Transcriptional Regulation of the Human a2(I) Collagen Gene COMBINED ACTION OF UPSTREAM STIMULATORY AND INHIBITORY CIS-ACTING ELEMENTS*

This study identifies three regions of the human a2(I) collagen promoter involved in the binding of nuclear factors. These regions include sequences from 2173 to 2155 (footprint I), 2133 to 2119 (footprint II), and 2101 to 272 (footprint III). A novel positive cis-element containing a TCCTCC motif was identified within footprint II. In addition, we demonstrated that a pyrimidine-rich region within footprint I is a binding site for a transcriptional repressor, and a CCAATmotif within footprint III is a binding site for a transcriptional activator. Comparative functional analysis of the cis-acting elements within the proximal 350 base pairs of this promoter, including previously characterized Sp1 binding sites at 2300, indicates that constitutive activity of this promoter is regulated equivalently by the three positive cis-acting elements at 2300, 2125, and 280. Mutations in the repressor site at 2160 increase constitutive activity by 4–6-fold. However, simultaneous mutations of the repressor site and the cis-regulatory element at either the 2300 or 2125 sites result in no increase in constitutive transcription activity suggesting interaction between the activators and repressor elements. In contrast, simultaneous mutation of the CCAAT motif and the repressor site results in about a 4-fold increase, suggesting that activation via the CCAAT motif may be independent of this repressor.

This study identifies three regions of the human ␣2(I) collagen promoter involved in the binding of nuclear factors. These regions include sequences from ؊173 to ؊155 (footprint I), ؊133 to ؊119 (footprint II), and ؊101 to ؊72 (footprint III). A novel positive cis-element containing a TCCTCC motif was identified within footprint II. In addition, we demonstrated that a pyrimidine-rich region within footprint I is a binding site for a transcriptional repressor, and a CCAAT motif within footprint III is a binding site for a transcriptional activator. Comparative functional analysis of the cis-acting elements within the proximal 350 base pairs of this promoter, including previously characterized Sp1 binding sites at ؊300, indicates that constitutive activity of this promoter is regulated equivalently by the three positive cis-acting elements at ؊300, ؊125, and ؊80. Mutations in the repressor site at ؊160 increase constitutive activity by 4 -6-fold. However, simultaneous mutations of the repressor site and the cis-regulatory element at either the ؊300 or ؊125 sites result in no increase in constitutive transcription activity suggesting interaction between the activators and repressor elements. In contrast, simultaneous mutation of the CCAAT motif and the repressor site results in about a 4-fold increase, suggesting that activation via the CCAAT motif may be independent of this repressor.
Collagen type I, the most abundant mammalian collagen, consists of two ␣1(I) chains and one ␣2(I) chain, which are coordinately expressed (1,2). The expression of type I collagen is strictly regulated during development and is tissue-specific (2). Excessive deposition of type I collagen is characteristic of many fibrotic disorders (3) and most likely results from transcriptional activation of collagen genes in response to cytokines and other factors present in the prefibrotic/inflammatory lesions.
Previous studies have characterized several responsive elements and cognate transcription factors involved in the regulation of collagen type I genes in murine fibroblasts (1). The most extensively studied transcription factor CBF is a heterotrimer consisting of subunits denoted A, B, and C, all of which are necessary for DNA binding (4,5). CBF is a transcriptional activator of mouse ␣1(I) and ␣2(I) collagen genes (4,5), where it binds to CCAAT motifs located between Ϫ80 and Ϫ84 in the ␣2(I) promoter and Ϫ96 and Ϫ100 in the ␣1(I) promoter (6,7). Another transcriptional activator of the murine ␣2(I) collagen promoter is a member of the CTF/NF1 family, which binds between Ϫ315 and Ϫ295 in the ␣2(I) promoter (6) and mediates TGF-␤ 1 stimulation of this promoter (8). A third site in the mouse ␣2(I) collagen promoter at Ϫ250 also contributes to basal promoter activity, but a cognate transcription factor has not yet been characterized (6). Both ␣1(I) and ␣2(I) collagen promoters in the mouse are negatively regulated by a transcriptional repressor termed IF1 that binds to two adjacent sites between Ϫ90 and Ϫ170 in the ␣1(I) promoter and between Ϫ165 and Ϫ155 in the ␣2(I) promoter (9,10). In addition, a novel factor of unknown function has been shown to interact with the region between Ϫ419 and Ϫ399 in the mouse ␣2(I) promoter (11).
Initial studies of the human ␣2(I) promoter have indicated that the Ϫ376 to Ϫ108-bp promoter segment is sufficient to direct a high level of transcription in human fibroblasts (12). However, transcriptional regulation of this promoter in human fibroblasts differs in some respects from transcriptional regulation of the murine ␣2(I) collagen promoter. For example, it has been demonstrated that CTF/NF1 does not bind to the human ␣2(I) collagen promoter (13). TGF-␤ stimulation of human ␣2(I) collagen promoter is mediated by a multiprotein complex that interacts with two distinct promoter segments (Ϫ330 to Ϫ286 and Ϫ271 to Ϫ255) termed TbRE (13). One of the proteins in this complex has been identified as Sp1 (13). Interestingly, TbRE also mediates inhibitory effects of tumor necrosis factor-␣ on transcriptional regulation of the human ␣2(I) collagen promoter (14). In addition, Sp1 mediates basal activity of the human ␣2(I) collagen promoter by binding to three responsive elements located between bp Ϫ303 and Ϫ271 (15). Two recent studies have also suggested the involvement of AP1 family members in mediating TGF-␤ effects in the mouse and human ␣2(I) collagen promoters (16,17). The human ␣2(I) collagen promoter region downstream from the Sp1 and TbRE binding sites has not been characterized. However, previous studies have indicated that a human ␣2(I) collagen promoter segment with a deleted or mutated Sp1-responsive element still maintained low levels of promoter activity suggesting the presence of additional regulatory sites in the downstream promoter region (15). Moreover, several response elements were mapped in the corresponding region of the murine ␣2(I) collagen promoter (6). It is, therefore, important to characterize this region in the human ␣2(I) collagen promoter.
In this study we have undertaken a systematic analysis of the Ϫ234 to Ϫ34-bp segment of the human ␣2(I) collagen promoter. Using in vitro and in vivo footprinting analyses we demonstrate that this promoter region contains three protein-DNA binding sites. Functional analyses of these sites suggest complex interactions of the repressor(s) and activator(s) in regulating constitutive transcriptional levels of the human ␣2(I) promoter.

MATERIALS AND METHODS
Cell Culture-Human fibroblasts were obtained from foreskins of healthy newborns, following institutional approval and informed consent. Primary explant cultures were established in 25-cm 2 culture flasks in DMEM supplemented with 10% fetal calf serum (FCS), 2 mM L-glutamine, and 50 g/ml amphotericin. Fibroblast cultures independently isolated from different individuals were maintained as monolayers at 37°C in 90% air, 10% CO 2 and studied between the third and sixth subpassages. IMR-90 (ATCC) fibroblasts were used in parallel with human foreskin fibroblasts. Both fibroblast types produced similar results in all assays.
Transient Transfections and Chloramphenicol Acetyltransferase Assays-Human fibroblasts were grown to 90% confluence in 100-mm dishes in DMEM with 10% FCS. Monolayers were washed and cells were transfected by the calcium phosphate technique (17) with 20 g of various promoter-chloramphenicol acetyltransferase constructs. pSV-␤galactosidase control vector (Promega) was co-transfected to normalize for transfection efficiency. After incubation overnight, the medium was replaced with DMEM containing 1% FCS. Incubation was then continued for 48 h. Cells were harvested in 0.25 M Tris-HCl, pH 8, and fractured by freeze-thawing. Extracts were normalized for protein contents as measured by Bio-Rad reagents and incubated with butyryl-CoA and [ 14 C]chloramphenicol for 90 min at 37°C, an assay condition predetermined to be within the linear range of chloramphenicol acetyltransferase activity for these samples. Butyrated chloramphenicol was extracted using organic solvent (2:1 mixture of tetramethylpentadecane and xylene) and quantitated by scintillation counting. Each experiment was performed in duplicate.
DNase I Footprinting and DNA Mobility Shift Assays-Nuclear extracts were prepared according to Andrews and Faller (18). Briefly, confluent cells from five 150-mm dishes were washed with phosphatebuffered saline and scraped into 1 ml of cold buffer A (10 mM HEPES-KOH, pH 7.9, at 4°C, 1.5 mM MgCl 2 , 10 mM KCl, 1 mM dithiothreitol, 0.2 mM phenylmethylsulfonyl fluoride, 1 mM sodium orthovanadate). The cells were allowed to swell on the ice for 10 min and then vortexed for 10 s. The tube was centrifuged for 3 min and supernatant was discarded. The pellet was resuspended in 80 l of cold buffer C (20 mM HEPES-KOH, pH 7.9, 25% glycerol, 420 mM NaCl, 1.5 mM MgCl 2 , 0.2 mM EDTA, 1 mM dithiothreitol, 0.7 mM phenylmethylsulfonyl fluoride, 1 mM sodium orthovanadate) and incubated on ice for 20 min for high salt extraction. Cellular debris was removed by centrifugation for 2 min at 4°C, and the supernatant fraction was stored at Ϫ80°C until use. The protein concentration of the extracts was determined using the Bio-Rad reagent.
Radioactive probes were generated by polymerase chain reaction (202-mer wild type and mutated) using [␥-32 P]ATP end-labeled primers. For DNase I footprinting, the binding reaction was performed for 30 min on ice in 50 l of 10 mM HEPES, pH 7.9, 50 mM NaCl, 0.5 mM dithiothreitol, 1 mM MgCl 2 , 4% glycerol, 12,000 cpm of labeled probe, 2 g of poly(dI-dC)⅐poly(dI-dC), and nuclear extract containing 120 g of protein. After subsequent addition of 5 l of 10 mM MgCl 2 , 5 mM CaCl 2 and incubation for 1 min at room temperature, 0.02 (for conditions without nuclear extract) to 3 units of DNase I (purchased from Boehringer Mannheim) were added. Digestion with DNase I was continued for 1 min at room temperature and terminated by adding 140 l of 192 mM sodium acetate, 32 mM EDTA, 0.14% SDS, and 64 g/ml yeast RNA. After phenol/chloroform extraction and subsequent precipitation with ethanol, the digested probe was dissolved in 3 l of 95% formamide containing 10 mM EDTA, 0.3% bromphenol blue, and 0.3% xylene cyanol and was electrophoresed on 8% polyacrylamide/urea gel along with Maxam-Gilbert GϩA sequencing reactions as size markers.
For DNA mobility shift assay, the binding reaction was performed for 30 min in 12-20 l of binding buffer containing 10,000 cpm of labeled probe, 2 g of poly(dI-dC)⅐poly(dI-dC), and nuclear extracts containing 3-5 g of protein.
In some assays double-stranded competitors were added. Oligonucleotides used as competitors were purchased from Operon Technologies. Separation of free radiolabeled DNA from DNAprotein complexes was carried out on a 5% nondenaturing polyacrylamide gel. Electrophoresis was carried out in 0.5 ϫ Tris borate electrophoresis buffer at 200 V at 4°C. Autoradiography was performed by overnight exposure to Kodak X-OMAT XAR2 film with intensifying screens at Ϫ80°C.
In Vivo Footprinting-Human fibroblasts were grown to confluence in DMEM, 10% FCS. The medium was then replaced with DMEM, 10% FCS containing 0.1% dimethyl sulfate (DMS) and incubated for 2 min. Cells were rinsed once with phosphate-buffered saline at 37°C, followed by three washes with phosphate-buffered saline for 30 s each with gentle shaking at 37°C. Cells were lysed on the plates using 1.5 ml of lysing buffer (300 mM NaCl, 50 mM Tris-Cl, pH 8.0, 25 mM EDTA, pH 8.0, 200 g/ml proteinase K, 0.2% SDS) and, after scraping, incubated for 4 h at 37°C. DNA was extracted twice with phenol, twice with phenol/chloroform/isoamyl alcohol (25:24:1), once with a mixture of chloroform/isoamyl alcohol (24:1), and once with ether. DNA was precipitated once each with isopropyl alcohol and ethanol and resuspended in TE buffer (10 mM Tris-HCl, pH 7.5, 1 mM EDTA). DNA for in vitro DMS treatment (naked) was prepared identically, with omission of DMS treatment. For the in vitro DMS treatment 25 l of 1% DMS was added to 175 l of DNA (100 g), and the mixture was incubated at room temperature for 2 min, followed by the addition of 50 l of ice-cold DMS stop buffer (1.5 M sodium acetate, pH 7.0, 1 M ␤-mercaptoethanol, 100 g/ml of yeast tRNA). Stop buffer was also added to 200 l of DNA from in vivo DMS treatment, and both DNA samples were precipitated with the addition of 750 l of ethanol in a dry ice bath for 30 min, and the pellets were washed with 75% ethanol after centrifugation in the microcentrifuge. The pellets were dissolved in 200 l of 1 M piperidine and incubated at 90°C for 30 min. Piperidine was removed by lyophilization, and DNA samples were resuspended in TE buffer, precipitated with isopropyl alcohol and ethanol, and washed with 75% ethanol. The DNA was dissolved in TE buffer, and the concentration was adjusted to 0.4 g/l.

Multiple Nuclear Factors Bind to the Human ␣2(I) Collagen
Promoter-We have previously analyzed the Ϫ353 to Ϫ234 region of the human ␣2(I) collagen promoter and have demonstrated the presence of three binding sites for a transcription factor related to Sp1 in this promoter region (15). To characterize DNA-protein interactions in the promoter region downstream from the Sp1 binding sites, we performed a DNase I protection assay using a promoter fragment from bp Ϫ235 to Ϫ34 and the nuclear extract from human fibroblasts. Three protected DNA segments were observed, which included two strongly protected regions located between bp Ϫ173 and Ϫ155 (footprint I), bp Ϫ133 and Ϫ119 (footprint II), and a broad weakly protected region between bp Ϫ101 and Ϫ72 (footprint III) (Fig. 1).
In Vivo Footprinting of the Human ␣2(I) Collagen Promoter-We next asked whether the protein-DNA interactions identified by in vitro assays also occur in intact cells. We used in vivo footprinting to analyze protein binding to the bp Ϫ234 to Ϫ34 region of the ␣2(I) collagen promoter in human fibroblasts. As shown in Fig. 2A differences were observed in the protein-DNA binding patterns between naked DNA treated with DMS in vitro (lane 1) and chromosomal DNA treated with DMS in vivo (lane 2) on the mRNA coding strand in the promoter region between bp Ϫ179 and Ϫ148 and bp Ϫ133 and Ϫ119. It appears that in vivo footprints correlate with the footprints I and II obtained in the DNase I protection assay (Fig. 1). A second set of primers was used to visualize binding to the noncoding strand (Fig. 2B). Protection and hypersensitivity sites were observed between bp Ϫ81 and Ϫ74 that correlate with the footprint III in Fig. 1. Thus, the three footprints identified by DNase I protection assays colocalized with in vivo footprints. The footprint III contains an inverted CCAAT motif between bp Ϫ84 and Ϫ80 that has been previously shown to bind the heterotrimeric transcription factor CBF in the mouse ␣2(I) collagen promoter (6). The sequences corresponding to footprints I and II have not been rigorously analyzed in either mouse or human ␣2(I) collagen promoters.
The Function of the CCAAT Motif Is Conserved between the Mouse and the Human ␣2(I) Collagen Promoters-To test binding to the CCAAT motif in the human ␣2(I) collagen promoter, we employed a mobility shift assay using a 202-bp DNA fragment (from Ϫ235 to Ϫ34) and human fibroblast nuclear extracts. Two strong protein-DNA complexes (denoted 2 and 3 in Fig. 3, lane 2)) and two relatively weak complexes (denoted 1 and 4, Fig. 3, lane 2) were observed. Formation of one of the complexes (denoted 3) was abolished by the addition of excess cold 21-mer containing the CCAAT motif (CAGCCCTCCCAT-TGGTGGAGG) (lane 4) but not by the 21-mer containing a single substitution mutation within the CCAAT motif (CAGC-CCTCCCTTTGGTGGAGG) (lane 5). Such mutation has been previously shown to abolish binding of CBF to the mouse collagen promoter (6). Furthermore, the smallest complex did not form when a 202-mer containing the same substitution mutation in the CCAAT motif was used as a probe (lane 6). These data indicate that a factor from human fibroblasts, probably a human homolog of murine CBF, binds specifically to the CCAAT motif in the human ␣2(I) collagen promoter. Previous studies in the mouse ␣2(I) collagen promoter have demonstrated that the CCAAT motif at position Ϫ80 contributes significantly to the constitutive activity of this promoter (6). We therefore tested the effects of substitution mutation in the CCAAT motif (CCAAA) on the activity of the human promoter. The substitution mutation introduced into the Ϫ353 bp ␣2(I) collagen promoter fragment linked to the CAT reporter gene resulted in a 90% reduction of the basal promoter activity when compared with the wild type promoter (set arbitrarily at 100%) (Fig. 4). Thus, similar to the mouse promoter, the CCAAT motif in the human promoter is the binding site for an activator, most likely a human homolog of CBF.
The TCCTCC Motif within Footprint II Mediates Binding of a Transcriptional Activator-To characterize binding of the nuclear factors to the promoter region corresponding to footprint II, we performed DNA mobility shift assay using the promoter fragment from Ϫ135 to Ϫ116 and nuclear extracts from foreskin fibroblasts. Three specific protein-DNA complexes were observed (Fig. 5B, lane 2). To further characterize DNA sequences involved in formation of these DNA-protein complexes, a series of oligonucleotides containing substitution mutations (see Fig. 5A) were used as competitors in the DNA mobility shift assay. M3 and M4 oligonucleotides, containing substitution mutations in the TCCTCC motif located in the middle of the protected region, failed to inhibit formation of the DNA-protein complexes, while formation of all three complexes was abolished by an excess of M1 and M6 oligonucleotides and reduced by an excess of M2 and M5 oligonucleotides. These results indicate that the TCCTCC motif located between bp Ϫ123 and Ϫ128 mediates binding of the nuclear proteins to this promoter region; and the flanking regions may be important for forming stable DNA-protein complexes.
To analyze the contribution of the TCCTCC motif to the basal collagen promoter activity, we introduced substitution mutations into TCC motifs using a Ϫ353 promoter construct. TCC motifs were mutated separately in the same fashion as oligonucleotides M3 and M4 used in the gel shift assay (Fig. 5). The promoter constructs carrying substitution mutations were analyzed by transient transfection assays (Fig. 4). Mutations in either TCC motif, when compared with the wild type promoter, resulted in significant reduction of the basal promoter activity, although mutation of the proximal site had a reproducible stronger effect (10.9 Ϯ 2.9%) than mutation of the distal site (17.8 Ϯ 2.8%).
Transcriptional Repressor Binds to the Footprint I in the Human ␣2(I) Collagen Promoter-To test binding to the promoter region corresponding to footprint I, we performed a mobility shift assay using the promoter fragment from bp Ϫ176 to bp Ϫ153 and nuclear extract from newborn foreskin fibroblasts. Three specific protein-DNA complexes were observed (Fig. 6B, lane 2). To further characterize DNA sequences involved in formation of these DNA-protein complexes, a series of oligonucleotides containing substitution mutations (see Fig.   6A) were used as competitors in the DNA mobility shift assay. M4 and M5 oligonucleotides, containing substitution mutations in the "TCCCCC" motif located between Ϫ164 and Ϫ159 in the promoter region, failed to inhibit formation of the DNAprotein complexes, while formation of both complexes was abolished by an excess of M1, M2, and M6 oligonucleotides and reduced by an excess of M3 oligonucleotide. These results indicate that the pyrimidine-rich sequence CCTTCCCCC within footprint I mediates binding of the nuclear proteins to this promoter region.
To analyze the contribution of this pyrimidine-rich motif to the basal promoter activity, we introduced substitution mutations into the TCCCCC motif using a Ϫ353 promoter construct. Proximal and distal triplets were mutated separately in the same fashion as oligonucleotides M5 and M4 used in the gel shift assay (Fig. 6A). The promoter constructs carrying substitution mutations were analyzed together with the wild type promoter by transient transfection assays. Mutations in either triplet, when compared with the wild type promoter, resulted in a significant increase in the basal promoter activity (Fig. 4). Mutation in the Ϫ164 to Ϫ162 (M4) site increased promoter activity to 580 Ϯ 140%, and mutation in the Ϫ161 to Ϫ159 (M5) site increased promoter activity to 624 Ϯ 183%. Based on this analysis we concluded that the pyrimidine-rich motif within footprint I is a binding site for the transcriptional repressor(s).
Functional Analysis of the Cis-regulatory Elements in the Human ␣2(I) Collagen Promoter-Previous analyses of the human ␣2(I) collagen promoter have demonstrated a strong positive cis-regulatory element located between Ϫ319 and Ϫ267 (15). This element, composed of three GC-rich motifs, has been characterized as a binding site for transcription factor Sp1. Thus, three positive and one negative cis-acting regulatory elements are present within the 350-bp fragment of the human ␣2(I) promoter. The three positive cis-acting elements at Ϫ300, Ϫ125, and Ϫ80 appear to contribute equally to promoter activity, as substitution mutation in each of these elements decreased promoter activity about 10-fold (15) (Fig. 4). To test the nature of the interactions between the Sp1 response element and the two proximal positive response elements at Ϫ125 and Ϫ80, we generated double mutants in which two elements were simultaneously altered (Fig. 7). The activities of these mutants were compared by transient transfection assays in human foreskin fibroblasts. Double mutations in Sp1 and CBF sites or Sp1 and "TCC" sites further diminished promoter activity about 2-fold to 5.6 Ϯ 2.1 and 4.5 Ϯ 1.9%, respectively (Fig. 7).
We then asked whether mutation in the repressor binding site is capable of reversing the inhibitory effects of mutations in the activators' sites. To test this, we constructed three double mutants in which substitution mutation in the Ϫ161 to Ϫ159 position (Fig. 6A, M5) was introduced simultaneously with substitution mutations in the Sp1, CBF, or "TCC" sites. As shown in Fig. 7, double mutants containing either Sp1 or "TCC" mutations exhibited low promoter activity similar to the basal activity of the corresponding single mutant. Interestingly, activity of the double mutant of the repressor and CBF binding sites exhibited about 4-fold higher basal activity than the single mutant in the CBF site (9.3 Ϯ 3 versus 36 Ϯ 7.1, p Ͻ 0.01). These data suggest that activation of the human ␣2(I) promoter via the CCAAT cis-regulatory element and repression via the TCCCCC element are independent of each other. On the other hand, activation either via Sp1 binding sites or the TCCTCC motif may involve interaction between the activator and repressor elements. DISCUSSION Previous studies of the human ␣2(I) collagen promoter have demonstrated that the 350-bp upstream promotor segment directs high constitutive promoter activity (12,15). The Ϫ303 to Ϫ255 region has been shown to contain the binding site for the transcription factor Sp1 and also the TGF-␤ response element (13,15). In this study we have identified three additional regions of the human ␣2(I) collagen promoter that bind nuclear factors by using in vitro and in vivo footprinting assays (Figs. 1  and 2). These regions include sequences from Ϫ173 to Ϫ155 (footprint I), Ϫ133 to Ϫ119 (footprint II), and Ϫ101 to Ϫ72 (footprint III) (Fig. 8).
Footprint II contains a novel cis-regulatory element that is not present in the mouse ␣2(I) collagen promoter. Functional analysis of this response element indicates that the sequence TCCTCC (between Ϫ128 and Ϫ123) contributes significantly to the basal promoter activity (Fig. 4). The nature of the binding proteins is presently unknown. A similar responsive element has been found in the osteonectin promoter, and a cognate factor termed "GGA factor" has been demonstrated by UV cross-linking to be a single 40-kDa protein (20). Binding to the TCCTCC motif in the human ␣2(I) collagen promoter appears to be more complex as indicated by the three DNA-protein complexes in the gel shift assay (Fig. 5). Purification of the transcription factors from human nuclear extracts that bind to the TCCTCC motif will facilitate further analysis of this novel responsive element.
Footprint III contains an inverted CCAAT motif, which was previously shown to bind CBF, an activator of the murine ␣2(I) and ␣1(I) collagen promoters (6). In this study we have demonstrated specific binding to the CCAAT motif in the human ␣2(I) promoter (Fig. 3). Functional analysis of the CCAAT motif in human fibroblasts indicates that a transcriptional activator, possibly a human homolog of CBF, binds to this sequence (Fig.  4). Analysis of footprint I reveals that a pyrimidine-rich motif located between Ϫ164 and Ϫ159 is a binding site for a transcriptional repressor (Figs. 4 and 6). The homologous regions in murine ␣1(I) and ␣2(I) collagen promoters also have been shown to bind a transcriptional repressor termed IF1 (10,21). Thus, the CCAAT motif and the pyrimidine-rich element between Ϫ164 and Ϫ159 are functionally conserved between human and mouse promoters. On the other hand, the Sp1 binding sites at Ϫ300 (15) and the TCCTCC motif within footprint II appear to be specific for the human ␣2(I) collagen promoter. Interestingly, we have not observed any protection in either in vitro or in vivo footprinting analyses in the Ϫ250-bp region in the human ␣2(I) collagen promoter (data not shown). The corresponding region has been shown to mediate activation of the murine ␣2(I) collagen promoter (6) and could be the locus of additional differences between the regulatory mechanisms of the human and mouse collagen genes.
Functional analysis of the four response elements identified in the 350-bp region of the human ␣2(I) collagen promoter reveals a complex interaction between the positive and negative regulatory elements. Mutation in the Sp1 binding sites decreases promoter activity by a factor of 10 (15). Likewise, mutations in the CCAAT and TCCTCC motifs result in a similar reduction of promoter activity (Fig. 4), indicating that these three response elements contribute equivalently to the constitutive promoter activity. This is further demonstrated by double mutants in either Sp1 and CCAAT sites or Sp1 and TCCTCC sites, which reduce the promoter activity by an additional factor of 2 ( Fig. 7), suggesting that proteins recognizing these sites have an additive effect in activating transcription. It is possible that combined actions of the three independent positive cis-regulatory elements are necessary to provide high levels of expression of the collagen type I gene in fibroblasts.
Previous studies demonstrated that activation of transcription by Sp1 involves direct interaction with one of the subunits of the transcription factor TFIID, the 110-kDa polypeptide TAFII 110 (22). A similar interaction was recently reported for CBF (5). Future studies are needed to elucidate possible interactions of CBF, Sp1, and other activators of collagen transcription with the basal transcription machinery and with each other in the regulation of the human ␣2(I) collagen promoter. Stimulation of transcription by CBF, Sp1, and other activators is counteracted by the action of the transcriptional repressor. Mutations in the TCCCCC motif located between Ϫ164 and Ϫ159 in the collagen promoter increase constitutive transcription levels by 5-6-fold. While the nature and mechanism of action of this repressor are presently unknown, our data suggest that it may work by interfering with the activity of the factors bound to the Ϫ300 and Ϫ125 sites. Interestingly, CBF seems to work independently of this repressor. These results indicate that under different physiological conditions, different combinations of available factors may dictate the expression levels of the ␣2(I) collagen gene. For example, the CBF-like factor may provide a pathway for bypassing the repressor activity or for finetuning the expression level.
In conclusion, this report presents a detailed analysis of the cis-regulatory elements of the human ␣2(I) collagen promoter. Characterization of the nuclear proteins interacting with these sequences is currently under way, but much work remains to be done to fully understand the basal and cytokine-stimulated transcriptional regulatory mechanism of the collagen type I gene in human fibroblasts. The in vivo footprinting assay pro- FIG. 7. Activities of the ␣2(I) collagen promoter constructs carrying double substitution mutations. Substitution mutations were introduced into the plasmids containing the Ϫ353 to ϩ58 ␣2(I) collagen promoter fragment cloned upstream from the chloramphenicol acetyltransferase reporter gene. At Ϫ300, all three GC boxes were mutated as described previously (15); at Ϫ160, the M5 mutation was used to construct double mutants (see text); at Ϫ125, the M4 mutation was used to construct double mutants (see text); at Ϫ80, the CCAAT motif was mutated as described in the text. These plasmids were used in transient transfection assays as described under "Materials and Methods." The diagram on the left indicates the mutated responsive elements (black boxes). The bar graph on the right shows the promoter activity of each mutated construct relative to the wild type Ϫ353 promoter construct, which was arbitrarily set at 100. The means Ϯ S.E. for separate experiments are shown at the right. The number of experiments used to calculate the mean is shown in parentheses. Asterisks indicate statistically significant results relative to the wild type promoter (p Ͻ 0.01, Mann-Whitney U test). ϱ indicates statistically significant results relative to the single mutant in the CCAAT motif (p Ͻ 0.01). vides an important tool to confirm in vitro observations in living cells.