ERp57 Modulates STAT3 Signaling from the Lumen of the Endoplasmic Reticulum*

ERp57 is an endoplasmic reticulum (ER) resident thiol disulfide oxidoreductase. Using the gene trap technique, we created a ERp57-deficient mouse model. Targeted deletion of the Pdia3 gene, which encodes ERp57, in mice is embryonic lethal at embryonic day (E) 13.5. β-Galactosidase reporter gene analysis revealed that ERp57 is expressed early on during blastocyst formation with the highest expression in the inner cell mass. In early stages of mouse embryonic development (E11.5) there is a relatively low level of expression of ERp57. As the embryos developed, ERp57 became highly expressed in both the brain and the lungs (E15.5 and E18.5). The absence of ERp57 has no impact on ER morphology; expression of ER-associated chaperones and folding enzymes, ER stress, or apoptosis. ERp57 has been reported to interact with STAT3 (signal transducer and activator of transcription)-DNA complexes. We show here that STAT3-dependent signaling is increased in the absence of ERp57 and this can be rescued by expression of ER-targeted ERp57 but not by cytoplasmic-targeted protein, indicating that ERp57 affects STAT3 signaling from the lumen of the ER. ERp57 effects on STAT3 signaling are enhanced by ER luminal complex formation between ERp57 and calreticulin. In conclusion, we show that ERp57 deficiency in mouse is embryonic lethal at E13.5 and ERp57-dependent modulation of STAT3 signaling may contribute to this phenotype.

The endoplasmic reticulum (ER) 5 is involved in many cellular functions including protein synthesis and modification, reg-ulation of Ca 2ϩ homeostasis, phospholipid and steroid synthesis, and regulation of the response to cellular stress (1)(2)(3). To carry out these diverse functions, the ER is equipped with many chaperone proteins and folding enzymes (3). For example, calreticulin, calnexin, and the oxidoreductase ERp57 are components of the folding machinery involved in quality control of newly synthesized glycoproteins (3). ERp57 forms complexes with calnexin and calreticulin to assist in chaperone functions to ensure that newly synthesized proteins are correctly folded (3). Targeted deletion of ERp57 in B cells in mice results in normal B cell development and proliferation as well as antibody production (4). However, there is aberrant assembly of the peptide loading complex indicating that ERp57 is involved in the assembly of the peptide loading complex and this protein contributes both qualitatively and quantitatively to major histocompatibility complex (MHC) class I antigen presentation in vivo (4). Interestingly, small interfering RNA studies also demonstrate that ERp57 might be critical for oxidative folding of immunoglobulin heavy chain but not important for peptide loading of class I molecules (5).
Aside from its role as a folding enzyme in quality control in the secretory pathway, ERp57, also known as glucose-regulated protein 58 or Grp58 (6 -13) has been reported to affect STAT3 (signal transducers and activators of transcription) signaling via interaction with STAT3 (6, 14 -17). STATs are a family of cytoplasmic proteins with Src homology 2 domains that act as signal messengers and transcription factors as a part of the Janus kinase-STAT pathway (18). Upon activation, STAT proteins become phosphorylated on a specific tyrosine residue by activated Janus kinases and subsequently dimerize and translocate to the nucleus where they bind to specific DNA sequences to regulate the expression of target genes (18,19). Among the STAT proteins, STAT3 has been implicated in transduction of the cellular signals that are involved in the regulation of cardiac growth, development, and hypertrophy (20). Although it is not clear how STAT3 activity is negatively regulated by ER resident ERp57, it has been reported that ERp57 may sequester the inactive and active STAT3 preventing its interaction with DNA and consequently activation of STAT3-dependent genes (6,12).
In this study we created ERp57-deficient mice and ERp57 Ϫ/Ϫ cell lines. ERp57 deficiency was embryonic lethal at embryonic day 13.5 (E13.5) and ERp57-deficient cells had significantly increased STAT3 activity. We showed that the ER, but not cytoplasmic, form of ERp57 is responsible for inhibition of STAT3 activity. Furthermore, ERp57-dependent modulation of STAT3 was enhanced by ER luminal interactions between ERp57 and calreticulin. Our results suggest that, in vivo, ERp57 and STAT3 may not interact and that the observed modulation of STAT3 activity may be due to ERp57-dependent signaling from the ER.

EXPERIMENTAL PROCEDURES
Generation of ERp57-deficient Mice-Gene trapping with the trap vector pGT1TMpfs was used to generate the Pdia3 gene disrupted embryonic stem cells from the Gene Trap Resource (BayGenomics, University of San Francisco, San Francisco, CA). The Pdia3 gene encodes ERp57. Parental cell lines (CGR8 and E14Tg2A) were established from delayed blastocysts on gelatinized tissue culture dishes in embryonic stem cell medium containing leukemia inhibitory factor (21). Embryonic stem cells were cultured in medium containing Glasgow minimal essential medium, 2 mM glutamine, 1 mM sodium pyruvate, non-essential amino acids, 10% fetal calf serum, 100 nM ␤-mercaptoethanol, and 1000 units of leukemia inhibitory factor (22). Embryonic stem cells were microinjected into 3.5-day-old C57BL/6J blastocysts to generate chimeric mice (23). Chimeric males were analyzed for germline transmission by mating with C57BL/6J females, and the progeny were identified by PCR analysis, ␤-galactosidase staining, and Western blot analysis. All animal experimental procedures were approved by the Animal Welfare Program at the Research Ethics Office, University of Alberta, and conformed to relevant regulatory standards.
Genotype Analysis of ERp57-deficient Mice-Genomic DNA from heterozygous mice (carrying the LacZ gene) was isolated using the DNeasy Blood and Tissue kit (Invitrogen). Genomic DNA was purified using a PCR purification kit (Qiagen), and then DNA fragments were ligated using the Rapid DNA Ligation Kit (Boehringer). First, PCR amplification was performed using primers specific to the inserted vectors: INVF1, 5Ј-GTT-CCCAACGATCAAGGCGAG-3Ј; and INVR1, 5Ј-AAGCCAT-ACCAAACGACGAGCG-3Ј. The product from the PCR was used as a template for a second PCR with specific (nested) primers: INVF2, 5Ј-TCAAGGCGAGTTACATGATCCC-3Ј, and INVR2, 5Ј-CGAGCGTGACACCACGATGC-3Ј. Products of the inverse PCR-driven amplification were analyzed by agarose gel electrophoresis, purified using a gel purification kit (Qiagen), and sequenced. Once the integration site was identified, a protocol was designed for genotyping wild-type, heterozygote, and homozygote ERp57-deficient embryos (Fig. 1). Embryos were harvested at embryonic days 10.5-15.5 followed by isolation of genomic DNA. The following PCR primers were used for genotyping wild-type embryos: forward primer (F2), 5Ј-GGACAGTTTTGAGCTGCCAT-3Ј (hybridizes to the intron within the insertion of the vector site) and reverse primer (R3), 5Ј-TCTCCATTATCATCGTACTCC-3Ј (hybridizes to intron 4 after the vector insertion site). To identify heterozygous embryos the following primers were used: forward primer (F1), 5Ј-TCAAGGCGAGTTACATGATCCC-3Ј (hybridizes to the end of the inserted vector) and reverse primer R3.
Cell Culture, Plasmid DNA, DNA Cloning, and Lentivirus Transduction-Mouse embryonic fibroblasts were isolated from day 11 wild-type and ERp57-deficient embryos, immortalized, and maintained in culture as described previously (23). When indicated, cells were treated with 1 M thapsigargin or 2.5 M brefeldin A for 16 h.
Immunohistochemistry and Electron Microscopy-For immunostaining, ERp57 Ϫ/Ϫ , ERp57 Ϫ/Ϫ -ERp57 ER , and ERp57 Ϫ/Ϫ -ERp57 cyt mouse embryonic fibroblasts were grown on glass coverslips. Cells were fixed with 3.7% paraformaldehyde in PBS for 20 min at room temperature (2). For immunostaining of ERp57, cells were permeabilized with 0.1% Triton X-100 in PBS for 20 min at room temperature and washed twice with PBS. Cells were incubated in PBS containing 1% bovine serum albumin for 30 min at room temperature and then incubated with anti-ERp57 (1:100) in PBS containing 1% bovine serum albumin for 1 h at room temperature. The secondary antibody used was rabbit-conjugated fluorescein isothiocyanate (Invitrogen) at a dilution of 1:100 in PBS containing 1% bovine serum albumin for 45 min at room temperature (30). All coverslips were costained with Alexa Fluor 546-concanavalin A at a dilution of 1:1000 (Sigma). The coverslips were mounted onto glass slides and fluorescent signals visualized using spinning disk confocal microscopy (WaveFX from Quorum Technologies, Guelph, Canada) set up on an Olympus IX-81 inverted stand (Olympus, Markham, Canada). Images were acquired through a ϫ60 objective (N.A. 1.42) with an EMCCD camera (Hamamatsu, Japan). The fluorescent dyes, fluorescein isothiocyanate and Texas Red, were excited by 491-and 561-nm laser lines, respectively (Spectral Applied Research, Richmond Hill, Canada). Z-slices (0.25 m) were acquired using Volocity (Improvision) through the cells using a piezo z-stage (Applied Scientific Instrumentation, Eugene, OR).
For electron microscopy analysis, cells were harvested, spun down, and fixed at 4°C for 4 h in a freshly prepared solution containing 2.5% glutaraldehyde and 2% paraformaldehyde in 100 mM cacodylate, pH 7.2 (31). Samples were processed for electron microscopy and examined with a Hitachi Transmission Electron Microscope H-7000.
Subcellular Fractionation-Wild-type, ERp57 Ϫ/Ϫ , and ERp57 Ϫ/Ϫ -ERp57 cyt cells harvested in a buffer containing of 250 mM sucrose, 20 mM Hepes, pH 7.0, 10 mM KCl, 1.5 mM MgCl 2 , 1 mM EDTA, 1 mM EGTA, 1 mM dithiothreitol, 200 M phenylmethylsulfonyl fluoride, 100 M benzamidine, and protease inhibitors (26). The cellular suspension was incubated on ice for 10 min, then subjected to 40 strokes with a Dounce homogenizer followed by a 20-min incubation on ice. The sample was centrifuged at 720 ϫ g for 5 min, and the supernatant containing cytoplasm, mitochondria, and membranes was centrifuged at 10,000 ϫ g for 10 min. The supernatant containing cytoplasm and membranes was then subjected to a high speed centrifuge at 100,000 ϫ g for 1 h to separate the cytoplasmic fraction from the pellet containing the ER-enriched membrane vesicles. Membrane pellet was suspended in 1 ml of RIPA buffer containing 50 mM Tris, pH 7.2, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% Triton X-100, 0.1% SDS, 0.5% sodium deoxycholate, 200 M phenylmethylsulfonyl fluoride, 100 M benzamidine, and protease inhibitors (26). Prior to SDS-PAGE proteins were precipitated with ice-cold acetone. Proteins were transferred to nitrocellulose membrane followed by Western blot analysis with rabbit anti-ERp57 antibodies at a dilution of 1:1,000.
FACS Analysis-Samples were analyzed on BD FACScan single laser flow cytometry (BD Bioscience) equipped with a 488-nm filter. Data were collected from 10,000 cells and analyzed using CellQuest software. Cells at confluence of 80 -90% were harvested by scraping into PBS, re-suspended in 100 l of 0.1% formaldehyde in PBS, and incubated for 30 min with rab-ERp57 and STAT3 Regulation FEBRUARY 26, 2010 • VOLUME 285 • NUMBER 9 bit anti-ERp57 antibodies at a concentration of 1:150. Pelleted cells were than washed 3 times with PBS containing 2% fetal bovine serum followed by addition of fluorescein isothiocyanate-labeled anti-rabbit (Alexa Fluor 488; Invitrogen) secondary antibody. Cells were washed 3 times with PBS with fetal bovine serum and once with PBS alone. Staining with secondary antibodies alone was used as a negative control. All experiments were carried out at least three times.
Luciferase Reporter Gene Assay-Generation of crt Ϫ/Ϫ , crt Ϫ/Ϫ -CRT G242A , and crt Ϫ/Ϫ -CRT E234R mouse embryonic fibroblasts was described previously (32). ERp57 Ϫ/Ϫ and ERp57 Ϫ/Ϫ -ERp57 ER cells were transfected with ER stress luciferase reporter vector (pRL-XFL, a generous gift from Dr. R. J. Kaufman (33)) as described previously (2). Twenty four hours after transfection, cells were treated with G242A , and crt Ϫ/Ϫ -CRT E234R cells were co-transfected with a luciferase reporter gene under control of the STAT3 promoter (pLucTKS3, a kind gift from Dr. J. Jung and was described previously (34,35)) with a Renilla luciferase reporter ratio of 50:1, respectively (2). After 48 h cells were harvested, lysed, and collected in Passive Lysis Buffer (Promega) followed by measurement of luciferase activity using the Dual Luciferase Assay Kit (Promega) and a Berthold-Lumat 9507 luminometer (2). Relative light units were normalized to the Renilla luciferase under the control of the cytomegalovirus promoter (2,33). Four independent transfection experiments were carried out. Statistics were performed using two-sample unpaired t tests.
Miscellaneous Procedures-Protein concentration was measured spectrophotometrically using a Bio-Rad procedure (2). Ca 2ϩ fluorescence measurements were carried out using 0.5 M Fura2-AM (22). Cells were treated with 600 nM bradykinin and 300 nM thapsigargin (36). Each trial was calibrated with 1 M ionomycin and 4 mM CaCl 2 to obtain a maximum value and with 30 mM EGTA, 25 mM Tris-HCl, pH 7.4, and 0.4% Triton X-100 to obtain a minimum value . Fluorescence was measured at ex ϭ 340 nm, ex ϭ 380 nm, and em ϭ 510 nm. Ca 2ϩ concentrations were calculated from the fluorescence values. Fig. 1A summarizes the results of the gene trapping strategy used to generate ERp57-deficient mice. The Pdia3 gene, which encodes Introns are represented as blocks. B, PCR analysis of genomic DNA isolated from wild-type (wt), heterozygote ERp57 ϩ/Ϫ , and homozygote ERp57 Ϫ/Ϫ embryos. Primers R3 and F1 were used for detection of wild-type alleles and R3 and F2 for mutated alleles. C, Western blot analysis with anti-ERp57 antibodies (upper panel). The lower protein band corresponds to a nonspecific immunoreactivity. Anti-glyceraldehyde-3-phosphate dehydrogenase (GAPDH) antibodies were used to normalize for protein loading (lower panel). D, hematoxylin and eosin staining of embryonic day 12.5 wild-type (wt) and ERp57-deficient embryos (ERp57 Ϫ/Ϫ ). ERp57, was interrupted between introns 4 and 5 (nucleotides 14,083-14,084) with the pGT1.8TM cassette containing neomycin resistance and ␤-galactosidase genes (Fig. 1A). PCR analysis of genomic DNA using the specific sets of primers depicted in Fig. 1A allowed for determination of the genotypes of mice. Analysis of wild-type mice showed only a 797-base pair (bp) DNA product corresponding to the wildtype allele (Fig. 1B, primers F2 and R3) and no DNA product with primers F1 and R3 (Fig. 1B). As expected, ERp57 Ϫ/Ϫ mice showed a 1800-bp DNA product with primers F2 and FIGURE 2. Activation of the ERp57 promoter during mouse embryogenesis. Expression of the ERp57 gene was detected as ␤-galactosidase activity from the ␤-galactosidase reporter gene inserted in the ERp57 gene within the ERp57 ϩ/Ϫ mouse (see Fig. 1). A, activation of the ERp57 promoter in mouse blastocysts. High activation of the ERp57 gene was detected in the inner cell mass (icm) of the blastocysts but not in the blastocoels (b) or trophoblasts (t). B, activation of the ERp57 promoter in lung, liver, and spine but not the heart of E13.5 ERp57 ϩ/Ϫ mouse embryos. C, high activity of the ERp57 promoter in the lung, liver, gut, and spinal cord in an E15.5 ERp57 ϩ/Ϫ mouse embryo. D, high activity of the ERp57 promoter in the brain, lungs, and liver in an E18.5 an ERp57 ϩ/Ϫ mouse embryo. E, ␤-galactosidase staining of a wild-type E15.5 control embryo. FEBRUARY 26, 2010 • VOLUME 285 • NUMBER 9

JOURNAL OF BIOLOGICAL CHEMISTRY 6729
R3 and no product with primers F1 and R3 (Fig. 1B). PCRdriven amplification of genomic DNA isolated from ERp57 ϩ/Ϫ mice produced both the 797-and 1,800-bp DNA products, indicating the presence of one copy of the wildtype allele and one copy of the targeted knock-out allele (Fig.  1B). Western blot analysis revealed that interruption of both

ERp57 and STAT3 Regulation
alleles of the ERp57 gene resulted in no detectable expression of ERp57 protein (Fig. 1C). Chimeric mice were crossed with wild-type C57BL6 females to generate the first generation of heterozygotes. ERp57 ϩ/Ϫ B6/CD-1 mice were indistinguishable from wild-type animals. ERp57 ϩ/Ϫ males were intercrossed to ERp57 ϩ/Ϫ females to generate homozygote (ERp57 Ϫ/Ϫ ) gene knock-out mice. We were unable to obtain any viable ERp57 Ϫ/Ϫ pups from this cross. Viable embryos were obtained at E13.5 or earlier ( Table  1). Analysis of embryos at or after E13.5 showed a deficiency in the number of ERp57 Ϫ/Ϫ embryos indicating that a significant fraction of ERp57 Ϫ/Ϫ embryos died earlier than E13.5. We concluded that the ERp57 gene knock-out was embryonic lethal and that ERp57 is essential for survival.
Developmental Activation of the ERp57 Gene-Histological analysis of E12.5 ERp57 Ϫ/Ϫ embryos revealed that although there were no obvious gross morphological differences between ERp57 Ϫ/Ϫ and wild-type embryos, the ERp57 Ϫ/Ϫ embryos were markedly smaller than wild-type (Fig. 1D). To gain additional insight into the potential role of ERp57 during embryonic development, we carried out ␤-galactosidase reporter gene analysis. The transgenic animals used in this study were generated by insertion of a gene cassette containing the ␤-galactosidase reporter gene (Fig. 1A). To quantify transcriptional activation of the ERp57 gene we monitored expression of the ␤-galactosidase gene to evaluate ERp57 promoter activity in blastocysts and during embryonic development. Fig. 2A shows high expression and activity of ␤-galactosidase in blastocysts 3 days post-coitum indicating high expression of ERp57 during the early stages of blastocyst formation. High activation of the ERp57 gene was confined to the inner cell mass ( Fig. 2A, icm) with relatively decreased activation of the gene in the blastocoel ( Fig. 2A, b) and trophoblast ( Fig. 2A, t). Next, ERp57 ϩ/Ϫ and wild-type embryos were harvested at different gestational stages and stained for ␤-galactosidase activity. Activation of the ERp57 gene, as reported by high activity of ␤-galactosidase, was observed as early as E13.5 (Fig.  2B). At E13.5 there was high activation in the lung and vertebra but a relatively low activity of the gene in the liver and heart (Fig.  2B). At E15.5 ␤-galactosidase reporter gene activity was the highest in the lungs with elevated activation also observed in the liver, vertebra, and intestine (Fig. 2C). However, the activity was strikingly absent in the heart, thymus, and neurological tissue including the brain (Fig. 2C). At later stages of embryonic  FEBRUARY 26, 2010 • VOLUME 285 • NUMBER 9 development (E18.5), the pattern of activation of the ERp57 gene was significantly changed. Low activity of the ERp57 promoter was maintained in the heart, thymus, and skeletal muscle and high activity was observed in the lungs and liver (Fig. 2D). At E18.5 we now observed a significant increase in activation of the ERp57 gene in the brain, which was absent in the E13.5 (Fig. 2D). These data suggest that ERp57 may play a direct role during embryonic development, specifically in lung, liver, and vertebrae, and in the later developmental stages, in the brain.

ERp57 and STAT3 Regulation
The ER in ERp57-deficient Mouse Fibroblasts-To determine the effects of ERp57 deficiency on ER functions, we isolated mouse embryonic fibroblasts from ERp57-deficient and wildtype embryos. Some of the ERp57-deficient cells were transfected with expression vectors encoding full-length, ER-targeted ERp57, or cytoplasm-targeted ERp57 with no signal sequence and were designated ERp57 Ϫ/Ϫ -ERp57 ER or ERp57 Ϫ/Ϫ -ERp57 cyt , respectively. As expected, Western blot analysis revealed that wild-type, ERp57 Ϫ/Ϫ -ERp57 ER , or ERp57 Ϫ/Ϫ -ERp57 cyt cells contained immunoreactive ERp57 (Fig. 3A). Fig. 3B shows that wild-type and ERp57 Ϫ/Ϫ -57 ER cells expressed ERp57 and that the protein was localized to an ER-like network. As expected, ERp57 was localized to the cytoplasmic compartment in ERp57-deficient cells expressing ERp57 without signal sequence (Fig. 3B, ERp57 Ϫ/Ϫ -ERp57 cyt ). Cytoplasmic localization of ERp57 without signal sequence (ERp57 cyt ) was further confirmed by biochemical fractionation of ERp57-deficient cells expressing ERp57 without signal sequence (ERp57 Ϫ/Ϫ -ERp57 cyt ) (Fig. 3C). Finally, we carried out FACS analysis of different cells lines and showed that there was a small level of immunoreactive ERp57 detected on the cell surface of cell lines used in this study (Fig. 3D).
Morphologically, the ER appeared intact in all cell lines as judged by staining with Texas Redconjugated concanavalin A or by electron microscopy analysis (Fig. 3,  B and E). Western blot analysis demonstrated that there were no significant differences in the expression of ER chaperone proteins calreticulin, calnexin, Grp78/BiP, and Grp94 in the absence of ERp57 (Fig.  3F). Next we examined the impact of the absence of ERp57 on other oxidoreductase folding enzyme family members. Again there was no significant change in expression of PDI and or the PDI-like family of protein ERp19 in the absence of ERp57 (Fig. 3F). A slight increase in the level of ERp41 and decrease in the level of ERp54 was seen in the absence of ERp57 (Fig. 3F). Thus, the absence of ERp57 had no major impact on the expression of other ER chaperones or folding enzymes.
ER Ca 2ϩ Homeostasis in the Absence of ERp57-The ER is the major Ca 2ϩ store of the cell and calreticulin-or Grp94-deficient cells have impaired Ca 2ϩ buffering and Ca 2ϩ homeostasis (36 -43). We examined, therefore, whether ERp57 deficiency had any effect on ER Ca 2ϩ homeostasis. Wild-type, ERp57 Ϫ/Ϫ , and ERp57 Ϫ/Ϫ -ERp57 ER mouse embryonic fibroblasts were stimulated with thapsigargin, an inhibitor of sarcoplasmic/endoplasmic reticulum Ca 2ϩ -ATPase-type Ca 2ϩ pumps or with bradykinin,apotentactivatoroftheinositol1,4,5-trisphosphatedependent Ca 2ϩ release channel located in the ER. When cells were stimulated with bradykinin, the peak amplitude and the duration of enhanced cytoplasmic Ca 2ϩ concentration was comparable in wild-type, ERp57 Ϫ/Ϫ , and ERp57 Ϫ/Ϫ -ERp57 ER cells (Fig. 4A). In ERp57 Ϫ/Ϫ and ERp57 Ϫ/Ϫ -ERp57 ER we observed a slight increase in Ca 2ϩ concentration in response to thapsigargin when compared with wild-type (Fig. 4B), suggesting that there might be minor changes in ER Ca 2ϩ capacity in the absence of ERp57. This might be due to an ERp57-dependent regulation of sarcoplasmic/endoplasmic reticulum Ca 2ϩ -ATPase 2b activity (44). The Unfolded Protein Response (UPR) in ERp57-deficient Cells-Next, we examined if the absence of ERp57 led to any induction of ER stress and consequently the initiation of the UPR. The UPR was examined by splicing analysis of mRNA encoding X-box-binding protein (Xbp1), which is cleaved and activated by ER stress-activated inositol-requiring enzyme-1 (2). As a control, ER stress was induced using a classical ER stress inducer, thapsigargin (2). In addition, we induced UPR/ER stress in wild-type and ERp57 Ϫ/Ϫ cells with brefeldin A (BFA), an ER-Golgi trafficking inhibitor (45).
RT-PCR analysis was carried out to test for expression of Grp78/BiP mRNA and splicing of Xbp1, both markers of UPR. Non-stimulated wild-type and ERp57 Ϫ/Ϫ cells did not have an increased level of Grp78/BiP mRNA nor was there any observable splicing of Xbp1 mRNA (Fig. 5A). As expected, thapsigargin induced complete splicing of Xbp1 (Fig. 5A, Xbp1s) as well as a significant increase in the level of Grp78/BiP mRNA (Fig.  5A). When wild-type and ERp57 Ϫ/Ϫ cells were treated with BFA, there was incomplete splicing of Xbp1 with both unspliced (Fig. 5A, Xbp1 us) and spliced (Fig. 5A, Xbp1s) forms of Xbp1 mRNA detected. Interestingly, both the level of Grp78/ BiP mRNA and the spliced form of Xbp1 were increased in ERp57 Ϫ/Ϫ compared with wild-type cells (Fig. 5A), indicating that ERp57 Ϫ/Ϫ cells were more sensitive to BFA-induced ER stress.
Next, we quantified Xbp1 splicing using a luciferase reporter system developed by Kaufman et al. (33). In this system, only the spliced form of the Xbp1 is in-frame with the luciferase reporter gene, therefore, high luciferase activity reports splicing of Xbp1 and increased ER stress (33). Fig. 5B shows that there was a small but significant increase in the UPR in ERp57 Ϫ/Ϫ cells when compared with wild-type cells (Fig. 5B) and this was not rescued by expression of the recombinant ERp57 (Fig. 5B). Thapsigargin treatment significantly enhanced luciferase activity in wild-type, ERp57 Ϫ/Ϫ , and ERp57 Ϫ/Ϫ -ERp57 ER cells (Fig.  5B). There was an 8-fold increase in UPR activation upon treatment with thapsigargin when compared with non-stimulated counterparts (Fig. 5B). Although there was a significant increase in ER stress in the absence of ERp57, this did not result in increased apoptosis in ERp57 Ϫ/Ϫ cells (data not shown) indicating that ERp57-deficient cells experienced a tolerable level

. Endoplasmic reticulum Ca 2؉ homeostasis in ERp57-deficient cells and cells expressing ER targeted ERp57.
Cytoplasmic Ca 2ϩ concentration was measured in the wild-type and ERp57-deficient cells as described under "Experimental Procedures." A, Ca 2ϩ released by bradykinin. In the absence of ERp57 (white bar) there was a decrease in bradykinin Ca 2ϩ release but this not significantly different from wild-type (black bar). Bradykinin release is decreased in cells supplied with ER localized ERp57 (ERp57 Ϫ/Ϫ -ERp57 ER ; gray bar) but this was not significantly different from wild-type cells. Data are presented as the mean Ϯ S.D., n ϭ 4. B, Ca 2ϩ released by thapsigargin. In the absence of ERp57 (ERp57 Ϫ/Ϫ , white bar) and ERp57 Ϫ/Ϫ cells transfected with expression vector encoding ERp57 (ERp57 Ϫ/Ϫ -ERp57 ER ; gray bar), there was significantly more Ca 2ϩ release. Data are presented as the mean Ϯ S.D., n ϭ 4. Two-sample, unpaired t test was performed. #, p ϭ 0.01 versus wild-type and *, p ϭ 0.0065 versus wild-type.  (2). Spliced (Xbp1s) and unspliced (Xbp1 us) forms of Xbp are indicated with arrows. RT-PCR of GAPDH mRNA was carried out to normalize for loading of agarose gels. B, quantitative analysis of splicing of the Xbp1 mRNA in wild-type and ERp57-deficient cells (ERp57 Ϫ/Ϫ ). Cells were transfected with DNA plasmid reporting Xbp1 splicing. ER stress was induced with 1 M thapsigargin (TG) or 50 M H 2 O 2 . Renilla luciferase and firefly luciferase activities were measured and the relative ratio of firefly luciferase to Renilla luciferase activity in each cell lysate reported. Data are presented as the mean Ϯ S.D., n ϭ 9. RLU, relative light units. Two-sample, unpaired t test was performed. #, p ϭ 0.01 versus wild-type and *, p ϭ 0.02 versus wild-type. FEBRUARY 26, 2010 • VOLUME 285 • NUMBER 9 JOURNAL OF BIOLOGICAL CHEMISTRY 6733 of ER stress (46). ERp57 is one of the targets of oxidative stress induced by H 2 O 2 (47), therefore, we also used H 2 O 2 to induce ER stress in wild-type and ERp57 Ϫ/Ϫ cells. H 2 O 2 was not a very potent inducer of ER stress as there was only a slight increase in ER stress in cells treated with H 2 O 2 (Fig. 5C). We concluded that ERp57 deficiency did not have a significant effect on ER morphology; expression of ER-associated chaperones and folding enzymes, ER stress, or apoptosis.

ERp57 and STAT3 Regulation
cally targeted protein (Fig. 6C, ERp57 Ϫ/Ϫ -ERp57 cyt ). As expected, expression of the recombinant GFP in ERp57-deficient cells did not have any effect on STAT3 transcriptional activity. Fig. 3D shows that there was a detectable level of immunoreactive ERp57 at the cell surface. However, we do not believe that this small fraction of ERp57 contributes to the modulation of STAT3 transcriptional activity because addition of purified ERp57 to cell cultures did not affect STAT3 transcriptional activity in the ERp57-deficient cells (supplemental Fig. S1). We concluded that ER-targeted ERp57 was the most effective in influencing STAT3 signaling, suggesting that ERp57 modulates STAT3 activity from the lumen of the ER.
Calreticulin Enhances ERp57 Modulation of STAT3 Signaling-ERp57 forms functional complexes with calreticulin to assist in folding and post-translational modification of newly synthesized proteins (3). We asked whether calreticulin may influence ERp57-dependent STAT3 activity. To do this we employed crt Ϫ/Ϫ cell lines expressing wildtype calreticulin or calreticulin mutants defective in binding ERp57 (32). In this study, we used a loss of ERp57 binding calreticulin-E239R mutant and a calreticulin-G242A mutant that has no loss in ERp57 binding (32). Western blot analysis showed that STAT3 or phospho-STAT3 (Tyr 706 ) expression was not affected by expression of calreticulin or calreticulin mutants in crt Ϫ/Ϫ cells (Fig. 7). Surprisingly, quantitative analysis of STAT3 activity using the luciferase reporter gene system showed high activity of STAT3 in crt Ϫ/Ϫ cells compared with wildtype counterparts (Fig. 7B), indicating that calreticulin may also affect STAT3 signaling. Fig. 7B shows that STAT3 was not affected in the absence of calnexin, a homologue of calreticulin. Expression of wild-type calreticulin (Fig. 7B, crt Ϫ/Ϫ -CRT) or a calreticulin mutant interacting with ERp57 (Fig. 7B, crt Ϫ/Ϫ -CRT G242A ) resulted in full recovery of STAT3 activity back to the level observed in wild-type cells (Fig. 7B, wt). In contrast, expression of the calreticulin mutant with a loss of ERp57 binding had no effect on STAT3 activity (Fig. 7B, crt Ϫ/Ϫ -CRT E239R ), which remained high and at the same level as observed in calreticulin-deficient cells (Fig. 7B, crt Ϫ/Ϫ ). Taken together, these results indicate that ERp57-dependent modulation of STAT3 from the lumen of the ER requires calreticulin capable of forming complexes with ERp57.

DISCUSSION
In this study, we showed that deletion of the Pdia3 gene, which encodes ERp57, was embryonic lethal at E13.5. Biochemical and cell biological analysis of ERp57-deficient cells indicated that the absence of ERp57 has no impact on ER morphology, expression of ER-associated chaperones and folding enzymes,ERstress,orapoptosis.However,weshowthatSTAT3dependent signaling was increased in the absence of ERp57 and  FEBRUARY 26, 2010 • VOLUME 285 • NUMBER 9 this could be rescued by expression of the ER-targeted ERp57, indicating that ERp57 affects STAT3 signaling from the lumen of the ER. This effect of ERp57 on the STAT3 pathway is further enhanced by an ER luminal complex formation between ERp57 and calreticulin. Increased STAT3 activity in the absence of ERp57 may contribute to embryonic lethality of ERp57 Ϫ/Ϫ mice.

ERp57 and STAT3 Regulation
The ERp57 gene is differentially regulated during embryonic development indicating that the protein may play a direct role in development of specific tissues. This is in line with earlier observations of the expression of ER-associated ERp57 partner proteins, calreticulin and calnexin (2,23). Calreticulin is predominately expressed in the heart during early stages of embryonic development and calreticulin-deficient mice die at E14.5 from impaired cardiac development (23). In contrast the calnexin gene is highly activated in neuronal tissue (2) and calnexin-deficient mice are born with neurological abnormalities (49). In this study, we showed that the ERp57 gene was highly active in the inner cell mass of blastocysts, an origin of fetus formation (50) and in the lung, liver, and brain during early stages of embryogenesis, suggesting that impaired development of these tissues may, at least in part, contribute to ERp57 Ϫ/Ϫ embryonic lethality. The high expression of ERp57 in the lungs seen in this study is of interest because it has recently been reported that ERp57 is decreased in a neonatal rat model of hyperoxia-induced lung injury and cultured cells (47). Small interfering RNA knockdown of ERp57 in human endothelial cells resulted in cellular protection against hyperoxia and tunicamycin-induced caspase-3 activation apoptosis, suggesting that ERp57 has a role in the regulation of apoptosis (47).
Garbi et al. (4) carried out a targeted deletion of the ERp57 gene in mouse B cells. They also reported no ERp57 Ϫ/Ϫ offspring, indicating that ERp57 deficiency might be embryonic lethal. Here we investigated embryonic development in the absence of ERp57 and showed that ERp57 Ϫ/Ϫ was embryonic lethal at E13.5. Interestingly, deletion of ERp57 in B cells in mice results in normal B cell development, proliferation, and antibody production indicating that ERp57 is not required for glycoprotein folding in B cells (4). However, there is aberrant assembly of the peptide loading complex, showing that ERp57 is involved in the assembly of the peptide loading complex and contributes both qualitatively and quantitatively to MHC class I antigen presentation in vivo. This supports earlier observations that ERp57 is critically involved in the early folding events of MHC class I heavy chain (5,14). The role of ERp57 in MHC class I biogenesis and assembly is further supported by ERp57 small interfering RNA studies (5,51,52).
In addition to its role as a folding enzyme in quality control in the secretory pathway, ERp57 is reported to affect STAT3 signaling (12). This may be due to formation of inhibitory complexes between the ERp57 and STAT3 protein either in the cytoplasm or nucleus (6,15,48,53,54). Furthermore, in avian and mammalian cells, DNA-protein cross-linking experiments indicate that ERp57 is nuclear and interacts directly with DNA (15). In vitro experiments in HeLa cells showed that ERp57 preferentially binds DNA sequences having characteristic scaffold/matrix-associated regions (55). Finally, the protein has also been suggested to be present on the cell surface and in the cytoplasm (6, 14 -17). Using FACS analysis we detected some immunoreactive ERp57 on the surface of cells used in this study. However, it is unlikely that cell surface ERp57 affects STAT3 transcriptional activity as addition of extracellular ERp57 did not affect STAT3 function (supplemental Fig. S1). The conundrum then is: how does ERp57 gain access to the cytoplasm/nucleus to influence STAT3 activity when its N-terminal signal sequence should direct it to the ER and its C-terminal QDEL ER retrieval motif should keep it there? To address this question, we took advantage of available ERp57-deficient cells and tested, for the first time, STAT3 signaling in the absence of ERp57. Analysis of STAT3-dependent expression of the luciferase reporter gene in ERp57-deficient cells revealed that ERp57 indeed affects STAT3 signaling. Most importantly, we showed that the ER but not the cytoplasmic form of ERp57 was effective in inhibition of STAT3 activity. This effect is specific to ERp57, because transfection of ER-targeted GFP or calnexin did not have any effect on the STAT3 activity. Fig. 8 shows a model of ERp57 function in the lumen of the ER. ERp57 is an ER resident folding enzyme involved in oxidative folding of glycoproteins in the ER, biosynthesis of the MHC class I, and as a component of the loading complex (14). In this study, we showed that ERp57 also affects STAT3 signaling from the lumen of the ER (Fig. 8). It is not clear at present how ERp57 affected STAT3 signaling, but our work indicates that the inhibitory function of ERp57 was significantly enhanced by interaction with another ER resident luminal protein, calreticulin (Fig. 8). The two proteins form functional complexes involved in folding and post-translational modification of FIGURE 8. A model of the relationship between STAT3, ERp57, and calreticulin. This model shows cross-talk between ERp57, calreticulin, and STAT3 signaling. ERp57 forms functional complexes with calnexin and calreticulin to promote protein folding, disulfide bond formation, and isomerization. ERp57 is also critical for MHC class I biosynthesis and assembly. A model shows that ERp57, from the lumen of the ER, also affects STAT3 signaling and functions as a STAT3 inhibitor. ERp57-dependent modulation of STAT3 is enhanced by complex formation between ERp57 and calreticulin.
newly synthesized (glyco)proteins (42). Our results indicate that ERp57-dependent modulation of STAT3 activity also depends on the formation of ERp57-calreticulin complexes in the lumen of the ER. Interestingly, ERp57-calreticulin complex formation might also be important for cell surface targeting of these ER proteins (56).
The molecular mechanisms of ERp57-dependent signaling from the ER are not presently clear, however, several other cellular processes are controlled by "ER signaling." For example, cholesterol homeostasis is controlled by ER-nuclear signaling via sterol-regulated proteolysis of ER membrane-bound transcription factors called sterol regulatory element-binding proteins (57). The UPR is also regulated by ER-associated kinases, which transduce their signals from the ER to affect gene expression in the nucleus (58). One critical aspect of Ca 2ϩ homeostasis is involved in store-operated Ca 2ϩ influx resulting from ERdependent activation of the plasma membrane Ca 2ϩ channel (59). Transcriptional activity of steroid receptors is also modulated by ER luminal proteins (60). Here we report, for the first time, that the STAT3-dependent pathway may also be modulated by signaling from the ER.