Characterization of the c-Myb-responsive Region and Regulation of the Human Type I Collagen α2 Chain Gene by c-Myb*

We have characterized the role of c-Myb and B-Myb in the regulation of human type I collagen α2 chain gene expression in fibroblastic cells. We have identified four Myb-binding sites (MBSs) in the promoter. Transactivation assays on wild type and mutant promoter-reporter constructs demonstrated that c-Myb, but not B-Myb, can transactivate the human type I collagen α2 chain gene promoter via the MBS-containing region. Electrophoretic mobility shift assay experiments showed that c-Myb specifically binds to each of the four MBS; however, the mutagenesis of site MBS-4 completely inhibited transactivation by c-Myb, at least in the full-length promoter. In agreement with these results, c-myb−/− mouse embryo fibroblasts (MEFs) showed a selective lack of expression of type I collagen α2 chain gene but maintained the expression of fibronectin and type III collagen. Furthermore, transforming growth factor-β induced type I collagen α2 chain gene expression in c-myb−/− MEFs, implying that the transforming growth factor-β signaling pathway is maintained and that the absence of COL1A2 gene expression in c-myb−/− MEFs is a direct consequence of the lack of c-Myb. The demonstration of the importance of c-Myb in the regulation of the type I collagen α2 chain gene suggests that uncontrolled expression of c-Myb could be an underlying mechanism in the pathogenesis of several fibrotic disorders.

The myb oncogene family is composed of c-myb, A-myb, and B-myb genes, each encoding a distinct nuclear protein that displays transcription factor activity (1). The myb genes are structurally related, and initially they were believed to be expressed only in hematopoietic cells, where they were found to play a pivotal role in the regulation of growth and development (2)(3)(4)(5). However, the different myb family members have been reported to be expressed and to function in some other cell types. Hence, although little information is available relating to A-Myb, it has been demonstrated that c-Myb and B-Myb are expressed in epithelial cells (6) and fibroblasts (7). In the latter cells, the role of the Myb proteins has not been completely elucidated, and despite their structural homology and similar patterns of expression, B-Myb and c-Myb may exert opposite effects through the repression or activation of the same gene(s) (9 -13). With regard to c-Myb, it has been demonstrated that it can induce insulin-like growth factor-1-independent growth (14) and that it can control fibroblast proliferation through the regulation of the intracellular Ca 2ϩ concentration (15) as well as the expression of cell cycle-related genes, such as proliferating cell nuclear antigen (16).
c-Myb protein can activate or repress gene transcription by binding directly to the promoters of genes as in the case of mim-1 (17), cdc2 (18), c-myc (19,20), CD4 (21), the human T-cell lymphotrophic virus, type I, long terminal repeat (22), and c-myb itself (23). However, despite intensive investigations, not many genes regulated by c-Myb have been identified so far.
Following the finding of abnormal expression of the c-myb gene in quiescent scleroderma fibroblasts, a disease characterized by an augmented production of extracellular matrix protein, we have speculated that c-Myb but not B-Myb could be involved in the up-regulation of the type I collagen promoter (8). However, the employment of animal and not human type I collagen promoters, the use of the 3T3 fibroblast cell line rather than human fibroblasts, and the lack of identification and characterization of the region of the promoter involved in c-Myb binding has left open the issue of whether c-Myb can modulate collagen gene expression (24).
In view of the potential relevance of this finding for the understanding of the pathogenesis of scleroderma and of fibrotic disease in general, we have therefore decided to elucidate more exhaustively the role of c-Myb and B-Myb in the expression of the human type I collagen gene. We have cloned the type I collagen ␣2 chain (COL1A2) promoter and screened it for the presence of potential Myb-binding sites (MBSs). 1 Transactivation of the COL1A2 promoter by c-Myb and B-Myb has then been investigated in human fibroblasts and human fibroblast cell lines, and the regions of the promoter that are critical for Myb-mediated transactivation have been determined. Finally, the requirement for c-Myb in the expression of the type I collagen ␣2 chain gene has been demonstrated by comparison of embryonic fibroblasts derived from wild type and c-myb Ϫ/Ϫ mice and of normal human skin fibroblasts overexpressing c-Myb or a dominant negative c-Myb derivative.

EXPERIMENTAL PROCEDURES
Computer Analysis of the Putative Myb-binding Sites Contained in the COL1A2 Promoter-The whole DNA sequence of the COL1A1 and COL1A2 promoters (GenBank TM accession numbers J03559, U06669, and AF004877, respectively) was subjected to computer analysis and screened for putative MBSs using the software MatInspector version 2.2 (25). The computer analysis utilized matrices derived from the published MBSs consensus sequence (26 -28), and results were expressed in matrix similarity, where a value of 1 corresponds to complete homology.
Cloning of the Human COL1A2 Promoter-The 5Ј-flanking region of human COL1A2 gene, spanning from Ϫ2430 to ϩ5 bp (according to the GenBank TM accession number AF004877) containing a CAAT-binding site (Ϫ220) and a TATA box (Ϫ170), was cloned from genomic DNA of human fibroblasts using a PCR approach. Briefly, fibroblast genomic DNA was extracted from monolayer cultures of normal human fibroblasts using a commercial kit (Qiagen) according to the manufacturer's instructions. Genomic DNA (250 ng) was amplified in a final volume of 50 l containing 50 mM KCl, 10 mM Tris-HCl, 1.5 mM MgCl 2 , 0.2 mM each dNTP, and 15 pmol of forward and reverse primers (COL-F, TTACCACCCTGAGTCATTTTGC, and COL-B, GCACTTAGACATG-CAGACTCCT), and 2 units of LA-Taq polymerase (Takara). PCR (20 cycles of 1 min at 95°C, 1 min at 55°C, and 3 min at 70°C) was carried out in a PTC-200 thermal cycler (MJ Research). The resulting DNA fragment of 2430 bp was extracted from a 1% agarose gel, purified using Geneclean kit (Qiagen) and cloned within the multiple cloning site region (mcsr) of the pT-Adv vector (Clontech) using standard ligation procedure. Both strands of the COL1A2 promoter inserted in the recombinant plasmid pTCOL1A2-P were extensively sequenced to exclude the possibility of random mutations inserted by PCR. The COL1A2 promoter was further digested from pTCOL1A2-P using SacI and EcoRV, gel-purified, and subcloned between the SacI and SmaI sites upstream of the firefly luciferase gene of the Promega vectors pGL3-Basic (pGbC1A2-P) and pGL3-Enhancer (pGECA2-P) in which the SV40 enhancer sequences are located downstream of the mcsr. Briefly, 100 ng of COL1A2 promoter DNA was mixed with 30 ng of vector in 66 mM Tris-HCl, pH 7.6, 6.6 mM MgCl 2 , 10 mM DTT, 0.1 mM ATP, 2 M Hexamino-CoCl, and 450 units of T4 DNA ligase. The ligation mix was incubated at 14°C overnight, and 3 l of the mix were used to transform DH5-␣ competent bacteria.
Plasmids-The pGbC1A2-P and p-GECA-P recombinant plasmids were derived as described above. The pGL-KHK plasmid was created by digesting KHK-CAT (24) with BamHI and HindIII. The resulting DNA fragment containing the KHK synthetic promoter was purified by gel electrophoresis and blunted in a mix containing 100 M of each dNTP, 0.01% BSA, 33 mM Tris acetate, pH 7.9, 66 mM potassium acetate, 10 mM magnesium acetate, 0.5 mM DTT, and 4 units of T4 DNA polymerase. The blunted KHK promoter was then subcloned into the SmaI site of the pGL3-Basic mcsr, upstream from the firefly luciferase gene.
The pQCM plasmid, used for the production of the c-Myb recombinant protein, was obtained by digesting pSGC-myb, carrying the fulllength coding region of the human c-myb gene, with NcoI and BglII and subcloning the resulting c-myb gene fragment (ϩ1 to ϩ1200 bp) upstream and in-frame with a His 6 tag into the NcoI/BglII sites of the pQE-60 plasmid mcsr (Qiagen, Hilden, Germany). This c-myb DNA fragment contained the sequences encoding the R1-R3 domains that are necessary for c-Myb DNA binding activity. The pSGC and pSGB plasmids, containing the entire coding region of the c-myb and B-myb genes, respectively, and the pCys 130 plasmid, carrying the entire coding region of c-myb gene mutated at codon 130, have already been described (24). The pGREMyb plasmid, containing the full coding region of the c-myb gene, and the pGREMen plasmid, in which sequences encoding the R2 and R3 domains of c-Myb and the Drosophila engrailed gene encoding the alanine-rich repressor domain (15), are linked (a kind gift of Prof. M. Simons, Angiogenesis Research Center, Cardiovascular Division, Boston).
All recombinant plasmids were grown and further purified using the Endofree maxiprep method (Qiagen, Hilden, Germany).
Nested Deletions of the Human COL1A2 Promoter-The pTCOL1A2-P plasmid was used in PCRs to obtain the promoter deletions. Nested primers starting from different regions of the COL1A2 promoter (del1F-1500, AGCCTTTCAAACCTAGGGCCTG; del 2F-1269, GCCTCAGCAAAGGCAAGCTAG; del3F-950, TGGAGCCCTCCACCC-TACAA; del4F-575, GGACAGCTCCTGCTTTATCG; and del5F-290, TTCGCTCCCTCCTCTGCGCCC) and a backward primer (GCCCATC-TGCAGAATTCGGCTT) were used in PCRs to generate five different DNA fragments spanning from Ϫ1500 to Ϫ290. Because the first two MBSs are closely located at position Ϫ1025 and Ϫ1045, it was not possible to separate them efficiently, and therefore they were both deleted in the pGLD-1269 plasmid. Template DNA (250 ng) was amplified in a final volume of 50 l containing 50 mM KCl, 10 mM Tris-HCl, 2 mM MgCl 2 , 0.2 mM each dNTP, 15 pmol of each primer, and 2 units of LA-Taq polymerase (Takara, Otsu, Shiga, Japan). PCRs (1 min at 95°C, 1 min at 55°C, and 3 min at 70°C) were carried out in an MJ Research PTC-200 thermal cycler for 20 cycles. Each fragment was gel-purified and subcloned into the mcsr of pT-Adv vector (Clontech, Palo Alto, CA) and thereafter into the SacI-SmaI sites of the pGL3Basic plasmid mcsr to obtain the pGLD-1500, pGLD-1269, pGLD-950, pGLD-575, and pGLD-290 recombinant plasmids, respectively. The pGLD1095 plasmid was obtained by the digestion of pGbC1A2-P with NcoI, and the resulting COL1A2 promoter fragment, spanning from position Ϫ1095 to the NcoI site of the pGL3-Basic vector, was subcloned within the NcoI site of the pGL3-Basic vector mcsr. All plasmids were extensively sequenced to check for mutations introduced by PCR.
Site-directed Mutagenesis-The pGbC1A2-P and pGL-D1095 plasmids, containing only one MBS at position Ϫ1000, were used as templates for site-directed mutagenesis. The sequences of the primers used are as follows: MBM-1, CGGTCTCCAGGTCGATATCAGTCGTGT-CGGAGTGCCAG; MBM-2, CGGTCTCCAGGTCACTAGTAGTCGTGT-CGGAGTGCCAG; MBM-3, CGGTCCCAGGTCCGCGGTAGTCGTGTC-GGAGTGCCAG, both in the sense and antisense orientations and each incorporating a restriction enzyme (EcoRV, SpeI, and SacII, respectively) for the screening of recombinant mutated plasmids. Only the MBM-3 primers were used to mutagenize the MBS-4 of the pGbC1A2-P plasmid (containing the full-length COL1A2 promoter). The site-directed mutagenesis was achieved using the "Ex-Site PCR-based Sitedirected Mutagenesis" kit (Stratagene, La Jolla, CA). Each primer was phosphorylated at 37°C for 30 min in 1ϫ kinase buffer (Takara, Otsu, Shiga, Japan) containing 10 units of T4 polynucleotide kinase (Takara, Otsu, Shiga, Japan). 30 pmol of phosphorylated sense and antisense primers were then used in a mix containing 400 ng of template DNA, 5 units of Exsite DNA polymerase blend, 0.2 mM dNTPs, 20 mM Tris-HCl, pH 8.8, 10 mM KCl, 10 mM (NH 4 ) 2 SO 4 , 2 mM Mg SO 4 , 0.1% Triton X-100, 0.1 mg/ml BSA. The PCR was carried out in an MJ Research PTC-200 Thermal Cycler. At the end of the reaction, 10 units of DpnI restriction enzyme and 2.5 units of Pfu polymerase were added, and the mix was incubated at 37°C for 2 h and 72°C for 1 h. An aliquot of PCR products was further ligated at 14°C overnight with 5 mM rATP and 4 units of T4 DNA ligase in 10 mM Tris-HCl, pH 8.8, 5 mM KCl, 5 mM (NH 4 ) 2 SO 4 , 0.1% Triton X-100, 0.1 mg/ml BSA. Ten microliters of the reaction were used to transform XL-1-Blue competent bacteria. All recombinant plasmids were screened for the inserted mutations using the above-mentioned restriction enzymes, and thereafter both strands were extensively sequenced using a CEQ-2000 DNA sequencing instrument (Beckman Instruments, Palo Alto, CA).
Cells and Cell Cultures-Normal human skin fibroblasts (NSF) were obtained from punch biopsies taken from the forearms of healthy subjects. Primary explant cultures were established in 25-cm 2 culture flasks in minimum Eagle's medium containing 10% FCS, 2 mM glutamine, penicillin (100 units/ml), streptomycin (100 g/ml), and amphotericin B (0.25 mg/ml). Minimum Eagle's medium with these supplements is hereafter referred to as "culture medium." Normal fibroblast human cell lines WS-1, HUDE, and HFL-1 were purchased from ATTC and grown in Dulbecco's modified Eagle's medium containing 10% FCS, 2 mM glutamine, penicillin (100 units/ml), streptomycin (100 g/ml), and amphotericin B (0.25 mg/ml). Normal human fibroblastic cell lines transfected with pGreMyb and pGreMen plasmids were grown in culture medium containing 0.2 M dexamethasone (15). Monolayer cultures were maintained at 37°C in 5% CO 2 . Fibroblasts at the fifth passage were used for all experiments. c-myb knock-out (c-myb Ϫ/Ϫ ) (29) embryonic fibroblasts (MEFs) were cultured in Dulbecco's modified Eagle's medium containing 20% FCS, 2 mM glutamine, penicillin (100 units/ml), streptomycin (100 g/ml), and amphotericin B (0.25 mg/ml). Fibroblasts at the fourth passage were used for Northern blot analysis. For the stimulation experiments of wt and c-myb Ϫ/Ϫ MEFs, recombinant TGF-␤ was used at a final concentration of 2 ng/ml.
Cell Transfection-For transfection experiments, confluent fibroblasts were harvested with trypsin and plated in 60-mm dishes in culture medium. After 24 h, the medium was discarded, replaced with fresh culture medium, and the cells transfected. Transfection experi-ments were carried out in triplicate using a liposomal method (Effectene, Qiagen) in the presence of an Enhancer and a DNA condensation buffer (according to the manufacturer's instructions). Plasmid DNA was purified with Endofree maxi-kit (Qiagen) to remove bacterial endotoxins. Cells were transfected with a mix containing 0.7 g of reporter plasmid Ϯ 1.5 g of effector plasmid, 8 l of Enhancer, 25 l/g DNA of Effectene, and culture medium containing 10% FCS. To control transfection efficiency, the pRLSV reporter plasmid (Promega), coding for Renilla luciferase, was used in all experiments at a concentration of 0.05 g/60-mm dish. After 12 h, the medium was changed, and 24 h later the cells were harvested and lysed in 100 l of PBL buffer (Promega). Following the manufacturer's instructions, 20 l of the supernatant were used in the "Dual Reporter Luciferase Assay" in which the activities of firefly and Renilla luciferases are sequentially measured from a single sample using two different enzymatic substrates. The luciferase activities of the samples were measured with a TD-20/20 luminometer (Turner design), and the ratio of Renilla to firefly luciferase values was used to normalize all of the transfection experiments.
Western Blot Analysis-Cleared cell lysates, obtained from NSF cells that had been cultured and transfected as described above, were precipitated with 20% trichloroacetic acid, and aliquots containing the same amount of protein were subjected to SDS-PAGE on 10% gels, according to standard procedures. The transfer from the gels to nitrocellulose membranes (Bio-Rad) was achieved using a Trans Blot Cell apparatus (Bio-Rad), and blotting was performed at 4°C, 30 V overnight in 25 mM Tris, 192 mM glycine, 20% (v/v) methanol, 3.5 mM SDS. To ensure that comparable amounts of proteins had been transferred to the nitrocellulose membranes, proteins were revealed by staining with 0.05% (v/v) Ponceau S (Sigma) for 1 min. The blots were then rinsed in TBS (50 mM Tris, 170 mM NaCl, 0.2% (v/v) Tween 20, pH 7.5) and incubated for 4 h in Blotto solution (TBS containing 5% non-fat dried milk), before incubating overnight at 4°C with rabbit anti-c-Myb (Geneka) diluted 1:1000 in Blotto solution. The blots were rinsed in TBS with several changes and then incubated for 60 min at room temperature in a horseradish peroxidase-labeled donkey anti-rabbit IgG (Amersham Biosciences) diluted 1:1000 in TBS. After further washing, bound antibodies were detected using enhanced chemiluminescence detection reagents (Amersham Biosciences). Images were analyzed using a Molecular Images FX (Bio-Rad).
Production of the Recombinant Myb-HIS Fusion Protein-pQCM and the control plasmid pQE40 (Qiagen) were used to transform pREP-MC5 chemicompetent bacteria (Qiagen, Hilden, Germany). Single bacterial colonies were picked from LB plates, grown in an orbital shaker at 37°C in NZYC broth, and induced with 0.5 mM isopropylthiogalactoside for 5 h. An aliquot of the culture was lysed in 8 M urea and subjected to Western blot analysis to check the expression of recombinant proteins using the Penta-HIS antibody (a mouse monoclonal IgG1 antibody raised against the His 6 tag coupled to horseradish peroxidase, Qiagen, Hilden, Germany). Western blot analyses were carried out as described above and following the manufacturer's experimental conditions with minor variations. The batch purification of Myb-HIS recombinant protein was achieved using an extraction method under native conditions. Briefly, bacterial pellets were resuspended in 5 ml/g of B-PER II reagent (Pierce) containing 20 mM imidazole and 0.1 mM phenylmethylsulfonyl fluoride, briefly vortexed, and then incubated at room temperature for 30 min. Thereafter, the solution was centrifuged at 4°C for 30 min, and the cleared lysate was recovered. 2 ml of 50% nickel-nitrilotriacetic acid slurry matrix (Qiagen, Hilden, Germany) were added to the lysate and gently mixed by shaking at 4°C for 1 h. The lysate/ nickel-nitrilotriacetic acid mixture was loaded onto a column, washed twice with 50 mM NaH 2 PO 4 , 300 mM NaCl, and Myb-HIS recombinant protein was eluted with 50 mM NaH 2 PO 4 , 300 mM NaCl, 250 mM imidazole. The purity of Myb-HIS recombinant protein in the eluate was assessed by Western blot analyses using the Penta-HIS antibody described above.
Electrophoretic Mobility Shift Assay (EMSA)-EMSA experiments were carried out using the Myb-HIS recombinant protein and four double-stranded oligonucleotides (MBS-1, CCTCTCCCTAGTAGGGAG-TGGAGGGTTGGATGGAGGCGGC; MBS-2, TGGAGGCGGCCAGAG-AAGAGGGAAGTTGGGTGCTGGGGAGAGAGTTAACA; MBS-3, GGA-CCGGGGGGCTCACGGGAGGGTTGAAGGGTCCAGCTC; and MBS-4, GTTCTCGGTCTCCAGGTCGGTTGGAGTCGTGTCGGACTGC), the sequences of which (listed from 5Ј to 3Ј) are complementary to COL1A2 promoter regions that contain each of the four MBS (see Fig. 1A). The gel-purified KHK promoter, obtained by the digestion of the pGL-KHK plasmid with HindIII and BamHI, was used as a cold competitor. Anti-c-Myb polyclonal antibody (Geneka, Montreal, Quebec, Canada), raised against a peptide derived from amino acid residues 2-16 of human c-Myb, was used to block the binding activity of Myb-HIS recombinant protein to its MBS. The Penta-HIS antibody described above was used to supershift the DNA-Myb-HIS complex. The complementary single strand oligonucleotides were annealed at 95°C for 3 min, at 55°C for 2 min, and at 37°C for 15 min in 10 mM Tris-HCl, pH 7.5, 10 mM MgCl 2 , 50 mM NaCl. Three hundred nanograms of doublestranded oligonucleotides were then labeled with [␣-32 P]dCTP (3000 Ci/mmol) using 2.5 units of Klenow DNA enzyme and purified on a silica column (Qiagen). The binding reaction was carried out at room temperature for 30 min using 25,000 cpm of labeled DNA, 1 g of poly(dI-dC), and 10 g of Myb-HIS in 50 mM NaCl/20 mM Hepes, pH 8, 1 mM EDTA, 10 mM DTT, 0.5% non-fat dried milk, 5% glycerol in a final volume of 12 l. For supershift and blocking experiments the binding reaction was 2 l of Penta-HIS or anti-c-Myb polyclonal antibody, respectively, and was preincubated at 4°C for 15 min in the absence of the target DNA. After the incubation period, 1.5 l of loading buffer (250 mM Tris-HCl, pH 7.5, 40% glycerol) were added, and the samples were immediately loaded onto a 6% acrylamide Retardation Gel (NOVEX). Gel electrophoresis was carried out in a cold room at 100 V for 2 h. The gel was then fixed with 10% acetic acid, 10% methanol, dried, and exposed using the FX-Molecular Imager system (Bio-Rad).
Northern Analysis-The plasmid used for the detection of the type I collagen ␣2 chain mRNA (Hf32) and type III collagen were a kind gift of Dr. C. M. Lapiere (Laboratoire de Biologie des Tissue Conjonctifs, University of Liege, Belgium). The fibronectin and TGF-␤ cDNAs were purchased from ATCC. The probe was labeled with [ 32 P]dCTP by standard random priming procedures to specific activities of 5 ϫ 10 8 cpm/g DNA. Total cellular RNA was extracted according to the guanidine isothiocyanate-cesium chloride method. Eight micrograms of total RNA were loaded in each lane of a 1% agarose-formaldehyde gel and transferred to nylon membranes by standard blotting procedures. RNA was fixed to the membranes by baking or UV cross-linking. Prehybridization was performed for 7 h at 60°C in 1 M NaCl, 1% SDS, 10% dextran sulfate, and 100 mg/ml denatured salmon sperm DNA. Hybridization was performed overnight at 60°C in the same buffer by adding 1.2 ϫ 10 6 cpm/ml of labeled probe. The filters were washed twice for 10 min at 60°C with 2ϫ SSC, 1% SDS, 15 min with 0.5ϫ SSC, 0.5% SDS and finally twice at room temperature with a large volume of 0.5ϫ SSC. The blots were briefly dried, exposed in screen cassette Bio-Rad, and hybridization signals obtained using a Molecular Images FX (Bio-Rad).
Semiquantitative Analysis of c-myb RNA by PCR in MEFs-Total cellular RNA was extracted from MEFs as described above. Two micrograms of total RNA were directly reverse-transcribed in 5 mM MgCl 2 , 50 mM KCl, 10 mM Tris-HCl, pH 8.3, 2.5 M random hexamers, 1 mM dNTP, 1 unit/ml RNase inhibitor, and 2.5 units/ml Moloney murine leukemia virus-reverse transcriptase. The samples were incubated for 10 min at room temperature and then for 45 min at 42°C. Three microliters of the reverse transcription reactions were amplified by PCR in 2 mM MgCl 2 , 50 mM KCl, 10 mM Tris-HCl, 0.2 mM each dNTP, 2.5 units/ml Taq DNA polymerase, and 5 ng/ml of each primer: c-myb 1, ACTCAACTGCCCAATGAAGTCG; c-myb 2, TTCCTGTTCCACCTT-GCG; actin 1, ATCGTGGGCCGCCCTAGGCACCA; actin 2, TTGGCCT-TAGGGTTCAG GGGG. PCR was performed in an MJ-PT200 thermal cycler using the following cycle: 30 min at 95°C, 30 min at 58°C, and 30 min at 72°C, for 18 cycles. The predicted sizes of the RT-PCR products were 445 and 244 bp, respectively. An aliquot of each reaction was run in a 1.5% ultrapure agarose gel (Invitrogen) and transferred to nylon membranes by standard Southern blotting procedures.

RESULTS
The Promoter of the Human Type I Collagen ␣2 Chain Gene Contains Four Putative Myb-binding Sites-In a previous report (8), we have shown that c-Myb can up-regulate rat ␣1(I) collagen and mouse ␣2(I) collagen promoters by 6 -10-fold, whereas B-Myb was inactive. Here we have tried to ascertain whether c-Myb could also up-regulate human type I collagen gene expression. By using the computerized software MatInspector version 2.2, we have analyzed the DNA sequence of the promoter of the human type I collagen ␣2 chain (COL1A2-P), and we found four putative MBSs. These are located between positions Ϫ1400 to Ϫ1000 (defined as a "Myb-Responsive Region" (M-RR)) and display a high degree of homology with the published consensus sequence, the degree of matrix similarity ranging from 0.87 to 0.93, where 1 corresponds to complete homology (Fig. 1A).
When we extended the software analysis to the DNA sequence of the human COL1A1 gene promoter, from the transcription start site to Ϫ2000 bp, we did not find myb-binding sites. Thus, we have focused our investigation on the regulation of COL1A2 gene expression by c-myb.

c-Myb, but Not B-Myb, Up-regulates the Promoter of the Gene Encoding the ␣2 Chain of Human Type I Collagen in Human
Fibroblasts-To demonstrate that c-Myb is able to transactivate the human COL1A2 promoter, we co-transfected the reporter plasmid pGbC1A2-P (containing the COL1A2 promoter inserted upstream from the firefly luciferase gene, Fig. 1B) and either the pSGC or pSGB CMV promoter-driven plasmids which express human c-Myb or B-Myb, respectively. Western blot analysis confirmed that c-Myb was efficiently expressed from the pSGC vector (Fig. 2B). In a different set of experiments, the pGECA-2 plasmid, in which the COL1A2 promoter is inserted upstream an SV40 enhancer cassette, was used as the reporter vector (Fig. 1C). The transfection assays demonstrated that c-Myb, but not B-Myb, up-regulates expression driven by the COL1A2 promoter by 6 -8-fold in the WS-1 and HUDE cell lines and in NSF. A stronger degree of transactivation (10 -15-fold) was obtained in the HFL-1 cell line ( Fig. 2A).
Furthermore, in the same cell types we performed co-transfection assays using the expression plasmids pSGC and pSGB and the control reporter plasmid pGL-KHK, which carries the KHK synthetic promoter inserted upstream from the luciferase gene. In this case, both c-Myb and B-Myb were able to stimulate luciferase expression by 8-fold (Fig. 2C). Moreover, cotransfection assays using pSGC and pSGB together with the reporter plasmid pGbC1A2-P demonstrated that B-Myb could partially inhibit transactivation of the COL1A2 promoter driven by c-Myb (Fig. 2D).
The strength of the Myb-induced transactivation was also evaluated by transfecting the pSGC or pSGB expression vectors together with the reporter pGECA2-P in which the COL1A2 promoter was linked to SV40 enhancer sequences. Stimulation of this promoter/enhancer combination by c-Myb was equivalent to that seen with the pGbC1A2-P reporter (data not shown), thus demonstrating that enhancer region(s) are not required for the effect of c-Myb on the COL1A2 promoter. NcoI, position of the restriction enzyme site NcoI used to obtain the pGL-D1095 plasmid (see "Experimental Procedures" for more details). The table shows from left to right: the nucleotide sequences of the four MBSs (the core sequence regions are in boldface and underlined), the nucleotide position within the COL1A2 promoter, and the values of matrix similarities to the MBS consensus sequences, where a value of 1 corresponds to complete homology. B, the recombinant plasmid pGbC1A2-P obtained by cloning the COL1A2 promoter between the SacI-SmaI sites of the pGL3-Basic mcsr, upstream the firefly luciferase gene (LUCϩ). C, the recombinant plasmid pGECA2-P obtained by cloning the COL1A2 promoter between the SacI-SmaI sites of the pGL3-Enhancer mcsr, upstream the firefly luciferase gene (LUCϩ) and an SV40 enhancer cassette (SV40-Enh).

FIG. 2. Transient transactivation of the COL1A2 promoter by c-Myb or B-Myb in human fibroblasts. A, co-transfection of human fetal lung (HFL), human dermal embryonic cells (HUDE and WS1)
, and NSF with the COL1A2 promoter-reporter (COL1A2-P) and c-Myb or B-Myb expression vectors. The results represent the mean of three separate experiments. All transfections also included the pRL-SV vector that encodes Renilla luciferase. The ratios of Renilla to firefly luciferase activities obtained with the "Dual Reporter Luciferase Assay" were used to normalize all the transfection experiments. B, Western blot analysis of NSF transfected with COL1A2 promoter-reporter (COL1A2-P) Ϯ the c-Myb expression vector pSGC. Affinity-purified rabbit anti-c-Myb polyclonal antibody was used as the primary antibody, and Western blot detection was carried out following the manufacturer's instructions. K562, nuclear extract prepared from K562 human chronic myelogenous leukemia cell line used as positive control for c-Myb. C, co-transfection of NSF with pGL-KHK plasmid (carrying the KHK synthetic promoter) and c-Myb or B-Myb expression vectors. The results represent the mean of three distinct experiments. Normalization for transfection efficiency was performed as described in A. D, co-transfection of NSF with COL1A2 promoter-reporter (COL1A2-P) together with c-Myb (pSGC) and B-Myb (pSGB) expression vectors at different molar ratios (e.g. 1:1 ϭ 1 g/1 g). The results represent the mean of three distinct experiments. Normalization for transfection efficiency was performed as described in A.
These results clearly prove that the human COL1A2 promoter is strongly and specifically transactivated by c-Myb in several human fibroblastic cell types, whereas B-Myb can potentially modulate this transactivation through partial inhibition of the positive effect of c-Myb.
Transactivation of the Human Type I Collagen ␣2 Chain Gene Promoter Requires the Presence of the 400-bp Myb-responsive Region (M-RR)-Next we investigated the role that the four MBSs of the 400-bp COL1A2 promoter Myb-responsive region (M-RR) might play in c-Myb-mediated transactivation. We created a set of pGL plasmids that carry different deletions of COL1A2 promoter from Ϫ1500 to Ϫ290 in order to remove progressively each MBS (Fig. 3A). Transfection assays utilizing the pSGC expression vector together with promoter constructs containing deletions to Ϫ950, Ϫ575, and Ϫ250 demonstrated that one or more MBSs were sufficient to allow c-Myb-dependent activation but that the loss of all four MBSs completely ablated the response to the transactivator (Fig. 3B).
These results prove that the presence of the 400-bp M-RR in the COL1A2 promoter is necessary for c-Myb-dependent transactivation and suggest that each MBS could be important in mediating the effect.

Mutagenesis of MBS-4 or Mutation of c-Myb Abrogates Mybdependent Transactivation of the Human Type I Collagen ␣2
Chain Gene Promoter-To show further that c-Myb acts upon the MBS in the COL1A2 promoter, mutagenesis experiments were performed on MBS-4 at position Ϫ1000. MBS-4 (GGT-TGG) was mutated in the context of the following: (i) plasmid pGbC1A2-P, which contains all four MBS, to CCGCGG (MBM-3); and (ii) plasmid pGLD-1095, which contains only MBS-4, to sequences GATATC, ACTAGT, or CCGCGG (MBM-1, MBM-2, or MBM-3, respectively). In transfection assays, c-Myb expressed from pSGC was unable to stimulate luciferase expression from the mutated pGLD-1095-based plasmids, demonstrating the requirement for the integrity of MBS-4 (Fig. 4A). Moreover, mutation of MBS-4 alone in the context of the fulllength COL1A2 promoter strongly decreased up-regulation by c-Myb (2.3 times less than on the wt promoter, Fig. 4A).
To demonstrate that c-Myb is directly involved in transactivation of the COL1A2 promoter, we used an expression vector, pCys130, which encodes a mutated c-Myb protein containing a Cys to Ser mutation at position 130 that abrogates the binding ability of c-Myb to DNA (30,31). The mutated c-Myb protein was unable to activate luciferase expression from either the wild type pGLD-1095 or the full-length COL1A2 promoter (Fig.  4B).
These data demonstrate that both MBS integrity and fully functional c-Myb protein are necessary for the c-Myb-mediated transactivation of the COL1A2 promoter. Furthermore, of the four MBS, MBS-4 seems to play the pivotal role in COL1A2 promoter transactivation by c-Myb.
c-Myb Protein Binds to the MBSs DNA Sequences in the Promoter of the Human Type I Collagen ␣2 Chain Gene Promoter-In view of results described above, we have investigated the DNA binding capacity of c-Myb to the MBSs DNA sequences that are present in the M-RR of the COL1A2 promoter. For this purpose we produced a c-Myb recombinant protein, truncated at position ϩ1200, conjugated to a His 6 tag. The MBS core sequence present in each construct is indicated, and the mutated residues are shown in lowercase letters. B, co-transfection of NSF with the full-length COL1A2 promoter-reporter construct (COL1A2-P) and the expression vector pSGC encoding either wild type c-Myb or the version mutated at Cys-130. Results represent the mean of three distinct experiments. Normalization for transfection efficiency was performed as described in the legend to Fig. 2A. This protein, Myb-HIS, contains the three repeat domains (R1, R2, and R3) that are necessary for DNA binding ability (32)(33)(34). Gel retardation experiments (EMSA) demonstrated that Myb-HIS is able to bind to the KHK synthetic promoter (Fig. 5A) and to a labeled double-stranded oligonucleotide containing the sequence of MBS-4 (Fig. 5B). The Myb-HIS-DNA complexes were similar to those obtained using nuclear extracts from the K562 cell line (Fig. 5, A and B, 2nd lane). The binding activity was abrogated by incubating Myb-HIS with a 100-fold molar excess of cold competitor (Fig. 5, A and B, 4th lane) and with an anti-c-Myb polyclonal antibody raised against the region of the protein that contains the R1-R3 domains (Fig. 5, A and B, 5th  lane). The same results were obtained using three different double-strand oligonucleotides, each one containing one of the three other MBS (MBS1-3) from the COL1A2 promoter (data not shown).
To confirm the binding specificity of Myb-HIS to the MBS, we carried out binding reactions using the Penta-HIS antibody, raised against the His 6 tag of the recombinant protein, to supershift the Myb-HIS-DNA complex. The reason for using this antibody was that the available anti-c-Myb antibodies were directed against the COOH terminus of the protein, which is deleted in our recombinant Myb-HIS, and because the histidine tag does not usually participate in the function of recombinant proteins (35)(36)(37)(38)(39)(40). Moreover, the Penta-HIS antibody easily detected the recombinant protein in Western blot analysis (data not shown). Gel retardation experiments demonstrated that the Penta-HIS antibody was able to supershift the Myb-HIS-DNA complex (Fig. 5C, 4th lane). Binding reactions between Myb-HIS protein and double-stranded oligonucleotides containing the three mutated MBS described above (MBM1-3) demonstrated that the mutated sites could no longer specifically bind Myb-HIS (Fig. 5C, 5th to 7th lanes). These data demonstrate that the Myb-HIS recombinant protein, containing the R1-R3 DNA binding domains conjugated to a His 6 tag, can specifically bind to both the KHK synthetic promoter and to several double-stranded oligonucleotides that contain the COL1A2 promoter MBSs and that the Myb-HIS-DNA complex can be supershifted by the Penta-HIS antibody. Specificity of binding is indicated by the fact that binding can be blocked by the following: (a) a molar excess of cold competitor; (b) an anti-c-Myb antibody directed against the R1-R3 DNA binding domains; and (c) mutation of the MBSs. All the above evidence clearly demonstrates that Myb-HIS binds to the human type I collagen ␣2 chain gene promoter, requiring specific interaction between the DNA-binding domain of c-Myb and the MBS.
The Lack of Transcription of Type I Collagen ␣2 Chain mRNA in c-myb Ϫ/Ϫ Embryonic Fibroblasts Can Be Restored by Ectopic Expression of c-Myb-To confirm the relationship between c-Myb and type I collagen ␣2 chain gene expression, we studied mouse embryonic fibroblasts derived from embryos homozygous for an inactivating mutation in the c-myb gene (c-myb Ϫ/Ϫ MEFs). Although c-myb Ϫ/Ϫ embryos do not survive beyond day 16 of gestation because of a severe impairment of fetal hematopoiesis, MEFs can be grown in cultures for several weeks (15) and are suitable for the study of extracellular matrix gene expression. The results of a representative Northern blot analysis of mRNA from MEFs c-myb Ϫ/Ϫ cells (out of a total of five experiments) are shown in Fig. 6A. Cells lacking c-Myb expressed very low levels of ␣2(I) collagen mRNA when compared with wild type cells, a faint band being visible only after long exposures of the autoradiograms. This defect is specific because the same cells maintained the expression of other important extracellular matrix proteins, such as type III collagen and fibronectin (Fig. 6A). The transfection of the c-myb Ϫ/Ϫ MEFs with a plasmid encoding c-Myb restored the expression of the ␣2(I) collagen mRNA to a level comparable with that of the wt MEFs (Fig. 6B), implying that the lack of the constitutive expression of ␣2(I) collagen gene is caused by the lack of c-Myb expression and not by a global impairment of other signaling pathways. This latter conclusion is further supported by the normal expression of the ␣2(I) collagen gene, but not c-myb, when the c-myb Ϫ/Ϫ MEFs are stimulated by TGF-␤ (Fig.  6C). Wild type MEFs exhibited increases in both c-myb and COL1A2 gene expression after TGF-␤ stimulation (Fig. 6C), thus mimicking the response seen in human dermal fibroblasts (24).
Normal Human Skin Fibroblasts Expressing a Dominant Negative Version of c-Myb Lose the Ability to Express the Type I Collagen Gene-Finally, to validate and confirm in human cells the data obtained in c-myb Ϫ/Ϫ MEFs, we transfected NSF with a plasmid constitutively expressing wild type c-Myb (pGREMyb) or a dominant negative form of c-Myb (pGREMen). NSF transfected with pGREMyb displayed an augmented expression of the type I collagen gene, whereas they lost the ability to express the gene when transfected with pGREMen (Fig. 7).
This demonstrates that c-Myb also plays a crucial role in the regulation of the type I collagen gene in human cells.

DISCUSSION
Type I collagen is a heterotrimeric molecule consisting of two ␣1(I) chains and one ␣2(I) chain. Fibroblasts and osteoblasts are the major collagen-producing cells in tissues, such as skin and bone, that present large amounts of type I collagen (41). Fibroblasts also appear to be the collagen-releasing cells in those tissues that contain smaller amounts of type I collagen. Hence, because the genes encoding the ␣1(I) and ␣2(I) chains are expressed in several cell types, at distinct stages of development, and under various physiologic conditions, their regulation is consequently complex (42)(43)(44). Interactions between a number of cis-regulatory elements and sequence-specific transacting factors are involved in the stage-specific and tissuespecific regulation of type I collagen gene transcription. Common cis-acting elements present in the genes encoding both the ␣1 and ␣2 polypeptides have been demonstrated to be responsible for their co-regulation (46).
Recent studies have examined collagen gene transcriptional regulation using the human and mouse ␣1(I) and ␣2(I) gene promoter regions (COL1A1 and COL1A2) and have led to the identification of several cis-elements and transcription factors that control constitutive expression. Thus, Ets factors (45), CCAAT-binding factor (46,47), K-ROX (48,49), NF-B (50), Sp-1 (51,52), and SMADs (53,54) have all been found to regulate type I collagen gene expression. These transcription factors may act alone or in concert, as is seen for example in the case of regulation elicited by the TGF-␤ signaling pathway which has a positive effect on COL1A2 promoter activity in cooperation with Sp-1 (55,56).
At present the mechanisms that regulate the expression of collagen genes in pathologic fibroblasts are not known. Cells isolated from fibrotic tissues have an activated phenotype. Both the COL1A1 and COL1A2 promoters exhibit severalfold higher activity when studied in scleroderma fibroblasts as compared with fibroblasts derived from healthy controls (57,58). In activated Ito cells derived from cirrhotic liver increased COL1A1 mRNA levels correlate with increased Sp-1 binding to the promoter (59); however, the specific cis-elements and transacting factors responsible for abnormal regulation of promoter activity in fibrotic disease have not been conclusively elucidated.
Previous work by our group (8,24) has shown that c-Myb, a transcription factor involved in differentiation and proliferation of hematopoietic cells, is expressed by scleroderma fibroblasts cultured in serum-deprived medium and is able to transactivate mouse and rat type I collagen gene promoters. Because a detailed study of the transcriptional regulation of the COL1A2 promoter by c-Myb may shed light on the pathogenesis of scleroderma and could lead to possible therapeutic strategies against this incurable disease, we felt it important to investigate more thoroughly the relationship between c-Myb and COL1A2 promoter regulation.
Previous studies (44,52) have identified several functional cis-acting elements in the 350-bp proximal region of the human COL1A2 gene promoter, whereas the data presented here demonstrate that c-Myb stimulates transcription by binding to an M-RR containing four MBSs located between Ϫ1400 and Ϫ1000 bp upstream of the initiation site. Multiple MBSs are often present in promoters regulated by c-Myb (33), so that the clustering of the COL1A2 promoter MBSs implies that the M-RR could play a key role in the regulation of COL1A2 gene expression by c-Myb. Consistent with this hypothesis, we have found that deletion of the M-RR abrogates the ability of c-Myb to regulate the COL1A2 promoter in transactivation assays. A, Northern blot analysis of RNA prepared from wild type (wt) and c-myb Ϫ/Ϫ MEFs. Eight micrograms of total RNA were loaded in each lane of a 1% agarose-formaldehyde gel. The membrane was hybridized successively with labeled cDNA probes for the ␣(2)-type collagen gene (COL1A2), fibronectin, the type III collagen gene (COL3), and ␤-actin. Autoradiographic exposure was for 24 h, except for COL1A2 gene (72 h). B, Northern blot analysis of RNA prepared from wild type (wt) and c-myb Ϫ/Ϫ MEFs transfected Ϯ the pGREMyb plasmid, expressing inducible full-length c-Myb (c-myb/ϩ). Cells were cultured in the presence of 0.2 M dexamethasone. C, the upper part shows Northern blot analysis of RNA prepared from wild type (wt) and c-myb Ϫ/Ϫ MEFs Ϯ TGF-␤ (2 ng/ml). The membrane was hybridized with radiolabeled cDNA probes for the ␣(2)-type collagen gene (COL1A2) and ␤-actin. The lower part is an RT-PCR analysis of c-myb RNA and a corresponding control reaction using primers specific for ␤-actin. The sequences of the four potential MBSs suggest that each could be involved in the specific binding of c-Myb to the COL1A2 promoter. Indeed, using gel shift experiments we have shown that c-Myb binds to each MBS; however, mutagenesis experiments demonstrated that MBS-4 plays a pivotal role in up-regulation of the COL1A2 promoter by c-Myb. In fact, the presence of the MBS-4 in the COL1A2 promoter appears to be sufficient to allow strong transactivation by c-Myb. This finding could be explained by the fact that the different means by which c-Myb activates its target gene promoters via MBS do not necessarily require cooperative interactions between multiple c-Myb molecules (60). However, it is noteworthy that the COL1A2 promoter with a mutation only in MBS-4 is still transactivated to some extent by c-Myb suggesting that MBS-1, MBS-2, and MBS-3 are perhaps important in cooperation with MBS-4.
We have shown that the likelihood that B-Myb cannot transactivate the COL1A2 promoter in mouse fibroblasts (24) holds true in human cells. Indeed, although c-Myb activates both the COL1A2 and synthetic KHK promoters, B-Myb, which should be capable of binding the same MBS, only activates the KHK construct and can partially inhibit the transactivation of the COL1A2 promoter that is driven by c-Myb. Two possible explanations for this distinction are as follows: (i) that B-Myb works as a repressor of type I collagen gene expression, as demonstrated by our experiments and by others (12); and (ii) that c-Myb may regulate human type I collagen expression in cooperation with an as yet unknown protein that does not operate in conjunction with B-Myb.
Our most direct evidence for a close link between c-Myb and COL1A2 gene expression has been provided by showing the lack of type I collagen expression in MEFs derived from c-myb Ϫ/Ϫ embryos. That the absence of c-Myb has a specific effect on the COL1A2 gene was apparent because the expression of type III collagen and fibronectin was normal in c-myb Ϫ/Ϫ MEFs.
Because the constitutive but not the TGF-␤-induced expression of the COL1A2 gene is abrogated in c-myb Ϫ/Ϫ MEFs, our data imply that TGF-␤-responsive cellular pathways are not affected in these cells. Furthermore, as already demonstrated in human dermal fibroblasts (24), wild type MEFs stimulated by TGF-␤ showed an increase in c-myb gene expression, implying that c-Myb may regulate both constitutive and cytokineinduced expression of the type I collagen gene. However, although TGF-␤ induces c-myb expression, c-myb is not obligatory for COL1A2 induction by this cytokine because TGF-␤ induces COL1A2 in the absence of c-myb in c-myb Ϫ/Ϫ MEFs.
Thus, in summary, c-myb is not the only transcription factor in the pathway that leads to collagen following TGF-␤ stimulation and c-myb, as shown in unstimulated cells (Fig. 7C), may be under the control of factors distinct from TGF-␤. These findings raise intriguing questions about the role of c-myb in the pathogenesis of SSc where it is overexpressed in quiescent SSc fibroblasts, and TGF-␤ is considered a key pathogenetic factor, and these findings suggest that the scenario is more complex than so far known, with distinct intracellular (collagen transcription factors) and extracellular factors (cytokines, growth factors, etc.) involved.
It is noteworthy that we have confirmed the results obtained in mouse embryonic fibroblasts lacking the c-myb gene by expressing a dominant negative c-Myb protein in NSF. As already demonstrated in 3T3 fibroblasts, such a dominant negative c-Myb protein suppresses transcription normally stimulated by endogenous c-Myb and blocks cell cycle progression in these cells (15). Induction of the dominant negative c-Myb in NSF led to the abrogation of COL1A2 gene expression, confirming that c-Myb plays a crucial role in the regulation of this gene transcription in both mouse (24) and, as shown here, human fibroblasts. In this regard, it is noteworthy that although the regions recognized by c-Myb in the mouse and human COL1A2 gene promoter have a different location, they display a high degree of homology. Furthermore, c-Myb modulates the expression of type I collagen genes in a species-specific manner, as shown in the case of other transcription factors (61). In fact, although multiple MBSs are present in the rat promoter of COL1A1 2 and in the promoter of mouse and human COL1A2 genes, no MBSs have been found in the human COL1A1 promoter. It is unlikely that this finding can be ascribed to the limited extension of the published sequences of the human COL1A1 promoter (62,63), because we have analyzed a region spanning from the transcription start site to Ϫ2000 bp which has the size of the promoter region of COL1A2 and rat COL1A1 genes where MBSs are present.
Thus, in summary, it can be speculated that in humans c-myb modulates the expression of type I collagen gene expression acting preferentially on the COL1A2 gene and that the precise mechanisms employed by c-Myb have evolved in a species-specific way (64).
In conclusion, in linking c-Myb to the expression of type I collagen in human fibroblasts, the present work emphasizes the physiological role of c-Myb in human fibroblasts and its potential importance in fibrotic conditions such as a scleroderma, in which the augmented production of collagen could be maintained by the deregulated expression of this gene (8). Further studies will be necessary to elucidate more clearly the role by c-Myb in fibrotic disorders.