER Stress-HeLa cells, mouse embryonic fibroblasts (MEFs) derived from wild type or IRE1␣ knock-out mice (38), were grown at 37°C in a humidified 5% CO 2 , 95% atmosphere in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum, 4.5 g/liter glucose, and 2 mM glutamine. To induce ER stress, we incubated actively growing cells for 8 h (unless otherwise indicated) with 300 nM thapsigargin (TG), 2 mM DTT, or 10 M MG132 (Sigma) before harvesting.
Construction of Plasmids-A 1084-bp fragment spanning from nt Ϫ874 to ϩ202 of the human BiP/Grp78 promoter (EMBL Data Library accession number X59969) (40) and a 1036-bp fragment corresponding to the LMAN1/ERGIC-53 gene (NCBI ref/ NT_025028) from nt Ϫ1003 to ϩ33 were isolated by PCR amplification from genomic DNA extracted from HeLa cells, and PCR fragments were gel-purified and cloned by using the pGEM-T Easy Vector system (Promega). The Bip/Gpr78 promoter region spanning from nt Ϫ838 to Ϫ16 was PCR-amplified with synthetic oligonucleotides having KpnI-XhoI flanking restriction sites and subcloned in the pGL3Basic vector (Promega) to drive the expression of firefly luciferase reporter gene. Similarly, the ERGIC-53 promoter region (nt Ϫ1000 to Ϫ1) was ligated to the KpnI-HindIII site of the pGL3Basic vector (Promega) upstream from the luciferase reporter gene. Progressive deletion of the 5Ј-flanking region of the ERGIC-53 promoter was obtained by PCR amplification using forward primers containing the 5Ј-KpnI site and common reverse primers with the 3Ј-HindIII site. The Ϫ149 to Ϫ1 and Ϫ149 to Ϫ63 fragments were obtained by PCR, digested, and inserted in the KpnI-BglII site of the pGL3 promoter vector (Promega), which contains the SV40 minimal promoter upstream from the luciferase coding sequence (SV40Luc). Synthetic oligonucleotides corresponding to the ERSE-I consensus sequence, the Ϫ32 to Ϫ1 and the Ϫ65 to Ϫ31 region of the ERGIC-53 ERSE, and cassette and point mutations of ERGIC-53 ERSE were synthesized, in vitro annealed, and inserted in the KpnI-BglII site of the pGL3 promoter vector and controlled by sequencing. The pCGN-ATF6-(1-373) and pCGN-ATF6-(1-273) expression vectors (31,41) were kindly provided by A. S. Lee. The pCGN-ATF6-(1-373)m1 was already described (25).
Transfection Experiments for Transient Expression of Reporter Plasmids-Actively growing HeLa and MEF cells were seeded onto 60-mm plates at ϳ20 -30% of confluence and  A, activity of luciferase reporter gene driven by the undeleted ERGIC-53 promoter (1036 bp) or by deletion mutants lacking each of the three ERSE-like sequences. HeLa cells transiently transfected with the recombinant ERGIC-53 promoter were unexposed (C) or exposed to 300 nM thapsigargin (TG) for 8 h. Luciferase activity obtained with the reporter vector driven by the undeleted Bip/Grp78 promoter served as positive control. B, activity of the Ϫ149 to Ϫ1 deletion mutant assayed in cell extracts of unexposed cells (C) or HeLa cells exposed to 300 nM TG, 2 mM DTT, and 10 M MG132 for 8 h. The histograms represent relative luciferase activity normalized for ␤-galactosidase activity and protein concentration in the cell lysates. -Fold induction is calculated as the ratio between the enzyme activity of induced cells and noninduced cells. Values represent the average of at least three independent experiments performed in duplicate. Statistical analysis was performed using Student's t test (n ϭ 6-8;*,p Ͻ 0.001; **, p Ͻ 0.005 relative to the control, respectively). C, real time RT-PCR analysis of ERGIC-53 and ER chaperone gene Grp94 mRNAs accumulation in response to ER stress. Total RNA fractions obtained from HeLa cells exposed for 4 -8 h to 300 nM TG were analyzed by RT-PCR to measure the level of the indicated mRNAs. The values were normalized to GAPDH mRNA. The relative -fold induction was calculated as the ratio of treated cells divided by untreated cells according to the 2 Ϫ⌬⌬CT method. Each value represents the mean Ϯ S.D. of three independent experiments. measure LacZ reporter gene activity (␤-galactosidase), we incubated cell extracts for 1 h at 37°C in ␤-galactosidase assay buffer (200 mM sodium phosphate buffer, pH 7.3, 2 mM MgCl 2 , 100 mM ␤-mercaptoethanol, 1.33 mg/ml o-nitrophenyl-␤-Dgalactopyranoside). The reaction was blocked by adding 1 M sodium carbonate, and absorbance was measured at 420 nm. We measured luciferase activity with a Berthold luminometer in 40 l of cell lysate supplemented with 100 l of luciferase assay reagent (Promega). "Relative luciferase activity" is defined as the luciferase-to-␤-galactosidase activity ratio and normalized for the protein concentration of each sample. "-Fold induction" is defined as the ratio between the enzyme activities of treated cells with respect to that of untreated cells. In all experiments, values are reported as the average and S.D. of at least three independent experiments carried out in duplicate. Statistical analysis was performed using Student's t test (n ϭ 6 -8).
Chromatin Immunoprecipitation Assays-1 ϫ 10 6 HeLa cells were treated or not for 2-4 h with 300 nM TG for the UPR induction. For the ChIP analysis of overexpressed ATF6␣, HeLa cells were transfected with 6 g of the pCGN-ATF6-(1-373) and pCGN-ATF6-(1-273) expression vectors (31,41) and harvested 24 h after transfection. Experiments were performed according to the instructions of the manufacturer (Upstate Biotechnology, Inc.). Briefly, cells were exposed to 1% formaldehyde for 10 min at 37°C to obtain protein-DNA cross-linking.
The nuclear fraction was sonicated to obtain chromatin fragments of 200 -1000 bp; an aliquot (5% of the total volume) was removed from each sample and used as the input fraction. Chromatin was precleared by preincubation with a DNA salmon sperm/protein A-agarose 50% slurry (Upstate Biotechnology) for 1 h at 4°C. The agarose was centrifuged, and the precleared chromatin supernatant was then incubated with the indicated antibodies overnight at 4°C. The protein-DNA-antibody complexes were collected by the addition of the salmon sperm DNA-protein A-agarose (2 h at 4°C) and washed, and protein-DNA cross-linking was reversed (4 h at 65°C). DNA was purified by phenol/chloroform extraction and ethanol precipitation, and aliquots (25%) of the purified materials underwent PCR (5 min at 94°C; 1 min at 94°C, 1 min at 52°C, 1 min at 72°C for 40 cycles; 5 min at 72°C). The following primers were used in the PCR: 5Ј-TGAACCAATGGGACCAGC-3Ј and 5Ј-CTCACCGTCGCCTACTCG-3Ј to amplify a 254-bp fragment of the human Bip/Grp78 promoter spanning from nt Ϫ267 to nt Ϫ13; 5Ј-AAGCGAAGGTTGGAGTCC-3Ј and 5Ј-CGCCATCTTGGATTCTCC-3Ј to amplify a 271-bp fragment of the LMAN1/ERGIC-53 promoter extending from nt Ϫ226 to nt ϩ45.
Preparation of Nuclear Extracts and Electrophoresis Mobility Shift Assays-5-6 ϫ 10 6 HeLa cells were either treated for 2 h with 300 nM TG to induce ER stress or transiently transfected with 6 g of the pCGN-ATF6-(1-373) expression vector and harvested 48 h after transfection. Cells were washed with cold phosphate-buffered saline and harvested by scraping. The cell pellet was resuspended in extraction buffer containing 10 mM HEPES, pH 7.9, 10 mM KCl, 1.5 mM MgCl 2 , 0.1 mM EGTA, 0.5 mM DTT, 0.5 mM PMSF, 10 g/ml aprotinin, 10 g/ml leupeptin, passed through a needle, kept on ice for 45 min, and centrifuged (15 min at 14,000 rpm at 4°C). The nuclear pellet was then resuspended in high salt extraction buffer containing 10 mM HEPES, pH 7.9, 0.4 mM NaCl, 1.5 mM MgCl 2 , 0.1 mM EGTA, 0.5 mM DTT, 0.5 mM phenylmethylsulfonyl fluoride, 10 g/ml aprotinin, 10 g/ml leupeptin and incubated for 45 min at 4°C. The nuclear extract supernatant was obtained by centrifugation (30 min at 14,000 rpm at 4°C), protein concentration was determined, and 5-g aliquots were stored at Ϫ80°C until used. Double-stranded synthetic oligonucleotides were radiolabeled using [␥-32 P]ATP (3000 Ci mmol Ϫ1 ; Amersham Biosciences) and T4 polynucleotide kinase (Fermentas). The binding reaction was carried out for 20 min at room temperature with 5 g of nuclear proteins in 4% glycerol, 1 mM MgCl 2 , 0.5 mM EDTA, pH 8.0, 0.5 mM DTT, 50 mM Tris-HCl, pH 7.5, 0.250 g/l poly(dI-dC) containing 30,000 cpm of radiolabeled probe and a 50 -100-fold molar excess of unlabeled competitor oligonucleotide when indicated. For supershift experiments, 4 g of specific anti-NFY/CBF-A and 10 g of either anti-YY1-or anti-HA-specific antibody was added to the binding reaction and incubated for 30 min before the addition of the radiolabeled probe. DNA-protein complexes were separated by 5% nondenaturing polyacrylamide gel and revealed by PhosphorImager analysis (Typhoon 8600; Amersham Biosciences).

Isolation of the ERGIC-53 Minimal Inducible Promoter-To
identify ERSE elements in the ERGIC-53 promoter, we looked for transcription factor binding sites (TESS analysis) on about 1000 bp of the 5Ј-region of the ERGIC-53 gene. Three Synthetic oligonucleotides corresponding to the ERGIC-53 ERSE or to ERGIC-53 ERSE mutants (mut 1-5) were inserted in the PGL3 promoter vector, and the basal and thapsigargin-dependent luciferase activities (n ϭ 6 -8; *, p Ͻ 0.001) were measured as described in the legend to Fig. 2. Synthetic oligonucleotides corresponding to the consensus ERSE-I or to its inactive form served as control. Mutated nucleotide sequences are underlined. B, effect of single point mutations (mut 6 -17) in the Ϫ34 to Ϫ45 segment of the ERGIC-53 ERSE. Mutated fragments were inserted in the PGL3 promoter vector, and the corresponding reporter activity was assayed as reported in the legend to Fig. 2. The activity of a double substitution (mut 18) is also shown. The graphs represent the average Ϯ S.D. of three independent experiments. Statistical analysis was performed using Student's t test (n ϭ 6; *, p Ͻ 0.001; **, p Ͻ 0.025; #, p Ͻ 0.01, respectively).

TABLE 1 Conservation of the ERSE ERGIC-53 element
5Ј-CCAAT motifs scored high as binding sites for the transcription factor NFY/CBF. This result suggested the presence of three putative ERSEs located at positions Ϫ334 to Ϫ316, Ϫ277 to Ϫ259, and Ϫ62 to Ϫ44 (Fig. 1A). Alignment of the three ERGIC-53 ERSE-like sequences with ERSE of human genes responsive to ER stress (Fig. 1B) revealed differences in the downstream region (excluding the 5Ј-CCAAT box). In addition, five sites had a high affinity for transcription factors of the SP family of proteins (Fig. 1A), and the ERGIC-53 promoter lacked known transcription initiation sequences (Fig. 1A).
Thus, we attempted to isolate the ERGIC-53 minimal inducible promoter by measuring the activity of the firefly luciferase reporter gene driven by deletion mutants lacking the three ERSE-like sequences we had identified ( Fig. 2A). Expression vectors were transfected in HeLa cells treated or not with TG. The 1036-bp region containing all three putative ERSE enhanced luciferase expression 3.61 Ϯ 0.98-fold in TG-treated cells ( Fig. 2A, right). Deletions of the Ϫ335 to Ϫ317 and Ϫ278 to Ϫ259 ERSE-like sequences reduced the basal activity in untreated cells, but the -fold induction was retained in the Ϫ149 to Ϫ1 region ( Fig. 2A, right) that contains the Ϫ62 to Ϫ35 ERSE-like motif (Fig. 1A). With the three intact ERSE-Is, TG caused a 6.2 Ϯ 1.12-fold increase in the expression of the control Grp78/BiP promoter. Thus, the minimal inducible region was retained in the Ϫ149 to Ϫ1 region. This region could initiate transcription in response to various ER stress inducers (Fig. 2B), giving a -fold induction of the reporter gene of 4.2 Ϯ 0.49 with TG, 3.24 Ϯ 0.62 with the reducing agent DTT, and 2.14 Ϯ 0.38 with proteasome inhibitor MG132. In contrast, the undeleted Grp78/ BiP promoter enhanced luciferase activity in response to TG and DTT by 6.79 Ϯ 0.72-and 6.72 Ϯ 0.78-fold, respectively and by 4.98 Ϯ 0.52-fold in response to MG132. Finally, the activity of the Ϫ149 to Ϫ1 ERGIC-53 promoter was consistent with the -fold ERGIC-53 mRNA induction by TG as measured by real time RT-PCR analysis (Fig. 2C), indicating that the Ϫ149 to Ϫ1 region ( Fig. 2A, right) that contains a single ERSE-like is sufficient to ensure higher ERGIC-53 mRNA levels in response to TG treatment.
UPR Regulates ERGIC-53 Transcription by Means of a Single ERSE-To identify the ERSE within the Ϫ149 to Ϫ1 region, we first compared the luciferase activity of this region with that of the Ϫ149 to Ϫ63 deletion mutant that lacks the putative ERSE (Fig. 3A). Deletion mutants were placed upstream from the SV40Luc vector to drive the expression of the luciferase reporter gene. As expected, reporter assays confirmed that the Ϫ149 to Ϫ1 region conferred thapsigargin-mediated activation of the SV40Luc vector and enhanced the activity . ERGIC-53 ERSE is a general sensor of ER stress. A, activity of the consensus ERSE and its mutated form (mut ERSE) assayed in cell extracts derived from unexposed cells (C) or HeLa cells exposed to 300 nM TG, 2 mM DTT, or 10 M MG132 for 8 h. B, activity of ERGIC-53 ERSE and its mutated form (mut 4) assayed in cell extracts derived from HeLa cells untreated (C) or exposed for 8 h to 300 nM TG, 2 mM DTT, or 10 M MG132. The histograms represent relative luciferase activity normalized for ␤-galactosidase activity and protein concentration in the cell lysates. Induction -fold is calculated as the ratio between the luciferase activity of induced cells with respect to noninduced cells. Plotted values represent the average of at least three independent experiments performed in duplicate (n ϭ 6 -8; *, p Ͻ 0.001; **, p Ͻ 0.005, respectively).

The ERGIC-53 ERSE
of reporter gene by 2.50 Ϯ 0.08-fold. Conversely, the ERSElacking mutant (Ϫ149 to Ϫ63) was inactive in thapsigargintreated cells, which indicates that the putative ERGIC-53 ERSE is located within the Ϫ63 to Ϫ1 region of the promoter (Fig. 3A). Thus, we mapped the boundaries of the ERGIC-53 ERSE, by investigating the capacity of the Ϫ65 to Ϫ31 and Ϫ32 to Ϫ1 deletion mutants to confer TG-mediated activation to the SV40Luc reporter vector (Fig. 3B). We also compared the activity of ERGIC-53 promoter deletions with that of the reporter gene under the control of the ERSE-I consensus sequence (20,21). The results of luciferase assays showed that the ERGIC-53 ERSE spans from nucleotide Ϫ65 to nucleotide Ϫ31 of the promoter. In addition, the -fold induction of the SV40Luc vector driven by the ERGIC-53 ERSE (2.36 Ϯ 0.12) was comparable with that driven by the control vector bearing a single copy consensus ERSE-I (2.69 Ϯ 0.27). The ERGIC-53 ERSE sequence identified is highly conserved and is located at a similar position along the promoter region of several mammalian ERGIC-53 genes as shown by BLAST analysis (42) carried out on the NCBI genomic data base (available on the World Wide Web). This suggests that ERGIC-53 could be similarly regulated by ER stress in different species (Table 1).
Mutational Analysis Reveals the Novel Structure of the ERGIC-53 ERSE-To define the sequence required for the ER stress response of the ERGIC-53 ERSE (Fig. 3C), we inserted oligonucleotides bearing the substitutions shown in Fig. 4A in the SV40Luc reporter plasmid and determined their capacity to confer the ER stress response to the minimal SV40 promoter. We also performed these experiments using reporter vectors bearing a single copy of the consensus or a mutated ERSE-I (20, 21). As expected, mutation of the entire 5Ј-CCAAT box (mut 1) impaired ERSE activity. Interestingly, nucleotide replacement of the Ϫ48 to Ϫ35 segment (mut 2-4) inhibited the response to TG. Inhibition was greater when we replaced the segment between nucleotides Ϫ44 and Ϫ35 (mut 2 and 4), which indicates that this sequence is crucial for TG-dependent activation of ERGIC-53. Mutation in the segment spanning nucleotides Ϫ55 to Ϫ49 (mut 5) did not result in a similar reduction (Fig.  4A). To verify these findings, we introduced single point mutations into each nucleotide of the Ϫ44 to Ϫ35 sequence (Fig. 4B). Replacement of any of the nucleotides within this sequence impaired -fold induction, albeit to different degrees (Fig. 4B, mut 7-16). In particular, the Ϫ44 G-to-C transition (mut 7) resulted in a -fold induction decrease comparable with that of the Ϫ48 to Ϫ44 cassette mutant (mut 3), whereas the Ϫ35 C-to-T transversion (mut 16) almost completely abolished induction. Replacement of the outer nucleotides (mut 6 and 17) had little effect on the ER stress-dependent activity of the ERGIC-53 ERSE.
Therefore, the sequence of the ERGIC-53 ERSE crucial for UPR-mediated gene induction is constituted by a leader 5Ј-CCAAT-3Ј sequence followed by a novel Ϫ48 to Ϫ35 sequence (5Ј-CCCTGTTGGCCATC-3Ј), both of which are required to confer full inducibility in response to ER stress. The ERGIC-53 ERSE is a general sensor of ER stress. The induction of the ERGIC-53 ERSE reporter gene transfected in HeLa cells was 2.61 Ϯ 0.32-fold with TG, 2.42 Ϯ 0.35 with DTT, and 1.88 Ϯ 0.12 with MG132 (Fig. 5). These values were comparable The numbers with arrows indicate the relative position of the PCR primers used to reveal DNA protein interaction. B, sonicated chromatin obtained from either steady-state (C) or thapsigargin-treated (TG) HeLa cells was immunoprecipitated by anti-IgG as control or by the indicated specific antibodies recognizing different basal and ER stress-activated transcription factors and revealed by PCR amplification, as described under "Experimental Procedures." C, HeLa cells transfected with empty (Ϫ), HA-ATF6(373), or HA-ATF6(273) expression vector were cross-linked with formaldehyde subjected to ChIP assay with 10 g of an anti-HA epitope-specific antibody. PCRs were performed as described above.

The ERGIC-53 ERSE
with those obtained with the ERGIC-53 deletion mutant Ϫ149 to Ϫ1 despite the different reporter vector (Fig. 2B). mut 4 did not respond to any of the ER stress inducers, thus confirming the regulatory role of the 3Ј region (5Ј-CCATC-3Ј) of the ERGIC-53 ERSE.
NFY/CBF and ATF6␣-YY1 Interact in Vivo with the ERGIC-53 Promoter-We next asked whether transcription factors involved in the ER stress response could interact with the ERGIC-53 promoter. Accordingly, we examined DNAprotein complexes from untreated and TG-stressed cells using chromatin immunoprecipitation. The presence of endogenous factors NFY, Sp1, ATF6␣, YY1, or XBP1 in the immunoprecipitates was revealed by PCR amplification of a 271-bp fragment of the ERGIC-53 promoter extending from nucleotide Ϫ226 to nucleotide ϩ45, containing a single ERGIC-53 ERSE (Fig. 6A). As a control, we examined the 254-bp fragment of the human Bip/Grp78 promoter spanning from nucleotide Ϫ267 to nucleotide Ϫ13, which harbors the three ERSEs (Fig. 6A). Untreated cells retained endogenous NFY, YY1, and Sp1 bound on ERGIC-53, and NFY binding was more evident in untreated cells on the control Bip/Grp78 promoter, which contains multiple copies of ERSE (Fig. 6B). Consequent to TG-induced ER stress, and on both promoters, NFY binding was more pronounced, whereas Sp1 and YY1 binding was slightly higher than in uninduced cells. In TG-stressed cells, there was endogenous ATF6␣ binding on the ERGIC-53 promoter and on the control gene Bip/Grp78. In contrast, induction of ER stress raises the interaction of endogenous XBP1 with Bip/Grp78 but not with the proximal region of the ERGIC-53 gene (Fig.  6B, bottom). To verify ATF6␣ binding to the ERGIC-53 promoter, we examined DNA-protein complexes from HeLa cells transiently transfected with plasmid vectors expressing two distinct HA-tagged ATF6␣ nuclear forms (Fig. 6C): the active HA-tagged form ATF6(373) and the deletion mutant ATF6(273) that lacked the B-ZIP domain (31). Chromatin immunoprecipitation with anti-HA antibody revealed binding of the ATF6(373) form to the ERGIC-53 promoter and to the control gene Bip/Grp78 (Fig. 6C). Instead, immunoprecipitation with anti-HA antibody did not reveal binding of the HA-tagged ATF6(273)-deleted form, which confirms recruitment of active ATF6␣ at the ERGIC-53 promoter (Fig. 6C).
ATF6␣ but Not XBP1 Stimulates ERGIC-53 ERSE Activity-In transiently transfected HeLa cells, the overexpression of the HA-tagged ATF6␣ nuclear form was able to stimulate, in the absence of ER stress induction, the activity of the ERGIC-53 minimal promoter (Ϫ149 to Ϫ1) by 5.97 Ϯ 0.59-fold and the control plasmid Grp78/BiP-driven reporter gene by 15.28 Ϯ 1.35-fold (Fig. 7A). To define the sequence required for the ATF6␣-dependent transactivation, we compared the effect of the HA-tagged ATF6(373) expression on the wild-type and mutant (mut 1-5) ERGIC-53 ERSE (Fig. 7B) reporter gene induction. Overexpressed ATF6(373) enhanced the ERGIC-53 ERSE luciferase activity by about 18-fold and enhanced the expression of the ERSE-I control reporter, albeit to a lesser extent (about 12-fold). As we would expect, mutation of the CCAAT box (mut 1) strongly reduced ERGIC-53 ERSE activity. Nucleotide replacement of the Ϫ48 to Ϫ35 segment (mut 2-4) inhibited the response to ATF6(373), and the inhibition was greater for mut 2, which confirms the importance of this sequence for ER stress activation of ERGIC-53 (Fig. 4B). We verified further the involvement of ATF6␣ by analyzing the effect of the overexpression of the ATF6␣ dominant negative form ATF6-(1-373)m1, which is able to prevent the activity of endogenous ATF6␣ (25). Expression of the ATF6-(1-373)m1 construct completely inhibited the TG induction of the ERGIC-53 minimal promoter and ERGIC-53 ERSE-driven reporter genes; a similar result was observed with the control Bip/Grp78 and ERSE-I reporter (Fig. 7C).
Finally, we tested the requirement of XBP1 for the transcriptional activation of ERGIC-53 in MEFs derived from wild-type or IRE1␣ knock-out mice (IRE1 Ϫ/Ϫ MEF) (Fig. 8A). TG induction of the ERGIC-53 promoter and of the ERGIC-53 ERSE reporter plasmid was similar in wild-type and IRE1 Ϫ/Ϫ MEF cells, which indicates that ERGIC-53 transcription occurred independently of the IRE1/XBP1 pathway. This conclusion was confirmed by the analysis of ERGIC-53 mRNA level in response to TG induction in wild-type and in IRE1 Ϫ/Ϫ MEF cells (Fig.  2C). Real time RT-PCR assays showed that wild-type and IRE1 Ϫ/Ϫ MEF cells have a similar -fold induction of ERGIC-53 transcripts in response to TG treatment. Differently, reporter and RT-PCR assays revealed a significant decrease in Bip/ Grp78 and Grp94 -fold induction upon ER stress, thereby confirming direct involvement of XBP1 in the ERSE-I-mediated transactivation. Thus, our results show that ATF6␣ requires the Ϫ48 to Ϫ35 segment of the ERSE to activate the ERGIC-53 promoter, and, unlike XBP1, it is essential for the transcriptional regulation of ERGIC-53.
The ERGIC-53 ERSE Retains Binding Sites for Nuclear Factors NFY/CBF and YY1-The ERGIC-53 ERSE contained two distinct regulatory regions: a high affinity NFY/CBF binding site located at the 5Ј-end and a downstream 5Ј-CCCTGTT-GGCCATC-3Ј sequence containing an NFY/CBF-like site in the reverse orientation (5Ј-GTTGG-3Ј). To examine the interaction of nuclear proteins with these motifs, we carried

The ERGIC-53 ERSE
out mobility shift assays using wild-type or mutated forms of ERGIC-53 ERSE as probes (Fig. 9A). In both uninduced and TG-induced cell extracts (data not shown), there were two distinct ERSE-binding complexes (EBC-I and EBC-II) that specifically interacted with the ERGIC-53 ERSE (Fig. 9B and supplemental materials). Preincubation with anti-NFY/CBF antibody resulted in the supershift of EBC-II in all samples, suggesting that NFY/CBF is the major component of the slower migrating complex (Fig. 9B). Preincubation with anti-YY1 antibody resulted in inhibition of the faster migrating complex, suggesting that YY1 is the major component of EBC-I (Fig. 9C). The ERGIC-53 ERSE mut 1, in which the 5Ј-CCAAT-3Ј motif was mutated, retained the EBC-II interaction, which indicates that NFY/CBF could interact with the inner NFY/CBF-like. Mutation of the 5Ј-CCCTGT-TGG-3Ј sequence (mut 2 and mut 3) impaired EBC-I activity, which suggests that YY1 interacts with the 5Ј-CCCTGT-TGG-3Ј sequence. Furthermore, EBC-II binding to mut 4 ( Fig. 9B) was severely reduced, thereby confirming that the pentameric 5Ј-CCATC-3Ј motif plays a critical role within the 5Ј-CCCTGTTGGCCATC-3Ј of the ERSE. In conclusion, the results of in vitro binding assays strongly support the concept that the ERGIC-53 ERSE is constituted by the 5Ј-CCAAT box and two distinct motifs located 9 nt downstream, the 5Ј-CCCTGTTGG and the CCATC-3Ј part, both of which are important for the ERSE induction.

DISCUSSION
In an attempt to identify the transcription mechanisms that control the ERGIC-53 gene, we have studied the transcription regulation of the ERGIC-53 promoter. Here we show that a single cis-acting element, the ERGIC-53 ERSE, which has a novel ERSE organization, enhances the UPR-dependent expression of the gene, and favors the formation of a transcriptional complex constituted by ER stress factors.
The Novel ERSE Structure of the ERGIC-53 ERSE-The ERSE found in the ERGIC-53 promoter region is distinct from all other ERSEs and UPREs so far identified in the promoter regions of UPR-regulated genes (20 -25). In the newly identified ERGIC-53 ERSE, the NFY/CBF CCAAT site, which is a feature of both ERSE-I and -II elements, is functionally coupled to the CCCTGTTGGCCATC ER stress regulatory sequence, located 9 nucleotides downstream from the CCAAT site and equally important for UPR-mediated regulation of the gene.
The CCAAT site of the ERGIC-53 ERSE constitutively binds NFY/CBF that in all ERSE types previously reported serves as the "foundation" upon which the UPR-induced ER stress factors are assembled (20, 26 -30). In our study, replacement of the CCAAT domain impaired the functional activity of the ERGIC-53 ERSE, which suggests that NFY/ CBF exerts a function in the ER stress-induced activation of ERGIC-53 similar to that exerted in other ERSE types. Interestingly, we showed that the downstream CCCTGTTGGC-CATC sequence is equally important for the UPR induction. TESS analysis of the sequence revealed, in the reverse orientation of the DNA sequence, a low affinity binding site for the basal transcription factor NFY/CBF (GTTGG). We found that NFY/CBF could interact with the inner NFY/CBF-like. Interestingly, a double substitution mutant that reconstituted the NFY/CBF consensus binding site (CCAATCG) did not enhance ERGIC-53 ERSE activity in response to ER stress (Fig. 4B, mut 18), which suggests that the NFY/CBF protein is not involved in the UPR regulation of that sequence. The downstream sequence could be divided into two parts: the 5Ј-end part CCCTGTTGG that is required for the interaction of YY1 and the 3Ј-end part CCATC that possibly represents the binding site for accessory and as yet unidentified regulatory protein(s).
The ERGIC-53 ERSE Is the Binding Site for Basic Factors NFY/CBF and YY1 and Is Activated by the ATF6 Pathway of the UPR-The results of our ChIP assays showed that ERGIC-53 is a target gene for nuclear factors commonly involved in UPR-mediated activation of gene expression. Moreover, we found that not only NFY/CBF but also YY1 is recruited by the ERGIC-53 promoter, and both proteins are immunoprecipitated particularly well in the nuclei of ERstressed cells. YY1 is a co-activator of the ER stress response in mammalian cells (28 -31), and our results suggest that it plays a similar role during UPR-mediated activation of ERGIC-53. We also found that SP proteins constitutively interact with the ER stress-responsive region of the ERGIC-53 gene, presumably by recognizing a high affinity site in the promoter region analyzed. This finding is compatible with the report that nuclear factors of the SP family are essential for the stress-induced response of Grp78 in which they constitutively bind ERSE sequences (43).
Our experiments show that, differently from the Grp78 promoter, ER stress-induced XBP1 factor does not enter the ERGIC-53 regulatory region. This result, together with our previous finding that, in response to nitric oxide-induced ER stress, ERGIC-53 mRNA accumulated independently of XBP1 activation (8), suggested that XBP1 is not required for ERGIC-53 activation during ER stress. Results of ERGIC-53 expression analyses performed in IRE1 knocked-out cells clearly show that ERGIC-53 induction does not rely on the IRE1␣-XBP1 pathway, suggesting that the UPR modulates ERGIC-53 expression selectively during the ER stress response.
Earlier evidence showed that induction of ERGIC-53 during the UPR was dependent on the activation of the ERSE binding factor ATF6␣ (6). Now we show that ATF6␣ is FIGURE 8. Analysis of ERGIC-53 gene expression in IRE1␣-null MEFs in response to ER stress. A, wild type or IRE1␣ Ϫ/Ϫ MEFs cells were transfected with the indicated reporter vector, treated or not for 8 h with 300 nM TG, and then assayed for luciferase activities. The values reported represent the average Ϯ S.D. of at least three independent experiments performed in duplicate. Statistical analysis was performed using Student's t test (n ϭ 6 -8; *, p Ͻ 0.001; **, p Ͻ 0.005, respectively, relative to control cells). B, real time RT-PCR analysis of ERGIC-53 and ER chaperone gene Grp94 mRNA accumulation in response to ER stress. Total RNA fractions obtained from either wild type or IRE1␣ Ϫ/Ϫ MEF cells exposed or not for 8 h to 300 nM TG were analyzed by RT-PCR to measure the level of the indicated mRNAs. The values were normalized to c-Abl mRNA. The relative -fold induction was calculated as the ratio of treated cells divided by untreated cells according to the 2 Ϫ⌬⌬CT method. Each value represents the mean Ϯ S.D. of three independent experiments. context, we found that ER stress rapidly stimulated similar modifications of the histone acetylation pattern at both the ERGIC-53 and Bip/Grp78 promoter. 4 In particular, we observed a rapid increase of H3 acetylation and decrease of the H4 acetylation on both promoters. Therefore, we believe that the observed modifications of histone acetylation patterns could favor the formation of the transactivation complex at the ERGIC-53 promoter during UPR. In addition, the recruitment of co-activators that either interact with other transcription factors or modify histones represents a critical event for the achievement of an open chromatin state that has been well described for, among other functions, the regulation of gene expression mediated by YY1 (44,45), which we show to play a pivotal role in the ERGIC-53 ERSE complex.
Because the ERSE sequence is highly conserved, ERGIC-53 could be regulated by the same mechanism in diverse species (Table 1). Our model is consistent with the requirements for conservation of the ER stress response and could explain how the newly identified ERGIC-53 ERSE is a general sensor for ER stress.
A putative ERSE sequence is present in the promoter of MCFD2, a protein that is functionally related to ERGIC-53. To gain further insight into the role of UPR in the control of genes involved in post-ER functions, it would be interesting to establish whether UPR regulates MCFD2 transcription by a similar mechanism. A crucial question is the function of ERGIC-53 and MCFD2 during the UPR. We shall address this issue studying the effect of the ER stress on the intracellular trafficking and interaction properties of the two proteins.