Regulation of the Transforming Growth Factor (cid:98) -responsive Transcription Factor CTF-1 by Calcineurin and Calcium/ Calmodulin-dependent Protein Kinase IV*

Transforming growth factor (cid:98) (TGF- (cid:98) ) is a pluripotent peptide hormone that regulates various cellular activi- ties, including growth, differentiation, and extracellular matrix protein gene expression. We previously showed that TGF- (cid:98) induces the transcriptional activation domain (TAD) of CTF-1, the prototypic member of the CTF/ NF-I family of transcription factors. This induction correlates with the proposed role of CTF/NF-I binding sites in collagen gene induction by TGF- (cid:98) . However, the mechanisms of TGF- (cid:98) signal transduction remain poorly understood. Here, we analyzed the role of free calcium signaling in the induction of CTF-1 transcriptional activity by TGF- (cid:98) . We found that TGF- (cid:98) stimulates calcium influx and mediates an increase of the cytoplasmic calcium concentration in NIH3T3 cells. TGF- (cid:98) induction of CTF-1 is inhibited in cells pretreated with thapsigargin, which depletes the endoplasmic re- ticulum calcium stores, thus further arguing for the potential relevance of calcium mobilization in TGF- (cid:98) action. Consistent with this possibility, expression of a constitutively active form of the calcium/calmodulin-de-pendent phosphatase calcineurin or of

Members of the transforming growth factor ␤ (TGF-␤) 1 su-perfamily are small peptide hormones that regulate growth, proliferation, differentiation, apoptosis, wound healing, and gene expression of responsive cells in a variety of tissues and organisms (reviewed in Ref. 1). The extreme diversity of TGF-␤ actions is further complicated by the fact that TGF-␤ often elicits opposite responses, depending on the cell type and the assay conditions. The importance of TGF-␤ in cell physiology and pathophysiology is emphasized by the identification of several fibrotic disorders that may arise from increased TGF-␤ sensitivity such as arthritis (2), as well as by the increased tumorigenicity of several epithelial cells that have lost TGF-␤ responsiveness (3).
TGF-␤ binds specific receptors at the cell surface, which are often composed of two dimers of distinct transmembrane serine/threonine kinase receptor chains referred to as type I (T␤RI) and type II (T␤RII) TGF-␤ receptors. Following ligand binding, T␤RII is thought to phosphorylate and activate the T␤RI kinase, which in turn propagates the signal to downstream substrates. Genetic and biochemical analyses in nematode, fruit fly, and primate cells led to the identification of several putative TGF-␤ signal transducers, including various proteins interacting with the TGF-␤ receptor complex and a family of related factors collectively referred to as the Smad proteins (1,4). Some of the Smads were shown to influence known aspects of TGF-␤ signaling, which led to the proposal that they may act as bona fide TGF-␤ effectors. For example, Smad3 and Smad4 are necessary and sufficient for specific TGF-␤-mediated growth and transcriptional responses in Mv1Lu cells (5), whereas Smad2 participates in the transcriptional induction of the Mix.2 gene, a well known target of the activin-like members of the TGF-␤ superfamily (6). Thus, individual Smads appear to play a pivotal role in relaying the signals of the various TGF-␤ ligands. Nevertheless, their specific functions remain unknown.
Recently, a possible role for calcium in TGF-␤ action has been proposed (7,8). For instance, T␤RI was shown to interact with FKBP12 (9), the prototype of the FKBP class of immunophilins (see Ref. 10 for review). FKBP12 is of interest, as it has the potential to inhibit the activity of the calcium-dependent protein phosphatase calcineurin (11). In addition, FKBP12 and calcineurin may regulate the function of the inositol 1,4,5trisphosphate and ryanodine receptors (IP 3 R and RyR), the major endoplasmic reticulum (ER) receptor channels controlling calcium stores (12,13). However, whether calcineurin and * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
We have previously shown that TGF-␤ specifically induces the activity of the transcriptional activation domain (TAD) of CTF-1, the prototypic member of the CTF/NF-I family of proline-rich transcription factors (14). Interestingly, the TGF-␤responsive domain of the CTF-1 TAD mediates histone H3 interaction and alters nucleosomal structure, suggesting that TGF-␤ may induce gene expression by chromatin remodeling. In addition, the TGF-␤-responsive domain mediates tumor necrosis factor ␣ (TNF-␣) repression of basal and TGF-␤-induced CTF-1 transcriptional activity (15). This antagonistic regulation of CTF-1 activity by TGF-␤ and TNF-␣ correlates with the opposing effects of the two growth factors on the regulation of some collagen genes (16,17), which may be relevant for the process of wound healing.
Here, we analyzed the relevance of calcium signaling in the induction of CTF-1 transcriptional activity by TGF-␤. We show that TGF-␤ stimulates calcium influx and mediates an increase in the free cytoplasmic calcium concentration in NIH3T3 cells. TGF-␤ induction of the CTF-1 TAD is inhibited in cells pretreated with thapsigargin, which interferes with calcium homeostasis, while expression of constitutive forms of calcineurin or of the calcium/calmodulin-dependent kinase IV (CaMKIV) induces CTF-1 transcriptional activity. Thus, calcium signaling may be relevant for the regulation of CTF-1 by extracellular stimuli.

MATERIALS AND METHODS
Plasmids-pG5BCAT, p␣CAT-⌬87-3xAd, p␣CAT-⌬55, and p3TP-Lux have been described previously (14,18). The CMV-driven mammalian expression vectors for the CTF-1 fusion protein GAL 399 -499 and its derivatives GAL 399 -486, GAL 486 -499, GAL 479 -499 (3xM), the CTF-2 fusion vector GAL 399 -430 and GAL-AP2 were derived after subcloning appropriate fragments of the corresponding SV40-based vectors (14), in pCMV5 (19). The untagged version of ⌬CNA, a deletion mutant of the mouse CN␣4 calcineurin catalytic subunit lacking the autoinhibitory and the calmodulin-binding domains, was expressed from the previously described SR␣-driven vector (pSR␣-⌬CaM-AI, Ref. 20). ⌬CNA was tagged at the C terminus with the hemagglutinin HA1 epitope after subcloning a 1.3-kilobase pair cDNA fragment encoding the first 398 amino acids of the protein in pCMV-HA, a modified pCMV5 vector 2 to yield pCMV-⌬CNA-HA. (The addition of the tag had no detectable effect on ⌬CNA activity, hence pSR␣-⌬CaM-AI and pCMV-⌬CNA-HA were used interchangeably in these studies). The ⌬CNA mutant used in Fig. 2 was constructed in pCMV-HA using overlapextension PCR mutagenesis and pSR␣-⌬CaM-AI as template. It contains several point mutations that convert codons 349, 350, 356, and 357 of the CN␣4 cDNA in glutamate, thereby abolishing CNB binding, as shown previously (E mutant in Ref. 21), as well as a carboxylterminal HA tag, as above. A 0.5-kilobase pair cDNA fragment encoding the regulatory B subunit of calcineurin was cloned from total rat brain RNA by reverse transcription PCR and appropriate primers designed after the published sequence (GeneBank™ accession number LO3554), digested with Asp-718 and EagI, and inserted between the cohesive sites of pCMV5 to yield pCMV-CNB. Oligonucleotide-mediated PCR 2 A. Alevizopoulos, unpublished data.

FIG. 1. TGF-␤ regulates calcium dynamics in NIH3T3 cells.
A, TGF-␤ stimulates calcium influx. 7 ϫ 10 4 serum-starved NIH3T3 cells growing in 24-well plates were washed twice with PSS buffer containing 0.12 mM CaCl 2 and then stimulated with ethanol carrier or 5 ng/ml of human TGF-␤ for the indicated times at 37°C. The cells were subsequently labeled with 1 Ci of 45 Ca 2ϩ /well at 37°C for 5 more min in the continued presence or absence of TGF-␤, as indicated. The cells were washed four times with cold PSS buffer containing 1.2 mM CaCl 2 , lysed in 1% SDS, and the 45 Ca 2ϩ content of the lysate was measured by liquid scintillation counting. For each time point, bars represent the meansϮstandard deviation of three independent influx experiments done in quadruplicate. Results are expressed as percentage of specific 45 Ca 2ϩ influx relative to that obtained in the absence of TGF-␤, which corresponds to a mean of 344 cpm 45 Ca 2ϩ /min and was arbitrarily set to 100. The star indicates a statistically significant value (p Ͻ 0.05, Student's t test). B, TGF-␤ leads to an increase in the cytoplasmic free calcium concentration ([Ca 2ϩ ] c ) of NIH3T3 cells. 3 ϫ 10 5 serum-starved NIH3T3 cells growing on glass coverslips were loaded with fura-2 and subsequently processed for fura-2 fluorescence, as described under "Materials and Methods." TGF-␤ was added 0.5 min after the beginning of the recordings, as indicated by the arrow. The trace represents measurements of [Ca 2ϩ ] c recorded from a single field of about five cells and is representative of four independent experiments with similar results. C, extracellular calcium is required for the TGF-␤-dependent increase of [Ca 2ϩ ] c . Fura-2-loaded NIH3T3 cells received 10 mM EGTA and 5 ng/ml TGF-␤ 1 and 2 min after the beginning of the recordings respectively, as indicated by the arrows. The trace represents measurements of [Ca 2ϩ ] c in the presence of both agents, recorded from a single field of about five cells. The experiment was repeated twice with identical results. mutagenesis of an RSV-driven expression vector encoding a truncated 313-amino acid-long catalytic domain of the mouse calmodulin-dependent protein kinase IV (⌬CaMKIV, which lacks the calmodulin binding and the autoinhibitory domains of wild-type CaMKIV, Ref. 22) was used to construct a catalytically inactive ⌬CaMKIV kinase variant, after replacement of lysine 75 by glutamate (23). The SR␣-based mammalian expression vectors for the constitutive and the catalytically inactive variants of the ␥ isoform of the CaMKII kinase have been described previously (24). The CMV-CaMKI expression vector was constructed after subcloning of a 1.3-kilobase pair BamHI-EcoRI fragment encoding the complete rat cDNA sequence (from pGEX-2T-ATG 92 , Ref. 25) in the cohesive sites of pCDNA3 (Invitrogen). Correct PCR amplification of all recombinant clones was confirmed by dideoxy DNA sequencing.
Cell Lines and Transfections-NIH3T3 cells were grown in standard DMEM medium supplemented with 10% donor calf serum (Life Technologies, Inc.) and antibiotics. Cells were transiently transfected by electroporation, essentially as described (14). Briefly, 4.5 ϫ 10 6 cells were mixed with 70 g of total plasmid DNA and pulsed once at 960 microfarads and 250 V at room temperature, according to the instructions of the electroporator manufacturer (Bio-Rad). The contents of one pulsed cuvette were split in two, and cells were plated in DMEM plus 0.5% donor calf serum for 3-5 h. Cultures were then induced for a period of 15 h with either ethanol vehicle or 5 ng/ml of human TGF-␤1 (Nacalai Tesque, Kyoto, Japan). Cells were lysed in 1 ϫ reporter lysis buffer (Promega Corp.); CAT and luciferase activities were determined using standard procedures and normalized according to ␤-galactosidase activity from a co-transfected internal control plasmid (CMV␤gal, CLONTECH). As ⌬CaMKIV slightly (Ͻ2 ϫ) induces CMV␤gal activity, CAT activities in the experiments involving ⌬CaMKIV were normalized according to the total protein concentration.
Gel Mobility Shift Assays-For gel shift analysis of the endogenous CTF/NF-I or of the transiently expressed GAL4 fusion proteins, cells were lysed in extraction buffer (20 mM Tris, pH 7.5, 20% glycerol, 500 mM KCl, 1 mM dithiothreitol, and protease inhibitors) as described by Martinez et al. (26). Whole-cell lysates were normalized for protein concentration and incubated with end-labeled double-stranded DNA probes containing either the high-affinity CTF/NF-I binding site found within the first 50 base pairs of the Adenovirus origin of replication (27) or the 17-base pair GAL4 binding site (28). Protein⅐DNA complexes were separated from free probe on native polyacrylamide gels and revealed by autoradiography.
Calcium Influx and Fura-2 Fluorescence Studies-For 45 Ca 2ϩ influx studies, 7 ϫ 10 4 exponentially growing NIH3T3 cells were plated in 24-well plates and incubated in standard DMEM plus 0.5% donor calf serum overnight. Cells were washed twice with 0.5 ml of PSS buffer (145 mM NaCl, 5 mM KCl, 1 mM MgCl 2 , 10 mM glucose, 5 mM Hepes, pH 7.4) containing 0.12 mM CaCl 2 , and they were then induced with 5 ng/ml of TGF-␤ or ethanol carrier in 0.2 ml of the same buffer at 37°C. The cells were labeled with 1 Ci of 45 Ca/well at 37°C for the last 5 min of the incubation in the presence or absence of the growth factor. The assay was terminated by aspirating the labeling medium and rinsing the cells 4 ϫ with 0.5 ml of cold PSS buffer containing 1.2 mM CaCl 2 . The cells were then trypsinized, lysed in 0.25 ml of 1% SDS, and the 45 Ca 2ϩ content of the lysate was measured by liquid scintillation counting (Packard Tri Carb 4640).
The cytoplasmic free calcium concentration ([Ca 2ϩ ] c ) was measured using fura-2 fluorescence, essentially as described (29). Briefly, 3 ϫ 10 5 serum-starved NIH3T3 cells plated on glass coverslips were washed twice with PSS buffer containing 1.2 mM CaCl 2 and were subsequently loaded with 5 M fura-2/AM (Molecular Probes) for 40 min in the same buffer at room temperature and in the dark. The excess of fura-2 was removed by washing twice with PSS, 1.2 mM Ca 2ϩ , and the coverslip was inserted into a thermostatted chamber on a Nikon Diaphot inverted epifluorescence microscope, which is part of a PhoCal single cell fluorescence analyzer (Life Science Resources, Cambridge, UK). The cells were illuminated with alternating light of 340 and 380 nm from a rotating filter wheel. Emission was monitored at 510 nm from a field of about five cells, and the data were analyzed using PhoCal software. Calibrations were performed by treating the cells with 10 M ionomycin plus 10 mM calcium to obtain the maximal signal, followed by the addition of 10 mM EGTA to obtain the minimal signal. Background fluorescence, obtained by quenching the signal with 1 mM MnCl 2 , was subtracted from the signals. [Ca 2ϩ ] c was calculated based on the equation of Grynkiewicz et al. (30).

FIG. 2. Thapsigargin, a compound that interferes with cellular calcium homeostasis, inhibits TGF-␤ induction of the CTF-1 TAD.
A, NIH3T3 cells were transfected with a reporter construct containing five GAL4 binding sites in front of a TATA box and the cat reporter gene (G5BCAT), an internal control plasmid (CMV␤gal), and a CMV-driven expression vector coding for a GAL4 fusion protein containing either the entire CTF-1 TAD (amino acids 399 -499) or a portion of the AP-2 TAD (amino acids 31-76) (14). Five hours post-transfection, cells received 15 nM thapsigargin ("thapsi", Sigma) or Me 2 SO (DMSO) carrier in standard DMEM medium containing 2 mM Ca 2ϩ . Twenty minutes later, cells received 5 ng/ml TGF-␤ or were left untreated, as indicated. After overnight incubation in the presence of these agents, the cells were collected and CAT activities were determined and normalized to ␤-galactosidase activity. The mean values of three independent experiments are presented, expressed as normalized CAT activity relative to that obtained with GAL 399 -499 in the absence of stimuli, which was set to 100. B, equal protein amounts of whole-cell lysates of NIH3T3 cells that had been transfected with expression vectors encoding GAL 399 -499 or GAL-AP2 (lanes 2-9) or mock-transfected (lane 1) and subsequently treated with 15 nM thapsigargin and/or 5 ng/ml TGF-␤, exactly as denoted in part A, were analyzed for GAL4 DNA binding activity in a gel mobility shift assay. The relevant GAL4 fusion⅐DNA complexes are indicated by arrows, whereas the closed circle indicates complexes that may result from proteolytic degradation products of the GAL4 fusion proteins.

TGF-␤ Regulates Calcium Dynamics in NIH3T3 Cells-Pre-
vious studies indicated that the proline-rich transcriptional activation domain of CTF-1 mediates TGF-␤ induction in NIH3T3 fibroblasts (14) and that the type I TGF-␤ receptor interacts with the immunophilin FKBP12, a protein implicated in the regulation of calcium homeostasis (9,12). Therefore, we wished to evaluate the relevance of calcium signaling in CTF-1 regulation. First, we assessed whether TGF-␤ is capable of mobilizing cellular calcium, by performing 45 Ca 2ϩ influx studies in NIH3T3 cells (Fig. 1A). Serum-starved cells were treated with TGF-␤ or carrier for specific time periods, labeled with 45 Ca 2ϩ , and the radioactivity associated with the cells ( 45 Ca 2ϩ influx) was measured. Incubation of cells with TGF-␤ for periods under 7 min had no effect on 45 Ca 2ϩ influx. However, a significant increase in 45 Ca 2ϩ influx was observed in TGF-␤treated cultures after 7 min, whereas thereafter the calcium influx declined to intermediate levels (Fig. 1A). Thus, these results imply that TGF-␤ stimulates calcium influx in NIH3T3 cells.
As increased calcium entry should lead to elevated total free calcium concentrations ([Ca 2ϩ ] c ) in the cytoplasm of the cells, we measured [Ca 2ϩ ] c following TGF-␤ treatment of fura-2loaded NIH3T3 fibroblasts (Fig. 1B). Consistent with the results of the 45 Ca 2ϩ influx studies, treatment of the cells with TGF-␤ for 7 min resulted in a delayed but rapid elevation of [Ca 2ϩ ] c from 25 nM to approximately 150 nM. This peak value of [Ca 2ϩ ] c persisted for a period of 3-4 min, whereas significantly increased [Ca 2ϩ ] c levels could still be detected for prolonged periods (Fig. 1B). No [Ca 2ϩ ] c increase could be observed in vehicle-treated cells, even after prolonged incubations at 37°C (Ͼ1.5 h, results not shown). In addition, chelation of the extracellular calcium with EGTA prevented the TGF-␤-mediated [Ca 2ϩ ] c increase (Fig. 1C), indicating that the presence of external Ca 2ϩ is required for TGF-␤ action. These data altogether demonstrate that TGF-␤ mediates a potent and prolonged free Ca 2ϩ increase, at least in part by stimulating calcium entry in NIH3T3 cells, and they support a possible involvement of calcium-dependent events in TGF-␤-mediated signaling.
Growth factor-triggered activation of calcium-responsive signaling pathways depends on the integrity and content of the intracellular Ca 2ϩ pools, which mainly reside in the ER (reviewed in Ref. 31). ER calcium content can be irreversibly depleted following treatment of the cells with thapsigargin, a highly specific ER Ca 2ϩ pump inhibitor (32,33). We therefore tested whether treatment of NIH3T3 cells with this compound might affect TGF-␤ induction of the CTF-1 TAD. As shown previously (14,15), GAL 399 -499, a chimeric protein contain-ing the entire proline-rich transactivation domain of CTF-1 fused to the DNA binding and dimerization domain of the yeast protein GAL4, specifically activated transcription from GAL4 binding sites in NIH3T3 cells, and this basal activity was induced by TGF-␤ ( Fig. 2A). Interestingly, treatment of the cells with low concentrations (15 nM) of thapsigargin inhibited the TGF-␤-mediated induction of GAL 399 -499 transcriptional activity ( Fig. 2A). This effect was specific, as thapsigargin did not significantly alter the transcriptional activities of control GAL4 fusion proteins containing the TADs of other transcription factors such as AP2 and Oct2 (14). In addition, thapsigargin did not significantly affect GAL 399 -499 and GAL-AP2 expression, as measured by gel shift assays (Fig. 2B and results not shown). Thus, these data suggest that the induction of the CTF-1 TAD by TGF-␤ is a specific thapsigargin-sensitive process. Moreover, these data are consistent with the notion that an intact ER calcium pool may be necessary for TGF-␤ induction of the CTF-1 TAD, further arguing for the involvement of calcium signaling in TGF-␤ action.
Calcineurin Induces the Transcriptional Activation Domain of CTF-1 in a FK506-and Cyclosporin A-sensitive Fashion-Numerous effectors have been shown to mediate calcium signaling in mammalian cells (10). Of these, the calcium-dependent phosphatase calcineurin may be important for TGF-␤ signal transduction, as it is regulated by the immunophilin FKBP12, a protein recently shown to interact with the TGF-␤ type I receptor (9). We therefore asked whether overexpressed calcineurin might affect GAL 399 -499 induction by TGF-␤. Interestingly, expression of ⌬CNA (a truncated catalytic calcineurin subunit lacking both the autoinhibitory and the calcium/calmodulin-binding domains, Ref. 20) alone or in the presence of its heterodimeric partner CNB (the regulatory calcineurin subunit, Refs. 35 and 36) resulted in a significant increase of both the basal and the TGF-␤-induced GAL 399 -499 transcriptional activities (Fig. 3A). In contrast, expression of a mutated ⌬CNA (which is unable to heterodimerize with CNB, Ref. 21) or of other phosphatases such as PP2A or PTP1D did not induce the CTF-1 TAD (Fig. 3A and data not shown). In addition, four distinct GAL4 fusions, containing either the proline-rich TADs of the transcription factors CTF-2 or AP2 or the glutamine-rich TADs of Oct2 or Sp1 all failed to mediate calcineurin activation (Fig. 3A, right panel, and data not shown). Gel mobility shift assays indicated that calcineurin had no significant effect on the DNA-binding activity and/or the expression levels of GAL 399 -499 and GAL-AP2 (Fig. 3B). Thus, calcineurin expression leads to an increase of the transcriptional activity of TGF-␤-responsive CTF/NF-I species, such as CTF-1 but not CTF-2, in NIH3T3 fibroblasts.  (20,21), the wild-type rat calcineurin regulatory B subunit (CNB), or equimolar amounts of the parental empty vector, as indicated. Five hours post-transfection, cells received 5 ng/ml TGF-␤ or were left untreated, as indicated. CAT activities were determined 18 h later and normalized to ␤-galactosidase activity. The mean values of at least three independent experiments are presented, expressed as normalized CAT activity relative to that obtained with GAL 399 -499 in the absence of stimuli, which was set to 100. B, NIH3T3 cells were either transfected with expression vectors coding for GAL 399 -499, GAL-AP2, and/or ⌬CNA and CNB (lanes 2-13) or mock-transfected (lane 1), as in part A. Five hours post-transfection, cells received 5 ng/ml TGF-␤ or were left untreated, as indicated. The cells were collected after overnight incubation, and equal protein amounts of whole cell lysates were analyzed for GAL4 DNA binding activity in a gel mobility shift assay, as described under "Materials and Methods." The relevant GAL4 fusion⅐DNA complexes are indicated by arrows, whereas the closed circle indicates complexes that may result from proteolytic degradation products of the GAL4 fusion proteins. C, the immunosuppressants FK506 and cyclosporin A abolish calcineurin-mediated induction of the CTF-1 TAD. NIH3T3 cells were transfected with G5BCAT, CMV␤gal, and CMV-driven expression vectors for either GAL 399 -499 or GAL-AP2 and/or wild-type ⌬CNA, as indicated. Five hours post-transfection, cells were stimulated with Me 2 SO (DMSO), 2 g/ml FK506 (Fujisawa USA Inc.) or 1 g/ml of cyclosporin A (CsA, Novartis Pharma Inc.), and 20 min later they received 5 ng/ml TGF-␤ or carrier, as indicated. After overnight incubation in the presence of these agents, the cells were collected and CAT activities were determined and normalized to ␤-galactosidase activity. The mean values of at least three independent experiments are presented, expressed as normalized CAT activity relative to that obtained with GAL 399 -499 in the absence of stimuli, which was set to 100.
The enzymatic activity of calcineurin can be inhibited by the immunosuppressive drugs FK506 and cyclosporin A (CsA) complexed with their respective cellular receptors, such as the immunophilins FKBP12 and cyclophilin A respectively, but not by a rapamycin⅐FKBP12 complex (11). We thus tested the effect of these three immunosuppressants on GAL 399 -499 activation by ⌬CNA and TGF-␤ (Fig. 3C). GAL 399 -499 still mediated TGF-␤ induction in the presence of FK506, whereas a slightly reduced TGF-␤ induction was consistently observed in CsA-treated NIH3T3 fibroblasts. Significantly, treatment of the cells with either FK506 or CsA eliminated the ability of ⌬CNA to induce basal GAL 399 -499 activity (Fig. 3C). In contrast, rapamycin had no effect on GAL 399 -499 regulation, whereas GAL-AP2 and GAL-Oct2 activities were unaffected by treatment with either immunosuppressant (Fig. 3C and data  not shown). Thus, the enzymatic activity of calcineurin is required for GAL 399 -499 induction.
Calcium/Calmodulin-dependent Protein Kinase IV Is a Potent Activator of the CTF-1 TAD-Although calcineurin induces the basal transcriptional activity of CTF-1, it has little effect on TGF-␤ induction per se (Fig. 3), suggesting that calcineurin may not be a direct intermediate of TGF-␤ action. ⌬CNA may act indirectly when overexpressed, however, for example by increasing the activity of a calcium-sensitive component(s) of the TGF-␤ pathway. We therefore tested whether other known calcium-regulated signaling enzymes, such as specific calcium/ calmodulin-dependent kinases (CaMKs, reviewed in Ref. 34), might be involved in the induction of GAL 399 -499 transcriptional activity by TGF-␤. Suprisingly, expression of a truncated catalytic subunit of CaMKIV, lacking both the autoinhibitory and the calcium/calmodulin-binding domains (⌬CaMKIV), was found to induce basal GAL 399 -499 transcriptional activity more than 80-fold in NIH3T3 fibroblasts, and to a lesser extent in Mv1Lu, TA-1, and COS-7 cells (Fig. 4A, "⌬IV," and data not shown). In contrast, expression of a catalytically inactive ⌬CaMKIV mutant, or of the calcium/calmodulin-dependent protein kinases I and II (24,25), failed to induce the CTF-1 TAD. In addition, ⌬CaMKIV did not significantly affect GAL-DBD, GAL-AP2, GAL-Sp1, and GAL-Oct2 basal transcriptional activities, whereas it increased only slightly the expression  (23), a CMV vector expressing wild-type rat CaMKI, and/or equimolar amounts of the corresponding parental vectors without insert, as indicated. Five hours post-transfection, cells received 5 ng/ml TGF-␤ or carrier, as indicated. After overnight incubation, the cells were collected, and CAT activities were determined. The mean values of two to three independent experiments are presented, expressed as normalized CAT activity relative to that obtained with GAL 399 -499 in the absence of stimuli, which was set to 100. B, the TGF-␤ responsive domain of CTF-1 mediates ⌬CaMKIV-regulated transcriptional induction. NIH3T3 cells were transfected with G5BCAT, CMV␤gal, a RSV-driven expression vector encoding wild-type ⌬CaMKIV or the parental vector devoid of the kinase sequences, and CMVdriven plasmids encoding one the following GAL4 DNA-binding domain (GAL-DBD)derived chimeric activators: GAL 399 -499, containing the wild-type CTF-1 TAD; GAL 399 -486, a GAL 399 -499 derivative lacking the CTF-1 TGF-␤-responsive domain (TRD); GAL 486 -499, containing the wildtype CTF-1 TRD (14); GAL 479 -499 (3xM), containing a mutated CTF-1 TRD, which cannot be phosphorylated (because of substitution of tyrosines 491 and 497 with phenylalanine and of the serine 495 with alanine, Ref. 14); and GAL 399 -430, containing the wild-type CTF-2 TAD, which naturally lacks the TRD (37). Five hours post-transfection, cells received 5 ng/ml TGF-␤ or carrier, as indicated. After overnight incubation, the cells were collected and CAT activities were determined. For each activator, the values of the bars correspond to the ratio of normalized CAT activity obtained in the presence of ⌬CaMKIV over that obtained in the presence of the empty RSV vector (⌬CaMKIVmediated fold activation) and are the means of two independent experiments. levels of all GAL4 fusion proteins tested (Fig. 4A and data not  shown). Thus, ⌬CaMKIV specifically induces the activity of the CTF-1 transcriptional activation domain.
TGF-␤ induction of CTF-1 transcriptional activity is medi-ated by the TGF-␤ responsive domain (TRD), which resides in the last 20 carboxyl-terminal amino acids of the CTF-1 TAD (14). To test whether the CTF-1 TRD might be also required for ⌬CaMKIV-mediated activation, we examined several previ- ously constructed GAL-CTF1 fusions (14) for ⌬CaMKIV responsiveness in NIH3T3 cells (Fig. 4B). As expected, the DNAbinding domain of GAL4 (GAL-DBD) did not mediate ⌬CaMKIV activation in this assay, while the activity of GAL 399 -499 was potently induced by ⌬CaMKIV. Interestingly, deletion of the CTF-1 TRD abolished kinase induction (Fig. 4B, GAL 399 -486), whereas its direct fusion to GAL-DBD was sufficient to confer full ⌬CaMKIV responsiveness (GAL 486 -499). A derivative mutated in all phosphorylation acceptor sites in the TRD could still be efficiently induced by ⌬CaMKIV (Fig.  4B, GAL 479 -499 3xM), as it is induced by TGF-␤ (14). In contrast, the CTF-2 TAD, which naturally lacks a TRD as a result of alternative splicing events (37), conferred little, if any, ⌬CaMKIVand TGF-␤-mediated induction (Fig. 4B, GAL 399 and Ref. 14). TGF-␤ was still able to increase the CaMKIVactivated levels of all TGF-␤-responsive GAL-CTF derivatives (for instance GAL 486 -499 and GAL 479 -499 3xM, Fig. 4B), even though the ability of TGF-␤ to induce GAL 399 -499 activity was reduced in cells overexpressing ⌬CaMKIV (from 8.8-to 1.8-fold, Fig. 4A). Thus, we conclude that the TGF-␤responsive domain mediates the ⌬CaMKIV-dependent transcriptional activation of CTF-1. Furthermore, similarly to TGF-␤ and calcineurin regulation, ⌬CaMKIV regulation is confined to specific CTF/NF-I species, such as CTF-1 but not CTF-2. Finally, our results indicate that TRD activation may occur in the absence of ⌬CaMKIV-mediated phosphorylation, implying that other ⌬CaMKIV-regulated proteins must be involved in this regulatory process.
Mouse CTF/NF-I Proteins Mediate Calcineurin-and CaMKIV-regulated Transcriptional Activation-To address the possibility that the endogenous CTF/NF-I polypeptides might be also induced by calcineurin and CaMKIV, we used a reporter promoter containing three high-affinity CTF/NF-I binding sites in front of the ␣-globin TATA box and the cat gene. As shown previously (15), this CTF/NF-I-responsive promoter is induced by TGF-␤ in transiently transfected NIH3T3 fibroblasts (Fig.  5A, p␣CAT-⌬87-3xAd). Interestingly, expression of either ⌬CNA or ⌬CaMKIV significantly induced basal CTF/NF-I transcriptional activity (Fig. 5A). ⌬CNA had no effect on CTF/NF-I induction by TGF-␤, whereas TGF-␤ induction was increased in cells expressing ⌬CaMKIV. In contrast, neither enzyme had any effect on the activity of a control promoter lacking CTF/ NF-I binding sites or on the activity and TGF-␤ induction of a distinct TGF-␤-responsive promoter (Fig. 5A, p␣CAT-⌬55 and p3TP-Lux). Gel mobility shift analysis of the endogenous CTF/ NF-I proteins indicated that ⌬CNA and ⌬CaMKIV did not significantly affect CTF/NF-I binding activity and/or expression levels (Fig. 5B). Thus, these data indicate that the transcriptional activity of endogenous mouse CTF/NF-I polypeptides is induced by expression of constitutive calcineurin and CaMKIV, which suggests that CTF/NF-I is a relevant transcriptional target for these calcium-regulated enzymes in vivo. However, as for the GAL4 fusion proteins (Fig. 4), ⌬CaMKIV did not prevent CTF/NF-I induction by TGF-␤, suggesting that CaMKIV is unlikely to be directly involved in this regulatory process.

DISCUSSION
Growth factors regulate gene expression by modulating the activity of target transcription factors through signal transduction cascades. Stimulation of collagen synthesis by mesenchymal cells is an important aspect of TGF-␤ regulatory action for tissue remodeling, wound healing, and the pathogenesis of some fibrotic disorders (1,2). A role for CTF/NF-I in collagen synthesis was initially proposed by Rossi et al. (16), who suggested that induction of the human collagen ␣2(I) promoter by TGF-␤ might involve CTF/NF-I binding sites (16). Indeed, TGF-␤ was subsequently demonstrated to induce the transcriptional activity of CTF-1, the prototypic member of the CTF/ NF-I family (14). Recently, a possible role for calcium in TGFaction was suggested, as the TGF-␤ type I receptor was shown to interact with FKBP12, a protein involved in the regulation of calcium homeostasis (9). Here, we investigated a possible role of free calcium signaling in the induction of CTF-1 transcriptional activity by TGF-␤.
Our results provide evidence for the potential relevance of calcium signaling in TGF-␤ action, as we found that (i) TGF-␤ induces calcium influx in NIH3T3 cells, (ii) it increases free cytoplasmic calcium concentration ([Ca 2ϩ ] c ) to about 150 nM, which is consistent with the requirements for the activation of These calcium ions may serve as the primary signal for the release of stored calcium from the endoplasmic reticulum (ER) in the cytoplasm of the cells, mediated by an intracellular calcium channel (RyR/IP 3 R). This leads to an increase in total cytoplasmic calcium concentration ([Ca 2ϩ ] c ), which can be prevented upon chelation of the extracellular calcium with EGTA. Increased [Ca 2ϩ ] c leads to the activation of calciumdependent signaling enzymes such as calcineurin (CN, pathway 1) and CaMKIV (pathway 2), which bind Ca 2ϩ and/or Ca 2ϩ complexed to calmodulin (CaM). These and other putative TGF-␤-regulated cascades (pathways 3 and 4) relay the TGF-␤ signals further downstream in the nucleus, where CTF-1 activation occurs (probably through the phosphorylation of a CTF-1-interacting protein(s), such as histone H3), leading to CTF-1-dependent chromatin remodeling (14) and eventually to the transcriptional induction of TGF-␤-responsive genes, such as collagen. FK506 complexed to FKBP12 binds CN and blocks its phosphatase activity, while thapsigargin inhibits the function of the ER calcium (SERCA) pumps and depletes the calcium pool, thereby blocking TGF-␤ action. The Ras/Raf pathway, which may mediate CTF-1 repression in response to TNF-␣, thereby antagonizing TGF-␤ induction (15), is also indicated. Note that the intracellular localization of CN and CaMKIV is hypothetical in this scheme. calcium-dependent signaling enzymes (31), (iii) TGF-␤ is unable to increase [Ca 2ϩ ] c in EGTA-pretreated cells, and (iv) TGF-␤ induction of CTF-1 transcriptional activity is prevented by thapsigargin, a compound that inhibits calcium signaling by specifically blocking calcium uptake in the ER (32,33). Altogether, these results indicate that TGF-␤ stimulates calcium influx in NIH3T3 cells, thus leading to an increase in the cytoplasmic calcium concentration that may in turn activate downstream calcium-dependent effectors such as calcineurin and CaMKIV (Fig. 6, pathways 1 and 2). Consistent with this possibility, constitutively active variants of CNA and CaMKIV specifically induce the basal transcriptional activity of both the CTF-1 TAD and the endogenous mouse CTF/NF-I proteins in NIH3T3 cells. However, these enzymes are unlikely to be directly involved in the TGF-␤ induction pathways, as (i) the calcineurin inhibitors FK506 and CsA do not inhibit TGF-␤ induction of the CTF-1 TAD, (ii) TGF-␤ still potentiates GAL 399 -499 transcriptional activity in ⌬CaMKIV and/or ⌬CNA co-expressing cells, and (iii) NIH3T3 cells do not express significant levels of the CaMKIV protein, although they do express CNA and CNB mRNAs (as estimated by reverse transcription PCR). 2 Our results rather imply that, if calcium signaling is indeed relevant for TGF-␤ induction of CTF-1, as suggested by the thapsigargin inhibitory effect, it may involve other calcium-regulated intermediates (Fig. 6, pathway 3). In any case, however, calcium signaling alone is unlikely to account for all of the TGF-␤ actions, as simple administration of calcium ionophores is unable to mimick TGF-␤ induction of CTF-1. 2 Therefore, we postulate the existence of yet an additional pathway linking CTF-1 activity to cellular signaling by TGF-␤ (Fig. 6, pathway 4).
⌬CaMKIV action (and also calcineurin-mediated CTF-1 induction, results not shown) specifically targets the previously identified TGF-␤-responsive domain of CTF-1 (14,15), and the catalytic activity of the kinase is required for this effect. This suggests that the kinase phosphorylates either the TRD itself or a TRD-interacting protein. However, phosphorylation-defective TRD variants still confer efficient TGF-␤ induction, as well as calcineurin-and ⌬CaMKIV-mediated activation. Thus, these effects must be mediated by a protein(s) interacting with the TGF-␤-responsive domain. One such protein could be histone H3, since it binds the CTF-1 TAD, and it has been proposed to be relevant for TGF-␤ induction (14). For instance, CaMKIV or a related kinase might regulate the interaction of the CTF-1 TAD with histone H3, in response to extracellular signaling. Consistent with this possibility, several growth factors and protein kinases have been found to induce histone H3 phosphorylation, and this effect has been correlated with changes in gene expression (38).
In summary, the results presented in this study identify CTF-1 as a novel molecular target of calcineurin and CaMKIV action, and they further argue for the potential relevance of calcium signaling in the regulation of extracellular matrix gene expression by TGF-␤. Analysis of CTF/NF-I regulation might thus provide insights in the control mechanisms of wound healing and extracellular matrix production by TGF-␤ and may thus help understand the role of TGF-␤ in the pathogenesis of fibrotic disorders.