Functional analysis of human alpha 1 ( I ) procollagen gene promoter . Differential activity in collagen-producing and-nonproducing cells and response to transforming growth factor beta 1

To gain a further understanding of the regulation of human type I collagen gene expression under physiologic and pathologic conditions, we characterized 5.3 kilobase pairs (kb) of the human alpha 1(I) procollagen gene promoter. A series of deletion constructs containing portions of the alpha 1(I) procollagen 5'-flanking region (with end points from -5.3 kb to -84 base pairs (bp)) ligated to the chloramphenicol acetyltransferase (CAT) reporter gene were transiently transfected into NIH/3T3 cells. Maximal CAT activity was observed with constructs having 5' end points from -804 to -174 bp. A further 5' deletion to -84 bp caused a marked reduction in CAT activity. Cells transfected with plasmids containing longer 5'-flanking fragments of the alpha 1(I) procollagen gene (-2.3 or -5.3 kb) showed reduced CAT activity compared with the -804 bp construct. The activity of the alpha 1(I) procollagen promoter was much lower in cells that do not normally express type I collagen (HeLa cells) compared with collagen-producing NIH/3T3 cells. The CAT activity of deletion constructs containing longer 5' regions than -84 bp was increased by approximately 2-fold in NIH/3T3 cells treated with transforming growth factor beta 1 (TGF beta 1). Electrophoretic mobility shift assays indicated that protein-DNA complex formation with a probe corresponding to the -170 to -80 bp fragment of the alpha 1(I) procollagen promoter was markedly enhanced in nuclear extracts prepared from TGF beta 1-treated fibroblasts as compared with untreated fibroblasts. The DNA binding activity stimulated by TGF beta 1 was specific for an Sp1-like sequence at positions -164 to -142 bp in the promoter. These results demonstrate that 1) there are both positive and negative cis-acting regulatory elements in the human alpha 1(I) procollagen promoter, 2) these regulatory regions function differently in collagen-producing and -nonproducing cells, 3) the alpha 1(I) procollagen promoter contains TGF beta 1-responsive sequences located between -174 and -84 bp from the transcription start site, and 4) TGF beta 1 caused marked stimulation of the DNA binding activity of a nuclear factor interacting with an Sp1-like binding site located within a region encompassing -164 to -142 bp of the alpha 1(I) procollagen promoter.

To gain a further understanding of the regulation of human type I collagen gene expression under physiologic and pathologic conditions, we characterized 5.3 kilobase pairs (kb) of the human d(1) procollagen gene promoter. A series of deletion constructs containing portions of the d(1) procollagen 5'-flanking region (with end points from -5.3 kb to -84 base pairs (bp)) ligated to the chloramphenicol acetyltransferase (CAT) reporter gene were transiently transfected into NIW3T3 cells. Maximal CAT activity was observed with constructs having 5' end points from -804 to -174 bp. A further 5' deletion to -84 bp caused a marked reduction in CAT activity. Cells transfected with plasmids containing longer S'-flanking fragments of the al(1) procollagen gene (-2.3 or -5.3 kb) showed reduced CAT activity compared with the -804 bp construct. The activity of the al(1) procollagen promoter was much lower in cells that do not normally express type I collagen (HeLa cells) compared with collagen-producing NW3T3 cells. The CAT activity of deletion constructs containing longer 5' regions than -84 bp was increased by = 2-fold in NIW3T3 cells treated with transforming growth factor p l (TGFf.31). Electrophoretic mobility shift assays indicated that protein-DNA complex formation with a probe corresponding to the -170 to -80 bp fragment of the al(1) procollagen promoter was markedly enhanced in nuclear extracts prepared from TGFpl-treated fibroblasts as compared with untreated fibroblasts. The DNA binding activity stimulated by TGFpl was specific for an Spl-like sequence at positions -164 to -142 bp in the promoter. These results demonstrate that 1) there are both positive and negative cis-acting regulatory elements in the human al(1) procollagen promoter, 2) these regulatory regions function differently in collagen-producing and -nonproducing cells, 3) the cul(1) procollagen promoter contains TGFpl-responsive sequences located between -174 and -84 bp from the transcription start site, and 4) TGFpl caused marked stimulation of the DNA binding activity of a nuclear factor interacting with an Spl-like binding site located within a region encompassing -164 to -142 bp of the al(1) procollagen promoter. * This work was supported by Grants AM 19106 and AR 42309 from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBankmIEMBL Data Bank with accession number(s) U06669. (1 To whom all correspondence should be addressed: Thomas Jefferson University, Rm ** Present address: Biomedical Sciences Div., Lawrence Livermore Normal fibroblasts modulate their biosynthetic activity to maintain a precise balance between the synthesis and degradation of their products during dynamic events of tissue remodeling such as development, differentiation, and repair. It has been suggested that abnormalities in these regulatory mechanisms may be responsible for the excessive extracellular matrix accumulation in a variety of fibrotic diseases such as systemic sclerosis and idiopathic pulmonary fibrosis. The collagens comprise a large family of widely distributed proteins that play a crucial role in the maintenance of the structural properties of the extracellular matrix. Despite the important structural and functional roles that the collagens play in normal tissues, the mechanisms that regulate their production are not completely understood. Variations in net collagen production occurring during growth and differentiation (1-31, viral (4-6) and chemical (7-9) transformation, cytokine and growth factor modulation (10)(11)(12)(13)(14), and spontaneous (15-17) and experimentally induced (18,19) fibrotic processes have been ascribed to fluctuations in the steady-state collagen mRNA levels. The regulatory mechanisms responsible for the maintenance of normal procollagen mRNA levels have not been completely elucidated. Most of the available evidence suggests that the principal mechanisms operate at the level of transcription, although translational control and changes in mRNAprocessing and stability may also play a role (reviewed in Refs. 20 and 21). The broad spectrum of regulatory signals that can influence collagen gene transcription suggests that the collagen gene promoters are responsive to various trans-acting pathways. Several putative regulatory elements that may determine the transcriptional efficiency of procollagen genes have been identified in their corresponding promoters. These include the consensus TATA and CCAAT motifs as well as additional regulatory elements , which are the potential targets for the action of promoter-specific transcription factors (26-31). Furthermore, the transcriptional activity of some procollagen gene promoters appears to be modulated by enhancer and silencer elements located 3' from the transcription start site (32,33). Detailed characterization of the cis-acting elements involved in modulation of collagen gene expression is crucial for understanding the physiologic and pathologic regulation of tissue collagen deposition. The purpose of the work reported here was to analyze the human crl(1) procollagen gene promoter in order to identify regulatory regions of the gene that may play a role in the modulation of its expression under normal and pathologic conditions. MATERIALS AND METHODS Construction ofPlasmids"Severa1 preliminary steps were necessary to prepare fragments with the appropriate restriction sites for ligations. The Hind111 site in pSV0 CAT' was changed to a KpnI site, and a 1.6 kb The abbreviations used are: CAT, chloramphenicol acetyltrans-KpnI-EamHI fragment was isolated. A 7 kb aUI) procollagen gene KpnI fragment extending from -5.3 to +1.7 kb was isolated from a human cosmid clone2 and subcloned into pUC19, and the KpnI site at -5.3 kb was changed to a NotI site to give pJH49. A HindIII-ThaI fragment, extending from -804 to +42 bp, was subcloned into the HindIII and SnaI sites of pUC19, excised as a HindIII-KpnI fragment, and subcloned into the HindIII and KpnI sites of Bluescript KS+ to give p804BS. The 4.5 kb NotI-Hind111 fragment from pJH49 (containing al(1) procollagen gene sequences from -5.3 kb to -804 bp) was then subcloned into the NotI and HindIII sites of p804BS to give p5.3kBS.
For the -5.3 and -804 bp constructs, the NotI-KpnI fragments from p5.3kBS and p804BS were ligated with the 1.6 kb KpnI-EarnHI CAT coding fragment into the NotI and BamHI sites of Bluescript KS+ to give p5.3kal CAT and p804al CAT, respectively. For the 2.3 kb construct, a 1.5 kb BarnHI-Hind111 fragment containing sequences from -2.3 kb to -804 bp was converted to a NotI-Hind111 fragment and inserted into the Not1 and HindIII sites of p804al CAT to give p2.3kal CAT. A series of deletions (from the HindIII site at -804 bp in p804al CAT toward the start of transcription site) was made following the exonuclease 111 digestion procedure of Henikoff (34). Exact deletion end points were determined by sequencing, and clones ending at -675, -463, -369, -174, and -84 bp were selected for analysis. A promoterless CAT plasmid, pOCAT, was prepared by removing the 846 bp HindIII-KpnI a1 procollagen promoter fragment from p804al CAT and religating.
The sequence from -2.3 kb to -804 bp of the al(1) procollagen gene was obtained from deletions made from the NotI site in p2.3kal CAT toward the initiation of transcription site (also following the exonuclease I11 digestion procedure). DNA sequencing of both strands was performed using the dideoxy chain termination procedure (35) with T7 polymerase (Sequenase, U. S. Biochemical Corp.) following the instructions provided by the supplier.
Cell ~ansfections-NIW3T3 cells (obtained from the ATCC) from subconfluent cultures were plated a t a density of 3 x lo5 celld60-mm dish and cultured in Dulbecco's modified Eagle's medium containing 10% calf serum. Transfections were performed 24 h later employing the calcium phosphate co-precipitation method, as described (36). The NIW 3T3 cells were transfected with 4 pg of al(1) procollagen promoter-CAT plasmid DNA and 0.2 pg of the pSV2AP plasmid DNA containing the SV40 early promoter and enhancer fused to a rat alkaline phosphatase cDNA (kindly provided by Dr. Kyong Yoon) (37) as an internal standard. m e r 4 h, the cultures were subjected to a 15% glycerol shock for 3 min at room temperature and then grown in fresh medium for 48 h before harvesting. HeLa cells (obtained from the ATCC) were plated in 60-mm plastic culture dishes (3 x lo5 cellddish), and 16 h later they were transfected with the plasmids with the same procedure as that used for NIW3T3 cells, except that HeLa cells were cultured in minimum Eagle's medium containing 10% fetal calf serum instead of calf serum following the glycerol shock.
Effects of TGFpI-To investigate the effects of TGFpl on the expression of the various al(1) procollagen promoter constructs, NIW3T3 cells were transfected with the deletion construct plasmids and the pSV2AP control plasmid as described above. Six hours aRer the transfections, the culture media were removed, and a fresh medium containing 10% of the serum substitute Serum Plus (Hazelton Biologics) and 12.5 ng/ml human recombinant TGFpl (NIH Bureau of Standards, or R & D Systems) was added to duplicate cultures. Control cells incubated in parallel received culture medium without TGFp1. The control and TGFp1treated cells were incubated for a n additional 48 h and then harvested for assays of CAT and alkaline phosphatase activities as described below.
Assays of CAT and Alkaline Phosphatase Actiuities-Cell extracts were prepared by mechanically detaching the cells in 1.5 ml of phosphate-buffered saline and by centrifugation a t 5000 x g for 3 min. The cell pellets were resuspended in 100 pl of 0.25 M Tris-HC1, pH 7.8,0.1% Triton X-100 and sonicated for 15 s. CAT activity was measured on 20-111 aliquots of supernatants following centrifugation of the extracts for 5 min in a Microfuge. The supernatants were heated for 10 min a t 60 "C prior to the assay of CAT activity by a liquid scintillation assay using 1 pCi of [3Hlacetyl-CoA/sample a s described (38). In some experiments, CAT activity in the cell extracts was determined employing [14C]chloramphenicol according to the method described by Gorman et al. (39). The conversion of chloramphenicol to acetylated forms was quantified by scraping the corresponding areas from the thin layer chromatography plates and measuring radioactivity by liquid scintillation spectroscopy. The two methods of assaying CAT activity yielded comparable ferase; kb, kilobase pairb); bp, base paids).
J. Hyland, unpublished observations. results. Alkaline phosphatase activity was assayed on 5 pl of the extracts as described by Yoon et al. (37). CAT activity in each sample was normalized to alkaline phosphatase activity to correct for differences in transfection efficiency. Electrophoretic Mobility Shift Assays-Nuclear extracts were prepared from confluent cultures of normal human skin fibroblasts that had been incubated with or without TGFpl (1 or 10 ng/ml) in the presence of 5% fetal calf serum for 48 h according to the method of Dignam et al. (40). Protein concentrations were determined by a dye binding assay (Bio-Rad), and the nuclear extracts were stored in 50-pl aliquots a t -70 "C until use. In order to ensure that fibroblast protein biosynthesis was stimulated by TGFPI, cultures were labeled with [14Clproline (1.5 pCi/ml) for 24 h prior to harvesting.
Only nuclear extracts prepared from cultures exhibiting a greater than 2-fold increase in ['4Clproline-labeled protein production over untreated cultures were used. In a typical experiment, a confluent 175-cm flask (lo7 cells) yielded -100 pg of crude nuclear protein.
Probes for electrophoretic mobility shiR assays were prepared by enzymatic digestion of the p804alCAT plasmid with NotI and KpnI. The resulting 431 bp fragment was gel-purified and further digested with NaeI and Hinfl. The resulting 219 bp NotI-Hinff (200-3) and 90 bp Hinff-NueI (200-2) fragments were purified and 5' end-labeled with [a-32PldCTP using the Klenow fragment of DNApolymerase I according to conventional procedures (41). Additional oligonucleotide probes were prepared by automated DNA synthesis (Applied Biosystems), and oligonucleotide S p l was obtained from Stratagene. Each oligonucleotide was annealed to its complementary strand and end-labeled with [y3'P1ATP using T4 polynucleotide kinase (Boehringer Mannheim). Competition studies were performed with molar excesses of unlabeled DNA fragments or double-stranded oligonucleotides. Electrophoretic mobility assays were performed using low ionic strength buffers as described (42). Binding reactions contained 5-10 pg of the nuclear extracts, 1-2 pg of double-stranded poly[d(I-C)I (Pharmacia LKB Biotechnology Inc.), and radiolabeled probes containing 50,000 cpm for a total of approximately 0.5 ng. The reaction mixtures were incubated for 30 min on ice in a buffer containing 60 m KCl, 10 m HEPES pH 7.9, 1 l l l~ dithiothreitol, 1 m EDTA, and 5% glycerol in a total volume of 10 p1. Protein-DNA complexes were resolved from the free probes in nondenaturing 5% polyacrylamide gels. Electrophoresis was performed in Tridglycine buffer (50 m Tris-HC1, 260 m glycine) at 120 V for 120 min. The gels were dried under vacuum and exposed to x-ray film with intensifying screens at -70 "C for 1648 h.

Nucleotide Sequence of -2.3 kb to -804 bp Region of al(Z)
Procollagen Gene-The nucleotide sequences of the human al(I) procollagen promoter and 5"flanking region to -804 bp have been previously reported (23). Our results extend the upstream sequence of the gene to the BarnHI site at -2.3 kb ( Fig. 1). The only species for which the al(1) procollagen sequence in this area has been reported to date is rat (43). A comparison of the nucleotide sequences of human and rat al(1) procollagen gene 5"flanking regions indicates a relatively low overall nucleotide identity (62%). However, there are several regions within the human and rat sequences (underlined in  experiments is shown in Fig. 2. The results indicate that CAT expression was maximal and constant in cells transfected with promoter constructs having 5' end points from -804 to -174 bp. A further 3' deletion to -84 bp caused a significant reduction in CAT activity, although even this short promoter was able to drive 10-fold higher CAT activity than the promoterless CAT plasmid pOCAT. Plasmids containing longer 5' fragments (extending to -2.3 or -5.3 kb) showed lower CAT activity than the -804 bp construct (50 and 20%, respectively). Comparison of Activity of d(I) Procollagen Promoter in Collagen-producing and -nonproducing Cells-In order to assess the activity of cul(1) procollagen promoter sequences in a cell line that does not normally exhibit high levels of type I collagen gene production, constructs with end points at -5.3 kb, -2.3 kb, and -804 bp were transfected into HeLa cells. As shown in Fig.  3, the CAT activity driven by these three constructs was markedly lower in HeLa cells than in NIW3T3 cells. In NIW3T3 cells, CAT activity driven by the -804 bp construct was about 26-fold higher than the CAT activity observed with the promot-erless construct pOCAT, whereas in HeLa cells the relative activity of the -804 bp construct was 10-fold lower. In contrast to NIW3T3 cells, transfection of constructs with longer 5' sequences into HeLa cells resulted in about 2-fold greater CAT activity than that obtained with the -804 bp plasmid. These results suggest that the positive and negative transcriptional regulatory regions of the cul(1) procollagen promoter function differently in collagen-producing and -nonproducing cells.

Localization of d(I) Procollagen Promoter Regions Respon-
sive to TGFPl-In order to localize regions within the al (1) procollagen promoter that are responsive to stimulation by TGFPl in the NIW3T3 cells, the effects of TGFPl on the CAT activity in cells transfected with promoter deletion constructs with 5' end points at -5.3 kb, -2. 3 kb, and -804 bp were examined. Incubation with TGFpl for 48 h resulted in an approximately 2-fold stimulation of CAT activity in cells transfected with each of the three constructs (Fig. 4), suggesting that TGFpl-responsive elements were located within the proximal region of the promoter 3' to the -804 bp end point. To further  identify TGFpl-responsive sequences within the proximal region of the al(1) procollagen promoter, NIW3T3 cells were transfected with constructs with 5' end points a t -675, -463, -369, -174, and -84 bp or with the promoterless construct pOCAT and incubated with TGFp1. The results indicate that TGFpl caused a greater than 2-fold increase in CAT activity driven by the constructs with 5"flanking regions longer than -84 bp. In contrast, TGFpl did not stimulate the CAT activity driven by the -84 bp promoter. A representative transfection experiment is shown in Fig. 5. These results indicated that TGFpl-responsive sequences are located between -174 and -84 bp of the al(1) procollagen promoter. changes in trans-acting protein-DNA interactions involving the al(1) procollagen promoter that were associated with TGFp1induced stimulation of collagen production, electrophoretic mobility shift assays were performed. For this purpose, nuclear extracts were prepared from untreated and TGFpl-treated fibroblasts. Two fragments (200-2 and 200-3) of the al(1) procollagen promoter region spanning the sequences from -389 to -80 bp relative to the transcription start site were used as probes (Fig. &I). Incubation of the 200-2 probe (-170 to -80 bp) with nuclear extracts from untreated or TGFpl-treated cells resulted in the formation of two complexes with retarded electrophoretic mobility (labeled R1 and R2), indicating the presence of nuclear DNA binding factorb) recognizing sequences within the probe (Fig. 7). The intensity of the R1 and R2 complexes determined by laser densitometry of the autoradiograms was increased 18-and 8-fold, respectively, when nuclear extracts prepared from fibroblasts that had been treated with TGFpl(10 ng/ml) were used (Fig. 7). Competition experiments indicated that a 4-fold molar excess of unlabeled 200-2 probe completely prevented the formation of the R1 complex and re-  duced by about 60% the formation of the R2 complex with the labeled 200-2 probe. A 25-fold molar excess of unlabeled 200-2 probe did not completely prevent the formation of the R2 complex (Fig. 7). These results suggest substantial differences in the DNA binding affinities of these nuclear proteins. Several DNA-protein complexes were detected when nuclear extracts were incubated with probe 200-3, spanning the sequences from

Binding of Nuclear Proteins from Control or TGFpl-treated Fibroblasts to Oligonucleotides with Spl-like Binding Sequences-A detailed analysis of the nucleotide sequence of the
-170 to -80 bp region of the human al(1) procollagen promoter revealed two regions of homology with consensus sequences recognized by the transcription factor Spl, here designated Spl.1 and Sp1. 2 (44, 45). To determine whether nuclear proteins from human fibroblasts may interact with these Sp-1-like elements in the al(1) procollagen promoter, electrophoretic mobility shift assays were performed with synthetic doublestranded oligonucleotides corresponding to the Spl.1 and Sp1.2 sequences as well as with oligonucleotides corresponding to promoter sequences that display homology with the AP-1 and NF"1 consensus binding elements. The location and nucleotide sequences of these potential regulatory elements in the al(1) procollagen promoter are shown in Fig. 6. Upon incubation of the Spl.1 probe with nuclear extracts prepared from untreated fibroblasts, two distinct DNA-protein complexes, designated R1 (upper, a duplex) and R2 (lower), were detected (Fig. 8A). In nuclear extracts prepared from fibroblasts treated with TGFp1, a reproducible 4-fold increase in the intensity of the R1 complex and a less consistent 30% increase in the intensity of the R2 complex were noted compared with the intensity of the corresponding complexes formed with nuclear extracts from untreated fibroblasts. Essentially no complex formation was detected when Sp1.2 was used as the probe (Fig. 8A). To determine whether the formation of the R1 and R2 complexes resulted from specific protein interactions with the Spl.1 probe, competition experiments with increasing amounts of unlabeled oligonucleotides containing the Spl.1 or Sp1.2 sequences or the Spl consensus binding site were performed. As shown in Fig. 8B, the formation of the R2 complex was completely prevented by the addition of excess unlabeled Spl.1 oligonucleotide, whereas excess Sp1.2 or consensus Spl oligonucleotides failed to compete for binding. In contrast, the formation of the R1 duplex was markedly reduced by each of the three competitor oligonucleotides. Incubation of the nuclear extracts with the AP-1.1 or NF-1.1 oligonucleotide probes resulted in the formation of single DNA-protein complexes (Fig.  8C). However, in contrast to results obtained with the Spl.1 probe, there were no differences in the intensities of protein-DNA complexes when nuclear extracts prepared from untreated or from TGFpl-treated fibroblasts were used. These results indicate that a nuclear trans-acting protein binds to the Spl.1 element at -164 to -142 bp on the antisense strand of the al(1) procollagen promoter and that this DNA binding activity is increased by treatment of human fibroblasts with TGFp1. An additional DNA-protein complex (R2) is formed with this Sp-llike element, but this binding activity is less consistently increased by TGFP1.

DISCUSSION
The mechanisms involved in the regulation of collagen production under normal or pathologic conditions are not completely understood (reviewed in Refs. 20 and 21). Although the synthesis of most proteins in eukaryotic cells appears to be regulated at a transcriptional level, post-transcriptional events, such as the regulation of the stability of newly synthesized mRNA, may play an important role under certain conditions (46,47). Studies of the transcriptional regulation of various collagen genes in human and rodent cells in vitro have demonstrated the presence of regulatory elements located immediately 5' upstream of the transcription initiation site (22)(23)(24)(25)(26). In addition, an enhancer element located within the first intron has been identified in the type I collagen genes (32)(33)(34). Sequences located far upstream of the initiation of the transcription site may also be involved in the regulation of expres-  (1) procollagen gene sequences encompassing -3700 to +1400 bp and found that a short segment of the promoter (220 bp upstream from the start of the transcription site) was sufficient for expression of the gene, whereas further upstream flanking sequences had a negative effect on transcription.
In order to identify upstream elements that may be involved in the regulation of transcription of the human al(1) procollagen gene, we determined the nucleotide sequence of the promoter region encompassing from -804 to -2292 bp. Comparison of the newly obtained sequence with that of the promoter region in the rat crl(1) procollagen gene (43) indicated less than 65% overall nucleotide identity. However, several regions with highly conserved sequences in the two species were found between -1900 and -1540 bp (underlined in Fig. l). The high degree of nucleotide sequence conservation between the two species in this region suggests that these sequences may have important regulatory functions. In the rat gene, this region contains two repeats that are variations of the palindromic of the human gene reported here demonstrated two identical palindromes localized at -1687 and -1063 bp. Other putative regulatory sequences were also identified in the newly obtained sequence. These included two S p l binding sites (GGGCGG) a t positions -2168 and -1614 bp and one NF-1 (half-site) binding sequence (GCCAA) at position -830 bp (reverse strand). gressively decreasing activity (Fig. 2). These results are similar to those reported by Rippe et al. (49) for the murine al(1) procollagen gene.
In order to examine if there were differences in the regulation of al(1) procollagen gene transcription in cells that constitutively produce high levels of collagen and cells that normally do not disdav exmession of interstitial collagen genes, we com--r m r . " . pared the expression of the promoter-CAT constructs following their transfection into NIW3T3 cells or into HeLa cells. Marked differences in the expression of the promoter were observed between these two types of cells. Transfection of noncollagen-producing HeLa cells resulted in relatively low levels of CAT activity, which, in marked contrast to NIW3T3 cells, was increased when promoter constructs containing -2.3 and -5.3 kb upstream sequences were examined (Fig. 3). The dif-" 17 bp Sp1.2 double-stranded oligonucleotides encompassing the human al(1) procollagen promoter sequences -164 to -142 bp or -93 to -77 bp, respectively, were end-labeled and used as probes in electrophoretic mobility shift assays. Nuclear extracts were prepared from normal human skin fibroblasts that had been treated with TGFpl (10 ng/ml) or left untreated for 48 h. Labeled probes were incubated with nuclear extracts under identical conditions, as described in the legend to Fig. 7.  1 and -175 to -166 bp and -102 to -85-bp,respectIvely, were used. The ferences in expression of the various constructs when transfected into collagen-producing and -nonproducing cell types suggest that the intracellular milieu plays an important role in the regulation of the rates at which the collagen genes are transcribed. Furthermore, these results suggest that cell-specific transcriptional factors that act on the upstream sequences of the al(1) procollagen gene may influence the expression of this gene.
To further investigate the functional role that the al(1) procollagen promoter may play during dynamic events requiring remodeling of the extracellular matrix, we examined the effects of TGFpl on the expression of the various promoter constructs. It has been suggested that this growth factor acting in autocrine and paracrine fashion plays a n important role in the regulation of extracellular matrix metabolism during development, differentiation, and repair (reviewed in Refs. 50 and 51). TGFp causes marked accumulation of collagen in vivo and in vitro (52,53). In animal models of hepatic and pulmonary fibrosis, TGFp mRNA is expressed in high levels in tissues prior to the development of fibrosis (54)(55)(56). Furthermore, recent observations implicate TGFp in the pathogenesis of various human diseases characterized by exaggerated fibrosis (57)(58)(59)(60).
In the present study, we found that TGFpl stimulated CAT activity driven by al(1) procollagen promoter constructs with 5' end points distal to -84 bp in transiently transfected NIW3T3 cells. These results indicate that TGFpl-responsive sequences in the human al(1) procollagen promoter are located between -174 and -84 bp from the transcription start site.
To determine if TGFpl induced alterations in DNA binding activity in fibroblasts, we examined DNA-protein complex formation in nuclear extracts prepared from TGFpl-treated and untreated cells by electrophoretic mobility shift analysis using DNA probes spanning the proximal region of the al(1) procollagen promoter. The results showed a marked increase in DNA binding activity that was specific for the probe encompassing nucleotides -170 to -80 bp in nuclear extracts from TGFp1treated cells. This region of the human al(1) procollagen promoter contains a binding site for S p l at -87 bp and an element with complete identity to the 3' portion of the canonical NF-1binding motif (GCCAA), located in reverse orientation at -95 bp. These overlapping potential binding sites have been shown to be important in the basal expression of the mouse al (1) procollagen gene (29). In addition, a sequence similar to the AP-1-binding site (5"GAGTCC) is located in reverse orientation at -165 bp, and a sequence of perfect identity with a binding site for an Spl-like factor from the GC2 cis-acting element of the rat growth hormone and the F2 element of the human growth hormone genes (GGGAGGAG) is found at -148 bp in reverse orientation (45). When sequences in the -170 to -80 bp region of the human al(1) procollagen promoter homologous to these consensus binding sites were examined in electrophoretic mobility shift assays, only the Spl.1 probe (corresponding to the sequence identical with GC2 and F2) demonstrated increased DNA-protein complex formation with nuclear extracts from TGFpl-treated when compared with untreated fibroblasts. The formation of the R1 complex was abolished or decreased in competition assays with all three Splrelated oligonucleotides, in contrast to the R2 complex, which was prevented only by the Spl.1 oligonucleotide competitor. These observations suggest that at least one component of the R1 complex may be a member of the Spl family of transcription factors. Of interest in this regard is the previous demonstration that Spl interacted with cis-acting elements within the first intron (34) or the promoter (61) of the al(1) procollagen gene. Spl, originally described as a factor required for SV40 transcription, binds to "GC boxes," which are found in the promoters of many mammalian genes (44). Although ubiquitously ex-pressed, recent evidence points to transcriptional regulation as an important role for S p l (62). Overexpression of S p l was shown to inhibit mouse al(1) procollagen promoter activity in transiently transfected NIW3T3 cells (29). However, when cotransfected into Drosophila melanogaster Schneider L2 cells, which are devoid of homologs of mammalian transcription factors, S p l overexpression caused potent trans-activation of the al(1) procollagen promoter (63).
In the rat al(1) and the mouse a2(I) procollagen genes, TGFP-responsive elements resembling the NF-1 consensus sequence have been described (64,65). In the rat al(1) procollagen gene, the putative "TGFp activating element" was reported to be located 1.6 kb upstream from the transcription start site (65). The element contains the 3' portion of the canonical NF-1 sequence GCCAAG also found in the mouse a2(I) promoter. However, in contrast to the mouse a2(1) procollagen gene, stimulation of rat al(1) procollagen gene expression by TGFpl does not apear to involve NF-1 binding to its cognate DNA element (66). Analysis of the nucleotide sequence of the human al(1) procollagen gene indicates that a NF-1-like sequence similar to the rat TGFp activating element is located at -1718 bp (5'-TGCCCACGGCCAGC). However, our results indicate that deletion of a 1.5 kb fragment including this element did not prevent stimulation of promoter activity by TGFpl (Fig. 41, suggesting that in NIW3T3 cells this NF-1-like element is not involved in TGFpl-induced transcriptional activation of the al(1) procollagen gene. The nature of trans-acting factors, which are implicated in the modulation of the expression of diverse TGFP-responsive genes, is not completely understood. Recent studies suggest that, depending on the cell type and the gene that is regulated, a variety of distinct nuclear proteins may be involved (67). Moreover, transcriptional regulation by TGFp is likely to be a complex process, associated with the combinatorial interactions of ubiquitous and inducible transacting factors. The identification of cell-specific DNA binding factors and of the stimulatory and inhibitory cis-acting elements in the promoter of the human al(1) procollagen gene that are targets for these factors will permit a better understanding of the complex mechanisms that modulate the transcriptional activity of collagen genes during physiologic processes of development and repair as well as in a variety of diseases characterized by excessive collagen production.