Calreticulin Regulates Transforming Growth Factor-β-stimulated Extracellular Matrix Production*

Background: Endoplasmic reticulum (ER) stress is associated with fibrotic diseases, although the mechanisms are not completely understood. Results: The ER stress protein calreticulin regulates TGF-β stimulated extracellular matrix through control of intracellular calcium and NFAT signaling. Conclusion: Calreticulin is necessary for TGF-β stimulated extracellular matrix production. Significance: These findings identify calreticulin as a mechanistic link between ER stress and fibrosis. Endoplasmic reticulum (ER) stress is an emerging factor in fibrotic disease, although precise mechanisms are not clear. Calreticulin (CRT) is an ER chaperone and regulator of Ca2+ signaling up-regulated by ER stress and in fibrotic tissues. Previously, we showed that ER CRT regulates type I collagen transcript, trafficking, secretion, and processing into the extracellular matrix (ECM). To determine the role of CRT in ECM regulation under fibrotic conditions, we asked whether CRT modified cellular responses to the pro-fibrotic cytokine, TGF-β. These studies show that CRT−/− mouse embryonic fibroblasts (MEFs) and rat and human idiopathic pulmonary fibrosis lung fibroblasts with siRNA CRT knockdown had impaired TGF-β stimulation of type I collagen and fibronectin. In contrast, fibroblasts with increased CRT expression had enhanced responses to TGF-β. The lack of CRT does not impact canonical TGF-β signaling as TGF-β was able to stimulate Smad reporter activity in CRT−/− MEFs. CRT regulation of TGF-β-stimulated Ca2+ signaling is important for induction of ECM. CRT−/− MEFs failed to increase intracellular Ca2+ levels in response to TGF-β. NFAT activity is required for ECM stimulation by TGF-β. In CRT−/− MEFs, TGF-β stimulation of NFAT nuclear translocation and reporter activity is impaired. Importantly, CRT is required for TGF-β stimulation of ECM under conditions of ER stress, as tunicamycin-induced ER stress was insufficient to induce ECM production in TGF-β stimulated CRT−/− MEFs. Together, these data identify CRT-regulated Ca2+-dependent pathways as a critical molecular link between ER stress and TGF-β fibrotic signaling.

Endoplasmic reticulum (ER) 3 stress is emerging as a factor in fibroproliferative diseases including pulmonary fibrosis, dia-betic nephropathy, hypertension-associated cardiac fibrosis, and atherosclerosis (1)(2)(3)(4)(5)(6)(7)(8). ER stress can be induced by glucose, glucosamine, and oxidative stress, factors known to induce fibroproliferative remodeling in multiple tissues (5,9,10). Chemical chaperones that reduce ER stress, such as valproate and 4-phenylbutyrate, reduce atherosclerosis and fibrosis (3,(11)(12)(13) and also inhibit TGF-␤-induced type I collagen production independent of Smad reporter activity (14). In addition, induction of ER stress by tunicamycin exacerbates lung fibrosis in the bleomycin model of pulmonary fibrosis, although tunicamycin alone did not induce fibrosis (1). Enhanced ER stress in alveolar epithelial cells facilitates epithelial to mesenchymal transition, a process that occurs in some forms of fibrosis (15). Despite growing evidence for ER stress as factor in fibrosis, the mechanisms by which ER stress predisposes to or exacerbates fibrosis are not clear. In the lung, ER stress-induced alveolar epithelial cell apoptosis is thought to be a significant factor in the development of fibrosis (1,6,15). However, ER stress is also associated with fibroproliferative remodeling in tissues such as the diabetic vasculature where apoptosis is not a significant initiating factor (3,4). This suggests that ER stress can drive pathways that promote fibrosis through additional mechanisms.
Calreticulin (CRT) is a 46-kDa ER protein that regulates cellular responses to stress through its roles in the unfolded protein response and its chaperone activity (16,17). In addition, CRT also is important in ER Ca 2ϩ buffering and regulation of downstream Ca 2ϩ -dependent signaling pathways such as calcineurin and NFAT (nuclear factor of activated T cells) (17). CalreticulinϪ/Ϫ mouse embryonic fibroblasts (MEFs) have decreased ER Ca 2ϩ stores and impaired agonist-induced Ca 2ϩ release from the ER, whereas cells overexpressing CRT have enhanced Ca 2ϩ binding depots within thapsigargin-sensitive ER stores (18,19). Impaired ER Ca 2ϩ release in the absence of CRT leads to defects in downstream Ca 2ϩ /calcineurin signaling with reduced NFAT and MEF2C (myocyte enhancer factor 2c) nuclear translocation (20,21). CRTϪ/Ϫ mice can be rescued from embryonic lethality by constitutively active calcineurin, which induces myocyte enhancer factor 2C and NFAT translocation to the nucleus providing evidence that CRT is an upstream modulator of calcineurin signaling (20,22,23).
Our laboratory recently demonstrated that CRT regulates transcription of multiple extracellular matrix (ECM) proteins in a Ca 2ϩ -dependent manner and that it has post-transcriptional effects on collagen trafficking and matrix assembly (24). CRT regulation of fibronectin is also thought to involve ER Ca 2ϩ (25,26). Interestingly, CRT expression is increased in multiple models of fibrosis including bleomycin-induced pulmonary fibrosis, the UUO (unilateral ureteral obstruction) model of renal fibrosis, and in chronic diseases of fibroproliferative remodeling, such as atherosclerosis (2)(3)(4). Cardiac-specific overexpression of CRT during development results in interstitial fibrosis, although the mechanisms have not been defined (27). CRT expression is up-regulated by factors that are known to induce both ER stress and fibrosis, including glucose, oxidative stress, cigarette smoke, hypoxia, and TGF-␤ (2, 3, 28 -30).
Given our previous findings that CRT regulates fibronectin and type I collagen transcription and the known role of CRT in modulating calcineurin-NFAT activity, we asked whether CRT might play a role in regulating TGF-␤ stimulation of ECM proteins (20,23,24). These studies show that CRT is required for cellular responsiveness to TGF-␤. CRT mediates TGF-␤ responsiveness through regulation of TGF-␤-stimulated Ca 2ϩ release and NFAT activity. Furthermore, CRT is required for TGF-␤ stimulation of ECM in tunicamycin-treated cells. Together, these data provide evidence that CRT is a critical regulator of TGF-␤-mediated ECM production and establish a new mechanism by which ER stress contributes to fibrosis.
Cells-Wild type MEFs, CRTϪ/Ϫ MEFs, calreticulinϪ/Ϫ MEFs stably transfected with the pcDNA3 expression vector to express rabbit HA-tagged CRT were gifts from Dr. Marek Michalak (University of Alberta, Edmonton, Alberta, Canada). CalreticulinϪ/Ϫ MEFs stably transfected with HA-tagged CRT lacking the TSP1 binding domain (amino acids 19 -36) were generated as described previously (44). Mouse L fibroblasts (parental cells and CRT overexpressors) were provided by Dr. Michal Opas (University of Toronto). These cell lines were engineered to overexpress (CRT overexpressors) CRT by 1.6fold compared with parental cells as described previously (45). Rat lung fibroblasts (RFL6) stably expressing an empty vector (pcDNA3.1, Invitrogen) were generated as previously described (gift of Dr. James Hagood) (46). Human IPF lung fibroblasts were provided by Dr. Victor Thannickal (University of Alabama at Birmingham). L fibroblasts were maintained in high (4.5 g/liter) glucose DMEM with 10% FBS in the presence of 100 g/ml G418 sulfate (Cellgro).
Quantitative Real Time PCR-Cells were grown overnight in complete media containing 10% FBS, starved in low (0.5%) serum medium overnight, and treated with TGF-␤ or other compounds. After treatment, RNA was harvested with TRIzol reagent and isolated according to the manufacturer's specifications. Quantitative real time PCR was performed using standard protocols with an Opticon instrument (MJ Research, model CFD-3200). Primers for mouse type I collagen (Col1A1) (catalog #PPM03845F-200), fibronectin (catalog #QT00135758), CRT (catalog #QT00101206), and S9 (catalog #PPM03695A) were obtained from Qiagen and verified by melting curve analysis. Transcript levels were assayed using SYBR Green from Qiagen. Results were calculated using the ⌬⌬CT method and are expressed as the mean Ϯ S.D. of three samples each assayed in triplicate as indicated in Fig. 1, 3, 4, 8, and 9 legends. Results are representative of at least two-three separate experiments.
Immunoblotting for ECM Proteins-After treatment, cells were harvested using 1ϫ Laemmli lysis buffer (Bio-Rad) containing 1ϫ protease inhibitor mixture (Sigma catalog #p8340). After lysis, cells were sonicated for 7 s, 5% final ␤-mercaptoethanol was added, and samples were boiled at 100°C for 7 min. Samples were centrifuged, and equal volumes were loaded onto 4 -15 or 10% SDS-polyacrylamide gels. After separation by SDS-PAGE, samples were transferred onto a PVDF membrane at 100 V for 100 min. After transfer, membranes were blocked with 1% casein followed by application of the primary antibody. Membranes were washed in TBS-Tween, and secondary antibody was applied for 1 h at room temperature. Membranes were washed with TBS-Tween and developed using Western Lightning Chemiluminescence Reagent Plus (PerkinElmer Life Sciences). Membranes were stripped and reprobed with rabbit anti-␤-tubulin or rabbit anti-␤-actin IgG to normalize for cell protein. Densitometric analysis of immunoblots was performed using the NIH Image J program. Data are expressed as the mean band density normalized for cell protein from at least three separate experiments.
Deoxycholate Extraction of the Extracellular Matrix Fraction-Deoxycholate (DOC) extractions of detergent-soluble and -insoluble fractions were performed similarly to previous reports (47). Briefly, wild type and CRTϪ/Ϫ MEFs were plated in full serum (10%) DMEM for 24 h, switched to low serum (0.5%) DMEM with 20 M ascorbic acid, and treated with 100 pM TGF-␤ for 24 h. After 24 h, wells were rinsed with PBS and harvested by scraping with 300 l of 4% DOC (4% DOC in 20 mM Tris-HCL, pH 8.8, with 1ϫ protease inhibitor). Lysates were homogenized with a 27 1 ⁄ 2-gauge needle and tumbled overnight at 4°C. Precipitates containing the DOC-insoluble portion were pelleted at 13,500 ϫ g and washed 3 times with 4% DOC solution. The supernatant containing the DOC soluble fraction was removed, placed into a new tube, and centrifuged twice as above. The DOC-insoluble pellet was resuspended in 30 l of Laemmli buffer.
Soluble Collagen Assays-Wild type and CRTϪ/Ϫ MEFs were cultured for 48 h in DMEM with 10% FBS, switched to DMEM with 0.5% FBS and 20 M ascorbic acid, and treated with or without 10 pM TGF-␤ for 72 h. Cells were dosed daily with TGF-␤ and ascorbic acid in 0.5% FBS. Conditioned medium was collected in the presence of protease inhibitor mixture (Sigma) and centrifuged at 15,000 ϫ g for 5 min to remove cellular debris. Soluble collagen in the medium was measured using the Sircol assay (Biocolor) as described by the manufacturer.
siRNA Transfection of Thy-1 (Ϫ) Rat Lung Fibroblasts and Human IPF Lung Fibroblasts-Non-targeting siRNA (SI03650325) or rat CRT siRNA (SI04449004) were purchased from Qiagen and resuspended in RNase-free water. One million Thy-1 (Ϫ) rat lung fibroblasts were transfected via nucleofection using the MEF1 Nucleofector kit from Amaxa Biosys-tems (Amaxa GmbH, Lonza) in an Amaxa Nucleofector II using program A-023. Transfected cells were cultured in DMEM with 10% FBS for 24 h. Medium was switched to low serum (0.5% FBS) medium for 2 h followed by treatment with 100 pM TGF-␤ for 24 h. Cells were washed with D-PBS (Cellgro) and lysed with 1ϫ Laemmli buffer containing 1ϫ protease inhibitor.
Human IPF lung fibroblasts were transfected with non-targeting siRNA or human CRT siRNA (SI02654589) resuspended in RNase-free water. One million human IPF lung fibroblasts were transfected via nucleofection using the primary fibroblasts nucleofector kit from Amaxa Biosystems (Amaxa) in an Amaxa Nucleofector II using program A-023. Transfected cells were cultured in DMEM with 10% FBS for 48 h. Medium was switched to low serum (0.5% FBS) DMEM for 6 h followed by treatment with 2 M ascorbic acid and 100 pM TGF-␤ for 24 h. Cells were washed with D-PBS and lysed with 1ϫ Laemmli buffer containing 1ϫ protease inhibitor.
Tunicamycin Experiment-Wild type and CRTϪ/Ϫ MEFs were plated with or without 0.01 g/ml tunicamycin in DMEM with 10% FBS containing 20 M ascorbic acid for 24 h as previously described (24). Cells were treated with or without 100 pM TGF-␤ for 24 h. Cells were washed with D-PBS and lysed with 1ϫ Laemmli lysis buffer containing 1ϫ protease inhibitor.
Immunofluorescence-Wild type and CRTϪ/Ϫ MEFs were plated on glass coverslips in a 24-well plate in DMEM with 10% FBS for 24 h and then switched to low serum (0.5% FBS) medium overnight. The next day cells were treated with TGF-␤, fixed with 4% paraformaldehyde for 10 min, and permeabilized with 0.1% Triton X for 3 min. Cells were washed with D-PBS and blocked for 1 h with filtered, sterile 1% casein. Primary antibody was added as follows: rabbit anti-phospho-Smad 2 or 3 at a 1:150 dilution; mouse anti-NFATc3 at a 1:100 dilution in 1% casein solution overnight at 4°C. Cells were washed with PBS, and the appropriate secondary Alex-aFluor488 antibody (1:500 dilution) was added for 1 h at room temperature. After washing, cells were incubated with 4 g/ml Hoechst for 5 min. Images were obtained using a Nikon Eclipse TE2000-U inverted microscope equipped for epifluorescence with a Nikon camera or a Zeiss LSM 710 confocal microscope. Non-immune IgG and secondary antibody alone were used as negative controls. Images in a particular experiment were obtained using a uniform exposure time, and images were adjusted uniformly.
Ca 2ϩ Release Assay-Cells were plated in 24-well plates in complete (10% FBS) medium overnight. The next day low serum (0.5%) FBS medium containing 5 M Fluo-4 AM was added to the cells. Dye was loaded into the cells for 20 min at 37°C. The plate was allowed to equilibrate at 37°C for 5 min. After equilibration, cells were treated with either TGF-␤ (100 pM) or ionomycin (1 M) as indicated. Cells were excited at 485 nm, and emission was measured at 520 nm every 10 s for 30 min.
Reporter Assays-Wild type and CRTϪ/Ϫ MEFs were transfected with 2 g of NFAT or Smad firefly luciferase reporter constructs with control renilla luciferase purchased from SABiosciences. Cells were transfected using the MEF 1 Nucleofector kit from Amaxa Biosystems in Amaxa Nucleofector II using program A-023. Transfected cells were cultured in com-plete (10% FBS) medium overnight, and GFP expression was confirmed the next morning. Cells were starved in low serum (0.5% FBS) medium and treated as indicated. After treatment, cells were lysed with 1ϫ lysis buffer (Promega), and firefly and renilla luciferase activity were measured using a Dual Glo Luciferase kit from Promega according to the manufacturer's instructions. Luciferase reporter activity is normalized to the renilla luciferase control. Luciferase reporter construct data are representative of at least three individual experiments each performed in triplicate.
Statistics-Data were analyzed for statistical significance using one-way analysis of variance with Holm-Sidak post hoc analyses (Sigma Stat). p Ͻ 0.05 was considered significant.

Calreticulin Is Required for TGF-␤-mediated Stimulation of Collagen and Fibronectin Transcript and Protein-Wild type
and CRTϪ/Ϫ MEFs were treated with TGF-␤, and levels of transcript were compared by RTQ-PCR. TGF-␤ stimulated a significant increase in fibronectin and COL1A1 transcript in wild type MEFs (Fig. 1, A and B). In contrast, TGF-␤ failed to stimulate an increase in either COL1A1 or fibronectin transcript in MEFs lacking CRT (Fig. 1, A and B). Levels of COL1A1 and fibronectin transcript were significantly increased 4 h post stimulation and were maintained for up to 24 h in wild type cells (Fig. 1, C and D). The lack of response in CRTϪ/Ϫ MEFs is not due to a delay in response to TGF-␤, as no increase was observed over a 24-h period with COL1A1 or 12 h for fibronectin ( Fig. 1, C and D).
The failure of TGF-␤ to stimulate type I collagen and fibronectin transcription in CRTϪ/Ϫ MEFs correlates with a lack of protein stimulation as measured in cell lysates 24 h after TGF-␤ treatment ( Fig. 2A). In addition, the increased ECM expression in TGF-␤-treated wild type cells is due to increased protein synthesis and not incorporation of serum fibronectin into the extracellular matrix, as differential extraction of the DOC cell soluble from DOC insoluble extracellular matrix (48) shows increased fibronectin in the cellular fraction after TGF-␤ treatment (Fig. 2B). Similarly, TGF-␤ failed to stimulate an increase in secreted soluble collagen in the conditioned medium of CRTϪ/Ϫ MEFs (Fig. 2C). In these studies both wild type and CRTϪ/Ϫ MEFs were stimulated with TGF-␤ in the presence of 20 M ascorbic acid. We showed previously that CRT is a collagen chaperone and that CRTϪ/Ϫ MEFs have reduced ER to Golgi trafficking and secretion of collagen, which is corrected by ascorbic acid (24). Ascorbic acid increases collagen transcript stability and is a cofactor for proline hydroxylation of procollagen, which enhances translation and secretion efficiency (49,50).
To test whether re-expression of CRT in the CRTϪ/Ϫ cells can rescue responsiveness to TGF-␤, CRTϪ/Ϫ MEFs stably transfected with HA-tagged CRT were stimulated with TGF-␤ and transcript measured 4 h post stimulation (44). Similar to wild type cells, TGF-␤ induced a significant increase in COL1A1 and fibronectin transcript production in CRTϪ/Ϫ MEFs stably transfected with HA-tagged CRT (Fig. 3A).
Because cell surface CRT can act as a receptor for the matricellular protein thrombospondin 1 (TSP1) and stimulate collagen production, we examined whether TSP1 binding to cell surface CRT might be involved in regulating cellular responsiveness to TGF-␤. TSP1 binding to cell surface CRT does not appear to be important for cellular responsiveness to TGF-␤, as CRTϪ/Ϫ MEFs stably expressing CRT lacking the TSP1 binding site were able to respond to TGF-␤ (Fig. 3B) (44,47).
Knockdown of CRT in Fibrogenic Lung Fibroblasts Inhibits the Ability of TGF-␤ to Stimulate ECM-We next determined whether CRT expression is important for TGF-␤-induced ECM production in fibroblasts known to be highly responsive to TGF-␤. Thy-1 (Ϫ) lung fibroblasts predominate in the fibrotic foci of lungs with idiopathic pulmonary fibrosis, and rat thy-1 (Ϫ) fibroblasts have robust production of ECM in response to TGF-␤ (46, 51, 52). SiRNA knockdown of CRT to 60% of control levels in the Thy-1 (Ϫ) rat lung fibroblasts blocked the ability of TGF-␤ to stimulate type I collagen and fibronectin protein (Fig. 4, A-D). Similar results were obtained by siRNA knockdown of human CRT in lung fibroblasts isolated from IPF patients (Fig. 4E). In these studies, knockdown of CRT to 35% of control levels reduced base-line ECM levels and attenuated TGF-␤ stimulated collagen I and fibronectin protein as compared to cells transfected with non-targeting siRNA.
TGF-␤ Induces a Greater Stimulation of ECM in Cells Overexpressing CRT-CRT and other ER stress response proteins are increased in several models of fibrosis, including bleomy-   cin-induced lung fibrosis and unilateral ureteral obstruction renal fibrosis (2). Therefore, we asked whether overexpression of CRT correlates with enhanced ECM production in response to TGF-␤. L-fibroblasts overexpressing CRT (ϳ1.6-fold increase in CRT) have increased collagen I and fibronectin transcript as compared with parental L-fibroblasts at base line (24). TGF-␤ treatment induced a greater increase in fibronectin and collagen I protein as compared with TGF-␤-stimulated parental cells (Fig. 5, A-C).
ER Stress in the Absence of CRT Is Not Sufficient to Stimulate ECM Production in Response to TGF-␤-Increased ER stress exacerbates response to fibrogenic stimuli in vivo and in vitro (1,14,53). Because our data indicate that CRT is required for TGF-␤-stimulated ECM production, we asked whether increased ER stress in the absence of CRT is sufficient to stimulate ECM production or whether CRT is a critical component of ER stress-induced ECM production. Wild type and CRTϪ/Ϫ MEFs were treated with TGF-␤ in the presence or absence of the ER stress inducer tunicamycin. In the presence of tunicamycin, both wild type and CRTϪ/Ϫ MEFs showed increased levels of the ER stress response protein GRP78, although tunicamycin increased GRP78 to a greater extent in CRTϪ/Ϫ MEFs as compared with wild type cells (Fig. 6, A and B). In the presence and absence of tunicamycin, wild type MEFs increased collagen I levels when treated with TGF-␤, although TGF-␤ stimulation of collagen I was not enhanced in tunicamycin-treated cells as compared with cells treated with TGF-␤ alone (Fig. 6, A and C). However, enhanced ER stress due to tunicamycin treatment was not able to overcome the failure of CRTϪ/Ϫ MEFs to stimulate ECM in response to TGF-␤ (Fig. 6,  A and C). Interestingly, tunicamycin did not increase CRT expression in wild type cells (data not shown). These data show that tunicamycin-induced ER stress is not sufficient for induction of ECM by TGF-␤ in the absence of CRT.
TGF-␤-dependent Smad Signaling Is Active in Wild Type and CRTϪ/Ϫ MEFs-TGF-␤ signals ECM production primarily through Smad-dependent pathways, although other pathways can mediate TGF-␤ signaling (33,36,54). To determine if Smad signaling is impaired in the CRTϪ/Ϫ MEFs, we examined wild type and CRTϪ/Ϫ MEFS for Smad 3 phosphorylation after TGF-␤ stimulation. Phosphorylated Smad 3 was detected in both TGF-␤ treated wild type and CRTϪ/Ϫ MEFs, suggesting that TGF-␤ engagement of its signaling receptors and receptor Smad phosphorylation are not defective in CRTϪ/Ϫ MEFs (Fig. 7, A). Because CRTϪ/Ϫ MEFs are defective in their ability to induce MEFC2 nuclear translocation (21), we asked whether phosphorylated Smad 2/3 can be imported into the nucleus in CRTϪ/Ϫ MEFs treated with TGF-␤. Immunofluorescence staining of TGF-␤ stimulated wild type and CRTϪ/Ϫ cells showed similar nuclear translocation of phosphorylated Smad 3 (Fig. 7B). Smad 2 phosphorylation and nuclear translocation were also similarly stimulated by TGF-␤ in CRTϪ/Ϫ MEFs (data not shown). Furthermore, Smad binding to Smad binding DNA elements is not deficient in the absence of CRT as TGF-␤ is able to induce Smad 2/3/4 reporter activity in wild type and CRTϪ/Ϫ MEFs (Fig. 7C). LY364947, an ALK5 (activin like kinase 5) TGF-␤R1 inhibitor, blocked TGF-␤ stimulation of Smad 2/3/4 reporter activity by both wild type and CRTϪ/Ϫ MEFs in this assay, suggesting that TGF-␤ receptor signaling is not deficient in CRTϪ/Ϫ MEFs (Fig. 7C). We also confirmed that Smad signaling is important for TGF-␤ stimulation of ECM in wild type MEFs, as treatment of wild type MEFs with LY364947 significantly impaired TGF-␤-induced collagen and -fibronectin transcript (data not shown). Finally, levels of active and total TGF-␤ were not decreased in the conditioned medium of CRTϪ/Ϫ MEFs (data not shown).
CRT Is Required for TGF-␤ Induction of Cytosolic Ca 2ϩ -TGF-␤ can stimulate the slow release of intracellular Ca 2ϩ , which activates the calcineurin/NFAT pathway to increase fibronectin expression in mesangial cells (37,41,(55)(56)(57). Because CRT regulates both ER Ca 2ϩ and calcineurin/NFAT activity (17,21,22), we asked whether TGF-␤ stimulation of Ca 2ϩ release was altered in CRTϪ/Ϫ MEFs. Wild type and CRTϪ/Ϫ MEFs were treated with TGF-␤ or the Ca 2ϩ ionophore, ionomycin, as a positive control (58). Ca 2ϩ release was measured over time using the Ca 2ϩ binding fluorescent dye Fluo-4 AM. With ionomycin, both wild type and CRTϪ/Ϫ MEFS induced a large increase in cytosolic Ca 2ϩ , although Ca 2ϩ re-uptake occurred more slowly in CRTϪ/Ϫ MEFs (Fig.  8, A and B). In contrast, TGF-␤ increased cytoplasmic Ca 2ϩ levels in wild type but not in CRTϪ/Ϫ MEFs (Fig. 8, A and B), suggesting that CRT is required for TGF-␤-dependent Ca 2ϩ signaling.
To determine whether TGF-␤ stimulation of cytosolic Ca 2ϩ is important for stimulation of ECM, wild type MEFs were treated with TGF-␤ in the presence or absence of thapsigargin, a SERCA2b inhibitor that blocks ER Ca 2ϩ re-uptake to effectively deplete ER-releasable Ca 2ϩ (59). Thapsigargin blocked the ability of TGF-␤ to stimulate collagen I and fibronectin transcript (Fig. 8, C and D), suggesting that TGF-␤ stimulation of cytosolic Ca 2ϩ release is required for induction of ECM proteins. Cyclopiazonic acid, a SERCA2b inhibitor that acts in a manner similar to thapsigargin (60), also inhibited TGF-␤stimulated ECM production (data not shown). To determine if increased cytoplasmic Ca 2ϩ is sufficient to support TGF-␤driven ECM expression in the absence of CRT, CRTϪ/Ϫ MEFs were treated with TGF-␤ in the presence of ionomycin. Despite an increase in cytoplasmic calcium with ionomycin, TGF-␤ was still unable to increase ECM transcript in CRT-deficient cells (Fig. 8E), suggesting that although CRT-mediated calcium regulation is critical for TGF-␤-driven ECM stimulation, calcium alone is not sufficient, and other CRT-dependent factors are likely important.
CRT Is Required for TGF-␤ Stimulation of NFAT Activity-The activity of the transcription factor NFAT is regulated by Ca 2ϩ -activated calcineurin, and NFAT activity can regulate TGF-␤-stimulated ECM production in the presence of sustained increases in cytoplasmic Ca 2ϩ (37,38). Given the importance of CRT for TGF-␤ stimulation of cytoplasmic Ca 2ϩ levels and the deficient NFAT activation in CRTϪ/Ϫ mouse embryos (20,23), we asked whether TGF-␤ stimulation of NFAT activity is defective in CRTϪ/Ϫ MEFs. TGF-␤ induces NFATc3 isoform nuclear translocation in wild type cells, whereas NFATc3 remained in the cytoplasm after TGF-␤ treatment of CRTϪ/Ϫ MEFs (Fig. 9A) (61). Furthermore, TGF-␤ stimulation NFAT reporter activity is also absent in CRTϪ/Ϫ MEFs (Fig. 9B).
NFAT Activity Is Required for TGF-␤-stimulated ECM Production-To determine if TGF-␤ stimulation of NFAT activity is important for production of type I collagen and fibronectin, we asked whether inhibition of NFAT activity blocked TGF-␤ stimulation of fibronectin and collagen transcription. We used a cell-permeable peptide, 11R-VIVIT, which specifically blocks calcineurin binding to the NFAT PXIXIT sequence and prevents calcineurin-dependent NFAT dephosphorylation and nuclear translocation (61,62). 11R-VIVIT blocked TGF-␤ stimulation of ECM in wild type MEFs (Fig. 9, C and D) and in human lung fibroblasts (data not shown). As expected, 11R-VIVIT had no effect on CRTϪ/Ϫ MEF ECM production by TGF-␤ (data not shown). Similarly, another NFAT inhibitor, A285222, which maintains NFAT in the cytoplasm regardless of calcineurin activity (63), showed a dose-dependent inhibition of TGF-␤ stimulated collagen and fibronectin transcript (data not shown). These results show that NFAT activity is necessary for TGF-␤ stimulation of ECM transcription in wild type MEFs and that CRT is required for TGF-␤ stimulation of NFAT activity.

DISCUSSION
ER stress is emerging as a significant factor in fibrotic disorders (1,2). Although ER stress-induced apoptosis is a contributing factor in fibrosis, a complete understanding of the mechanistic roles of ER stress in fibrosis remains to be defined. In our present studies we have shown that the ER stress-induced protein, CRT, is a critical regulator of fibrogenic responses to TGF-␤. We show that CRT regulation of cytosolic Ca 2ϩ and NFAT activity is required for TGF-␤ stimulation of collagen type I and fibronectin transcript (Fig. 10). Knock-out or knockdown of CRT abrogates cellular responsiveness to TGF-␤ even in the presence of active Smad 2/3 signaling. Consistent with these observations, cells with increased CRT expression exhibit a relative increase in responsiveness to TGF-␤. Importantly, these studies show that tunicamycin-induced ER stress is not sufficient to support TGF-␤ stimulation of ECM in the absence of CRT. These studies identify a critical mechanistic link between ER stress and fibrosis.
CRT expression is increased in a number of different fibrotic tissues, including lung, kidney, and diabetic atherosclerotic vasculature (2, 4), 4 suggesting an involvement in fibrotic processes. CRT acts as a collagen chaperone to mediate collagen ER-Golgi trafficking, processing, and incorporation into the ECM (24). In addition, CRT regulation of Src-dependent fibronectin expression and matrix deposition impacts collagen matrix assembly (24,25). Our current studies now show that CRT regulation of  . TGF-␤ stimulates Smad activity in wild type and CRT؊/؊ MEFs. A, wild type and CRTϪ/Ϫ MEFs were grown overnight in medium with 10% FBS, starved overnight in low (0.5%) serum medium, and treated with 100 pM TGF-␤ for 15, 30, or 60 min. Laemmli cell lysates containing phosphatase inhibitor were immunoblotted for phospho-Smad 3. Membranes were reprobed with antibody to Smad 2/3 or ␤-tubulin (data not shown) to normalize cell protein. The blot is representative of four separate experiments using 10 -400 pM TGF-␤. B, wild type and CRTϪ/Ϫ MEFs were grown overnight on glass coverslips in medium with 10% FBS, starved in low serum medium overnight, and treated with 100 pM TGF-␤ for 30 min. Cells were fixed, permeabilized, and incubated with antibody to phospho-Smad 3 followed by fluorescein-conjugated secondary antibody. Nuclei were stained with Hoechst. Results are representative of three separate experiments. Confocal images were obtained at an original magnification of 600ϫ. C, wild type and CRTϪ/Ϫ MEFs were transfected with the Smad 2/3/4 firefly luciferase reporter construct and the control renilla luciferase construct and kept in medium with 10% FBS overnight, switched to low serum DMEM, and treated with 100 pM TGF-␤ for 8 h. Some cells were treated with 100 pM TGF-␤ in the presence of 3 M LY364947. Cells were lysed with 1ϫ lysis buffer (Promega), and triplicate samples were combined. Luciferase reporter activity is normalized to the renilla luciferase control. Data represent the means of samples from three separate experiments Ϯ S.D. *, p Ͻ 0.05 versus untreated control.
Ca 2ϩ -dependent NFAT activity is required for the ability of TGF-␤ to stimulate ECM transcription.
TGF-␤ is a pro-fibrotic cytokine that drives expression of ECM proteins and integrin ECM receptors (64 -67). TGF-␤ signals through both the canonical Smad pathway and other non-Smad pathways (20 -24). Although not well studied, there is evidence that TGF-␤ can increase cytoplasmic Ca 2ϩ and activate calcineurin, which dephosphorylates NFAT, resulting in NFAT nuclear translocation (37,38). Despite these data, the mechanism by which TGF-␤ increases cytosolic Ca 2ϩ remains unclear. Our data suggest that CRT is essential for TGF-␤ to increase cytoplasmic Ca 2ϩ levels as cells lacking CRT were unable to increase cytoplasmic Ca 2ϩ when treated with TGF-␤. There are several mechanisms by which TGF-␤ increases intracellular Ca 2ϩ levels, including increasing inositol 3-phosphate levels to cause release of Ca 2ϩ from thapsigargin-sensitive stores in the ER, mediating translocation of type III inositol 3-phosphate receptors to the cell surface, stimulating H 2 O 2mediated Ca 2ϩ release, and via a c-Jun-dependent mechanism (39 -43). Rises in cytoplasmic Ca 2ϩ levels are typically regu-lated by ER-mediated Ca 2ϩ release or through store-operated Ca 2ϩ entry through channels present on the plasma membrane (68,69). CRT can regulate both ER-mediated Ca 2ϩ release and store-operated Ca 2ϩ entry (19, 20, 70 -72). The ability of thapsigargin to block responses to TGF-␤ suggests that CRT regulation of ER Ca 2ϩ stores might be involved, although TGF-␤ stimulation of wild type MEFs in medium lacking extracellular Ca 2ϩ failed to stimulate a rise in cytoplasmic Ca 2ϩ , suggesting that CRT might also regulate cytoplasmic Ca 2ϩ levels through regulation of store-operated Ca 2ϩ entry (data not shown). In addition, treatment of CRTϪ/Ϫ MEFs with TGF-␤ in the presence of ionomycin failed to make CRTϪ/Ϫ cells responsive to TGF-␤, showing that TGF-␤ stimulation of ECM is not simply a function of increased cytoplasmic calcium and suggesting that additional CRT-dependent functions are involved in regulating TGF-␤ stimulation of ECM.
TGF-␤ stimulation of NFAT activation and nuclear translocation were impaired in CRTϪ/Ϫ MEFs, suggesting that CRTmediated calcineurin activity is likely impaired in CRTϪ/Ϫ MEFs. Our data are consistent with prior reports demonstrat- ing that CRT is needed for efficient NFAT nuclear translocation (21,23). Furthermore, studies by Gooch et al. (37) showed that TGF-␤ activates calcineurin in a time-and dose-dependent manner and that calcineurin is required for TGF-␤-stimulated ECM in glomerular mesangial cells. These studies showed that TGF-␤ stimulation of fibronectin is dependent on NFAT binding to the fibronectin promoter, as cells expressing a dominant negative NFATc mutant are unable to induce ECM in response to TGF-␤ (38). NFAT signaling has also been shown to be important for myofibroblast induction and expression of collagens I and III by cardiac fibroblasts in response to mechanical stress and osteopontin by vascular smooth muscle cells in high glucose (73,74).
CRT can regulate cell behavior, ECM production, and wound healing from multiple cellular compartments, including at the cell surface and as an extracellular ligand (75,76). However, cell surface CRT, which can act as a co-receptor for TSP1 to stimulate collagen production, does not appear to be involved in mediating these TGF-␤ responses, as CRTϪ/Ϫ MEFs stably expressing CRT lacking the TSP1 binding sequence had normal responses to TGF-␤ (47).
The absence of CRT does not appear to negatively impact Smad 2/3 signaling as Smad 2/3 phosphorylation and nuclear translocation are unaffected. Because nuclear CRT can affect DNA binding of some nuclear hormones such as vitamin D, a nuclear receptor known to antagonize TGF-␤ signaling (45,77), we also examined the ability of TGF-␤ to stimulate Smad reporter activity in CRT-deficient cells. Although reporter activity in CRTϪ/Ϫ MEFs was reduced as compared with wild type cells, TGF-␤ was still able to stimulate Smad reporter activity 2.5-fold in CRTϪ/Ϫ MEFs, levels that have been shown to be sufficient for Smad-induced gene transcription in other systems (78,79). We also investigated whether the lack of CRT impacted non-Smad pathways regulated by TGF-␤ (32)(33)(34)(35)(36). In contrast to the negative effects of CRT deficiency on the calcium-NFAT pathway, the CRTϪ/Ϫ MEFs did not show alterations in levels of p-AKT, p-JNK, and p-ERK induced by TGF-␤ (data not shown). There was a trend toward increased phosphorylated p-38 MAPK in CRTϪ/Ϫ cells treated with TGF-␤, although these data did not reach statistical significance, and the importance remains unclear.
Recent literature suggests that ER stress is associated with fibrotic disorders such as diabetic atherosclerosis, pulmonary fibrosis, and diabetic nephropathy (1)(2)(3)(4)(5)(6)(7). Chronic stimuli such as high glucose, glucosamine, or oxidative stress can up-regulate ER stress proteins and are associated with enhanced matrix production and fibrotic disease (3-5, 80, 81). Chemical chaperones such as 4-phenylbutyric reduce the expression of ER stress proteins such as CRT and GRP78 and also reduce fibrotic remodeling due to high glucose in animal models of atherosclerosis and nephropathy (3)(4)(5). Despite the growing appreciation for a role for ER stress in fibrotic disease, knowledge regarding the mechanisms by which ER stress regulates fibrosis is limited (8,15). Although there is a clear role for ER stress-induced apoptosis in some models of fibrosis (6), there is also evidence for ER stress involvement in processes important for fibroproliferative remodeling that are independent of apoptosis (82,83). Knockdown of GRP78, another ER stress response protein, also reduces TGF-␤ or tunicamycin-stimulated collagen and ␣-smooth muscle actin production in human lung fibroblasts (82). 150-kDa oxygen-regulated protein (ORP150) mediates TGF-␤ myofibroblast induction and collagen production in human lung fibroblasts (83). The mechanisms by which GRP78 and ORP150 mediate responsiveness to TGF-␤ were not addressed in these reports. Our data now suggest that CRT is an important ER stress factor in fibrotic remodeling through regulation TGF-␤-stimulated ECM production. It is interesting that CRT mediates TGF-␤ responsiveness through its role as a regulator of calcium signaling rather than through its chaperone function, suggesting a unique role for CRT in the ER stress response. In contrast to these reports, Vonk et al. (84) showed that enhanced ER stress due to glucose and nutrient deprivation reduces collagen production in chondrocyte and dermal fibroblasts. Nonetheless, the majority of evidence suggests that enhanced ER stress is associated with exacerbated fibrotic outcomes (1-7, 10, 14, 15, 85). TGF-␤ has been shown to increase expression of ER stress response proteins, including CRT in the human IPF lung fibroblasts and rat vascular smooth muscle cells (data not shown) (82,83). However, we did not observe TGF-␤ stimulation of either GRP78 or CRT in MEFs or Thy-1 Ϫ/Ϫ rat lung fibroblasts, suggesting that basal levels of CRT expression are sufficient to mediate responses to TGF-␤. Furthermore, knockdown of CRT in Thy1Ϫ/Ϫ rat lung fibroblasts to 60% that of control levels was able to attenuate TGF-␤-stimulated matrix production, although base-line levels of matrix proteins were unaffected. The lack of effect on base-line matrix production might reflect a lower threshold of CRT required to maintain homeostatic levels of ECM expression in these cells. This would be consistent with our observations that knockdown of CRT to 35% that of control levels does reduce basal ECM expression in human lung fibroblasts. Interestingly, CRTϪ/Ϫ MEFs have increased basal levels of the ER stress proteins GRP78, GRP94 (glucose-regulated protein 94), calnexin, and protein disulfide isomerase (data not shown and Ref. 86) and reduced expression of multiple ECM proteins (24), suggesting that increased expression of other ER stress proteins cannot compensate for the lack of CRT in mediating ECM production in response to TGF-␤ (1). This conclusion is supported by our observation that TGF-␤ treatment of CRTϪ/Ϫ MEFs in the presence tunicamycin did not increase collagen I protein. These data suggest that enhanced levels of ER stress with TGF-␤ in the absence of CRT are not sufficient to drive fibrosis, implicating CRT as an important link between enhanced levels of ER stress and fibrotic disease. Furthermore, these data suggest that ER CRT might be a novel therapeutic target to attenuate fibrosis. In conclusion, these studies demonstrate that CRT is required for TGF-␤-stimulated ECM production and provide a link between enhanced ER stress and TGF-␤ stimulation of ECM.